UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Cellular mechanisms of neuronal swelling underlying cytotoxic edema Rungta, Ravi Logan 2014

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

Item Metadata

Download

Media
24-ubc_2014_september_rungta_ravi.pdf [ 15.56MB ]
Metadata
JSON: 24-1.0167572.json
JSON-LD: 24-1.0167572-ld.json
RDF/XML (Pretty): 24-1.0167572-rdf.xml
RDF/JSON: 24-1.0167572-rdf.json
Turtle: 24-1.0167572-turtle.txt
N-Triples: 24-1.0167572-rdf-ntriples.txt
Original Record: 24-1.0167572-source.json
Full Text
24-1.0167572-fulltext.txt
Citation
24-1.0167572.ris

Full Text

 CELLULAR MECHANISMS OF NEURONAL SWELLING UNDERLYING CYTOTOXIC EDEMA  by  Ravi Logan Rungta B.Sc., McGill University, 2006     A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in The Faculty of Graduate and Postdoctoral Studies (Neuroscience)   The University of British Columbia (Vancouver)  July, 2014   © Ravi Logan Rungta, 2014	  	   ii	  Abstract	  	   Cytotoxic brain edema is the principal cause of mortality following brain trauma and cerebral infarct yet the mechanisms underlying neuronal swelling are poorly understood.  This thesis aims at identifying cellular mechanisms of neuronal swelling that cause cytotoxic edema (chapter 3) and describes a novel method for highly efficient neuronal transfection using lipid nanoparticle delivery of siRNA in vitro and in vivo (chapter 2).   In chapter 2, we demonstrate that neurons accumulate lipid nanoparticles in an apolipoprotein E dependent fashion, resulting in very efficient uptake in cell culture (100%) with little apparent toxicity.  In vivo, lipid nanoparticle delivery of siRNA resulted in knockdown of target genes in either discrete regions around the injection site following intracortical injections or in more widespread areas following intracerebroventricular injections with no apparent toxicity or immune reactions from the lipid nanoparticles. Effective targeted knockdown was demonstrated by showing that lipid nanoparticle delivery of siRNA against GRIN1 (encoding GluN1 subunit of the NMDA receptor) selectively reduced synaptic NMDA receptor currents in vivo as compared to synaptic AMPA receptor currents.  Therefore, lipid nanoparticle delivery of siRNA rapidly manipulates expression of proteins involved in neuronal processes in vivo, possibly enabling development of gene therapies for neurological disorders. In chapter 3, we show that increasing intracellular sodium concentration ([Na+]i) by either activating voltage-gated sodium channels or NMDA receptors triggers a secondary Cl- influx that leads to neuronal swelling and death.  Cl- but not Ca2+ entry was 	   iii	  required for neuronal swelling and cell death.  Pharmacological analyses indicated that a DIDS-sensitive HCO3-/Cl- exchanger was responsible for the majority of the Cl- influx.  We used lipid nanoparticle-siRNA mediated knockdown (described in chapter 2) to determine the molecular identity of the Cl- influx pathway.  Neuronal swelling was attenuated in brain slices by siRNA-mediated knockdown of the Cl-, SO42-, HCO3- exchanger, SLC26A11, but not by knockdown of other HCO3-/Cl- exchangers examined. We conclude that cytotoxic brain edema can occur when sufficient Na+ entry into neurons results in Cl- entry via SLC26A11 to trigger subsequent neuronal swelling. 	   iv	  Preface	  Chapter	  2	  is	  based	  on	  a	  published	  manuscript:	  Rungta, R.L., Choi H.B., Lin, P.J.C., Ko, R, Ashby, D., Jay, N., Manoharan, M., Cullis, P.R., MacVicar, B.A. (2013) Lipid nanoparticle delivery of siRNA to silence neuronal gene expression in the brain. Molecular	   therapy	  Nucleic	   acids	  2:e136. I designed the study with BAM and input from PRC, PJCL and HBC.   I performed all live cell imaging, electrophysiology, injections and analyzed respective experiments. CHB performed cell culture experiments and biochemistry assays. PJCL encapsulated siRNA in lipid nanoparticles, RK performed PTEN immunostaining, DA performed ICV injections, NJ and MM designed and synthesized lipid.  I wrote the manuscript with BAM.  All coauthors edited the manuscript. Chapter 3 is based on a manuscript under review: Rungta, R.L., Choi H.B., Tyson, J.R., Lin, P.J.C., Cullis, P.R., Snutch, T.P., MacVicar, B.A., The	  cellular	  mechanisms	  of	  neuronal	  swelling	  underlying	  cytotoxic	  edema.	   	   I	  designed	   the	  study	  with	  BAM	  and	   input	   from	  HBC,	   JRT	  and	  TPS.	   	   I	  performed	  and	  analyzed	   all	   experiments	   except	   for	   cell	   death	   assays	   and	   qPCR	   analysis.	   HBC	  performed	  cell	  death	  assays.	  	  JRT	  performed	  qPCR	  analyses.	  	  JRT	  and	  TPS	  designed	  and	  tested	  siRNAs.	  	  PJCL	  performed	  siRNA-­‐LNP	  encapsulations	  under	  supervision	  of	  PRC.	  I wrote the manuscript with BAM.	  	  	  All coauthors edited the manuscript.  All	   experimental	   protocols	   were	   approved	   by	   the	   Committee	   on	   Animal	   Care,	  University	   of	   British	   Columbia	   and	   conducted	   in	   compliance	   with	   guidelines	  provided	  by	  the	  Canadian	  Council	  of	  Animal	  Care.	  	  	  	  Certificate	  numbers:	  A11-­‐0031,	  A11-­‐0088,	  A09-­‐0933 	   v	  Table	  of	  Contents	  Abstract	  ......................................................................................................................................	  ii	  Preface	  .......................................................................................................................................	  iv	  Table	  of	  Contents	  .....................................................................................................................	  v	  List	  of	  Tables	  .........................................................................................................................	  viii	  List	  of	  Figures	  ..........................................................................................................................	  ix	  List	  of	  Abbreviations	  ...............................................................................................................	  x	  Chapter	  1:	  General	  introduction	  ........................................................................................	  1	  1.1	  Research	  Hypotheses	  and	  Objectives	  .................................................................................	  1	  1.2	  Cerebral	  Edema	  ..........................................................................................................................	  2	  1.2.1	  Introduction	  to	  Cerebral	  Edema	  ....................................................................................................	  2	  1.2.2	  Vasogenic	  Brain	  Edema	  .....................................................................................................................	  5	  1.2.3	  Cytotoxic	  Brian	  Edema	  ......................................................................................................................	  6	  1.2.4	  Medical	  Management	  and	  Treatment	  of	  Cerebral	  Edema	  ..................................................	  7	  1.3	  Volume	  Regulation	  in	  the	  Central	  Nervous	  System	  (CNS)	  .........................................	  11	  1.3.1	  Water	  Transport	  in	  the	  Brain	  ......................................................................................................	  11	  1.3.2	  Volume	  Regulation	  During	  Apoptosis	  and	  Necrosis	  ..........................................................	  16	  1.3.3	  A	  Brief	  History	  of	  Excitotoxic	  Neuronal	  Swelling	  ................................................................	  21	  1.4	  Chloride	  Homeostasis	  ............................................................................................................	  25	  1.4.1	  Introduction	  to	  Chloride	  Homeostasis	  and	  Chloride	  Equilibrium	  ...............................	  25	  1.4.2	  Aberrant	  Chloride	  Regulation	  in	  CNS	  Disorders	  .................................................................	  26	  1.5	  Chloride	  Channels	  and	  Transporters	  in	  the	  CNS	  ...........................................................	  28	  1.5.1	  CLC	  family	  of	  Chloride	  Channels	  and	  Transporters	  ...........................................................	  28	  1.5.2	  GABAA	  Receptor	  Channels	  .............................................................................................................	  29	  1.5.3	  Volume-­‐Activated	  Anion	  Channels	  ............................................................................................	  31	  1.5.4	  Calcium-­‐Activated	  Anion	  Channels	  ...........................................................................................	  34	  1.5.5	  Cation-­‐Chloride	  Cotransporters	  .................................................................................................	  35	  1.5.6	  SLC4	  Family	  of	  Anion	  Exchangers	  and	  Transporters	  ........................................................	  37	  1.5.7	  SLC26	  Family	  of	  Anion	  Exchangers	  and	  Transporters	  .....................................................	  40	  Chapter	  2:	  Lipid	  Nanoparticle	  Delivery	  of	  siRNA	  to	  Silence	  Neuronal	  Gene	  Expression	  in	  the	  Brain.	  ......................................................................................................	  43	  2.1	  Introduction	  ..............................................................................................................................	  43	  2.2	  Materials	  and	  Methods	  ..........................................................................................................	  45	  2.2.1	  	  Synthesis	  of	  the	  Lipid,	  3-­‐(dimethylamino)propyl	  3,3-­‐bis(linoleyl)	  propionate,	  DMAP-­‐BLP,	  Compound	  8	  ..........................................................................................................................	  45	  2.2.2	  Lipid	  Nanoparticle	  Formulation	  and	  siRNA	  Encapsulation	  ............................................	  52	  2.2.3	  Hippocampal	  Neuronal	  Cultures	  and	  Lipid	  Nanoparticle	  Treatments.	  .....................	  55	  2.2.4	  Intracranial	  and	  Intracerebroventricular	  	  Injections	  ........................................................	  55	  2.2.5	  Brain	  Slice	  Preparation	  ...................................................................................................................	  56	  2.2.6	  Electrophysiology	  .............................................................................................................................	  57	  2.2.7	  Imaging	  ..................................................................................................................................................	  58	  2.2.8	  Immunohistochemistry	  ..................................................................................................................	  58	  2.2.9	  Immunocytochemistry	  ....................................................................................................................	  59	  	   vi	  2.2.10	  Western	  Blotting	  .............................................................................................................................	  60	  2.2.11	  Lactate	  dehydrogenase	  (LDH)	  Assay	  .....................................................................................	  