Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The cell cycle regulates mouse and human pancreas development Krentz, Nicole Angela Jane 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_february_krentz_nicole.pdf [ 27.11MB ]
Metadata
JSON: 24-1.0362895.json
JSON-LD: 24-1.0362895-ld.json
RDF/XML (Pretty): 24-1.0362895-rdf.xml
RDF/JSON: 24-1.0362895-rdf.json
Turtle: 24-1.0362895-turtle.txt
N-Triples: 24-1.0362895-rdf-ntriples.txt
Original Record: 24-1.0362895-source.json
Full Text
24-1.0362895-fulltext.txt
Citation
24-1.0362895.ris

Full Text

	THE	CELL	CYCLE	REGULATES	MOUSE	AND	HUMAN	PANCREAS	DEVELOPMENT		by	Nicole	Angela	Jane	Krentz		B.Sc.,	The	University	of	British	Columbia,	2011		A	THESIS	SUBMITTED	IN	PARTIAL	FULFILLMENT	OF	THE	REQUIREMENTS	FOR	THE	DEGREE	OF		DOCTOR	OF	PHILOSOPHY	in	THE	FACULTY	OF	GRADUATE	AND	POSTDOCTORAL	STUDIES	(Cell	and	Developmental	Biology)		THE	UNIVERSITY	OF	BRITISH	COLUMBIA	(Vancouver)			December	2017		©	Nicole	Angela	Jane	Krentz,	2017	  ii Abstract	 Diabetes	is	caused	by	a	loss	or	dysfunction	of	insulin-producing	pancreatic	b-cells.	A	potential	treatment	for	diabetes	is	to	replace	these	cells	through	transplantation.	As	there	is	a	shortage	of	donor	tissue,	efforts	to	generate	an	unlimited	source	of	functional	insulin-producing	b-cells	from	human	embryonic	stem	cells	(hESCs)	are	ongoing.	During	pancreas	development,	proliferating	pancreatic	progenitors	activate	Neurog3,	exit	the	cell	cycle,	and	differentiate.	The	overarching	goal	of	this	thesis	was	to	understand	the	role	of	the	cell	cycle	in	regulating	Neurog3	expression	and	endocrine	cell	fate.	First,	the	length	of	each	cell	cycle	phase	of	pancreatic	progenitors	was	measured	using	cumulative	EdU	labelling,	determining	an	increase	in	G1	length	in	Pdx1+	progenitors	from	4.5±0.4	to	7.2±0.8	hours	between	embryonic	day	(E)11.5	and	E13.5.	Next,	two	mouse	models	were	used	to	show	that	cell	cycle	lengthening	within	pancreatic	progenitors	stimulates	endocrine	differentiation.	Kras	heterozygous	loss-of-function	mice	have	increased	endocrine	cell	genesis	that	was	correlated	with	an	increase	in	progenitor	cell	cycle	length.	Ectopic	expression	of	the	cyclin-dependent	kinase	inhibitor	Cdkn1b	in	Sox9+	progenitor	cells	resulted	in	a	2.7-fold	increase	in	the	number	of	Neurog3+	cells.	As	Cdkn1b	is	an	inhibitor	of	G1-S	cyclin-dependent	kinases	(Cdks),	the	effect	of	directly	inhibiting	Cdk2,	Cdk4	and	Cdk6	on	endocrine	differentiation	was	investigated.	Treating	embryonic	pancreata,	ex	vivo,	for	24	hours	with	Cdk	inhibitors	resulted	in	a	3-fold	increase	in	the	number	of	Neurog3+	cells.	To	investigate	the	consequences	of	CDK	inhibition	on	human	endocrine	differentiation,	a	NEUROG3-2A-eGFP	(N5-5)	knock-in	reporter	CyT49	hESC	line	was	generated	using	CRISPR-Cas9.	CDK	inhibition	increased	the	number	of	GFP+	endocrine	progenitor	cells	1.7-fold.	These	findings	  iii suggest	that	G1	lengthening	is	required	for	normal	mouse	and	human	organogenesis	and	that	cyclin-dependent	kinases	act	directly	to	reduce	Neurog3	protein.	In	the	final	chapter,	single-cell	transcriptomics	was	used	to	profile	the	gene	expression	and	cell	populations	present	during	mouse	and	human	endocrine	development.	In	conclusion,	these	studies	show	that	progenitor	cell-cycle	G1	lengthening,	through	its	actions	on	stabilization	of	Neurog3,	is	an	essential	determinant	of	normal	endocrine	cell	genesis.		 	  iv Lay	summary	Diabetes	is	a	disease	of	uncontrolled	blood	glucose	levels	caused	by	the	loss	or	dysfunction	of	insulin-producing	b-cells.	Current	treatment	for	diabetes	is	to	replace	the	function	of	these	cells	through	insulin	therapy.	However,	it	is	difficult	to	control	blood	glucose	levels	with	insulin	therapy,	potentially	resulting	in	damage	to	nerves,	eyes,	and	other	organs.	Replacing	the	lost	or	damaged	b-cells	through	transplantation	reduces	the	risk	of	these	complications	but	is	limited	by	donor	availability.	Therefore,	there	is	a	great	deal	of	promise	and	interest	in	producing	alternative	sources	of	b-cells	for	transplantation.	This	requires	an	in	depth	understanding	of	how	b-cells	form	during	normal	fetal	development.	Using	mouse	and	human	models,	this	thesis	investigates	the	role	of	the	cell	cycle	in	b-cell	development.	This	study	may	improve	current	methods	in	generating	an	unlimited	source	of	b-cells	for	diabetes	treatment.		 	  v Preface	Animal	studies	were	reviewed	and	approved	by	the	University	of	British	Columbia	Committee	on	Animal	Care	under	protocols	A13-0184	and	A14-0163.		All	studies	in	this	thesis	were	conceived	and	designed	by	N.A.J.	Krentz	and	F.C.	Lynn.	N.A.J.	Krentz	performed	experiments,	analyzed	data,	and	wrote	the	manuscripts	included	in	this	thesis.	Technical	assistance	was	provided	by	the	following	people:		 Chapter	3:	M.	Tang,	A.	Watanabe,	M.	Yu,	and	A.	Branch	assisted	with	embryo	dissections,	paraffin	embedding,	sectioning,	and/or	staining.	R	pipeline	used	for	statistical	analysis	(Appendix	A)	was	provided	by	R.	White,	T.	Zhao	and	J.	Petkau.		 Chapter	4:		A.	Branch	and	A.	Sharon	assisted	with	timed	matings,	embryo	dissections,	paraffin	embedding,	sectioning,	and/or	staining.	C.	Nian	performed	western	blots.	F.	C.	Lynn	made	Sox9-rtTA	mice.		 Chapter	5:	F.C.	Lynn	designed	and	generated	constructs	for	OCT4	TALEN	and	OCT4	and	NEUROG3	CRISPR-Cas9	genome	editing	studies.	C.	Nian	performed	western	blots.		 Chapter	6:	F.C.	Lynn	designed	and	generated	constructs	for	FUCCI	transgenic	hESC	lines	and	made	human	induced	pluripotent	stem	cell	line.	C.	Nian	performed	western	blots.		 Chapter	7:	E.E.	Xu	generated	mouse	libraries	and	established	bioinformatic	pipeline	used	for	data	analysis	(Appendix	B	and	C).	E.E.	Xu,	B.H.	Hoffman	and	F.C.	Lynn	assisted	with	data	analysis.		Experiments	presented	in	Chapters	3,	4,	5,	&	6	have	been	published	in	Krentz	NA*,	van	Hoof	D*,	Li	Z,	Watanabe	A,	Tang	M,	Nian	C,	German	MS,	Lynn	FC.	(2017).	Phosphorylation	of	NEUROG3	links	endocrine	differentiation	to	the	cell	cycle	in	pancreatic	progenitors.	Developmental	Cell	41,	129-142.	(*	co-authors)	Experiments	presented	in	Chapter	5	have	been	published	in	Krentz	NA,	Nian	C,	Lynn	FC.	(2014).	TALEN/CRISPR-Mediated	eGFP	Knock-in	Add-On	at	the	OCT4	Locus	Does	Not	Impact	Differentiation	of	Human	Embryonic	Stem	Cells	towards	Endoderm.	PLoS	ONE	9(12):	e114275.		Research	collected	in	the	preparation	of	the	following	article	was	used	in	Chapter	1:	Krentz	N.A.J.,	Lynn	F.C.	(2016).	Using	CRISPR-Cas9	Genome	Editing	to	Enhance	Cell	Based	Therapies	for	the	Treatment	of	Diabetes	Mellitus.	In:	Turksen	K.	(eds)	Genome	Editing.	Springer,	Cham	  vi Table	of	contents	  Abstract	....................................................................................................................................................	ii	Lay	summary	.........................................................................................................................................	iv	Table	of	contents	.................................................................................................................................	vi	List	of	tables	...........................................................................................................................................	x	List	of	figures	.........................................................................................................................................	xi	List	of	abbreviations	.........................................................................................................................	xiv	Acknowledgements	...........................................................................................................................	xvi	Dedication	.........................................................................................................................................	xviii	Chapter	1:	Introduction	......................................................................................................................	1	1.1	 The	role	of	the	pancreas	in	regulating	glucose	metabolism	...............................................	1	1.2	 Diabetes	mellitus	...................................................................................................................................	1	1.3	 Overview	of	mouse	pancreas	development	...............................................................................	5	1.3.1	 Gastrulation	and	the	formation	of	the	definitive	endoderm	germ	layer	.............	5	1.3.2	 Specification	of	endoderm	to	the	pancreatic	lineage	...................................................	6	1.3.3	 Primary	transition	and	growth	of	the	pancreatic	buds	..............................................	7	1.3.4	 Secondary	transition	and	endocrine	cell	differentiation	...........................................	9	1.4	 Differences	in	mouse	and	human	pancreas	development	...............................................	10	1.5	 Human	embryonic	stem	cells	........................................................................................................	11	1.6	 Using	human	embryonic	stem	cells	as	a	model	of	human	pancreas	development	11	1.7	 Generating	hESC	reporter	lines	using	genome	editing	technology	..............................	12	1.8	 TALEN	technology	.............................................................................................................................	13	1.9	 CRISPR-Cas9	technology	.................................................................................................................	14	1.10	 Introduction	to	the	cell	cycle	.........................................................................................................	15	1.11	 The	role	of	cell	cycle	proteins	during	pancreas	development	........................................	17	1.12	 G1	lengthening	and	differentiation	............................................................................................	18	1.13	 Cell	cycle	regulation	of	Neurog	family	members	..................................................................	19	1.14	 Thesis	investigation	..........................................................................................................................	19	Chapter	2:	Materials	and	methods	................................................................................................	21	2.1	 Animals	...................................................................................................................................................	21	  vii 2.2	 EdU	cumulative	labelling,	embryo	collection	and	tissue	preparation	........................	21	2.3	 Immunostaining	and	imaging	.......................................................................................................	22	2.4	 EdU	quantification	and	calculation	of	cell	cycle	lengths	...................................................	23	2.5	 Explant	dissection,	culture	and	in	situ	immunostaining	...................................................	23	2.6	 Maintenance	of	pluripotent	cells	.................................................................................................	24	2.7	 Generation	of	induced	pluripotent	stem	cells	.......................................................................	24	2.8	 Generation	of	OCT4-eGFP	pluripotent	stem	cell	lines	........................................................	25	2.9	 Generation	of	NEUROG3-2A-eGFP	pluripotent	stem	cell	line	..........................................	26	2.10	 Generation	of	FUCCI	transgenic	pluripotent	stem	cell	line	..............................................	27	2.11	 In	vitro	differentiation	of	hESC	.....................................................................................................	27	2.12	 RNA	isolation	and	RT-PCR	analysis	............................................................................................	29	2.13	 Western	blot	analysis	.......................................................................................................................	29	2.14	 Flow	cytometry	and	FACS	..............................................................................................................	30	2.15	 Immunocytochemical	analyses	....................................................................................................	31	2.16	 Nanostring	nCounter	SPRINT™	protocol	.................................................................................	32	2.17	 Preparing	cells	for	single	cell	RNA-sequencing	....................................................................	32	2.18	 Generating	scRNA-sequencing	libraries	..................................................................................	33	2.19	 Bioinformatic	analysis	of	scRNA-sequencing	data	..............................................................	33	2.20	 Statistical	analyses	.............................................................................................................................	34	Chapter	3:	Changes	in	cell	cycle	parameters	during	early	mouse	pancreatic	development	........................................................................................................................................	39	3.1	 Background	...........................................................................................................................................	39	3.2	 Results	.....................................................................................................................................................	41	3.2.1	 Increasing	cell	cycle	length	during	early	mouse	pancreatic	development	......	41	3.2.2	 Changes	in	cell	cycle	length	between	tip	and	trunk	progenitor	cells	................	44	3.2.3	 The	pancreatic	epithelium	has	a	population	of	cells	that	do	not	undergo	S-phase	during	early	development	......................................................................................	46	3.2.4	 Neurog3+	cells	mainly	arise	from	cycling	progenitor	population	......................	49	3.3	 Discussion	..............................................................................................................................................	51	Chapter	4:	Altering	cell	cycle	length	in	vivo	changes	endocrine	cell	differentiation	...	55	  viii 4.1	 Background	...........................................................................................................................................	55	4.2	 Results	.....................................................................................................................................................	57	4.2.1	 Alterations	in	KRAS	signalling	changes	cell	cycle	length	........................................	57	4.2.2	 Acute	ectopic	expression	of	Cdkn1b	at	E12.5	increases	the	formation	of	Neurog3+	endocrine	progenitors	.....................................................................................	60	4.2.3	 Long	term	upregulation	of	Cdkn1b	from	E10.5	to	E12.5	increases	the	number	of	Gcg+	endocrine	cells	..........................................................................................................	61	4.2.4	 Long	term	upregulation	of	Cdkn1b	from	E12.5	to	E14.5	increases	the	number	of	Gcg+	and	Ins+	endocrine	cells	.......................................................................................	63	4.2.5	 Inhibition	of	cyclin	dependent	kinases	(CDKs)	increases	Neurog3+	cells	during	mouse	pancreas	development	............................................................................	64	4.3	 Discussion	..............................................................................................................................................	66	Chapter	5:	Using	genome	editing	to	generate	reporter	human	embryonic	stem	cell	lines	.........................................................................................................................................................	69	5.1	 Background	...........................................................................................................................................	69	5.2	 Results	.....................................................................................................................................................	71	5.2.1	 Generation	of	OCT4-eGFP-2A-Puro	hESC	lines	using	genetically	engineered	nucleases	.....................................................................................................................................	71	5.2.2	 OCT4-eGFP	reporter	lines	have	normal	OCT4	protein	expression	and	stem	cell	phenotype	...........................................................................................................................	74	5.2.3	 OCT4-eGFP	reporter	lines	form	the	definitive	endoderm	germ	layer	..............	76	5.2.4	 eGFP	expression	mirrors	OCT4	protein	levels	in	OCT4	reporter	lines	.............	80	5.2.5	 Ability	to	differentiate	into	pancreatic	progenitors	is	maintained	in	OCT4	reporter	lines	.............................................................................................................................	81	5.2.6	 Generation	and	characterization	of	NEUROG3-2A-eGFP	knock-in	reporter	CyT49	hESC	lines	.....................................................................................................................	85	5.2.7	 Characterization	of	NEUROG3-2A-eGFP	reporter	lines	............................................	86	5.3	 Discussion	..............................................................................................................................................	87	Chapter	6:	CDK	inhibition	stabilizes	NEUROG3	during	human	embryonic	stem	cell	differentiation	.....................................................................................................................................	90	  ix 6.1	 Background	...........................................................................................................................................	90	6.2	 Results	.....................................................................................................................................................	92	6.2.1	 Neurog3	expression	peaks	between	stage	5	and	6	of	hESC	differentiations	.	92	6.2.2	 CDK	inhibition	increases	the	proportion	of	cells	in	G1	and	the	expression	of	pancreas-specific	genes	........................................................................................................	94	6.2.3	 CDK	inhibition	increases	the	number	of	GFP+	endocrine	progenitors	.............	96	6.2.4	 CDK	inhibition	during	iPSC	differentiation	increases	expression	of	human	NEUROG3	....................................................................................................................................	97	6.3	 Discussion	..............................................................................................................................................	99	Chapter	7:	Single	cell	transcriptomics	of	mouse	and	human	endocrine	progenitor	cells	.......................................................................................................................................................	101	7.1	 Introduction	.......................................................................................................................................	101	7.2	 Results	..................................................................................................................................................	103	7.2.1	 Strategy	for	generating	quality	controlled,	single	cell	transcriptome	data	from	mouse	embryonic	pancreas	..................................................................................	103	7.2.2	 Identification	of	cell	types	in	E15.5	pancreas	...........................................................	106	7.2.3	 Characterization	of	the	endocrine	cell	transcriptome	in	E15.5	pancreas	....	113	7.2.4	 Characterization	of	endocrine	cell	population	at	E18.5	.......................................	116	7.2.5	 Single	cell	transcriptome	of	NEUROG3-lineage	during	hESC	differentiation	.......................................................................................................................................................	120	7.2.6	 Discovery	of	non-endocrine	cells	in	GFP+	S6D1	cells	............................................	126	7.3	 Discussion	...........................................................................................................................................	129	Chapter	8:	Conclusions	..................................................................................................................	133	8.1	 Research	summary	.........................................................................................................................	133	8.2	 Future	directions	.............................................................................................................................	137	8.3	 Final	thoughts	...................................................................................................................................	140	References	.........................................................................................................................................	141	Appendix	A:	R	pipeline	for	determining	SEM	for	TC	and	TS	.............................................	159	Appendix	B:	Analysis	of	single	cell	RNA-sequencing	data	using	Scater	R	pipeline	...	164	Appendix	C:	Analysis	of	single	cell	RNA-sequencing	data	using	Seurat	pipeline	.......	166	  x List	of	tables	Table	1:	Antibody	list	...........................................................................................................................................	35	Table	2:	List	of	TAQMAN	qPCR	primers	......................................................................................................	36	Table	3:	List	of	target	sequences	for	Nanostring	Sprint	.......................................................................	37	Table	4:	Changes	in	cell	cycle	length	during	mouse	pancreatic	development	............................	44	Table	5:	Targeting	efficiency	of	TALEN/CRISPR-mediated	OCT4-eGFP-2A-Puro	CyT49	hESC	lines	..............................................................................................................................................................................	74	Table	6:	Targeting	efficiency	of	CRISPR-mediated	NEUROG3-2A-eGFP	CyT49	hESC	lines	....	86	Table	7:	Metrics	of	scRNA-seq	libraries	from	mouse	E15.5,	mouse	E18.5	and	human	S6D1.....................................................................................................................................................................................	104		  xi List	of	figures	Figure	1:	Specification	of	the	pancreas	from	the	posterior	foregut	....................................................	7	Figure	2:	The	primary	transition	and	formation	of	dorsal	and	ventral	pancreatic	buds	..........	8	Figure	3:	Positive	regulators	of	cell	cycle	progression	..........................................................................	15	Figure	4:	Progression	through	the	G1-phase	of	the	cell	cycle	............................................................	16	Figure	5:	Cumulative	EdU	labelling	measures	increases	in	cell	cycle	length	during	pancreas	development	............................................................................................................................................................	42	Figure	6:	The	length	of	the	G1-phase	of	the	cell	cycle	increases	in	Pdx1+	pancreatic	progenitors	during	early	mouse	embryonic	development	..................................................................	43	Figure	7:	Tip	progenitor	cells	have	shorter	G1	length	than	trunk	progenitors	from	E11.5	to	E13.5	............................................................................................................................................................................	45	Figure	8:	Population	of	EdU-negative	epithelial	cells	remain	after	17	hours	of	EdU	labelling........................................................................................................................................................................................	47	Figure	9:	EdU-negative	pancreatic	epithelial	cells	express	pancreatic	progenitor	markers	49	Figure	10:	Most	Neurog3+	cells	arise	from	EdU+	cells	.........................................................................	50	Figure	11:	KrasLSL-G12D	mutations	lead	to	altered	cell	cycle	length	during	mouse	pancreatic	development	............................................................................................................................................................	58	Figure	12:	There	is	no	change	in	the	growth	fraction	in	KrasG12D	embryos	.................................	60	Figure	13:	Ectopic	expression	of	Cdkn1b	in	Sox9+	progenitors	increases	Neurog3+	cells	at	E12.5	............................................................................................................................................................................	61	Figure	14:	Ectopic	expression	of	Cdkn1b	from	E10.5-E12.5	increases	Chga	and	Gcg	endocrine	cell	formation	....................................................................................................................................	62	Figure	15:	Ectopic	expression	of	Cdkn1b	from	E12.5-E14.5	increases	endocrine	cell	formation	...................................................................................................................................................................	64	Figure	16:	Cdk	inhibition	increases	the	number	of	Neurog3+	cells	in	mouse	embryonic	explants	......................................................................................................................................................................	65	Figure	17:	Targeting	strategy	using	genetically	engineered	nucleases	to	generate	OCT4-eGFP-2A-Puro	hESC	lines	....................................................................................................................................	73	Figure	18:	Knock-in	add	on	of	eGFP	does	not	impact	native	OCT4	expression	in	targeted	hESC	lines	..................................................................................................................................................................	75	  xii Figure	19:	Stem	cell	characteristics	are	retained	in	OCT4-eGFP-2A-Puro	reporter	hESC	lines........................................................................................................................................................................................	76	Figure	20:	Differentiation	of	the	definitive	endoderm	germ	layer	is	unaffected	by	the	addition	of	eGFP	into	the	OCT4	locus	............................................................................................................	78	Figure	21:	OCT4	and	SOX17	expression	in	GFP+	and	GFP-	cells	........................................................	79	Figure	22:	Western	blot	analysis	of	OCT4	and	eGFP	expression	during	differentiation	of	hESCs	to	definitive	endoderm	and	primitive	gut	tube	..........................................................................	80	Figure	23:	eGFP	expression	decreases	upon	differentiation	towards	pancreatic	progenitor	cells	..............................................................................................................................................................................	82	Figure	24:	Addition	of	eGFP	does	not	affect	the	efficiency	of	pancreatic	progenitor	formation	during	in	vitro	differentiation	protocol	..................................................................................	84	Figure	25:	Gene	expression	analysis	of	pancreatic	endoderm	markers.	.......................................	84	Figure	26:	Generation	of	NEUROG3-2A-eGFP	knock-in	reporter	CyT49	hESC	lines	.................	85	Figure	27:	Validation	of	Neurog3-2A-eGFP	CyT49	hESC	reporter	lines	.........................................	86	Figure	28:	The	number	of	GFP+	cells	peaks	during	the	transition	from	stage	5	to	stage	6	..	92	Figure	29:	Gene	expression	analysis	of	progenitor	and	endocrine	markers	in	CyT49	N5-5	hESC	line.	...................................................................................................................................................................	94	Figure	30:	Generation	of	FUCCI-3	CyT49	transgenic	hESC	line	.........................................................	95	Figure	31:	CDK	inhibition	increases	expression	of	pancreas	specific	genes	in	the	G1-phase	of	the	cell	cycle	........................................................................................................................................................	96	Figure	32:	CDK	inhibition	increases	NEUROG3	protein	expression	and	activation	of	downstream	target	genes	...................................................................................................................................	97	Figure	33:	CDK2/4/6	inhibition	increases	NEUROG3	protein	expression	in	hiPSC-derived	endocrine	progenitor	cells	.................................................................................................................................	98	Figure	34:	Strategy	for	isolating	mouse	pancreatic	progenitors	and	endocrine	cells	during	embryogenesis	.....................................................................................................................................................	103	Figure	35:	Pipeline	to	generate	sequenced	libraries	...........................................................................	105	Figure	36:	Analysis	of	single	cells	in	E15.5	pancreas	reveals	13	clusters	..................................	107	Figure	37:	Expression	of	macrophage	markers	in	E15.5	pancreatic	cells	.................................	107	Figure	38:	Single	cell	expression	of	pancreatic	genes	at	E15.5	......................................................	109	  xiii Figure	39:	Single	cell	expression	of	cell	cycle	genes	in	E15.5	mouse	pancreas	.......................	110	Figure	40:	Single	cell	expression	of	genes	highly	expressed	in	mesenchyme	clusters	........	112	Figure	41:	Expression	of	mesenchymal	and	epithelial	markers	in	E15.5	embryos	...............	113	Figure	42:	tSNE	plot	of	clusters	in	endocrine	cells	at	E15.5	............................................................	114	Figure	43:	The	top	ten	genes	expressed	in	E15.5	yellow	and	green	clusters	...........................	115	Figure	44:	tSNE	plots	of	yellow	populations	of	cells	at	E18.5	.........................................................	116	Figure	45:	The	top	ten	genes	expressed	in	E18.5	yellow	clusters	.................................................	117	Figure	46:	tSNE	plots	of	green	populations	of	cells	at	E18.5	...........................................................	119	Figure	47:	Single	cell	expression	of	endocrine	cell	genes	in	E18.5	green	cells	........................	120	Figure	48:	tSNE	plot	of	single	GFP+	cells	at	S6D1	of	N5-5	hESC	differentiation	.....................	122	Figure	49:	Single	cell	expression	of	endocrine-specific	genes	........................................................	123	Figure	50:	Expression	of	key	b-cell	genes	in	single	GFP+	S6D1	cells	...........................................	125	Figure	51:	Expression	of	MAF	genes	in	S6D1	GFP+	cells	...................................................................	125	Figure	52:	Expression	of	pancreas-specific	genes	in	S6D1	GFP+	cells	........................................	126	Figure	53:	Gene	expression	profile	of	liver	cluster	in	S6D1	GFP+	cells	......................................	127	Figure	54:	Top	nine	genes	expressed	in	unknown	cluster	of	S6D1	GFP+	cells	.......................	128	Figure	55:	Model	for	cell	cycle	regulation	of	Neurog3	via	Cdk	phosphorylation	...................	136		 xiv List	of	abbreviations		 APPM	 Amyloid	precursor	protein	modulator	bHLH	 Basic	helix-loop-helix	Cdh1	 E-cadherin	CDK	 Cyclin-dependent	kinase	CDKi	 CDK	inhibition	Chga	 Chromogranin	A	CKI	 Cyclin-dependent	kinase	inhibitors	CRISPR	 Clustered	Regularly	Interspaced	Short	Palindromic	Repeats	DE	 Definitive	endoderm	DSB	 Double	stranded	break	DT	 Double	transgenic	E	 Embryonic	day	EdU	 Ethynyl-2’-deoxyuridine	EP	 Endocrine	progenitor	ESC	 Embryonic	stem	cell	EtOH	 Ethanol	Gcg	 Glucagon	GEMs	 Gel	Bead-In-Emulsions	GF	 Growth	fraction	gRNA	 Guide	RNA	HDR	 Homology-directed	repair	hESC	 Human	embryonic	stem	cells	Indels	 Insertions	and	deletions	Ins	 Insulin	iPSC	 Induced	pluripotent	stem	cells	MODY	 Maturity	onset	diabetes	of	the	young	Neurog3	 Neurogenin	3	Pak3	 p21	protein-activated	kinase	3	PAM	 Protospacer	adjacent	motif	Pdx1	 Pancreas/duodenum	homeobox	1	PFA	 Paraformaldehyde	PP-cells	 Pancreatic	polypeptide	cells	QC	 Quality	control	qPCR	 Quantitative	PCR	RVD	 Repeat	variable	domain	RTK	 Receptor	tyrosine	kinase	scRNA-seq	 Single-cell	RNA-sequencing	Shh	 Sonic	hedgehog	ST	 Single	transgenic	STZ	 streptozotocin	TALEN	 Transcription	Activator	Like	Effector	Nucleases	TBP	 TATA-binding	protein	TC	 Total	cell	cycle	length	 xv TFs	 Transcription	factor	TG1	 G1-phase	length	TG2/M	 G2/M-phase	length	TS	 S-phase	length	Vim	 Vimentin	Wpc	 Weeks	post	conception	ZFN	 Zinc	finger	nuclease	α-cells	 Alpha	cells	β-cells	 Beta	cells	δ-cells 	 Delta	cells	e-cells		 Ghrelin	cells	f	 macrophage		 	 xvi Acknowledgements			 While	my	doctoral	training	was	at	times	challenging,	it	was	also	a	very	fulfilling	and	rewarding	experience,	due	in	large	part	to	the	great	people	who	were	part	of	it.	I	really	appreciate	each	and	every	one	of	you	for	making	the	past	six	years	some	of	the	best	of	my	life.	 First,	thank	you	to	my	supervisor,	Dr.	Francis	Lynn,	for	providing	a	supportive	training	environment	that	allowed	me	to	follow	my	scientific	curiosity;	for	continually	offering	me	opportunities	and	new	experiences;	for	believing	in	my	scientific	ambitions	and	encouraging	me	to	reach	my	full	potential.	I	am	incredibly	fortunate	to	have	been	trained	by	such	a	great	mentor	and	scientist.	I	also	want	to	acknowledge	the	great	number	of	mentors	who	have	supported	me	throughout	my	research	career.	To	my	committee	members,	Dr.	Christopher	Maxwell,	Dr.	Timothy	Kieffer,	and	Dr.	Brad	Hoffman:	thank	you	for	your	guidance,	constructive	criticism	and	support	during	my	studies.	A	special	thanks	to	Dr.	Pamela	Hoodless,	Dr.	Juan	Hou,	and	members	of	the	Hoodless	Lab	for	a	wonderful	first	lab	experience	and	for	sparking	my	love	of	developmental	biology.	I	want	to	thank	Dr.	Jim	Johnson	and	Dr.	Bruce	Verchere	for	their	guidance	and	support	over	the	years.		 Many	thanks	to	members	of	the	Lynn	Lab,	Dr.	Paul	Sabatini,	Dr.	Eric	Xu,	Thilo	Speckmann,	Dr.	Shugo	Sasaki,	Samantha	Yoon,	Alex	Kadhim,	and	Shannon	Sproul,	for	providing	a	fun	environment	filled	with	laughter,	confusing	conversations	and	friendship.	A	special	thank	you	to	Cuilan	Nian,	Mei	Tang,	and	Elizabeth	Lin,	whose	efforts	are	invaluable	to	the	success	of	the	Lynn	Lab.	I	am	grateful	for	the	members	of	the	Verchere,	Hoffman,	Luciani,	Gibson,	Kieffer	and	Johnson	laboratories	who	shared	their	reagents,	equipment,	 xvii and	expertise.	Thank	you	to	the	undergraduate	students	that	I	co-supervised	during	my	PhD	studies:	Akie	Watanabe,	Anna	Branch,	Milo	Yu,	Andrew	Sharon,	Supun	Kotteduwa,	and	Michelle	Lee.	I	hope	you	enjoyed	your	experience	in	the	laboratory	as	much	as	I	appreciated	your	efforts.	I	also	want	to	acknowledge	all	staff	at	BC	Children’s	Hospital	Research	Institute	who	provided	research	or	administrative	support,	including	Angel	Lam	and	Meg	Hughes	(Diabetes	Research	Program),	Lisa	Xu	(Flow	Cytometry),	Jingsong	Wang	(Imaging	Core),	Dawn	McArthur	and	Tamara	English	(Research	and	Technology	Development),	Ashley	Biggerstaff	and	Jennifer	Myers	(Research	Education).		 I	want	to	thank	my	entire	family	for	their	encouragement,	love,	and	support.	Mom	and	Dad,	none	of	this	would	be	possible	without	you.	You	have	always	believed	in	me	and	gave	me	the	confidence	to	follow	my	dreams.	Thank	you	to	my	big	brother	Chris,	whose	love	of	music	and	literature	inspired	me	to	find	my	passion.	I	am	grateful	for	the	strong	women	in	my	life	who	taught	me	that	I	can	do	anything:	Anna	Krentz,	Winnifred	Krentz,	Rina	Lot,	Lucie	Matich,	Heather	Riley,	and	Kim	Fier.	Finally,	to	my	partner,	Nathan	Fier:	you	are	my	favourite.	Thank	you	for	patiently	waiting	“just	ten	minutes	while	I	quickly	feed	my	cells”,	for	always	making	me	laugh,	and	for	being	my	best	friend.		 	 xviii Dedication	      To	my	Mom	and	Dad.			 1 Chapter	1: Introduction	1.1 The	role	of	the	pancreas	in	regulating	glucose	metabolism		 The	mammalian	pancreas	can	be	characterized	as	having	both	an	exocrine	and	endocrine	function.	The	exocrine	pancreas	includes	acinar	and	ductal	cells,	forms	up	to	90%	of	the	pancreas	and	produces	products	that	are	critical	for	digestion,	such	as	proteases,	lipases,	amylases,	and	bicarbonate.	The	endocrine	pancreas	is	located	in	the	islets	of	Langerhans.	Within	the	islets	there	are	glucagon-producing	α-cells,	insulin-producing	β-cells,	somatostatin-producing	δ-cells,	ghrelin-producing	 e-cells	and	pancreatic	polypeptide-producing	PP-cells.		 In	addition	to	secreting	products	critical	for	digestion,	the	pancreas	plays	an	important	role	in	controlling	blood	glucose	levels.	When	blood	glucose	is	low,	glucagon	is	secreted	by	α-cells,	stimulating	the	liver	to	undergo	glycogenolysis	to	convert	stored	glycogen	into	glucose	and	gluconeogenesis	to	generate	glucose	from	amino	acids.	When	blood	glucose	is	high,	the	β-cells	secrete	insulin,	which	acts	on	several	peripheral	tissues	to	restore	blood	glucose	back	to	baseline	levels.	This	occurs	by	both	promoting	the	uptake	of	glucose	into	hepatocytes,	adipose	cells,	and	muscle	cells	and	by	suppressing	the	breakdown	of	glycogen,	the	release	of	amino	acids,	and	the	release	of	free	fatty	acids	by	the	liver,	muscle,	and	adipose	tissue,	respectively.	The	opposing	actions	of	insulin	and	glucagon	are	essential	for	the	proper	control	of	blood	glucose	levels.	1.2 Diabetes	mellitus		 In	2015,	it	was	estimated	that	diabetes	affected	415	million	people	worldwide	and	is	expected	to	continue	to	increase	to	642	million	adults	by	20401.	Diabetes	mellitus	is	a	group	of	metabolic	disorders	that	are	characterized	by	elevated	blood	glucose	levels	due	to	 2 the	defects	in	insulin	secretion	and/or	insulin	action.	The	prolonged	periods	of	hyperglycemia	can	lead	to	devastating	secondary	complications	including:	retinopathy,	neuropathy,	nephropathy	and	cardiovascular	disease2.		 The	first	recorded	description	of	diabetes	was	by	the	Greek	physician	Aretaeus	the	Cappadocian3.	The	symptoms	of	excessive	thirst,	frequent	urination,	and	weight	loss	led	Aretaeus	to	call	it	diabetes,	which	is	Greek	for	‘siphon’	or	‘to	run	through’.	While	diabetes	as	a	disease	was	recognized	in	antiquity,	the	underlying	mechanism	of	diabetes	was	not	discovered	until	the	19th	and	20th	centuries.	Hypothesizing	that	the	pancreas	plays	a	role	in	the	digestion	of	fats,	Oskar	Minkowski	and	Joseph	von	Mering	performed	pancreatectomy	on	dogs	and	found	that	this	leads	to	diabetes4.	They	also	found	that	you	can	prevent	hyperglycemia	following	pancreatectomy	by	subcutaneously	implanting	a	small	piece	of	pancreas,	clearly	identifying	an	important	role	of	the	pancreas	in	diabetes	pathogenesis4.	While	Paul	Langerhans	described	the	cells	within	the	pancreas	that	later	became	known	as	the	‘islets	of	Langerhans’	in	18695,	their	critical	role	in	glucose	metabolism	was	not	appreciated	for	another	half	century.	In	the	early	20th	century,	many	researchers	were	studying	the	link	between	pancreatic	islets	and	diabetes.	This	work	led	to	the	conclusion	that	the	islets	produced	an	internal	secretion	that	was	critical	for	controlling	blood	glucose	levels,	but	isolating	this	extract	was	proving	to	be	difficult	due	to	the	digestive	enzymes	produced	by	the	pancreas6.		 The	idea	that	led	to	the	successful	isolation	of	insulin	is	credited	to	Fredrick	Banting,	a	Canadian	surgeon.	In	1920,	Moses	Barron	published	an	autopsy	report	on	a	patient	who	died	of	diabetic	ketoacidosis	and	coma.	In	this	report,	he	outlined	a	patient	with	pancreatic	lithiasis,	an	obstruction	of	the	pancreatic	duct	by	calcium	carbonate	stones,	that	had	 3 pancreatic	atrophy	of	the	acinar	tissue,	leaving	numerous	islets	intact7.	Inspired	by	the	work	of	Barron,	Banting	hypothesized	that	performing	pancreatic	duct	ligation	on	dogs	would	also	cause	the	acinar	cells	to	atrophy,	allowing	for	the	isolation	of	pancreatic	secretion	from	the	islets.	In	collaboration	with	Charles	Best	and	John	Macleod,	Banting	successfully	isolated	insulin	from	canines	and	used	it	to	treat	the	diabetes	of	depancreatized	dogs,	providing	evidence	that	insulin	therapy	can	be	used	to	treat	diabetes8.	With	the	help	of	James	Collip,	they	purified	insulin	at	a	quality	that	would	allow	for	testing	in	humans	and	were	able	to	treat	their	first	patient,	a	14-year	old	boy	named	Leonard	Thompson,	with	insulin	in	19219.	The	isolation	of	insulin	by	Banting,	Best,	Collip	and	Macloed	was	awarded	a	Novel	Prize	and	changed	the	course	of	diabetes	treatment	by	transforming	a	fatal	disease	to	one	that	can	be	managed	with	exogenous	insulin.		 While	insulin	therapy	revolutionized	the	face	of	diabetes,	mimicking	the	function	of	the	endogenous	pancreas	is	difficult	and	can	result	in	large	fluctuations	in	blood	glucose	levels.	This	led	to	work	on	cell	replacement	therapies	for	the	treatment	of	diabetes,	either	through	whole	pancreas	or	islet	transplantation.	In	1966,	the	first	whole	pancreas	transplant	was	performed	and	was	able	to	treat	diabetes	without	the	need	for	exogenous	insulin10.	With	better	surgical	techniques	and	immunosuppressive	drugs,	the	American	Diabetes	Association	recommended	that	pancreas	transplant	be	considered	as	an	effective	alternative	treatment	for	type	1	diabetes11.	However,	as	this	is	a	major	surgery	that	carries	significant	risk12,	whole	pancreas	transplant	is	usually	reserved	as	a	treatment	for	a	subset	of	patients	with	difficult	to	control	diabetes.		 As	diabetes	is	caused	by	a	loss	of	the	insulin-producing	b-cells,	another	cell	replacement	therapy	is	islet	cell	transplantation.	