61	  2.2.12	  DiI	  Uptake	  Assay	  .............................................................................................................................	  62	  2.2.13	  Tumor	  necrosis	  factor-­‐α	  (TNF-­‐α)	  ELISA	  ..............................................................................	  62	  2.2.14	  siRNA	  Sequences	  and	  Chemistry	  .............................................................................................	  63	  2.2.15	  Statistical	  Analysis	  .........................................................................................................................	  64	  2.3	  Results	  .........................................................................................................................................	  64	  2.3.1	  Encapsulation	  of	  siRNA	  in	  Lipid	  Nanoparticles	  ...................................................................	  64	  2.3.2	  Lipid	  Nanoparticle	  Mediated	  Neuronal	  Gene	  Silencing	  in	  vitro	  ....................................	  67	  2.3.3	  Uptake	  of	  Lipid	  Nanoparticles	  by	  Neurons	  is	  Apolipoprotein	  E	  (ApoE)-­‐Dependent.	  ..............................................................................................................................................................................	  70	  2.3.4	  Lipid	  Nanoparticle	  Mediated	  Neuronal	  Gene	  Silencing	  in	  vivo	  .....................................	  72	  2.3.5	  Time	  Course	  and	  Distance	  Analysis	  of	  PTEN	  Knockdown	  in	  vivo	  ................................	  77	  2.3.6	  Lipid	  Nanoparticles	  are	  Capable	  of	  Widespread	  Distribution	  and	  Knockdown	  After	  ICV	  Administration	  ..........................................................................................................................	  80	  2.3.7	  Lipid	  Nanoparticle	  Mediated	  Knockdown	  of	  GluN1	  in	  Cell	  Culture	  ............................	  83	  2.3.8	  Functional	  Knockdown	  of	  NMDAR	  Currents	  in	  vivo	  ..........................................................	  85	  2.4	  Discussion	  ..................................................................................................................................	  88	  Chapter	  3:	  Mechanisms	  of	  neuronal	  chloride	  loading	  underlying	  cytotoxic	  edema.	  .......................................................................................................................................	  93	  3.1	  Introduction	  ..............................................................................................................................	  93	  3.2	  Materials	  and	  Methods	  ..........................................................................................................	  95	  3.2.1	  Slice	  Preparation.	  ..............................................................................................................................	  95	  3.2.2	  Imaging	  ..................................................................................................................................................	  96	  3.3.3	  Fluorescence	  Lifetime	  Imaging	  (FLIM)	  ...................................................................................	  96	  3.2.4	  Dye	  Loading	  Protocols	  ....................................................................................................................	  97	  3.2.5	  LDH	  Assay	  .............................................................................................................................................	  98	  3.2.6	  Intracranial	  Injections	  .....................................................................................................................	  99	  3.2.7	  Lipid	  Nanoparticle	  Encapsulation	  of	  siRNA	  ..........................................................................	  99	  3.2.8	  Quantitative	  PCR	  (qPCR)	  .............................................................................................................	  102	  3.2.9	  Gene	  Knock-­‐Down	  Dicer-­‐Substrate	  RNAs	  (DsiRNAs)	  .....................................................	  103	  3.2.10	  Drugs	  ..................................................................................................................................................	  104	  3.2.11	  Data	  Collection,	  Analysis	  and	  Statistics	  ..............................................................................	  105	  3.3	  Results	  ......................................................................................................................................	  106	  3.3.1	  	  Neuronal	  Swelling	  is	  Caused	  by	  Prolonged	  Increases	  in	  Intracellular	  Na+	  and	  is	  Independent	  of	  Ca2+.	  ..................................................................................................................................	  106	  3.3.2	  Na+	  Influx	  is	  Correlated	  with	  a	  Secondary	  Cl-­‐	  Influx	  that	  is	  Required	  for	  Neuronal	  Swelling.	  .........................................................................................................................................................	  114	  3.3.3	  Na+	  and	  Cl-­‐	  Dependent	  Neuronal	  Swelling	  Causes	  Death	  ..............................................	  118	  3.3.4	  Neuronal	  Swelling	  and	  Death	  Show	  the	  Pharmacological	  Profile	  of	  a	  HCO3-­‐	  /	  Cl-­‐	  Exchanger	  ......................................................................................................................................................	  119	  3.3.5	  Identification	  of	  SLC26A11	  as	  the	  Predominant	  Cl-­‐	  Influx	  Pathway	  Underlying	  Na+	  Dependent	  Cytotoxic	  Neuronal	  Swelling	  .........................................................................................	  126	  3.4	  Discussion	  ...............................................................................................................................	  133	  Chapter	  4:	  Conclusions	  .....................................................................................................	  138	  4.1	  Summary	  of	  Research	  Findings	  .......................................................................................	  138	  4.1.1	  Lipid	  Nanoparticle-­‐siRNA	  Delivery	  Mediates	  Targeted	  Knockdown	  of	  Neuronal	  Gene	  Expression	  in	  vitro	  and	  in	  vivo.	  .................................................................................................	  138	  	   vii	  4.1.2	  Fluorescence	  Lifetime	  Imaging	  of	  CoroNa	  Reports	  Intracellular	  Na+	  Concentration,	  Independent	  of	  CoroNa	  Dye	  Concentration.	  ...................................................	  139	  4.1.3	  Neuronal	  Swelling	  is	  Dependent	  on	  Na+	  and	  Cl-­‐	  Influx	  but	  Independent	  of	  Ca2+	  142	  4.1.4	  	  Identification	  of	  SLC26A11	  as	  the	  Predominant	  Neuronal	  Cl-­‐	  Influx	  Pathway	  that	  Causes	  Cytotoxic	  Edema	  ..........................................................................................................................	  143	  4.2	  Limitations	  and	  Future	  Directions.	  ................................................................................	  145	  4.2.1	  	  Enhancement	  of	  Lipid	  Nanoparticle	  Delivery	  Systems	  for	  Neuronal	  Transfection	  ............................................................................................................................................................................	  145	  4.2.2	  Does	  SLC26A11	  Regulate	  Neuronal	  Cl-­‐	  Homeostasis?	  ....................................................	  147	  4.2.3	  What	  is	  the	  Relative	  Contribution	  of	  SLC26A11	  Mediated	  Cl-­‐	  Entry	  to	  Cytotoxic	  Edema	  in	  vivo?	  .............................................................................................................................................	  148	  4.2.4	  Do	  Other	  Neuronal	  Cl-­‐	  Entry	  Pathways	  Contribute	  to	  Cytotoxic	  Edema?	  ...............	  149	  4.2.5	  Development	  of	  Specific	  Inhibitors	  Against	  SLC26A11	  ..................................................	  151	  4.3	  Clinical	  significance	  .............................................................................................................	  152	  4.3.1	  Lipid	  Nanoparticle-­‐siRNA	  Systems	  to	  Treat	  Psychiatric	  and	  Neurological	  Disorders	  ........................................................................................................................................................	  152	  4.3.2	  Treatment	  of	  Cytotoxic	  Brain	  Edema	  .....................................................................................	  153	  References:	  ...........................................................................................................................	  156	  	   viii	  List	  of	  Tables	  Table	  2.1	  	   LNP-­‐siRNA	  properties	  ……………………………………………………….……....…54	  Table	  3.1	   Characterization	   of	   LNP-­‐siRNA	   systems	  ……………………………….……101	  Table	  3.2	  	   Pharmacology	   of	   antagonists	   that	   inhibit	   chloride	   channels	   and	  chloride	   transporters…………………………………………………………………121	  	   ix	  List	  of	  Figures	  Fig.	  1.1	  	   Schematic	  demonstrating	  cytotoxic	  and	  vasogenic	  cerebral	  edema...…4	  Fig.	  1.2	  	   Molecular	  mechanism	  of	  water	  transport	  across	  cell	  membranes…...15	  Fig.	  1.3	  	   The	  morphological	  features	  of	  apoptosis,	  oncosis	  and	  necrosis….…...17	  Fig.	  2.1	  	   Schematic	  of	  LNP-­‐siRNA	  formulation	  process	  employing	  the	  staggered	  herringbone	  micromixer……………………………………………………………….66	  Fig.	  2.2	  	   LNP-­‐siRNA	   systems	   mediate	   knockdown	   of	   target	   gene	   in	   neuron	  cultures………………………………………………………………………………………69	  Fig.	  2.3	  	   LNPs	  are	  taken	  up	  by	  neurons	  in	  an	  ApoE	  dependent	  manner………...71	  Fig.	  2.4	  	   LNP-­‐siRNA	  systems	  mediate	  knockdown	  of	  target	  gene	  in	  vivo……….75	  Fig.	  2.5	  	   Distance	  profile	  and	  time	  course	  of	  protein	  knockdown	  in	  vivo……….78	  Fig.	  2.6	  	   Lipid	  nanoparticles	  (LNPs)	  are	  capable	  of	  widespread	  distribution	  and	  knockdown	   after	   intracerebroventricular	   administration………...81	  Fig.	  2.7	  	   LNP-­‐GluN1	   siRNA	   results	   in	   knockdown	   of	   the	   NMDAR	   obligatory	  subunit	   GluN1…………………………………………………………………………….84	  Fig.	  2.8	  	   Selective	   knockdown	   of	   GluN1	   protein	   in	   vivo	   results	   in	   functional	  disruption	   of	   NMDAR	   synaptic	   currents…………………………………….87	  Fig.	  3.1	  	   Neuronal	   swelling	   is	   caused	   by	   prolonged	   increases	   in	   intracellular	  Na+	  and	  is	  independent	  of	  Ca2+.	  …………………………………...…………..…..108	  Fig.	  3.2	  	   NMDAR	  activation	  triggers	  neuronal	  swelling	  that	  requires	  Na+	  influx,	  but	   that	   is	   independent	   of	   Ca2+	   influx………………………………………..110	  Fig.	  3.3	  	   Dye	   dilution	   results	   in	   decreases	   fluorescence	   intensity	   as	   neurons	  swell…………………………………………………………………………………………112	  Fig.	  3.4	  	   Biexponetial	   decay	   of	   CoroNa	   fluorescence	   indicates	   multiple	  fluorescence	   components………………………………………………….113	  Fig.	  3.5	  	   Na+	  influx	  is	  correlated	  with	  a	  secondary	  Cl-­‐	  influx	  that	  is	  required	  for	  neuronal	  swelling	  and	  causes	  cell	  death……………………………………….116	  Fig.	  3.6	  	   Neuronal	  swelling	  shows	  the	  pharmacological	  profile	  of	  a	  HCO3-­‐	  /	  Cl-­‐	  exchanger.……………………………………………………………………………...122	  Fig.	  3.7	  	   DIDS	  blocks	  Na+	  and	  Cl-­‐	  dependent,	  Ca2+	  independent	  cell	  death…...125	  Fig.	  3.8	  	   SLC26A	   and	   SLC4A	   gene	   families,	   siRNA-­‐mediated	   knockdown	   and	  expression	  profiles…….……………………………………………………..…………128	  Fig.	  3.9	  	   siRNA-­‐mediated	   knockdown	   of	   individual	   SLC4A	   family	   members	  does	  not	  alter	  the	  magnitude	  of	  neuronal	  swelling………………………130	  Fig.	  3.10	  	   Cl-­‐	   influx	   via	   SLC26A11	   causes	   cytotoxic	   neuronal	   edema	   following	  increased	   [Na+]i…………………………….…………………………………………….131	  	  	  	   x	  List	  of	  Abbreviations	  AE	   	   Anion	  exchanger	  ACSF	   	   Artificial	  cerebrospinal	  fluid	  AMPA	   	   α-­‐Amino-­‐3-­‐hydroxy-­‐5-­‐methyl-­‐4-­‐isoxazolepropionic	  acid	  ANOVA	   Analysis	  of	  variance	  ApoE	   	   Apolipoprotein	  E	  AQP	   	   Aquaporin	  ASO	   	   Antisense	  oligonucleotide	  ATP	   	   Adenosine	  triphosphate	  AVD	   	   Apoptotic	  volume	  decrease	  BBB	   	   Blood	  brain	  barrier	  BCECF	  	   2’,7’-­‐bis(2-­‐carboxymethyl),5(and-­‐6)carboxyfluorescein	  BDNF	   	   Brain-­‐derived	  neurotrophic	  factor	  CaCC	   	   Calcium-­‐activated	  chloride	  channels	  CNS	   	   Central	  nervous	  system	  CSF	   	   Cerebrospinal	  fluid	  DiI	   	   1,1'-­‐dioctadecyl-­‐3,3,3',3'-­‐tetramethylindocarbocyanine	  perchlorate	  DIDS	   	   4,4'-­‐Diisothiocyano-­‐2,2'-­‐stilbenedisulfonic	  acid	  dsRNA	  	   Double-­‐stranded	  RNA	  EAAT	   	   Excitatory	  amino	  acid	  transporter	  EGTA	   	   Ethylene	  glycol	  tetraacetic	  acid	  EPSC	   	   Excitatory	  postsynaptic	  current	  Em	   	   Resting	  membrane	  potential	  ECl-­‐	   	   Chloride	  equilibrium	  potential	  FLIM	   	   Fluorescence	  lifetime	  imaging	  GABA	   	   Gamma-­‐aminobutyric	  acid	  GAT	   	   GABA	  transporter	  HEK	   	   Human	  embryonic	  kidney	  ICP	   	   Intracranial	  pressure	  ICV	   	   Intracerebroventricular	  IRF	   	   Instrument	  response	  function	  I-­‐V	   	   Current-­‐voltage	  KBAT	   	   Kidney	  brain	  anion	  transporter	  KO	   	   Knock	  out	  LDH	   	   Lactate	  dehydrogenase	  LNP	  	   	   Lipid	  nanoparticle	  Luc	   	   Luciferase	  MCAO	   	   Middle	  cerebral	  artery	  occlusion	  MMP	   	   Matrix	  metalloproteinase	  MQAE	   	   N-­‐(Ethoxycarbonylmethyl)-­‐6-­‐methoxyquinolinium	  bromide	  NDCBE	   Na+	  driven	  Cl-­‐/HCO3-­‐	  exchanger	  NMDA	  	   N-­‐methyl-­‐D-­‐aspartate	  NMDG	  	   N-­‐methyl-­‐D-­‐glucamine	  NPPB	   	   5-­‐Nitro-­‐2-­‐(3-­‐phenylpropylamino)benzoic	  acid	  	   xi	  OGD	   	   Oxygen	  glucose	  deprivation	  PEG	   	   Polyethylene	  glycol	  PTEN	   	   Phosphatase	  and	  tensin	  homolog	  1	  R	   	   Receptor	  RNAi	   	   RNA	  interference	  RVD	   	   Regulatory	  volume	  decrease	  SBFI	   	   Sodium-­‐binding	  benzofuran	  isophthalate	  SD	   	   Spreading	  depression	  	  siRNA	   	   Small	  interfering	  RNA	  shRNA	  	   Short	  hairpin	  RNA	  TBI	   	   Traumatic	  brain	  injury	  TNF-­‐α	  	   Tumor	  necrosis	  factor-­‐α	  TPSLM	   Two	  photon	  scanning	  laser	  microscopy	  TRP	   	   Transient	  receptor	  potential	  VAAC	   	   Volume-­‐activated	  anion	  channel	  VGCC	   	   Voltage-­‐gated	  calcium	  channel	  VGSC	   	   Voltage-­‐gated	  sodium	  channel	  VRAC	   	   Volume-­‐regulated	  anion	  channel	  VSOR	   	   Volume-­‐sensitive	  outward	  rectifying	  Cl-­‐	  channel	   1	  Chapter	  1:	  General	  introduction	  	  1.1	  Research	  Hypotheses	  and	  Objectives	  	  Hypotheses:	  	  1)	   Lipid	   nanoparticle	   (LNP)	   delivery	   of	   siRNA	   is	   a	   highly	   efficient	   and	   non-­‐toxic	  method	  to	  attenuate	  neuronal	  gene	  expression	  in	  vitro	  and	  in	  vivo.	  	  2)	  Prolonged	   increases	   in	   intracellular	  sodium	  ([Na+]i)	   cause	  a	  secondary	   influx	  of	  Cl-­‐	   via	   an	   unknown	   pathway	   that	   is	   required	   for	   neuronal	   swelling	   underlying	  cytotoxic	  brain	  edema.	  	  Aim	  1:	  Develop	  new	  methodology	  for	  rapid	  transfection	  of	  neurons	  in	  vitro	  and	  in	  vivo.	  	  	  Aim	  2:	  Examine	  the	  interrelationship	  between	  neuronal	  [Na+]i,	  [Cl-­‐]i	  and	  volume	  in	  order	   to	   investigate	   roles	   for	   Cl-­‐	   entry	   pathways	   that	   contribute	   to	   neuronal	  swelling.	  	  	  Aim	  3:	  Use	  newly	  developed	  LNP	  transfection	  method	  to	  screen	  candidate	  proteins,	  and	   identify	   the	   predominant	   neuronal	   chloride	   entry	   pathway(s)	   underlying	  cytotoxic	  neuronal	  edema.	  	   2	  1.2	  Cerebral	  Edema	  	  1.2.1	  Introduction	  to	  Cerebral	  Edema	  	   Cerebral	   Edema	   is	   the	   process	   by	   which	   excess	   water	   accumulates	   in	   the	  brain	   and	   causes	   an	   increase	   in	   intracranial	   pressure	   (ICP).	   	   Cerebral	   edema	   is	  potentially	   lethal	   and	   can	   be	   caused	   by	   a	   variety	   of	   intracranial	   insults	   such	   as	  hemorrhage,	   ischemic	   stroke,	   infection,	   neoplasm,	   tumor,	   traumatic	   brain	   injury	  (TBI)	   or	   hypoxia	   (Klatzo,	   1987;	   Quagliarello	   and	   Scheld,	   1992;	   Kimelberg,	   1995;	  Marmarou	  et	   al.,	   2000;	  Wick	  and	  Kuker,	   2004;	  Marmarou	  et	   al.,	   2006;	  Raslan	  and	  Bhardwaj,	   2007;	   Simard	   et	   al.,	   2007).	   	   Brain	   edema	   was	   first	   classified	   as	   either	  vasogenic	  or	  cytotoxic	  by	  Igor	  Katzo	  in	  1967	  (Klatzo,	  1967)	  and	  was	  later	  expanded	  to	   include	   interstitial	   and	   osmotic	   edema	   (Fishman,	   1975).	   	   Whereas	   vasogenic	  edema	   is	   characterized	   by	   water	   entry	   into	   the	   brain	   from	   the	   blood,	   due	   to	  breakdown	   of	   the	   blood	   brain	   barrier	   (BBB),	   cytotoxic	   edema	   is	   caused	   by	  water	  entry	  from	  the	  extracellular	  to	  intracellular	  space	  of	  neurons	  or	  astrocytes	  (Klatzo,	  1987;	   Donkin	   and	   Vink,	   2010)	   (Figure	   1.1).	   	   Interstitial	   edema	   occurs	   in	  hydrocephalus	   and	   refers	   to	   the	   abnormal	  movement	   of	   cerebrospinal	   fluid	   (CSF)	  from	  the	  ventricles	   into	  the	   interstitial	  space,	  and	  differs	   from	  vasogenic	  edema	  in	  that	  CSF	   contains	   relatively	   low	   levels	   of	   proteins	   compared	   to	   the	  blood	   (Redzic,	  2011).	  	  Osmotic	  edema	  is	  driven	  by	  an	  osmotic	  imbalance	  between	  the	  plasma	  and	  CSF	  and	  differs	  from	  vasogenic	  edema	  in	  that	  water	  flux	  into	  the	  brain	  occurs	  across	  an	  intact	  BBB.	  	  	   3	  	   The	  total	  brain	  volume	  is	  made	  up	  of	  3	  compartments;	  the	  parenchyma,	  the	  blood,	  and	  the	  CSF.	  	  Therefore,	  total	  brain	  volume	  can	  be	  calculated	  by	  the	  following	  equation:	   (Brain	  Volume)total=	   (Cellular	  volume)	  +	   (Blood	  volume)	  +	   (CSF	  volume)	  (Kimelberg,	  2004;	  Wakeland	  and	  Goldstein,	  2005).	  Brain	  edema	  therefore	  occurs	  if	  any	   of	   these	   components	   increase	   in	   volume	   without	   an	   equal	   volume	   decrease	  within	  another	  component.	  	  The	  skull,	  a	  rigid	  encasement,	  imposes	  a	  fixed	  boundary	  beyond	   which	   the	   brain	   is	   not	   able	   to	   swell	   and	   there	   is	   very	   little	   room	   for	  expansion	  of	  tissue	  volume	  without	  causing	  significant	  brain	  damage	  (Simard	  et	  al.,	  2007).	  Cerebral	  edema	  can	  be	  fatal	  if	  herniation	  of	  the	  brain	  towards	  the	  brain	  stem	  occurs,	   resulting	   in	   loss	  of	   the	   respiratory	   centers	   and	  death.	   	  Although	  vasogenic	  and	   cytotoxic	   edema	   can	   theoretically	   occur	   in	   isolation,	   cerebral	   edema	   is	   most	  often	  composed	  of	  both	  cytotoxic	  and	  vasogenic	  components	  occurring	  at	  different	  time	   points.	   	   The	   initial	   disturbance	   that	   triggers	   the	   edema	   is	   often	   complex,	  sometimes	   generating	   both	   types	   of	   edema	   in	   parallel	   and	   even	   when	   edema	  originates	  as	  either	  purely	  cytotoxic	  or	  vasogenic,	  one	  type	  of	  edema	  will	  ultimately	  drive	   the	   other	   (Kimelberg,	   1995;	   Marmarou	   et	   al.,	   2006;	   Papadopoulos	   and	  Verkman,	  2007;	  Simard	  et	  al.,	  2007;	  Donkin	  and	  Vink,	  2010).	  	  	  	  	   4	  	  Figure	  1.1:	  Schematic	  demonstrating	  cytotoxic	  and	  vasogenic	  cerebral	  edema.	  	  Cytotoxic	  edema	  is	  essentially	  a	  water	  compartment	  shift	  with	  no	  change	   in	  tissue	  water	   content	   or	   volume.	   In	   contrast,	   vasogenic	   edema	   increases	   tissue	   water	  content,	  leading	  to	  swelling.	  Tissue	  swelling	  thus	  requires	  a	  vascular	  contribution	  if	  it	   is	   to	   occur.	   	   From	   (Donkin	   and	   Vink,	   2010),	   reprinted	   with	   permission	   from	  Wolters	  Kluwer	  Health.	  	   5	  1.2.2	  Vasogenic	  Brain	  Edema	  	   Vasogenic	   brain	   edema	   is	   caused	   by	   breakdown	   of	   the	   BBB	   that	   occurs	  following	  either	  physical	  disruption	  or	  the	  release	  of	  vasogenic	  compounds	  (Donkin	  and	   Vink,	   2010).	   	   	   The	   BBB	   is	   made	   up	   of	   intercellular	   tight	   junctions	   between	  endothelial	   cells	   that	   line	   the	   blood	   vessels	   and	   capillaries	   of	   the	   central	   nervous	  system.	  	  	  Specialized	  transport	  systems	  therefore	  must	  exist	  on	  the	  membrane	  of	  the	  endothelial	   cell	   to	   move	   polar	   substances	   such	   as	   glucose	   from	   the	   blood	   to	   the	  brain	  and	  vice	  versa.	  	  In	  the	  healthy	  brain,	  water	  is	  efficiently	  kept	  out	  of	  the	  brain	  by	  the	  osmotic	  force	  imposed	  from	  retained	  ions	  and	  plasma	  proteins	  in	  the	  blood	  (Kimelberg,	  1995).	  	  	  When	  BBB	  breakdown	  occurs,	  solutes	  are	  no	  longer	  retained	  in	  the	  vessel;	  the	  blood	  filtrate,	  including	  plasma	  proteins	  are	  driven	  into	  the	  brain	  by	  hydrostatic	  pressure	  and	  water	  is	  no	  longer	  retained	  in	  the	  vessel	  (Nagashima	  et	  al.,	  1990).	   	   	   	   In	  humans	  vasogenic	  edema	  primarily	  affects	  white	  matter	  due	  to	  higher	  compliance	   associated	   with	   the	   multiple	   unconnected	   parallel	   axonal	   tracts	  (Marmarou,	  2007).	  Several	  inflammatory	  and	  vasoactive	  factors	  have	  been	  reported	  to	   cause	   breakdown	   of	   the	   BBB	   (Abbott,	   2000).	   	   Neurogenic	   inflammation	   that	  occurs	   following	   cytotoxic	   edema	   has	   been	   shown	   to	   result	   in	   the	   release	   of	  vasoactive	   agents	   that	   can	   increase	   vascular	   permeability	   such	   as	   such	   as	  bradykinins	  and	  substance	  P	  (Donkin	  and	  Vink,	  2010).	  	  