Following	the	successful	isolation	of	rat	 4 islets13,	proof	of	concept	experiments	showing	the	treatment	of	streptozotocin	(STZ)-induced	diabetes	in	rats	using	islet	transplantation	were	performed	in	197214.	These	pre-clinical	studies	led	to	a	great	interest	in	the	clinical	application	of	islet	transplantation	for	diabetes	treatment.	In	patients	with	chronic	pancreatitis,	autotransplantation	of	the	patient’s	own	islets	following	pancreatectomy	prevented	the	resulting	diabetes15.	Unfortunately,	there	was	limited	success	in	maintaining	normoglycemia	in	patients	following	alloislet	transplantation16-19,	due	to	the	inhibitory	effect	of	the	immunosuppressive	drugs	on	b-cell	function20-23.	Therefore,	the	success	of	islet	transplantation,	as	measured	by	transplant	function	and	ability	to	maintain	normoglycemia,	was	limited.	In	2000,	Shapiro	and	colleagues	outlined	the	‘Edmonton	Protocol’,	a	specific	immunosuppressive	regimen	that	resulted	in	seven	islet	transplant	recipients	remaining	insulin-independent	for	one-year	post-transplant24.	Five-year	follow	up	of	these	patients	suggested	that,	while	the	median	period	of	insulin	independence	was	only	15	months,	the	risk	of	hypoglycemic	events	was	dramatically	reduced25.	Together,	these	results	suggest	that	islet	transplantation	is	an	effective	treatment	for	diabetes,	especially	those	with	“brittle”	diabetes	who	have	difficulty	detecting	hypoglycemia.		 Unfortunately,	the	shortage	of	human	donor	tissue	prevents	pancreas	and	islet	transplantation	from	becoming	a	widespread	treatment	for	diabetes.	Thus,	there	is	great	interest	in	generating	an	unlimited	source	of	large	numbers	of	transplantable	β-cells.	One	potential	approach	is	the	differentiation	of	human	embryonic	stem	cells	(hESC).	To	produce	mature	cells	for	transplantation	in	culture,	we	need	a	detailed	understanding	of	the	developmental	network	that	regulates	the	formation	of	pancreatic	β-cells.	 5 1.3 Overview	of	mouse	pancreas	development	Mouse	endocrine	development	can	be	divided	into	four	stages:	1)	gastrulation	and	the	formation	of	the	definitive	endoderm	germ	layer;	2)	specification	of	endoderm	to	the	pancreatic	lineage;	3)	primary	transition	and	the	growth	of	pancreatic	buds;	and	4)	secondary	transition	and	the	differentiation	of	endocrine	cells.	1.3.1 Gastrulation	and	the	formation	of	the	definitive	endoderm	germ	layer	For	the	first	three	to	four	days	of	mammalian	development,	the	fertilized	egg	undergoes	many	cycles	of	rapid	cell	division	known	as	cleavage	to	generate	the	blastocyst.	The	blastocyst	is	made	up	of	two	groups	of	cells:	the	trophectoderm	and	the	inner	cell	mass.	The	trophectoderm	gives	rise	to	the	extraembryonic	tissues	that	are	important	for	providing	the	embryo	with	nutrition	and	the	inner	cell	mass	of	pluripotent	cells	gives	rise	to	all	cell	types	of	the	embryo	proper.	The	first	major	step	towards	organogenesis	is	a	process	called	gastrulation	where	the	ectoderm,	mesoderm,	and	definitive	endoderm	germ	layers	are	formed	from	the	cells	of	the	inner	cell	mass.	The	three	germ	layers	differ	in	the	types	of	cells	they	produce.	Of	interest	for	this	thesis	is	the	definitive	endoderm	as	it	forms	the	digestive	tube	and	associated	organs,	including	the	pancreas.	In	the	mouse,	endoderm	formation	begins	between	embryonic	day	(E)6.5	and	7.5	in	a	process	that	requires	instructive	signals	from	the	adjacent	germ	layers26.	Studies	in	mice	identified	Tgfb27,28	and	Wnt29,30	signalling	pathways	to	be	important	for	the	generation	of	definitive	endoderm.	More	specifically,	definitive	endoderm	formation	requires	high	levels	of	Nodal,	a	Tgfb	family	member,	while	low	levels	allow	mesoderm	formation31,32.	Successful	completion	of	gastrulation	results	in	the	formation	of	the	body	plan	and	the	generation	of	the	three	germ	layers.	 6 1.3.2 Specification	of	endoderm	to	the	pancreatic	lineage	Following	gastrulation	is	the	process	of	organogenesis	where	the	pancreas	is	derived	from	the	definitive	endoderm	germ	layer.	It	forms	from	the	posterior	region	of	the	foregut	and	begins	to	develop	as	dorsal	and	ventral	buds.	The	specification	of	the	pancreas	involves	many	instructive	cues	from	closely	associated	tissues.	This	includes	retinoic	acid	from	the	paraxial	mesoderm	that	is	required	for	the	formation	of	the	dorsal	pancreas33.	The	mesoderm-derived	notochord	represses	sonic	hedgehog	(Shh)	signalling	through	the	production	of	Activin	b2	and	Fgf2	(Figure	1A)34,35.	The	dorsal	aorta	also	plays	an	important	role	in	dorsal	bud	formation	through	the	secretion	of	an	unknown	factor	(Figure	1B)36,37.	All	these	signals	together	result	in	the	activation	of	the	transcription	factor	Pancreatic	and	Duodenal	Homeobox	1	(Pdx1).	Expression	of	Pdx1	is	found	in	two	distinct	regions:	the	dorsal	pancreatic	bud	starting	at	E8.5	and	the	ventral	pancreatic	bud	at	E9.5	(Figure	1B)38.	Pdx1-/-	mice	develop	early	pancreatic	buds	but	fail	to	develop	a	fully	functioning	pancreas,	indicating	that	Pdx1	is	necessary	for	pancreas	development39.  7   Figure	1:	Specification	of	the	pancreas	from	the	posterior	foregut	(A)	At	E8.0,	Fgf2	and	Activin	b2	are	secreted	by	the	notochord	and	repress	Shh	signaling	in	the	region	of	the	dorsal	endoderm	that	will	form	the	pancreas.	(B)	At	E8.5,	signals	from	the	dorsal	aorta	and	the	vitelline	vein	are	important	for	the	induction	of	Pdx1	(green).	 1.3.3 Primary	transition	and	growth	of	the	pancreatic	buds	The	acquisition	of	commitment	to	the	pancreatic	lineage	as	a	“protodifferentiated”	cell	occurs	between	E8.5-10.5	and	is	known	as	the	“primary	transition”.	At	this	point,	the	pancreatic	epithelium	begins	to	bud	off	the	primitive	gut	tube	with	the	dorsal	bud	becoming	visible	before	the	ventral	(Figure	2A).	At	E9.5,	the	buds	begin	to	form	(Figure	2B)	 8 and	by	E11.0	microlumen	are	visible	within	the	pancreatic	buds	(Figure	2C)40,41.	By	E12.5,	the	epithelium	is	segregated	into	Pdx1+	Cpa1+	multipotent	tip	progenitor	cells	(Figure	2D;	red)	and	the	Pdx1+	Cpa1-	bipotent	trunk	progenitor	cells	(Figure	2D;	blue)42.	These	progenitor	populations	are	differentially	specified	in	that	the	tip	cells	are	competent	to	give	rise	to	acinar,	ductal	and	endocrine	cells	while	the	trunk	cells	are	fated	to	the	endocrine	and	ductal	lineages	only42.	While	there	are	some	Glucagon+	cells	that	are	formed	during	the	primary	transition,	the	majority	of	cells	within	the	epithelium	remain	undifferentiated	and	proliferate	to	expand	the	progenitor	cell	pool.	This	is	of	great	importance	as	the	number	of	progenitor	cells	at	this	stage	of	development	dictates	the	final	size	of	the	organ43.	  Figure	2:	The	primary	transition	and	formation	of	dorsal	and	ventral	pancreatic	buds	(A)	The	dorsal	and	ventral	pancreas	are	morphologically	distinct	from	the	gut	tube	and	express	Pdx1	(green).	(B)	At	E9.5,	the	pancreatic	buds	begin	to	form	and	express	Pdx1	(green).	(C)	At	E11.0,	there	are	microlumen	that	form	within	the	Pdx1+	bud	(green).	(D)	The	pancreatic	epithelium	becomes	branched	and	is	segregated	into	the	Pdx1+Cpa1+	tip	progenitor	cells	(red)	and	the	Pdx1+Cpa1-	trunk	progenitor	cells	(blue).	The	tip	cells	can	form	acinar,	ductal	and	endocrine	cells	while	the	trunk	cells	can	only	form	ductal	and	endocrine	cells.  9 1.3.4 Secondary	transition	and	endocrine	cell	differentiation		 The	majority	of	endocrine	differentiation	occurs	in	a	process	known	as	the	“secondary	transition”.	This	process	involves	the	basic	helix-loop-helix	(bHLH)	transcription	factor	Neurogenin	3	(Neurog3).	First	identified	for	its	expression	in	neural	crest	development,	Neurog3	is	the	only	Neurog	family	member	expressed	in	the	pancreas44.	Neurog3	mRNA	is	transiently	expressed	as	early	as	E9.5,	with	the	peak	of	expression	at	E15.545-48.	Lineage	tracing	revealed	that	Neurog3+	cells	give	rise	to	all	endocrine	cell	types	during	mammalian	pancreas	development49,50,	suggesting	that	it	plays	an	important	role	in	endocrine	cell	differentiation.		 Neurog3-/-	mice	have	a	complete	loss	of	all	pancreatic	endocrine	cells	and	consequently	die	of	diabetes	1-3	days	after	birth46.	Neurog3+	endocrine	progenitors	are	unipotent,	meaning	that	an	individual	progenitor	will	only	give	rise	to	one	endocrine	cell	type51.	This	occurs	through	a	“competence	model”	where	the	first	Neurog3+	cells	become	Glucagon+	 a-cells,	then	Insulin+	b-cells	and	Pancreatic	Polypeptide+	PP-cells	before	finally	generating	Somatostatin+	d-cells52.	At	a	single	cell	level,	Neurog3	protein	is	found	transiently,	lasting	less	than	24	hours53.	Consistent	with	the	rapid	turnover	of	Neurog3	protein,	pulse-chase	studies	have	found	that	the	transition	from	a	Sox9+	progenitor	to	a	committed	endocrine	cell	is	~12.3	hours54.		 Strict	control	over	Neurog3	protein	is	critical	for	proper	pancreas	development.	For	instance,	ectopic	expression	of	Neurog3	in	chick	endoderm	or	mouse	pancreatic	endoderm	induces	endocrine	cell	differentiation47,55.	Using	lineage	tracing,	most	pancreatic	progenitors	that	express	high	levels	of	Neurog3	will	become	endocrine	cells49,52.	If	Neurog3	expression	is	absent	or	reduced,	these	progenitors	will	adopt	a	ductal	cell53,56,57	or	acinar	 10 cell	fate50.	Consistent	with	these	studies,	it	was	determined	that	before	E12.5	Neurog3-induced	cells	can	give	rise	to	acinar	and	ductal	cells	but	after	E12.5	the	capacity	to	form	acinar	cells	is	lost53.	Taken	together,	these	results	suggest	that	the	endocrine	cell	fate	requires	high	expression	of	Neurog3.	1.4 Differences	in	mouse	and	human	pancreas	development	Pancreas	development,	which	takes	approximately	ten	days	in	mice,	occurs	over	the	course	of	several	months	in	humans58.	As	with	mouse	pancreas	development,	human	pancreas	development	also	progresses	through	the	four	main	stages	of	development:	gastrulation,	specification,	progenitor	expansion	and	endocrine	differentiation.	Additionally,	many	of	the	transcription	factors	that	are	known	to	be	important	in	mouse	development	also	play	a	role	in	human	development.	For	example,	homozygous	and	heterozygous	mutations	in	PDX1	are	associated	with	human	pancreas	agenesis	and	the	monogenic	hereditary	forms	of	diabetes	called	Maturity	Onset	Diabetes	of	the	Young	(MODY)59,60	and	permanent	neonatal	diabetes61.	However,	there	are	some	key	differences	between	mouse	and	human	pancreas	development.	In	mouse,	Pdx1	is	detected	in	the	dorsal	endoderm	during	pancreas	specification	while	human	PDX1	is	only	detected	after	the	notochord	and	aorta	have	separated	from	the	dorsal	foregut62.	NKX2-2	is	not	detected	in	early	human	pancreas	progenitors	but	is	widely	detected	across	the	pancreatic	epithelium	during	mouse	development.	Finally,	endocrine	differentiation,	which	occurs	in	two	waves	of	Neurog3	expression	during	mouse	development,	has	only	one	phase	of	NEUROG3	expression	during	human	pancreas	development62.	Taken	together,	these	data	suggest	that	while	mouse	 11 studies	are	very	informative	it	is	important	to	also	use	human	models	of	pancreas	development,	for	example	the	differentiation	of	human	embryonic	stem	cells.	1.5 Human	embryonic	stem	cells	Embryonic	stem	cells	(ESC)	are	derived	from	the	inner	cell	mass	of	blastocyst	embryos.	Given	the	right	culture	conditions,	hESCs	are	capable	of	unlimited	expansion	in	vitro	and	are	able	to	form	all	three	germ	layers63.	Owing	to	these	two	properties,	there	is	a	great	potential	for	hESCs	to	generate	an	unlimited	source	of	specialized	cell	types	for	the	treatment	of	diseases,	including	diabetes.	This	involves	directing	the	differentiation	of	hESCs	through	the	normal	stages	of	development	towards	insulin-producing	b-cells.	1.6 Using	human	embryonic	stem	cells	as	a	model	of	human	pancreas	development		 The	first	step	in	generating	endocrine	cells	from	hESCs	is	the	formation	of	the	definitive	endoderm	germ	layer.	Based	on	studies	on	the	formation	of	mouse	definitive	endoderm,	high	levels	of	the	TGFb	family	member	ACTIVIN	A	and	short	term	exposure	to	WNT3A	was	used	to	derive	definitive	endoderm	from	hESCs64-67.	Following	definitive	endoderm	formation,	the	differentiation	then	continues	through	the	primitive	gut	tube	and	posterior	foregut	to	the	generation	of	PDX1+	cells.	There	are	several	protocols	that	efficiently	produce	PDX1+	cells64,66,67.	Consistent	within	these	protocols	is	the	activation	and/or	inhibition	of	several	pathways	that	are	known	to	be	important	for	the	pancreas	formation.	For	instance,	the	addition	of	hedgehog-signaling	inhibitor	KAAD-cyclopamine	mimics	the	important	role	of	the	notochord	during	mammalian	development	(Figure	1)34,35.	Also,	retinoic	acid	is	added	during	this	stage	of	the	differentiation	protocol	and	is	important	for	the	generation	of	Pdx1+	cells	during	murine	development33.	Continued	 12 differentiation	of	the	pancreatic	progenitor	cells	in	vivo	following	transplantation	into	mice	resulted	in	the	formation	of	many	endocrine	cell	types	that	were	able	to	prevent	the	rise	in	blood	glucose	levels	following	STZ-treatment68,	suggesting	that	they	represent	a	bona	fide	pancreatic	progenitor	population.	Building	upon	this	work,	two	papers	were	published	in	2014	that	documented	the	generation	of	hESC-derived	b-like	cells69,70.	These	cells	share	some	functional	similarities	to	human	b-cells	but	are	not	fully	responsive	to	glucose	in	vitro.	Nevertheless,	upon	transplantation	under	the	kidney	capsule	in	mice,	they	rapidly	generate	endocrine	cells	that	are	able	to	reverse	diabetes69.	There	are	now	several	clinical	trials	underway	to	assess	the	safety	of	using	hESC-derived	cells	for	the	treatment	of	diabetes.	This	milestone	results	from	decades	of	research	by	developmental	biologists	and	stem	cell	researchers	to	find	the	signalling	pathways	that	are	important	for	the	formation	of	endocrine	cells.	1.7 Generating	hESC	reporter	lines	using	genome	editing	technology		 hESC	differentiation	is	an	attractive	model	to	investigate	developmental	biology	questions	in	a	human	system.	However,	these	protocols	often	result	in	heterogeneous	populations,	making	it	important	to	be	able	to	easily	isolate	the	cells	of	interest.	This	can	be	accomplished	by	creating	cell-type	specific	reporter	hESC	lines.	Early	strategies	to	generate	reporter	hESC	lines	used	constitutive71-76	or	truncated	promoter-driven	transgenes71,77,78.	However,	variation	in	copy	number	and	the	location	of	the	integration	site	may	affect	expression	of	the	reporter	genes	or	disrupt	endogenous	gene	expression.	In	addition,	there	is	the	potential	of	transgene	silencing	during	differentiation	to	specialized	cell	types79.	Another	strategy	is	to	replace	one	allele	with	the	reporter	gene.	Unfortunately,	this	can	result	in	haploinsufficiency	that	may	impair	differentiation.	To	mitigate	the	potential	for	 13 altering	targeted	gene	expression,	the	reporter	gene	can	be	knocked-in	downstream	but	in-frame	with	the	protein	of	interest.	This	will	result	in	marker	expression	that	is	driven	by	the	endogenous	promoter,	making	it	more	specific	and	less	likely	to	suffer	from	gene	silencing.		 Editing	the	genome	of	hESCs	was	previously	difficult	due	to	the	requirement	for	very	large	homology	arms	and	the	low	rate	of	homologous	recombination	in	hESCs80.	However,	genome	editing	is	fast	becoming	a	reality	in	hESCs	with	the	recent	advent	of	highly	efficiency	genome	editing	technologies,	such	as	Transcription	Activator	Like	Effector	Nucleases	(TALENs)	and	Clustered	Regularly	Interspaced	Short	Palindromic	Repeats	(CRISPR)-CRISPR-Associated	protein	(Cas)81-83.	These	technologies	use	sequence-specific	nucleases	to	create	a	double	stranded	break	in	the	DNA,	dramatically	increasing	the	frequency	of	homologous	recombination	through	homology	directed	repair.	1.8 TALEN	technology		 TALENs	are	genetically	engineered	proteins	with	a	FokI	nuclease	domain	fused	to	a	DNA-binding	domain	of	TALE	repeats.	These	repeats	are	based	on	naturally	occurring	Xanthomonas	proteobacteria	proteins	that	facilitate	the	pathogenic	colonization	by	binding	to	DNA	and	altering	transcription	in	the	host	plant	cells84.	TALENs	use	a	‘protein-DNA	code’	to	drive	specificity	for	where	the	FokI	nuclease	will	generate	a	double	stranded	break.	Driving	this	specificity	are	TALE	repeats	comprised	of	a	34-amino	acid	sequence.	Located	at	amino	acid	position	12	and	13	is	the	Repeat	Variable	Domain	(RVD).	The	amino	acids	of	this	RVD	dictates	the	deoxynucleotide	it	binds	to:	NG	is	highly	specific	for	deoxythymidine,	HD	for	deoxycytidine,	NI	for	deoxyadenosine,	and	NH	for	deoxyguanosine85.	 14 1.9 CRISPR-Cas9	technology		 First	described	in	Escherichia	coli	in	1987,	the	structure	of	a	repetitive	DNA	sequence86	containing	24-47	bp	long	repeats	with	unique	intervening	24-72	bp	long	spacer	sequences	was	discovered87.	Found	exclusively	in	prokaryotes88,	the	significance	of	these	repeats	was	a	mystery	but	the	discovery	of	genes	for	DNA	helicase	and	polymerase	suggested	they	are	involved	in	DNA	repair89.	When	three	independent	research	groups	discovered	that	the	unique	spacer	sequences	came	from	invading	bacteriophage	genomes	in	200590-92,	it	was	proposed	that	they	are	part	of	a	bacterial	adaptive	immune	system93-95.	When	foreign	DNA	is	detected	in	prokaryotic	cells,	short	fragments	of	the	DNA	are	integrated	into	the	host	chromosome	at	the	5’	end	of	the	CRISPR	locus93,96,97,	acting	as	a	genetic	record	of	previous	encounters	with	foreign	DNA93,97,98.	These	sequences	defend	the	host	against	invading	pathogens	via	an	RNA	interference-like	mechanism96.		 Recognizing	the	significance	of	a	RNA-based	DNA	cleavage	system,	the	CRISPR-Cas	system	of	Streptococcus	pyogenes	was	quickly	adapted	for	genome	editing.	This	uses	a	guide	RNA	(gRNA)	that	recognizes	its	target	sequence	by	Watson-Crick	base	pairing96,99.	The	gRNA	binds	to	the	genome	and	brings	the	Cas9	endonuclease,	which	requires	the	protospacer	adjacent	motif	(PAM)	sequence	5’-NGG-3’	to	generate	a	double	stranded	break	(DSB).	Repair	of	the	DSB	can	occur	in	two	ways:	non-homologous	end-joining	(NHEJ)	that	can	result	in	missense	or	nonsense	mutations	(indels)	when	the	DSB	is	in	an	open	reading	frame	or	homology-directed	repair	(HDR).	HDR	can	be	used	to	generate	cell	type	specific	reporter	hESC	lines	when	a	DNA	template	containing	a	fluorescent	protein	and	homology	arms	is	provided.	The	resulting	reporter	hESC	lines	can	be	used	to	answer	biological	questions	in	a	human	model,	including	the	role	of	the	cell	cycle	on	pancreas	development.	 15 1.10 Introduction	to	the	cell	cycle		 The	cell	cycle	can	be	divided	into	four	phases:	two	gap	phases	(G1	and	G2),	S-phase	where	DNA	replication	occurs	and	M-phase	where	the	nucleus	and	cytoplasm	divides.	In	mammalian	cells,	progression	through	the	cell	cycle	is	controlled	by	the	positive	and	negative	regulators,	cyclins/cyclin-dependent	kinases	(Cdks)	and	cyclin-dependent	kinase	inhibitors,	respectively	(Figure	3).		Figure	3:	Positive	regulators	of	cell	cycle	progression	Transition	through	G1	is	regulated	by	CyclinD/Cdk4/6	in	early	G1	and	CyclinE/Cdk2	in	late	G1.	In	S-phase,	Cdk2	complexes	with	CyclinA	to	phosphorylate	proteins	involved	in	DNA	replication.	The	G2-M	transition	is	governed	by	CyclinA/Cdk1	while	CyclinB/Cdk1	is	required	for	mitosis.	 The	progression	through	the	G1-phase	of	the	cell	cycle	is	of	interest	as	it	is	during	this	stage	that,	in	response	to	extracellular	cues,	a	cell	will	either	commit	to	another	round	of	cell	division	or	exit	the	cell	cycle	and	differentiate.	In	early	G1,	CyclinD/Cdk4/6	phosphorylates	and	inactivates	the	tumor	suppressor	proteins	(pRb,	p107,	p130)	and	this	 16 leads	to	the	release	of	E2F	factors	(Figure	4).	During	late	G1,	CyclinE/Cdk2	phosphorylates	and	activates	E2F,	which	is	then	able	to	drive	transcription	of	genes	involved	in	DNA	synthesis	and	cell	cycle	progression.	The	activity	of	Cdks	are	inhibited	by	cyclin-dependent	kinases	inhibitors	(CKIs).	Most	CKIs	inhibit	either	the	early	or	late	G1	Cyclin/Cdk	complex,	except	for	Cdkn1b,	which	can	inhibit	both.	  Figure	4:	Progression	through	the	G1-phase	of	the	cell	cycle	In	early	G1	CyclinD/Cdk4/6	complexes	phosphorylate	pocket	proteins	(pRb,	p107	and	p30;	orange),	resulting	in	the	release	of	E2F.	During	late	G1,	unbound	E2F	is	phosphorylated	and	activated	by	CyclinE/Cdk2	complexes,	allowing	for	E2F-driven	activation	of	genes	involved	in	DNA	replication,	apoptosis	and	cell	cycle	progression.	Several	cyclin	dependent	kinases	inhibitors	phosphorylate	and	inactivate	Cdk2/4/6,	preventing	the	transition	from	G1-S.  There	are	at	least	20	different	Cdks	in	the	mammalian	genome,	most	of	which	have	redundant	roles100,101.	The	only	essential	cell	cycle	kinase	is	mitotic	Cdk1	as	Cdk1-/-	 17 embryos	die	before	E3.5102.	Other	studies	on	mice	null	for	a	single	Cdk	are	able	to	survive	to	birth,	but	often	have	meiotic	phenotypes	relating	to	sterility103.	Interestingly,	Cdk4-/-	mice	develop	diabetes	caused	by	a	reduced	b-mass	due	to	defects	in	postnatal	b-cell	proliferation104.	1.11 The	role	of	cell	cycle	proteins	during	pancreas	development	The	number	of	progenitor	cells	is	regulated	by	the	balance	between	proliferation	and	differentiation,	which	dictates	the	final	size	of	the	adult	organ	and,	therefore,	its	physiological	function43.	To	date	there	are	several	cell	cycle	genes	that	have	been	implicated	in	β-cell	proliferation,	but	have	limited	roles	during	development,	possibly	due	to	the	redundancy	of	cell	cycle	proteins.	In	addition	to	its	role	in	b-cell	proliferation,	Cdk4	is	important	for	proliferation	of	Pdx1+	pancreatic	progenitor	cells.	Cdk4-/-	embryos	have	reductions	in	EdU+	Pdx1+	cells105	that	may	be	driven	by	defects	in	pancreatic	mesenchyme	development106-108.	Conversely,	reductions	in	the	CKI	Cdkn1c/p57	results	in	increased	proliferation	of	Pdx1+	progenitor	cells109.	Together,	these	data	suggest	that	cell	cycle	proteins	play	an	important	role	in	regulating	the	proliferation	of	pancreatic	progenitor	cells.		 In	addition	to	its	role	in	proliferation,	the	cell	cycle	is	also	required	during	differentiation	to	drive	cell	cycle	exit.	In	the	context	of	endocrine	cell	differentiation,	this	is	thought	to	be	driven	by	Neurog3-directed	upregulation	of	Cdkn1a49,51,110.	Supporting	the	role	of	Neurog3	in	cell	cycle	exit,	there	are	several	convincing	studies	that	demonstrate	committed	endocrine	progenitors,	labelled	by	Neurog3-Cre	activity,	are	slowly	or	non-dividing	cells49,51.	More	recently,	P21	protein-activated	kinase	3	(Pak3)	was	proposed	to	act	further	downstream	of	Neurog3	to	promote	cell	cycle	arrest	through	repression	of	 18 CyclinD1111.	While	Neurog3	is	known	to	regulate	cell	cycle	proteins	to	drive	cell	cycle	exit,	whether	the	cell	cycle	regulates	activity	of	Neurog3	protein,	and	therefore	endocrine	cell	differentiation,	is	unknown.	1.12 G1	lengthening	and	differentiation		 There	are	several	correlations	between	G1	lengthening	and	differentiation	that	support	the	role	of	the	cell	cycle	in	regulating	progenitor	cell	differentiation.	For	example,	hESCs	have	a	short	cell	cycle	length	due	to	a	truncated	G1-phase	that	is	thought	to	be	due	to	the	promotion	of	G1	progression	by	elevated	CYCLIND2/CDK4	mRNA	and	decreased	expression	of	CKIs112,113.	Interestingly,	G1	lengthening	occurs	during	hESC	differentiation112	and	hESCs	isolated	in	the	G1-phase	have	an	increased	propensity	for	differentiation	compared	with	those	cells	isolated	in	S-	and	G2-phases114.	Consistent	with	this	correlation	between	G1	length	and	differentiation,	CDK2	inhibition	of	hESC	arrests	cells	in	G1,	leading	to	differentiation	to	extra-embryonic	lineages115.	Taken	together,	these	data	suggest	that	the	cell	cycle,	in	particular	G1	lengthening,	is	involved	in	the	early	cell	fate	decisions	of	hESCs.		 In	addition	to	its	role	in	early	development,	the	cell	cycle	is	thought	to	be	involved	in	neurogenesis,	where	G1	lengthening	is	correlated	with	the	switch	from	neural	progenitor	cell	expansion	to	differentiation116.	For	instance,	in	the	pseudostratified	ventricular	epithelium	where	the	majority	of	neocortical	neurons	originate,	the	length	of	G1	increases	from	3.2	hours	to	12.4	hours	between	E11	to	E16	in	mice117.	It	has	been	demonstrated	the	lengthening	G1	by	inhibition	of	Cdk2/CyclinE1	is	sufficient	to	trigger	premature	neural	differentiation118.	Similarly,	shortening	G1	by	overexpression	of	Cdk4/CyclinD1	prevents	neural	differentiation	while	lengthening	G1	by	knockdown	of	Cdk4/CyclinD1	drives	neural	 19 differentiation119.	The	CKI	Cdkn1b/p27	delays	the	transition	from	G1-S	by	inhibiting	Cdk2/CyclinE	and	is	proposed	to	act	as	a	cell	intrinsic	timer	for	oligodendrocyte	differentiation	by	a	mechanism	that	senses	its	accumulation	over	consecutive	cell	cycles120.	Overexpression	of	Cdkn1b	during	corticogenesis	induces	neurogenesis121,	consistent	with	G1	lengthening	being	required	for	differentiation.	1.13 Cell	cycle	regulation	of	Neurog	family	members		 The	correlation	between	G1	lengthening	and	differentiation	suggests	that	the	cell	cycle	itself	may	directly	regulate	differentiation	by	altering	the	stability	of	obligatory,	lineage-establishing	transcription	factors.	For	example,	the	neurogenic	transcription	factor,	Neurog2,	is	hyperphosphorylated	by	Cdk1/2,	leading	to	ubiquitin-mediated	proteolysis	that	prevents	differentiation	and	maintains	the	progenitor	pool122.	Therefore,	the	CKI	Cdkn1b/p27Xic1	promotes	neurogenesis	by	reducing	the	ubiquitin-mediated	proteasomal	degradation123-125	of	Xenopus126	and	mouse	neurogenic	transcription	factors121.	As	G1	is	lengthened	there	is	a	reduction	in	Cdk	expression	resulting	in	the	reduced	phosphorylation	of	Neurog2122,	leading	to	increases	in	the	downstream	targets	of	Neurog2	and	driving	differentiation127.	In	this	way,	cell	cycle	proteins	directly	regulate	progenitor	cell	differentiation	by	controlling	expression	of	critical	transcription	factors.	Whether	a	similar	paradigm	regulates	Neurog3	and	endocrine	cell	differentiation	is	unknown.	1.14 Thesis	investigation		 Regenerative	medicine	approaches	for	the	treatment	of	diabetes	are	becoming	increasingly	close	to	the	goal	of	generating	functional	b-cells	in	vitro69,70.	This	milestone	was	made	possible	by	basic	research	into	how	cell	fate	decisions	are	made	during	pancreas	development	and	the	pathways	that	govern	this	process.	While	progress	has	been	made	in	 20 generating	endocrine	cells	from	hESCs,	improvements	can	be	made	by	understanding:	1)	what	induces	a	subset	of	the	pancreatic	endodermal	cells	to	become	Neurog3+	endocrine	progenitor	cells	and	2)	how	closely	the	hESC-derived	endocrine	progenitor	cells	resemble	the	cells	that	form	in	vivo	during	development.	The	objective	of	this	thesis	was	to	understand	whether	the	cell	cycle	regulates	the	formation	of	Neurog3+	endocrine	progenitors	and	to	characterize	the	single	cell	transcriptomes	of	both	mouse	and	hESC-derived	Neurog3+	endocrine	progenitor	cells.	Despite	elegant	studies	in	the	nervous	system	and	hematopoiesis,	whether	the	cell	cycle	regulates	pancreas	endocrine	differentiation	has	not	been	explored.	First,	the	lengths	of	each	phase	of	the	cell	cycle	were	measured	during	early	mouse	pancreas	development	(Chapter	3).	Second,	the	consequence	of	changing	cell	cycle	length	was	determined	using	two	mouse	models	(Chapter	4).	Third,	to	understand	the	role	of	the	cell	cycle	during	hESC	differentiation,	a	strategy	to	generate	cell	type	specific	reporter	hESC	lines	was	created	using	genome	editing	technology	(Chapter	5).	Fourth,	the	effect	of	CDK	inhibition	on	endocrine	cell	formation	during	hESC	differentiations	was	investigated	(Chapter	6).	Finally,	the	gene	expression	signature	of	single	mouse	and	hESC-derived	endocrine	progenitors	was	examined	(Chapter	7).	Together,	these	data	provide	insight	into	how	Neurog3+	endocrine	progenitors	form	and	inform	upon	potential	methods	to	improve	current	hESC	differentiation	protocols.		 	 21 Chapter	2: Materials	and	methods	2.1 Animals	Mice	were	housed	on	a	12-hour	light-dark	cycle	in	a	climate-controlled	environment	according	to	protocols	(A13-0184	and	A14-0163)	approved	by	the	University	of	British	Columbia	Animal	Care	Committee.	CD-1	mice	were	obtained	from	Charles	River.	Rosa26mT/mG	(Stock	No:	007576)128	and	Neurog3-Cre	(Stock	No:	005667)	mice	were	purchased	from	Jackson	Laboratory.	KrasLSL-G12D:	B6.129S4-Krastm4Tyj/J	(Stock	No.	008179)129	and	tetO-Cdkn1b:	C57BL/6(tetO-Cdkn1b)1Scpr/J	(Stock	No.	017613)130	mice	were	purchased	from	the	Jackson	Laboratory.	Frozen	embryos	of	Pdx1-Cre:	B6.Cg-Tg(Pdx1-cre)89.1Dam/Mmucd49	mice	were	obtained	from	the	MMRRC	(Stock	No.	015970-UCD)	and	were	rederived	by	the	BC	Preclinical	Research	Consortium	(Vancouver	BC,	Canada).	Sox9-rtTA	transgenic	mice	(C57BL/6-BAC-Sox9-rTTA)	were	made	by	recombining	the	reverse	tetracycline	transactivator	(Clontech;	pTet-On)	into	the	Sox9	coding	region	of	a	202kb	bacterial	artificial	chromosome	(RP23-36D5).	Transgenic	animals	were	generated	by	the	UCSF	transgenic	core	using	previously	published	approaches131	and	have	been	validated	herein.	To	induce	transgene	expression	in	Sox9-rtTA;	tetO-Cdkn1b	pregnant	dams	were	injected	intraperitoneally	with	1	mg	of	doxycycline	as	described	with	single	transgenic	littermate	embryos	serving	as	controls.	2.2 EdU	cumulative	labelling,	embryo	collection	and	tissue	preparation		 EdU	cumulative	labelling	studies	were	carried	out	using	the	CD1	mouse	strain	as	using	a	previously	described	protocol132.	Briefly,	noon	on	the	morning	of	the	discovery	of	a	vaginal	plug	was	considered	E0.5.	Beginning	at	9	am	on	either	gestational	day	(G)11.5,	G12.5	or	G13.5,	pregnant	dams	were	injected	intraperitoneally	with	1	mg	5-Ethynyl-2’- 22 deoxyuridine	(EdU;	Sigma-Aldrich)	for	the	first	dose	followed	by	0.25	mg	for	all	subsequent	doses.	Injections	continued	every	1.5	hours	until	a	labelling	maximum	was	reached	(10.5	hours).	Embryos	were	collected	by	hysterectomy	0.5	hours	after	the	last	injection	of	EdU.	The	pancreas	was	dissected	along	with	the	stomach,	spleen,	and	gut	before	fixation	in	4%	paraformaldehyde	(PFA;	Sigma-Aldrich)	for	6-12	hours.	Tissues	were	then	embedded	in	paraffin	using	the	following	dehydration	gradient:	50%	ethanol	(EtOH)	for	24	hours;	70%	EtOH	for	24	hours;	2x30	minutes	in	95%	EtOH;	3x30	minutes	in	100%	EtOH;	2x30	minutes	in	xylene;	2x1	hour	in	paraffin.	2.3 Immunostaining	and	imaging		 For	fluorescent	staining	and	EdU	detection	of	fixed	tissue,	paraffin	slides	were	deparaffinized	and	rehydrated	in	the	following	gradient:	2x5	minutes	xylene;	3x2	minutes	100%	EtOH;	2x2	minutes	95%	EtOH;	2x2	minutes	70%	EtOH;	2x2	minutes	50%	EtOH;	2x2	minutes	H2O.	Following	rehydration,	antigen	retrieval	was	performed	for	20	minutes	in	a	95°C	citrate	buffer	(10	mM	sodium	citrate,	0.05%	Tween-20,	pH	6.0).	Slides	were	washed	in	PBS	and	permeabilized	with	0.5%	Triton	X-100	(Sigma-Aldrich)	in	PBS	before	EdU	detection	was	performed	by	incubating	slides	for	1h	with:	100	mM	Tris	pH	8.5	(ThermoFisher	Scientific),	1	mM	CuSO4	(Sigma-Aldrich),	30	 µM	Alexa	Fluor®	594	azide	triethylammonium	(ThermoFisher	Scientific),	100	mM	ascorbate	(Sigma-Aldrich)	in	H20133.	After	three	washes	in	PBS,	slides	were	blocked	for	30	minutes	each	in	blocking	buffer	(5%	horse	serum	in	PBS)	and	1:50	anti-mouse	IgG	(Jackson	ImmunoResearch)	in	blocking	buffer.	Primary	antibodies	were	incubated	at	4°C	overnight	at	the	dilutions	listed	in	Table	1.	Secondary	antibodies	from	Jackson	ImmunoResearch	were	incubated	at	room	temperature	in	blocking	buffer	for	1-2	hours	at	the	following	dilutions:	anti-FITC	(1:250)	 23 and	anti-Cy3	(1:450).	To-pro	Iodide-3	was	used	to	stain	nuclei	(Life	Technologies;	1:10,000).	Slides	were	mounted	using	SlowFade	Antifade	Reagents	(Thermo	Fisher	Scientific)	and	were	imaged	with	a	20x/0.75	objective	using	a	Leica	TCS	SP8	confocal	system.	2.4 EdU	quantification	and	calculation	of	cell	cycle	lengths		 At	least	8	sections	spaced	across	the	pancreas	of	4-13	embryos	from	2	dams	were	analyzed	at	each	time	point.	In	total,	over	100	embryos	from	62	pregnant	dams	were	used	to	quantify	more	than	50,000	pancreatic	progenitors.	Custom	CellProfiler	pipelines	were	used	for	image	analyses134.	To	determine	the	lengths	of	the	G1-,	S-,	G2/M-phases	of	the	cell	cycle,	the	percent	of	EdU+	progenitor	cells	was	plotted	against	the	labelling	time	and	the	data	was	fit	using	least	squares	linear	regression.	To	measure	the	length	of	G2/M	(TG2/M),	the	time	for	all	EdU-labelled	cells	to	exit	S,	transit	through	G2	and	enter	mitosis	(pHH3+)	after	a	single	pulse	of	EdU	was	determined	using	least	squares	linear	regression.	2.5 Explant	dissection,	culture	and	in	situ	immunostaining		 The	dorsal	pancreas	of	CD1	E11.5	embryos	was	dissected	as	previously	described135	and	transferred	to	Millicell®	EZ	slides	(EMD	Millipore)	covered	in	Geltrex	(ThermoFisher	Scientific)	diluted	1:1	in	DMEM/F12.	Explants	recovered	overnight	in	culture	medium:	DMEM/F12	(Corning),	10%	FBS	(ThermoFisher	Scientific),	GlutaMAX	(ThermoFisher	Scientific),	1x	Hyclone	Penicillin/Streptomycin	(GE	Healthcare	Life	Sciences),	and	1x	Insulin-Transferrin-Selenium	(ThermoFisher	Scientific).	The	following	morning,	media	was	topped	up	with	CDKi	for	24	hrs:	2.5	µM	CDK4/6	inhibitor	PD-0332991	(Sigma-Aldrich),	1	µM	CDK2	inhibitor	II	(EMD	Millipore)	and	1	µM	CDK2	inhibitor	III	(EMD	Millipore).	Explants	were	then	washed	with	PBS	and	fixed	in	4%	PFA	for	30	minutes.	After	removal	of	 24 PFA,	explants	were	washed	three	times	with	PBS	and	blocking	solution	(PBS,	0.1%	Triton	X-100,	and	3%	horse	serum)	was	added	for	a	minimum	of	1	hr.	Primary	antibodies	(mouse	anti-PDX1	[DSHB;	1:100]	and	rabbit	anti-SOX9	[Millipore;	1:500])	in	blocking	solution	were	added	to	explants	and	incubated	overnight	at	4°C.	The	following	morning,	after	three	PBS	washes	explants	were	incubated	with	secondary	antibodies	in	blocking	solution	for	1	hr:	anti-mouse	FITC	(1:250),	anti-rabbit	594	(1:450)	and	nuclei	dye	To-Pro-3	iodide	(1:10,000).	Explants	were	imaged	on	Leica	SP8	confocal	using	z-stacks	to	image	whole	explant	and	images	were	quantified	using	custom	CellProfiler	pipelines.	2.6 Maintenance	of	pluripotent	cells	Undifferentiated	CyT49	hESCs	(ViaCyte,	Inc.	San	Diego	CA)	were	maintained	on	Geltrex	(ThermoFisher	Scientific;	diluted	1:100	with	DMEM/F12)	in	10/10	media	[DMEM/F12,	10%	XenoFree	KnockOut™	Serum	Replacement	(ThermoFisher	Scientific),	1x	MEM	non-essential	amino	acids	(ThermoFisher	Scientific),	1x	GlutaMAX,	1x	penicillin/streptomycin,	10	nM	β-mercaptoethanol	(Sigma-Aldrich),	supplemented	with	10	ng/mL	ACTIVIN	A	(E-biosciences)	and	10	ng/mL	HEREGULIN-β1	(Peprotech)]65,136.	Cells	were	split	every	two-three	days	and	plated	at	a	density	of	1x106	per	60	mm	plate.	2.7 Generation	of	induced	pluripotent	stem	cells	Human	iPSCs	were	generated	in	house	from	male	human	PBMCs	by	infecting	erythroid	progenitors	(Erythroid	Progenitor	Reprogramming	Kit;	StemCell	Technologies)	with	Sendai	virus	as	outlined	in	the	manufacturer’s	protocol	(Life	Technologies;	Cytotune	2.0).	Cells	were	plated	on	Geltrex	after	48	hours	and	subsequently	transitioned	to	ReproTeSR	from	days	4–7	post-infection	(StemCell	Technologies).	At	3	weeks	post-infection,	iPSC	clones	were	picked	into	and	maintained	in	mTeSR-E8	(StemCell	 25 Technologies)	in	Geltrex-coated	96-well	plates.	Clones	were	passaged	using	ReLeSR	(Stem	Cell	Technologies)	every	4–6	days	until	passage	15.	Sendai	virus	transgene	expression	was	then	analysed,	and	found	to	be	absent,	using	Taqman	(Life	Technologies)	and	pluripotency	assessed	by	immunostaining	and	qPCR.	2.8 Generation	of	OCT4-eGFP	pluripotent	stem	cell	lines	Transcription	activator-like	effector	nuclease	(TALEN)s	were	generated	in	house	using	the	TALE	toolbox	(pTALEN_v2)137.	Guanine	binding	was	encoded	by	the	repeat-variable	diresidue	Asn-His	(NH)	as	described138.	TALEN	binding	sites	flanked	the	stop	codon	of	the	OCT4	gene	with	the	forward	TALEN	designed	to	bind	to	the	sequence:	5’-	TCTGGGCTCTCCCATGCATT-3’	and	the	reverse	TALEN	to	the	sequence:	5’-	TCCCCCATTCCTAGAAGGGC-3’.	The	CRISPR/Cas	vector	was	based	on	a	px458	(Addgene;	plasmid	48138);	however,	the	Cbh	promoter	was	exchanged	for	a	full-length	CAGGS	promoter	in	order	to	maximize	hESC	expression	(pCCC).	The	gRNA	(AGAGTGGTGACGGAGACAGG;	score	0.6)	was	designed	using	the	algorithm	reported	by	Doench	et	al.139	and	was	cloned	into	the	BbsI	sites	of	pCCC	to	generate	pCCC-LL488	as	described	by	Ran	et	al.140.	The	targeting	vector	was	obtained	from	Addgene	(plasmid	31939)	and	has	been	previously	described141.	CyT49	hESC	were	cultured	in	10/10	media	with	1	μM	Y-27632	dihydrochloride	(Tocris	Bioscience)	for	2	hours	prior	to	electroporation.	Cells	were	washed	with	PBS	before	trypsinization	with	Accutase	(STEMCELL	Technologies)	for	5	minutes	at	37°C.	Following	detachment,	cells	were	centrifuged	at	200xg	for	5	minutes	before	being	washed	three	times	in	100-pellet	volumes	of	PBS.	107	cells	were	resuspended	in	Embryomax®	Electroporation	Buffer	(Millipore),	transferred	to	a	0.4	cm	cuvette	with	40	μg	of	OCT4-eGFP-2A-Puro	donor	 26 plasmid	and	15	μg	of	each	TALEN	encoding	plasmid	(or	15	μg	pCCC-LL488),	and	electroporated	using	Bio-Rad	Gene	Pulser	II	system	(250	V,	500	μF,	time	constants	<13	ms).	 After	electroporation,	cells	were	resuspended	in	10/10	media	with	1	μM	Y-27632	dihydrochloride	and	plated	onto	a	10	cm	Geltrex-coated	tissue	culture	dish	(BD	Biosciences).	Media	was	replaced	daily	with	10/10	and	cells	were	allowed	to	recover	for	four	days	before	selecting	with	0.25	μg/mL	puromycin	(Sigma-Aldrich).	Colonies	were	picked	into	a	96-well	Geltrex-coated	plate	within	10	days	of	electroporation	by	manually	scraping	and	pipetting	the	colony	off	the	plate	and	into	a	well	with	100	μL	of	10/10.	Once	clones	were	close	to	confluent,	cells	were	replica	plated	onto	three	plates:	one	to	genotype,	one	to	freeze	down	and	one	to	expand	the	correctly	targeted	clones.	Genomic	DNA	was	extracted	using	QuickExtract	(Epicentre)	and	the	following	primers	were	used	to	genotype:	LL317:	5’F	CTCAGTTCTGCTGGGATAAG;	LL318:	5’R	GTCTTGTAGTTGCCGTCGTC;	LL319:	3’F	GCAACCTCCCCTTCTACGAG;	LL320:	3’R	CTTACACCAAGCCAAACTATTG.	2.9 Generation	of	NEUROG3-2A-eGFP	pluripotent	stem	cell	line	To	generate	NEUROG3-2A-eGFP	knock-in	hESC	line,	the	CRISPR/Cas9	system	was	used	as	previously	described142.	pCCC	contains	a	full-length	CAGGS	promoter	to	replace	the	Cbh	promoter	of	px458,	improving	expression	in	hESC	CyT49	cells.	The	gRNA	(GGGTCGCTCCTCCAGCGACG;	score	0.73)	was	designed	using	the	algorithm	reported	by	Doench	et	al139	and	was	cloned	into	the	BbsI	sites	of	pCCC	to	generate	pCCC-LL502/3	as	described	by	Ran	et	al140.	The	targeting	vector	was	based	on	Addgene	#31938	and	contained	~800	bp	NEUROG3	homology	arms.	Electroporation,	selection,	and	picking	of	clones	was	performed	as	previously	described	in	section	2.8	using	40	µg	of	donor	and	15	µg	 27 of	CRISPR/Cas	vectors.	Genomic	DNA	was	extracted	using	QuickExtract	(Epicentre)	and	the	following	primers	were	used	to	genotype:	LL529:	5’F	GGTAGAAAGGTAATATTTGGAGGCCT;	LL530:	5’R	CTGAACTTGTGGCCGTTTACG;	LL351:	3’F	ATCAGCAGCCTCTGTTCCAC;	LL352:	3’R	GGAGTCATCTTGCCAAGGCT.	2.10 Generation	of	FUCCI	transgenic	pluripotent	stem	cell	line	To	generate	FUCCI	transgenic	CyT49,	106	cells	were	plated	into	6-well	plate	and	4	µg	of	FUCCI	construct	was	transfected	using	Lipofectamine	3000	(Life	Technologies).	Cells	were	selected	for	integration	using	0.25	μg/mL	puromycin	and	clones	were	picked	based	on	fluorescence	expression.	mKO-hCDT-2A-mAG-hGEM-2A-Puro	was	cloned	into	a	custom	pCAGGS	expression	vector.	2.11 In	vitro	differentiation	of	hESC	Protocol	1:	The	differentiation	protocol	was	adapted	from	Schulz	et	al.136.	Briefly,	to	produce	definitive	endoderm	cells	were	treated	with	ACTIVIN	A	(100	ng/mL;	eBioscience),	WNT3A	(25	ng/mL;	R&D),	and	1:5000	Insulin-Transferrin-Selenium	in	RPMI	(0.5x	penicillin/streptomycin,	1x	GlutaMAX;	Hyclone)	for	24	hours.	Cells	were	in	the	same	media	supplemented	with	0.2%	defined	FBS	and	without	WNT3a	for	another	48	hours.	To	generate	primitive	gut	tube,	cells	were	first	treated	with	KGF	(25	ng/mL;	R&D),	TGF-β	RI	kinase	inhibitor	(2.5	μM;	EMB	Bioscience),	0.2%	defined	FBS	and	1:1000	ITS.	Cells	were	then	treated	as	previously	but	without	TGF-β	RI	kinase	inhibitor	for	48	hours.	To	produce	posterior	foregut,	cells	were	treated	for	36	hours	in	TTNPB	(3	nM;	Sigma),	cyclopamine-KAAD	(0.25	μM;	Toronto	Research	Chemicals),	Noggin	(50	ng/mL;	R&D),	0.5x	B27	(Gibco)	in	DMEM	High	Glucose	(0.5x	penicillin/streptomycin,	1x	GlutaMAX;	Hyclone).	Finally,	to	produce	pancreatic	progenitors	and	endocrine	precursors,	cells	were	treated	for	36	hours	 28 in	Noggin	(50	ng/mL;	R&D),	KGF	(50	ng/mL;	R&D),	EGF	(50	ng/mL;	R&D)	in	DMEM	High	Glucose.	Protocol	2:	CyT49	or	hIPSCs	were	plated	onto	Geltrex	(1:100)-coated	12	well	plates	at	a	density	of	5e5	in	10/10	media	[DMEM/F12,	10%	Xenofree-KOSR,	GlutaMAX,	P/S,	10	ng/mL	ACTIVIN	A	and	10	ng/mL	HEREGULIN-1b	(Peprotech)].	Differentiations	began	48	hours	post-seeding	using	a	modified	version	of	Rezania	et	al69.	Briefly,	cells	were	rinsed	with	1	x	PBS	and	then	basal	culture	media	(MCDB	131	medium	(US	Biological	Life	Sciences),	1.5	g/L	sodium	bicarbonate	(Sigma-Aldrich),	1	x	GlutaMAX	(ThermoFisher	Scientific),	1	x	P/S	(ThermoFisher	Scientific))	with	10	mM	final	glucose	(Sigma-Aldrich),	0.5%	BSA	(Sigma-Aldrich),	100	ng/mL	ACTIVIN	A,	and	3	μM	of	CHIR-99021	(Sigma-Aldrich)	was	added	for	1	day	only.	For	the	following	two	days,	cells	were	treated	with	the	same	media	without	CHIR-99021	compound	to	generate	definitive	endoderm	(Stage	1).	On	day	four,	cells	were	cultured	in	basal	media	with	0.5%	BSA,	10	mM	glucose,	0.25	mM	ascorbate	(Sigma-Aldrich)	and	50	ng/mL	of	KGF	(R&D	or	StemCell	Technologies)	for	2	days	to	generate	primitive	gut	tube	(Stage	2).	To	produce	posterior	foregut	(Stage	3),	cells	were	treated	for	three	days	with	basal	media	with	10	mM	final	glucose	concentration,	2%	BSA,	0.25	mM	ascorbate,	50	ng/mL	of	KGF,	0.25	μM	SANT-1	(Tocris	Biosciences),	1	μM	retinoic	acid	(Sigma-Aldrich),	100	nM	LDN193189	(EMD	Millipore),	1:200	ITS-X,	and	200	nM	a-Amyloid	Precursor	Protein	Modulator	(APPM;	EMD	Millipore).	For	stage	4,	cells	were	treated	with	basal	media	with	10	mM	glucose,	2%	BSA,	0.25	mM	ascorbic	acid,	2	ng/mL	of	KGF,	0.25	μM	SANT-1,	0.1	μM	retinoic	acid,	200	nM	LDN193189,	1:200	ITS-X,	and	100	nM	APPM	for	3	days	to	generate	pancreatic	progenitors.	