The	  release	  of	  such	  factors	  is	  exacerbated	   if	   necrotic	   cell	   death	   occurs.	   	   Matrix	   metalloproteinases	   (MMPs)	  represent	   another	   interesting	   group	   of	   proteins	   that	  may	   contribute	   to	   vasogenic	  edema	   (Candelario-­‐Jalil	   et	   al.,	   2009).	   	   MMPs	   can	   degrade	   extracellular	   matrix	  	   6	  proteins	  including	  tight	  junctions	  of	  the	  BBB.	  	  MMP	  inhibitors	  have	  been	  reported	  to	  reduce	   BBB	   injury,	   cerebral	   edema	   and	   cell	   death	   in	   animal	   models	   of	   TBI	   and	  ischemia	  (Candelario-­‐Jalil	  et	  al.,	  2009;	  Homsi	  et	  al.,	  2009;	  Tejima	  et	  al.,	  2009),	  and	  transgenic	  mice	  lacking	  MMP-­‐9	  have	  been	  shown	  to	  be	  protected	  from	  ischemia	  and	  traumatic	   brain	   injury	   with	   reduced	   edema	   and	   improved	   functional	   outcome	  (Wang	  et	  al.,	  2000;	  Asahi	  et	  al.,	  2001).	  	  1.2.3	  Cytotoxic	  Brian	  Edema	  	  	   Cytotoxic	   brain	   edema	   is	   the	   principal	   cause	   of	   mortality	   from	   stroke	   and	  traumatic	   brain	   injury	   (TBI)	   although	   the	   cellular	   mechanisms	   remain	   largely	  unknown	  (Klatzo,	  1987;	  Marmarou	  et	  al.,	  2000;	  Rosenblum,	  2007;	  Donkin	  and	  Vink,	  2010).	   	   	  The	  cellular	  process	  of	  cell	  swelling	  termed,	  cytotoxic	  brain	  edema	  begins	  with	   the	   extracellular	   to	   intracellular	   movement	   of	   Na+	   and	   other	   cations	   into	  neurons	   and/or	   astrocytes	   (Liang	   et	   al.,	   2007).	   	   This	   occurs	   via	   transmembrane	  cation	   channels	   and	   transporters	   in	   combination	   with	   the	   failure	   of	   energy	  dependent	  extrusion	  mechanisms.	   	  The	   influx	  of	  cations	   then	   leads	   to	  a	  secondary	  influx	  of	  anions	  (most	  likely	  Cl-­‐)	  and	  osmotically	  obliged	  water	  thereby	  causing	  cell	  swelling.	   	   Although	   it	   is	   possible	   in	   principle	   for	   cytotoxic	   brain	   edema	   to	   occur	  without	  a	  change	  in	  brain	  volume,	  and	  solely	  a	  decrease	  in	  extracellular	  volume	  this	  is	   not	   the	   case	   in	   vivo.	   	   The	   depletion	   of	   extracellular	   ions	   will	   form	   new	  concentration	   gradients	   across	   the	   BBB	   and	   thereby	   trigger	   the	   transport	   of	   ions	  from	  the	  blood	  into	  the	  brain	  in	  order	  to	  replace	  them	  along	  with	  water	  that	  causes	  	   7	  an	   increase	   in	   brain	   volume	   (Liang	   et	   al.,	   2007;	   Simard	   et	   al.,	   2007).	   	   In	   addition,	  regulatory	  volume	  decrease	  (RVD)	  mechanisms	  are	  often	  activated	  in	  neurons	  and	  astrocytes	   in	   an	   attempt	   to	   compensate	   for	   and	   counteract	   cell	   volume	   increases.	  	  RVD	   in	   astrocytes	   is	   associated	   with	   the	   swelling-­‐induced	   activation	   of	   channels	  such	  as	  the	  volume	  regulated	  anion	  channel	  (VRAC)	  that	  can	  release	  osmolytes	  such	  as	   taurine	   into	   the	   extracellular	   space	   (Phillis	   et	   al.,	   1997;	   Choe	   et	   al.,	   2012).	   	   As	  these	   osmolytes	   are	   not	   normally	   present	   in	   the	   extracellular	   space,	   they	   may	  contribute	   to	  new	  osmotic	   gradients	   across	   the	  BBB	   that	  will	   thereby	  draw	  water	  into	  the	  brain	  from	  the	  blood.	  	  	  Regardless	   of	  whether	   brain	   edema	   originates	   from	   cytotoxic	   or	   vasogenic	  mechanisms,	   the	   increased	   intracranial	   pressure	   often	   leads	   to	   constriction	  of	   the	  cerebral	  vasculature	  and	  a	  resultant	  decrease	  in	  blood	  flow	  causing	  local	   ischemia.	  	  The	   decrease	   in	   energy	   supply	   to	   neurons	   and	   astrocytes	   results	   in	   failure	   to	  maintain	   their	   cationic	   gradients	   via	   adenosine	   triphosphate	   (ATP)	   driven	   pumps	  such	   as	   the	   Na+/K+	   ATPase	   and	   generation	   or	   exacerbation	   of	   cytotoxic	   edema	  (Kimelberg,	  1995).	  	  	  	  	  1.2.4	  Medical	  Management	  and	  Treatment	  of	  Cerebral	  Edema	  	   As	  of	  today	  the	  treatment	  of	  cytotoxic	  edema	  is	  mainly	  symptomatic,	  with	  the	  major	  focus	  being	  on	  decreasing	  intracranial	  pressure	  (ICP)	  (Unterberg	  et	  al.,	  2004).	  	  Osmotherapy,	  defined	  as	  the	  elevation	  of	  plasma	  osmolarity,	  is	  undertaken	  to	  draw	  water	   from	   the	  extracellular	  CSF	  of	  brain	  back	  across	   the	  BBB	   into	   the	  blood	   in	   a	  	   8	  process	  called	  brain	  dehydration.	  	  	  The	  first	  attempts	  at	  manipulating	  brain	  volume	  by	  osmotic	  manipulations	  were	  done	  almost	  100	  years	  ago	  (Weed	  and	  McKibbens,	  1919).	   However,	   initial	   clinical	   attempts	   were	   largely	   unsuccessful	   due	   to	   the	  adverse	  effects	  of	  the	  concentrated	  urea	  or	  human	  plasma	  proteins	  that	  were	  used	  as	  an	  osmolyte	  (Paczynski,	  1997;	  Raslan	  and	  Bhardwaj,	  2007).	   	   In	  1960	  the	  use	  of	  mannitol	  was	  introduced	  and	  remains	  the	  osmolyte	  of	  choice	  today	  (Shenkin	  et	  al.,	  1962;	  Fink,	  2012).	  	  Although	  osmotherapy	  has	  been	  used	  in	  the	  clinical	  setting	  as	  an	  effective	  method	  for	  decreasing	  brain	  volume	  and	  ICP	  (Shenkin	  et	  al.,	  1962;	  Diringer	  and	  Zazulia,	  2004),	  it’s	  usefulness	  remains	  controversial	  and	  poorly	  understood	  for	  the	  several	  reasons.	  	  First,	  in	  cases	  of	  vasogenic	  edema	  with	  severe	  BBB	  disruption	  mannitol	   can	   accumulate	   in	   the	   injured	   brain	   and	   thereby	   worsen	   the	   edema,	   as	  previously	   reported	   (Kaufmann	   and	   Cardoso,	   1992;	   Bereczki	   et	   al.,	   2000),	   which	  may	   contribute	   to	   a	   rebound	   effect	   (Palma	   et	   al.,	   2006).	   	   Second,	   mannitol	  administration	   in	   certain	   cases	   has	   been	   reported	   to	   shrink	   only	   the	   non-­‐swollen	  regions	   of	   the	   brain	   (Hartwell	   and	   Sutton,	   1993;	   Diringer	   and	   Zazulia,	   2004).	  Although,	   this	   will	   still	   reduce	   ICP	   it	   offers	   limited	   therapeutic	   benefit	   to	   the	  damaged	   tissue.	   	   	   Third,	   healthy	   cells	   individually	   compensate	   for	   the	  decrease	   in	  volume	  by	  triggering	  unknown	  regulatory	  volume	  increase	  (RVI)	  mechanisms,	  and	  this	  may	  contribute	  to	  a	  rebound	  effect	  when	  mannitol	  is	  withdrawn	  (McManus	  and	  Soriano,	  1998).	   	  It	  is	  interesting	  to	  note	  that	  mannitol	  also	  exerts	  biological	  effects,	  and	  it	  is	  possible	  that	  some	  of	  the	  reported	  positive	  actions	  could	  be	  a	  result	  of	  off	  target	  effects	  such	  as	  free	  radical	  scavenging	  (Shen	  et	  al.,	  1997).	  	   9	  Controlled	  hyperventilation	  can	  be	  used	  to	  decrease	  ICP	  by	  causing	  cerebral	  vasoconstriction	   and	   thereby	   a	   reduction	   the	   blood	   volume.	   	   However,	  hyperventilation	  can	  only	  be	  used	  acutely	  as	  it	  reduces	  brain	  tissue	  PO2	  (Rabinstein,	  2006)	  and	  may	  cause	  secondary	  ischemic	  damage	  (Muizelaar	  et	  al.,	  1991),	   limiting	  it’s	   usability.	   	   Barbiturates	   also	   act	   to	   decrease	   ICP	   by	   decreasing	   the	   cerebral	  blood	   volume	   as	   a	   result	   of	   lowering	   the	   metabolic	   demands	   of	   the	   brain.	  	  Barbituates	   are	   used	   in	   cases	  where	   the	   increases	   in	   ICP	   are	   life	   threatening	   and	  uncontrollable	   (Eisenberg	   et	   al.,	   1988;	   Raslan	   and	   Bhardwaj,	   2007),	   however,	  whether	  they	  improve	  functional	  outcome	  remains	  controversial	  (Ward	  et	  al.,	  1985;	  Schwab	  et	  al.,	  1997;	  Rabinstein,	  2006).	  	  	  	  Glucocorticoids	  are	  used	  primarily	  in	  the	  treatment	  of	  vasogenic	  edema	  that	  accompanies	   tumors,	   and	   preceding	   neurosurgical	   manipulation	   (French	   and	  Galicich,	   1964;	   Rabinstein,	   2006).	   	   Glucocorticoids	   decrease	   tight-­‐juntion	  permeability,	  thereby	  increasing	  vascular	  water	  retention	  across	  the	  BBB,	  however	  the	   precise	   mechanisms	   of	   action	   are	   unknown	   (Raslan	   and	   Bhardwaj,	   2007).	  	  Unfortunately	   in	   clinical	   trials	   for	   ischemic	   stroke,	   hemorrhage	   and	   head	   injury,	  corticosteroids	  had	  no	  beneficial	  effect	  in	  treating	  edema,	  and	  may	  have	  even	  been	  harmful	   in	   some	   cases	   (Dearden	   et	   al.,	   1986;	  Qizilbash	   et	   al.,	   2002;	  Roberts	   et	   al.,	  2004;	   Gomes	   et	   al.,	   2005).	   Steroids	   have	   no	   beneficial	   effect	   in	   the	   treatment	   of	  cytotoxic	  brain	  edema.	  Hypothermia	   lowers	   the	   metabolic	   demand	   of	   the	   brain	   and	   has	   been	  proven	   to	   reduce	   ICP	   in	   animal	   models	   and	   human	   patients	   (Choi	   et	   al.,	   2012a;	  Yenari	  and	  Han,	  2012).	  The	  reduction	  in	  ICP	  is	  in	  part	  due	  to	  a	  reduction	  in	  cerebral	  	   10	  blood	   volume,	   but	  may	   also	   involve	   other	  mechanisms	   such	   as	   a	   reduction	   in	   the	  release	  of	   inflammatory	   factors	   that	   increase	  BBB	  permeability	   and	  by	  decreasing	  the	   rate	   of	   cellular	   energy	   consumption.	   Despite	   promising	   results	   of	   many	  observational	   studies	  and	  phase	   II	   clinical	   trials	   (Polderman,	  2008),	   several	  phase	  III	  trials	  failed	  to	  demonstrate	  any	  benefit	  of	  hypothermia	  following	  TBI	  (Clifton	  et	  al.,	  2001;	  Hutchison	  et	  al.,	  2008).	  	  In	  addition	  to	  the	  potential	  systemic	  complications	  a	  major	  concern	  of	  these	  studies	  was	  that	  rewarming	  occurred	  after	  24	  to	  48	  hours,	  the	   time	  when	  brain	  edema	   is	  maximal	   (Choi	   et	   al.,	   2012a).	   	   	   Currently,	   cooling	   is	  generally	   only	   used	   to	   treat	   brain	   edema	  when	   other	  methods	   to	   reduce	   ICP	   are	  unsuccessful.	   	   Further	   studies	   to	   identify	   the	   optimal	   conditions	   and	   timing	   of	  hypothermia	   in	   the	   treatment	   of	   brain	   edema	   are	   needed	   to	   validate	  widespread	  clinical	   use.	   	   Hypothermia	   may	   be	   further	   beneficial	   when	   combined	   with	   other	  neuroprotective	  strategies.	  Decompressive	   craniectomy,	   the	   removal	   of	   a	   portion	   of	   the	   skull,	   is	   a	  controversial	  surgery	  used	  to	  reduce	  ICP	  in	  life	  threatening	  cases	  of	  cerebral	  edema.	  	  Although	  it	  is	  clear	  that	  craniectomies	  reduce	  ICP	  and	  can	  be	  life	  saving,	  a	  beneficial	  functional	  outcome	  on	  survivors	  remains	  unproven,	  which	  may	  partly	  be	  explained	  by	  age	  of	  the	  patient	  and	  the	  time	  of	  surgery	  (Taylor	  et	  al.,	  2001;	  Rabinstein	  et	  al.,	  2006;	  Plesnila,	  2007).	  Decompressive	   craniectomy	   is	  usually	   reserved	   for	  patients	  who	   fail	   to	   respond	   to	   all	   other	   therapeutic	   measures,	   a	   time	   delay	   which	  unfortunately	  may	  decrease	  the	  efficacy	  of	  the	  operation	  (Rabinstein,	  2006;	  Plesnila,	  2007).	  	  	  	   11	  Although	   treatments	   (outlined	   above)	   exist	   to	   reduce	   the	   ICP	   elevations	  associated	   with	   cytotoxic	   edema,	   they	   remain	   largely	   empirical,	   and	   a	   better	  understanding	   of	   the	   underlying	   cellular	   and	   molecular	   mechanisms	   is	  needed	   to	   identify	   new	   and	   more	   effective	   forms	   of	   treatment	   (Rabinstein,	  2006;	  Marmarou,	  2007;	  Raslan	  and	  Bhardwaj,	  2007).	  	  1.3	  Volume	  Regulation	  in	  the	  Central	  Nervous	  System	  (CNS)	  	  1.3.1	  Water	  Transport	  in	  the	  Brain	  	   Aquaporins	   (AQPs),	   discovered	   in	   1992	   (Preston	   et	   al.,	   1992),	   are	  transmembrane	   proteins	   that	   function	   as	   water	   channels.	   AQP1,	   AQP4	   and	   AQP9	  isoforms	   are	   expressed	   in	   the	   brain,	   with	   AQP4	   showing	   the	   highest	   levels	   of	  expression	   (Amiry-­‐Moghaddam	   and	   Ottersen,	   2003).	   	   AQP4	   expression	   is	   highly	  polarized	  and	  is	  found	  to	  be	  enriched	  on	  astrocyte	  endfeet	  that	  are	  in	  direct	  contact	  with	  blood	  vessels	  (Nielsen	  et	  al.,	  1997;	  Rash	  et	  al.,	  1998),	  suggesting	  involvement	  in	  cerebral	   water	   and	   volume	   homeostasis.	   AQP4	   expression	   has	   been	   shown	   to	   be	  colocalized	  with	  Kir4.1	  (Nagelhus	  et	  al.,	  1999),	  and	  deletion	  of	  AQP4	  in	  mice	  results	  in	  increased	  seizure	  duration	  (Binder	  et	  al.,	  2006a;	  Binder	  et	  al.,	  2006b),	  suggesting	  a	   possible	   role	   for	   AQP4	   water	   influx	   coupled	   to	   local	   K+	   buffering	   following	  neuronal	  activation.	  	  AQP4	  KO	  mice	  have	  recently	  been	  shown	  to	  reduce	  the	  rate	  at	  which	   interstitial	   solutes	   are	   cleared	   from	   the	   brain	   (Iliff	   et	   al.,	   2012;	   Rangroo	  Thrane	   et	   al.,	   2013),	   and	   AQP4	   may	   therefore	   be	   important	   for	   the	   clearance	   of	  	   12	  metabolites	   from	  the	  brain	  during	  sleep	  (Xie	  et	  al.,	  2013).	   	  However,	  AQPs	  are	  not	  the	  only	  transmembrane	  proteins	  capable	  of	  water	  transport	  but	  rather	  increase	  the	  energetic	  efficiency	  and	  rate	  of	  water	  transport	  (Zeuthen,	  2010).	  	  This	  may	  explain	  the	  lack	  of	  an	  observable	  phenotype	  first	  described	  in	  AQP4	  KO	  mice	  (Manley	  et	  al.,	  2000).	   	   	   In	   line	  with	   this,	   a	   quadruple	   knockout	   of	   all	   AQPs	   found	   present	   in	   the	  osmoregulatory	   tissues	   of	   caenorhabditis	   elegans	   had	   no	   effect	   on	   survival,	  development,	  growth,	  fertility	  or	  movement	  in	  normal	  or	  hypertonic	  media	  (Huang	  et	   al.,	   2007).	   	   The	   role	   of	   AQPs	   seems	   to	   generally	   become	   more	   clear	   under	  conditions	  of	  stress,	  for	  example	  the	  inability	  of	  AQP1	  KO	  mice	  to	  concentrate	  urine	  only	  becomes	  apparent	  after	  36	  hours	  of	  water	  deprivation	  (Ma	  et	  al.,	  1998).	  	  	  Due	  to	  the	  high	  expression	  of	  AQP4	  in	  astrocyte	  endfeet	  at	  the	  interface	  of	  astrocytes	  with	  endothelial	   cells	   forming	   the	   blood	   brain	   barrier,	   it	   seems	   likely	   that	   AQP4	   may	  regulate	  the	  rate	  of	  water	  entry	  into	  the	  brain.	  	  This	  was	  confirmed	  in	  AQP4	  KO	  mice	  following	   intraperitoneal	   (IP)	  distilled	  water	   infusions	  equal	   to	  20%	  of	   the	  mouse	  body	  weight.	   	  Whereas	  only	  8%	  of	  wild	   type	  mice	   survived,	   60%	  of	   the	  AQP4	  KO	  mice	   survived	   the	   insult	   (Manley	   et	   al.,	   2000).	   The	   same	   study	   also	   showed	  decreased	  hemispheric	  brain	  enlargement	  (a	  marker	  of	  cerebral	  edema)	  in	  AQP4	  KO	  mice	  24	  hours	  after	  permanent	  middle	  cerebral	  artery	  occlusion	  (MCAO).	  	  However,	  targeting	   aquaporins	   for	   the	   treatment	   of	   cerebral	   edema	   remains	   controversial	  since	   aquaporins	   are	   also	   a	   pathway	   for	   water	   efflux	   from	   the	   brain	   (Iliff	   et	   al.,	  2012).	  	  In	  fact	  deletion	  of	  AQP4	  severely	  worsens	  edema	  and	  outcome	  in	  models	  of	  vasogenic	   brain	   edema	   (Papadopoulos	   et	   al.,	   2004a;	   Papadopoulos	   et	   al.,	   2004b).	  	  	  Additionally,	   the	   initial	   analysis	   on	   hemispheric	   volume	   and	   neurologic	   scores	   by	  	   13	  Manley	  et	  al.	  was	  only	  done	  24	  hours	  following	  MCAO	  (Manley	  et	  al.,	  2004),	  whereas	  maximal	   edema	   may	   have	   yet	   to	   occur	   (Simard	   et	   al.,	   2007).	   	   In	   a	   more	   recent	  analysis	   of	   AQP4	   KO	   mice	   following	   a	   transient	   MCAO	   (40min)	   model,	   AQP4	   KO	  mice	  showed	  more	  severe	  neurological	  deficits	  and	  a	  striking	  mortality	  rate	  of	  88%	  compared	  to	  24%	  for	  WT	  mice	  (Zeng	  et	  al.,	  2012).	   	  Although,	  APQ4	  inhibition	  may	  decrease	  the	  rate	  of	  water	  entry	  across	  an	  intact	  BBB	  following	  cytotoxic	  edema,	  due	  to	   their	   dual	   role	   in	   water	   extrusion,	   it	   is	   doubtful	   that	   that	   the	   generation	   AQP	  antagonists	  will	  serve	  much	  use	  in	  the	  treatment	  of	  brain	  edema.	  This	  is	  because	  the	  primary	  goal	  of	  edema	  treatment	  is	  to	  decrease	  ICP	  which	  is	  not	  realized	  by	  AQP4	  deletion	  that	  exacerbates	  vasogenic	  edema	  that	  often	  succeeds	  the	   initial	  cytotoxic	  edema.	   Additionally,	   as	   the	   driving	   force	   for	   the	   generation	   of	   cytotoxic	   edema	   is	  imbalanced	   ionic	  concentrations,	  AQP4	  inhibition	  will	   likely	  only	  reduce	  the	  speed	  of	  edema	  generation,	  but	  not	  the	  final	  extent	  of	  the	  volume	  increase.	  Unlike	  astrocytes,	  neurons	  do	  not	  express	  aquaporin	  water	  channels	  (Amiry-­‐Moghaddam	   and	   Ottersen,	   2003).	   However,	   this	   does	   not	   mean	   that	   they	   do	   not	  transport	  water.	  Water	  transport	   is	  a	  general	   feature	  of	  many	  co-­‐transporters	  (Zeuthen	  and	  Zeuthen,	  2007;	  Zeuthen,	  2010;	  Zeuthen	  and	  Macaulay,	  2012)	  many	  of	  which	   are	   expressed	   on	   the	   neuronal	   plasma	   membrane	   such	   as	   cation-­‐chloride	  cotransporters,	   glutamate	   transporters,	   gamma-­‐aminobutyric	   acid	   (GABA)	  transporters	   and	  monocarboxylate	   transporters.	   	   There	   are	   two	  distinct	  modes	   of	  water	   transport	   that	   have	   been	   reported	   by	   transporters;	   secondary	   active	  cotransport	  and	  passive	  water	   transport	   (Figure	   1.2).	   	   	   Secondary	  co-­‐transport	   is	  water	   transport	   that	  occurs	  along	  with	   the	   transporter’s	   substrate.	  The	  amount	  of	  	   14	  transported	  water	  can	  be	  quite	  significant,	  for	  example	  NKCC1	  transports	  590	  water	  molecules	  per	  transport	  cycle	  (Hamann	  et	  al.,	  2005),	  and	  the	  lactate-­‐H+	  transporter,	  MCT1,	   transports	  500	  water	  molecules	  per	   transport	  cycle	  (Hamann	  et	  al.,	  2003)	   .	  	  An	  important	  characteristic	  of	  secondary	  co-­‐transport	  is	  that	  it	  does	  not	  require	  an	  osmotic	  driving	  force,	  and	  can	  actually	  transport	  water	  “uphill”	  against	  the	  osmotic	  gradient;	   water	   is	   always	   transported	   in	   the	   direction	   of	   substrates.	   	   	   Some	  transporters	   also	   display	   passive	   transport,	   which	   is	   defined	   by	   water	   transport	  through	   the	   pore	   of	   the	   protein	   analogous	   to	   AQP	   mediated	   transport	   (Zeuthen,	  2010).	   	   This	   mode	   of	   transport	   does	   not	   apply	   to	   all	   co-­‐transporters	   capable	   of	  water	   transport,	   and	   has	   been	   most	   thoroughly	   described	   in	   Na+	   coupled	  neurotransmitter	  transporters	  such	  as	  excitatory	  amino	  acid	  transporters	  (EAATs)	  and	  GABA	   transporters	   (GATs)	   (Loo	  et	   al.,	   1999;	  MacAulay	  et	   al.,	   2001;	  Loo	  et	   al.,	  2002;	  MacAulay	  et	  al.,	  2002).	   	  The	  osmotic	  water	  permeability	  per	   transporter	   for	  EAAT1	  and	  GAT1	  are	  0.2*10-­‐14	  cm3s-­‐1	  and	  0.7*10-­‐14	  cm3s-­‐1	  respectively	  versus	  4*10-­‐14	  cm3s-­‐1	  for	  AQP1	  (Zeuthen,	  2010).	  	  	  	  Given,	  the	  aforementioned	  studies	  on	  passive	  water	  transport,	  the	  observation	  that	  in	  contrast	  to	  astrocytes,	  neurons	  do	  not	  swell	  following	  a	  small	  (~13%)	  hyposomotic	  stimulation	  remains	  mysterious	  (Andrew	  et	  al.,	   2007).	   	   	   This	   result	   was	   attributed	   to	   the	   fact,	   that	   neurons	   do	   not	   express	  functional	  water	  channels.	  	  Another	  possibility	  in	  line	  with	  this	  hypothesis	  could	  be	  that	  as	  the	  water	  permeability	  of	  neurons	  is	  much	  lower	  than	  that	  of	  the	  aquaporin	  expressing	  astrocytes,	  volume	  regulatory	  mechanism	  are	  able	  to	  compensate	  for	  the	  water	  entry	  triggered	  by	  osmotic	  insults	  on	  a	  similar	  or	  more	  rapid	  time	  course.	  	  	  	   15	  	  	  Figure	  1.2:	  Molecular	  mechanism	  of	  water	   transport	  across	   cell	  membranes.	  Water	  crosses	  membranes	  by	  diffusion	   in	  the	   lipid	  bilayer,	  by	  osmosis	   in	  channels	  and	   by	   cotransport	   in	   cotransporters	   and	   uniporters.	   Diffusion	   and	   osmosis	   are	  driven	   by	   the	   water	   chemical	   potential	   difference.	   The	   cotransporters	   and	   the	  uniporters	  function	  as	  molecular	  water	  pumps	  in	  which	  free	  energy	  contained	  in	  the	  substrate	   gradient	   can	   be	   transferred	   to	   the	   transport	   of	   water;	   i.e.,	   a	   downhill	  transport	   of	   substrate	   can	   energize	   an	   uphill	   transport	   of	   water.	   Some	  cotransporters,	  such	  as	  the	  KCC,	  employ	  only	  cotransport;	  others,	  such	  as	  the	  SGLT1	  and	   the	   EAAT1,	   employ	   both	   cotransport	   and	   osmosis.	   From	   (Zeuthen,	   2010),	  reprinted	  with	  permission	  from	  Springer.	  	  	  	   16	  1.3.2	  Volume	  Regulation	  During	  Apoptosis	  and	  Necrosis	  	   The	   morphological	   changes	   that	   occur	   during	   apoptosis	   and	   necrosis	   are	  drastically	  different.	  	  Whereas	  necrosis	  is	  characterized	  by	  karyolysis	  (dissolution	  of	  the	  nucleus),	  cell	  swelling	  and	  membrane	  blebbing,	  apoptosis	  is	  accompanied	  by	  cell	  shrinkage,	  pyknosis	   (chromatin	   condensation)	  and	  karyorrhexis	   (fragmentation	  of	  the	  nucleus)	  (Figure	  1.3).	  	  