Cells	were	maintained	as	planar	cultures	and	media	was	changed	to	basal	media	with	20	mM	glucose,	2%	BSA,	0.25	μM	 29 SANT-1,	0.05	μM	retinoic	acid,	100	nM	LDN193189,	1:200	ITS-X,	1	μM	T3	(Sigma-Aldrich),	10	μM	Repsox	(Sigma-Aldrich),	and	10	μM	zinc	sulfate	(Sigma-Aldrich)	for	3	days	to	generate	pancreatic	endocrine	precursors	(Stage	5).	Finally,	cells	were	treated	for	two	days	in	stage	6	media:	basal	media	with	20	mM	final	glucose	concentration,	2%	BSA,	100	nM	LDN193189,	1:200	ITS-X,	1	μM	T3,	10	μM	Repsox,	10	μM	zinc	sulfate,	100	nM	gamma	secretase	inhibitor	XX	(EMD	Millipore).	2.12 RNA	isolation	and	RT-PCR	analysis	Cells	were	lysed	using	TRIzol	(ThermoFisher	Scientific).	For	1	mL	of	TRIzol,	200	μL	of	chloroform	was	added	to	lysed	cells	and	vortexed	well	before	centrifugation	at	12,000xg	for	15	minutes.	Next,	the	clear	aqueous	phase	was	transferred	to	a	new	tube	and	RNA	was	precipitated	using	500	μL	of	isopropanol	and	1	μL	of	glycogen	if	necessary.	Following	centrifugation	at	12,000xg	for	10	minutes,	the	RNA	pellet	was	washed	twice	in	70%	EtOH	in	DEPC-H2O	before	resuspension	in	44	μL	of	water.	For	DNAse	treatment,	TURBO	DNA-free	Kit	(Qiagen)	was	used	according	to	manufactures	protocol.	Gene	expression	analysis	was	determined	using	ΔΔCT	relative	to	the	housekeeping	gene,	TATA-binding	protein	(TBP).	For	a	list	of	TAQMAN	primers	used,	see	Table	2.	2.13 Western	blot	analysis	Lysis	buffer	(95°C)	was	used	to	lyse	cells	and	protein	was	denatured	by	boiling	at	95°C	for	10	minutes	before	sonication	(S-4000	with	cuphorn;	Misonix)	for	2	minutes	(80%).	Cells	were	then	centrifuged	at	10,000xg	for	5	minutes	at	20°C	and	supernatant	was	collected.	Lysates	were	subjected	to	standard	SDS-PAGE	followed	by	blotting	onto	nitrocellulose	membrane	(Biorad).	Blots	were	then	blocked	with	5%	skim	milk	powder	in	Tris-buffered	saline	with	Tween	20	(0.1%)	and	probed	with	rabbit	anti-human	OCT4	(Cell	 30 Signaling;	1:1000),	anti-GFP	(MBL;	1:1000)	or	mouse	anti-GAPDH	(Sigma;	1:125,000)	overnight	at	4°C	in	blocking	buffer.	The	next	day	membranes	were	probed	with	horseradish	peroxidase-conjugated	secondary	antibodies	at	1:10,000	(Jackson	ImmunoResearch)	for	1	hour	and	visualized	with	ECL	Prime	(GE	Biosciences).	For	detection	of	Neurog3	during	development,	embryonic	pancreata	from	five	E15.5	CD1	embryos	were	pooled	and	lysed	in	non-reducing	sample	buffer	(62.5	mM	Tris	Buffer	pH	6.8,	1	mM	sodium	vanadate	(activated),	1	mM	sodium	fluoride,	2%	w/v	SDS,	10%	glycerol)	with	protease	inhibitors	and	analyzed	by	western	blot	as	described	above	using	guinea	pig	anti-NEUROG3	(see	Table	1).	2.14 Flow	cytometry	and	FACS	Cells	were	rinsed	once	with	PBS	and	detached	from	the	plate	using	Accutase	(Life	Technologies).	Cells	were	centrifuged	for	5	minutes	at	200xg	before	being	resuspended	in	4%	PFA	and	fixed	for	15	minutes.	Subsequently,	cells	were	rinsed	twice	in	PBS	before	analyses	on	a	BD	FACSCalibur	flow	cytometer	for	GFP	expression.	To	sort	for	eGFP+	and	GFP–	populations,	cells	were	trypsinized,	washed	in	PBS	and	sorted	directly	into	TRIzol	(Life	Technologies)	using	a	BD	FACS	Aria.	To	determine	endogenous	OCT4	expression,	cells	were	fixed	as	above,	permeabilized	in	0.5%	Triton	X-100	and	stained	overnight	at	4°C	with	rabbit	anti-human	OCT4	antibody	(1:100;	Cell	Signaling).	The	next	morning	cells	were	rinsed	three	times	in	PBS	before	incubation	with	secondary	antibody	anti-rabbit	FITC	(1:250;	Jackson	Immunoresearch)	for	1	hour	at	room	temperature.	Cells	were	analyzed	on	a	BD	FACSCalibur	using	appropriate	unstained	and	secondary	antibody	only	controls.	To	determine	CXCR4+,	cells	were	fixed	in	4%	PFA	for	15	minutes,	washed	three	times	in	PBS	 31 and	incubated	with	CXCR4-PE	antibody	(1:20;	R&D)	for	45	minutes.	After	cells	were	washed	well,	they	were	analyzed	using	a	BD	FACSCalibur.	CDKi	treatment	(2.5	µM	CDK4/6	inhibitor	PD-0332991,	1	µm	CDK2	inhibitor	II	and	1	µm	CDK2	inhibitor	III)	was	carried	out	for	24	hours	in	Stage	5	or	6	using	Differentiation	Protocol	2.	For	flow	cytometric	analysis,	cells	were	treated	with	Accutase	to	generate	single	cells	and	fixed	in	4%	PFA	for	15	min	before	analysis	for	GFP	expression	on	BD	Canto	or	for	FUCCI	expression	on	BD	Fortessa.	2.15 Immunocytochemical	analyses	Cells	were	grown	and	differentiated	on	Geltrex-coated	35	mm	optical	dishes	(MatTek).	On	the	day	of	collection,	cells	were	rinsed	once	in	PBS	before	fixation	in	4%	PFA	for	15	minutes.	Cells	were	permeabilized	with	0.5%	Triton	X-100	in	PBS	for	30	minutes,	blocked	for	30	minutes	in	5%	horse	serum	in	PBS	and	stained	with	primary	antibodies	overnight	at	4°C:	mouse	anti-OCT4	antibody	(1:100;	Cell	Signaling),	mouse	anti-SOX2	(1:100;	Cell	Signaling),	mouse	anti-NANOG	(1:100;	Cell	Signaling),	rabbit	anti-SOX9	(1:500;	Millipore),	and	mouse	anti-NKX6-1	(1:100;	DSHB).	The	following	morning,	dishes	were	washed	three	times	with	PBS	and	stained	with	secondary	antibodies	for	1	hour:	anti-mouse	Dy-488	(1:250;	Jackson	Immunoresearch),	anti-mouse	Dy-594	(1:450;	Jackson	Immunoresearch),	anti-rabbit	Dy-594	(1:450;	Jackson	Immunoresearch),	and	TO-PRO®	Iodide	(1:10,000;	Life	Technologies).	Images	were	taken	using	63x	oil	immersion	objective	on	a	Leica	TCS	SP8	confocal	microscope.	 32 2.16 Nanostring	nCounter	SPRINT™	protocol	Cells	were	lysed	in	350	µL	of	Buffer	RLT	(Qiagen).	Custom	nCounter	Gene	Expression	Reporter	CodeSet	and	Capture	ProbeSet	were	designed	(Table	3).	nCounter	assays	were	set	up	according	to	manufactures	instructions.	Briefly,	10	µL	of	Reporter	Codeset,	10	µL	of	hybridization	buffer,	3.5	µL	of	H2O,	and	1.5	µL	of	total	cell	lysate	was	incubated	at	65°C	for	12	hours.	Immediately	following	hybridization	the	sample	cartridge	was	loaded	with	the	30	µL	of	each	sample	and	run	on	the	nCounter	SPRINT	Profiler.	Data	was	analyzed	using	nSolver	Analysis	Software	3.0	and	normalized	to	reference	genes.	2.17 Preparing	cells	for	single	cell	RNA-sequencing	For	mouse	studies,	Neurog3-Cre;	Rosa26mTmG	embryos	were	collected	on	E15.5	and	E18.5	and	dissected	on	ice.	For	library	generation,	one	embryo	from	each	time	point	was	used.	To	generate	single	cells,	embryonic	pancreases	were	incubated	in	2	mL	of	pre-warmed	37°C	0.25%	Trypsin	with	mild	agitation	for	8	or	20	minutes	for	E15.5	and	E18.5	pancreases,	respectively.	To	stop	digestion,	1	mL	of	cold	FBS	and	2	mL	of	cold	PBS	were	added	and	mixed	by	inversion	to	stop	digestion,	followed	by	pipette	filtering	with	a	40	µm	nylon	filter.	Cells	were	then	centrifuged	at	4°C	for	5	minutes	at	200xg.	After	aspirating	the	supernatant,	cells	were	resuspended	in	cold	2%	FBS	in	PBS,	placed	on	ice,	and	immediately	sorted	into	mTomato+,	mTomato+mGreen+	(yellow),	and	mGreen+	fractions	using	a	Beckman	Coulter	MoFlo	Astrios	(Mississauga,	ON,	Canada)	into	20%	FBS	in	PBS.	For	hESC	studies,	N5-5	cells	were	differentiated	using	protocol	2	and	cells	were	collected	following	three	days	at	stage	5	(S6D1).	Cells	were	washed	once	with	PBS	before	500	µL	of	Accutase	was	added	per	well	of	a	12-well	plate.	Following	5	minutes	at	37°C,	500	µL	of	2%	BSA	MCDB	media	was	added	to	each	well	and	cells	were	transferred	to	a	15	mL	 33 conical	tube.	Cells	were	centrifuged	for	5	minutes	at	200xg,	washed	once	with	PBS,	and	resuspended	in	350	µL	of	ice	cold	PBS.	GFP+	cells	were	sorted	into	stage	5	media	with	10	µM	Y-27632	dihydrochloride	using	Beckman	Coulter	MoFlo	Astrios.	2.18 Generating	scRNA-sequencing	libraries	The	10x	Genomics	ChromiumTM	controller	and	Single	Cell	3’	Reagent	Kits	v2	(Pleasanton,	CA,	USA)	were	used	to	generate	single	cell	libraries.	Briefly,	cells	were	counted	following	FACS	and	cell	suspensions	were	loaded	for	a	Targeted	Cell	Recovery	of	3000	cells	per	channel.	The	microfluidics	platform	was	used	to	barcode	single	cells	using	Gel	Bead-In-Emulsions	(GEMs).	RT	is	performed	within	GEMs,	resulting	in	barcoded	cDNA	from	single	cells.	The	full	length,	barcoded	cDNA	is	PCR	amplified	followed	by	enzymatic	fragmentation	and	SPRI	double	sided	size	selection	for	optimal	cDNA	size.	End	repair,	A-tailing,	Adaptor	Ligation	and	PCR	are	performed	to	generate	the	final	libraries	that	have	P5	and	P7	primers	compatible	with	Illumina	sequencing.	The	libraries	were	pooled	and	sequenced	using	an	Illumina	NextSeq	500	platform	with	a	150	cycle	High	Output	v2	kit	in	paired-end	format	with	26	bp	Read	1,	8	bp	I5	Index,	and	85	bp	Read	2.	2.19 Bioinformatic	analysis	of	scRNA-sequencing	data	10X	Genomics’	Cell	Ranger	software	was	used	to	de-multiplex	single-cell	samples	prepared	by	the	Chromium	v2	kit,	convert	raw	read	data	to	FASTQ	format,	and	generate	gene	count	data	per	cell.	Cell	Ranger	performed	sequence	alignment	via	STAR	aligner143	to	either	the	ENSEMBL	reference	build	transcriptome	GRCm38	for	mouse	or	GRCh38	for	human,	followed	by	count	and	single-cell	barcode	calling	(https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/algorithms/overview).	ENSEMBL	FASTQ	files	and	 34 transcriptome	annotation	GTF	files	were	also	appended	with	Rosa26mT/mG	mTomato	and	mGreen	sequences	for	mouse	and	eGFP	sequence	for	human	to	enable	alignment	and	count	information.	2.20 Statistical	analyses	Statistical	analyses	were	performed	using	Prism	6	(GraphPad	Software)	or	RStudio.	All	data	are	presented	as	mean	±	SEM.	Statistically	significant	differences	were	assessed	using	ANOVA	followed	by	Tukey	tests,	t-test	or	non-parametric	tests	as	appropriate,	with	p	value	of	<	0.05	deemed	significant.	Segmental	linear	regression	using	1000	iterations	was	used	to	mathematically	model	EdU	cumulative	labelling,	to	determine	y-intercept,	slope,	and	plateau.	Custom	R	package	was	generated	with	statistical	support	from	R.	While,	T.	Zhao	and	J.	Petkau	at	the	University	of	British	Columbia	to	calculate	SEM	for	TC,	TG1,	TS	and	TG2/M	(see	Appendix	A).		 	 35 Table	1:	Antibody	list		 Antibody	 Source	 Concentration	 Catalog	Donkey	polyclonal	anti-mouse	IgG	 Jackson	ImmunoResearch	 1:50	 Cat#715-001-003;	RRID:AB_2307338	Goat	polyclonal	anti-Cpa1	 R&D	systems	 1:250	 Cat#AF2765;	RRID:AB_2085841	Guinea	pig	anti-NEUROG3	 Schwitzgebel	et	al.47	 1:500	 German	Lab	Guinea	pig	polyclonal	anti-Insulin	 DAKO	 1:1000	 Cat#A056401-2;	RRID:AB_2617169	Mouse	monoclonal	anti-Glucagon	 Sigma-Aldrich	 1:2000	 Cat#G2654;	RRID:AB_259852	Mouse	monoclonal	anti-Neurog3	 Developmental	Studies	Hybridoma	Bank	 1:100	 Cat#F25A1B3;	RRID:	AB_528401	Mouse	monoclonal	anti-Pdx1	 Developmental	Studies	Hybridoma	Bank	 1:100	 Cat#F109-D12;	RRID:	AB_1157903	Rabbit	polyclonal	anti-ChromograninA	 Thermo	Scientific	 1:400	 Cat#RB-9003-P1;	RRID:	AB_149730	Rabbit	polyclonal	anti-Cdkn1b	 Cell	Signaling	Technology	 1:500	 Cat#2552S;	RRID:AB_10693314	Rabbit	polyclonal	anti-pHH3	 Cell	Signaling	Technology	 1:1000	 Cat#9701	Rabbit	polyclonal	anti-Sox9	 Millipore	 1:500	 Cat#AB5535;	RRID:AB_2239761	Sheep	polyclonal	anti-NEUROG3	 R&D	systems	 1:500	 Cat#AF3444;	RRID:AB_2149527	Hamster	monoclonal	anti-Mucin	 Thermo	Scientific	 1:500	 Cat#HM-1630	Rabbit	polyclonal	anti-GFP	 MBL	 1:1000	 Cat#598	Mouse	monoclonal	anti-Nkx6-1	 Developmental	Studies	Hybridoma	Bank	 1:100	 Cat#F55A10	Rabbit	polyclonal	anti-Oct4	 Cell	Signaling	Technology	 1:1000	 Cat#2750	Mouse	polyclonal	anti-Nanog	 Cell	Signaling	Technology	 1:100	 Cat#3580	Rabbit	polyclonal	anti-Sox2	 Cell	Signaling	Technology	 1:100	 Cat#3579			 	 36 Table	2:	List	of	TAQMAN	qPCR	primers		Gene	Name	 Probe	 Primer	1	 Primer	2	 Dye	CER1	 CCCTTCAGCCAGACTATAACCCACG	 CTTGCCCATCAAAAGCCATG	 CCAGGAAAATGAACAGACCCG	 FAM/ZEN/IABKFQ	GSC	 CTGGCCCGGAAAGTGCACC	 TCTTCCAGGAGACCAAGTACC	 GATGAGGACCGCTTCTGC	 FAM/ZEN/IABKFQ	INS	 CGGCGGGTCTTGGGTGTGTA	 CTAGTGTGCGGGGAACG	 CACGCTTCTGCAGGGAC	 FAM/ZEN/IABKFQ	NEUROD1	 CGCCAGTTTCACCATTTCCGGG	 TCCTTCGATAGCCATTCACATC	 GCTGCCTTTTGTAAACACGAC	 FAM/ZEN/IABKFQ	NEUROG3	 AGCGGTCCTTCCCCAGAGC	 TCTATTCTTTTGCGCCGGTAG	 GCAGGTCACTTCGTCTTCC	 FAM/ZEN/IABKFQ	NKX6-1	 TGCTTCTTCCTCCACTTGGTCCG	 TCGTTTGGCCTATTCGTTGG	 TGTCTCCGAGTCCTGCTTC	 FAM/ZEN/IABKFQ	OCT4	 CCCCCTGTCCCCCATTCCTAGA	 TCTCCCATGCATTCAAACTGAG	 CCTTTGTGTTCCCAATTCCTTC	 FAM/ZEN/IABKFQ	PDX1	 CGCTTGTTCTCCTCCGGCTCC	 TGAAGTCTACCAAAGCTCACG	 GGAACTCCTTCTCCAGCTCTA	 FAM/ZEN/IABKFQ	SOX17	 TCCACGACTTGCCCAGCATCTT	 AGAATCCAGACCTGCACAAC	 GCCGGTACTTGTAGTTGGG	 FAM/ZEN/IABKFQ	SOX7	 CCTTCCACGACTTTCCCAGCATGT	 CACAACGCCGAGCTCAG	 GGCCGGTACTTGTAGTTCG	 FAM/ZEN/IABKFQ	TBP	 TGGGATTATATTCGGCGTTTCGGGC	 GAGAGTTCTGGGATTGTACCG	 ATCCTCATGATTACCGCAGC	 FAM/ZEN/IABKFQ	 37 Table	3:	List	of	target	sequences	for	Nanostring	Sprint		 Gene	Name	 Target	Sequence	CHGA	 CTGCGCCGGGCAAGTCACTGCGCTCCCTGTGAACAGCCCTATGAATAAAGGGGATACCGAGGTGATGAAATGCATCGTTGAGGTCATCTCCGACACACTT	CHGB	 ACAGGGAGGAAGCTAGGTTTCAAGATAAACAATATAGCTCCCATCACACAGCTGAAAAGAGGAAGAGATTAGGGGAACTGTTCAACCCATACTACGACCC	GAPDH	 CACTCCTCCACCTTTGACGCTGGGGCTGGCATTGCCCTCAACGACCACTTTGTCAAGCTCATTTCCTGGTATGACAACGAATTTGGCTACAGCAACAGGG	GCG	 TGGACTCCAGGCGTGCCCAAGATTTTGTGCAGTGGTTGATGAATACCAAGAGGAACAGGAATAACATTGCCAAACGTCACGATGAATTTGAGAGACATGC	GUSB	 CCGATTTCATGACTGAACAGTCACCGACGAGAGTGCTGGGGAATAAAAAGGGGATCTTCACTCGGCAGAGACAACCAAAAAGTGCAGCGTTCCTTTTGCG	IAPP	 ATTCTCTCATCTACCAACGTGGGATCCAATACATATGGCAAGAGGAATGCAGTAGAGGTTTTAAAGAGAGAGCCACTGAATTACTTGCCCCTTTAGAGGA	INS	 GGGTCCCTGCAGAAGCGTGGCATTGTGGAACAATGCTGTACCAGCATCTGCTCCCTCTACCAGCTGGAGAACTACTGCAACTAGACGCAGCCCGCAGGCA	MAFA	 AGCTGCCCAGCAGCCCGCTGGCCATCGAGTACGTCAACGACTTCGACCTGATGAAGTTCGAGGTGAAGAAGGAGCCTCCCGAGGCCGAGCGCTTCTGCCA	MAFB	 GGCGGCGAGGCATAGTCCCGAGAAGTCACCAAGGCCATCTGGAGACTCCTGGCTTTCTGAACTTTGCGCGTTAAGCCGGGACAGCTGCTTTGCTGCCCGG	NEUROD1	 GTGCCCAGCTCAATGCCATATTTCATGATTAGAGGCACGCCAGTTTCACCATTTCCGGGAAACGAACCCACTGTGCTTACAGTGACTGTCGTGTTTACAA	NEUROG3	 CGCGCGAAGTGGGCATTGCAAAGTGCGCTCATTTTAGGCCTCCTCTCTGCCACCACCCCATAATCTCATTCAAAGAATACTAGAATGGTAGCACTACCCG	NKX2-2	 TAATTATTATTATGGAGTCGAGTTGACTCTCGGCTCCACTAGGGAGGCGCCGGGAGGTTGCCTGCGTCTCCTTGGAGTGGCAGATTCCACCCACCCAGCT	NKX6-1	 CTGGCCTGTACCCCTCATCAAGGATCCATTTTGTTGGACAAAGACGGGAAGAGAAAACACACGAGACCCACTTTTTCCGGACAGCAGATCTTCGCCCTGG	PAX4	 CTGCCCATTGTCCTTACCGTCCTGCCCATACAGACTGTGGCTCCTTCCTCCTTCCTGTGATTGCTCCCTCCTGTGTGGACGTTGCCTGGCCCTGCCTCGA	PAX6	 GAACATCCTTTACCCAAGAGCAAATTGAGGCCCTGGAGAAAGAGTTTGAGAGAACCCATTATCCAGATGTGTTTGCCCGAGAAAGACTAGCAGCCAAAAT	PDX1	 GGGAGCCGAGCCGGGCGTCCTGGAGGAGCCCAACCGCGTCCAGCTGCCTTTCCCATGGATGAAGTCTACCAAAGCTCACGCGTGGAAAGGCCAGTGGGCA	PPY	 TATAATGCCACCTTCTGTCTCCTACGACTCCATGAGCAGCGCCAGCCCAGCTCTCCCCTCTGCACCCTTGGCTCTGGCCAAAGCTTGCTCCCTGCTCCCA	RFX6	 TGCCCAGATTGCCAGACCAGCTCTCTTTGACCAGCATGTCGTTAATTCTATGGTGTCTGATATTGAAAGGGTTGATTTGAACAGCATTGGCTCTCAAGCC	SOX2	 CTTAAGCCTTTCCAAAAAATAATAATAACAATCATCGGCGGCGGCAGGATCGGCCAGAGGAGGAGGGAAGCGCTTTTTTTGATCCTGATTCCAGTTTGCC	SOX4	 GTTCACGGTCAAACTGAAATGGATTTGCACGTTGGGGAGCTGGCGGCGGCGGCTGCTGGGCCTCCGCCTTCTTTTCTACGTGAAATCAGTGAGGTGAGAC	 38 Gene	Name	 Target	Sequence	SOX9	 CAGTGGCCAGGCCAACCTTGGCTAAATGGAGCAGCGAAATCAACGAGAAACTGGACTTTTTAAACCCTCTTCAGAGCAAGCGTGGAGGATGATGGAGAAT	SST	 AGCTGCTGTCTGAACCCAACCAGACGGAGAATGATGCCCTGGAACCTGAAGATCTGTCCCAGGCTGCTGAGCAGGATGAAATGAGGCTTGAGCTGCAGAG	TBP	 ACAGTGAATCTTGGTTGTAAACTTGACCTAAAGACCATTGCACTTCGTGCCCGAAACGCCGAATATAATCCCAAGCGGTTTGCTGCGGTAATCATGAGGA	UCN3	 GTGATCTGTCACTTTTATATACACACAGGGGAGGGGACCGTTTCCATAGAGAGGGAATATCACAGCCCACTTAGGAACAATACCGGAGAAGCAGGAGCCG			 	 39 Chapter	3: Changes	in	cell	cycle	parameters	during	early	mouse	pancreatic	development	3.1 Background	The	cell	cycle	plays	an	important	role	during	development	by	controlling	the	rate	of	cellular	division	and	consequently	the	growth	of	the	organism.	The	cell	cycle	is	divided	into	four	phases:	S-phase,	where	DNA	replication	occurs;	M-phase,	where	the	nucleus	and	cytoplasm	undergo	division;	and	two	gap	phase	(G1	and	G2)	that	separate	successive	rounds	of	DNA	replication	and	cellular	division.	Progression	through	the	cell	cycle	is	controlled	by	positive	(Cyclins/Cdks)	and	negative	cell	cycle	regulators	(CKIs).	Thus,	the	rate	through	which	a	progenitor	progresses	through	the	cell	cycle	is	dependent	on	the	balance	between	Cdks	and	CKIs.		An	important	stage	of	the	cell	cycle	with	regards	to	differentiation	is	the	G1-phase.	That	is	because	it	is	during	the	G1-phase	of	the	cell	cycle	that	a	cell	commits	to	another	round	of	cellular	division	or	exits	the	cell	cycle	and	differentiates144.	As	cells	have	the	ability	to	respond	to	external	cues	during	the	G1-phase,	it	is	not	surprising	that	it	is	from	this	stage	that	cells	initiate	differentiation145.	Supporting	the	role	of	the	G1-phase	in	progenitor	cell	differentiation,	there	is	a	correlation	between	increasing	time	spent	in	the	G1-phase,	or	G1	lengthening,	and	differentiation.	For	instance,	the	pluripotent	cells	of	an	early	embryo	have	a	truncated	cell	cycle	that	results	in	their	rapid	expansion	and	proliferation.	This	unique	cell	cycle	structure	means	that	~50-60%	of	the	cell	cycle	is	devoted	to	the	S-phase	and	the	G1-phase	is	very	short.	This	abbreviated	pluripotent	cell	cycle	is	conserved	across	many	species,	including	Xenopus146,	Danio	rerio147,	Drosophila148,	rat149,	and	mouse150.	However,	as	pluripotent	stem	cells	differentiate	cell	cycle	length	increases	and	the	G1- 40 phase	elongates151.	This	remodeling	of	the	cell	cycle	is	thought	to	be	due	to	changes	in	the	expression	of	cell	cycle	proteins150,152.	In	addition	to	changes	in	G1	length	during	pluripotent	stem	cell	differentiation,	G1	lengthening	also	correlates	with	differentiation	of	tissue	specific	progenitor	cells	including	neural	and	hematopoietic	stem	cells153.	For	example,	during	mouse	neurogenesis	there	is	a	change	in	G1	length	from	3.2	hours	to	12.4	hours	between	E11	and	E16117.	Together,	there	are	several	stem	and	progenitor	cells	that	undergo	changes	in	G1	length	around	the	time	of	their	differentiation,	suggesting	that	the	cell	cycle	may	play	a	role	in	progenitor	cell	differentiation.	While	there	is	a	correlation	between	cell	cycle	lengthening	and	differentiation	in	several	developmental	populations,	whether	this	correlation	exists	during	mouse	pancreas	development	is	unknown.	To	address	this	hypothesis,	the	in	vivo	lengths	of	each	cell	cycle	phase	during	mouse	pancreas	development	was	measured	using	cumulative	EdU	labelling.	From	this	analysis,	an	increase	in	G1	length	was	measured	across	three	progenitor	populations.	 41 3.2 Results	3.2.1 Increasing	cell	cycle	length	during	early	mouse	pancreatic	development	As	cell	cycle	lengthening	has	been	correlated	with	differentiation	of	embryonic,	neural	and	hematopoietic	stem	cells153,	I	first	set	out	to	understand	whether	a	similar	paradigm	exists	in	mouse	pancreas	development.	To	accomplish	this,	cumulative	EdU	labelling	was	used	to	experimentally	determine	the	length	of	cell	cycle	phases	in	mouse	pancreatic	progenitors	between	E11.5	and	E13.5	(Figure	5)154.	This	approach	requires	serial	injections	of	EdU	to	label	all	S-phase	cells	in	vivo.	The	time	required	for	a	cell	that	started	in	S-phase	to	complete	the	rest	of	the	cell	cycle	(G2-M-G1)	and	re-enter	S-phase	(Figure	6A)	is	measured	by	the	time	to	reach	maximal	EdU	labelling,	i.e.	the	growth	fraction	(GF;	Figure	6B),	and	equals	the	cell	cycle	length	(TC)	less	the	S-phase	length	(TS)	(X0	=	TC-TS;	Figure	6B).	Segmental	linear	regression	establishes	the	y-intercept	(y-int	=	TS/TC*GF;	Figure	6B)	and	slope	(m	=	GF/TC).	Using	these	data,	TS	=	y-int/m	and	TC	=	y-int/m	+	X0.	To	measure	the	length	of	G2/M	(TG2/M),	the	time	for	all	EdU-labelled	cells	to	transit	through	G2	and	enter	mitosis	(pHH3+)	following	a	single	bolus	of	EdU	is	experimentally	determined	(Figure	6A&E).	The	length	of	G1	(TG1)	can	then	be	calculated	by	TC	-TS	-	TG2/M.	  42   Figure	5:	Cumulative	EdU	labelling	measures	increases	in	cell	cycle	length	during	pancreas	development	(A)	Immunofluorescence	images	of	time	dependent	incorporation	of	EdU	into	pancreatic	progenitors	(Pdx1;	red	&	EdU;	green)	at	E11.5,	E12.5	and	E13.5	after	0.5,	3.5,	6.5,	8	and	11	hours	of	EdU	labelling.	(B)	Immunofluorescent	staining	for	Pdx1	(blue),	EdU	(red),	and	pHH3	(green)	in	E11.5,	E12.5	and	E13.5	embryos	sacrificed	at	1,	1.5	and	2	hours	following	a	single	injection	of	EdU;	arrowheads:	pHH3+EdU+	and	arrows:	pHH3+EdU-	pancreatic	(outlined)	epithelial	cells.	Scale	bars	=	50	μm.	 43 		Figure	6:	The	length	of	the	G1-phase	of	the	cell	cycle	increases	in	Pdx1+	pancreatic	progenitors	during	early	mouse	embryonic	development	(A)	Schematic	describing	experimental	approach	used	to	measure	the	lengths	of	the	G1	(TG1),	S	(TS)	and	G2/M	(TG2/M)	cell	cycle	phases	with	cumulative	EdU	labelling	of	cells	in	S-phase	(light	blue)	and	the	M-phase	marker	pHH3	(grey).	The	lengths	of	the	G1,	S	and	G2/M-phases	of	the	cell	cycle	were	determined	from	measurements	of	TC-TS	(green),	TS	(blue)	and	TG2/M	(red).	(B)	Cumulative	EdU	labelling	of	Pdx1+	cells	at	E11.5.	Green	arrow	(TC-TS)	indicates	the	length	of	time	it	takes	to	reach	plateau	(dotted	black	line	=	Growth	Fraction	[GF]).	y-intercept	represents	(TS/TC)*GF.	(C)	Cumulative	EdU	labelling	of	Pdx1+	cells	at	E12.5.	(D)	Cumulative	EdU	labelling	of	Pdx1+	cells	at	E13.5.	(E)	The	proportion	of	pHH3+	cells	labelled	with	EdU	after	1.0,	1.5,	2.0	and	2.5	hours	at	E11.5	(black),	E12.5	(blue),	and	E13.5	(green).	Red	arrow	indicates	the	length	of	G2/M	(TG2/M).	(F)	The	length	in	hours	of	the	G1	(black),	S	(grey),	and	G2/M	(white)	phases	of	the	cell	cycle	at	E11.5,	E12.5	and	E13.5	in	the	Pdx1+	progenitor	population.	Data	are	presented	as	mean	±	SEM.	n	=	4-13	embryos	from	2	dams	at	each	time	point.		Using	this	approach,	the	time	for	Pdx1	immunoreactive	(+)	pancreatic	progenitors	to	reach	the	GF	was	6.5±0.4,	8.0±0.5,	9.3±0.6	hours	at	E11.5,	E12.5	and	E13.5,	respectively	(Figure	6B-D).	Total	cell	cycle	length	increased	from	8.20±0.68	to	11.79±0.60	hours	while	 44 both	S-phase	length	(1.70±0.38	to	2.53±0.4)	and	G2/M	length	(1.96±0.10	to	2.07±0.05)	remained	similar	for	each	developmental	day	(Figure	6F	&	Table	4).	Therefore,	the	increases	in	total	cell	cycle	length	were	due	to	G1-phase	lengthening	from	4.54±0.40	to	7.19±0.48	hours	(Figure	6D	&	Table	4).	Table	4:	Changes	in	cell	cycle	length	during	mouse	pancreatic	development	Lengths	of	the	phases	of	the	cell	cycle	at	E11.5,	E12.5	and	E13.5	in	Pdx1+	pancreatic	epithelium,	Pdx1+Cpa1+	tip	progenitor	cells,	and	Pdx1+Cpa1-	trunk	progenitor	cells.	Data	are	presented	as	mean	±	SEM.		3.2.2 Changes	in	cell	cycle	length	between	tip	and	trunk	progenitor	cells	Close	examination	of	the	EdU	labelling	revealed	that	the	EdU+	cells	appeared	in	clusters	near	the	tips	prior	to	the	trunks	of	the	epithelium	(Figure	5)	suggesting	a	difference	in	cell	cycle	length	between	the	Pdx1+Cpa1+	tip	and	the	Pdx1+Cpa1-	trunk	progenitors42.	The	time	to	reach	the	GF	increased	from	E11.5	to	E13.5	in	both	tip	(5.7±0.3	to	7.7±0.3	hours)	and	trunk	cells	(6.4±0.4	to	9.4±0.6	hours),	suggesting	a	lengthening	of	the	cell	cycle	(Figure	7A-C).	Additionally,	on	each	developmental	day	the	time	to	reach	the	GF	differed	between	the	tip	and	trunk	cells,	resulting	in	a	G1-phase	length	that	is	~25%	longer	in	the	more	differentiated	trunk	progenitors	(4.41±0.43	to	7.36±0.64	hours)	than	in	tip		 TC		(hours)	 TG1		(hours)	 TS		(hours)	 TG2/M	(hours)	 Growth	Fraction		Pdx1+	E11.5	 8.20±0.68	 4.54±0.40	 1.70±0.38	 1.96±0.10	 0.75±0.02	E12.5	 9.73±0.86	 5.91±0.51	 1.75±0.46	 2.07±0.07	 0.73±0.01	E13.5	 11.79±0.60	 7.19±0.48	 2.53±0.48	 2.07±0.05	 0.78±0.01	Tip	E11.5	 8.41±0.74	 3.74±0.35	 2.71±0.49	 1.96±0.10	 0.94±0.02	E12.5	 10.54±0.66	 5.29±0.34	 3.18±0.41	 2.07±0.07	 0.94±0.02	E13.5	 11.57±0.66	 5.62±0.34	 3.89±0.40	 2.07±0.05	 0.95±0.02	Trunk	E11.5	 7.57±0.74	 4.41±0.43	 1.20±0.41	 1.96±0.10	 0.73±0.03	E12.5	 9.49±0.67	 6.24±0.44	 1.17±0.37	 2.07±0.07	 0.70±0.02	E13.5	 11.47±0.95	 7.36±0.64	 2.04±0.47	 2.07±0.05	 0.74±0.01	 45 progenitors	(3.74±0.35	to	5.62±0.34	hours)	(Figure	7D	&	Table	1).	Despite	the	increased	G1	length	in	trunk	cells,	their	total	cell	cycle	length	was	similar	to	tip	cells,	due	to	a	reduction	in	the	length	of	the	S-phase	(Figure	7D	&	Table	1).	Most	pancreatic	epithelial	cells	are	trunk	cells	(~80%)	and	this	proportion	remains	stable	from	E11.5	to	E13.5	(Figure	7E).	Taken	together,	these	findings	demonstrate	that	G1	lengthening	occurs	within	both	tip	and	trunk	progenitor	pools	during	pancreas	development.			Figure	7:	Tip	progenitor	cells	have	shorter	G1	length	than	trunk	progenitors	from	E11.5	to	E13.5	(A-C)	Cumulative	EdU	labelling	of	Pdx1+Cpa1+	tip	(blue)	and	Pdx1+Cpa1-	trunk	progenitors	(red)	at	(A)	E11.5,	(B)	E12.5,	and	(C)	E13.5.	Arrows	indicate	time	to	reach	GF.	(D)	Calculated	cell	cycle	parameters	for	tip	and	trunk	progenitors	at	E11.5,	E12.5	and	E13.5:	G1	(red	or	blue),	S	(grey),	and	G2/M	(white).	(E)	The	fraction	of	Pdx1+	cells	that	are	tip	(blue)	and	trunk	(red)	progenitors	at	E11.5,	E12.5	and	E13.5.	(F)	The	Growth	Fraction	(number	of	EdU+	cells	following	8	hours	of	labelling	for	E11.5	or	11	hours	of	EdU	labelling	for	E12.5	and	E13.5)	of	tip	and	trunk	progenitors	at	E11.5,	E12.5	and	E13.5.	Data	are	presented	as	mean	±	SEM.	n	=	4-13	embryos	from	2	dams	at	each	time	point.	 46 3.2.3 The	pancreatic	epithelium	has	a	population	of	cells	that	do	not	undergo	S-phase	during	early	development		 Along	with	differences	in	the	lengths	of	both	the	G1-	and	S-phases	between	tip	and	trunk	cells,	there	was	also	a	significant	difference	in	the	proportion	of	EdU	labelled	cells	or	the	growth	fraction	between	the	two	populations	(Figure	7	and	Table	4).	While	the	majority	(>90%)	of	tip	cells	transitioned	S-phase	and	incorporated	EdU	at	least	once	within	the	11	hour	EdU	labelling	period,	only	~70%	of	trunk	cells	did	(Figure	7).	To	ensure	that	this	did	not	result	from	labelling	for	too	short	of	a	time,	a	pregnant	dam	was	injected	with	EdU	every	1.5	hours	for	a	total	of	17	hours	before	sacrifice	and	embryo	collection.	Even	after	17	hours,	there	is	still	a	population	of	Pdx1+	EdU-negative	cells	at	E12.5	(Figure	8A).	Quantification	of	the	proportion	of	EdU+	tip	and	trunk	cells	at	E12.5	following	11	or	17	hours	of	EdU	labelling	revealed	no	differences	(Figure	8B&C),	suggesting	that	this	population	of	EdU-negative	cells	do	not	enter	S-phase	within	a	17-hour	period	at	E12.5.		 47 		Figure	8:	Population	of	EdU-negative	epithelial	cells	remain	after	17	hours	of	EdU	labelling	(A)	Immunofluorescent	images	of	Pdx1	(red)	and	EdU	(green)	following	17	hours	of	EdU	labelling	at	E12.5.	Pancreatic	epithelium	is	outlined	by	dotted	white	line.	(B-C)	The	proportion	of	Pdx1+Cpa1+	tip	cells	(B;	blue)	and	Pdx1+Cpa1-	trunk	cells	(C;	red)	labelled	with	EdU	following	either	11	or	17	hours	of	exposure.	Data	are	presented	as	mean	±	SEM.	n	=	7-13.	 To	identify	the	significance	of	the	pancreatic	epithelial	cells	that	are	not	undergoing	proliferation,	I	profiled	the	expression	of	several	known	markers	in	the	EdU-negative	cells.	First,	the	EdU-negative	cells	appear	in	clusters	within	the	trunk	of	the	epithelium	(Figure	9A),	suggesting	that	they	are	likely	to	be	cells	fated	to	the	ductal	or	endocrine	lineage.	As	some	endocrine	cells	express	Pdx1,	the	Neurog3-Cre;	Rosa26mT/mG	mice	were	used	to	mark	endocrine	cells	with	GFP.	After	labelling	for	11	hours	with	EdU	and	immunostaining	for	GFP,	the	majority	of	cells	in	the	endocrine	lineage	(GFP+	cells)	were	EdU-negative	(Figure	 48 9B),	consistent	with	reports	that	cell	cycle	exit	occurs	following	Neurog3	activation49,51,110.	Further,	most	Pdx1+	EdU-negative	cells	were	GFP-negative,	suggesting	that	this	EdU-negative	population	of	Pdx1+	cells	are	not	endocrine	cells	(Figure	9B).	While	most	duct	(CK19+)	cells	form	after	E14.5155,	the	EdU-negative	cells	could	be	differentiated	duct	cells.	To	address	this,	the	EdU-negative	cells	were	co-stained	for	mucin,	another	marker	of	the	ductal	cell	lineage.	As	mucin	staining	did	not	exclusively	mark	EdU-negative	cells	it	is	unlikely	that	these	EdU-negative	cells	are	duct	cells	(Figure	9C).	However,	it	is	possible	that	they	are	a	ductal	progenitor	cell,	a	population	that	there	are	no	specific	markers	for	at	this	point	in	development.	As	the	EdU-negative	cells	did	not	appear	to	be	a	terminally	differentiated	endocrine	or	ductal	cell,	I	next	investigated	progenitor	cell	markers.	The	Pdx1+	EdU-negative	cells	were	not	Neurog3+	endocrine	progenitor	cells	(Figure	9D),	consistent	with	the	fact	that	most	Neurog3+	cells	do	not	express	Pdx1156.	However,	Pdx1+	EdU-negative	cells	expressed	the	pancreatic	progenitor	markers	Nkx6-1	(Figure	9E)	and	Sox9	(Figure	9F).	In	sum,	these	observations	suggest	that	these	Sox9+	Pdx1+	EdU-negative	cells	remain	trunk	epithelial	cells	that	have	undergone	cell	cycle	arrest	or	are	slowly	cycling	but	have	yet	to	delaminate	and	differentiate.	 49 		Figure	9:	EdU-negative	pancreatic	epithelial	cells	express	pancreatic	progenitor	markers	Embryos	were	labelled	with	EdU	for	11	hours	at	12.5	and	clusters	of	EdU-	cells	are	outlined	by	the	dotted	line.	(A)	Many	Pdx1+	(red)	and	Cpa1-negative	(blue)	cells	remain	EdU-negative	(green)	after	11	hour	of	labelling.	(B)	Most	cells	of	the	endocrine	lineage	cells	EdU-negative	as	determined	by	GFP+	cells	(cyan)	in	Neurog3-Cre;	Rosa26mTmG	embryos.	(C)	Mucin	staining	(cyan)	labels	both	EdU+	and	EdU-negative	cells	within	the	pancreatic	epithelium.	(D)	Neurog3+	(arrowhead;	cyan)	cells	are	not	labeled	with	EdU.	Pdx1+	EdU-negative	cells	also	express	a	variety	of	progenitor	cell	markers	including	Nkx6-1	(E;	red)	and	Sox9	(F;	red).	Scale	bar	=	50	μm.	 3.2.4 Neurog3+	cells	mainly	arise	from	cycling	progenitor	population	To	address	whether	these	cells	are	slowly	cycling	or	cell	cycle	arrested	in	preparation	for	differentiation,	a	pulse-chase	labelling	experiment	was	performed	where	pregnant	dams	were	injected	with	EdU	for	11	hours	at	E12.5	and	embryos	were	examined	 50 24	hours	later.	Quantification	of	Neurog3+	cells	indicated	that	the	majority	(>80%)	of	Neurog3+	cells	at	E13.5	were	derived	from	actively	cycling	EdU+	cells	(Figure	10A).	In	addition,	the	proportion	of	EdU+	tip	and	trunk	progenitor	cells	does	not	significantly	change	after	24	hours,	suggesting	that	there	is	not	a	large	loss	in	the	progenitor	cell	pool	to	differentiation	over	this	24-hour	period	(Figure	10B&C).	  Figure	10:	Most	Neurog3+	cells	arise	from	EdU+	cells	(A)	Pregnant	dams	were	injected	with	EdU	for	11	hours	at	E12.5	before	embryos	were	collected	at	E13.5.	Quantification	of	the	fraction	of	Neurog3+	cells	that	are	EdU+	or	EdU-negative	following	pulse-chase	labelling	experiment.	(B-C)	The	fraction	of	EdU+	tip	(blue)	and	trunk	(red)	cells	does	not	markedly	change	following	11	hours	of	EdU	labelling	at	E12.5	between	embryos	collected	immediately	following	labelling	(E12.5)	or	after	24-hour	chase	period	(E13.5).		 	 51 3.3 Discussion		 In	this	chapter,	I	demonstrate	that	there	is	a	correlation	between	differentiation	and	G1	lengthening	in	mouse	pancreas	development.	This	is	supported	by	the	increase	in	G1-phase	length	from	4.6	to	7.2	hours	between	E11.5	and	E13.5.	These	findings	are	consistent	with	previous	estimations	of	total	cell	cycle	length	performed	in	cultured	embryonic	pancreas54,157.	Together,	this	work	extends	the	“cell	cycle	length	hypothesis”	from	neural	development116	and	adds	the	pancreas	to	the	growing	list	of	progenitor	cells	that	undergo	G1	lengthening.	This	hypothesis	suggests	that	there	is	a	mechanistic	link	between	cell	cycle	and	differentiation:	it	provides	the	time	required	for	a	progenitor	cell	to	accumulate	enough	of	a	cell	fate	determinant	(i.e.	Neurog3)	to	drive	differentiation.		 An	interesting	question	that	arises	from	this	work,	and	the	cell	cycle	length	hypothesis,	is	how	does	G1	lengthening	occur?	While	further	experiments	will	be	needed	to	directly	address	this	question,	it	likely	results	from	cell	autonomous	and/or	non-cell	autonomous	mechanisms157.	The	differentiation	of	pluripotent	stem	cells150	and	neural	progenitor	cells158	is	associated	with	increased	expression	of	CKIs,	suggesting	the	G1	lengthening	may	be	driven	by	CKI-dependent	inhibition	of	the	G1/S-phase	transition.	In	the	developing	pancreas,	expression	of	CKIs	p21	(Cdkn1a),	p27	(Cdkn1b),	and	p57	(Cdkn1c)	increases	during	early	pancreas	development	and	are	upregulated	in	differentiated	endocrine	cells159.	Interestingly,	during	oligodendrocyte	differentiation	there	is	accumulation	of	p27	following	each	cellular	division158.	Whether	a	similar	process	controls	cell	cycle	lengthening	in	pancreas	development	is	unknown.	Alternatively,	there	may	be	non-cell	autonomous	factors	that	control	G1	lengthening	during	pancreas	development.	For	example,	the	pancreatic	mesenchyme	secretes	Fgf10,	which	signals	the	Pdx1+	pancreatic	 52 epithelium	to	proliferate160.	Fgf10	is	expressed	by	the	mesenchyme	beginning	at	E10.5	and	is	downregulated	at	E12.5,	around	the	time	when	cell	cycle	lengthening	occurs160.	Evidence	supporting	a	potential	role	for	a	mesenchymal	secreted	factor	in	regulating	cell	cycle	length	is	that	the	tip	progenitors,	which	are	in	apposition	with	the	surrounding	mesenchyme,	have	a	shorter	G1	length	than	the	trunk	cells.	To	understand	the	mechanism(s)	that	endogenously	control	cell	cycle	length,	it	will	be	important	to	investigate	whether	there	are	changes	in	cell	cycle	length	in	models	that	have	impaired	endocrine	cell	differentiation.		 While	trunk	progenitor	cells	spend	more	time	in	the	G1-phase	of	the	cell	cycle	than	tip	progenitor	cells,	the	total	cell	cycle	length	between	tip	and	trunk	cells	is	similar.	This	results	from	the	tip	progenitors	dedicating	more	time	the	S-phase	than	trunk	cells	do.	This	difference	in	S-phase	length	was	also	noted	in	mammalian	cerebral	cortex	development	where	progenitor	cells	spend	more	time	in	S-phase	than	the	cells	committed	to	differentiation154.	While	proliferating	progenitor	cells	take	more	time	for	DNA	replication	than	committed	progenitor	cells,	it	does	not	account	for	the	3.3-fold	increase	in	S-phase	length.	To	understand	why	proliferating	progenitors	spend	more	time	in	S-phase,	the	authors	performed	RNA-sequencing	of	S-phase	cells	and	found	changes	in	several	genes	that	are	involved	in	DNA	replication	and	repair,	chromatin	remodeling	and	cell	cycle	regulation154.	Together,	these	results	suggest	that	progenitors	dedicate	more	time	to	the	S-phase	to	ensure	quality	control	during	DNA	replication.		 Recently,	an	elegant	study	using	live-cell	imaging	of	ex	vivo	cultures	of	pancreatic	explants	suggests	that	the	type	of	cell	division	a	progenitor	cell	undergoes	depends	on	when	during	the	G1-phase	the	inductive	event	occurs157.	The	authors	described	that	there	are	three	types	of	pancreatic	progenitor	cell	division:	symmetric	cell	division	producing	 53 two	progenitor	daughter	cells;	asymmetric	cell	division	producing	one	progenitor	and	one	differentiated	daughter	cell;	and	symmetric	cell	division	producing	two	differentiated	daughters	cells.	The	probability	that	a	progenitor	cell	is	primed	for	endocrine	differentiation	is	~20%	in	vivo.	Depending	on	when	during	the	cell	cycle	the	inductive	event	occurs	will	inform	the	type	of	cell	division	the	progenitor	cell	undergoes.	For	example,	if	the	inductive	event	occurs	in	late	G1,	the	mother	cell	is	already	committed	to	another	round	of	cell	division	and	will	give	rise	to	two	endocrine	daughter	cells.	However,	if	the	inductive	event	occurs	early	in	G1,	the	progenitor	will	exit	the	cell	cycle	and	differentiate,	giving	the	appearance	of	an	asymmetric	cell	division	for	the	mother	cell.	As	G1	lengthening	endogenously	occurs	during	pancreas	development,	this	may	increase	the	likelihood	of	the	inductive	event	happening	later	in	G1,	resulting	in	a	symmetric	cell	division	producing	two	endocrine	cells.	It	would	be	interesting	to	investigate	whether	there	are	more	symmetric	cell	divisions	during	the	later	stages	of	pancreas	development	as	a	result	of	the	increased	G1	length.		 Another	interesting	observation	from	the	cumulative	EdU	labelling	is	that	the	proportion	of	EdU	labelled	cells	or	the	growth	fraction	between	the	tip	and	trunk	progenitors	is	significantly	different.	While	the	majority	(>90%)	of	tip	cells	transitioned	S-phase	and	incorporated	EdU	at	least	once	within	the	11	hour	EdU	labeling	period,	only	~70%	of	trunk	cells	did.	One	potential	reason	for	the	difference	in	EdU-labelled	cells	between	the	two	progenitor	populations	could	be	due	to	a	technical	caveat.	It	is	possible	that	the	trunk	cells	have	reduced	access	to	EdU	due	to	differences	in	tissue	perfusion	and/or	increased	cellular	efflux	of	EdU	by	trunk	cells,	resulting	in	fewer	EdU-labelled	cells.	To	address	this	caveat,	future	experiments	should	be	aimed	at	confirming	the	EdU	labelling	 54 experiments	with	other	measures	of	the	cell	cycle,	such	as	pHH3	(M-phase	marker)	or	Ki67	(proliferation	marker).	Following	confirmation	of	the	presence	of	the	EdU-negative	cells,	future	studies	using	single	cell	RNA-sequencing	of	pancreatic	progenitors	at	E12.5	may	help	identify	molecular	markers	that	will	allow	for	further	investigation	of	their	role	in	pancreas	development.		 In	this	chapter,	the	pancreatic	progenitor	is	added	to	the	growing	list	of	cell	types	where	there	is	a	documented	correlation	between	G1	lengthening	and	differentiation.	These	findings	extend	the	“cell	cycle	length	hypothesis”	to	endocrine	cell	differentiation	and	suggests	that	the	cell	cycle	length	plays	an	important	role	in	pancreas	development.		 	 55 Chapter	4: Altering	cell	cycle	length	in	vivo	changes	endocrine	cell	differentiation	4.1 Background	The	correlation	between	G1	lengthening	and	differentiation	during	mouse	pancreas	development	suggests	that	the	cell	cycle	itself	may	regulate	differentiation.	To	address	this	hypothesis,	two	mouse	models	that	allow	manipulation	of	cell	cycle	length	in	vivo	were	used:	KrasLSL-G12D	and	tetO-Cdkn1b.	Kras	is	a	member	of	the	RAS	family	of	GTP-binding	proteins.	When	bound	to	GTP,	Kras	interacts	with	signalling	molecules	to	regulate	several	biological	processes	including	proliferation	and	differentiation161.	When	Kras	activity	is	no	longer	required,	GTP	is	hydrolyzed	to	GDP.	Point	mutations	that	reduce	this	hydrolysis	and	cause	Kras	to	remain	active	are	oncogenic	and	are	found	in	almost	every	pancreatic	tumour162,	including	the	most	common	mutation	(KrasG12D).	In	an	effort	to	elucidate	the	role	of	Kras	signalling	in	pancreatic	endocrine	tumours,	Chamberlain	et	al163	generated	Pdx1-Cre;	KrasLSL-G12D	mice	to	study	both	loss-of-function	(KrasLSL)	and	gain-of-function	(KrasG12D)	mutations.	In	these	mice,	they	noted	a	doubling	in	Neurog3+	area	in	loss-of-function	embryos	at	E13.5	and	a	reduction	in	Neurog3+	area	in	mice	expressing	the	constitutively	active	(G12D)	form	of	Kras.	Since	Kras	positively	regulates	cell	cycling	I	hypothesized	that	the	increased	Neurog3+	area	of	the	KrasLSL	mice	might	be	due	to	a	change	in	cell	cycle	length.	The	other	model	that	was	used	to	change	cell	cycle	length	during	embryonic	development	is	the	ectopic	expression	of	Cdkn1b	in	Sox9+	pancreatic	progenitors	(Sox9-rtTA;	tetO-Cdkn1b).	Cdkn1b	was	first	discovered	in	a	yeast	interaction	screen	to	identify	proteins	that	interact	with	CyclinD1/Cdk4164,	a	complex	that	is	required	for	the	transition	from	G1/S-phase	during	the	cell	cycle.	In	the	pancreas,	Cdkn1b	accumulates	in	terminally	 56 differentiated	endocrine	cells	and	is	thought	to	prevent	reentry	into	the	cell	cycle159.	In	addition,	Cdkn1b	plays	an	important	role	in	controlling	b-cell	proliferation.	For	instance,	Cdkn1b-/-	mice	have	increased	proliferation	and	b-cell	mass159	and	specific	reduction	of	Cdkn1b	in	Ins+	b-cells	increases	proliferation,	resulting	in	improved	glucose	tolerance	165,166.	Conversely,	ectopic	expression	of	Cdkn1b	in	b-cells	decreases	proliferation,	leading	to	a	reduced	b-cell	mass	and	glucose	intolerance165,166.	In	this	chapter,	Cdkn1b	is	ectopically	expressed	in	pancreatic	progenitor	cells	in	order	to	prevent	the	G1/S-phase	transition	and	cause	G1	lengthening.	One	of	the	mechanistic	ways	that	the	cell	cycle	may	regulate	endocrine	cell	differentiation	is	through	regulating	Neurog3	levels.	Supporting	this	hypothesis	is	that	the	changes	in	cell	cycle	length	occurs	in	both	tip	and	trunk	progenitors	before	the	upregulation	of	Neurog3	and	resulting	wave	of	endocrine	cell	differentiation.	Previous	work	identified	that	Neurog3	mRNA	is	more	abundant	than	protein	in	embryonic	pancreas167,168.	This,	in	combination	with	the	presence	of	multiple	forms	of	Neurog3	of	different	electrophoretic	mobilities169,	suggests	that	Neurog3	might	be	post-translationally	modified	by	cell	cycle	proteins.	Work	in	Xenopus	found	that	Neurog2,	another	member	of	the	Neurogenin	family	of	basic	helix-loop-helix	transcription	factors,	is	phosphorylated	by	the	cyclin	dependent	kinases	1/2	(Cdk1/2)122.	Hyperphosphorylation	of	Neurog2	in	rapid	cycling	cells	results	in	its	degradation,	maintaining	the	progenitor	state122.	In	this	chapter,	the	effect	of	changing	cell	cycle	length	in	pancreatic	progenitors	on	endocrine	cell	differentiation	was	investigated.	In	addition,	the	role	of	Cdk-dependent	regulation	of	Neurog3	was	investigated	as	the	mechanistic	link	between	G1	lengthening	and	endocrine	cell	differentiation.	 57 	4.2 Results	4.2.1 Alterations	in	KRAS	signalling	changes	cell	cycle	length	To	confirm	that	alterations	in	KRAS	signalling	affects	the	formation	of	Neurog3+	endocrine	progenitor	cells,	Pdx1-Cre+;	KrasLSL-G12D	(KrasG12D)	and	Pdx1-Cre-;	KrasLSL-G12D	(KrasLSL)	embryos	(Tuveson,	et	al.,	2004)	were	used	to	investigate	both	increased	(KrasG12D)	and	decreased	(KrasLSL)	Kras	signalling	(Figure	11A).	Quantification	of	the	number	of	Neurog3+	cells,	normalized	to	Sox9+	progenitors	at	E13.5,	confirmed	that	increased	Kras	signalling	reduces	the	number	of	endocrine	progenitors	1.7-fold	(KrasG12D;	Figure	11B).	Conversely,	decreasing	Kras	signalling	expands	the	Neurog3+	endocrine	progenitor	pool	1.2-fold	(KrasLSL	;	Figure	11B).	To	determine	if	Kras	signalling	changes	cell	cycle	length,	embryos	were	labelled	at	E12.5	for	3.5	hours	with	EdU	(Figure	11C).	This	length	of	EdU	labelling	was	chosen	as	it	falls	within	the	steep	portion	of	the	labelling	curve	at	E12.5	and	therefore	allows	facile	and	sensitive	determination	of	changes	in	cell	cycle	parameters	(Figure	6).	The	proportion	of	Pdx1+	EdU+	pancreatic	progenitors	in	littermate	control	mice	at	E12.5	was	0.45±0.03	after	3.5	hours	of	labelling,	while	KrasLSL	embryos	had	a	significant	reduction	in	the	number	of	EdU+	cells	(0.32±0.04)	(Figure	11D).	Conversely,	activations	in	Kras	signalling	increased	the	proportion	of	EdU+	cells	to	0.49±0.03	in	KrasG12D	embryos	at	E12.5	(Figure	11D).	 58   Figure	11:	KrasLSL-G12D	mutations	lead	to	altered	cell	cycle	length	during	mouse	pancreatic	development	(A)	Schematic	outlining	the	alleles	present	in	the	three	different	genotypes.