Necrosis	  ultimately	  results	  in	  the	  loss	  of	  cell	  membrane	  integrity	   and	   the	   release	   of	   cytoplasmic	   contents	   triggering	   an	   inflammatory	  response,	   whereas	   during	   apoptosis	   cellular	   contents	   are	   contained	   in	   apoptotic	  bodies	  and	  rapidly	  phagocytosed	  (Elmore,	  2007).	  Originally,	  necrosis	  was	  thought	  to	  be	  an	  uncontrollable	  passive	  process	  that	  did	  not	  require	  energy	  and	  was	  referred	  to	   as	   “accidental	   cell	   death”	   whereas	   apoptosis,	   an	   energy	   dependent	   process	  triggered	   by	   precise	   cellular	   mechanisms	   was	   referred	   to	   as	   “programmed	   cell	  death”.	   	   	  However,	   recent	   research	  has	   suggested	   that	   some	   forms	  of	  necrotic	   cell	  death	   may	   also	   be	   governed	   by	   a	   set	   of	   distinct	   molecular	   mechanisms,	   and	  classified	  as	  necroptosis	  (described	  bellow).	  	  	   17	  	  Figure	   1.3:	   The	   morphological	   features	   of	   apoptosis,	   oncosis	   and	   necrosis.	  	  From	  (Hail	  et	  al.,	  2006),	  reprinted	  with	  permission	  from	  Springer.	  	  	  	   18	  Morphologically,	   apoptosis	   is	   characterized	   by	   cell	   rounding,	   pyknosis	  (chromatin	  condensation),	  nuclear	  fragmentation,	  the	  formation	  of	  apoptotic	  bodies	  and	   cell	   shrinkage	   and	   was	   originally	   termed	   “shrinkage	   necrosis”	   (Kerr,	   1971;	  Kerr	   et	   al.,	   1972).	   	   The	   process	   of	   isotonic	   cell	   shrinkage	   that	   occurs	   during	  apoptosis	   is	   called	  apoptotic	   volume	   decrease	   	   (AVD).	   	   AVD	   is	   triggered	   by	   the	  efflux	  of	  potassium	  and	  chloride	  along	  with	  osmotically	  obliged	  water	  (Heimlich	  et	  al.,	  2004).	  The	  role	   for	  K+	  efflux	   in	   the	  generation	  of	  apoptosis	   is	  well	  established.	  	  Accumulating	   evidence	   suggests	   rather	   than	   playing	   a	   secondary	   passive	   role	  associated	  with	   cell	   death,	   AVD	  may	   actually	   play	   a	   key	   role	   in	   driving	   apoptosis.	  	  Blocking	   neuronal	   K+	   channels	   with	   TEA	   can	   inhibit	   apoptosis	   in	   neurons,	   even	  when	  increases	  in	  [Ca2+]i	  persist	  (Yu	  et	  al.,	  1997)	  and	  the	  K+	  ionophore,	  valinomycin	  can	   trigger	   apoptosis	   (Allbritton	   et	   al.,	   1988;	   Ojcius	   et	   al.,	   1991;	   Yu	   et	   al.,	   1997).	  	  However,	  the	  effects	  of	  potassium	  on	  apoptosis	  are	  complex,	  although	  K+	  efflux	  is	  a	  requirement	   for	   AVD,	   K+	   also	   inhibits	   some	   of	   the	   apoptotic	   machinery	   (Ki	   of	  caspase-­‐3	  and	  endonucleases	  for	  K+	  are	  around	  40mM	  and	  65mM	  [K+]	  respectively)	  (Hughes	  et	  al.,	  1997;	  Yu	  and	  Choi,	  2000).	  The	  ability	  to	  trigger	  apoptosis	  by	  inducing	  sustained	  shrinkage	  with	  hyperotonic	  solutions,	  led	  to	  the	  hypothesis	  that	  AVD	  may	  actually	  be	  an	  active	  process	  necessary	  for	  apoptosis	  (Bortner	  and	  Cidlowski,	  1996,	  1998).	   	   In	   support	  of	   this	  hypothesis	  AVD	  has	  been	  reported	   to	  occur	  early	   in	   the	  apoptotic	   cascade	  before	  DNA	   fragmentation,	   cytochrome	   c	   release	   and	   caspase-­‐3	  activation	  (Maeno	  et	  al.,	  2000;	  Okada	  et	  al.,	  2006)	  and	  pharmacological	  inhibition	  of	  K+	  or	  Cl-­‐	  channels	  can	  block	  both	  the	  AVD	  and	  the	  subsequent	  cell	  death	  triggered	  by	  apoptotic	   stimuli	   (Maeno	   et	   al.,	   2000;	   Porcelli	   et	   al.,	   2004;	  Okada	   et	   al.,	   2006).	  As	  	   19	  different	  cell	   types	  are	  endowed	  with	  different	   ion	  channels,	   the	   identity	  of	   the	  K+	  and	  Cl-­‐	  channels	  responsible	   for	  AVD	  most	   likely	  differ	   in	  different	  cell	   types.	   	  The	  most	   commonly	   reported	   Cl-­‐	   channel	   underlying	   AVD	   displays	   properties	   of	   the	  volume-­‐sensitive	   outward	   rectifying	   Cl-­‐	   channel	   (VSOR)	   also	   named	   the	   volume-­‐regulated	  anion	  channel	  (VRAC),	  which	  until	  very	  recently	  had	  no	  known	  molecular	  identity	  (Okada	  et	  al.,	  2006).	  	  Proposed	  candidates	  for	  the	  molecular	  identity	  of	  the	  Cl-­‐	  channel	  were	  CLC-­‐3	  (Duan	  et	  al.,	  1997)	  or	  the	  plasmalemmal	  voltage-­‐dependent	  anion	   channel	   (pl-­‐VDAC)	   (Elinder	   et	   al.,	   2005),	   however,	   these	   results	   remain	  controversial	  as	  apoptotic	  stimuli	  still	   trigger	  VSOR-­‐like	  currents	   in	  the	  absence	  of	  either	  channel	  (Wang	  et	  al.,	  2005;	  Sabirov	  et	  al.,	  2006).	  	  Two	  recent	  and	  independent	  studies	  have	  claimed	  to	  have	  identified	  a	  key	  component	  of	  the	  classically	  described	  VRAC	  channel	  as	  LRRC8A	  (Qiu	  et	  al.,	  2014;	  Voss	  et	  al.,	  2014),	  a	  gene	  that	  shares	  a	  common	   ancestor	   with	   pannexins	   (Abascal	   and	   Zardoya,	   2012).	   	   However,	   a	  potential	  role	   for	  LRRC8	  proteins	   in	  AVD	  and	  apoptosis	  has	  yet	   to	  be	   investigated.	  	  TMEM16A	   and	   B	   were	   recently	   identified	   as	   molecular	   candidates	   for	   calcium	  activated	   chloride	   channels	   (Schroeder	   et	   al.,	   2008),	   that	   can	   be	   activated	   by	   cell	  swelling	   and	   contribute	   to	   regulatory	   volume	   decrease	   (Almaca	   et	   al.,	   2009).	  Interestingly,	   TMEM16B	   was	   recently	   shown	   to	   be	   functionally	   expressed	   in	  pyramidal	  neurons	  of	  the	  hippocampus	  as	  an	  outwardly	  rectifying	  calcium	  activated	  Cl-­‐	   channel	   (Huang	  et	  al.,	  2012).	   	  Given	  TMEM16B’s	  ability	   to	  be	  activated	  by	  Ca2+	  entry	   through	  N-­‐methyl-­‐D-­‐aspartate	   	   (NMDA)	   receptors	   it	  would	  be	   interesting	   to	  test	   whether	   TMEM16B	   is	   activated	   during	   apoptosis	   and	   if	   it	   contributes	   to	   the	  AVD	  and	  resultant	  apoptotic	  cell	  death	  in	  neurons.	  	   20	  Necrosis	   is	   characterized	  morphologically	  by	  cell	   swelling	   and	  membrane	  blebbing.	  Necrosis	   refers	   to	   the	   degradative	   process	   that	   occurs	   after	   the	   cell	   has	  died	   and	   is	   therefore	   argued	   to	   be	   inappropriate.	   The	   term	  Oncosis	   (from	  onkos,	  meaning	  swelling)	  was	  coined	  by	  von	  Recklinghausen	  in	  1910	  to	  describe	  cell	  death	  by	  swelling,	  and	  is	  a	  more	  suitable	  term	  to	  describe	  the	  cellular	  processes	  leading	  to	  necrosis	   (Majno	   and	   Joris,	   1995).	   Contrary	   to	   cell	   shrinkage	   in	   apoptosis	   the	  mechanisms	   underlying	   cell	   swelling	   are	   less	   well	   described,	   which	   may	   be	  attributable	   to	   the	   fact	   that	   oncosis/necrosis	   has	   long	   been	   thought	   of	   as	   an	  irreversible	  process.	  	  The	  increase	  in	  cell	  volume	  that	  causes	  necrosis	  is	  thought	  to	  arise	  mainly	  from	  increases	  in	  intracellular	  Na+,	  due	  to	  defective	  outward	  pumping	  caused	   by	   ATP	   depletion	   or	   by	   influx	   via	   membrane	   transporters	   and	   channels	  (Barros	   et	   al.,	   2001).	   	   In	   hepatocytes	   for	   example,	   inhibition	   of	   intracellular	   Na+	  accumulation	   has	   shown	   to	   be	   protective	   against	   cellular	   swelling	   and	   the	  development	   of	   necrosis	   induced	   by	   inhibition	   of	   mitochondrial	   oxidative	  phosphorylation	  (Carini	  et	  al.,	  1995;	  Carini	  et	  al.,	  1999).	  	  However,	  few	  reports	  have	  investigated	  the	  underlying	  mechanisms	  that	  cause	  oncosis	  leading	  to	  necrosis.	  	  	  	  	  Although	   necrosis	   was	   originally	   described	   as	   passive	   cell	   death,	   recent	  advances	  have	   in	  part	  challenged	  this	  view,	  suggesting	   that	  at	   least	  some	   forms	  of	  necrosis	  may	  be	  executed	  via	  precise	  mechanisms;	  a	  programed	  cell	  death	  process	  leading	   to	   necrosis	   termed	   necroptosis	   (Zhang	   et	   al.,	   2009;	   Vandenabeele	   et	   al.,	  2010;	   Wu	   et	   al.,	   2012;	   Re	   et	   al.,	   2014).	   	   Morphologically	   necroptosis	   exhibits	  features	  of	  necrosis	   such	  as	  cell	   swelling	  and	  disruption	  of	   the	  plasma	  membrane.	  	  Necroptosis	   involves	   a	   unique	   molecular	   signaling	   cascade	   involving	   specific	  	   21	  proteins	   RIP1	   and	   RIP3	   and	   can	   be	   inhibited	   by	   necrostatins	   (Zhang	   et	   al.,	   2009;	  Vandenabeele	   et	   al.,	   2010).	   	   	   These	  new	   findings	   suggest	   that	   the	   cellular	   process	  leading	  to	  necrosis	  may	  involve	  distinct	  steps	  that	  could	  be	  targeted	  therapeutically.	  	  Compared	  to	  apoptosis,	  however,	  the	  mechanisms	  underlying	  necroptosis	  are	  only	  starting	   to	   be	   uncovered	   and	   the	   role	   of	   cell	   swelling	   in	   this	   process	   remains	  unknown.	   	   It	   would	   be	   interesting	   to	   test	   the	   following	   questions:	   1)	   Can	   cell	  swelling	   trigger	   necroptosis?	   and	   2)	   Can	   necroptosis	   be	   blocked	   by	   blocking	   cell	  swelling?	  	  1.3.3	  A	  Brief	  History	  of	  Excitotoxic	  Neuronal	  Swelling	  	  	   Following	   the	   observation	   by	   Lucas	   and	  Newhouse	   in	   1957	   that	   glutamate	  caused	  damage	  to	  the	  inner	  retina	  (Lucas	  and	  Newhouse,	  1957),	  John	  Olney	  showed	  that	  the	  food	  additive	  monosodium	  glutamate	  killed	  brain	  neurons	  in	  newborn	  mice	  and	   	   coined	   the	   term	   excitotoxicity	   (Olney,	   1969).	   	   He	   noted	   that	   the	   excessive	  activation	  of	  neurons	  by	  glutamate	  led	  to	  acute	  swelling	  of	  both	  the	  neuronal	  soma	  and	   their	   dendrites	   that	   preceded	   necrosis	   (Olney,	   1971).	   	   	   In	   1985,	   Rothman	  discovered	   that	   by	   removing	   chloride	   from	   the	   extracellular	   solution	   of	   neurons	  exposed	   to	   glutamate	  he	  was	   able	   to	   inhibit	   glutamate	   toxicity	   and	   the	   associated	  swelling	   in	   cultured	   hippocampal	   neurons	   (Rothman,	   1985).	   	   Rothman	   concluded	  that	  the	  pathophysiology	  of	  amino	  acid	  neurotoxicity	  was	  rather	  straightforward,	  in	  that	   substances	   that	   caused	   a	   steady	   depolarization	   led	   to	   passive	   chloride	   influx	  which	  along	  with	  cations	  resulted	  in	  water	  entry	  and	  cell	   lysis.	   	  However	  it	   is	  now	  	   22	  known	   that	  Cl-­‐	   is	  not	  passively	  distributed	  across	   the	  neuronal	  plasma	  membrane	  (Alvarez-­‐Leefmans	   and	   Delpire,	   2009),	   and	   therefore	   the	   Cl-­‐	   entry	   required	   for	  swelling	   must	   either	   involve	   active	   transport	   or	   activation	   of	   a	   Cl-­‐	   conductance	  pathway	   that	   is	   not	   open	   at	   rest.	   	  