(B)	Quantification	of	the	number	of	Neurog3+	cells	relative	to	Sox9+	trunk	cells	in	Control,	KrasLSL	and	KrasG12D	embryos	at	E13.5.	n	=	3.	**p	<	0.01,	***p	<	0.001	by	one-way	ANOVA	and	Tukey	post	test.	(C)	Pdx1	(red),	Cpa1	(cyan)	and	EdU	(green)	immunostaining	in	Control,	KrasLSL	and	KrasG12D	embryos	at	E12.5	following	3.5	hours	of	EdU	exposure.	Scale	bar	=	50	μm.	(D-F)	KrasLSL	embryos	had	reduced	EdU-labeled	Pdx1+	pancreatic	progenitor	cells	and	(E)	reduced	EdU+	tip	cells	(Pdx1+Cpa1+)	but	(F)	no	significant	changes	in	EdU+	trunk	cells	(Pdx1+	Cpa1-)	at	E12.5.	n	=	3	*p	<	0.05,	**p	<	0.01	by	one-way	ANOVA	and	Tukey	post	test.	(G)	The	total	number	of	E12.5	pancreatic	cells	did	not	change	with	Kras	expression	level.	n	=	3.	(H)	The	relative	proportion	of	tip	and	trunk	cells	in	Control,	KrasLSL	and	KrasG12D	embryos	at	E12.5	remained	constant.	n	=	3.
Data	are	presented	as	mean	±	SEM.	 As	most	Neurog3+	cells	are	derived	from	the	trunk	Pdx1+	Cpa1-negative	trunk	progenitor	cells,	I	next	investigated	whether	changes	in	Kras	signalling	affects	the	cell	cycle	length	of	tip	and	trunk	progenitor	cells.	EdU+	tip	cells	in	the	KrasLSL	decreased	from	0.75±0.07	in	control	embryos	to	0.45±0.03	while	increasing	to	0.81±0.10	in	KrasG12D	 59 (Figure	11E).	In	addition,	the	proportion	of	EdU+	trunk	cells	decreased	from	0.39±0.02	to	0.28±0.04	in	the	KrasLSL	embryos	while	KrasG12D	were	increased	to	0.44±0.02	compared	to	control	littermates	(Figure	11F).	Altering	Kras	signalling	had	no	effect	on	the	relative	proportions	of	tip	and	trunk	progenitor	cells	at	E12.5	(Figure	11G),	nor	did	it	change	the	number	of	pancreatic	nuclei	(Figure	11H).	As	Kras	signalling	regulates	cell	proliferation,	activating	mutations	in	Kras	may	increase	the	proportion	of	proliferating	cells,	or	the	growth	fraction.	To	ensure	that	driving	Kras	signalling	did	not	alter	the	growth	fraction,	the	number	of	EdU+	Pdx1+,	tip,	and	trunk	progenitor	cells	was	quantified	following	11	hours	of	EdU	labelling	at	E13.5.	Compared	to	control	embryos,	there	was	no	change	in	the	growth	fraction	for	Pdx1	(Figure	12A),	tip	(Figure	12B)	or	trunk	(Figure	12C)	progenitor	cells.	Overall,	these	data	support	the	hypothesis	that	manipulation	of	Kras	signalling	during	early	pancreas	development	alters	endocrine	cell	differentiation	by	changing	cell	cycle	length,	suggesting	that	cell	cycle	lengthening	is	important	for	pancreatic	progenitor	cell	differentiation.			 60 		Figure	12:	There	is	no	change	in	the	growth	fraction	in	KrasG12D	embryos	The	number	of	EdU+	Pdx1	(A),	tip	(B)	and	trunk	(C)	progenitors	in	control	and	KrasG12D	embryos	at	E13.5	following	11	hours	of	EdU	labelling.	n=3.	Data	are	presented	as	mean	±	SEM.		4.2.2 Acute	ectopic	expression	of	Cdkn1b	at	E12.5	increases	the	formation	of	Neurog3+	endocrine	progenitors		 To	more	directly	test	the	consequence	of	altering	G1	length,	Cdkn1b—a	cyclin-dependent	kinase	inhibitor	not	normally	expressed	in	Sox9-expressing	pancreatic	progenitor	cells	that	blocks	the	G1-S	transition110—	was	ectopically	expressed	in	a	doxycycline-inducible	manner	(Figure	13A).	Control	embryos	were	either	single	transgenic	(ST)	Sox9-rtTA	or	tetO-Cdkn1b	littermates	of	Cdkn1b-expressing	Sox9-rtTA;	tetO-Cdkn1b	double	transgenic	(DT)	embryos.	Pregnant	dams	were	injected	with	doxycycline	on	the	morning	of	P12.5	and	embryos	were	collected	for	analysis	9.5	hrs	later.	Using	immunofluorescence	staining,	the	upregulation	of	Cdkn1b	in	pancreatic	progenitors	(Pdx1+)	cells	was	confirmed	at	E12.5	(Figure	13B).	This	induction	of	Cdkn1b	was	sufficient	to	increase	the	number	of	Neurog3+	cells	2.7-fold	compared	to	ST	control	embryos	without	altering	total	cell	number	(Figure	13C&D).	 61 		Figure	13:	Ectopic	expression	of	Cdkn1b	in	Sox9+	progenitors	increases	Neurog3+	cells	at	E12.5	(A)	Schematic	of	transgenic	mice	used	to	ectopically	express	Cdkn1b	in	Sox9+	pancreatic	progenitors.	(B)	One	injection	of	doxycycline	(Dox)	at	E12.3	increased	Cdkn1b	(green)	expression	in	the	pancreatic	epithelium	(Pdx1+;	red)	of	double	transgenic	(DT)	but	not	in	single	transgenic	(ST)	mice	at	E12.7.
(C)	9.5	hr	of	Cdkn1b	induction	at	E12.5	significantly	increased	the	number	of	Neurog3+	cells	in	DT	embryos.	n	=	3.	***p	<	0.0001	by	unpaired	t	test.	(D)	Ectopic	expression	of	Cdkn1b	(9.5	hr)	did	not	alter	the	total	number	of	pancreatic	cells.	n	=	6.	Data	are	presented	as	mean	±	SEM.	Scale	bars	=	50	µm.	 4.2.3 Long	term	upregulation	of	Cdkn1b	from	E10.5	to	E12.5	increases	the	number	of	Gcg+	endocrine	cells	To	determine	whether	ectopic	Cdkn1b	expression	slowed	S-phase	transition,	doxycycline	was	given	from	P10.5-P12.5	and	EdU-labelling	was	carried	out	for	3.5	hrs	before	embryo	collection	at	E12.5	(Figure	14A).	The	proportion	of	EdU+	pancreatic	progenitor	cells	significantly	decreased	from	0.38±0.03	in	control	embryos	to	0.29±0.02	in	double	transgenic	embryos	(Figure	14B).	To	ensure	that	ectopic	Cdkn1b	expression	did	not	cause	cell	cycle	exit,	the	growth	fraction	was	quantified	using	11	hours	of	EdU	labelling.	No	differences	were	measured	between	control	ST	and	Cdkn1b-expressing	DT	embryos	(Figure	14C).	Notably,	48	hrs	of	Cdkn1b	expression	resulted	in	a	2.5-fold	increase	in	Chromogranin	A	(Chga)+	endocrine	cells	at	E12.5	(Figure	14D)	caused	by	an	increase	in	Glucagon	(Gcg;	α)+	cells	(Figure	14E)	without	altering	overall	cell	numbers	(Figure	14F).	 62 These	results	suggest	that	the	increase	in	Neurog3+	endocrine	progenitors	caused	endocrine	cell	formation	that	was	predominantly	of	the	α-cell	lineage.	  Figure	14:	Ectopic	expression	of	Cdkn1b	from	E10.5-E12.5	increases	Chga	and	Gcg	endocrine	cell	formation	(A)	At	E12.5,	Cdkn1b-expressing	(red)	pancreatic	epithelial	cells	(Pdx1+;	blue)	from	Dox-treated	(E10.5–12.5)	DT	embryos	incorporated	less	EdU	(green)	at	3.5	hr.	(B)	Significantly	fewer	Pdx1+EdU+		cells	were	counted	following	3.5	hr	of	labelling	of	DT	embryos.	n	=	7.	*p	<	0.05	by	unpaired	t	test.	(C)	Ectopic	Cdkn1b	expression	did	not	alter	the	fraction	of	EdU+	progenitor	cells	(the	GF)	after	11	hr	of	labelling,	indicating	that	cells	remained	in	the	cell	cycle.	n	=	3.	(D)	Chga+		cells	were	significantly	increased	at	E12.5	DT	embryos	following	48	hr	of	Dox.	n	=	5.	**p	<	0.01	by	Mann-Whitney	U	test.	(E)	Gcg+	cells	were	significantly	increased	at	E12.5	in	DT	embryos	following	48	hr	of	Dox.	n	=	6.	**p	<	0.01	by	unpaired	t	test.	(F)	Ectopic	expression	of	Cdkn1b	(E10.5–12.5)	did	not	reduce	total	pancreatic	cells	at	E12.5.	n	=	7.	Data	are	presented	as	mean	±	SEM.	Scale	bars	=	50	µm.	 63 4.2.4 Long	term	upregulation	of	Cdkn1b	from	E12.5	to	E14.5	increases	the	number	of	Gcg+	and	Ins+	endocrine	cells	As	competency	of	Neurog3+	cells	is	predominantly	down	the	α-cell	lineage	before	E12.552,	I	next	asked	if	Cdkn1b	expression	promotes	the	formation	of	other	endocrine	cell	types	later	in	pancreas	development	by	administering	doxycycline	from	P12.5-P14.5	(Figure	15A).	This	doxycycline	dosing	regimen	resulted	in	a	significant	decrease	in	the	number	of	EdU+	cells	from	0.36±0.01	to	0.22	±0.02	at	E14.5	following	3.5	hrs	of	EdU	labelling	(Figure	15B).	The	number	of	Chga+	endocrine	cells	increased	3.3-fold	at	E14.5	(Figure	15C).	In	addition,	the	numbers	of	Gcg+	(2.8-fold;	Figure	15D)	and	Ins+	(1.4-fold;	Figure	15E)	cells	increased,	without	affecting	pancreatic	size	(Figure	15F).	Taken	together,	these	data	suggest	that	ectopic	expression	of	Cdkn1b	increases	endocrine	cell	differentiation	in	a	pattern	that	is	consistent	with	the	lineage-specification	of	Neurog3+	endocrine	progenitor	cells.		 64 		Figure	15:	Ectopic	expression	of	Cdkn1b	from	E12.5-E14.5	increases	endocrine	cell	formation	(A)	Dox	treatment	(E12.5–14.5)	of	DT	embryos	increased	Cdkn1b	(red)	expression	in	subset	of	Pdx1+	(blue)	trunk	cells.	(B)	Dox-treated	(E12.5–14.5)	DT	embryos	showed	significantly	reduced	EdU+	labeling	after	3.5	hr	at	E14.5.	n	=	4.	***p	<	0.0001	by	unpaired	t	test.	(C)	More	Chga+	cells	were	present	in	Dox-treated	(E12.5–14.5)	DT	embryos	at	E14.5.	n	=	5.	**p	<	0.001	by	Mann-Whitney	U	test.	(D)	More	Gcg+	cells	were	present	in	Dox-treated	(E12.5–14.5)	DT	embryos	at	E14.5.	n	=	5.	**p	<	0.01	by	unpaired	t	test.	(E)	Ins+	cells	were	not	significantly	changed	in	Dox-treated	(E12.5–14.5)	DT	embryos.	n	=	9.	p	=	0.09	by	unpaired	t	test.	(F)	The	number	of	pancreatic	cells	per	slide	was	unchanged,	suggesting	that	growth	was	not	affected	by	48	hr	of	Cdkn1b	expression	at	E14.5.	n	=	5.	Data	are	presented	as	mean	±	SEM.	Scale	bars	=	50	µm.	4.2.5 Inhibition	of	cyclin	dependent	kinases	(CDKs)	increases	Neurog3+	cells	during	mouse	pancreas	development		 Ectopic	expression	of	Cdkn1b	increased	the	expression	of	Neurog3	and	subsequent	endocrine	cell	differentiation.	It	has	previously	been	demonstrated	that	Neurog3	message	expression	is	more	prevalent	than	its	protein	abundance	in	rapidly	cycling	pancreatic	 65 progenitors167,168.	At	its	peak	of	expression	in	the	mouse	embryonic	pancreas,	Neurog3	displays	multiple	distinct	electrophoretic	mobilities,	consistent	with	post-translation	modification	(Figure	16A).		 As	Cdkn1b	is	an	inhibitor	of	G1-S	Cdks	(i.e.	Cdk2,	Cdk4,	Cdk6),	I	hypothesized	that	Cdks	might	phosphorylate	Neurog3,	leading	to	its	degradation155.	To	test	this,	the	effect	of	Cdk	inhibition	on	Neurog3	expression	was	investigated.	Embryonic	pancreata	were	isolated	at	E11.5	and	treated	ex	vivo	for	24	hrs	with	Cdk4/6	inhibitor	PD-03329911	and	Cdk2	inhibitors	ii	and	iii	(CDKi).	This	resulted	in	a	3-fold	increase	in	the	number	of	Neurog3+	cells	without	altering	the	total	number	of	Sox9+	progenitor	cells	(Figure	16B&C),	as	assessed	by	immunofluorescence.	This	suggests	that	the	G1	lengthening	may	cause	endocrine	cell	differentiation	through	reducing	CDK-dependent	phosphorylation	of	Neurog3.		Figure	16:	Cdk	inhibition	increases	the	number	of	Neurog3+	cells	in	mouse	embryonic	explants	(A)	Western	blot	from	E15.5	mouse	pancreas	reveals	multiple	Neurog3	bands.	(B)	Mouse	pancreatic	explants	were	cultured	for	24	hours	starting	at	E11.5	with	Cdk2	and	4/6	inhibitors	(CDKi)	and	the	number	of	Neurog3+	cells/Sox9+	cells	and	(C)	the	total	Sox9+	cells	were	quantified	by	immunofluorescence	staining.	n	=	5.	**p	<	0.01	by	unpaired	t	test.	 	 66 4.3 Discussion		 In	this	chapter,	two	different	mouse	models	were	used	to	manipulate	cell	cycle	length	in	vivo.	First	the	KrasLSL-G12D	mouse,	most	commonly	used	to	model	pancreatic	ductal	adenocarcinoma,	was	used	to	both	lengthen	and	shorten	the	cell	cycle.	There	are	three	mammalian	Ras	genes,	Hras,	Kras,	Nras,	that	code	for	189	amino	acid	proteins.	These	proteins	have	high	sequence	conservation	in	the	N-terminal	effector	binding	domain,	which	is	important	for	interacting	with	all	downstream	targets	of	Ras.	The	Ras	genes	show	differential	gene	expression	in	different	organs,	suggesting	cell	type	specific	roles	for	each	Ras	protein170.	While	reductions	in	Nras171	or	Hras172	signalling	does	not	affect	embryonic	development,	Kras-/-	mice	are	embryonic	lethal	due	to	heart	defects173.	This	embryonic	defect	can	be	rescued	by	transgenic	expression	of	Hras	in	Kras-/-	embryos,	suggesting	that	there	is	functional	redundancy	between	the	three	Ras	genes174.		 Ras	family	proteins	are	expressed	in	all	animal	cell	lineages	and	play	important	roles	in	cell	growth,	differentiation,	and	survival.	This	is	thought	to	be	mediated	by	their	involvement	in	the	signal	transduction	pathway	downstream	of	many	receptor	tyrosine	kinases	(RTK).	RTKs	are	cell-surface	receptors	that	are	bound	to	and	activated	by	many	hormones	including	fibroblast	growth	factor,	epidermal	growth	factor	and	insulin.	Activation	of	RTKs	results	in	the	activation	of	Ras	and	the	downstream	signalling	cascade.	Ras	proteins	can	signal	through	many	pathways,	including	Raf/Mek/Erk	and	PI3K/Pdk1/Akt175.	As	Kras	signalling	activates	many	diverse	and	sometimes	opposing	pathways,	the	downstream	function	of	Kras	is	cell-type	dependent.	For	instance,	activating	mutations	in	Kras	leads	to	tumour	formation	in	pancreatic	ducts	and	acinar	cells176	while	mutating	Kras	in	endocrine	cells	suppresses	proliferation163.	 67 	 In	the	context	of	pancreas	development,	Kras	is	involved	in	controlling	the	rate	of	progenitor	cell	cycling,	and	therefore	endocrine	cell	formation.	Previous	reports	suggest	that	KrasLSL-G12D/+	loss-of-function	(KrasLSL)	embryos	have	increased	number	of	Neurog3+	endocrine	progenitors163.	In	this	chapter,	I	provide	evidence	that	reductions	in	Kras	signaling	increases	cell	cycle	length.	Notably,	the	fraction	of	EdU	labelled	tip	cells	at	E12.5	in	KrasLSL	equals	the	proportion	of	EdU	labelled	trunk	cells	in	the	wildtype	embryos,	suggesting	a	similar	cell	cycle	length.	This	may	explain	the	increases	in	the	number	of	Neurog3+	cells	as	the	tip	cells	have	slowed	their	cell	cycle	to	one	that	is	permissive	for	Neurog3	activation.	As	Kras	signalling	is	upstream	of	many	pathways,	further	studies	will	be	needed	to	elucidate	the	endogenous	role	of	Kras	signalling	during	pancreas	development.	Another	model	used	to	alter	cell	cycle	length	was	ectopic	expression	of	the	cyclin-dependent	kinase	inhibitor	Cdkn1b	in	Sox9+	pancreatic	progenitors.	Many	cell	cycle	kinase	inhibitors	are	found	to	be	upregulated	downstream	of	Neurog3,	including	Cdkn1a110	and	Cdkn1b159,	suggesting	that	the	main	role	of	these	proteins	is	to	regulate	cell	cycle	exit	during	endocrine	cell	differentiation.	Conversely,	at	E10.5	and	E11.5	Cdkn1c	is	the	most	highly	expressed	CKI	and	can	be	found	in	a	subset	of	Pdx1+	epithelial	cells,	an	expression	pattern	that	is	consistent	with	a	potential	role	for	Cdkn1c	in	regulating	progenitor	cell	proliferation109.	However,	very	few	epithelial	cells	express	Cdkn1c	after	E12.5,	suggesting	that	the	role	of	Cdkn1c	in	pancreatic	progenitors	is	temporally	limited109.	Ectopic	expression	of	Cdkn1b	in	Sox9-expressing	progenitor	cells	increases	the	number	of	Neurog3+	cells,	which	appears	to	be	mediated	by	inhibition	of	CDKs.	Whether	endogenous	CKIs	function	to	directly	regulate	Neurog3	in	vivo	will	need	to	be	further	studied.	 68 As	manipulating	Kras	signalling	and	ectopic	expression	of	Cdkn1b	both	increased	endocrine	cell	formation,	these	results	suggest	that	the	increase	cell	cycle	length	drives	pancreatic	endocrine	cell	differentiation.	This	is	consistent	with	other	studies	that	have	lengthened	G1	by	altering	expression	of	cell	cycle	proteins.	Treating	cultures	of	Xenopus177	or	mouse	embryos118	with	olomoucine,	a	synthetic	inhibitor	of	cyclin-dependent	kinases,	lengthens	G1	and	results	in	premature	neurogenesis.	Using	shRNA	against	CyclinD/Cdk4	in	embryonic	mouse	brains	also	lengthens	G1	and	increases	differentiation	of	neuronal	cells178.	Conversely,	overexpression	of	CyclinD/Cdk4	in	mouse178	or	overexpression	of	CyclinA2/Cdk2	in	Xenopus179	prevents	differentiation	and	expands	the	progenitor	pool.	In	this	chapter,	I	provide	evidence	that	inhibition	of	Cdks	increases	expression	of	Neurog3	and	endocrine	cell	formation	during	mouse	development.	Studies	in	Chapter	6	of	this	thesis	outline	the	role	of	Cdks	on	Neurog3	stability.	These	results	suggest	that	cyclin	dependent	kinases	play	important	roles	in	the	balance	between	proliferation	and	differentiation	of	progenitor	cells	across	many	species	and	cell	types,	including	the	pancreas.	 69 Chapter	5: Using	genome	editing	to	generate	reporter	human	embryonic	stem	cell	lines	5.1 Background		 I	next	set	out	to	investigate	the	link	between	the	cell	cycle	and	endocrine	differentiation	in	a	human	model	of	endocrine	cell	formation	using	embryonic	stem	cell	differentiations.	Embryonic	stem	cells	are	pluripotent	cells	located	in	the	inner	cell	mass	of	early	embryos	that	have	the	capacity	for	long-term	self-renewal	and	the	ability	to	form	all	cell	types	of	the	embryo	proper.	Thus,	hESC	differentiations	can	be	used	to	model	normal	pancreatic	development	and	endocrine	cell	formation.	Over	the	past	ten	years,	great	progress	has	been	made	in	generating	differentiation	protocols	that	mimic	the	stages	of	human	endocrine	cell	formation65,67-70.	However,	these	protocols	often	result	in	a	mixed,	heterogeneous	population	of	several	pancreatic	and	endocrine	cell	types.		 The	heterogeneous	nature	of	hESC	differentiations	means	that	reporter	lines	that	are	specific	to	the	cell	type	of	interest	are	needed	to	explore	biological	questions.	The	first	step	in	hESC	differentiation	to	b-like	cells	is	to	reduce	expression	of	pluripotency	genes	such	as	OCT4,	SOX2,	and	NANOG.	Owing	to	their	important	role	in	maintaining	pluripotency,	there	are	several	reporter	lines	for	OCT477,78,141,180,181,	SOX2182,	and	NANOG183.	Following	loss	of	pluripotency,	the	next	step	in	hESC	differentiation	to	b-like	cells	is	the	formation	of	definitive	endoderm	layer.	A	key	transcription	factor	involved	in	definitive	endoderm	formation	in	mice	is	Sox17184.	As	such,	an	hESC	reporter	line	that	marked	SOX17-expressing	cells	with	GFP	was	used	to	show	that	pancreas,	along	with	liver	and	intestinal	epithelium,	are	derived	from	SOX17+	cells185.	The	generation	of	pancreas-specific	reporter	lines	such	as	PDX1-eGFP186,	NKX6-1-GFP187	and	several	dual	reporter	lines188,	are	important	 70 resources	to	understand	the	specification	of	the	pancreas	from	definitive	endoderm	cells.	Finally	as	expression	of	Neurog3	is	critical	to	the	formation	of	both	mouse46	and	human189	endocrine	cells,	there	are	several	Neurog3	reporter	hESC	lines188,190,191.		 In	this	chapter,	I	outline	a	strategy	to	generate	cell	type	specific	reporter	CyT49	hESCs	lines	using	TALEN	and	CRISPR-Cas	genome	editing	technologies.	The	CyT49	line	was	used	owing	to	their	efficient	differentiation	to	the	definitive	endoderm	and	pancreatic	lineages65,58.	To	optimize	a	protocol	for	using	genome	editing	technologies	to	generate	reporter	lines,	I	first	targeted	the	OCT4	locus.	OCT4	is	an	important	member	of	the	pluripotency	network	as	Oct4-/-	mouse	embryos	are	unable	to	form	pluripotent	cells192.	In	addition	to	its	role	in	pluripotency,	Oct4	is	also	required	for	the	formation	of	all	embryonic	germ	layers	in	vitro	and	in	vivo193,	making	it	a	good	target	to	not	only	investigate	the	effect	of	genome	editing	on	pluripotency	but	also	on	differentiation.	Second,	a	NEUROG3-2A-eGFP	hESC	line	was	generated	using	CRISPR-Cas9.	This	reporter	line	marks	human	endocrine	progenitor	cells	derived	from	hESCs,	a	tool	that	will	be	important	for	the	studies	included	in	the	remainder	of	this	thesis.		 	 71 5.2 Results	5.2.1 Generation	of	OCT4-eGFP-2A-Puro	hESC	lines	using	genetically	engineered	nucleases	To	target	the	OCT4	locus,	a	pair	of	TALENs	were	designed	where	the	repeat	variable	domain	at	amino	acids	12-13	dictated	the	specific	nucleotide	each	individual	TALE	binds	to	(Figure	17A)85,194.	By	designing	the	TALEN	pair	to	bind	on	either	side	of	the	OCT4	stop	codon,	the	Fok1	nuclease	domains	homodimerize	and	generate	a	double	stranded	break	(DSB)	(Figure	17B).	This	DSB	can	either	be	repaired	through	the	error-prone	non-homologous	end	joining	to	generate	OCT4	loss-of-function	mutations,	or	through	homologous	recombination	of	a	provided	donor	plasmid.	To	fuse	eGFP-2A-Puro	downstream	of	the	last	exon	of	OCT4,	a	donor	plasmid	was	provided	as	a	template	for	homologous	recombination.	Applying	this	strategy,	52	puromycin-resistant	clones	from	two	electroporations	were	picked	and	characterized.	Sixteen	of	these	clones	(31%)	were	correctly	targeted	at	both	the	5’	and	3’	ends	as	determined	by	PCR	genotyping	(Table	5).	As	the	primer	pairs	used	to	amplify	the	5’	and	3’	regions	of	the	genomic	insertion	contained	one	primer	that	bound	outside	of	OCT4	donor	vector	homology	arms	and	a	second	primer	that	bound	within	sequences	not	contained	in	wild	type	cells	(Figure	17B&D),	these	experiments	correctly	distinguished	targeted	clones	from	those	with	the	random	genomic	insertions.	Further	sequence	analyses	of	the	obtained	PCR	products	of	three	correctly	targeted	clones,	OCT4-2,	OCT4-3,	and	OCT4-28,	confirmed	precise	insertion	of	the	reporter	gene	without	introduced	errors.	To	determine	if	the	insertion	was	found	in	one	or	both	alleles,	PCR	genotyping	was	used	to	distinguish	the	wildtype	allele	from	the	modified	allele,	indicating	 72 that	all	three	hESC	lines	were	heterozygous	for	the	insertion	(Figure	17C).	Taken	together,	these	results	demonstrate	that	this	new	TALEN	pair	can	drive	efficient	genomic	modification	downstream	of	OCT4	in	hESCs.								 73   Figure	17:	Targeting	strategy	using	genetically	engineered	nucleases	to	generate	OCT4-eGFP-2A-Puro	hESC	lines	(A)	The	structure	of	Xanthomonas	sp	TALE	protein.	Each	nucleotide-binding	module	is	comprised	of	a	34-amino	acid	sequence,	inside	of	which	is	embedded	one	of	four	repeat	variable	domains	(RVD).	The	sequence	of	this	di-amino	acid	RVD	dictates	the	deoxynucleotide-binding	cipher:	NG	is	highly	specific	for	deoxythymidine,	HD	for	deoxycytidine,	NI	for	deoxyadenosine,	and	NH	for	deoxyguanosine.	(B)	Schematic	overview	of	the	targeting	strategy	using	TALENs	to	knock	eGFP	onto	the	OCT4	coding	sequence.	Red	line	represents	the	stop	codon.	Regions	where	genotyping	PCR	primer	pairs	bind	are	highlighted	for	5’F,	5’R,	3’F	and	3’R.	(C)	Genotyping	PCR	for	i)	5’	arm	of	insertion	using	primers	5’F	and	5’R	(1180	bp)	ii)	3’	arm	of	insertion	using	primers	3’F	and	3’R	(1000	bp)	iii)	endogenous	allele	using	primers	5’F	and	3’R	for	parent	CyT49	line	and	three	of	the	generated	knock-in	lines	OCT4-2,	OCT4-3	and	OCT4-28.	(D)	Schematic	overview	of	the	CRISPR-Cas9	targeting	strategy.	Red	line	represents	the	stop	codon.	A	green	circle	represents	the	Cas9	endonuclease	with	the	tracrRNA	in	black.	The	genomic	protospacer	adjacent	motif	(PAM)	sequence	is	highlighted	in	red	type	and	the	guide	RNA	sequence	is	in	bold	type.	Genotyping	PCR	primer	pairs	are	the	same	as	for	TALEN	targeting	and	are	highlighted.	 74 As	OCT4	had	not	been	targeted	using	the	CRISPR-Cas	system	and	previous	reports	suggested	CRISPR-Cas9	is	more	efficient,	I	next	compared	the	efficiencies	of	TALEN	and	CRISPR-Cas9	technologies.	CRISPR-Cas9	is	an	RNA-guided	endonuclease	technology195,196	and	requires	two	distinct	components:	the	guide	RNA	(gRNA)	and	the	Cas9	endonuclease.	The	gRNA	binds	to	target	genomic	sequences	by	complementary	base	pairing	and	recruits	the	Cas9	endonuclease	to	generate	a	DSB	that	can	be	repaired	through	homologous	recombination	(Figure	17D).		 Early	attempts	at	targeting	the	OCT4	locus	using	the	protocol	outlined	by	Ran	et	al.140	were	unsuccessful,	with	only	one	of	177	puro-resistant	clones	from	three	separate	electroporations	correctly	targeted	at	the	5’	end.	Because	of	low	expression	from	the	Cbh	promoter	in	CyT49	cells,	a	new	CRISPR-Cas9	vector	was	generated	that	utilized	the	full	length	CAGGS	promoter	to	drive	Cas9	expression.	Using	this	expression	system,	15/32	(47%)	clones	were	correctly	targeted	at	both	the	5’	and	3’	ends	(Table	5)	and	one	of	these	clones	was	homozygous	for	the	insertion	while	all	others	were	heterozygous.	Table	5:	Targeting	efficiency	of	TALEN/CRISPR-mediated	OCT4-eGFP-2A-Puro	CyT49	hESC	lines  	5.2.2 OCT4-eGFP	reporter	lines	have	normal	OCT4	protein	expression	and	stem	cell	phenotype		 To	determine	if	the	knock-in	of	eGFP	to	the	C-terminus	of	OCT4	faithfully	reported	Donor	Plasmid	 Donor	Amount	 Nuclease	(Amount)	 Cell	Number	 Number	of	Clones	 Targeted	at	5’	 Targeted	at	3’	 Correctly	Targeted	 Targeting	Efficiency	(%)	Oct4-eGFP-2A-Puro	40	ug	 TALEN	(15	ug	each)	 10	million	 52	 32	 25	 16	 31	Oct4-eGFP-2A-Puro	40	ug	 CRISPR-Cas	(15	ug)	 10	million	 32	 31	 15	 15	 47	 75 expression,	immunofluorescence	for	OCT4	in	the	CyT49,	OCT4-2,	OCT4-3,	and	OCT4-28	hESC	lines	was	performed.	As	depicted	in	Figure	18,	OCT4	and	GFP	expression	completely	overlapped	in	these	three	cell	lines	indicating	that	GFP	faithfully	recapitulated	endogenous	OCT4	expression.	Further,	similar	OCT4	staining	intensities	between	targeted	and	parental	cells	suggested	that	the	targeting	did	not	affect	native	OCT4	expression	levels	(Figure	18).	Importantly,	all	three	of	the	reporter	lines	maintained	eGFP	expression	for	up	to	eight	passages	in	culture	confirming	the	stability	of	this	insertion	with	minimal	effects	on	maintenance	of	pluripotency.			Figure	18:	Knock-in	add	on	of	eGFP	does	not	impact	native	OCT4	expression	in	targeted	hESC	lines	Undifferentiated	cells	on	optical	dishes	were	fixed	in	4%	PFA,	permeabilized	using	0.5%	triton-X	and	stained	for	OCT4.	Images	were	obtained	on	a	Leica	SP8	confocal	microscope	and	native	eGFP	fluorescence	(green)	overlapped	completely	with	both	OCT4	immunostaining	(red)	and	nuclear	counterstain	using	TO-PRO	Iodide	(blue).	Scale	bars	=	50	µm.	 76 	 To	ensure	the	genomic	modification	did	not	alter	the	stem	cell	characteristics	of	these	cells,	immunofluorescent	staining	for	two	other	pluripotency	markers,	NANOG	and	SOX2	(Figure	19)	was	performed.	Similar	immunostaining	intensities	for	these	pluripotency	factors	were	observed	in	all	four	cell	lines,	suggesting	that	the	targeting	and	cloning	did	not	impact	pluripotency.			Figure	19:	Stem	cell	characteristics	are	retained	in	OCT4-eGFP-2A-Puro	reporter	hESC	lines	Cells	were	grown	on	optical	dishes	and	then	fixed	with	4%	PFA	and	permeabilized	with	0.5%	triton-X.	Immunostaining	was	carried	out	and	images	were	obtained	on	a	Leica	SP8	confocal	microscope.	Native	eGFP	fluorescence	(green)	overlapped	with	both	SOX2	or	NANOG	immunostaining	(red)	and	nuclear	counterstain	using	TO-PRO	Iodide	(blue).	Scale	bars	=	50	µm.	 5.2.3 OCT4-eGFP	reporter	lines	form	the	definitive	endoderm	germ	layer		 To	understand	if	OCT4	expression	levels	were	downregulated	at	similar	rates	 77 during	definitive	endoderm	(DE)	formation	in	the	targeted	lines,	cells	were	differentiated	to	DE	using	protocol	1	and	qPCR	was	performed.	During	DE	formation,	only	the	OCT4-2	line	showed	significantly	elevated	levels	of	OCT4	when	compared	to	CyT49	controls	(Figure	20A).	Next,	the	efficiency	of	DE	formation	was	determined	by	measuring	the	number	of	CXCR4	immunopositive(+)	cells	using	flow	cytometry197.	As	seen	in	Figure	20B,	there	were	no	differences	in	the	number	of	CXCR4+	cells	derived	from	the	OCT4-2	or	OCT4-28	lines;	however,	there	was	a	slight,	13%	reduction	in	CXCR4+	cells	in	the	OCT4-3	line.	The	OCT4-eGFP	lines	expressed	normal	levels	of	DE	markers	including:	CER1	(Figure	20C),	GSC	(Figure	20D),	and	SOX17	(Figure	20F).	Of	note,	none	of	these	markers	were	significantly	different	between	the	reporter	lines	and	CyT49,	except	for	a	2.8-fold	reduction	in	GSC	expression	again	in	OCT4-2	(Figure	20D).	SOX7	expression	was	measured	to	determine	if	there	was	any	change	in	the	formation	of	visceral	endoderm	(Figure	20E)	and	no	significant	differences	were	observed.	Taken	together,	these	results	suggest	that	knocking	eGFP	onto	OCT4	has	minimal	effects	on	the	formation	of	DE	beyond	previously	described	clonal	variation64.	 78   Figure	20:	Differentiation	of	the	definitive	endoderm	germ	layer	is	unaffected	by	the	addition	of	eGFP	into	the	OCT4	locus	hESCs	were	differentiated	into	definitive	endoderm	using	a	three-day	protocol,	cells	were	collected	and	expression	of	OCT4	(A),	CER1	(C),	GSC	(D),	SOX7	(E),	and	SOX17	(F)	were	analyzed	using	Taqman	qPCR.	All	genes	were	normalized	to	TATA	Binding	Protein	expression	(TBP).	(B)	hESC-derived	DE	cells	were	fixed	with	4%	PFA	and	stained	for	the	cell	surface	marker	CXCR4.	The	number	of	CXCR4+	DE	cells	was	detected	using	BD	FACSCalibur	in	the	CyT49,	OCT4-2,	OCT4-3	and	OCT4-28	lines.	Statistical	analysis	was	carried	out	using	a	one-way	ANOVA	followed	by	a	Dunnett	post-test.	n³3.	*p<0.05.		 To	demonstrate	the	utility	of	eGFP	to	enrich	for	OCT4	expression,	the	reporter	lines	were	differentiated	to	definitive	endoderm	using	protocol	1.	FACS	was	then	used	to	collect	the	eGFP+	and	eGFP-	cells	and	qPCR	was	carried	out	on	the	two	populations.	As	expected,	OCT4	expression	was	enriched	2.7-,	1.5-,	and	4.4-fold	in	the	eGFP+	cells	from	OCT4-2,	OCT4-3,	and	OCT4-28	lines,	respectively	(Figure	21A,	C,	E).	Surprisingly,	abundant	OCT4	 79 mRNA	was	observed	in	the	GFP-	fraction	of	OCT4-3	(Figure	21),	potentially	resulting	from	a	mixed	clone.	As	SOX17	is	important	in	hESCs	for	DE	formation198,	SOX17	expression	was	measured	and	found	to	be	enriched	19.3-,	118.1-,	and	83.3-fold	in	the	eGFP-	cells	from	OCT4-2,	OCT4-3,	and	OCT4-28	lines,	respectively	(Figure	21B,	D,	F).	This	provides	evidence	that	reporter	expression	driven	by	a	single	endogenous	promoter	is	sufficient	to	isolate	cells	via	FACS	and	that	these	cell	lines	could	be	used	to	assess	for	presence	of	OCT4-expressing	cells	in	mixed,	differentiating	cultures.		Figure	21:	OCT4	and	SOX17	expression	in	GFP+	and	GFP-	cells	Cells	were	trypsinized	on	the	second	day	of	differentiation	to	definitive	endoderm	and	the	GFP+	and	GFP-	populations	were	collected	into	TRIzol	using	the	BD	FACS	Aria.	RNA	was	isolated	and	cDNA	synthesized	before	carrying	out	qPCR	analysis	for	OCT4	and	SOX17	using	TBP	as	control	gene.	Statistical	analysis	was	performed	using	a	Student’s	t-test.	n³3.	*p<0.05,	**p<0.01,	***p<0.001.	 80 5.2.4 eGFP	expression	mirrors	OCT4	protein	levels	in	OCT4	reporter	lines		 To	determine	whether	the	stability	of	the	OCT4-eGFP	fusion	protein	is	similar	to	that	of	native	OCT4,	western	blot	analyses	of	OCT4	and	GFP	in	stem	cells	(Day	0),	definitive	endoderm	(Days	1-3)	and	posterior	foregut	(Days	4-6)	(Figure	22)	was	performed	in	hESC	differentiations	using	protocol	1.	The	decline	of	OCT4	protein	in	the	CyT49	parental	line	was	consistent	with	mRNA	expression	(c.f.	Figure	22	&	Figure	23).	Furthermore,	the	OCT4-eGFP	fusion	protein	was	downregulated	at	a	similar	rate	to	wild	type	OCT4	in	all	three	genetically	modified	reporter	lines	and	the	rate	of	OCT4	loss	is	consistent	with	the	FACS	data	(Figure	23).			Figure	22:	Western	blot	analysis	of	OCT4	and	eGFP	expression	during	differentiation	of	hESCs	to	definitive	endoderm	and	primitive	gut	tube	Protein	lysates	were	collected	on	Day	0	(hESC),	Days	1–3	(definitive	endoderm)	and	Days	4–6	(primitive	gut	tube)	and	the	expression	of	OCT4	(fusion	protein	indicated	by	arrowhead),	eGFP	and	the	control	protein	GAPDH	were	analyzed	in	the	CyT49,	OCT4-2,	OCT4-3	and	OCT4-28	hESC	lines	using	SDS-PAGE	followed	by	western	blotting.	 81 5.2.5 Ability	to	differentiate	into	pancreatic	progenitors	is	maintained	in	OCT4	reporter	lines		 To	determine	whether	fusion	of	GFP	downstream	of	OCT4	would	alter	OCT4	stability	and	possibly	the	dynamics	of	differentiation	towards	endodermally-derived	tissues,	OCT4	gene	expression	was	assessed	during	the	differentiation	towards	pancreas	using	protocol	1.	To	confirm	that	the	loss	of	eGFP	expression	in	the	reporter	lines	mirrors	the	decline	in	OCT4	expression	that	is	observed	in	the	parental	line	(Figure	23A-B),	I	characterized	eGFP	expression	during	the	differentiation	towards	pancreas	in	the	OCT4	reporter	lines	(Figure	23C-E).	In	undifferentiated	cells,	98.6%,	79.3%,	and	92.1%	of	cells	were	GFP+	in	OCT4-2,	OCT4-3,	and	OCT4-28	lines,	respectively.	Consistent	with	the	role	of	OCT4	in	endoderm	formation,	51.9%,	12.7%,	and	11.7%	GFP+	cells	were	detected	at	the	end	of	day	3	in	OCT-2,	OCT4-3,	and	OCT4-28	lines,	respectively.	Furthermore,	less	than	1%	of	cells	were	GFP+	in	hESC-derived	pancreatic	endoderm	at	day	12	from	any	of	the	reporter	lines,	which	is	consistent	with	the	number	of	OCT4+	cells	in	the	differentiated	CyT49	line	(Figure	23).	Thus,	in	agreement	with	data	presented	in	Figure	20	and	Figure	22,	OCT4-2	has	a	delayed	loss	of	OCT4	protein	upon	differentiation,	which	may	result	in	a	delayed	differentiation	of	endodermally-derived	progenitors.	 82 		Figure	23:	eGFP	expression	decreases	upon	differentiation	towards	pancreatic	progenitor	cells	(A)	The	mRNA	expression	of	OCT4	was	measured	during	the	differentiation	of	CyT49	cells.	Day	0	represents	undifferentiated	hESC	cells,	Days	1–3	are	cells	becoming	definitive	endoderm,	Days	4–6	are	cells	becoming	posterior	foregut,	Day	9	are	pancreatic	endoderm	cells	and	Day	12	are	pancreatic	progenitors	and	endocrine	cells.	(B)	CyT49	cells	were	fixed	with	4%	PFA,	permeabilized	in	0.5%	triton-X,	stained	using	rabbit	anti-OCT4	antibodies	followed	by	FITC-conjugated	donkey	anti-rabbit	antibodies.	The	number	of	FITC+	cells	was	measured	using	a	BD	FACSCalibur	during	several	days	of	the	differentiation	protocol.	The	number	of	GFP+	cells	was	measured	using	native	eGFP	fluorescence	and	a	BD	FACSCalibur	in	OCT4-2	(C),	OCT4-3	(D)	and	OCT4-28	(E)	fixed	cells.	n³3.	 83 To	determine	whether	the	targeted	clones	would	differentiate	towards	the	pancreatic	lineage	with	similar	efficiencies	as	the	parental	line,	immunocytochemical	analyses	for	NKX6-1199-202	and	SOX9203-206	were	performed	on	day	12	(Figure	24).	Despite	the	small	differences	in	loss	of	OCT4	expression	described	above,	there	was	no	appreciable	change	in	the	level	of	these	two	proteins	in	the	reporter	lines	compared	to	the	parental	line.	Supporting	this	protein	data,	the	expression	of	PDX1,	SOX9	or	NKX6-1	was	comparable	between	all	reporter	lines	using	qPCR	analysis	(Figure	25).	Taken	together,	these	findings	suggest	that	this	approach	is	amenable	for	the	future	creation	of	tissue	or	cell-type	specific	reporter	lines.						 84 		Figure	24:	Addition	of	eGFP	does	not	affect	the	efficiency	of	pancreatic	progenitor	formation	during	in	vitro	differentiation	protocol	Immunostaining	for	NKX6-1	(green)	and	SOX9	(red)	was	carried	out	on	day	12	in	parent	CyT49	hESCs	and	knock-in	OCT4-2,	OCT4-3	and	OCT4-28	hESC	lines	that	were	grown	and	differentiated	in	optical	dishes	using	protocol	1.	Images	were	obtained	on	a	Leica	SP8	confocal	microscope	using	TO-PRO	Iodide	(blue)	as	nuclear	counterstain.	Scale	bars	=	25	µm.			Figure	25:	Gene	expression	analysis	of	pancreatic	endoderm	markers.	No	change	in	gene	expression	of	NKX6.1	(A),	PDX1	(B),	and	SOX9	(C)	in	stage	4	OCT4-2,	OCT4-3,	and	OCT4-28	cells	differentiated	using	protocol	1.	 85 5.2.6 Generation	and	characterization	of	NEUROG3-2A-eGFP	knock-in	reporter	CyT49	hESC	lines		 To	specifically	mark	endocrine	progenitors	derived	from	hESCs,	NEUROG3-2A-eGFP	knock-in	reporter	CyT49	hESC	lines	were	generated	using	CRISPR-Cas9	(Figure	26).	These	hESC	reporter	lines	allow	for	the	easy	isolation	of	endocrine	progenitors	from	the	heterogeneous	population	of	cells	generated	during	stem	cell	differentiation.	Unlike	the	OCT4-eGFP	fusion	reporter	line,	a	2A	sequence	was	used	to	generate	separate	NEUROG3	and	eGFP	proteins,	reducing	the	potential	for	altered	NEUROG3	protein	stability.			Figure	26:	Generation	of	NEUROG3-2A-eGFP	knock-in	reporter	CyT49	hESC	lines	Schematic	overview	of	the	CRISPR-Cas9	targeting	strategy	for	the	NEUROG3	locus.	gRNA	sequence	(bold)	guides	the	Cas9	endonuclease	(green	circle)	to	generate	a	double	stranded	break	upstream	of	the	stop	codon	(red).	The	genomic	protospacer	adjacent	motif	(PAM)	sequence	is	highlighted	in	red	type.	Genotyping	PCR	primer	pairs	are	highlighted.	 	 Of	the	total	67	clones	that	were	picked,	22	were	correctly	targeted	at	both	the	5’	and	3’	ends,	resulting	in	a	targeting	efficiency	of	33%	(Table	6).	From	these	clones,	three	clonal	lines	(N2-2,	N4-7	and	N5-5)	were	picked	for	further	analysis.	All	three	of	these	lines	are	heterozygous	for	the	insertion	and	sequencing	revealed	no	disruption	of	the	wildtype	allele.	In	addition,	the	floxed	puro	cassette	was	removed	using	transfection	of	cre-expressing	plasmid	and	further	subcloning.	 86 Table	6:	Targeting	efficiency	of	CRISPR-mediated	NEUROG3-2A-eGFP	CyT49	hESC	lines		 Donor		Plasmid	 Donor	Amount	 CRISPR-Cas9	Amount	 Cell	Number	 Correctly	Targeted	 Total	#	of	Clones	 Efficiency	(%)	NEUROG3-2A-eGFP	PGK-Puro	 40	µg	 15	µg	 10	million	 22	 67	 33		5.2.7 Characterization	of	NEUROG3-2A-eGFP	reporter	lines	To	confirm	the	fidelity	of	the	GFP	fluorescence	to	the	endocrine	lineages,	NEUROG3	reporter	lines	were	differentiated	to	stage	6	using	protocol	2.	The	number	of	eGFP+	cells	were	comparable	between	all	three	lines	(Figure	27A).	Next,	GFP+	and	GFP-	cells	were	FACS	sorted	and	qPCR	gene	expression	analysis	was	performed.	As	expected,	GFP+	cells	had	significantly	increased	expression	of	NEUROG3	and	INS	(Figure	27B&	C)	compared	with	GFP-	and	CyT49	cells.	Conversely,	SOX9	expression	was	enriched	in	GFP-	cells	compared	to	GFP+	(Figure	27D).	Together,	these	studies	suggest	that	NEUROG3-2A-eGFP	reporter	lines	enrich	for	endocrine	progenitor	cells	from	a	heterogeneous	population.			Figure	27:	Validation	of	Neurog3-2A-eGFP	CyT49	hESC	reporter	lines	(A)	Reporter	hESC	lines	were	differentiated	to	stage	6	and	the	number	of	GFP+	(NEUROG3+)	cells	were	quantified	using	flow	cytometry.	n	=	7	from	two	differentiations.	(B)	NEUROG3	and	(C)	INS	transcription	was	increased	in	GFP+	cells	compared	to	GFP-	cells	in	Stage	6	endocrine	progenitors	in	N2-2,	N4-7	and	N5-5	hESC	lines.	Conversely,	SOX9	expression	was	higher	in	GFP-	cells	compared	to	GFP+	cells	(D).	*p	<	0.05	by	one-way	ANOVA	and	Tukey	post	test.	n	=	3	from	two	differentiations.	Data	are	presented	as	mean	±	SEM.	 	 87 5.3 Discussion	This	chapter	outlines	two	strategies	for	the	creation	of	cell-type	specific	hESC	reporter	lines	using	either	TALEN	or	CRISPR-Cas9	genome	editing	methodologies.	Prior	to	these	approaches,	genome	editing	was	difficult	in	hESCs	due	to	low	rates	of	homologous	recombination80.	Approaches,	such	as	those	described	herein,	have	greatly	improved	the	efficiency	of	generating	reporter	hESCs	in	three	ways:	1)	through	reducing	the	burden	of	generating	constructs	containing	long	homology	arms;	2)	by	simplifying	clone	screening	and	verification	processes;	and	3)	by	increasing	the	likelihood	of	homologous	recombination.	Using	TALEN	or	CRISPR-Cas9,	reporter	lines	can	be	generated	in	approximately	four	weeks	and	allow	for	enrichment	of	specific	cell	types	from	the	heterogeneous	population	of	cells	present	during	hESC	differentiations.	In	order	to	optimize	the	genome	editing	approaches,	the	OCT4	locus	was	targeted	first	as	two	other	studies	had	shown	the	utility	of	genome	editing	in	generating	mutations	at	the	OCT4	locus.	Previously,	Zinc	Finger	Nucleases	were	used	to	insert	eGFP	into	one	of	two	regions	of	OCT4,	both	of	which	disrupted	the	protein	coding	region	in	BG01	hESCs181.	More	recently,	TALENs	were	used	to	insert	eGFP	downstream	of	the	last	exon	of	OCT4	to	create	an	OCT4-eGFP	fusion	protein,	avoiding	the	disruption	of	the	protein	coding	gene141.	As	I	wanted	to	create	reporter	lines	that	did	not	disrupt	protein	function,	the	latter	approach	was	used	to	generate	OCT4-eGFP-2A-Puro	reporter	lines	using	both	TALEN	and	CRISPR-Cas9	systems.	OCT4,	a	POU	domain	transcription	factor	that	can	act	as	an	activator	and	repressor,	was	first	identified	as	a	central	member	of	the	pluripotency	network192,207.	As	such,	its	absence	causes	ESCs	to	differentiate	into	trophoblasts208.	In	human	pluripotent	cells,	OCT4	 88 interacts	with	both	NANOG	and	SOX2	to	activate	pluripotent	genes209.	Importantly	no	changes	in	the	expression	of	NANOG	or	SOX2	were	observed	in	the	targeted	cell	lines,	consistent	with	the	fact	that	the	OCT4-eGFP	fusion	protein	did	not	alter	the	pluripotent	nature	of	these	hESC	lines.	However,	OCT4	expression	was	observed	in	the	GFP-	fraction	from	OCT4-3	line.	This	could	reflect	a	mixed	clone	and	highlights	a	limitation	of	this	approach:	ensuring	lines	are	derived	from	a	single	cell	can	often	be	difficult.	In	addition	to	its	role	in	maintaining	embryonic	stem	cell	pluripotency,	OCT4	is	also	important	for	differentiation.	Oct4	expression	is	required	for	the	differentiation	of	all	embryonic	lineages	in	vitro	and	in	vivo193.	For	instance,	in	Danio	rerio	the	Oct4	homolog	pou5f1	is	essential	for	endoderm	formation210.	The	dual	role	of	Oct4	in	both	maintaining	pluripotency	and	establishing	endoderm	is	believed	to	be	driven	by	its	Sox	binding	partner.	Oct4	interacts	with	Sox2	at	“canonical”	binding	sites	to	maintain	pluripotency,	while	endoderm	specification	involves	Oct4	and	Sox17	binding	at	“compressed”	Sox/Oct	motif211.	It	also	appears	that	the	level	of	human	OCT4	expression	dictates	which	lineage	stem	cells	will	differentiate	towards:	reduced	expression	of	OCT4	promotes	the	mesoderm	lineage	while	elevated	OCT4	promotes	adoption	of	the	endoderm	lineage212.	The	critical	role	of	OCT4	in	endoderm	formation	suggests	that	any	changes	in	OCT4	expression	or	stability	in	hESC	reporter	cell	lines	may	alter	their	differentiation	potential,	especially	to	endodermally-derived	tissues.	In	the	three	OCT4-eGFP	reporter	lines	characterized	here	there	were	no	significant	differences	in	expression	of	the	DE	genes	CER1	and	SOX17	and	the	VE	gene	SOX7;	however,	there	were	significant	changes	in	GSC	in	OCT4-2.	Interestingly,	the	significant	decrease	in	GSC	expression	was	concurrent	with	a	significant	increase	in	OCT4	expression.	As	OCT4	can	act	as	a	transcriptional	repressor	it	is	possible	that	GSC	 89 expression,	and	endoderm	formation,	is	repressed	by	prolonged	expression	of	OCT4.	Even	with	the	slight	differences	in	the	expression	profile	of	the	DE	generated	from	these	reporter	lines,	they	formed	pancreatic	progenitors	with	similar	efficiencies,	suggesting	that	this	strategy	does	not	dramatically	disrupt	differentiation.	Having	validated	that	knocking	in	a	fluorescent	protein	to	a	locus	does	not	impact	function,	I	next	set	out	to	use	CRISPR-Cas9	to	generate	an	endocrine	progenitor	reporter	line:	NEUROG3-2A-eGFP.	Previously,	other	NEUROG3-2A-GFP	reporter	lines	were	generated	using	Zinc	Finger	Nucleases	in	the	H1	and	H9	hESC	lines213.	Transplanting	200,000	NEUROG3-GFP+	cells	under	the	kidney	capsule	of	immunodeficient	mice	showed	that	NEUROG3-expressing	cells	can	give	rise	to	single	hormone	positive	endocrine	cells213.	In	addition,	there	was	no	evidence	of	exocrine,	ductal,	liver,	intestinal,	or	neuronal	cells	within	the	graft213.	Accordingly,	the	N5-5	CyT49	hESC	reporter	line	allows	for	the	enrichment	of	NEUROG3+	endocrine	progenitors,	based	on	elevated	gene	expression	of	NEUROG3	and	INS	in	GFP+	cells.	Together,	these	results	highlight	the	effectiveness	of	NEUROG3	reporter	lines	in	isolating	endocrine	progenitor	cells	from	a	heterogeneous	population.	In	summary,	this	work	has	shown	that	the	CyT49	hESC	line	are	amenable	to	genomic	modification	using	two	genome-editing	technologies.	These	studies	add	to	the	growing	body	of	literature	that	shows	nuclease-mediated	genome	engineering	is	a	powerful	approach	to	generate	reporter	hESC	lines	that	can	be	used	to	answer	biological	questions	in	a	human	model	of	pancreas	development.	 	 90 Chapter	6: CDK	inhibition	stabilizes	NEUROG3	during	human	embryonic	stem	cell	differentiation	6.1 Background	Most	of	what	is	known	about	b-cell	development	comes	from	studies	done	in	mice.	However,	difficulty	may	arise	when	trying	to	translate	these	findings	to	human	development	due	to	subtle	differences	between	species.	In	fact,	there	are	examples	of	studies	performed	in	mice	where	the	mouse	does	not	recapitulate	the	human	phenotype.	For	instance,	humans	with	haploinsufficiency	for	GATA6	have	neonatal	diabetes	due	to	a	complete	lack	of	pancreas214.	However,	pancreas-specific	deletion	of	Gata6	in	mice	results	in	phenotypically	normal	pancreatic	development	and	function215.	With	recent	advances	in	genome	editing	technologies,	such	as	those	described	in	Chapter	5,	it	is	now	possible	to	study	mechanisms	of	human	pancreas	development	using	hESC	differentiations	as	a	model.	Using	this	approach,	the	role	of	GATA6	in	human	pancreas	development	was	recently	corroborated216.	In	this	chapter,	I	investigate	the	role	of	the	cell	cycle	in	regulating	human	NEUROG3	using	differentiations	of	endocrine	progenitor-specific	reporter	hESCs.	While	Neurog3-null	mice	do	not	form	endocrine	cells46,	whether	human	NEUROG3	is	also	required	for	endocrine	cell	formation	was	controversial.	The	first	homozygous	mutation	in	NEUROG3	was	identified	in	a	patient	with	congenital	malabsorptive	diarrhea217.	However,	as	this	patient	did	not	present	with	diabetes	until	late	childhood,	it	was	hypothesized	that	NEUROG3	may	not	be	essential	for	human	pancreas	development218.	This	hypothesis	was	also	supported	by	several	other	patient	specific	mutations	in	NEUROG3	that	did	not	lead	to	diabetes218,219.	To	investigate	the	role	of	NEUROG3	in	human	pancreas	development,	McGrath	and	colleagues	generated	a	NEUROG3-/-	hESC	line	using	 91 CRISPR-Cas9189.	This	loss-of-function	hESC	line	forms	pancreas	progenitors	efficiently	but,	due	to	a	lack	of	NEUROG3	protein,	does	not	generate	any	endocrine	cells189.	However,	endocrine	cells	are	able	to	form	with	an	89%	reduction	in	NEUROG3	using	shRNA189,	suggesting	that	the	residual	NEUROG3	is	able	to	support	endocrine	development.	Consistently,	reductions	of	over	90%	in	NEUROG3	mRNA	is	required	to	prevent	endocrine	cell	development	in	the	intestine220.	Taken	together,	these	results	suggest	that	human	NEUROG3	is	required	for	endocrine	cell	formation.	Cell	cycle	proteins	regulate	endocrine	cell	formation	during	mouse	pancreas	development.	In	this	chapter,	I	aim	to	understand	whether	CDKs	can	also	regulate	human	pancreas	development	using	hESC	differentiations	as	a	model.	First,	expression	of	NEUROG3	was	characterized	using	the	N5-5	NEUROG3-2A-eGFP	reporter	hESC	line.	investigate	whether	CDK	inhibition	increases	NEUROG3	protein	expression.	Second,	the	effect	of	CDKi	treatment	on	cell	cycle	length	was	investigated	using	the	FUCCI	cell	cycle	reporter	hESC	line.	Lastly,	the	effect	of	CDK	inhibition	on	the	endocrine	progenitor	population	was	studied.		 	 92 6.2 Results	6.2.1 Neurog3	expression	peaks	between	stage	5	and	6	of	hESC	differentiations		 To	determine	when	Neurog3	expression	peaks	during	hESC	differentiation,	the	number	of	GFP+	cells	was	profiled	during	the	differentiation	(protocol	2)	of	the	N5-5	hESC	line	using	flow	cytometry	(Figure	28).	No	GFP+	cells	were	detected	at	the	start	of	stage	3	(S3D1)	or	stage	4	(S4D1).	Beginning	in	stage	5	(S5D1)	the	first	GFP+	cells	were	detected	at	an	average	of	5.2%	and	increased	to	32.8%	following	24	hours	in	stage	6	(S6D2).	After	nine	days	in	stage	6	(S6D9),	the	number	of	GFP+	cells	decreased	to	9.56%.	Rare	GFP+	cells	remain	following	two	weeks	in	stage	6	and	stage	7.	These	data	indicate	that	the	peak	of	NEUROG3	expression	is	during	the	transition	from	stage	5	to	stage	6.			Figure	28:	The	number	of	GFP+	cells	peaks	during	the	transition	from	stage	5	to	stage	6	The	number	of	GFP+	cells	was	quantified	using	a	flow	cytometry	on	S3D1,	S4D1,	S5D1,	S5D2,	S5D3,	S6D1,	S6D2,	S6D3,	S6D4,	S6D9.	These	cells	were	differentiated	using	protocol	2.	Individual	wells	of	a	differentiation	are	represented	by	open	circles.	For	each	time	point,	2-7	wells	from	at	least	two	differentiations	were	analyzed.			 To	further	understand	the	kinetics	of	gene	expression	during	stage	5	and	6,	gene	expression	analysis	for	a	panel	of	pancreas	specific	genes	was	performed	using	Nanostring	 93 (Figure	29A&B).	For	this	experiment,	differentiation	experiments	from	two	consecutive	passages	(P31	and	P32)	using	CyT49	N5-5	cells	were	collected	daily	beginning	at	the	start	of	stage	5	(S5D1)	and	continuing	48	hours	into	stage	6	(S6D3).	Importantly,	the	patterns	of	gene	expression	were	consistent	between	the	two	differentiation	experiments	and	only	varied	in	expression	level.	Pancreatic	markers	SOX9,	RFX6,	SOX4	and	PDX1	remained	expressed	at	a	mid-level	throughout	stage	5	and	6	(Figure	29A).	In	the	transition	from	the	end	of	stage	4	(S5D1)	to	24	hours	in	stage	5	(S5D2),	many	genes	were	immediately	upregulated	before	maintaining	consistent	expression	during	stage	5	and	6,	including	PAX4	and	NKX2-2	(Figure	29A).	Expression	of	the	key	endocrine	lineage	gene	NEUROG3	peaked	at	the	transition	from	stage	5	to	stage	6	(Figure	29A),	consistent	with	previous	GFP	expression	analysis	(Figure	28).	As	expected,	expression	of	NEUROD1,	the	downstream	target	of	NEUROG3,	increased	in	stage	6	after	the	peak	of	NEUROG3	(Figure	29A).	Finally,	expression	of	key	endocrine	genes,	such	as	GCG,	INS,	CHGA,	and	SST,	were	highly	upregulated	during	stage	5	and	6	(Figure	29B).	Taken	together,	these	results	highlight	the	suitability	of	the	N5-5	line	in	modeling	human	pancreas	development.		 94 		Figure	29:	Gene	expression	analysis	of	progenitor	and	endocrine	markers	in	CyT49	N5-5	hESC	line.	CyT49	N5-5	hESC	cells	were	differentiated	to	stage	5	and	6	using	protocol	2.	Starting	at	the	beginning	of	stage	5	(S5D1),	samples	were	collected	daily	and	lysed	using	Buffer	RLT.	Gene	expression	analysis	was	performed	for	progenitor	(A)	and	endocrine	(B)	markers	using	a	custom	Nanostring	nCounter	XT	CodeSet	panel	and	normalized	to	housekeeping	genes.	Data	are	presented	as	log2(normalized	counts).	Hierarchical	clustering	was	performed	using	Euclidean	distance	metrics.		6.2.2 CDK	inhibition	increases	the	proportion	of	cells	in	G1	and	the	expression	of	pancreas-specific	genes	To	investigate	the	role	of	the	cell	cycle	and	CDK	inhibition	on	human	endocrine	cell	differentiation,	a	transgenic	CyT49	hESC	cell-cycle	reporter	line	using	fluorescence	ubiquitination-based	cell-cycle	indicator	(FUCCI)	was	generated	(Figure	30A)221.	This	line	(FUCCI-3)	allows	for	the	facile	isolation	of	cells	in	G1	(red)	versus	S-G2-M	(green)	based	on	fluorescence	(Figure	30B).	 95 		Figure	30:	Generation	of	FUCCI-3	CyT49	transgenic	hESC	line	(A)	Schematic	of	transgenic	FUCCI	allele.	CAGGS	promoter	drives	expression	of	mKO-hCDT	and	mAG-hGEM	separated	by	a	2A	peptide.	(B)	Cells	in	G1	express	the	red	mKO	fluorescent	protein	while	cells	in	S-G2-M	express	the	green	mAG	fluorescent	protein.		 Treating	stage	6	cells	derived	from	the	FUCCI-3	line	with	inhibitors	of	CDK2/4/6	(CDKi)	increased	the	proportion	of	cells	in	G1	from	85%	to	93%,	suggesting	the	CDKi	treatment	slows	the	G1-phase	(Figure	31A).	FACS	was	next	used	to	isolate	G1	and	S-G2-M	cells	following	CDKi	and	gene	expression	analyses	were	performed,	indicating	that	expression	of	NEUROG3,	NEUROD1,	and	PDX1	are	significantly	upregulated	in	G1	compared	with	S-G2-M	cells	(Figure	31B-D).	In	addition,	CDKi	treatment	significantly	increased	the	expression	of	both	NEUROG3	and	PDX1	in	G1-phase	cells	compared	with	control	vehicle-treated	cells	(Figure	31B-D).		 96 		Figure	31:	CDK	inhibition	increases	expression	of	pancreas	specific	genes	in	the	G1-phase	of	the	cell	cycle	(A)	The	number	of	cells	in	G1	increases	with	CDK	inhibition.	(B-D)	The	expression	of	NEUROG3,	NEUROD1,	and	PDX1	is	increased	in	G1	compared	to	S-G2-M.	The	expression	of	NEUROG3	(B)	and	PDX1	(D)	is	further	increased	in	G1	with	CDK	inhibition	n	=	3	wells	from	two	differentiations.	*p	<	0.05,	**p	<	0.01,	****p	<	0.0001	by	one-way	ANOVA	and	Tukey	post	hoc	test.		6.2.3 CDK	inhibition	increases	the	number	of	GFP+	endocrine	progenitors		 As	CDK	inhibition	increased	the	number	of	cells	in	G1,	I	next	asked	if	CDKi	increased	the	number	of	NEUROG3+	cells.	CDKi	treatment	of	human	endocrine	progenitors	derived	from	N5-5	using	protocol	2	significantly	increased	the	number	of	GFP+	cells	by	1.7-fold	as	determined	by	flow	cytometry	(Figure	32A).	Consistent	with	previous	findings	that	phosphorylation	of	the	Neurogenin	family	member	Neurog2	reduces	its	transcriptional	activity222,	the	expression	of	NEUROG3	and	NEUROD1,	which	are	direct	targets	of	NEUROG3,	and	PDX1,	which	is	a	b/d-cell	marker,	were	all	increased	in	NEUROG3-lineage	cells	following	CDKi	treatment	(Figure	32B-D).	 97 		Figure	32:	CDK	inhibition	increases	NEUROG3	protein	expression	and	activation	of	downstream	target	genes	(A)	eGFP+	cells	were	quantified	using	a	NEUROG3-2A-eGFP	hESCs	(N5-5)	differentiated	to	stage	6	prior	to	24	hour	CDKi	treatment.	n	=	9	wells	from	two	differentiations.	*	p	<	0.05	by	Mann-Whitney	U	test.	(B)	NEUROG3	expression	in	FACS	purified	GFP+	N5-5	cells	following	CDKi	treatment	at	stage	6.	n	=	3	wells	from	two	differentiations.	*p	<0.05	by	one-way	ANOVA	and	Tukey	post	hoc	test.	(C-D)	CDKi	treatment	increased	expression	of	NEUROD1	(C)	and	PDX1	(D)	in	FACS	purified	GFP+	cells.	n	=	3	wells	from	two	differentiations.	*p	<	0.05,	**p	<	0.01,	***p	<	0.001	by	one-way	ANOVA	and	Tukey	post	hoc	test.		6.2.4 CDK	inhibition	during	iPSC	differentiation	increases	expression	of	human	NEUROG3		 To	ensure	that	the	increased	number	of	GFP+	cells	following	CDKi	reflects	an	increase	in	NEUROG3,	the	expression	of	NEUROG3	protein	following	CDKi	was	assayed.	Treating	human	pancreatic	progenitors	derived	from	human	induced	pluripotent	stem	cells	(iPSCs)	with	CDKi	for	24	hours	increased	the	expression	of	NEUROG3	protein	4.6-fold	(Figure	33A&B).	Interestingly,	unlike	in	the	mouse	where	the	predominant	form	of	NEUROG3	is	unphosphorylated	(Figure	16A),	in	hIPSCs	the	hyperphosphorylated	form	of	NEUROG3	is	the	most	abundant	(Figure	33A;	arrowhead).	 98   Figure	33:	CDK2/4/6	inhibition	increases	NEUROG3	protein	expression	in	hiPSC-derived	endocrine	progenitor	cells	(A)	Western	blot	of	protein	lysates	from	three	control	and	treatment	wells	of	one	differentiation	of	human	iPSCs.	Inhibitors	of	CDK2/4/6	(CDKi)	were	added	to	cultures	for	24	hours	at	the	start	of	stage	6.	Arrowhead	represents	hyperphosphorylated	form	and	arrow	is	unphosphorylated.	(B)	Stage	6	endocrine	progenitor	cells	derived	from	human	iPSCs	were	treated	for	24	hours	with	CDKi	and	NEUROG3	protein	was	quantified	by	western	blot.	n	=	3	wells.	**	p	<	0.01	by	unpaired	t	test.		 	 99 6.3 Discussion	Pancreatic	endocrine	cell	development	is	initiated	by	Neurog3	activation	within	bipotent	trunk	progenitor	cells.	In	the	mouse,	the	expression	of	Neurog3	is	biphasic,	with	an	initial	increase	from	E8.5	to	E11	and	a	second	peak	at	E15.5	during	the	“secondary	transition”167.	However,	during	human	development	there	is	only	one	wave	of	endocrine	differentiation	that	occurs	near	the	end	of	embryogenesis	and	appears	to	mirror	the	“secondary	transition”	of	the	mouse62.	Data	from	human	fetal	tissue	shows	that	NEUROG3	is	detected	from	8-21	weeks	post	conception	(wpc)	with	the	peak	of	NEUROG3-dependent	endocrine	differentiation	between	10-17	wpc223.	Using	the	N5-5	NEUROG3-2A-eGFP	reporter	line,	the	expression	of	GFP	was	found	to	peak	at	the	transition	from	stage	5	to	stage	6	during	hESC	differentiation.	In	addition,	the	bulk	gene	expression	pattern	of	several	pancreas	specific	genes	during	hESC	differentiation	was	found	to	mirror	mouse	pancreas	development.	Maturity	markers,	such	as	MAFA	and	UCN3,	were	expressed	at	low	levels	throughout	the	two	stages	of	hESC	differentiations,	consistent	with	the	immature	population	of	cells.	Importantly,	expression	of	endocrine	hormones,	GCG,	INS,	and	CHGA,	increased	during	this	time	period.	In	Chapter	7,	the	single	cell	transcriptome	of	hESC-derived	endocrine	cells	will	be	performed,	allowing	for	further	elucidation	of	the	gene	expression	changes	during	differentiation.	In	mouse	explants,	CDKi	increased	the	number	of	Neurog3+.	In	this	chapter,	CDKi	treatment	was	found	to	increase	the	number	of	cells	in	G1-phase	using	FUCCI	reporter	hESC	lines.	Inhibition	of	CDK2/4/6	using	small	molecules	also	increased	the	protein	expression	of	NEUROG3	in	hiPSC	differentiations,	suggesting	that,	as	in	the	mouse,	G1	lengthening	and	CDKi	treatment	drives	human	endocrine	cell	formation.	Interestingly,	a	 100 difference	in	the	phosphorylation	of	NEUROG3	was	noted	between	mouse	and	hESC-derived	endocrine	progenitors.	While	in	the	mouse	the	predominant	form	of	Neurog3	protein	is	the	smaller,	less	phosphorylated	form,	hESC-derived	endocrine	progenitors	express	the	hyperphosphorylated	form	of	NEUROG3.	This	suggests	that,	while	current	differentiation	protocols	induce	high	protein	expression	of	NEUROG3,	it	is	mainly	hyperphosphorylated.	As	CDKi	increases	the	transcriptional	activity	of	NEUROG3,	based	on	the	upregulation	of	NEUROD1,	preventing	this	hyperphosphorylation	of	NEUROG3	during	hESC	differentiations	may	improve	the	efficiency	of	endocrine	cell	formation.		 	 101 Chapter	7: Single	cell	transcriptomics	of	mouse	and	human	endocrine	progenitor	cells	7.1 Introduction	During	mouse	and	human	pancreas	development,	pancreatic	progenitors	become	restricted	to	the	endocrine	cell	fate	before	differentiating	to	hormone	producing	cells.	This	process	involves	many	transcription	factors	(TFs)	that	drive	the	changes	in	gene	expression	necessary	for	endocrine	cell	genesis.	Genetic	loss-of-function	studies	in	mice	have	identified	a	role	of	these	individual	TFs	in	the	formation	of	specific	islet	cell	types.	We	have	begun	to	map	the	TF	cascade	that	regulates	the	formation	of	endocrine	cells,	including	the	b-cell224.	However,	our	understanding	of	fate	decisions	during	endocrine	cell	formation	is	based	on	studies	looking	at	the	whole	population	of	progenitors,	using	technologies	such	as	bulk	RNA-sequencing	and	often	only	in	mouse	cells.	The	gene	expression	of	individual	human	and	mouse	cells	during	terminal	differentiation	is	unknown.	A	promising	method	to	understand	gene	expression	changes	at	the	single	cell	level	is	single	cell	RNA-sequencing	(scRNA-seq).	Following	the	first	publication	in	2009225,	commercial	platforms	and	lower	sequencing	costs	have	made	scRNA-seq	a	feasible	technology	for	many	biologists.	Recently,	several	studies	have	investigated	the	single	cell	transcriptome	of	healthy	and	T2D	human	islets226-230.	From	these	studies,	we	have	begun	to	appreciate	the	cell-type	specific	gene	expression	changes	that	occur	during	diabetes	progression,	the	differences	between	mouse	and	human	islets,	and	the	identity	of	novel	islet	and	pancreatic	cell	types.	Two	recent	studies	have	begun	characterization	of	the	single	cell	transcriptome	of	mouse	and	human	endocrine	progenitors.	The	first	investigated	the	single	cell	gene	 102 expression	of	E13.5	embryonic	pancreatic	cells	but	very	few	endocrine	progenitors	were	sequenced231.	The	second	performed	single	cell	qPCR	on	500	cells	during	several	stages	of	hESC	differentiation	towards	b-like	cells191.	In	this	chapter,	I	aim	to	identify	cell	populations	during	mouse	and	human	pancreas	development	using	scRNA-seq.	To	isolate	pure	populations	of	endocrine	progenitors	cells	reporter	mice	and	hESC	lines	were	used.	Characterization	of	these	populations	will	aid	efforts	to	generate	an	unlimited	source	of	insulin-producing	b-cells	for	diabetes	treatment. 103 7.2 Results		7.2.1 Strategy	for	generating	quality	controlled,	single	cell	transcriptome	data	from	mouse	embryonic	pancreas	To	isolate	progenitor	populations	during	mouse	embryogenesis,	two	mouse	lines	were	used:	the	Neurog3-Cre	and	Rosa26mTmG	mouse	lines	(Figure	34).	In	Neurog3-Cre;	Rosa26mTmG	embryos,	all	cells	of	the	pancreas	are	labelled	with	a	membrane-targeted	Tomato	red	fluorescent	protein	(mTomato).	Upon	activation	of	the	Neurog3	promoter,	Cre	recombinase	will	remove	the	floxed	mTomato	cassette,	resulting	in	expression	of	a	membrane-targeted	enhanced	green	fluorescent	protein	(mGreen).	Cells	that	have	recently	activated	Neurog3	will	express	both	mTomato	and	mGreen,	making	these	cells	yellow232	(Figure	34).	Cells	that	are	further	along	the	endocrine	cell	lineage	will	express	mGreen	only.	Using	this	strategy,	three	progenitor	populations	can	be	isolated:	the	pancreatic	progenitor	(red),	the	endocrine	progenitor	cells	(Neurog3+;	yellow)	and	maturing	endocrine	lineage	cells	(green)	(Figure	34).	  Figure	34:	Strategy	for	isolating	mouse	pancreatic	progenitors	and	endocrine	cells	during	embryogenesis	Schematic	overview	of	the	two	mouse	lines	used	to	isolate	cell	populations	during	pancreas	development.	Using	this	strategy,	pancreatic	progenitors	are	mTomato+,	endocrine	progenitors	are	mTomato+	and	mGreen+	(yellow),	and	endocrine	cells	are	mGreen+.  104 Using	this	approach,	the	three	populations	were	isolated	from	E15.5	and	E18.5	embryos	and	single	cell	libraries	were	generated	using	10x	Genomics	Chromium™	Single	Cell	3’	Kit.	In	total,	6,528	E15.5	and	6,693	E18.5	mouse	pancreatic	cells	were	sequenced	using	Illumina	NextSeq500	(Table	7).	At	E15.5,	5,111	red	pancreatic	progenitors	and	1,417	yellow	and	green	endocrine	cells	(Table	7).	Because	cells	of	the	endocrine	lineage	are	rare	during	this	point	in	development,	the	yellow	and	green	populations	were	pooled	and	sequenced	as	one	library.	At	E18.5,	2,727	red	progenitor	cells,	577	yellow	endocrine	progenitor	cells,	and	3,389	green	endocrine	cells	were	sequenced	(Table	7).	Table	7:	Metrics	of	scRNA-seq	libraries	from	mouse	E15.5,	mouse	E18.5	and	human	S6D1	 Species	 Stage	 Sample	name	 #	of	cells	sequenced	 Post-norm	mean	reads/cell	 Median	genes/cell	Mouse	 E15.5	 Total	cells	 6,528	 55,573	 2,687	Red	 5,111	 60,852	 2,733	Yellow	&	green	 1,417	 55,110	 2,692	E18.5	 Total	cells	 6,693	 50,918	 2,264	Red	 2,727	 52,531	 2,381	Yellow	 577	 56,172	 2,391	Green	 3,389	 54,758	 2,289	Human	 S6D1	 GFP+	 4,631	 54,922	 2,242	 Following	sequencing,	data	were	analyzed	using	publically	available	software	programs	and	R	pipelines	(Figure	35).	First,	cellranger	mkfastq	(10x	Genomics)	generates	FASTQ	files	from	the	raw	sequencing	data,	storing	the	nucleotide	sequence	and	its	corresponding	quality	score	in	a	text-based	format	for	further	analysis.	Next,	cellranger	count	uses	the	FASTQ	file	to	perform	sequence	alignment,	filter	sequences	based	on	quality	score,	and	generate	single	cell	gene	counts.	As	an	optional	step,	cellranger	aggr	can	be	used	to	combine	data	from	multiple	samples.	This	was	used	to	merge	all	E15.5	and	E18.5	libraries	into	E15.5	total	cells	and	E18.5	total	cells	datasets,	respectively.	 105 As	minimal	filtering	is	performed	in	cellranger,	I	next	used	two	R	pipelines	to	filter	out	cells	that	did	not	meet	the	quality	control	(QC)	standard.	The	first	pipeline	is	called	Scater	(https://bioconductor.org/packages/release/bioc/html/scater.html)	and	is	a	single	cell	analysis	pipeline	that	places	a	great	emphasis	on	quality	control233.	This	was	used	to	filter	low-quality	cells	based	on	counts	(transcripts/gene)	or	genes	(genes/cell)	greater	than	3	standard	deviation	away	from	the	mean.	Scater	discards	cells	based	on	the	total	number	of	expressed	genes,	removing	doublets	and	debris,	and	removes	low-abundance	genes	or	genes	with	high	dropout	rate	based	on	expression	level.	This	QC	dataset	was	then	analyzed	using	the	Seurat	pipeline	(http://satijalab.org/seurat/),	another	R	toolkit	for	single	cell	genomics234.	Seurat	was	used	to	remove	common	sources	of	variation	including	number	of	genes	(each	cell	must	express	a	minimum	of	500	genes),	number	of	counts	(each	gene	must	be	expressed	in	a	minimum	of	three	cells),	and	cell	cycle	phase.	Finally,	unsupervised	clustering	was	performed	using	Seurat	to	cluster	cells	based	on	gene	expression	and	to	identify	unique	cell	types	within	the	population	(Figure	35). 		Figure	35:	Pipeline	to	generate	sequenced	libraries	Following	sequencing	on	the	NextSeq®	500	Illumina	Sequencer,	cellranger	mkfastq	software	was	used	to	generate	FASTQ	files.	To	perform	alignment,	filtering	and	generate	UMI	counts,	cellranger	count	software	was	used.	As	an	optional	step,	cellranger	aggr	can	be	used	to	combine	data	from	multiple	samples	into	one	data	set	for	further	analysis.	This	step	was	used	to	generate	E15.5	total	cells,	E18.5	total	cells,	E18.5	yellow	&	green	datasets.	Using	the	R	package	scater,	cells	were	filtered	based	on	number	of	genes	expressed	and	genes	were	filtered	based	on	number	of	transcripts	expressed.	Finally,	in	the	R	package	seurat,	variation	based	on	genes	expressed,	transcript	number,	or	cell	cycle	phase	was	regressed	out	before	further	analysis.  106 7.2.2 Identification	of	cell	types	in	E15.5	pancreas		 First,	the	cells	from	the	E15.5	embryo	were	examined	and	13	individual	clusters	were	identified	based	on	unsupervised	k-means	clustering	(Figure	36).	Using	the	top	ten	highly	expressed	genes	within	each	cluster,	the	identity	of	the	cells	was	inferred.	Acinar	cells	(8.6%)	were	named	based	on	expression	of	Cpa1	and	Cpa2	(Figure	36).	Next	to	the	acinar	cells	were	the	Sox9-expressing	bipotent	trunk	progenitor	cells.	The	population	of	duct	cells	were	identified	by	Krt19	expression	(Figure	36).	The	endocrine	progenitors	(EP;	8.1%)	were	recognized	by	the	expression	of	Neurog3,	Cdkn1a,	Nkx2-2,	and	Pax4.	Closely	associated	with	the	endocrine	progenitors	were	the	endocrine	cells	(11.5%),	expressing	many	hormones	including	Gcg,	Ins1,	Ins2,	Iapp,	Pyy,	and	Gast.	There	was	also	a	rare	population	(2.4%)	of	CD45+	(Ptprc)	and	F4/80+	(Adgre1)	macrophages	(Figure	37).	Finally,	most	the	sequenced	cells	at	E15.5	were	found	in	four	clusters	that	appeared	to	be	mesenchymal	cells	(49.8%)	(Figure	36).	 107 		Figure	36:	Analysis	of	single	cells	in	E15.5	pancreas	reveals	13	clusters	Within	the	E15.5	pancreatic	cells	there	were	13	clusters,	including	duct	and	acinar	cells.	The	endocrine	population	separates	into	a	progenitor	cell	(EP)	and	hormone	producing	cells	(endocrine).	In	addition,	a	small	population	of	macrophage	cells	were	identified.	Most	of	the	sequenced	cells	at	E15.5	were	found	within	four	clusters	that	express	markers	of	mesenchyme.			Figure	37:	Expression	of	macrophage	markers	in	E15.5	pancreatic	cells	Cells	within	the	macrophage	cluster	expressed	(A)	Ptprc	(CD45)	and	(B)	Adgre1	(F4/80).		 108 	 To	further	support	the	cell	types	classifications	based	on	cluster	gene	expression,	I	next	profiled	the	expression	of	several	key	pancreatic	genes	in	individual	cells	at	E15.5	(Figure	38).	Pdx1,	a	marker	of	both	pancreatic	epithelium	and	b-cells,	was	found	in	many	cells	of	trunk,	endocrine	progenitor	and	endocrine	population.	Interestingly,	a	subset	of	the	acinar	cells	also	expressed	Pdx1	(Figure	38A).	This	is	in	discrepancy	with	Pdx1	protein	expression,	which	is	minimally	found	in	acinar	and	duct	cells.	The	trunk	cells	expressed	Nkx6-1,	Pdx1	and	Sox9	and	a	few	rare	cells	expressed	Neurog3	(Figure	38A-C&J).	Both	Ins1	and	Ins2	are	coexpressed	in	cells	within	the	endocrine	population	and	were	exclusive	from	those	cells	expressing	Gcg	(Figure	38D-F).	The	ductal	cell	population	had	high	expression	of	Krt19	and	a	few	cells	within	the	trunk	and	endocrine	progenitor	populations	also	expressed	Krt19	(Figure	38I).	Finally,	Cpa1	and	Cpa2	were	highly	expressed	within	the	acinar	cells	(Figure	38G-H).	 109 		Figure	38:	Single	cell	expression	of	pancreatic	genes	at	E15.5	Single	cell	expression	of	pancreatic	progenitor	markers	(A)	Pdx1,	(B)	Sox9,	and	(J)	Nkx6-1	and	endocrine	progenitor	marker	(C)	Neurog3.	Endocrine	hormones	(D)	Ins1,	(E)	Ins2,	and	(F)	Gcg	mark	the	b-	and	a-cell	lineages,	respectively.	Acinar	cell	enzymes	(G)	Cpa1	and	(H)	Cpa2	and	highly	expressed	in	a	subset	of	cells	while	ductal	cell	marker	(I)	Krt19	is	also	upregulated	in	a	population	of	cells.	  110   Figure	39:	Single	cell	expression	of	cell	cycle	genes	in	E15.5	mouse	pancreas	(A)	E15.5	mouse	pancreatic	cells	clustered	into	seven	different	cell	types.	(B)	The	cell	cycle	phase	of	each	individual	cell	was	classified	based	on	the	expression	of	S-phase	and	G2/M-phase	genes.	tSNE	plots	of	single	cell	gene	expression	for	cell	cycle	genes	(C)	Cdk1,	(D)	Cdk2,	(E)	Cdk4,	(F)	Cdk6,	(G)	Gadd45a,	and	(H)	Neurog3.			 As	I	previously	identified	a	role	for	cell	cycle	proteins	in	regulating	Neurog3	protein,	I	next	investigated	the	single	cell	expression	of	several	key	cell	cycle	genes	in	E15.5	trunk,	 111 endocrine	progenitors	(EP)	and	endocrine	cells	(Figure	39A).	Using	Seurat,	individual	cells	were	classified	as	G1-,	S-,	or	G2/M-phase	based	on	gene	expression	(Figure	39B).	Most	endocrine	progenitors	and	endocrine	cells	were	in	G1-phase	(Figure	39B),	consistent	with	previous	reports	that	these	populations	do	not	undergo	cell	division	frequently49,51,110.	In	contrast,	the	trunk	population	consisted	of	G1-,	S-,	and	G2/M-phase	cells	(Figure	39B).	To	understand	the	role	of	Cdks	during	mouse	pancreas	development,	the	single	cell	gene	expression	of	Cdk1,	Cdk2,	Cdk4,	and	Cdk6	was	characterized	(Figure	39C-F).	Cdk1	was	found	in	trunk	and	endocrine	progenitor	cells	that	were	in	the	G2/M-phase	(Figure	39C).	Cdk2	and	Cdk6	were	found	in	a	small	proportion	of	trunk	and	endocrine	progenitor	cells	while	Cdk4	was	the	most	abundantly	expressed	(Figure	39D-F).	Consistent	with	my	previous	findings	that	Cdks	negatively	regulate	Neurog3	protein,	Cdk1,	Cdk2,	and	Cdk6	were	not	highly	expressed	in	Neurog3-expressing	endocrine	progenitor	cells	(Figure	39H).	Interestingly,	the	cell	cycle	gene	Growth	arrest	and	DNA-damage-inducible	protein	alpha	(Gadd45a)	was	found	to	be	highly	expressed	in	Neurog3-expressing	trunk	and	endocrine	progenitor	cells,	suggesting	it	may	play	a	role	in	the	Neurog3+	cells	(Figure	39G).	 112   Figure	40:	Single	cell	expression	of	genes	highly	expressed	in	mesenchyme	clusters	Single	cell	expression	pattern	of	(A)	Notch2,	(B)	Itm2a,	(C)	Ptn,	(D)	Ncam1,	(E)	Tgfbi,	(F)	Lpar1,	(G)	Tshz2,	(H)	H19,	(I)	Akap12,	(J)	Cbx3,	(K)	Nr2f2,	(L)	Kdelr2,	(M)	Col1a1,	and	(N)	Rpgrip1	in	E15.5	embryonic	pancreas.	 113 Many	of	the	genes	that	were	highly	expressed	in	the	four	mesenchyme	clusters	are	the	same,	including	Notch2,	Itm2a,	Ptn,	Ncam1,	Tgfbi,	Lpar1,	Tshz2,	H19,	Akap12,	Cbx3,	Nr2f2,	Kdelr2,	Col1a1,	and	Rpgrip1	(Figure	40A-N).	These	genes	were	also	expressed	in	some	acinar,	ductal,	and	endocrine	cells.	Due	to	their	ubiquitous	expression	throughout	many	cells	it	was	difficult	to	assign	a	role	to	this	cell	type	based	on	the	top	ten	genes.	To	confirm	their	identity	as	mesenchyme,	the	single	cell	expression	of	the	“mesenchymal”	marker	Vimentin	(Vim)	and	the	epithelial	marker	E-cadherin	(Cdh1)	was	determined	(Figure	41).	Cdh1	expression	was	restricted	to	the	trunk	and	endocrine	cells,	while	the	Vim	was	clearly	enriched	in	the	mesenchymal	cells.			Figure	41:	Expression	of	mesenchymal	and	epithelial	markers	in	E15.5	embryos	(A)	Vim,	a	marker	of	mesenchyme,	is	found	in	many	cells	including	mesenchyme,	trunk,	and	duct	cells.	Importantly,	these	clusters	do	not	express	the	epithelial	marker	(B)	Cdh1.  7.2.3 Characterization	of	the	endocrine	cell	transcriptome	in	E15.5	pancreas	There	were	seven	clusters	of	E15.5	yellow	and	green	cells	that	can	be	divided	into	acinar,	endocrine	progenitors,	Chga-expressing	immature	endocrine	cells,	a-,	b-,	d-cells,	and	macrophages	(Figure	42).	Endocrine	progenitors	(EP;	28.7%)	express	Neurog3	along	with	tdTomato,	suggesting	that	these	cells	recently	activated	Cre	recombinase	(Figure	43).	Other	highly	expressed	genes	in	this	cluster	include	Mdk,	Gadd45a,	Pax4,	and	Tgm7	(Figure	 114 43).	The	a-cells	(26.4%)	express	Gcg,	Tmem27,	Slc38a5,	and	Peg10	(Figure	42	and	Figure	43).	The	b-cells	(18.4%)	express	Ins1,	Ins2,	Pdx1,	and	Iapp	(Figure	42	and	Figure	43).	Interestingly,	there	appear	to	be	a	few	rare	macrophage	cells	(1.2%)	that	have	a	similar	expression	profile	to	the	f	cluster	found	in	Figure	36.			Figure	42:	tSNE	plot	of	clusters	in	endocrine	cells	at	E15.5	Clustering	of	E15.5	yellow	and	green	cells	revealed	7	clusters.	These	clusters	can	be	identified	as	endocrine	progenitors	(EP;	28.7%),	Chga	and	Chgb-expressing	immature	endocrine	cells	(Chg;	16.7%),	alpha	cells	(a;	26.4%),	beta	cells	(b;	18.4%),	ghrelin+	cells	(e;	3.9%),	acinar	cells	(4.7%),	and	macrophages	(f;	1.2%).	  115   Figure	43:	The	top	ten	genes	expressed	in	E15.5	yellow	and	green	clusters	Heatmap	showing	single	cell	expression	of	the	top	ten	genes	in	the	seven	E15.5	yellow	and	green	cell	clusters.	Columns	represent	clusters	and	are	identified	by	colour	panel	located	on	the	bottom.	 116 7.2.4 Characterization	of	endocrine	cell	population	at	E18.5	Having	examined	endocrine	cell	types	at	E15.5,	I	next	aimed	to	characterize	endocrine	cells	at	E18.5.	As	there	are	more	endocrine	cells	at	E18.5,	the	yellow	and	green	cells	were	sequenced	separately,	allowing	for	the	identification	of	endocrine	progenitor	cells	(yellow)	from	their	descendants	(green).	