Within	  months	   of	   Rothman’s	   discovery,	   Dennis	  Choi	  discovered	  another	  pathway	  by	  which	  neurons	  could	  die	  following	  excitotoxic	  insults	   that	  depended	  on	   the	   influx	  of	  calcium	  and	  not	  sodium	  (Choi,	  1985).	   	   	   In	  a	  subsequent	  paper	  Choi	  elegantly	  demonstrated	  that	  there	  were	  in	  fact	  two	  separate	  and	  distinguishable	   cell	  death	  pathways	   following	  excitotoxic	   insults	   (Choi,	  1987).	  	  The	   first	   component	   was	   marked	   by	   neuronal	   swelling,	   occurred	   early,	   and	   was	  dependent	  on	  extracellular	  Na+	  and	  Cl-­‐,	   but	  was	   independent	  of	   extracellular	  Ca2+.	  	  The	   second	   component	  was	  marked	   by	   gradual	   neuronal	   disintegration,	   occurred	  late,	   was	   dependent	   on	   extracellular	   Ca2+,	   and	   could	   be	   mimicked	   by	   a	   Ca2+	  ionophore.	   	   	   The	   mechanisms	   governing	   the	   entry	   of	   chloride	   remain	   largely	  unknown,	   although	   a	   partial	   reduction	   in	   neuronal	   swelling	   was	   observed	   by	  blocking	   GABAA	   receptors	   in	   cell	   culture	   (Hasbani	   et	   al.,	   1998).	   	   	   The	   role	   of	  GABAARs	   in	   excitotoxic	   neuronal	   swelling	   are	  discussed	   in	   section	  1.5.2,	   however,	  reports	   in	   situ	   have	   shown	   that	   the	   majority	   of	   Cl-­‐	   influx	   occurs	   via	   other	  unidentified	  mechanisms	  (Allen	  et	  al.,	  2004;	  Pond	  et	  al.,	  2006)	  	  1.2.4	  Cortical	  Spreading	  Depression	  and	  the	  Anoxic	  Depolarization	  	   Discovered	   by	   Leao	   in	   1944,	   spreading	   depression	   (SD)	   is	   a	   wave	   of	  depolarization	  that	  propagates	  in	  all	  directions	  across	  the	  brain	  at	  a	  velocity	  of	  2-­‐5	  mm/min	  	  (Leao,	  1944).	   	  Spreading	  depression	  also	  called	  spreading	  depolarization	  	   23	  and	  the	  closely	  related	  anoxic	  depolarization	  are	  characterized	  by	  a	  rapid	  and	  nearly	  complete	   depolarization	   of	   a	   population	   of	   neurons	   that	   occurs	   with	   a	   massive	  redistribution	   of	   intracellular	   and	   extracellular	   ions	   coincident	   with	   neuronal	  volume	  changes	  and	  cytotoxic	  edema	  (Somjen,	  2001;	  Dreier,	  2011;	  Lauritzen	  et	  al.,	  2011).	   	   	  Simultaneous	  with	  the	  near	  complete	  breakdown	  of	  ionic	  gradients	  occurs	  the	  swelling	  of	  neurons	  and	  distortion	  of	  dendritic	  spines	  (Takano	  et	  al.,	  2007;	  Zhou	  et	  al.,	  2010;	  Zhou	  et	  al.,	  2013).	  	  	  The	  neuronal	  swelling	  that	  occurs	  is	  coincident	  with	  a	   sharp	   peak	   in	   the	   intrinsic	   optical	   signal	   (Zhou	   et	   al.,	   2010),	   that	   occurs	   at	   the	  onset	   of	   SD.	   Electrophysiological	   recording	   in	   humans	   have	   shown	   unequivocally	  that	  SD	  occurs	  during	  a	  variety	  of	  brain	  injuries	  such	  as,	  subarachnoid	  hemorrhage,	  ischemic	   stroke	   and	   traumatic	   brain	   injury	   and	   that	   the	  occurrence	  of	   SD	   is	   often	  associated	  with	  worsened	  outcomes	  (Dreier,	  2011;	  Lauritzen	  et	  al.,	  2011).	  	  Although	  the	   cellular	   mechanisms	   triggering	   SD	   and	   the	   SD	   like	   wave	   during	   the	   anoxic	  depolarization	   may	   be	   caused	   by	   different	   mechanisms,	   the	   ionic	   changes	   and	  neuronal	   swelling	   that	   occur	   are	   similar	   (Somjen,	   2001;	   Dreier,	   2011).	   	   	   	   A	   key	  difference	  between	  the	  transient	  spreading	  depolarization	  and	  the	  prolonged	  anoxic	  depolarization	   is	   the	  availability	  of	  ATP	   that	   is	   required	   to	   restore	   ionic	   gradients	  and	  repolarize	  the	  neuron.	   	   	  During	  ischemic	  insults,	  ATP	  levels	  drop	  drastically	  in	  the	   ischemic	   core,	  which	   leads	   to	   failure	   of	   the	  Na+/K+	  ATPase.	   	   Extracellular	  Na+	  decreases	   and	   extracellular	   K+	   increases,	   thereby	   decreasing	   and	   possibly	   even	  reversing	  the	  function	  of	  Na+	  and	  K+	  driven	  transporters	  such	  as	  those	  required	  for	  the	  uptake	  of	  glutamate	  (Rossi	  et	  al.,	  2000).	  	  The	  increases	  in	  extracellular	  glutamate	  activates	   ionotropic	   glutamate	   receptors,	   thereby	   further	   increasing	   intracellular	  	   24	  Na+	   and	   causing	   a	   viscous	   feed-­‐forward	   cycle	   resulting	   in	   the	   resultant	   anoxic	  depolarization	   (Somjen,	   2001,	   2002),	   which	   also	   involves	   the	   activation	   of	   non-­‐selective	  cation	  channels	  (Aarts	  et	  al.,	  2003;	  Thompson	  et	  al.,	  2006).	  	  The	  situation	  in	  the	   penumbra	   is	   different	   due	   to	   relatively	   preserved	   ATP	   compared	   to	   the	  penumbra	  region	   (Welsh	  et	  al.,	  1991;	  Hossmann,	  1994).	  Transient	  depolarizations	  are	  still	  observed	  in	  the	  penumbra	  resembling	  more	  similarity	  to	  the	  SD	  than	  to	  the	  anoxic	   depolarization	   (Ginsberg	   and	   Pulsinelli,	   1994;	   Dreier,	   2011),	   possibly	  triggered	  by	   increases	   in	   extracellular	  K+	   or	   glutamate	   released	   closer	   to	   the	   core	  (Nedergaard	   and	   Hansen,	   1993;	   Hossmann,	   1996).	   	   Very	   short	   and	   transient	   SDs	  such	  as	  those	  correlated	  with	  migraine	  aura	  do	  not	  cause	  cell	  death,	  however	  if	  the	  duration	  of	  SD	  is	  prolonged,	  brain	  damage	  is	  often	  observed	  (Dreier,	  2011).	  	  During	  ischemic	   events,	   the	   number	   and	   duration	   of	   SDs	   observed	   in	   the	   penumbra	   is	  positively	  correlated	  with	  cell	  death	  and	  infract	  volume	  (Mies	  G.,	  1993;	  Dijkhuizen	  et	  al.,	   1999;	   Hartings	   et	   al.,	   2003).	   	   Additionally,	   artificially	   triggered	   spreading	  depolarizations	   outside	   the	   penumbra	   but	   that	   propagated	   into	   the	   penumbra	  caused	   an	   enlargement	   of	   the	   necrotic	   core	   (Back	   et	   al.,	   1996;	   Busch	   et	   al.,	   1996;	  Takano	  et	  al.,	  1996).	  	  	  	  	  	  	   25	  1.4	  Chloride	  homeostasis	  	  1.4.1	  Introduction	  to	  Chloride	  Homeostasis	  and	  Chloride	  Equilibrium	  	  	   The	   initial	   concept	   that	   Cl-­‐	   was	   passively	   distributed	   across	   the	   plasma	  membrane	  came	  about	  because	  it	  was	  dominated	  by	  work	  done	  in	  skeletal	  muscle	  where	  Cl-­‐	  permeability	  is	  extremely	  high,	  in	  fact	  Cl-­‐	  permeability	  is	  more	  than	  twice	  the	  permeability	  of	  K+	  (Hodgkin	  and	  Horowicz,	  1959;	  Alvarez-­‐Leefmans	  and	  Delpire,	  2009),	  therefore	  setting	  resting	  membrane	  potential	  (Em)	  very	  close	  to	  the	  chloride	  equilibrium	  potential	  (ECl-­‐).	  	  The	  discovery	  that	  activation	  of	  GABAAR	  ligand	  gated	  Cl-­‐	  channels	   hyperpolarized	   neurons	   (Boistel	   and	   Fatt,	   1958;	   Kuffler	   and	   Edwards,	  1958)	  while	  depolarizing	  other	  types	  of	  neurons	  such	  as	  sensory	  neurons	  (De	  Groat	  et	  al.,	  1972)	  made	  it	  clear	  that	  ECl-­‐	  was	  not	  set	  at	  Em	  and	  therefore	  that	  chloride	  must	  be	  actively	  transported	  across	  the	  plasma	  membrane.	  	  Perhaps	  the	  best	  example	  of	  this	   is	   demonstrated	   by	   the	   fact	   that	   immature	   pyramidal	   neurons	   show	  depolarizing	   GABAergic	   responses	   that	   then	   switch	   to	   hyperpolarizing	   responses	  later	  in	  development	  (Ben-­‐Ari	  et	  al.,	  2007),	  although	  it	  should	  be	  pointed	  out	  that	  in	  addition	  to	  Cl-­‐,	  GABAARs	  are	  also	  permeable	  to	  HCO3-­‐,	  and	  therefore	  [HCO3-­‐]	  can	  also	  affect	  EGABA	  (Rivera	  et	  al.,	  2005).	  	  The	  predominant	  transporters	  known	  to	  govern	  Cl-­‐	  transport	   in	   neurons	   and	   set	   up	   ECl-­‐	   are	   the	   cation-­‐chloride	   cotransporters,	   KCC2	  and	  NKCC1	  (Blaesse	  et	  al.,	  2009)(see	  section	  1.4.5).	  There	  also	  may	  be	  a	  role	  for	  Na+	  dependent	   and	   Na+	   independent	   anion	   exchangers	   that	   exchange	   Cl-­‐	   for	   HCO3-­‐	  (Grichtchenko	   et	   al.,	   2001;	   Payne	   et	   al.,	   2003;	   Gonzalez-­‐Islas	   et	   al.,	   2009).	  	  	   26	  Additionally,	   a	   recently	   discovered	   Cl-­‐,	   HCO3-­‐,	   SO42-­‐	   exchanger	   was	   shown	   to	   be	  highly	  expressed	  in	  neurons	  throughout	  the	  brain,	  although	  a	  functional	  role	  for	  this	  transporter	  in	  neurons	  has	  not	  yet	  been	  tested	  (Rahmati	  et	  al.,	  2013).	  	  	  	  	  In	  addition	  to	   [Cl-­‐]i	   regulation	   by	   transporters,	   	   a	   recent	   study	   has	   suggested	   that	   the	   large	  impermeable	  intracellular	  anions	  that	  make	  up	  the	  bulk	  of	  the	  anionic	  intracellular	  milieu	  play	   a	   significant	   role	   in	   setting	  ECl-­‐	  (Glykys	   et	   al.,	   2014).	   	   This	  may	   in	  part	  explain	   how	   neurons	   display	   subcellular	   differences	   in	   EGABA	   despite	   a	   relatively	  uniform	   distribution	   of	   cation-­‐chloride	   cotransporters	   (Doyon	   et	   al.,	   2011).	  	  Importantly,	   it	   is	   these	   same	   large	   anions	   that	   also	   explain	  why	   small	   changes	   in	  intracellular	   Cl-­‐	  are	   tightly	   linked	   to	   cell	   volume	   changes;	   unlike	   the	   entry	   of	   Na+	  cations	   that	   can	   be	   balanced	   by	   efflux	   of	   K+,	   sustained	   increases	   in	   [Cl-­‐]i	   create	   a	  Gibbs-­‐Donnan	  imbalance,	  and	  therefore	  either	  combined	  cation	  and	  water	  influx	  or	  anion	  efflux	  must	  occur.	  	  	  1.4.2	  Aberrant	  Chloride	  Regulation	  in	  CNS	  Disorders	  	  	   In	   the	   adult	   brain,	   equilibrium	   between	   excitation	   and	   inhibition	   is	   an	  essential	   feature	   for	   proper	   function,	   and	   disequilibrium	   is	   often	   a	   central	   theme	  underlying	  brain	  disorders	  such	  as	  epilepsy,	  neuropathic	  pain,	  addiction	  as	  well	  as	  psychiatric	  illnesses.	  	  	  	  GABAA	  mediated	  inhibition	  is	  of	  utmost	  importance	  in	  dampening	  excitation,	  and	  agents	   that	  block	  GABAA	  receptors	  generate	  seizures.	   	   In	  adult	  neurons	  of	   the	  cortex	  and	  hippocampus,	  if	  [Cl-­‐]i	   increases	  and	  causes	  the	  equilibrium	  potential	  for	  	   27	  chloride	   (ECl-­‐)	   to	   become	   more	   depolarized	   relative	   to	   Em,	   activation	   of	   GABAA	  receptors	   will	   result	   in	   Cl-­‐	   efflux	   and	   depolarizing	   responses.	   	   Temporal	   lobe	  epilepsy	   is	   associated	   with	   perturbed	   chloride	   homeostasis	   (Cohen	   et	   al.,	   2002),	  possibly	  due	  to	  decreased	  KCC2	  expression	  as	  observed	  in	  a	  minority	  of	  pyramidal	  neurons	   from	   human	   epileptic	   tissue	   showing	   depolarized	   GABAA	   receptor	  mediated	  post-­‐synaptic	  events	  (Huberfeld	  et	  al.,	  2007).	  	  