In	the	E18.5	yellow	sample,	there	are	six	clusters	(Figure	44).	Endocrine	progenitors	(Figure	44;	EP)	were	named	based	on	expression	of	tdTomato,	Neurog3,	and	Nkx2-2	(Figure	45).	The	endocrine	cells	expressed	many	hormones	including	Ins1,	Ins2,	Gcg	and	Sst	(Figure	45).	There	were	a	few	macrophage	cells	that	expressed	C1qa,	C1qb,	C1qc,	Lyz2,	and	Apoe	(f;	4%)	(Figure	45	and	Figure	45).	Separate	from	the	endocrine	population	were	a	population	of	Notch-responsive	(Notch)	cells	based	on	expression	of	Hes1	and	Spp1	(Figure	44	and	Figure	45).	There	was	also	a	population	of	cells	that	expressed	Cryab	and	Plp1	that	may	be	glial	cells	(21.9%)235.	Finally,	the	Dlk	cells	expressed	several	collagen	genes	including	Col3a1,	Col1a2,	and	Col1a1	(Figure	45).  Figure	44:	tSNE	plots	of	yellow	populations	of	cells	at	E18.5	At	E18.5	there	are	six	distinct	clusters	of	yellow	cells.	The	endocrine	progenitors	(EP;	7.6%)	are	found	close	to	the	multi-hormone	expressing	endocrine	cells	(22.4%).	There	are	a	population	of	Notch-expressing	cells	(Notch;	28.5%)	and	Notch	ligand	Dlk-expressing	(Dlk;	15.5%)	progenitor	populations.	There	are	a	few	rare	macrophage	cells	(f;	4.0%).  117   Figure	45:	The	top	ten	genes	expressed	in	E18.5	yellow	clusters	Heatmap	showing	single	cell	expression	of	the	top	ten	genes	per	cluster	in	E18.5	yellow	cells.	Columns	represent	clusters	and	are	identified	by	colour	panel	located	on	the	bottom.  118 Next,	the	identity	of	the	10	clusters	of	endocrine	cells	in	E18.5	green	population	was	investigated	(Figure	46).	Most	these	cells	were	in	the	b-cell	cluster	(b)	and	expressed	genes	involved	in	glucose	metabolism	including	Slc2a2	and	G6pc2,	suggesting	a	more	functional	b-cell	than	those	found	in	the	E18.5	yellow	population	(Figure	47).	These	cells	also	expressed	Ins1	and	Ins2	at	a	high	level	(Figure	47).	There	were	three	clusters	closely	associated	with	the	b-cell	population	that	also	expressed	Ins1	and	Ins2	(Figure	47).	One	of	these	clusters	(MafB)	had	high	expression	of	Pdx1,	Mafb,	Cryba2,	Nkx6-1	and	Gadd45a,	suggesting	they	may	be	an	immature	b-cell	population.	In	addition,	cells	in	cluster	S	had	high	expression	of	genes	specific	to	the	S-phase,	such	as	Cdk1,	Topa2,	suggesting	that	this	is	a	small	(1%)	population	of	b-cells	undergoing	DNA	replication.	In	the	M	cluster	(2.5%),	the	cells	expressed	the	G2/M	gene	Ccnb1,	the	histone	genes	Hist1h2bc,	Hist1h1c,	H2af2,	and	the	kinetochore	protein	Spc25,	suggesting	that	these	cells	are	undergoing	mitosis.		 In	addition	to	b-cells,	there	were	other	endocrine	cell	types	that	were	easily	identified	based	on	their	single	cell	gene	expression.	There	were	a-cells	that	expressed	both	Gcg	and	Arx	and	the	d-cells	expressed	Hhex	and	Sst	(Figure	46	and	Figure	47).	The	pancreatic	polypeptide	cells	(PP)	expressed	Ppy	and	Etv1228	while	Ghrl	cells	expressed	Isl1,	Mdk	and	Cdkn1a	(Figure	46	and	Figure	47).	 119   Figure	46:	tSNE	plots	of	green	populations	of	cells	at	E18.5	At	E18.5,	there	are	ten	clusters	of	green	cells.	The	endocrine	cells	are	found	in	b	(26.8%),	a	(18.7%),	d	(17.3%)	and	PP	(7.6%).	There	is	also	a	population	of	b-cells	that	are	undergoing	DNA	replications	(S;	1.0%)	and	mitosis	(M;	2.5%).	The	Mafb	(15%)	cluster	appear	to	be	immature	b-cells.	  120   Figure	47:	Single	cell	expression	of	endocrine	cell	genes	in	E18.5	green	cells	The	expression	of	alpha	cell	markers	(A)	Arx	and	(E)	Gcg	are	specific	to	a	cluster.	(C)	Ins1	and	(D)	Ins2	are	highly	expressed	in	all	cells	of	b	cluster.	They	are	also	found	in	cells	of	clusters	S,	M	and	Mafb.	There	are	a	few	rare	cells	that	express	(B)	Neurog3	and	(J)	Cdkn1a,	mainly	within	the	Ghrl.	Expression	of	(F)	Hhex	and	(G)	Sst,	markers	of	the	delta	cell	lineage,	are	found	in	d	cluster.	(H)	Ppy,	a	marker	of	PP	cells,	is	most	highly	expressed	in	PP	cluster.	 7.2.5 Single	cell	transcriptome	of	NEUROG3-lineage	during	hESC	differentiation		 Lastly,	the	single	cell	transcriptome	of	GFP+	cells	from	NEUROG3-2A-eGFP	hESC	reporter	lines	was	profiled.	As	I	previously	found	that	the	number	of	GFP+	cells	peaked	at	the	end	of	stage	5	(Figure	28),	N5-5	cells	were	differentiated	using	protocol	2	and	collected	 121 for	scRNA-seq	before	the	transition	to	stage	6	(S6D1).	In	total,	4,631	GFP+	cells	were	sequenced	(Table	7).	Data	analysis	resulted	in	nine	clusters	that	can	be	classified	into	six	cell	types:	endocrine	progenitors	(EP)	(1750	cells),	polyhormonal	endocrine	(1478	cells),	delta	(424	cells),	duct	(255	cells),	liver	(385	cells),	and	an	unknown	cell	type	(205	cells)	(Figure	48).		 Further	examination	of	the	expression	of	endocrine	specific	genes	supported	these	cell	classifications.	While	all	cells	expressed	GFP	protein,	only	a	few	cells	expressed	NEUROG3	transcript	and	were	mainly	found	in	the	EP	population	(Figure	49A).	The	discrepancy	between	NEUROG3	transcript	and	GFP	protein	suggests	that	the	GFP	protein	also	marks	cells	that	do	not	have	NEUROG3	protein.	The	number	of	cells	expressing	NEUROG3	transcript	is	greater	than	that	of	GFP,	likely	resulting	from	the	monoallelic	expression	of	the	GFP	reporter	(Figure	49B).	NEUROD1,	a	direct	target	of	NEUROG3236,	was	widely	expressed	throughout	the	EP,	endocrine	and	delta	cell	populations	(Figure	49C).	Interestingly,	CDKN1A,	the	downstream	target	of	NEUROG3	that	reinforces	cell	cycle	exit	during	murine	endocrine	cell	differentiation110,	is	not	abundantly	expressed	in	hESC-derived	endocrine	progenitors	or	endocrine	cells	(Figure	49D).	 122   Figure	48:	tSNE	plot	of	single	GFP+	cells	at	S6D1	of	N5-5	hESC	differentiation	There	are	nine	clusters	of	cells	in	S6D1	GFP+	cells.	These	clusters	can	be	further	divided	into	several	cell	types	including	endocrine	progenitors	(EP),	polyhormonal	cells	(endocrine),	SST-expressing	endocrine	cells	(delta),	duct	cells,	and	liver	cells.			 The	endocrine	cell	population	was	made	up	of	hormone+	cells,	many	of	which	co-expressed	multiple	hormones	including	GCG,	SST,	and	INS	(Figure	49E,	H,	J).	To	understand	if	these	cells	are	of	the	a-cell	lineage,	the	expression	of	ARX	was	examined	and	found	that	almost	all	cells	in	the	endocrine	cluster	expressed	ARX	(Figure	49F).	ARX	knockout	hESC	lines	have	reduced	numbers	of	a-,	PP,	and	b-cells	and	re-expression	of	ARX	restores	insulin	expression	in	KO	cells237,	suggesting	ARX	is	required	for	the	formation	of	other	endocrine	cells	beyond	a-cells.	The	d-cell	population	highly	expressed	the	hormone	SST	and	the	transcription	factor	HHEX	(Figure	49G).	However,	most	of	the	cells	in	this	cluster	also	 123 expressed	INS,	ARX,	and/or	GHRL.	As	Ghrl+	cells	have	been	identified	as	a	potential	source	of	 a-,	PP-	and	rare	b-cells	during	mouse	development238,	the	expression	of	GHRL	throughout	the	EP,	endocrine	and	delta	population	suggests	that	these	populations	include	progenitor	cells	along	with	the	multihormonal	immature	endocrine	cells.			Figure	49:	Single	cell	expression	of	endocrine-specific	genes	Single	cell	expression	of	eGFP	(A)	and	NEUROG3	(B)	indicate	that	most	GFP+	cells	are	not	endocrine	progenitors.	(C)	NEUROD1	is	widely	expressed	throughout	the	endocrine	population	while	CDKN1A	(D)	is	rarely	found.	a-cell	markers	GCG	(E)	and	ARX	(F)	are	co-expressed	in	rare	cells	within	the	endocrine	cluster.	In	the	d-cell	cluster	there	is	robust	expression	of	both	HHEX	(G)	and	SST	(H).	(I)	GHRL	is	found	within	some	cells	of	the	EP,	endocrine	and	delta	cell	populations	where	INS	(J)	is	more	widely	expressed.	 124  Of	the	three	hormones,	INS	was	the	most	abundantly	expressed	and	can	be	detected	in	EP	and	endocrine	cells	(Figure	49J).	To	understand	if	any	of	these	cells	are	of	the	b-cell	lineage,	I	next	profiled	several	b-cell	specific	genes.	UCN3	is	a	marker	of	mature	human	a-	and	b-cells239	but	was	found	in	very	few	cells	at	S6D1	(Figure	50A),	suggesting	that	the	endocrine	cells	are	immature.	GCK	is	an	enzyme	necessary	for	glucose-stimulated	insulin	secretion.	There	were	several	cells	that	express	GCK	(Figure	50B),	many	of	which	co-expressed	INS.	While	GCK	phosphorylates	glucose	committing	it	to	glycolysis,	G6PC2	is	the	phosphatase	that	counteracts	this.	A	subset	of	the	endocrine	cells	expressed	G6PC2240,	although	not	as	many	as	GCK	(Figure	50C).	The	biosynthesis	of	INS	protein	involves	the	processing	of	precursors	by	the	enzymes	PCSK1	and	PCSK2241,242.	In	S6D1	cells,	PCSK2	transcript	was	more	abundant	than	PCSK1,	but	most	INS-expressing	cells	did	not	express	either	enzyme	(Figure	50D-E),	partially	explaining	the	defects	in	GSIS	of	hESC-derived	b-like	cells.	SLC30A8,	a	zinc	transporter	that	may	be	important	for	insulin	secretion,	was	expressed	in	a	subset	of	cells	(Figure	50F).	Taken	together,	these	results	suggest	that	most	INS-expressing	cells	are	immature	endocrine	cells.	Consistent	with	this	finding,	many	hESC-derived	endocrine	cells	expressed	MAFB	but	almost	none	expressed	MAFA	(Figure	51).	The	switch	in	expression	from	MafB	to	MafA	is	an	important	step	in	the	differentiation	of	mouse	b-cells243.	  125   Figure	50:	Expression	of	key	b-cell	genes	in	single	GFP+	S6D1	cells	(A)	The	b-cell	maturity	marker	UCN3	is	very	rarely	expressed.	(B)	GCK	and	(C)	G6PC2,	members	of	the	glucose	sensing	pathway,	are	expressed	in	some	cells,	mainly	in	the	endocrine	cluster.	Two	enzymes	involved	in	INS	protein	processing,	(D)	PCSK1	and	(E)	PCSK2,	are	expressed	in	a	scattering	of	EP	and	endocrine	cells.	The	zinc	transporter	(F)	SLC30A8	is	found	in	a	subset	of	the	endocrine	population.	   Figure	51:	Expression	of	MAF	genes	in	S6D1	GFP+	cells	MAFB	is	found	in	many	GFP+	S6D1	cells	while	almost	none	express	MAFA.	  126  Finally,	I	examined	the	expression	of	a	few	genes	that	are	markers	of	both	immature	progenitor	cells	as	well	as	terminally	differentiated	cells.	SOX9	is	restricted	to	the	ductal	cell	population	(Figure	52A).	There	were	some	NKX6-1-expressing	cells	in	the	endocrine	population	(Figure	52B).	NKX2-2	is	found	throughout	the	EP	and	endocrine	cell	clusters,	along	with	PDX1	(Figure	52C-D).	The	finding	that	the	GFP+	endocrine	cells	are	an	immature	population	is	likely	due	to	the	GFP	reporter	marking	only	cells	that	are	recent	descendants	of	NEUROG3+	endocrine	progenitor.	It	would	be	interesting	to	repeat	these	studies	using	a	lineage-tracing	model	that	would	allow	for	the	isolation	of	both	endocrine	progenitor	and	endocrine	cells,	as	was	performed	in	the	mouse	(Figure	34).	   Figure	52:	Expression	of	pancreas-specific	genes	in	S6D1	GFP+	cells	(A)	SOX9	expression	is	restricted	to	the	ductal	cell	population.	(B)	NKX6-1	is	mainly	found	in	the	EP	population,	while	(C)	NKX2-2	is	ubiquitously	expressed	throughout	the	EP	and	endocrine	population.	(D)	PDX1	is	found	in	some	GFP+	cells	at	S6D1.	 7.2.6 Discovery	of	non-endocrine	cells	in	GFP+	S6D1	cells		 Of	the	4,631	GFP+	cells,	845	cells	clustered	into	three	populations	separate	from	the	endocrine	cells.	Based	on	gene	expression,	these	cells	did	not	appear	to	belong	to	the	endocrine	lineage.	To	understand	whether	they	are	epithelial	or	mesenchymal	cells,	the	expression	of	CDH1	and	VIM	was	investigated	(Figure	53A-B).	While	some	non-endocrine	cells	expressed	the	epithelial	marker	CDH1,	there	was	a	cluster	of	cells	that	strongly	 127 expressed	the	“mesenchymal”	marker	VIM.	Using	the	top	ten	genes	expressed	in	this	cluster,	this	cluster	was	identified	as	liver	cells	expressing	alpha-fetoprotein	(AFP),	fibrinogen	beta	chain	(FGB),	fibrinogen	gamma	chain	(FGG),	and	transthyretin	(TTR)	(Figure	53C-F).	Interestingly,	some	endocrine	progenitors	and	endocrine	cells	expressed	these	liver	genes.	Whether	these	liver	proteins	play	a	role	in	endocrine	cell	formation	or	if	these	genes	are	erroneously	expressed	is	unknown.	  Figure	53:	Gene	expression	profile	of	liver	cluster	in	S6D1	GFP+	cells	Most	cells	of	the	liver	cluster	do	not	express	(A)	CDH1	but	express	(B)	VIM,	suggesting	they	are	mesenchymal	cells.	The	expression	of	(C)	AFP,	(D)	FGB,	(E)	FGG,	and	(F)	TTR	is	upregulated	in	the	liver	cluster	compared	to	the	other	cell	types.			 There	was	a	population	of	GFP+	cells	that	were	not	identified	based	on	the	expression	of	the	top	nine	genes:	CXCL14,	CRABP2,	TFAP2B,	NR2F2,	PDGFC,	ARHGAP29,	CA3,	HNRNPA1	and	NPM1	(Figure	54).	CXCL14	is	a	chemoattractant	that	is	important	for	the	 128 migration	of	immune	cells244.	They	also	express	cellular	retinoic	acid	binding	protein	2	(CRABP2)	and	the	transcription	factor	TFAP2B.	NR2F2	is	a	transcription	factor	that	controls	the	develop	of	several	tissues	but	has	not	been	implicated	in	the	pancreas.	PDGFC,	or	platelet	derived	growth	factor	C,	is	expressed	in	many	development	tissues	but	was	not	found	in	pancreas245.	ARHGAP29	is	a	Rho	GTPase	activating	protein	and	CA3,	carbonic	anhydrase	III,	is	specific	to	muscle	cells.	HNRNPA1	is	part	of	the	heterogeneous	nuclear	ribonucleoproteins	that	are	involved	in	mRNA	metabolism	and	transport,	while	NPM1	nucleophosmin	is	involved	in	ribosome	biogenesis.	Using	publicly	available	dataset	for	single	cell	gene	expression	in	human	islets	revealed	that	none	of	the	top	nine	differentially	expressed	genes	in	this	cluster	are	specific	to	any	pancreatic	or	endocrine	cell	type230.	In	addition,	these	genes	did	not	overlap	with	any	cell	clusters	in	the	E15.5	and	E18.5	mouse	data.			Figure	54:	Top	nine	genes	expressed	in	unknown	cluster	of	S6D1	GFP+	cells	(A)	CXCL14,	(B)	CRABP2,	(C)	TFAP2B,	(D)	NR2F2,	(E)	PDGFC,	(F)	ARHGAP29	and	(G)	CA3	are	highly	specific	to	this	cluster	of	cells.	Two	other	genes,	(H)	HNRNPA1	and	(I)	NPM1,	are	highly	expressed	in	the	unknown	cluster	but	are	also	expressed	in	other	subtypes	of	cells. 129 7.3 Discussion		 Single	cell	RNA-sequencing	allows	for	the	identification	of	novel	cell	types,	discovery	of	cell	state	specific	genes,	and	the	appreciation	of	cellular	heterogeneity	within	a	population.	In	this	chapter,	scRNA-seq	was	used	to	generate	a	resource	of	single	cell	transcriptomes	from	6,528	E15.5	embryonic	pancreatic	cells,	6,693	E18.5	embryonic	pancreatic	cells,	and	4,631	hESC-derived	NEUROG3-2A-eGFP	cells.	Several	unique	observations	were	made	including	the	presence	of	macrophage	cells	during	embryonic	pancreas	development,	novel	genes	that	specify	individual	cell	types,	and	previously	unidentified	populations	of	cells	generated	in	hESC	differentiations.	There	were	175	and	122	macrophage	cells	in	E15.5	and	E18.5	embryonic	pancreas,	respectively.	Macrophages	are	found	in	many	tissues	during	development246	and	play	important	roles	in	tissue	remodeling.	During	mammary	gland	development,	F4/80+	macrophage	cells	are	found	localized	to	the	highly	proliferative	epithelial	structures	and	are	required	for	proper	gland	development247.	Using	a	genetic	mouse	model	that	has	a	deficiency	in	macrophages	(Csf1op/op),	it	was	shown	that	macrophages	play	an	important	role	in	the	formation	of	the	epithelial	tree	during	mammary	gland	development	and	that	this	defect	can	be	rescued	by	restoring	the	macrophage	population.	However,	the	role	of	macrophages	in	the	growth	of	epithelial	organs	appears	to	be	mostly	indirect,	either	by	facilitating	the	clearance	of	apoptotic	cells248	or	remodeling	of	extracellular	matrices249.	In	mouse	pancreas	development,	macrophages	are	present	as	early	as	E12.5250,251.	Treating	pancreatic	explants	with	M-CSF	increased	the	number	of	insulin+	cells,	which	is	thought	to	be	via	the	differentiation/activation	of	macrophage	precursors250.	Csf1op/op	mice	have	reduced	b-cell	mass	due	to	decreased	proliferation	of	b-cells	during	the	late	embryonic	 130 period251.	Interestingly,	one	of	the	roles	proposed	for	these	macrophages	is	to	instruct	pancreatic	epithelial	cells	to	delaminate,	exit	the	cell	cycle	and	differentiate252.	It	would	be	interest	to	compare	the	single	cell	transcriptome	of	epithelial	progenitors	in	models	of	macrophage	deficiency	and	to	determine	if	this	has	any	effect	of	cell	cycle	length. 	 Comparison	of	E15.5	and	E18.5	endocrine	progenitor	cells	resulted	in	a	list	of	genes	that	are	important	in	endocrine	progenitors.	These	include	Btg2	and	Gadd45a,	both	of	which	are	involved	in	cell	cycle	regulation	and	have	been	implicated	in	neural	development.	Gadd45	genes	are	involved	in	tissue	development	via	their	role	in	cell	cycle	exit	and	DNA	demethylation253.	Proneural	proteins,	such	as	Neurog2,	NeuroD,	and	Ascl1,	have	been	shown	to	activate	expression	of	Gadd45g254,255.	This	leads	to	Gadd45-dependent	cell	cycle	exit	by	upregulation	of	cell	cycle	inhibitor	Cdkn1a256	and	direct	interaction	with	Cdk1/CyclinB257,	as	has	been	shown	in	Xenopus	embryos	for	both	Gadd45a	and	Gadd45g258.	In	addition,	studies	in	mice	implicate	Gadd45b	in	reducing	proliferation	of	neural	precursors	and	DNA	demethylation	of	promoters	involved	in	adult	neurogenesis259.	While	Gadd45	proteins	are	implicated	in	pancreatic	cancer256,	their	role	in	pancreas	development	has	not	been	investigated.	The	expression	of	Gadd45a	in	Neurog3+	endocrine	progenitor	cells	suggest	that	it	may	play	a	role,	along	with	Cdkn1a,	in	regulating	cell	cycle	exit.	Future	studies	will	explore	the	potential	role	of	Gadd45a	in	DNA	demethylation	of	endocrine-specific	promoters	during	pancreatic	endocrine	genesis		 Btg2,	also	known	as	Tis21,	is	a	negative	regulator	of	the	cell	cycle	that	inhibits	transcription	of	CyclinD1,	preventing	the	G1-S	transition260,261.	Deletion	of	Btg2	in	the	adult	dentate	gyrus	shortens	G1	length	in	progenitor	cells	and	prevents	their	terminal	differentiation262.	This	is	thought	to	be	caused	in	part	by	the	direct	binding	of	Btg2	to	Id3	 131 promoter.	Id	proteins	bind	E	proteins,	which	are	obligate	heterodimerization	partners	of	bHLH	proteins.	By	sequestering	E	proteins	and	preventing	their	association	with	proneural	bHLH	TFs,	Id	acts	to	prevent	terminal	differentiation263.	It	will	be	interesting	to	investigate	the	role	of	Btg2	in	pancreas	development.	Based	on	literature	from	neurogenesis,	Btg2	may	also	act	to	inhibit	Id3	transcription,	allowing	for	the	activation	of	proendocrine	genes,	including	Neurog3.	The	liver,	like	the	pancreas,	is	derived	from	the	foregut	endoderm.	The	region	of	the	endoderm	that	gives	rise	to	the	liver	can	also	form	the	ventral	pancreas264.	One	of	the	factors	that	controls	the	decision	between	liver	and	pancreas	is	the	secretion	of	FGFs	by	the	cardiac	mesoderm	that	permits	the	formation	of	liver,	while	preventing	ventral	pancreas	formation265.	The	similar	developmental	origin	of	the	pancreas	and	the	liver	makes	the	unintended	generation	of	liver	cells	during	hESC	differentiations	towards	pancreas	likely.	However,	finding	liver	cells	downstream	of	NEUROG3	is	surprising.	It	is	possible	that,	like	in	the	mouse,	a	small	population	of	hESC-derived	pancreatic	endoderm	cells	have	low	transcription	of	NEUROG3	that	is	not	sufficient	to	induce	the	endocrine	lineage.	Using	a	NEUROG3	lineage	tracing	hESC	line,	it	would	be	interesting	to	investigate	the	plasticity	of	cells	that	activate	NEUROG3	transcription.	Our	data	from	Chapter	6	suggests	that	NEUROG3	protein	in	hESC	differentiations	is	hyperphosphorylated,	likely	resulting	in	rapid	degradation.	Efforts	to	stabilize	NEUROG3	protein	may	prevent	the	unintended	formation	of	other	endodermal	cell	types,	including	liver	cells.		 In	this	chapter,	the	single	cell	transcriptome	of	mouse	pancreatic	progenitors,	endocrine	progenitors,	and	endocrine	cells	at	E15.5	and	E18.5	as	well	as	NEUROG3-expressing	cells	derived	from	hESCs	was	characterized.	These	data	are	a	resource	for	 132 developmental	biologists	interested	in	studying	heterogeneity	in	the	developing	mouse	pancreas	and	for	stem	cell	researchers	aiming	to	improve	the	current	differentiation	protocols	for	generating	b-like	cells.  133 Chapter	8: Conclusions	8.1 Research	summary	Owing	to	the	central	role	of	insulin	producing	b-cells	in	glucose	metabolism	and	diabetes	pathogenesis,	research	efforts	have	been	focused	on	understanding	how	b-cells	form	during	embryonic	development.	These	studies	have	provided	insight	into	methods	of	generating	unlimited	sources	of	b-cells	for	cellular	replacement	therapies,	including	the	differentiation	of	hESCs266	and	transdifferentiation	of	alternative	cell	types267.		 Critical	to	the	regenerative	medicine	efforts	for	the	treatment	of	diabetes	is	understanding	how	Neurog3+	endocrine	progenitors	form.	There	are	several	lines	of	evidence	for	the	importance	of	Neurog3	in	the	formation	of	endocrine	cells.	Studies	in	mice	determined	that	Neurog3+	cells	are	endocrine	progenitors	and	are	required	for	the	formation	of	endocrine	cells49,50.	The	adoption	of	the	endocrine	cell	fate	requires	high	expression	of	Neurog349,52	and,	without	this,	the	cells	adopt	an	acinar	or	ductal	fate50,53,56,57.	Interestingly,	studies	on	b-cell	failure	in	type	2	diabetes	suggest	that	b-cells	can	“dedifferentiate”	back	into	a	Neurog3-expressing	progenitor	state268.	Together,	these	studies	highlight	the	importance	of	understanding	how	Neurog3	is	regulated,	not	only	for	improving	b-cell	differentiation	protocols	but	also	for	understanding	b-cell	plasticity	during	diabetes	pathogenesis.		 During	development,	proliferating	progenitor	cells	transition	to	a	differentiated,	cell	cycle	arrested	fate.	This	change	from	proliferation	to	cell	cycle	exit	suggests	that	the	cell	cycle	itself	may	be	involved	in	the	differentiation	of	progenitor	cells.	An	example	of	this	comes	from	neurogenesis	where	the	Neurogenin	family	member,	Neurog2,	is	phosphorylated	by	cell	cycle	proteins	Cdk1/2,	resulting	in	ubiquitin-mediated	proteolysis	 134 of	Neurog2122.	Thus,	Cdks	maintain	the	progenitor	pool	by	preventing	Neurog2-mediated	differentiation.	In	this	thesis,	the	role	of	the	cell	cycle	in	regulating	mouse	and	human	pancreas	development	was	investigated.	There	were	four	objectives	to	address	this	hypothesis:	1. To	characterize	cell	cycle	length	during	mouse	development.	2. To	determine	if	cell	cycle	lengthening	causes	endocrine	cell	differentiation.	3. To	identify	the	mechanistic	link	between	cell	cycle	length	and	endocrine	cell	formation.	4. To	profile	the	heterogeneity	of	mouse	and	hESC-derived	endocrine	progenitor	cells	using	single	cell	transcriptomics.		 To	address	these	questions,	I	first	asked	whether	there	is	a	change	in	cell	cycle	length	during	pancreas	development.	This	was	accomplished	by	using	cumulative	EdU	labelling.	From	this,	a	specific	increase	in	G1-phase	length	from	E11.5	to	E13.5	was	found.	In	addition,	there	was	a	spatial	difference	in	the	length	of	G1	such	that	tip	progenitor	cells	spend	less	time	in	G1	than	the	trunk	progenitors.	Whether	this	difference	in	G1	length	is	responsible	for	their	altered	differentiation	potential	remains	to	be	determined.	A	population	of	epithelial	cells	that	do	not	transit	from	G1	to	S	was	identified,	the	significance	of	which	is	unknown	as	most	Neurog3+	cells	arise	from	cells	that	have	recently	divided.	This	is	consistent	with	reports	using	live	cell	imaging	that	found	~70%	of	Neurog3+	cells	arise	from	progenitors	that	have	recently	undergone	cell	division157.	This	work	adds	the	pancreas	to	the	growing	list	of	cell	types	whose	differentiation	is	correlated	with	G1	lengthening	during	embryogenesis.	 135 The	“cell	cycle	length	hypothesis”	suggests	that	increasing	cell	cycle	length	provides	the	time	required	to	accumulate	enough	of	a	cell	fate	determinant	to	cause	differentiation.	In	the	context	of	the	endocrine	cell	differentiation,	this	cell	fate	determinant	is	Neurog3.	To	investigate	the	consequences	of	altering	cell	cycle	length,	two	mouse	models	were	used.	Increasing	the	time	pancreatic	progenitors	spend	in	G1	resulted	in	more	Neurog3+	endocrine	progenitors	that	then	differentiated	to	a-	and	b-cells.	Interestingly,	increasing	G1	length	resulted	in	more	cells	of	the	endocrine	lineage	that	is	consistent	with	the	progenitors	developmental	competency52.	Thus,	cell	cycle	lengthening	alone	cannot	alter	progenitor	cell	specification.	Combined	with	studies	reported	in	Chapter	3,	these	in	vivo	studies	suggest	that	G1	lengthening	permits	efficient	Neurog3	expression	and	endocrine	differentiation.	Further,	ectopic	expression	of	Cdkn1b,	a	G1-S	cyclin-dependent	kinase	inhibitor,	suggests	that	the	activities	of	G1-S	Cdks	in	the	progenitor	cells	prevents	efficient	endocrine	differentiation	and	maintains	progenitor	cell	state.	The	presence	of	multiple	forms	of	Neurog3	in	mouse	pancreatic	progenitors	suggests	that	phosphorylation	of	Neurog3	by	cyclin-dependent	kinases	is	the	mechanistic	link	between	G1	lengthening	and	Neurog3	stabilization.	This	is	supported	by	the	increased	number	of	Neurog3+	cells	in	embryos	treated	with	Cdk2/4/6	inhibitors.		 Using	data	from	Chapters	3	and	4	I	provide	the	following	model	for	how	the	cell	cycle	regulates	pancreatic	progenitor	cell	differentiation	(Figure	55).	In	proliferating	progenitor	cells,	a	relatively	short	time	is	spent	in	G1,	resulting	in	high	activity	of	G1-S	Cdks.	These	kinases	phosphorylate	Neurog3,	presumably	resulting	in	the	degradation	of	Neurog3	and	the	maintenance	of	the	progenitor	cell	fate.	During	development,	there	is	an	endogenous	lengthening	of	the	G1-phase,	leading	to	lowered	activity	of	Cdks.	This	prevents	 136 the	hyperphosphorylation	of	Neurog3,	allowing	it	to	drive	endocrine	cell	differentiation.	Together,	this	suggests	that	the	cell	cycle	plays	an	important	role	in	pancreas	development	by	directly	regulating	expression	of	Neurog3190,269,270.	  Figure	55:	Model	for	cell	cycle	regulation	of	Neurog3	via	Cdk	phosphorylation	In	rapidly	dividing	pancreatic	progenitor	cells,	there	is	high	activity	of	Cdks.	These	kinases	phosphorylate	Neurog3,	resulting	in	its	degradation.	During	development,	there	is	a	lengthening	of	G1-phase	from	4.5	hours	at	E11.5	to	7.2	hours	at	E13.5.	This	change	in	cell	cycle	length	would	result	in	less	activity	of	Cdks,	reduced	phosphorylation	of	Neurog3,	and	differentiation	to	the	endocrine	cell	fate.		 In	support	of	this	model,	two	recent	studies	have	identified	that	Neurog3	is	phosphorylated	by	cyclin-dependent	kinases190,270.	Using	HeLa	cells,	a	series	of	FLAG-tagged	human	NEUROG3	mutants	identified	that	the	phosphorylation	sites	are	located	between	amino	acids	145-187	towards	the	C-terminal	end	of	the	protein190.	In	this	region,	 137 there	are	nine	serine	residues	that	could	be	targeted	by	the	serine-threonine	cyclin-dependent	kinases.	To	investigate	which	serines	are	phosphorylated,	a	series	of	human	NEUROG3	mutants	with	one,	seven	or	eight	of	the	nine	serine	residues	in	this	region	were	mutated	to	alanine.	These	experiments	identified	that	phosphorylation	on	serine	183	(S183)	leads	to	sequential	hyperphosphorylation	of	NEUROG3190.	In	addition,	Azzarelli	et	al.270	mutated	six	serine	residues	previously	implicated	in	Neurog3	stability271	and	found	that	mutating	all	six	residues	(6S-A	Neurog3)	in	Xenopus	embryos	prevents	Neurog3	phosphorylation270.	Using	a	Neurog3	promoter	to	drive	expression	of	either	wildtype	Neurog3	protein	as	a	control	or	6S-A	Neurog3	protein	during	mouse	development	is	sufficient	to	increase	expression	of	Gcg+	and	Sst+	cells	at	E16.5270.	Having	identified	cell	cycle	length	as	a	regulator	of	endocrine	differentiation,	I	next	set	out	to	profile	the	single	cell	transcriptome	of	mouse	and	human	endocrine	progenitors	in	an	effort	to	identify	other	potential	regulators	of	endocrine	differentiation.	From	this,	I	have	generated	a	database	of	markers	that	are	specific	for	pancreatic	mesenchyme,	macrophages,	acinar,	duct,	bipotent	trunk,	endocrine	progenitor,	and	endocrine	cells	at	both	E15.5	and	E18.5.	Using	lineage	tracing,	I	was	also	able	to	isolate	newly	specified	endocrine	cells	from	more	mature	hormone-expressing	endocrine	cells.	In	addition,	the	transcriptome	of	hESC-derived	endocrine	cells	was	characterized,	data	that	can	be	used	to	identify	potential	ways	to	improve	current	hESC	differentiation	protocols.	8.2 Future	directions	While	I	have	measured	a	change	in	cell	cycle	length	during	early	mouse	pancreas	development,	I	was	unable	to	investigate	beyond	E13.5	as	there	are	no	specific	markers	of	progenitor	cells	past	this	developmental	time	point.	In	addition,	cumulative	EdU	labelling	 138 only	allows	for	the	measurement	of	cell	cycle	lengths	at	a	population	level.	As	there	are	differences	between	tip	and	trunk	progenitors,	it	is	probable	that	cell	cycle	length	is	heterogeneous	even	within	a	progenitor	population.	It	would	be	interesting	to	calculate	cell	cycle	lengths	at	a	single	cell	level	using	Pdx1-Cre;	R26Fucci2aR,	a	mouse	model	that	would	allow	for	bicistronic	expression	of	the	FUCCI	cell	cycle	indicator	in	pancreatic	cells272.	Using	this	model	in	combination	with	the	time	lapse	imaging	and	manual	tracking	of	Neurog3+	cells157,	the	cell	cycle	lengths	of	individual	progenitor	cells	and	the	identity	of	their	progeny	could	be	determined.	Based	on	our	findings	and	others157,190,270,	I	hypothesize	that	those	progenitors	with	longer	G1	length	will	give	rise	to	Neurog3+	daughter	cells,	while	those	progenitors	with	a	short	G1	length	will	give	rise	to	two	Pdx1+	daughter	cells.		 How	cell	cycle	lengthening	occurs	during	development	remains	unanswered.	It	is	likely	caused	by	increased	expression	of	cell	cycle	inhibitors;	however,	what	drives	upregulation	of	cell	cycle	inhibitors	as	development	proceeds	is	unknown.	Future	experiments	using	scRNA-seq	of	E12.5	pancreatic	progenitors	may	provide	some	insights	into	this	question.	In	addition,	this	study	should	identify	the	~30%	of	epithelium	that	is	EdU-negative	in	the	cumulative	labelling	studies.	Analysis	of	this	population	of	cells	at	the	single	cell	level	may	offer	the	identity	of	these	cells,	allowing	for	further	investigation	into	their	role	in	pancreas	development.		 Data	highlighting	the	role	of	Cdks	in	the	phosphorylation	of	Neurog3	was	generated	using	pharmacological	inhibitors.	Even	though	these	inhibitors	were	used	at	a	moderate	concentration,	it	is	possible	that	there	were	off-target	effects	and	is	a	limitation	of	our	experiment.	However,	the	same	effect	was	measured	by	ectopic	expression	of	Cdkn1b,	supporting	the	role	of	Cdks	in	the	regulation	of	Neurog3.	It	would	be	interesting	to	 139 generate	hESC	mutant	lines	that	have	the	potential	phosphorylation	sites	of	NEUROG3	mutated	to	alanines.	As	most	NEUROG3	protein	in	hESC	differentiations	is	hyperphosphorylated,	this	phosphomutant	line	should	result	in	increased	endocrine	cell	formation.	Most	studies	evaluating	the	efficiency	of	hESC	differentiations	have	used	gene	expression	or	immunopositivity	as	a	measure	of	b-like	cells.	However,	my	findings	suggest	that	the	post-translational	modification	of	pro-endocrine	proteins	can	affect	their	function.	In	addition,	scRNA-seq	data	suggests	that	most	INS+	cells	do	not	have	insulin	processing	enzymes	or	other	proteins	required	for	glucose	stimulated	insulin	secretion.	Thus,	this	thesis	supports	the	call	for	deeper	phenotyping	of	hESC	derived	b-like	cells273.		 Single	cell	transcriptome	profiling	of	endocrine	progenitor	cells	revealed	two	potential	genes,	Gadd45a	and	Btg2,	that	are	implicated	in	regulating	the	cell	cycle.	Follow	up	studies	should	be	aimed	at	investigating	the	role	of	these	proteins	in	cell	cycle	lengthening,	interaction	with	Id	proteins	to	indirectly	regulate	Neurog3,	and	DNA	demethylation	of	genes	that	drive	endocrine	cell	formation.	In	addition,	further	study	of	genes	expressed	in	the	trunk	progenitor	cluster	in	E15.5	embryos	might	yield	discovery	of	novel	genes	involved	in	the	cell	fate	decisions	between	ductal	and	endocrine	cells.		 While	reporter	mouse	and	hESC	lines	were	used	to	enrich	for	populations	of	interest	before	scRNA-seq,	there	are	limitations	to	this	model.	Analysis	of	the	NEUROG3-2A-eGFP	reporter	line	indicates	that	GFP	protein	remains	on	after	Neurog3	transcription	is	turned	off.	A	fusion	reporter	line	would	make	GFP	expression	more	specific	for	the	NEUROG3+	cells	but	could	potentially	alter	NEUROG3	stability.	Another	approach	is	to	use	a	lineage-tracing	approach	in	the	hESCs.	This	would	allow	for	a	more	direct	comparison	of	mouse	 140 and	human	Neurog3+	endocrine	progenitor	cells	and	their	downstream	descendants,	information	that	could	be	used	to	improve	current	hESC	differentiation	protocols.	8.3 Final	thoughts	The	data	presented	in	this	thesis	provide	mechanistic	evidence	for	how	the	cell	cycle	regulates	mouse	and	human	pancreas	development.	It	also	provides	single	cell	transcriptome	data	for	both	murine	and	hESC-derived	endocrine	progenitor	cells.	Together,	this	work	offers	insight	into	potential	methods	of	improving	regenerative	medicine	approaches	for	the	treatment	of	diabetes.		 141 References		1.	 Federation,	I.	D.	IDF	Diabetes	Atlas,	Seventh	edition.	(2015).	2.	 Fowler,	M.	J.	Microvascular	and	Macrovascular	Complications	of	Diabetes.	Clinical	Diabetes	29,	116–122	(2011).	3.	 Laios,	K.,	Karamanou,	M.,	Saridaki,	Z.	&	Androutsos,	G.	Aretaeus	of	Cappadocia	and	the	first	description	of	diabetes.	Hormones	(Athens)	11,	109–113	(2012).	4.	 Mering,	von,	J.	&	Minkowski,	O.	Diabetes	mellitus	nach	Pankreasexstirpation.	(1890).	5.	 Sakula,	A.	Paul	Langerhans	(1847-1888):	a	centenary	tribute.	Journal	of	the	Royal	Society	of	Medicine	81,	414–415	(1988).	6.	 Stylianou,	C.	&	Kelnar,	C.	The	introduction	of	successful	treatment	of	diabetes	mellitus	with	insulin.	Journal	of	the	Royal	Society	of	Medicine	102,	298–303	(2009).	7.	 Loriaux,	D.	L.	Moses	Barron	(1883–1974).	A	Biographical	History	of	Endocrinology	(2016).	8.	 Banting,	F.	G.,	Best,	C.	H.,	Collip,	J.	B.,	Campbell,	W.	R.	&	Fletcher,	A.	A.	Pancreatic	Extracts	in	the	Treatment	of	Diabetes	Mellitus.	Canadian	Medical	Association	Journal	12,	141–146	(1922).	9.	 Banting,	F.	&	Best,	C.	H.	The	internal	secretion	of	the	pancreas.	The	Journal	of	Laboratory	and	Clinical	Medicine	7,	251–266	(1922).	10.	 Kelly,	W.	D.,	Lillehei,	R.	C.,	Merkel,	F.	K.,	Idezuki,	Y.	&	Goetz,	F.	C.	Allotransplantation	of	the	pancreas	and	duodenum	along	with	the	kidney	in	diabetic	nephropathy.	Surgery	61,	827–837	(1967).	11.	 Association,	A.	D.	Pancreas	and	islet	transplantation	in	type	1	diabetes.	Diabetes	Care	(2006).	12.	 Troppmann,	C.	Complications	after	pancreas	transplantation.	Current	Opinion	in	Organ	Transplantation	15,	112–118	(2010).	13.	 Lacy,	P.	E.	&	Kostianovsky,	M.	Method	for	the	isolation	of	intact	islets	of	Langerhans	from	the	rat	pancreas.	Diabetes	16,	35–39	(1967).	14.	 Ballinger,	W.	F.	&	Lacy,	P.	E.	Transplantation	of	intact	pancreatic	islets	in	rats.	Surgery	(1972).	15.	 Najarian,	J.	S.,	Sutherland,	D.	E.	R.,	Baumgartner,	D.,	Burke,	B.,	Rynasiewicz,	J.	J.,	Matas,	A.J.	&	Goetz,	F.C.	Total	or	near	total	pancreatectomy	and	islet	autotransplantation	for	treatment	of	chronic	pancreatitis.	Ann.	Surg.	192,	526–542	(1980).	16.	 Scharp,	D.	W.,	Lacy,	P.	E.,	Santiago,	J.	V.,	McCullough,	C.	S.,	Weide,	L.	G.,	Falqui,	L.,	Marchetti,	P.,	Gingerich,	R.	L.,	Jaffe,	A.	S.,	Cryer,	P.	E.,	Anderson,	C.	B.	&	Flye,	M.	W.	Insulin	Independence	After	Islet	Transplantation	Into	Type	I	Diabetic	Patient.	Diabetes	39,	515–518	(1990).	17.	 Warnock,	G.,	Kneteman,	N.,	Ryan,	E.,	Rabinovitch,	A.	&	Rajotte,	R.	Long-term	follow-up	after	transplantation	of	insulin-producing	pancreatic	islets	into	patients	with	type	1	(insulin-dependent)	diabetes	mellitus.	Diabetologia	35,	89–95	(1992).	18.	 Gores,	P.	F.,	Najarian,	J.	S.,	Stephanian,	E.	&	Lloveras,	J.	J.	Insulin	independence	in	type	I	diabetes	after	transplantation	of	unpurified	islets	from	single	donor	with	15-deoxyspergualin.	The	Lancet	341,	19–21	(1993).	19.	 Alejandro,	R.,	Lehmann,	R.,	Ricordi,	C.,	Kenyon,	N.	S.,	Angelico,	M.	C.,	Burke,	G.,	Esquenazi,	V.,	Nery,	J.,	Betancourt,	A.	E.,	Kong,	S.	S.,	Miller,	J.	&	Mintz,	D.	H.	Long- 142 term	function	(6	years)	of	islet	allografts	in	type	1	diabetes.	Diabetes	46,	1983–1989	(1997).	20.	 Robertson,	R.	P.	Cyclosporin-Induced	Inhibition	of	Insulin	Secretion	in	Isolated	Rat	Islets	and	HIT	Cells.	Diabetes	35,	1016–1019	(1986).	21.	 Nielsen,	J.	H.,	Mandrup-Poulsen,	T.	&	Nerup,	J.	Direct	Effects	of	Cyclosporin	A	on	Human	Pancreatic	β-cells.	Diabetes	35,	1049–1052	(1986).	22.	 Gold,	G.,	Qian,	R.	L.	&	Grodsky,	G.	M.	Insulin	biosynthesis	in	HIT	cells.	Effects	of	glucose,	forskolin,	IBMX,	and	dexamethasone.	Diabetes	37,	160–165	(1988).	23.	 Johnson,	J.	D.,	Ao,	Z.,	Ao,	P.,	Li,	H.,	Dai,	L-J.,	He,	Z.,	Tee,	M.,	Potter,	K.	J.,	Klimek,	A.	M.,	Meloche,	R.,	M.,	Thompson,	D.	M.,	Verchere,	C.	B.	&	Warnock,	G.	L.	Different	effects	of	FK506,	rapamycin,	and	mycophenolate	mofetil	on	glucose-stimulated	insulin	release	and	apoptosis	in	human	islets.	Cell	Transplant	18,	833–845	(2009).	24.	 Shapiro,	A.,	Lakey,	J.	&	Ryan,	E.	Islet	transplantation	in	seven	patients	with	type	1	diabetes	mellitus	using	a	glucocorticoid-free	immunosuppressive	regimen.	New	England	Journal	of	Medicine	343,	230–238	(2000).	25.	 Ryan,	E.	A.,	Paty,	B.	W.,	Senior,	P.	A.,	Bigam,	D.,	Alfadhli,	E.,	Kneteman,	N.	M.,	Lakey,	J.	R.	T.	&	Shapiro,	A.	M.	J.	Five-year	follow-up	after	clinical	islet	transplantation.	Diabetes	54,	2060	(2005).	26.	 Wells,	J.	M.	&	Melton,	D.	A.	Early	mouse	endoderm	is	patterned	by	soluble	factors	from	adjacent	germ	layers.	Development	127,	1563–1572	(2000).	27.	 Brennan,	J.,	Lu,	C.	C.,	Norris,	D.	P.,	Rodriguez,	T.	A.,	Beddington,	R.	S.	P.	&	Robertson,	E.	J.	Nodal	signalling	in	the	epiblast	patterns	the	early	mouse	embryo.	Nature	411,	965–969	(2001).	28.	 Conlon,	F.,	Lyons,	K.,	Takaesu,	N.	&	Barth,	K.	A	primary	requirement	for	nodal	in	the	formation	and	maintenance	of	the	primitive	streak	in	the	mouse.	Development	120,	1919–1928	(1994).	29.	 Liu,	P.,	Wakamiya,	M.,	Shea,	M.	J.,	Albrecht,	U.,	Behringer,	R.	R.	&	Bradley,	A.	Requirement	for	Wnt3	in	vertebrate	axis	formation.	Nat	Genet	22,	361–365	(1999).	30.	 Kelly,	O.	G.,	Pinson,	K.	I.	&	Skarnes,	W.	C.	The	Wnt	co-receptors	Lrp5	and	Lrp6	are	essential	for	gastrulation	in	mice.	Development	131,	2803–2815	(2004).	31.	 Lowe,	L.	A.,	Yamada,	S.	&	Kuehn,	M.	R.	Genetic	dissection	of	nodal	function	in	patterning	the	mouse	embryo.	Development	128,	1831–1843	(2001).	32.	 Vincent,	S.	D.,	Dunn,	N.	R.,	Hayashi,	S.,	Norris,	D.	P.	&	Robertson,	E.	J.	Cell	fate	decisions	within	the	mouse	organizer	are	governed	by	graded	Nodal	signals.	Genes	&	Development	17,	1646–1662	(2003).	33.	 Martín,	M.,	Gallego-Llamas,	J.,	Ribes,	V.,	Kedinger,	M.,	Niederreither,	K.,	Chambon,	P.,	Dollé,	P.	&	Gradwohl,	G.	Dorsal	pancreas	agenesis	in	retinoic	acid-deficient	Raldh2	mutant	mice.	Developmental	Biology	284,	399–411	(2005).	34.	 Kim,	S.	K.,	Hebrok,	M.	&	Melton,	D.	A.	Notochord	to	endoderm	signaling	is	required	for	pancreas	development.	Development	124,	4243–4252	(1997).	35.	 Hebrok,	M.,	Kim,	S.	K.	&	Melton,	D.	A.	Notochord	repression	of	endodermal	Sonic	hedgehog	permits	pancreas	development.	Genes	&	Development	12,	1705–1713	(1998).	36.	 Cleaver,	O.	&	Krieg,	P.	A.	VEGF	mediates	angioblast	migration	during	development	of	the	dorsal	aorta	in	Xenopus.	Development	125,	3905–3914	(1998).	37.	 Lammert,	E.,	Cleaver,	O.	&	Melton,	D.	Induction	of	pancreatic	differentiation	by	 143 signals	from	blood	vessels.	Science	294,	564–567	(2001).	38.	 Jorgensen,	M.	C.,	Ahnfelt-Ronne,	J.,	Hald,	J.,	Madsen,	O.	D.,	Serup,	P.	&	Hecksher-Sorensen,	J.	An	Illustrated	Review	of	Early	Pancreas	Development	in	the	Mouse.	Endocrine	Reviews	28,	685–705	(2007).	39.	 Ahlgren,	U.	&	Jonsson,	J.	The	morphogenesis	of	the	pancreatic	mesenchyme	is	uncoupled	from	that	of	the	pancreatic	epithelium	in	IPF1/PDX1-deficient	mice.	Development	(1996).	40.	 Kesavan,	G.,	Sand,	F.	W.,	Greiner,	T.	U.,	Johansson,	J.	K.,	Kobberup,	S.,	Wu,	X.,	Brakebusch,	C.	&	Semb,	H.	Cdc42-mediated	tubulogenesis	controls	cell	specification.	Cell	139,	791–801	(2009).	41.	 Villasenor,	A.,	Chong,	D.	C.,	Henkemeyer,	M.	&	Cleaver,	O.	Epithelial	dynamics	of	pancreatic	branching	morphogenesis.	Development	137,	4295–4305	(2010).	42.	 Zhou,	Q.,	Law,	A.	C.,	Rajagopal,	J.,	Anderson,	W.	J.,	Gray,	P.	A.	&	Melton,	D.	A.	A	multipotent	progenitor	domain	guides	pancreatic	organogenesis.	Developmental	Cell	13,	103–114	(2007).	43.	 Stanger,	B.	Z.,	Tanaka,	A.	J.	&	Melton,	D.	A.	Organ	size	is	limited	by	the	number	of	embryonic	progenitor	cells	in	the	pancreas	but	not	the	liver.	Nature	445,	886–891	(2007).	44.	 Sommer,	L.,	Ma,	Q.	&	Anderson,	D.	J.	neurogenins,	a	novel	family	of	atonal-related	bHLH	transcription	factors,	are	putative	mammalian	neuronal	determination	genes	that	reveal	progenitor	cell	heterogeneity	in	the	developing	CNS	and	PNS.	Mol.	Cell.	Neurosci.	8,	221–241	(1996).	45.	 Apelqvist,	Å.,	Li,	H.,	Sommer,	L.,	Beatus,	P.,	Anderson,	D.	J.,	Honjo,	T.,	Hrabe	de	Angelis,	M.,	Lendahl,	U.	&	Edlund,	H.	Notch	signalling	controls	pancreatic	cell	differentiation.	Nature	400,	877–881	(1999).	46.	 Gradwohl,	G.,	Dierich,	A.,	LeMeur,	M.	&	Guillemot,	F.	neurogenin3	is	required	for	the	development	of	the	four	endocrine	cell	lineages	of	the	pancreas.	Proceedings	of	the	National	Academy	of	Sciences	97,	1607–1611	(2000).	47.	 Schwitzgebel,	V.	M.,	Scheel,	D.	W.,	Conners,	J.	R.,	Kalamaras,	J.,	Lee,	J.	E.,	Anderson,	D.	J.,	Sussel,	L.,	Johnson,	J.	D.	&	German,	M.	S.	Expression	of	neurogenin3	reveals	an	islet	cell	precursor	population	in	the	pancreas.	Development	127,	3533–3542	(2000).	48.	 Jensen,	J.,	Heller,	R.	S.,	Funder-Nielsen,	T.,	Pedersen,	E.	E.,	Lindsell,	C.,	Weinmaster,	G.,	Madsen,	O.	D.	&	Serup,	P.	Independent	development	of	pancreatic	alpha-and	beta-cells	from	neurogenin3-expressing	precursors:	a	role	for	the	notch	pathway	in	repression	of	premature	differentiation.	Diabetes	49,	163–176	(2000).	49.	 Gu,	G.,	Dubauskaite,	J.	&	Melton,	D.	A.	Direct	evidence	for	the	pancreatic	lineage:	NGN3+	cells	are	islet	progenitors	and	are	distinct	from	duct	progenitors.	Development	(2002).	50.	 Schonhoff,	S.	E.,	Giel-Moloney,	M.	&	Leiter,	A.	B.	Neurogenin	3-expressing	progenitor	cells	in	the	gastrointestinal	tract	differentiate	into	both	endocrine	and	non-endocrine	cell	types.	Developmental	Biology	270,	443–454	(2004).	51.	 Desgraz,	R.	&	Herrera,	P.	L.	Pancreatic	neurogenin	3-expressing	cells	are	unipotent	islet	precursors.	Development	136,	3567–3574	(2009).	52.	 Johansson,	K.	A.,	Dursun,	U.,	Jordan,	N.,	Gu,	G.,	Beermann,	F.,	Gradwohl,	G.	&	Grapin-Botton,	A.	Temporal	Control	of	Neurogenin3	Activity	in	Pancreas	Progenitors	 144 Reveals	Competence	Windows	for	the	Generation	of	Different	Endocrine	Cell	Types.	Developmental	Cell	12,	457–465	(2007).	53.	 Beucher,	A.,	Martin,	M.,	Spenle,	C.,	Poulet,	M.,	Collin,	C.	&	Gradwohl,	G.	Competence	of	failed	endocrine	progenitors	to	give	rise	to	acinar	but	not	ductal	cells	is	restricted	to	early	pancreas	development.	Developmental	Biology	361,	277–285	(2012).	54.	 Bankaitis,	E.	D.,	Bechard,	M.	E.	&	Wright,	C.	V.	E.	Feedback	control	of	growth,	differentiation,	and	morphogenesis	of	pancreatic	endocrine	progenitors	in	an	epithelial	plexus	niche.	Genes	&	Development	29,	2203–2216	(2015).	55.	 Grapin-Botton,	A.,	Majithia,	A.	R.	&	Melton,	D.	A.	Key	events	of	pancreas	formation	are	triggered	in	gut	endoderm	by	ectopic	expression	of	pancreatic	regulatory	genes.	Genes	&	Development	15,	444–454	(2001).	56.	 Magenheim,	J.,	Klein,	A.	M.,	Stanger,	B.	Z.,	Ashery-Padan,	R.,	Sosa-Pineda,	B,	Gu,	G.	&	Dor,	Y.	Ngn3+	endocrine	progenitor	cells	control	the	fate	and	morphogenesis	of	pancreatic	ductal	epithelium.	Developmental	Biology	359,	26–36	(2011).	57.	 Wang,	S.,	Yan,	J.,	Anderson,	D.	A.,	Xu,	Y.,	Kanal,	M.	C.,	Cao,	Z.,	Wright,	C.	V.	E.	&	Gu,	G.	Neurog3	gene	dosage	regulates	allocation	of	endocrine	and	exocrine	cell	fates	in	the	developing	mouse	pancreas.	Developmental	Biology	339,	26–37	(2010).	58.	 McKnight,	K.	D.,	Wang,	P.	&	Kim,	S.	K.	Deconstructing	Pancreas	Development	to	Reconstruct	Human	Islets	from	Pluripotent	Stem	Cells.	Cell	Stem	Cell	6,	300–308	(2010).	59.	 Stoffers,	D.	A.,	Ferrer,	J.,	Clarke,	W.	L.	&	Habener,	J.	F.	Early-onset	type-ll	diabetes	mellitus	(MODY4)	linked	to	IPF1.	Nat	Genet	17,	138–139	(1997).	60.	 Stoffers,	D.	A.,	Zinkin,	N.	T.,	Stanojevic,	V.,	Clarke,	W.	L.	&	Habener,	J.	F.	Pancreatic	agenesis	attributable	to	a	single	nucleotide	deletion	in	the	human	IPF1	gene	coding	sequence.	Nat	Genet	15,	106–110	(2004).	61.	 Nicolino,	M.,	Claiborn,	K.	C.,	Senée,	V.,	Boland,	A.,	Stoffers,	D.	A.	&	Julier,	C.	A	novel	hypomorphic	PDX1	mutation	responsible	for	permanent	neonatal	diabetes	with	subclinical	exocrine	deficiency.	Diabetes	59,	733–740	(2010).	62.	 Jennings,	R.	E.,	Berry,	A.	A.,	Kirkwood-Wilson,	R.,	Roberts,	N.	A.,	Hearn,	T.,	Salisbury,	R.	J.,	Blaylock,	J.,	Hanley,	K.	P.	&	Hanley,	N.	A.	Development	of	the	human	pancreas	from	foregut	to	endocrine	commitment.	Diabetes	62,	3514–3522	(2013).	63.	 Thomson,	J.	A.,	Itskovitz-Eldor,	J.,	Shapiro,	S.	S.,	Waknitz,	M.	A.,	Swiergiel,	J.	J.,	Marshall,	V.	S.	&	Jones,	J.	M.	Embryonic	stem	cell	lines	derived	from	human	blastocysts.	Science	282,	1145–1147	(1998).	64.	 D'Amour,	K.	A.,	Bang,	A.	G.,	Eliazer,	S.,	Kelly,	O.	G.,	Agulnick,	A.	D.,	Smart,	N.	G.,	Moorman,	M.	A.,	Kroon,	E.,	Carpenter,	M.	K.	&	Baetge,	E.	E.	Production	of	pancreatic	hormone–expressing	endocrine	cells	from	human	embryonic	stem	cells.	Nat	Biotechnol	24,	1392–1401	(2006).	65.	 D'Amour,	K.	A.,	Agulnick,	A.	D.,	Eliazer,	S.,	Kelly,	O.	G.,	Kroon,	E.	&	Baetge,	E.	E.	Efficient	differentiation	of	human	embryonic	stem	cells	to	definitive	endoderm.	Nat	Biotechnol	23,	1534–1541	(2005).	66.	 Nostro,	M.	C.,	Sarangi,	F.,	Ogawa,	S.,	Holtzinger,	A.,	Corneo,	B.,	Li,	X.,	Micallef,	S.	J.m	Park,	I-H.,	Basford,	C.,	Wheeler,	M.	B.,	Daley,	G.	Q.,	Elefanty,	A.	G.,	Stanley,	E.	G.	&	Keller,	G.	Stage-specific	signaling	through	TGFβ	family	members	and	WNT	regulates	patterning	and	pancreatic	specification	of	human	pluripotent	stem	cells.	 145 Development	138,	1445–1445	(2011).	67.	 Rezania,	A.,	Riedel,	M.	J.,	Wideman,	R.	D.,	Karanu,	F.,	Ao,	Z.,	Warnock,	G.	L.	&	Kieffer,	T.	J.	Production	of	functional	glucagon-secreting	α-cells	from	human	embryonic	stem	cells.	Diabetes	60,	239–247	(2011).	68.	 Kroon,	E.,	Martinson,	L.	A.,	Kadoya,	K.,	Bang,	A.	G.,	Kelly,	O.	G.,	Eliazer,	S.,	Young,	H.,	Richardson,	M.,	Smart,	N.	G.,	Cunningham,	J.,	Agulnick,	A.	D.,	D'Amour,	K.	A.,	Carpenter,	M.	K.	&	Baetge,	E.	E.	Pancreatic	endoderm	derived	from	human	embryonic	stem	cells	generates	glucose-responsive	insulin-secreting	cells	in	vivo.	Nat	Biotechnol	26,	443–452	(2008).	69.	 Rezania,	A.,	Bruin,	J.	E.,	Arora,	P.,	Rubin,	A.,	Batushansky,	I.,	Asadi,	A.,	O'Dwyer,	S.,	Quiskamp,	N.,	Mojiban,	M.,	Albrecht,	T.,	Yang,	Y.	H.	C.,	Johnson,	J.	D.	&	Kieffer,	T.	J.	Reversal	of	diabetes	with	insulin-producing	cells	derived	in	vitro	from	human	pluripotent	stem	cells.	Nat	Biotechnol	32,	1121–1133	(2014).	70.	 Pagliuca,	F.	W.,	Millman,	J.	R.,	Gurtler,	M.,	Segel,	M.,	Van	Dervort,	A.,	Ryu,	J.	H.,	Peterson,	Q.	P.,	Greiner,	D.	&	Melton,	D.	A.	Generation	of	functional	human	pancreatic	β	cells	in	vitro.	Cell	159,	428–439	(2014).	71.	 Eiges,	R.,	Schuldiner,	M.,	Drukker,	M.,	Yanuka,	O.,	Itskovitz-Eldor,	J.	&	Benvenisty,	N.	Establishment	of	human	embryonic	stem	cell-transfected	clones	carrying	a	marker	for	undifferentiated	cells.	Current	Biology	11,	514–518	(2001).	72.	 Vallier,	L.,	Rugg-Gunn,	P.	J.,	Bouhon,	I.	A.,	Andersson,	F.	K.,	Sadler,	A.	J.	&	Pedersen,	R.	A.	Enhancing	and	diminishing	gene	function	in	human	embryonic	stem	cells.	Stem	Cells	22,	2–11	(2004).	73.	 Ren,	C.	P.,	Zhao,	M.,	Shan,	W-J.,	Yang,	X-Y.,	Yin,	Z-H.,	Jiang,	X-J.,	Zhang,	H-B.	&	Yao,	K-T.	Establishment	of	human	embryonic	stem	cell	line	stably	expressing	Epstein-Barr	virus-encoded	nuclear	antigen	1.	Acta	Biochim.	Biophys.	Sin.	(Shanghai)	37,	68–73	(2005).	74.	 Liu,	Y.-P.,	Dovzhenko,	O.	V.,	Garthwaite,	M.	A.,	Dambaeva,	S.	V.,	Durning,	M.,	Pollastrini,	L.	M.	&	Golos,	T.	G.	Maintenance	of	pluripotency	in	human	embryonic	stem	cells	stably	over-expressing	enhanced	green	fluorescent	protein.	Stem	Cells	and	Development	13,	636–645	(2004).	75.	 Siemen,	H.,	Hohenstein	Elliott,	K.	A.,	Lock,	L.	F.	&	Donovan,	P.	J.	Nucleofection	of	human	embryonic	stem	cells.	Stem	Cells	and	Development	14,	378–383	(2005).	76.	 Lakshmipathy,	U.,	Pelacho,	B.,	Sudo,	K.,	Linehan,	J.	L.,	Coucouvanis,	E.,	Kaufman,	D.	S.	&	Verfaillie,	C.	M.	Efficient	transfection	of	embryonic	and	adult	stem	cells.	Stem	Cells	22,	531–543	(2004).	77.	 Thyagarajan,	B.,	Liu,	Y.,	Shin,	S.,	Lakshmipathy,	U.,	Scheyhing,	K.,	Xue,	H.,	Ellerstrom,	C.,	Strehl,	R.,	Hyllner,	J.,	Rao,	M.	S.	&	Chesnut,	J.	D.	Creation	of	Engineered	Human	Embryonic	Stem	Cell	Lines	Using	phiC31	Integrase.	Stem	Cells	26,	119–126	(2008).	78.	 Gerrard,	L.,	Zhao,	D.,	Clark,	A.	J.	&	Cui,	W.	Stably	Transfected	Human	Embryonic	Stem	Cell	Clones	Express	OCT4-Specific	Green	Fluorescent	Protein	and	Maintain	Self-Renewal	and	Pluripotency.	Stem	Cells	23,	124–133	(2005).	79.	 MacArthur,	C.	C.,	Xue,	H.,	Van	Hoof,	D.,	Lieu,	P.	T.,	Dudas,	M.,	Fontes,	A.,	Swistowski,	A.,	Touboul,	T.,	Seerke,	R.,	Laurent,	L.	C.,	Loring,	J.	F.,	German,	M.	S.,	Zeng,	X.,	Rao,	M.	S.,	Lakshmipathy,	U.,	Chesnut,	J.	D.	&	Liu,	Y.	Chromatin	Insulator	Elements	Block	Transgene	Silencing	in	Engineered	Human	Embryonic	Stem	Cell	Lines	at	a	Defined	Chromosome	13	Locus.	Stem	Cells	and	Development	21,	191–205	(2012).	 146 80.	 Leavitt,	A.	D.	&	Hamlett,	I.	Homologous	Recombination	in	Human	Embryonic	Stem	Cells:	A	Tool	for	Advancing	Cell	Therapy	and	Understanding	and	Treating	Human	Disease.	Clinical	and	Translational	Science	4,	298–305	(2011).	81.	 Peters,	D.	T.,	Cowan,	C.	A.	&	Musunuru,	K.	Genome	editing	in	human	pluripotent	stem	cells.	StemBook,	ed.	(2013).	82.	 Wei,	C.,	Liu,	J.,	Yu,	Z.,	Zhang,	B.,	Gao,	G.	&	Jiao,	R.	TALEN	or	Cas9	-	Rapid,	Efficient	and	Specific	Choices	for	Genome	Modifications.	Journal	of	Genetics	and	Genomics	40,	281–289	(2013).	83.	 Gaj,	T.,	Gersbach,	C.	A.	&	Barbas,	C.	F.,	III.	ZFN,	TALEN,	and	CRISPR/Cas-based	methods	for	genome	engineering.	Trends	in	Biotechnology	31,	397–405	(2013).	84.	 Boch,	J.	&	Bonas,	U.	Xanthomonas	AvrBs3	Family-Type	III	Effectors:	Discovery	and	Function.	Annu.	Rev.	Phytopathol.	48,	419–436	(2010).	85.	 Boch,	J.,	Scholze,	H.,	Schornack,	S.,	Landgraf,	A.,	Hahn,	S.,	Kay,	S.,	Lahaye,	T.,	Nickstadt,	A.	&	Bonas,	U.	Breaking	the	Code	of	DNA	Binding	Specificity	of	TAL-Type	III	Effectors.	Science	326,	1509–1512	(2009).	86.	 Ishino,	Y.,	Shinagawa,	H.,	Makino,	K.,	Amemura,	M.	&	Nakata,	A.	Nucleotide	sequence	of	the	iap	gene,	responsible	for	alkaline	phosphatase	isozyme	conversion	in	Escherichia	coli,	and	identification	of	the	gene	product.	Journal	of	Bacteriology	169,	5429–5433	(1987).	87.	 Jansen,	R.,	Embden,	J.	D.	A.	V.,	Gaastra,	W.	&	Schouls,	L.	M.	Identification	of	genes	that	are	associated	with	DNA	repeats	in	prokaryotes.	Molecular	Microbiology	43,	1565–1575	(2002).	88.	 Horvath,	P.	&	Barrangou,	R.	CRISPR/Cas,	the	immune	system	of	bacteria	and	archaea.	Science	327,	167–170	(2010).	89.	 Makarova,	K.	S.,	Aravind,	L.,	Grishin,	N.	V.,	Rogozin,	I.	B.	&	Koonin,	E.	V.	A	DNA	repair	system	specific	for	thermophilic	Archaea	and	bacteria	predicted	by	genomic	context	analysis.	Nucleic	Acids	Research	30,	482–496	(2002).	90.	 Bolotin,	A.,	Quinquis,	B.,	Sorokin,	A.	&	Ehrlich,	S.	D.	Clustered	regularly	interspaced	short	palindrome	repeats	(CRISPRs)	have	spacers	of	extrachromosomal	origin.	Microbiology	151,	2551–2561	(2005).	91.	 Pourcel,	C.,	Salvignol,	G.	&	Vergnaud,	G.	CRISPR	elements	in	Yersinia	pestis	acquire	new	repeats	by	preferential	uptake	of	bacteriophage	DNA,	and	provide	additional	tools	for	evolutionary	studies.	Microbiology	151,	653–663	(2005).	92.	 Mojica,	F.	J.	M.,	Díez-Villaseñor,	C.,	García-Martínez,	J.	&	Soria,	E.	Intervening	sequences	of	regularly	spaced	prokaryotic	repeats	derive	from	foreign	genetic	elements.	J.	Mol.	Evol.	60,	174–182	(2005).	93.	 Barrangou,	R.,	Fremaux,	C.,	Deveau,	H.,	Richards,	M.,	Boyaval,	P.,	Moineau,	S.,	Romero,	D.	A.	&	Horwath,	P.	CRISPR	Provides	Acquired	Resistance	Against	Viruses	in	Prokaryotes.	Science	315,	1709–1712	(2007).	94.	 Deveau,	H.,	Barrangou,	R.,	Garneau,	J.	E.,	Labonté,	J.,	Fremaux,	C.,	Boyaval,	P.,	Romero,	D.	A.,	Horvath,	P.	&	Moineau,	S.	Phage	response	to	CRISPR-encoded	resistance	in	Streptococcus	thermophilus.	Journal	of	Bacteriology	190,	1390–1400	(2008).	95.	 Horvath,	P.,	Romero,	D.	A.,	Coûté-Monvoisin,	A-C.,	Richards,	M.,	Deveau,	H.,	Moineau,	S.,	Boyaval,	P.,	Fremaux,	C.	&	Barrangou,	R.	Diversity,	Activity,	and	Evolution	of	CRISPR	Loci	in	Streptococcus	thermophilus.	Journal	of	Bacteriology	190,	1401– 147 1412	(2008).	96.	 Makarova,	K.	S.,	Grishin,	N.	V.,	Shabalina,	S.	A.,	Wolf,	Y.	I.	&	Koonin,	E.	V.	A	putative	RNA-interference-based	immune	system	in	prokaryotes:	computational	analysis	of	the	predicted	enzymatic	machinery,	functional	analogies	with	eukaryotic	RNAi,	and	hypothetical	mechanisms	of	action.	Biol.	Direct	1,	7	(2006).	97.	 Garneau,	J.	E.,	Dupuis,	M-E.,	Villion,	M.,	Romero,	D.	A.,	Barrangou,	R.,	Boyaval,	P.,	Fremaux,	C.,	Horvath,	P.,	Magadán,	A.	H.	&	Moineau,	S.	The	CRISPR/Cas	bacterial	immunesystem	cleaves	bacteriophage	andplasmid	DNA.	Nature	468,	67–71	(2010).	98.	 Heler,	R.,	Marraffini,	L.	A.	&	Bikard,	D.	Adapting	to	new	threats:	the	generation	of	memory	by	CRISPR-Cas	immune	systems.	Molecular	Microbiology	93,	1–9	(2014).	99.	 Jinek,	M.,	Chylinski,	K.,	Fonfara,	I.,	Hauer,	M.,	Doudna,	J.	A.	&	Charpentier,	E.	A	Programmable	Dual-RNA-Guided	DNA	Endonuclease	in	Adaptive	Bacterial	Immunity.	Science	337,	816–821	(2012).	100.	 Malumbres,	M.	&	Barbacid,	M.	Cell	cycle,	CDKs	and	cancer:	a	changing	paradigm.	Nat	Rev	Cancer	9,	153–166	(2009).	101.	 Malumbres,	M.	Cyclin-dependent	kinases.	Genome	Biol.	15,	122–10	(2014).	102.	 Santamaria,	D.,	Barrière,	C.,	Cerqueira,	A.,	Hunt,	S.,	Tardy,	C.,	Newton,	K.,	Cáceres,	J.	F.,	Dubus,	P.,	Malumbres,	M.	&	Barbacid,	M.	Cdk1	is	sufficient	to	drive	the	mammalian	cell	cycle.	Nature	448,	811–815	(2007).	103.	 Satyanarayana,	A.	&	Kaldis,	P.	Mammalian	cell-cycle	regulation:	several	Cdks,	numerous	cyclins	and	diverse	compensatory	mechanisms.	Oncogene	28,	2925–2939	(2009).	104.	 Rane,	S.,	Dubus,	P.,	Mettus,	R.	&	Galbreath,	E.	Loss	of	Cdk4	expression	causes	insulin-deficient	diabetes	and	Cdk4	activation	results	in	β-islet	cell	hyperplasia.	Nat	Genet	(1999).	105.	 Kim,	S.	Y.	&	Rane,	S.	G.	The	Cdk4-E2f1	pathway	regulates	early	pancreas	development	by	targeting	Pdx1+	progenitors	and	Ngn3+	endocrine	precursors.	Development	138,	1903–1912	(2011).	106.	 Landsman,	L.,	Nijagal,	A.,	Whitchurch,	T.	J.,	VanderLaan,	R.	L.,	Zimmer,	W.	E.,	MacKenzie,	T.	C.	&	Hebrok,	M.	Pancreatic	Mesenchyme	Regulates	Epithelial	Organogenesis	throughout	Development.	PLoS	Biol	9,	e1001143	(2011).	107.	 Duvillié,	B.,	Attali,	M.,	Bounacer,	A.,	Ravassard,	P.,	Basmaciogullari,	A.	&	Scharfmann,	R.	The	mesenchyme	controls	the	timing	of	pancreatic	β-cell	differentiation.	Diabetes	55,	582–589	(2006).	108.	 Gittes,	G.	K.,	Galante,	P.	E.,	Hanahan,	D.,	Rutter,	W.	J.	&	Debase,	H.	T.	Lineage-specific	morphogenesis	in	the	developing	pancreas:	role	of	mesenchymal	factors.	Development	122,	439–447	(1995).	109.	 Georgia,	S.,	Soliz,	R.,	Li,	M.,	Zhang,	P.	&	Bhushan,	A.	p57	and	Hes1	coordinate	cell	cycle	exit	with	self-renewal	of	pancreatic	progenitors.	Developmental	Biology	298,	22–31	(2006).	110.	 Miyatsuka,	T.,	Kosaka,	Y.,	Kim,	H.	&	German,	M.	S.	Neurogenin3	inhibits	proliferation	in	endocrine	progenitors	by	inducing	Cdkn1a.	Proceedings	of	the	National	Academy	of	Sciences	108,	185–190	(2011).	111.	 Piccand,	J.,	Meunier,	A.,	Merle,	C.,	Jia,	Z.	&	Barnier,	J.	V.	Pak3	promotes	cell	cycle	exit	and	differentiation	of	beta-cells	in	the	embryonic	pancreas	and	is	necessary	to	maintain	glucose	homeostasis	in	adult	mice.	Diabetes	(2013).	 148 112.	 Becker,	K.	A.,	Ghule,	P.	N.,	Therrien,	J.	A.,	Lian,	J.	B.,	Stein,	J.	L.,	Van	Wijnen,	A.	J.	&	Stein,	G.	S.	Self-renewal	of	human	embryonic	stem	cells	is	supported	by	a	shortened	G1	cell	cycle	phase.	J.	Cell.	Physiol.	209,	883–893	(2006).	113.	 Becker,	K.	A.,	Stein,	J.	L.,	Lian,	J.	B.,	Van	Wijnen,	A.	J.	&	Stein,	G.	S.	Establishment	of	histone	gene	regulation	and	cell	cycle	checkpoint	control	in	human	embryonic	stem	cells.	J.	Cell.	Physiol.	210,	517–526	(2007).	114.	 Sela,	Y.,	Molotski,	N.,	Golan,	S.,	Itskovitz-Eldor,	J.	&	Soen,	Y.	Human	Embryonic	Stem	Cells	Exhibit	Increased	Propensity	to	Differentiate	During	the	G1	Phase	Prior	to	Phosphorylation	of	pRB.	Stem	Cells	30,	N/A–N/A	(2012).	115.	 Neganova,	I.,	Zhang,	X.,	Atkinson,	S.	&	Lako,	M.	Expression	and	functional	analysis	of	G1	to	S	regulatory	components	reveals	an	important	role	for	CDK2	in	cell	cycle	regulation	in	human	embryonic	stem	cells.	Oncogene	28,	20–30	(2008).	116.	 Salomoni,	P.	&	Calegari,	F.	Cell	cycle	control	of	mammalian	neural	stem	cells:	putting	a	speed	limit	on	G1.	Trends	in	Cell	Biology	20,	233–243	(2010).	117.	 Takahashi,	T.,	Nowakowski,	R.	S.	&	Caviness,	V.	S.	The	cell	cycle	of	the	pseudostratified	ventricular	epithelium	of	the	embryonic	murine	cerebral	wall.	Journal	of	Neuroscience	15,	6046–6057	(1995).	118.	 Calegari,	F.	&	Huttner,	W.	B.	An	inhibition	of	cyclin-dependent	kinases	that	lengthens,	but	does	not	arrest,	neuroepithelial	cell	cycle	induces	premature	neurogenesis.	Journal	of	Cell	Science	116,	4947	(2003).	119.	 Calegari,	F.,	Haubensak,	W.,	Haffner,	C.	&	Huttner,	W.	B.	Selective	lengthening	of	the	cell	cycle	in	the	neurogenic	subpopulation	of	neural	progenitor	cells	during	mouse	brain	development.	Journal	of	Neuroscience	25,	6533–6538	(2005).	120.	 Durand,	B.	&	Raff,	M.	A	cell-intrinsic	timer	that	operates	during	oligodendrocyte	development.	Bioessays	22,	64–71	(2000).	121.	 Nguyen,	L.,	Besson,	A.,	Heng,	J.	I-T.,	Schuurmans,	C.,	Teboul,	L,	Parras,	C.,	Philpott,	A.,	Roberts,	J.	M.	&	Guillemot,	F.	p27kip1	independently	promotes	neuronal	differentiation	and	migration	in	the	cerebral	cortex.	Genes	&	Development	20,	1511–1524	(2006).	122.	 Ali,	F.,	Hindley,	C.,	McDowell,	G.,	Deibler,	R.,	Jones,	A.,	Kirschner,	M.,	Guillemot,	F.	&	Philpott,	A.	Cell	cycle-regulated	multi-site	phosphorylation	of	Neurogenin	2	coordinates	cell	cycling	with	differentiation	during	neurogenesis.	Development	138,	4267–4277	(2011).	123.	 Vosper,	J.	M.	D.,	Fiore-Heriche,	C.	S.,	Horan,	I.,	Wilson,	K.,	Wise,	H.	&	Philpott,	A.	Regulation	of	neurogenin	stability	by	ubiquitin-mediated	proteolysis.	Biochem.	J.	407,	277–284	(2007).	124.	 Vosper,	J.	M.	D.,	McDowell,	G.	S.,	Hindley,	C.	J.,	Fiore-Heriche,	C.	S.,	Kucerova,	R,	Horan,	I.	&	Philpott,	A.	Ubiquitylation	on	canonical	and	non-canonical	sites	targets	the	transcription	factor	neurogenin	for	ubiquitin-mediated	proteolysis.	Journal	of	Biological	Chemistry	284,	15458–15468	(2009).	125.	 Roark,	R.,	Itzhaki,	L.	&	Philpott,	A.	Complex	regulation	controls	Neurogenin3	proteolysis.	Biology	Open	1,	1264–1272	(2012).	126.	 Vernon,	A.	E	&	Philpott,	A.	A	single	cdk	inhibitor,	p27Xic1,	functions	beyond	cell	cycle	regulation	to	promote	muscle	differentiation	in	Xenopus.	Development	130,	71–83	(2003).	127.	 Hindley,	C.	&	Philpott,	A.	Co-ordination	of	cell	cycle	and	differentiation	in	the	 149 developing	nervous	system.	Biochem.	J.	444,	375–382	(2012).	128.	 Muzumdar,	M.	D.,	Tasic,	B.,	Miyamichi,	K.,	Li,	L.	&	Luo,	L.	A	global	double-fluorescent	Cre	reporter	mouse.	genesis	45,	593–605	(2007).	129.	 Jackson,	E.	L.,	Willis,	N.,	Mercer,	K.,	Bronson,	R.	T.,	Crowley,	D.,	Montoya,	R.,	Jacks,	T.	&	Tuveson,	D.	A.	Analysis	of	lung	tumor	initiation	and	progression	using	conditional	expression	of	oncogenic	K-ras.	Genes	&	Development	15,	3243–3248	(2001).	130.	 Pruitt,	S.	C.,	Freeland,	A.,	Rusiniak,	M.	E.,	Kunnev,	D.	&	Cady,	G.	K.	Cdkn1b	overexpression	in	adult	mice	alters	the	balance	between	genome	and	tissue	ageing.	Nat	Comms	4,	1–12	(2013).	131.	 Shimajiri,	Y.,	Kosaka,	Y.,	Scheel,	D.	W.,	Lynn,	F.	C.,	Kishimoto,	N.,	Wang,	J.,	Zhao,	S.	&	German,	M.	S.	A	mouse	model	for	monitoring	islet	cell	genesis	and	developing	therapies	for	diabetes.	Disease	Models	&	Mechanisms	4,	268–276	(2011).	132.	 Takahashi,	T.,	Nowakowski,	R.	S.	&	Caviness,	V.	S.,	Jr.	Cell	cycle	parameters	and	patterns	of	nuclear	movement	in	the	neocortical	proliferative	zone	of	the	fetal	mouse.	Journal	of	Neuroscience	13,	820–833	(1993).	133.	 Salic,	A.	&	Mitchison,	T.	J.	A	chemical	method	for	fast	and	sensitive	detection	of	DNA	synthesis	in	vivo.	Proceedings	of	the	National	Academy	of	Sciences	105,	2415–2420	(2008).	134.	 Carpenter,	A.	E.,	Jones,	T.	R.,	Lamprecht,	M.	R.,	Clarke,	C.,	Kang,	I.	H.,	Friman,	O.,	Guertin,	D.	A.,	Chang,	J.	H.,	Lindquist,	R.	A.,	Moffat,	J.,	Golland,	P.	and	Sabatini,	D.	M.	CellProfiler:	image	analysis	software	for	identifying	and	quantifying	cell	phenotypes.	Genome	Biol.	7,	R100	(2006).	135.	 Petzold,	K.	M.	&	Spagnoli,	F.	M.	A	System	for	ex	vivo	Culturing	of	Embryonic	Pancreas.	JoVE	e3979–e3979	(2012).	136.	 Schulz,	T.	C.,	Young,	H.	Y.,	Agulnick,	A.	D.,	Babin,	M.	J.,	Baetge,	E.	E.,	Bang,	A.	G.,	Bhoumik,	A.,	Cepa,	I.,	Cesario,	R.	M.,	Haakmeester,	C.,	Kadoya,	K.,	Kelly,	J.	R.,	Kerr,	J.,	Martinson,	L.	A.,	McLean,	A.	B.,	Moorman,	M.	A.,	Payne,	J.	K.,	Richardson,	M.,	Ross,	K.	G.,	Sherrer,	E.	S.,	Song,	X.,	Wilson,	A.	Z.,	Brandon,	E.	P.,	Green,	C.	E.,	Kroon,	E.	J.,	Kelly,	O.	G.,	D'Amour,	K.	A.	&	Robins,	A.	J.	A	Scalable	System	for	Production	of	Functional	Pancreatic	Progenitors	from	Human	Embryonic	Stem	Cells.	PLoS	ONE	7,	e37004	(2012).	137.	 Sanjana,	N.	E.,	Cong,	L.,	Zhou,	Y.,	Cunniff,	M.	M.,	Feng,	G.	&	Zhang,	F.	A	transcription	activator-like	effector	toolbox	for	genome	engineering.	Nat	Protoc	7,	171–192	(2012).	138.	 Le	Cong,	Zhou,	R.,	Kuo,	Y.-C.,	Cunniff,	M.	&	Zhang,	F.	Comprehensive	interrogation	of	natural	TALE	DnA-binding	modules	and	transcriptional	repressor	domains.	Nat	Comms	3,	968–6	(2012).	139.	 Doench,	J.	G.,	Hartenian,	E.,	Graham,	D.	B.,	Tothova,	Z.,	Hegde,	M.,	Smith,	I.,	Sullender,	M.,	Ebert,	B.	L.,	Xavier,	R.	J.	&	Root,	D.	E.	Rational	design	of	highly	active	sgRNAs	for	CRISPR-Cas9-mediated	gene	inactivation.	Nat	Biotechnol	1–8	(2014).	140.	 Ran,	F.	A.,	Hsu,	P.	D.,	Wright,	J.,	Agarwala,	V.,	Scott,	D.	A.	&	Zhang,	F.	Genome	engineering	using	the	CRISPR-Cas9	system.	Nat	Protoc	8,	2281–2308	(2013).	141.	 Hockemeyer,	D.,	Wang,	H.,	Kiani,	S.,	Lai,	C.	S.,	Gao,	Q.,	Cassady,	J.	P.,	Cost,	G.	J.,	Zhang,	L.,	Santiago,	Y.,	Miller,	J.	C.,	Zeitler,	B.,	Cherone,	J.	M.,	Meng,	X.,	Hinkley,	S.	J.,	Rebar,	E.	J.,	Gregory,	P.	D.,	Urnov,	F.	D.	&	Jaenisch,	R.	Genetic	engineering	of	human	pluripotent	cells	using	TALE	nucleases.	Nat	Biotechnol	29,	731–734	(2011).	 150 142.	 Krentz,	N.	A.	J.,	Nian,	C.	&	Lynn,	F.	C.	TALEN/CRISPR-Mediated	eGFP	Knock-In	Add-On	at	the	OCT4	Locus	Does	Not	Impact	Differentiation	of	Human	Embryonic	Stem	Cells	towards	Endoderm.	PLoS	ONE	9,	e114275–21	(2014).	143.	 Dobin,	A.,	Davis,	C.	A.,	Schlesinger,	F.,	Drenkow,	J.,	Zaleski,	C.,	Jha,	S.,	Batut,	P.,	Chaisson,	M.	&	Gingeras,	T.	R.	STAR:	ultrafast	universal	RNA-seq	aligner.	Bioinformatics	29,	15–21	(2012).	144.	 Blomen,	V.	A.	&	Boonstra,	J.	Cell	fate	determination	during	G1	phase	progression.	Cell.	Mol.	Life	Sci.	64,	3084–3104	(2007).	145.	 Mummery,	C.	L.,	van	Rooijen,	M.	A.,	van	den	Brink,	S.	E.	&	de	Laat,	S.	W.	Cell	cycle	analysis	during	retinoic	acid	induced	differentiation	of	a	human	embryonal	carcinoma-derived	cell	line.	Cell	Differ.	20,	153–160	(1987).	146.	 Murray,	A.	W.	&	Kirschner,	M.	W.	Cyclin	synthesis	drives	the	early	embryonic	cell	cycle.	Nature	339,	275–280	(1989).	147.	 Yarden,	A.	&	Geiger,	B.	Zebrafish	cyclin	E	regulation	during	early	embryogenesis.	Dev.	Dyn.	206,	1–11	(1996).	148.	 Edgar,	B.	A.	&	Lehner,	C.	F.	Developmental	control	of	cell	cycle	regulators:	a	fly's	perspective.	Science	274,	1646–1652	(1996).	149.	 Mac	Auley,	A.,	Werb,	Z.	&	Mirkes,	P.	E.	Characterization	of	the	unusually	rapid	cell	cycles	during	rat	gastrulation.	Development	117,	873–883	(1993).	150.	 Stead,	E.,	White,	J.,	Faast,	R.,	Conn,	S.	&	Goldstone,	S.	Pluripotent	cell	division	cycles	are	driven	by	ectopic	Cdk2,	cyclin	A/E	and	E2F	activities.	Oncogene	(2002).	151.	 White,	J.	Cell	cycle	control	of	embryonic	stem	cells.	Stem	Cell	Rev	and	Rep	(2005).	152.	 Faast,	R.,	White,	J.,	Cartwright,	P.,	Crocker,	L.,	Sarcevic,	B.	&	Dalton,	S.	Cdk6–cyclin	D3	activity	in	murine	ES	cells	is	resistant	to	inhibition	by	p16INK4a.	Oncogene	23,	491–502	(2004).	153.	 Lange,	C.	&	Calegari,	F.	Cdks	and	cyclins	link	G1	length	and	differentiation	of	embryonic,	neural	and	hematopoietic	stem	cells.	Cell	Cycle	9,	1893–1900	(2010).	154.	 Arai,	Y.,	Pulvers,	J.	N.,	Haffner,	C.,	Schilling,	B.,	Nusslein,	I.,	Calegari,	F.	&	Huttner,	W.	B.	Neural	stem	and	progenitor	cells	shorten	S-phase	on	commitment	to	neuron	production.	Nat	Comms	2,	154	(2011).	155.	 Wescott,	M.	P.,	Rovira,	M.,	Reichert,	M.,	von	Burstin,	J.,	Means,	A.,	Leach,	S.	D.	&	Rustgi,	A.	K.	Pancreatic	ductal	morphogenesis	and	the	Pdx1	homeodomain	transcription	factor.	Molecular	Biology	of	the	Cell	20,	4838–4844	(2009).	156.	 Lynn,	F.	C.,	Skewes-Cox,	P.,	Kosaka,	Y.,	McManus,	M.	T.,	Harfe,	B.	D.	&	German,	M.	S.	MicroRNA	Expression	Is	Required	for	Pancreatic	Islet	Cell	Genesis	in	the	Mouse.	Diabetes	56,	2938–2945	(2007).	157.	 Kim,	Y.	H.,	Larsen,	H.	L.,	Rué,	P.,	Lemaire,	L.	A.,	Ferrer,	J.	&	Grapin-Botton,	A.	Cell	cycle-dependent	differentiation	dynamics	balances	growth	and	endocrine	differentiation	in	the	pancreas.	PLoS	Biol	13,	e1002111	(2015).	158.	 Durand,	B.,	Gao,	F.	B.	&	Raff,	M.	Accumulation	of	the	cyclin-dependent	kinase	inhibitor	p27/Kip1	and	the	timing	of	oligodendrocyte	differentiation.	EMBO	J	16,	306–317	(1997).	159.	 Georgia,	S.	&	Bhushan,	A.	p27	Regulates	the	transition	of	beta-cells	from	quiescence	to	proliferation.	Diabetes	55,	2950–2956	(2006).	160.	 Bhushan,	A.,	Itoh,	N.,	Kato,	S.,	Thiery,	J.	P.,	Czernichow,	P.,	Bellusci,	S.	&	Scharfman,	R.	Fgf10	is	essential	for	maintaining	the	proliferative	capacity	of	epithelial	 151 progenitor	cells	during	early	pancreatic	organogenesis.	Development	128,	5109–5117	(2001).	161.	 Wennerberg,	K.,	Rossman,	K.	L.	&	Der,	C.	J.	The	Ras	superfamily	at	a	glance.	Journal	of	Cell	Science	118,	843–846	(2005).	162.	 Jones,	S.,	Zhang,	X.,	Parsons,	D.	W.,	Lin,	J.	C-H.,	Leary,	R.	J.,	Angenendt,	P.,	Mankoo,	P.,	Carter,	H.,	Kamiyama,	H.,	Jimeno,	A.,	Hong,	S-M.,	Fu,	B.,	Lin,	M-T.,	Calhoun,	E.	S.,	Kamiyama,	M.,	Walter,	K.,	Nikolskaya,	T.,	Nikolsky,	Y.,	Hartigan,	J.,	Smith,	D.	R.,	Hidalgo,	M.,	Leach,	S.	D.,	Klein,	A.	P.,	Jaffee,	E.	M.,	Goggins,	M.,	Maitra,	A.,	Iacobuzio-Donahue,	C.,	Eshleman,	J.	R.,	Kern,	S.	E.,	Hruban,	R.	H.,	Karchin,	R.,	Papadopoulos,	N.,	Parmigiani,	G.,	Vogelsetin,	B.,	Velculescu,	V.	E.	&	Kinzler,	K.	W.	Core	signaling	pathways	in	human	pancreatic	cancers	revealed	by	global	genomic	analyses.	Science	321,	1801–1806	(2008).	163.	 Chamberlain,	C.	E.,	Scheel,	D.	W.,	McGlynn,	K.,	Kim,	H.,	Miyatsuka,	T.,	Wang,	J.,	Nguyen,	V.,	Zhao,	S.,	Mavropoulos,	A.,	Abraham,	A.	G.,	O'Neill,	E.,	Ku,	G.	M.,	Cobb,	M.	H.,	Martin,	G.	R.	&	German,	M.	S.	Menin	determines	K-RAS	proliferative	outputs	in	endocrine	cells.	J.	Clin.	Invest.	124,	4093–4101	(2014).	164.	 Toyoshima,	H.	&	Hunter,	T.	P27,	a	Novel	Inhibitor	of	Gl	Cyclin-Cdk	Protein	Kinase	Activity,	Is	Related	to	p21.	Cell	78,	67–74	(1994).	165.	 Rachdi,	L.,	Balcazar,	N.,	Elghazi,	L.,	Barker,	D.	J.,	Krits,	I.,	Kiyokawa,	H.	Bernal-Mizrachi,	E.	Differential	effects	of	p27	in	regulation	of	beta-cell	mass	during	development,	neonatal	period,	and	adult	life.	Diabetes	55,	3520–3528	(2006).	166.	 Uchida,	T.,	Nakamura,	T.,	Hashimoto,	N.,	Matsuda,	T.,	Kotani,	K.,	Sakaue,	H.,	Kido,	Y.,	Hayashi,	Y.,	Nakayama,	K.	I.,	White,	M.	F.	&	Kasuga,	M.	Deletion	of	Cdkn1b	ameliorates	hyperglycemia	by	maintaining	compensatory	hyperinsulinemia	in	diabetic	mice.	Nature	Medicine	11,	175–182	(2005).	167.	 Villasenor,	A.,	Chong,	D.	C.	&	Cleaver,	O.	Biphasic	Ngn3	expression	in	the	developing	pancreas.	Dev.	Dyn.	237,	3270–3279	(2008).	168.	 Bechard,	M.	E.,	Bankaitis,	E.	D.,	Hipkens,	S.	B.,	Ustione,	A.,	Piston,	D.	W.,	Yang,	Y-P.,	Magnuson,	M.	A.	&	Wright,	C.	V.	E.	Precommitment	low-level	Neurog3	expression	defines	a	long-lived	mitotic	endocrine-biased	progenitor	pool	that	drives	production	of	endocrine-committed	cells.	Genes	&	Development	1–15	(2016).	169.	 Wang,	S.,	Jensen,	J.	N.,	Seymour,	P.	A.,	Hsu,	W.,	Dor,	Y.,	Sander,	M.,	Magnuson,	M.	A.,	Serup,	P.	&	Gu,	G.	Sustained	Neurog3	expression	in	hormone-expressing	islet	cells	is	required	for	endocrine	maturation	and	function.	Proceedings	of	the	National	Academy	of	Sciences	106,	9715–9720	(2009).	170.	 Leon,	J.,	Guerrero,	I.	&	Pellicer,	A.	Differential	expression	of	the	ras	gene	family	in	mice.	Molecular	and	Cellular	Biology	7,	1535–1540	(1987).	171.	 Umanoff,	H.,	Edelmann,	W.,	Pellicer,	A.	&	Kucherlapati,	R.	The	murine	N-ras	gene	is	not	essential	for	growth	and	development.	Proceedings	of	the	National	Academy	of	Sciences	92,	1709–1713	(1995).	172.	 Ise,	K.,	Nakamura,	K.,	Nakao,	K.,	Shimizu,	S.,	Harada,	H.,	Ichise,	T.,	Miyoshi,	J.,	Gondo,	Y.,	Ishikawa,	T.,	Aiba,	A.	&	Katsuki,	M.	Targeted	deletion	of	the	H-ras	gene	decreases	tumor	formation	in	mouse	skin	carcinogenesis.	Oncogene	19,	2951–2956	(2000).	173.	 Koera,	K.,	Nakamura,	K.,	Nakao,	K.,	Miyoshi,	J.,	Toyoshima,	K.,	Hatta,	T.,	Otani,	H.,	Aiba,	A.	&	Katsuki,	M.	K-ras	is	essential	for	the	development	of	the	mouse	embryo.	Oncogene	15,	1151–1159	(1997).	 152 174.	 Nakamura,	K.,	Ichise,	H.,	Nakao,	K.,	Hatta,	T.,	Otani,	H.,	Sakagami,	H.,	Kondo,	H.	&	Katsuki,	M.	Partial	functional	overlap	of	the	three	ras	genes	in	mouse	embryonic	development.	Oncogene	27,	2961–2968	(2007).	175.	 Karnoub,	A.	E.	&	Weinberg,	R.	A.	Ras	oncogenes:	split	personalities.	Nat	Rev	Mol	Cell	Biol	9,	517–531	(2008).	176.	 di	Magliano,	M.	P.	&	Logsdon,	C.	D.	Roles	for	KRAS	in	Pancreatic	Tumor	Development	and	Progression.	Gastroenterology	144,	1220–1229	(2013).	177.	 Sabherwal,	N.,	Thuret,	R.,	Lea,	R.,	Stanley,	P.	&	Papalopulu,	N.	aPKC	phosphorylates	p27Xic1,	providing	a	mechanistic	link	between	apicobasal	polarity	and	cell-cycle	control.	Developmental	Cell	31,	559–571	(2014).	178.	 Lange,	C.,	Huttner,	W.	B.	&	Calegari,	F.	Cdk4/CyclinD1	Overexpression	in	Neural	Stem	Cells	Shortens	G1,	Delays	Neurogenesis,	and	Promotes	the	Generation	and	Expansion	of	Basal	Progenitors.	Cell	Stem	Cell	5,	320–331	(2009).	179.	 Richard-Parpaillon,	L.,	Cosgrove,	R.	A.,	Devine,	C.,	Vernon,	A.	E.	&	Philpott,	A.	G1/S	phase	cyclin-dependent	kinase	overexpression	perturbs	early	development	and	delays	tissue-specific	differentiation	in	Xenopus.	Development	131,	2577–2586	(2004).	180.	 Zwaka,	T.	P.	&	Thomson,	J.	A.	Homologous	recombination	in	human	embryonic	stem	cells.	Nat	Biotechnol	21,	319–321	(2003).	181.	 Hockemeyer,	D.,	Soldner,	F.,	Beard,	C.,	Gao,	Q.,	Mitalipova,	M.,	DeKelver,	R.	C.,	Katibah,	G.	E.,	Amora,	R.,	Boydston,	E.	A.,	Zeitler,	B.,	Meng,	X.,	Miller,	J.	C.,	Zhang,	L.,	Rebar,	E.	J.,	Gregory,	P.	D.,	Urnov,	F.	D.	&	Jaenisch,	R.	Efficient	targeting	of	expressed	and	silent	genes	in	human	ESCs	and	iPSCs	using	zinc-finger	nucleases.	Nat	Biotechnol	27,	851–857	(2009).	182.	 Brafman,	D.	A.,	Moya,	N.,	Allen-Soltero,	S.,	Fellner,	T.,	Robinson,	M.,	McMillen,	Z.	L.,	Gaasterland,	T.	&	Willert,	K.	Analysis	of	SOX2-Expressing	Cell	Populations	Derived	from	Human	Pluripotent	Stem	Cells.	Stem	Cell	Reports	1,	464–478	(2013).	183.	 Fischer,	Y.,	Ganic,	E.,	Ameri,	J.,	Xian,	X.,	Johannesson,	M.	&	Semb,	H.	NANOG	Reporter	Cell	Lines	Generated	by	Gene	Targeting	in	Human	Embryonic	Stem	Cells.	PLoS	ONE	5,	e12533	(2010).	184.	 Kanai-Azuma,	M.,	Kanai,	Y.,	Gad,	J.	M.,	Tajima,	Y.,	Taya,	C.,	Kurohmaru,	M.,	Sanai,	Y.,	Yonekawa,	H.,	Yazaki,	K.,	Tam,	P.	P.	L.	&	Hayashi,	Y.	Depletion	of	definitive	gut	endoderm	in	Sox17-null	mutant	mice.	Development	129,	2367–2379	(2002).	185.	 Wang,	P.,	Rodriguez,	R.	T.,	Wang,	J.,	Ghodasara,	A.	&	Kim,	S.	K.	Targeting	SOX17	in	Human	Embryonic	Stem	Cells	Creates	Unique	Strategies	for	Isolating	and	Analyzing	Developing	Endoderm.	Cell	Stem	Cell	8,	335–346	(2011).	186.	 Zhu,	Z.,	Verma,	N.,	González,	F.,	Shi,	Z.-D.	&	Huangfu,	D.	A	CRISPR/Cas-Mediated	Selection-free	Knockin	Strategy	in	Human	Embryonic	Stem	Cells.	Stem	Cell	Reports	4,	1–27	(2015).	187.	 Nostro,	M.	C.,	Sarangi,	F.,	Yang,	C.,	Holland,	A.,	Elefanty,	A.	G.,	Stanley,	E.	G.,	Greiner,	D.	L.	&	Keller,	G.	Efficient	Generation	of	NKX6-1+	Pancreatic	Progenitors	from	Multiple	Human	Pluripotent	Stem	Cell	Lines.	Stem	Cell	Reports	4,	591–604	(2015).	188.	 Liu,	H.,	Yang,	H.,	Zhu,	D.,	Sui,	X.,	Li,	J.,	Liang,	Z.,	Xu,	L.,	Chen,	Z.,	Yao,	A.,	Zhang,	L.,	Zhang,	X.,	Yi,	X.,	Liu,	M.,	Xu,	S.,	Zhang,	W.,	Lin,	H.,	Xie,	L.,	Lou,	J.,	Zhang,	Y.,	Xi,	J.	&	Deng,	H.	Systematically	labeling	developmental	stage-specific	genes	for	the	study	of	pancreatic	β-cell	differentiation	from	human	embryonic	stem	cells.	Cell	Research	 153 24,	1181–1200	(2014).	189.	 McGrath,	P.	S.,	Watson,	C.	L.,	Ingram,	C.,	Helmrath,	M.	A.	&	Wells,	J.	M.	The	Basic	Helix-Loop-Helix	Transcription	Factor	NEUROG3	Is	Required	for	Development	of	the	Human	Endocrine	Pancreas.	Diabetes	64,	2497–2505	(2015).	190.	 Krentz,	N.	A.	J.,	van	Hoof,	D.,	Li,	Z.,	Watanabe,	A.,	Tang,	M.,	Nian,	C.,	German,	M.	S.	&	Lynn,	F.	C.	Phosphorylation	of	NEUROG3	Links	Endocrine	Differentiation	to	the	Cell	Cycle	in	Pancreatic	Progenitors.	Developmental	Cell	41,	129–142.e6	(2017).	191.	 Petersen,	M.	B.	K.,	Azad,	A.,	Ingvorsen,	C.,	Hess,	K.,	Hansson,	M.,	Grapin-Botton,	A.	&	Honoré,	C.	Single-Cell	Gene	Expression	Analysis	of	a	Human	ESC	Model	of	Pancreatic	Endocrine	Development	Reveals	Different	Paths	to	β-Cell	Differentiation.	Stem	Cell	Reports	1–16	(2017).	192.	 Nichols,	J.,	Zevnik,	B.,	Anastassiadis,	K.,	Niwa,	H.,	Klewe-Nebenius,	D.,	Chambers,	I.,	Scholer,	H.	&	Smith,	A.	Formation	of	pluripotent	stem	cells	in	the	mammalian	embryo	depends	on	the	POU	transcription	factor	Oct4.	Cell	95,	379–391	(1998).	193.	 Radzisheuskaya,	A.,	Chia,	G.	L.	B.,	dos	Santos,	R.	L.,	Theunissen,	T.	W.,	Castro,	L.	F.	C.,	Nichols,	J.	&	Silva,	J.	C.	R.	A	defined	Oct4	level	governs	cell	state	transitions	of	pluripotency	entry	and	differentiation	into	all	embryonic	lineages.	Nat	Cell	Biol	15,	579–590	(2013).	194.	 Moscou,	M.	J.	&	Bogdanove,	A.	J.	A	Simple	Cipher	Governs	DNA	Recognition	by	TAL	Effectors.	Science	326,	1501–1501	(2009).	195.	 Mali,	P.,	Yang,	L.,	Esvelt,	K.	M.,	Aach,	J.,	Guell,	M.,	DiCarlo,	J.	E.,	Norville,	J.	E.	&	Church,	G.	M.	RNA-guided	human	genome	engineering	via	Cas9.	Science	339,	823–826	(2013).	196.	 Cong,	L.,	Ran,	F.	A.,	Cox,	D.,	Lin,	S.,	Barretto,	R.,	Habib,	N.,	Hsu,	P.	D.,	Wu,	X.,	Jiang,	W.,	Marraffini,	L.	A.	&	Zhang,	F.	Multiplex	genome	engineering	using	CRISPR/Cas	systems.	Science	339,	819–823	(2013).	197.	 McGrath,	K.	E.,	Koniski,	A.	D.,	Maltby,	K.	M.,	McGann,	J.	K.	&	Palis,	J.	Embryonic	expression	and	function	of	the	chemokine	SDF-1	and	its	receptor,	CXCR4.	Developmental	Biology	213,	442–456	(1999).	198.	 Séguin,	C.	A.,	Draper,	J.	S.,	Nagy,	A.	&	Rossant,	J.	Establishment	of	Endoderm	Progenitors	by	SOX	Transcription	Factor	Expression	in	Human	Embryonic	Stem	Cells.	Cell	Stem	Cell	3,	182–195	(2008).	199.	 Schaffer,	A.	E.,	Freude,	K.	K.,	Nelson,	S.	B.	&	Sander,	M.	Nkx6	Transcription	Factors	and	Ptf1a	Function	as	Antagonistic	Lineage	Determinants	in	Multipotent	Pancreatic	Progenitors.	Developmental	Cell	18,	1022–1029	(2010).	200.	 Henseleit,	K.	D.,	Nelson,	S.	B.,	Kuhlbrodt,	K.,	Hennings,	J.	C.,	Ericson,	J.	&	Sander,	M.	NKX6	transcription	factor	activity	is	required	for	alpha-	and	beta-cell	development	in	the	pancreas.	Development	132,	3139–3149	(2005).	201.	 Sander,	M.,	Sussel,	L.,	Conners,	J.	&	Scheel,	D.	Homeobox	gene	Nkx6.	1	lies	downstream	of	Nkx2.	2	in	the	major	pathway	of	beta-cell	formation	in	the	pancreas.	Development	127,	5533–5540	(2000).	202.	 Rezania,	A.,	Bruin,	J.	E.,	Xu,	J.,	Narayan,	K.,	Fox,	J.	K.,	O'Neil,	J.	J.	&	Kieffer,	T.	J.	Enrichment	of	human	embryonic	stem	cell-derived	NKX6.1-expressing	pancreatic	progenitor	cells	accelerates	the	maturation	of	insulin-secreting	cells	in	vivo.	Stem	Cells	31,	2432–2442	(2013).	203.	 Lynn,	F.	C.,	Smith,	S.	B.,	Wilson,	M.	E.,	Yang,	K.	Y.,	Nekrep,	N.	&	German,	M.	S.	Sox9	 154 coordinates	a	transcriptional	network	in	pancreatic	progenitor	cells.	Proceedings	of	the	National	Academy	of	Sciences	104,	10500–10505	(2007).	204.	 Seymour,	P.	A.,	Freude,	K.	K.,	Tran,	M.	N.,	Mayes,	E.	E.,	Jensen,	J.,	Kist,	R.,	Scherer,	G.	&	Sander,	M.	SOX9	is	required	for	maintenance	of	the	pancreatic	progenitor	cell	pool.	Proceedings	of	the	National	Academy	of	Sciences	104,	1865–1870	(2007).	205.	 Seymour,	P.	A.,	Shih,	H.	P.,	Patel,	N.	A.,	Freude,	K.	K.,	Xie,	R.,	Lim,	C.	J.	&	Sander,	M.	A	Sox9/Fgf	feed-forward	loop	maintains	pancreatic	organ	identity.	Development	139,	3363–3372	(2012).	206.	 McDonald,	E.,	Li,	J.,	Krishnamurthy,	M.,	Fellows,	G.	F.,	Goodyer,	C.	G.	&	Wang,	R.	SOX9	regulates	endocrine	cell	differentiation	during	human	fetal	pancreas	development.	International	Journal	of	Biochemistry	and	Cell	Biology	44,	72–83	(2011).	207.	 Takahashi,	K.,	Tanabe,	K.,	Ohnuki,	M.,	Narita,	M.,	Ichisaka,	T.,	Tomoda,	K.	&	Yamanaka,	S.	Induction	of	pluripotent	stem	cells	from	adult	human	fibroblasts	by	defined	factors.	Cell	131,	861–872	(2007).	208.	 Niwa,	H.,	Miyazaki,	J.-I.	&	Smith,	A.	G.	Quantitative	expression	of	Oct-3/4	defines	differentiation,	dedifferentiation	or	self-renewal	of	ES	cells.	Nat	Genet	24,	372–376	(2000).	209.	 Boyer,	L.	A.,	Lee,	T.	I.,	Cole,	M.	F.,	Johnstone,	S.	E.,	Levine,	S.	S.,	Zucker,	J.	P.,	Guenther,	M.	G.,	Kumar,	R.	M.,	Murray,	H.	L.,	Jenner,	R.	G.,	Gifford,	D.	K.,	Melton,	D.	A.,	Jaenisch,	R.	&	Young,	R.	A.	Core	Transcriptional	Regulatory	Circuitry	in	Human	Embryonic	Stem	Cells.	Cell	122,	947–956	(2005).	210.	 Reim,	G.,	Mizoguchi,	T.,	Stainier,	D.	Y.,	Kikuchi,	Y.	&	Brand,	M.	The	POU	domain	protein	spg	(pou2/Oct4)	is	essential	for	endoderm	formation	in	cooperation	with	the	HMG	domain	protein	casanova.	Developmental	Cell	6,	91–101	(2004).	211.	 Aksoy,	I.,	Jauch,	R.,	Chen,	J.,	Dyla,	M.,	Divakar,	U.,	Bogu,	G.	K.,	Teo,	R.,	Ng,	C.	K.	L.,	Herath,	W.,	Lili,	S.,	Hutchins,	A.	P.,	Robson,	P.,	Kolatkar,	P.	R.	&	Stanton,	L.	W.	Oct4	switches	partnering	from	Sox2	to	Sox17	to	reinterpret	the	enhancer	code	and	specify	endoderm.	EMBO	J	32,	938–953	(2013).	212.	 Rodriguez,	R.	T.,	Velkey,	J.	M.,	Lutzko,	C.,	Seerke,	R.,	Kohn,	D.	B.,	O'Shea,	K.	S.	&	Firpo,	M.	T.	Manipulation	of	OCT4	levels	in	human	embryonic	stem	cells	results	in	induction	of	differential	cell	types.	Exp.	Biol.	Med.	(Maywood)	232,	1368–1380	(2007).	213.	 Cai,	Q.,	Bonfanti,	P.,	Sambathkumar,	R.,	Vanuytsel,	K.,	Vanhove,	J.,	Gysemans,	C.,	Debiec-Rychter,	M.,	Raitano,	S.,	Heimberg,	H.,	Ordovas,	L.	&	Verfaillie,	C.	M.	Prospectively	Isolated	NGN3-Expressing	Progenitors	From	Human	Embryonic	Stem	Cells	Give	Rise	to	Pancreatic	Endocrine	Cells.	Stem	Cells	Translational	Medicine	3,	489–499	(2014).	214.	 Allen,	H.	L.,	Flanagan,	S.	E.,	Shaw-Smith,	C.,	De	Franco,	E.,	Akerman,	I.,	Caswell,	R.,	International	Pancreatic	Agenesis	Consortium,	Ferrer,	J.,	Hattersley,	A.	T.	&	Ellard,	S.	GATA6	haploinsufficiency	causes	pancreatic	agenesis	in	humans.	Nat	Genet	44,	20–22	(2011).	215.	 Carrasco,	M.,	Delgado,	I.,	Soria,	B.,	Martín,	F.	&	Rojas,	A.	GATA4	and	GATA6	control	mouse	pancreas	organogenesis.	J.	Clin.	Invest.	122,	3504–3515	(2012).	216.	 Tiyaboonchai,	A.,	Cardenas-Diaz,	F.	L.,	Ying,	L.,	Maguire,	J.	A.,	Sim,	X.,	Jobaliya,	C.,	Gagne,	A.	L.,	Kishore,	S.,	Stanescu,	D.	E.,	Hughes,	N.,	De	Leon,	D.	D.,	French,	D.	L.	&	Gadue,	P.	GATA6	Plays	an	Important	Role	in	the	Induction	of	Human	Definitive	 155 Endoderm,	Development	of	the	Pancreas,	and	Functionality	of	Pancreatic	β	Cells.	Stem	Cell	Reports	8,	589–604	(2017).	217.	 Wang,	J.,	Cortina,	G.,	Wu,	V.,	Tran,	R.,	Cho,	J-H.,	Tsai,	M-J,	Bailey,	T.	J.,	Jamrich,	M.,	Ament,	M.	E.,	Treem,	W.	R.,	Hill,	I.	D.,	Vargas,	J.	H.,	Gershman,	G.,	Farmer,	D.	G.,	Reyen,	L.	&	Martín,	M.	G.	Mutant	neurogenin-3	in	congenital	malabsorptive	diarrhea.	New	England	Journal	of	Medicine	355,	270–280	(2006).	218.	 Rubio-Cabezas,	O.,	Codner,	E.,	Flanagan,	S.	E.,	Gómez,	J.	L.,	Ellard,	S.	&	Hattersley,	A.	T.	Neurogenin	3	is	important	but	not	essential	for	pancreatic	islet	development	in	humans.	Diabetologia	57,	2421–2424	(2014).	219.	 Sayar,	E.,	Islek,	A.,	Yilmaz,	A.,	Akcam,	M.,	Flanagan,	S.	E.	&	Artan,	R.	Extremely	rare	cause	of	congenital	diarrhea:	Enteric	anendocrinosis.	Pediatrics	International	55,	661–663	(2013).	220.	 Spence,	J.	R.,	Mayhew,	C.	N.,	Rankin,	S.	A.,	Kuhar,	M.,	Vallance,	J.	E.,	Tolle,	K.,	Hoskins,	E.	E.,	Kalinichenko,	V.	V.,	Wells,	S.	I.,	Zorn,	A.	M.,	Shroyer,	N.	F.	&	Wells,	J.	M.	Directed	differentiation	of	human	pluripotent	stem	cells	into	intestinal	tissue	in	vitro.	Nature	470,	105–109	(2010).	221.	 Sakaue-Sawano,	A.,	Kurokawa,	H.,	Morimura,	T.,	Hanyu,	A.,	Hama,	H.,	Osawa,	H.,	Kashiwagi,	S.,	Fukami,	K.,	Miyata,	T.,	Miyoshi,	H.,	Imamura,	T.,	Ogawa,	M.,	Masai,	H.	&	Miyawaki,	A.	Visualizing	spatiotemporal	dynamics	of	multicellular	cell-cycle	progression.	Cell	132,	487–498	(2008).	222.	 Hindley,	C.,	Ali,	F.,	McDowell,	G.,	Cheng,	K.,	Jones,	A.,	Guillemot,	F.	&	Philpott,	A.	Post-translational	modification	of	Ngn2	differentially	affects	transcription	of	distinct	targets	to	regulate	the	balance	between	progenitor	maintenance	and	differentiation.	Development	139,	1718–1723	(2012).	223.	 Salisbury,	R.	J.,	Blaylock,	J.,	Berry,	A.	A.,	Jennings,	R.	E.,	De	Krijger,	R.,	Hanley,	K.	P.	&	Hanley,	N.	A.	The	window	period	of	NEUROGENIN3	during	human	gestation.	Islets	6,	e954436–5	(2014).	224.	 Cano,	D.	A.,	Soria,	B.,	Martín,	F.	&	Rojas,	A.	Transcriptional	control	of	mammalian	pancreas	organogenesis.	Cell.	Mol.	Life	Sci.	71,	2383–2402	(2013).	225.	 Tang,	F.,	Barbacioru,	C.,	Wang,	Y.,	Nordman,	E.,	Lee,	C.,	Xu,	N.,	Wang,	X.,	Bodeau,	J.,	Tuch,	B.	B.,	Siddiqui,	A.,	Lao,	K.	&	Surani,	A.	mRNA-Seq	whole-transcriptome	analysis	of	a	single	cell.	Nat	Meth	6,	377–382	(2009).	226.	 Wang,	Y.	J.,	Schug,	J.,	Won,	K-J.,	Liu,	C.,	Naji,	A.,	Avrahami,	D.,	Golson,	M.	L.	&	Kaestner,	K.	H.	Single	cell	transcriptomics	of	the	human	endocrine	pancreas.	Diabetes	65,	3028-3038	(2016).	227.	 Lawlor,	N.,	George,	J.,	Bolisetty,	M.,	Kursawe,	R.,	Sun,	L.,	Sivakamasundari,	V.,	Kycia,	I.,	Robson,	P.	&	Stitzel,	M.	L.	Single-cell	transcriptomes	identify	human	islet	cell	signatures	and	reveal	cell-type-specific	expression	changes	in	type	2	diabetes.	Genome	Research	27,	208–222	(2017).	228.	 Li,	J.,	Klughammer,	J.,	Farlik,	M.,	Penz,	T.,	Spittler,	A.,	Barbieux,	C.,	Berishvili,	E.,	Bock,	C.	&	Kubicek,	S.	Single-cell	transcriptomes	reveal	characteristic	features	of	human	pancreatic	islet	cell	types.	EMBO	reports	17,	178–187	(2016).	229.	 Xin,	Y.,	Kim,	J.,	Okamoto,	H.,	Ni,	M.,	Wei,	Y.,	Adler,	C.,	Murphy,	A.	J.,	Yancopoulos,	G.	D.,	Lin,	C.	&	Gromada,	J.	RNA	Sequencing	of	Single	Human	Islet	Cells	Reveals	Type	2	Diabetes	Genes.	Cell	Metabolism	24,	608–615	(2016).	230.	 Segerstolpe,	Å.,	Palasantza,	A.,	Eliasson,	P.,	Andersson,	E-M.,	Andréasson,	A-C.,	Sun,	 156 X.,	Picelli,	S.,	Sabirsh,	A.,	Clausen,	M.,	Bjursell,	M.	K.,	Smith,	D.	M.,	Kasper,	M.,	Ammala,	C.	&	Sandberg,	R.	Single-Cell	Transcriptome	Profiling	of	Human	Pancreatic	Islets	in	Health	and	Type	2	Diabetes.	Cell	Metabolism	24,	593–607	(2016).	231.	 Stanescu,	D.	E.,	Yu,	R.,	Won,	K.-J.	&	Stoffers,	D.	A.	Single	cell	transcriptomic	profiling	of	mouse	pancreatic	progenitors.	Physiol.	Genomics	49,	105–114	(2017).	232.	 Xu,	E.	E.,	Krentz,	N.	A.	J.,	Tan,	S.,	Chow,	S.	Z.,	Tang,	M.,	Nian,	C.	&	Lynn,	F.	C.	SOX4	cooperates	with	neurogenin	3	to	regulate	endocrine	pancreas	formation	in	mouse	models.	Diabetologia	58,	1013–1023	(2015).	233.	 McCarthy,	D.	J.,	Campbell,	K.	R.,	Lun,	A.	T.	L.	&	Wills,	Q.	F.	Scater:	pre-processing,	quality	control,	normalization	and	visualization	of	single-cell	RNA-seq	data	in	R.	Bioinformatics	33,	1179–1186	(2017).	234.	 Butler,	A.	&	Satija,	R.	Integrated	analysis	of	single	cell	transcriptomic	data	across	conditions,	technologies,	and	species.	bioRxiv	1–18	(2017).	235.	 Baron,	M.,	Veres,	A.,	Wolock,	S.	L.,	Faust,	A.	L.,	Gaujoux,	R.,	Vetere,	A.,	Ryu,	J.	H.,	Wagner,	B.	K.,	Shen-Orr,	S.	S.,	Klein,	A.	M.,	Melton,	D.	A.	&	Yanai,	I.	A	Single-Cell	Transcriptomic	Map	of	the	Human	and	Mouse	Pancreas	Reveals	Inter-	and	Intra-cell	Population	Structure.	Cell	Systems	3,	346–360.e4	(2016).	236.	 Huang,	H.	P.,	Liu,	M.,	El-hodiri,	H.	M.,	Chu,	K.,	Jamrich,	M.	&	Tsai,	M.	Regulation	of	the	Pancreatic	Islet-Specific	GeneBETA2	(neuroD)	by	Neurogenin	3.	Molecular	and	Cellular	Biology	20,	3292–3307	(2000).	237.	 Gage,	B.	K.,	Asadi,	A.,	Baker,	R.	K.,	Webber,	T.	D.,	Wang,	R.,	Itoh,	M.,	Hayashi,	M.,	Miyata,	R.,	Akashi,	T.	&	Kieffer,	T.	J.	The	Role	of	ARX	in	Human	Pancreatic	Endocrine	Specification.	PLoS	ONE	10,	e0144100	(2015).	238.	 Arnes,	L.,	Hill,	J.	T.,	Gross,	S.,	Magnuson,	M.	A.	&	Sussel,	L.	Ghrelin	Expression	in	the	Mouse	Pancreas	Defines	a	Unique	Multipotent	Progenitor	Population.	PLoS	ONE	7,	e52026	(2012).	239.	 van	der	Meulen,	T.,	Xie,	R.,	Kelly,	O.	G.,	Vale,	W.	W.,	Sander,	M.	&	Huising,	M.	O.	Urocortin	3	Marks	Mature	Human	Primary	and	Embryonic	Stem	Cell-Derived	Pancreatic	Alpha	and	Beta	Cells.	PLoS	ONE	7,	e52181	(2012).	240.	 Pound,	L.	D.,	Oeser,	J.	K.,	O'Brien,	T.	P.,	Wang,	Y.,	Faulman,	C.	J.,	Dadi,	P.	K.,	Jacobson,	D.	A.,	Hutton,	J.	C.,	McGuinness,	O.	P.,	Shiota,	M.	&	O'Brien,	R.	M.	G6PC2:	a	negative	regulator	of	basal	glucose-stimulated	insulin	secretion.	Diabetes	62,	1547–1556	(2013).	241.	 Zhu,	X.,	Orci,	L.,	Carroll,	R.,	Norrbom,	C.,	Ravazzola,	M.	&	Steiner,	D.	F.	Severe	block	in	processing	of	proinsulin	to	insulin	accompanied	by	elevation	of	des-64,65	proinsulin	intermediates	in	islets	of	mice	lacking	prohormone	convertase	1/3.	Proceedings	of	the	National	Academy	of	Sciences	99,	10299–10304	(2002).	242.	 Zhu,	X.,	Zhou,	A.,	Dey,	A.,	Norrbom,	C.,	Carroll,	R.,	Zhang,	C.,	Laurent,	V.,	Lindberg,	I.,	Ugleholdt,	R.,	Holst,	J.	J.	&	Steiner,	D.	F.	Disruption	of	PC1/3	expression	in	mice	causes	dwarfism	and	multiple	neuroendocrine	peptide	processing	defects.	Proceedings	of	the	National	Academy	of	Sciences	99,	10293–10298	(2002).	243.	 Nishimura,	W.,	Kondo,	T.,	Salameh,	T.,	El	Khattabi,	I.,	Dodge,	R.,	Bonner-Weir,	S.	&	Sharma,	A.	A	switch	from	MafB	to	MafA	expression	accompanies	differentiation	to	pancreatic	β-cells.	Developmental	Biology	293,	526–539	(2006).	244.	 Lu,	J.,	Chatterjee,	M.,	Schmid,	H.,	Beck,	S.	&	Gawaz,	M.	CXCL14	as	an	emerging	immune	and	inflammatory	modulator.	Journal	of	Inflammation	1–8	(2015).	 157 245.	 Ding,	H.,	Wu,	H.,	Kim,	I.,	Tam,	P.	P.	L.,	Koh,	G.	Y.	&	Nagy,	A.	The	mouse	Pdgfc	gene:	dynamic	expression	in	embryonic	tissues	during	organogenesis.	Mechanisms	of	Development	96,	209–213	(2000).	246.	 Morris,	L.,	Graham,	C.	F.	&	Gordon,	S.	Macrophages	in	haemopoietic	and	other	tissues	of	the	developing	mouse	detected	by	the	monoclonal	antibody	F4/80.	Development	112,	517–526	(1991).	247.	 Gouon-Evans,	V.,	Rothenberg,	M.	E.	&	Pollard,	J.	W.	Postnatal	mammary	gland	development	requires	macrophages	and	eosinophils.	Development	127,	2269–2282	(2000).	248.	 Henson,	P.	M.	&	Hume,	D.	A.	Apoptotic	cell	removal	in	development	and	tissue	homeostasis.	Trends	in	Immunology	27,	244–250	(2006).	249.	 Condeelis,	J.	&	Pollard,	J.	W.	Macrophages:	Obligate	Partners	for	Tumor	Cell	Migration,	Invasion,	and	Metastasis.	Cell	124,	263–266	(2006).	250.	 Geutskens,	S.	B.	Macrophages	in	the	murine	pancreas	and	their	involvement	in	fetal	endocrine	development	in	vitro.	Journal	of	Leukocyte	Biology	78,	845–852	(2005).	251.	 Banaei-Bouchareb,	L.	Insulin	cell	mass	is	altered	in	Csf1op/Csf1op	macrophage-deficient	mice.	Journal	of	Leukocyte	Biology	76,	359–367	(2004).	252.	 Mussar,	K.,	Tucker,	A.,	McLennan,	L.,	Gearhart,	A.,	Jimenez-Caliai,	A.	J.,	Cirulli,	V.	&	Crisa,	L.	Macrophage/Epithelium	Cross-Talk	Regulates	Cell	Cycle	Progression	and	Migration	in	Pancreatic	Progenitors.	PLoS	ONE	9,	e89492–16	(2014).	253.	 Kaufmann,	L.	T.,	Gierl,	M.	S.	&	Niehrs,	C.	Gadd45a,	Gadd45b	and	Gadd45g	expression	during	mouse	embryonic	development.	Gene	Expression	Patterns	11,	465–470	(2011).	254.	 la	Calle-Mustienes,	de,	E.,	Glavic,	A.,	Modolell,	J.	&	Gomez-Skarmeta,	J.	L.	Xiro	homeoproteins	coordinate	cell	cycle	exit	and	primary	neuron	formation	by	upregulating	neuronal-fate	repressors	and	downregulating	the	cell-cycle	inhibitor	XGadd45-gamma.	Mechanisms	of	Development	119,	69–80	(2002).	255.	 Huang,	H.	S.,	Kubish,	G.	M.,	Redmond,	T.	M.,	Turner,	D.	L.,	Thompson,	R.	C.,	Murphy,	G.	G.	&	Uhler,	M.	D.	Direct	transcriptional	induction	of	Gadd45γ	by	Ascl1	during	neuronal	differentiation.	Mol.	Cell.	Neurosci.	44,	282–296	(2010).	256.	 Hildesheim,	J.	&	Fornace,	A.	J.	Gadd45a:	an	elusive	yet	attractive	candidate	gene	in	pancreatic	cancer.	Clinical	Cancer	Research	8,	2475–2479	(2002).	257.	 Zhan,	Q.,	Antinore,	M.	J.,	Wang,	X.	W.,	Carrier,	F.,	Smith,	M.	L.,	Harris,	C.	C.	&	Fornace,	A.	J.	Association	with	Cdc2	and	inhibition	of	Cdc2/Cyclin	B1	kinase	activity	by	the	p53-regulated	protein	Gadd45.	Oncogene	18,	2892–2900	(1999).	258.	 Kaufmann,	L.	T.	&	Niehrs,	C.	Gadd45a	and	Gadd45g	regulate	neural	development	and	exit	from	pluripotency	in	Xenopus.	Mechanisms	of	Development	128,	401–411	(2011).	259.	 Ma,	D.	K..,	Jang,	M-H.,	Guo,	J.	U.,	Kitabatake,	Y.,	Chang,	M.,	Pow-anpongkul,	N.,	Flavell,	R.	A.,	Lu,	B.,	Ming,	G.	&	Song,	H.	Neuronal	Activity-Induced	Gadd45b	Promotes	Epigenetic	DNA	Demethylation	and	Adult	Neurogenesis.	Science	323,	1074–1077	(2009).	260.	 Guardavaccaro,	D.,	Corrente,	G.	&	Covone,	F.	Arrest	of	G1-S	progression	by	the	p53-inducible	gene	PC3	is	Rb	dependent	and	relies	on	the	inhibition	of	cyclin	D1	transcription.	Molecular	and	Cellular	Biology	(2000).	261.	 Canzoniere,	D.,	Farioli-Vecchioli,	S.,	Conti,	F.,	Ciotti,	M.	T.,	Tata,	A.	M.,	Augusti-Tocco,	 158 G.,	Mattei,	E.,	Lakshmana,	M.	K.,	Krizhanovsky,	V.,	Reeves,	S.	A.,	Giovannoni,	R.,	Castano,	F.,	Servadio,	A.,	Ben-Arie,	N.	&	Tirone,	F.	Dual	control	of	neurogenesis	by	PC3	through	cell	cycle	inhibition	and	induction	of	Math1.	Journal	of	Neuroscience	24,	3355–3369	(2004).	262.	 Farioli-Vecchioli,	S.,	Saraulli,	D.,	Costanzi,	M.,	Leonardi,	L.,	Cinà,	I.,	Micheli,	L.,	Nutini,	M.,	Longone,	P.,	Oh,	S.	P.,	Cestari,	V.	&	Tirone,	F.	Impaired	Terminal	Differentiation	of	Hippocampal	Granule	Neurons	and	Defective	Contextual	Memory	in	PC3/Tis21	Knockout	Mice.	PLoS	ONE	4,	e8339–19	(2009).	263.	 Lyden,	D.,	Young,	A.	Z.,	Zagzag,	D.,	Yan,	W.,	Gerald,	W.,	O'Reilly,	R.,	Bader,	B.	L.,	Hynes,	R.	O.,	Zhuang,	Y.,	Manova,	K.	&	Benezra,	R.	Id1	and	Id3	are	required	for	neurogenesis,	angiogenesis	and	vascularization	of	tumour	xenografts.	Nature	401,	670–677	(1999).	264.	 Tremblay,	K.	D.	&	Zaret,	K.	S.	Distinct	populations	of	endoderm	cells	converge	to	generate	the	embryonic	liver	bud	and	ventral	foregut	tissues.	Developmental	Biology	280,	87–99	(2005).	265.	 Zaret,	K.	S.,	Wandzioch,	E.,	Watts,	J.,	Xu,	J.,	Smale,	S.	T.	&	Sekiya,	T.	Pioneer	factors,	genetic	competence,	and	inductive	signaling:	programming	liver	and	pancreas	progenitors	from	the	endoderm.	Cold	Spring	Harbor	Symposia	on	Quantitative	Biology	73,	119–126	(2008).	266.	 Kieffer,	T.	J.	Closing	in	on	Mass	Production	of	Mature	Human	Beta	Cells.	Cell	Stem	Cell	18,	699–702	(2016).	267.	 Aguayo-Mazzucato,	C.	&	Bonner-Weir,	S.	Pancreatic	β	Cell	Regeneration	as	a	Possible	Therapy	for	Diabetes.	Cell	Metabolism	1–11	(2017).	268.	 Talchai,	C.,	Xuan,	S.,	Lin,	H.	V.,	Sussel,	L.	&	Accili,	D.	Pancreatic	β	Cell	Dedifferentiation	as	a	Mechanism	of	Diabetic	β	Cell	Failure.	Cell	150,	1223–1234	(2012).	269.	 Cleaver,	O.	β	Cell	Renewal	versus	Differentiation:	Slow	and	Steady	Wins	the	Race.	Developmental	Cell	41,	223–225	(2017).	270.	 Azzarelli,	R.,	Hurley,	C.,	Sznurkowska,	M.	K.,	Rulands,	S.,	Hardwick,	L.,	Gamper,	I.,	Ali,	F.,	McCracken,	L.,	Hindley,	C.,	McDuff,	F.,	Nestorowa,	S.,	Kemp,	R.,	Jones,	K.,	Gottgens,	B.,	Huch,	M.,	Evan,	G.,	Simons,	B.	D.,	Winton,	D.	&	Philpott,	A.	Multi-site	Neurogenin3	Phosphorylation	Controls	Pancreatic	Endocrine	Differentiation.	Developmental	Cell	1–19	(2017).	271.	 Sancho,	R.,	Gruber,	R.,	Gu,	G.	&	Behrens,	A.	Loss	of	Fbw7	Reprograms	Adult	Pancreatic	Ductal	Cells	into	alpha,	delta,	and	beta	Cells.	Cell	Stem	Cell	15,	139–153	(2014).	272.	 Mort,	R.	L.,	Ford,	M.	J.,	Sakaue-Sawano,	A.,	Lindstrom,	N.	O.,	Casadio,	A.,	Douglas,	A.	T.,	Keighren,	M.	A.,	Hohenstein,	P.,	Miyawaki,	A.	&	Jackson,	I.	J.	Fucci2a:	A	bicistronic	cell	cycle	reporter	that	allows	Cre	mediated	tissue	specific	expression	in	mice.	Cell	Cycle	13,	2681–2696	(2014).	273.	 Johnson,	J.	D.	The	quest	to	make	fully	functional	human	pancreatic	beta	cells	from	embryonic	stem	cells:	climbing	a	mountain	in	the	clouds.	Diabetologia	1–11	(2016).	   159 Appendix	A:	R	pipeline	for	determining	SEM	for	TC	and	TS		##	READMe	before	using	the	functions	##	Part	I		##	In	this	document,	before	each	function,	you	can	find	a	general		##	description	of	what	the	function	is	doing,	what	inputs	the	function	##	requires	and	the	meaning	of	the	function	output.		##	To	get	the	quantities	of	interest,	you	should	use	the	function	##	"getTcTsCovMtx",	i.e.	getTcTsCovMtx	=	##	function(y,	s,	x0,	Sdy,	Sds,	Sdx0,	pis,	pix,	psx)		in	part	III		##	You	do	not	need	to	read	any	functions	in	Part	II	unless	you	want		##	to	know	the	details	of	these	functions.	They	are	all	helper	functions		##	that	are	used	in	the	"getTCTsCovMtx"	function.		##	At	the	end	of	the	document,	you	can	find	two	example	uses	of		##	getTcTsCovMtx(y,	s,	x0,	Sdy,	Sds,	Sdx0,	pis,	pix,	psx)	##	with	your	provided	sample	data.		##	The	output	of	the	function	is	a	list	with	5	elements.	##	1.	SDTc	is	the	standard	error	of	Tc.		##	2.	SDTs	is	the	standard	error	of	Ts.	##	3.	covTcTs	is	the	2-by-2	covariance	matrix	of	Tc	and	Ts.	Eg.SDTc	*	SDTc	=	covTcTs[1,1]	##	4.	corrMtx	is	the	2-by-2	correlation	matrix	of	Tc	and	Ts.	##	5.	correlation	represents	the	correlation	between	Tc	and	Ts.			##	Below	is	an	example	of	the	output	(details	appear	in	Part	III).			##	sampleOutput2	#	$SDTc	#	[1]	0.7511815		#	$SDTs	#	[1]	0.4118539		#	$covTcTs	#											[,1]						[,2]	#	[1,]	0.5642737	0.2776955	#	[2,]	0.2776955	0.1696236		#	$corrMtx	#											[,1]						[,2]	#	[1,]	1.0000000	0.8975958	#	[2,]	0.8975958	1.0000000	 160 	#	$correlation	#	[1]	0.8975958		##	Part	II		##	This	part	provides	all	the	helper	functions.	You	can	ignore	##	this	part	but	please	read	part	III.		##	getCorrMtx	is	used	to	obtain	the	correlation	matrix		##	based	on	the	output	of	the	original	software.	###	Input:		##	pis:	correlation	between	intercept1	and	slope1.	##	pix:	correlation	between	intercept1	and	x0.	##	psx:	correlation	between	slope1	and	x0.		###	Output:	##	a	3-by-3	symmetric	correlation	matrix	M,	with	diagonal	elements		##	of	one	and	the	order	of	the	three	variables	is	intercept1,	slope1	##	and	x0.	For	example,	M12	=	M21	=	pis,	M13	=	M31	=	pix,	M23	=	M32	=	psx.	getCorrMtx	=	function(pis,	pix,	psx){		M	=	matrix(0,	3,	3)		diag(M)	=1			M[1,2]	=	M[2,1]	=	pis		M[1,3]	=	M[3,1]	=	pix		M[2,3]	=	M[3,2]	=	psx		return(M)	}		##	getSDVector	function	is	a	helper	function	and	used	to	organize	the		##	standard	errors	of	the	intercept1,	slope,	and	x0	into	a	vector.		getSDVector	=	function(SdIntercept,	SdSlope,	SdX0){										result	=	c(SdIntercept,	SdSlope,	SdX0)					return(result)	}		##	getCovMtx	is	used	to	obtain	the	covariance	matrix	based	on	the		##	correlation	matrix	and	the	SD	of	each	random	variable.	getCovMtx	=	function(SdVector,	corrMtx){			result	=	diag(SdVector)	%*%	corrMtx	%*%	diag(SdVector)		return(result)	}		##	getJacobianMtx	is	used	to	obtain	the	partial	derivative	of	Tc	and		##	Ts	with	respect	to	intercept1,	slope1	and	x0.	It	is	a	2-by-3	matrix	##	used	to	pre-multiply	the	covariance	matrix	of	y,	s,	x0.		 161 ##	Input:	##	y:	intercept1	obtained	from	the	original	software	##	s:	slope1	obtained	from	the	original	software	##	Output:		##	a	2-by-3	matrix	containing	the	partial	derivatives	of	Tc	and		##	Ts	with	respect	to	intercept1,	slope1	and	x0.	getJacobianMtx	=	function(y,	s){			result	=	matrix(0,	2,	3)		result[1,1]	=	1/s		result[1,2]	=	-y/(s^2)		result[1,3]	=	1		result[2,1]	=	1/s		result[2,2]	=	-y/(s^2)		result[2,3]	=	0		return(result)	}		##	getTransformedCov	is	used	to	obtain	the	covariance	matrix	of	Tc	##	and	Ts	##	Input:		##	Jacobian:	matrix	of	Tc	and	Ts	##	covMtx:	covariance	matrix	of	intercept1,	slope1	and	x0.	##	Output:	the	2-by-2	covariance	matrix	for	Tc	and	Ts		getTransformedCov	=	function(Jacobian,	covMtx){			result	=	Jacobian	%*%	covMtx	%*%	t(Jacobian)		return(result)	}		##	getCorrMtxFromCovMtx	is	used	to	obtain	the	correlation	matrix	given	the	covariance	##	matrix		##	Input:	covariance	matrix	##	Output:	the	corresponding	correlation	matrix	getCorrMtxFromCovMtx	=	function(covarianceMtx){		D	=	sqrt(diag(covarianceMtx))		DInv	=	1/D		R	=	diag(DInv)	%*%	covarianceMtx	%*%	diag(DInv)		return(R)	}		##	Part	III	##	This	is	the	function	that	will	provide	the	quantities	of		##	interest.		 162 ##	getTcTsCovMtx	is	the	overall	function	that	returns	the	covariance		##	matrix	of	Tc	and	Ts.	It	takes	as	inputs	the		##	estimates	of	the	intercept1,	slope1	and	x0	as	y,	s,	x0,	their		##	standard	errors	Sdy,	Sds,	Sdx0	and	the	pairwise	correlation.	##	Input:		##	estimates	of	intercept1,	slope1	and	x0:	y,	s,	x0;	##	Standard	errors	of	intercept1,	slope1	and	x0:	Sdy,	Sds,	Sdx0;	##	Correlations:	##	Correlation	between	intercept1	and	slope1:	pis;	##	Correlation	between	intercept1	and	x0:	pix;	##	Correlation	between	slope1	and	x0:	psx.		getTcTsCovMtx	=	function(y,	s,	x0,	Sdy,	Sds,	Sdx0,	pis,	pix,	psx){			##	get	the	correlation	matrix	of	intercept1,	slope1	and	x0		corrMtx	=	getCorrMtx(pis,	pix,	psx)		##	organize	the	standard	error	for	intercept1,	slope1	and	x0	into	a	vector		sdVector	=	getSDVector(Sdy,	Sds,	Sdx0)		##	get	the	covariance	matrix	of	intercept1,	slope1	and	x0		covMtx	=	getCovMtx(sdVector,	corrMtx)			##	get	Jacobian	matrix	of	Tc	and	Ts	with	respect	to	intercept1,			##	slope1	and	x0		Jacobian	=	getJacobianMtx(y,	s)			##	get	covariance	matrix	of	Tc	and	Ts		covTcTs	=	getTransformedCov(Jacobian,	covMtx)		SDTc	=	sqrt(covTcTs[1,1])		SDTs	=	sqrt(covTcTs[2,2])			##	get	the	correlation	matrix	of	Tc	and	Ts		corrMtxTcTs	=	getCorrMtxFromCovMtx(covTcTs)		correlation	=	corrMtxTcTs[1,2]			result	=	list(SDTc	=	SDTc,	SDTs	=	SDTs,	 		covTcTs	=	covTcTs,	 corrMtx	=	corrMtxTcTs,	correlation	=	correlation)			return(result)	}		####	Example	use	of	the	getTcTsCovMtx	function	with	sample	data	from	the		####	provided	document	y	=	0.3003;	s	=	0.1109;	x0	=	5.704;	Sdy	=	0.03037;	Sds	=	0.009489;	Sdx0	=	0.3318;		pis	=	-0.8593;	pix	=	0.4384;	psx	=	-0.7614;	sampleOutput	=	getTcTsCovMtx(y,	s,	x0,	Sdy,	Sds,	Sdx0,	pis,	pix,	psx)	 163 	sampleOutput	#	$SDTc	#	[1]	0.7379336		#	$SDTs	#	[1]	0.4875647		#	$covTcTs	#											[,1]						[,2]	#	[1,]	0.5445460	0.3360871	#	[2,]	0.3360871	0.2377193		#	$corrMtx	#											[,1]						[,2]	#	[1,]	1.0000000	0.9341191	#	[2,]	0.9341191	1.0000000		#	$correlation	#	[1]	0.9341191		##	Another	example,	using	the	other	set	of	data	in	the	document	sampleOutput2	=	getTcTsCovMtx(y=0.1173,	s=0.09813,	x0=6.376,			Sdy=0.03028,	Sds=0.009462,	Sdx0=0.4225,	pis=-0.8593,	pix=0.5199,			psx=-0.8274)	sampleOutput2		#	#	sampleOutput2	#	$SDTc	#	[1]	0.7511815		#	$SDTs	#	[1]	0.4118539		#	$covTcTs	#											[,1]						[,2]	#	[1,]	0.5642737	0.2776955	#	[2,]	0.2776955	0.1696236		#	$corrMtx	#											[,1]						[,2]	#	[1,]	1.0000000	0.8975958	#	[2,]	0.8975958	1.0000000		#	$correlation	#	[1]	0.8975958	 	 164 Appendix	B:	Analysis	of	single	cell	RNA-sequencing	data	using	Scater	R	pipeline	#Part	1:	Scater	#Generate	QC	data	#Replace	“Directory”	with	appropriate	directory	name	#Required	packages	library(cellrangerRkit)	library(scater)	library(mvoutlier)		#load	in	data	set	1	scater_E15_aggr	<-	read10XResults(data_dir	=	"file	name")	#Add	unique	identifier	to	column	for	downstream	analysis	colnames(scater_E15_aggr)=paste("E15_aggr",colnames(scater_E15_aggr),sep="_")	#filter	for	genes	not	expressed	in	any	cells	keep_feature	<-	rowSums(counts(scater_E15_aggr)	>	0)	>	0	scater_filter_E15_aggr	<-	scater_E15_aggr[keep_feature,	]	#calculate	automatic	QC	metrics	with	an	SD	of	3	scater_filter_E15_aggr	<-	scater::calculateQCMetrics(scater_filter_E15_aggr,	nmads	=	3)	#histograms	of	raw	data	hist(scater_filter_E15_aggr$total_counts,	breaks	=	100)	dev.copy2eps(file="Directory/raw_histo_totalcount.eps",	width	=	8.5,	height	=	11)	dev.off()	hist(scater_filter_E15_aggr$total_features,	breaks	=	100)	dev.copy2eps(file="Directory/raw_histo_totalfeat.eps.eps",	width	=	8.5,	height	=	11)	dev.off()	#summary	of	cells	removed	due	to	filter	of	3SD	summary(scater_filter_E15_aggr$filter_on_total_counts)	#setup	manual	filters	of	3SD	on	genes	and	counts	scater_filter_E15_aggr$use	<-	(!scater_filter_E15_aggr$filter_on_total_features	&	!scater_filter_E15_aggr$filter_on_total_counts)	#output	use	filter	scater_filter_use	<-	summary(scater_filter_E15_aggr$use)	write(scater_filter_use,	file	=	"Directory/scater_filter_use.txt")	#run	automatic	filters	based	on	default	parameters	scater_filter_E15_aggr	<-scater::plotPCA(scater_filter_E15_aggr,	size_by	=	"total_features",	shape_by	=	"use",	pca_data_input	=	"pdata",	detect_outliers	=	TRUE,return_SCESet	=	TRUE)	dev.copy2eps(file="Directory/scater_filter_pca.eps",width	=	8.5,	height	=	11)	dev.off()	#output	automatic	filter	scater_filter_outlier	<-	summary(scater_filter_E15_aggr$outlier)	write(scater_filter_outlier,	file	=	"Directory/scater_filter_outlier.txt")	#merge	both	manual	and	automatic	filters	filter_genes_E15_aggr	<-	apply(counts(scater_filter_E15_aggr[	,	pData(scater_filter_E15_aggr)$use	&	!pData(scater_filter_E15_aggr)$outlier]),	1,	function(x)	length(x[x	>	1])	>=	2)	 165 #add	the	merged	filters	to	the	$use	column	dataset	fData(scater_filter_E15_aggr)$use	<-	filter_genes_E15_aggr	#output	filtered	genes	by	manual	and	auto	outlier	scater_filter_merged	<-	summary(scater_filter_E15_aggr$use	&	!scater_filter_E15_aggr$outlier)	write(scater_filter_merged,	file	=	"Directory/scater_filter_merged.txt")	#saves	the	above	file	saveRDS(scater_filter_E15_aggr,	file="Directory/scater_E18_aggr.rds")	#apply	the	filter	with	the	$use	settings	qc_E15_aggr	<-	scater_filter_E15_aggr[fData(scater_filter_E15_aggr)$use,	pData(scater_filter_E15_aggr)$use	&	!pData(scater_filter_E15_aggr)$outlier]	#histograms	of	qc	data	hist(qc_E15_aggr$total_counts,	breaks	=	100)	dev.copy2eps(file="Desktop/qc_histo_totalcount.eps",	width	=	8.5,	height	=	11)	dev.off()	hist(qc_E15_aggr$total_features,	breaks	=	100)	dev.copy2eps(file="Desktop/qc_histo_totalfeat.eps",	width	=	8.5,	height	=	11)	dev.off()	#add	logbase2	normalized	matrix	data	set_exprs(qc_E15_aggr,	"log2_counts")	<-	log2(counts(qc_E15_aggr)	+	1)	#add	downsampled	normalized	matrix	data	#downsampling	function	Down_Sample_Matrix	<-			function	(expr_mat)				{					min_lib_size	<-	min(colSums(expr_mat))					down_sample	<-	function(x)	{							prob	<-	min_lib_size/sum(x)							return(unlist(lapply(x,	function(y)	{									rbinom(1,	y,	prob)							})))					}					down_sampled_mat	<-	apply(expr_mat,	2,	down_sample)					return(down_sampled_mat)			}	#add	downsampling	data	to	the	SCESet	norm_counts(qc_E15_aggr)	<-	log2(Down_Sample_Matrix(counts(qc_E15_aggr))	+	1)		#saves	the	above	qced	file	saveRDS(qc_E15_aggr,	file="Directory/qc_E15_aggr.rds")	#saves	the	dim	data	of	raw	and	qc	for	compare	write(dim(scater_filter_E15_aggr),	file	=	"Directory/raw_dim_E15_aggr.txt",	sep	=	"\t")	write(dim(qc_E15_aggr),	file	=	"Directory/qc_dim_E15_aggr.txt",	sep	=	"\t")		 	 166 Appendix	C:	Analysis	of	single	cell	RNA-sequencing	data	using	Seurat	pipeline	#Replace	“Directory”	with	appropriate	directory	name	#Regress	out	cell	cycle		#Generate	QC	data	#load	in	CORRECT	cellcycle	txt	file		library(Seurat)	library(Matrix)	library(dplyr)		#load	in	CORRECT	cellcycle	txt	file	cc.genes	<-	readLines(con	=	"Directory/Mouse_CC.txt")	s.genes	<-	cc.genes[1:42]	g2m.genes	<-	cc.genes[43:96]		#Run	for	dataset1	qc_E15_aggr	<-	readRDS(file="Directory/qc_E15_aggr.rds")		seurat_regress_qc2_exprs_E15_aggr	<-	CreateSeuratObject(raw.data	=	get_exprs(qc_E15_aggr,	exprs_values	=	"exprs"),	project	=	"E15_aggr",	min.cells	=	3,	min.genes	=	500,	normalization.method	=	T)	seurat_regress_qc2_exprs_E15_aggr	<-	CellCycleScoring(object	=	seurat_regress_qc2_exprs_E15_aggr,	s.genes	=	s.genes,	g2m.genes	=	g2m.genes,	set.ident	=	TRUE)	seurat_regress_qc2_exprs_E15_aggr	<-	ScaleData(object	=	seurat_regress_qc2_exprs_E15_aggr,	vars.to.regress	=	c("nUMI",	"nGene",	"S.Score",	"G2M.Score"))		seurat_regress_qc2_exprs_E15_aggr	<-	FindVariableGenes(seurat_regress_qc2_exprs_E15_aggr,	mean.function	=	ExpMean,	dispersion.function	=	LogVMR,	x.low.cutoff	=	0.0125,	y.cutoff	=	0.5,	do.contour	=	F)	dev.copy2eps(file="	Directory/regress_qc2_exprs_E15_aggr_meanvarplot.eps",	width	=	8.5,	height	=	11)	dev.copy2pdf(file="Directory/regress_qc2_exprs_E15_aggr_meanvarplot.pdf",	width	=	8.5,	height	=	11)	dev.off()	length(seurat_regress_qc2_exprs_E15_aggr@var.genes)	write(length(seurat_regress_qc2_exprs_E15_aggr@var.genes),	file	=	"Directory/seurat_regress_qc2_exprs_E15_aggr_length.txt",	sep	=	"\t")	seurat_regress_qc2_exprs_E15_aggr	<-	RunPCA(seurat_regress_qc2_exprs_E15_aggr,	pc.genes	=	seurat_regress_qc2_exprs_E15_aggr@var.genes,	do.print	=	TRUE,	pcs.print	=	5,	genes.print	=	5)	PCAPlot(seurat_regress_qc2_exprs_E15_aggr,	1,	2)	dev.copy2eps(file="Directory/regress_qc2_exprs_E15_aggr_pca.eps",	width	=	8.5,	height	=	11)	 167 dev.copy2pdf(file="Directory/regress_qc2_exprs_E15_aggr_pca.pdf",	width	=	8.5,	height	=	11)	dev.off()	PCElbowPlot(seurat_regress_qc2_exprs_E15_aggr)	dev.copy2eps(file="Directory/regress_qc2_exprs_E15_aggr_elbowplot.eps",	width	=	8.5,	height	=	11)	dev.copy2pdf(file="Directory/regress_qc2_exprs_E15_aggr_elbowplot.pdf",	width	=	8.5,	height	=	11)	dev.off()	seurat_regress_qc2_exprs_E15_aggr	<-	JackStraw(seurat_regress_qc2_exprs_E15_aggr,	num.replicate	=	100,	do.print	=	FALSE)	JackStrawPlot(seurat_regress_qc2_exprs_E15_aggr,	PCs	=	1:20)	dev.copy2eps(file="Directory/regress_qc2_exprs_E15_aggr_jackstraw.eps",	width	=	8.5,	height	=	11)	dev.copy2pdf(file="Directory/regress_qc2_exprs_E15_aggr_jackstraw.pdf",	width	=	8.5,	height	=	11)	dev.off()	seurat_regress_qc2_exprs_E15_aggr	<-	FindClusters(seurat_regress_qc2_exprs_E15_aggr,	reduction.type	=	"pca",	dims.use	=	1:15,	resolution	=	0.6,	print.output	=	0,	save.SNN	=	T)	seurat_regress_qc2_exprs_E15_aggr	<-	RunTSNE(seurat_regress_qc2_exprs_E15_aggr,	dims.use	=	1:15,	do.fast	=	T)	TSNEPlot(seurat_regress_qc2_exprs_E15_aggr)	dev.copy2eps(file="Directory/regress_qc2_exprs_E15_aggr_TSNE.eps",	width	=	8.5,	height	=	11)	dev.copy2pdf(file="Directory/regress_qc2_exprs_E15_aggr_TSNE.pdf",	width	=	8.5,	height	=	11)	dev.off()	clusters_seurat_regress_qc2_exprs_E15_aggr	<-	FindAllMarkers(seurat_regress_qc2_exprs_E15_aggr,	only.pos	=	TRUE)	write.table(clusters_seurat_regress_qc2_exprs_E15_aggr,	file	=	"Directory/clusters_seurat_regress_qc2_exprs_E15_aggr.txt",	sep	=	"\t")	#	heatmap	the	clusters	clusters_seurat_regress_qc2_exprs_E15_aggr	%>%	group_by(cluster)	%>%	top_n(10,	avg_diff)	->	top10	DoHeatmap(seurat_regress_qc2_exprs_E15_aggr,	genes.use	=	top10$gene,	order.by.ident	=	TRUE,	slim.col.label	=	TRUE,	remove.key	=	TRUE,	cex.row	=	5	)	dev.copy2eps(file="Directory/seurat_regress_qc2_exprs_E15_aggr_heatmap.eps",	width	=	8.5,	height	=	11)	dev.copy2pdf(file="Directory/seurat_regress_qc2_exprs_E15_aggr_heatmap.pdf",	width	=	8.5,	height	=	11)	dev.off()	#	save	seuratobject	saveRDS(seurat_regress_qc2_exprs_E15_aggr,	file	=	"Directory/seurat_regressCC_qc2_exprs_E15_aggr.rds")		#	generate	TSNE	with	cell	cycle	phases	overlaid	 168 head(x	=	seurat_regress_qc2_exprs_E15_aggr@meta.data)	seurat_regress_qc2_exprs_E15_aggr	<-	SetAllIdent(seurat_regress_qc2_exprs_E15_aggr,	id	=	"Phase")	seurat_regress_qc2_exprs_E15_aggr	<-	SetAll	TSNEPlot(seurat_regress_qc2_exprs_E15_aggr)	dev.copy2eps(file="Directory/regress_qc2_exprs_E15_aggrCC_TSNE.eps",	width	=	8.5,	height	=	11)	dev.copy2pdf(file="Directory/regress_qc2_exprs_E15_aggrCC_TSNE.pdf",	width	=	8.5,	height	=	11)	dev.off()	     

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0362895/manifest

Comment

Related Items