Insults	  such	  as	  ischemia	  and	  TBI	   that	   cause	   increases	   in	   Cl-­‐	   concentration	   and	   cytotoxic	   edema,	   often	   result	   in	  seizures	   with	   patients	   responding	   poorly	   to	   anticonvulsants	   that	   act	   on	   GABAA	  receptors	  (Young	  et	  al.,	  1990).	   	  In	  a	  neonate	  (p5-­‐7)	  brain	  slice	  model	  of	  TBI,	  it	  was	  noted	   that	   the	   NKCC1	   inhibitor	   bumetanide	   was	   capable	   of	   decreasing	   the	   post-­‐traumatic	  levels	  of	  Cl-­‐	  (Dzhala	  et	  al.,	  2012),	  consistent	  with	  other	  studies	  implicating	  a	  role	  for	  NKCC1	  Cl-­‐	  transport	  in	  neonatal	  seizures	  (Dzhala	  et	  al.,	  2005;	  Dzhala	  et	  al.,	  2010).	  	  Identifying	  other	  mechanisms	  that	  cause	  neuronal	  Cl-­‐	  increases	  in	  the	  adult	  brain	   with	   low	   levels	   of	   NKCC1	   expression	   (Wang	   et	   al.,	   2002)	   may	   help	   in	  developing	   therapeutics	   to	   reduce	   the	   excitation/inhibition	   imbalance	   following	  trauma	  and	  ischemia.	  	  Aberrant	  Cl-­‐	  homeostasis	  within	  the	  spinal	  dorsal	  horn	  results	  in	  neuropathic	  pain,	  underlying	  tactile	  allodynia	  (Coull	  et	  al.,	  2003).	  	  Excitatory	  output	  of	  the	  dorsal	  horn	  is	  normally	  suppressed	  by	  a	  large	  network	  of	  inhibitory	  connections,	  however,	  following	   peripheral	   nerve	   injury	   a	   depolarizing	   shift	   in	   ECl-­‐	   occurs	   leading	   to	  decreased	  and	  often	  reversal	  of	  GABAA	  and	  glycine	  receptor	  responses.	  	  Surprisingly	  it	  was	   found	  that	  the	  mechanism	  underlying	  this	  shift	   in	  ECl-­‐	  was	   in	   fact	  caused	  by	  release	  of	  	  brain-­‐derived	  neurotrophic	  factor	  (BDNF)	  from	  ATP	  stimulated	  microglia	  	   28	  acting	  on	  TrkB	  receptors	   in	  spinal	   lamina	  I	  neurons	  (Coull	  et	  al.,	  2005).	   	  The	  ATP-­‐microglia-­‐BDNF-­‐TrkB	   signal	   causes	   down-­‐regulation	   of	   KCC2,	   the	   major	   Cl-­‐	  extrusion	   mechanism	   in	   adult	   neurons,	   thereby	   reducing	   inhibition	   via	   Cl-­‐	  accumulation	   and	   a	   depolarizing	   shift	   in	   ECl-­‐.	   	   It	   was	   also	   found	   that	   this	   same	  mechanism	  was	   implicated	   in	  morphine	  hyperalgesia,	  caused	  by	  μ	  opioid	  receptor	  driven	   expression	   of	   P2X4Rs	   in	   microglia	   (Ferrini	   et	   al.,	   2013).	   	   	   Whether	   this	  microglia-­‐BDNF-­‐TrkB-­‐KCC2	   signaling	   pathway	   contributes	   to	   Cl-­‐	   homeostasis	   in	  other	  brain	  regions	  and	  other	  neurological	  disorders	  remains	  to	  be	  tested.	  	  	  	  Similarly,	   opiate	   activation	   of	   the	  mesolimbic	   dopamine	   system	   causes	   the	  switching	  of	  GABAAR	  responses	  on	  ventral	  tegmental	  area	  GABAergic	  neurons	  from	  hyperpolarizing	  to	  depolarizing	  (Laviolette	  et	  al.,	  2004),	  underling	  the	  switch	  from	  an	  opiate-­‐nondependent	  to	  an	  opiate-­‐dependent	  state,	  likely	  via	  increases	  in	  BDNF	  (Vargas-­‐Perez	  et	  al.,	  2009).	  However,	  the	  source	  of	  BDNF	  in	  this	  circuit	  has	  not	  yet	  been	  described.	  	  1.5	  Chloride	  Channels	  and	  Transporters	  in	  the	  CNS	  	  1.5.1	  CLC	  family	  of	  Chloride	  Channels	  and	  Transporters	  	   	  	   The	  CLC	  family	  represents	  a	  highly	  conserved	  family	  of	  chloride	  transporters	  with	   9	   members	   found	   in	   the	   mammalian	   genome	   (Stauber	   et	   al.,	   2012).	   	   CLC	  proteins	  most	  commonly	  form	  homodiemers,	  however	  heterodimerization	  has	  been	  observed	  (Lorenz	  et	  al.,	  1996;	  Weinreich	  and	  Jentsch,	  2001).	  	  CLCs	  are	  classified	  as	  	   29	  double	   barrel	   channels	   and	   transporters	   in	   that	   each	   subunit	   encloses	   it’s	   own	  permeation	   pathway,	   with	   several	   members	   exhibiting	   electrogenic	   (1Cl-­‐/2H+)	  exchange	  rather	  than	  being	  Cl-­‐	  selective	  channels	  (Accardi	  and	  Miller,	  2004).	  	  CLC-­‐1,	  -­‐2,	   -­‐Ka,	   and	   -­‐Kb	   reside	   in	   the	   plasma	   membrane,	   whereas	   all	   other	   CLCs	   are	  predominantly	  found	  in	  intracellular	  organelles	  and	  vesicles.	  	  Of	  the	  CLCs	  shown	  to	  regulate	  Cl-­‐	  across	  the	  plasma	  membrane,	  only	  CLC-­‐2	  shows	  expression	  in	  the	  brain	  and	  has	  been	  detected	  in	  both	  neurons	  and	  glia	  (Sik	  et	  al.,	  2000).	  	  	  Interestingly,	  CLC-­‐2	  is	  activated	  by	  osmotically	  induced	  cell	  swelling	  (Grunder	  et	  al.,	  1992).	  	  However,	  CLC-­‐2	   is	   activated	  by	  hyperpolarization	  and	  CLC-­‐2	   currents	  display	   strong	   inward	  rectification,	  making	  it	  an	  unlikely	  Cl-­‐	  loader	  in	  neurons	  as	  	  hyperpolarization	  would	  lead	  to	  Cl-­‐	  efflux.	  	  In	  fact,	  CLC-­‐2	  has	  been	  shown	  to	  mediate	  exclusively	  Cl-­‐	  efflux	  from	  neurons	  (Staley	  et	  al.,	  1996;	  Rinke	  et	  al.,	  2010),	  and	  may	  aid	  in	  rapidly	  re-­‐setting	  ECl-­‐	  after	   Cl-­‐	   influx.	   	   Surprisingly	   input	   resistance	  was	   increased	   by	   over	   75%	   in	   CA1	  pyramidal	   neurons	   of	   CLC-­‐2	   KO	   mice,	   suggesting	   a	   large	   role	   for	   CLC-­‐2	   in	  modulating	  neuronal	  excitability	  (Rinke	  et	  al.,	  2010).	  	  CLC-­‐2	  is	  strongly	  inhibited	  by	  Zn2+,	   and	   poorly	   inhibited	   by	   nonspecific	   Cl-­‐	   channel	   inhibitors	   such	   as	   4,4'-­‐diisothiocyano-­‐2,2'-­‐stilbenedisulfonic	   acid	   (DIDS)	   and	   5-­‐Nitro-­‐2-­‐(3-­‐phenylpropylamino)benzoic	  acid	  (NPPB)	  (Clark	  et	  al.,	  1998;	  Stauber	  et	  al.,	  2012).	  	  1.5.2	  GABAA	  Receptor	  Channels	  	  	   Gamma-­‐aminobutyric	   acid	   (GABA)	   and	   glycine	   are	   the	   major	   inhibitory	  neurotransmitters	  of	  the	  CNS.	  	  In	  pyramidal	  neurons	  of	  the	  hippocampus	  and	  cortex,	  	   30	  however,	  fast	  inhibitory	  transmission	  is	  mediated	  by	  GABAARs,	  in	  contrast	  to	  the	  co-­‐transmission	  by	  GABA	  and	  glycine	   that	  occurs	   in	  other	  systems	  such	  as	   the	  spinal	  cord	   (Jonas	   et	   al.,	   1998).	   	   GABAARs	   are	   heteropentameric	   ion	   channels	   assembled	  from	  a	   large	   family	  of	   subunits,	   there	  exist	  8	   identified	   subunit	   families	  each	  with	  multiple	   subtypes	   and	   splice	   variants	   resulting	   in	   multiple	   different	   possible	  channel	   configurations	  with	  different	  properties.	   	  GABAARs	  are	  permeable	   to	  both	  Cl-­‐	  and	  HCO3-­‐	  with	  a	  ratio	  of	  around	  4:1	  (Kaila	  and	  Voipio,	  1987;	  Staley	  et	  al.,	  1995).	  	  Therefore,	  the	  direction	  of	  net	  current	  depends	  on	  the	  intracellular	  and	  extracellular	  concentrations	  of	  Cl-­‐	  and	  HCO3-­‐	  as	  well	  as	  the	  membrane	  potential	  of	  the	  cell.	  In	  most	  mature	   CNS	   neurons	   GABA	   receptor	   activation	   leads	   to	   Cl-­‐	   influx	   and	   therefore	  hyperpolarization,	  although	   in	   immature	  neurons	  and	  peripheral	  neurons	  GABAAR	  activation	  can	  be	  depolarizing	  as	  intracellular	  Cl-­‐	  concentration	  is	  maintained	  above	  electrochemical	   equilibrium	   (Ben-­‐Ari	   et	   al.,	   2007).	   	   For	   this	   reason	   changes	   in	  intracellular	   Cl-­‐	   can	  profoundly	   affect	   inhibitory	   transmission	   and	   is	   implicated	   in	  many	   CNS	   disorders	   (see	   section	   1.4.2).	   	   GABAA	   receptors	   are	   the	   target	   of	  many	  drugs	   with	   clinical	   importance,	   including	   barbituates,	   benzodiazepines,	   steroids,	  anesthetics	   and	   alcohol.	   	   Following	   excitotoxic	   insults	   such	   as	   ischemia,	   the	  extracellular	   GABA	   concentration	   increases	   due	   to	   a	   combination	   of	   vesicular	  release	   and	   reversal	   of	   GABA	   transporters	   (GATs)	   (Allen	   et	   al.,	   2004),	   possibly	  causing	   Cl-­‐	   influx.	   	   Indeed,	   GABAARs	   have	   been	   shown	   to	   mediate	   a	   portion	   of	  excitotoxic	  neuronal	  swelling	  in	  cell	  culture	  (Hasbani	  et	  al.,	  1998).	  	  In	  brain	  slices	  it	  was	  also	  shown	  that	  blocking	  GABAARs	  led	  to	  a	  small	  reduction	  in	  the	  initial	  rapid	  increase	  (<1	  min)	  of	  the	  intrinsic	  optical	  signal	  (a	  marker	  of	  cell	  swelling)	  following	  	   31	  oxygen	   glucose	   deprivation	   (OGD),	   however,	   blocking	   GABAARs	   had	   no	   effect	   on	  final	  intrinsic	  optical	  signal	  increase	  (>2	  min)(Allen	  et	  al.,	  2004).	  	  Additionally	  it	  was	  observed	   that	   calcium	   influx	   caused	   GABAARs	   to	   become	   completely	   inactivated	  within	  2	  minutes	  of	  the	  anoxic	  depolarization.	   	  These	  results	  suggest	  that	  although	  GABAARs	   can	   contribute	   to	   neuronal	   swelling	   via	   a	   transient	   increase	   in	   Cl-­‐,	   the	  majority	  of	   the	  Cl-­‐	   influx	   that	  occurs	  during	   ischemia	  and	  other	  excitotoxic	   insults	  occurs	   via	   another	   mechanism.	   	   Consistent	   with	   these	   results,	   Cl-­‐	   imaging	   in	  pyramidal	  neurons	  with	  the	  genetically	  encoded	  Cl-­‐	  indicator	  Clomeleon	  showed	  no	  effect	  of	  blocking	  GABAARs	  on	  the	  magnitude	  of	  the	  chloride	  increase	  during	  in	  vitro	  ischemia	  (Pond	  et	  al.,	  2006).	  	  1.5.3	  Volume-­‐Activated	  Anion	  Channels	  	  	  	  	   Volume-­‐activated	   anion	   channels	   (VAACs)	   represent	   the	   group	   of	   anion	  channels	  that	  are	  activated	  by	  increases	  in	  cell	  volume,	  originally	  identified	  based	  on	  patch	   clamp	   recordings	  of	   currents	   activated	   following	  hyposmotic	   stress.	   	   	   These	  can	  be	  separated	  into	  2	  main	  groups	  based	  on	  their	  electrophysiological	  properties;	  the	  maxi-­‐anion	   channel	   and	   the	  volume-­‐sensitive	   outwardly	   rectifying	   anion	  channel	  (VSOR).	  	  	  It	  is	  also	  plausible	  that	  various	  calcium-­‐activated	  anion	  channels	  may	  function	  as	  VAACs.	  The	  molecular	  identities	  of	  these	  channels	  are	  just	  starting	  to	   be	   identified,	   and	   are	   still	   a	   large	   topic	   of	   controversy	   (Alvarez-­‐Leefmans	   and	  Delpire,	  2009).	  	  	  	  The	  maxi-­‐anion	  channel	  is	  present	  in	  a	  variety	  of	  cell	  types	  and	  can	  be	   activated	  by	   swelling,	   or	  ATP	  depletion	   (Strange	   et	   al.,	   1996).	   	   The	  maxi	   anion	  	   32	  channel	   is	   characterized	   by	   a	   large	   (300-­‐400pS)	   single-­‐channel	   conductance	  (Sabirov	  and	  Okada,	  2009),	  however,	  it	  has	  been	  observed	  that	  the	  channel	  may	  also	  function	   in	  many	  sub-­‐conductance	  states.	   	  The	   fully	  open	  channel	  exhibits	  a	   linear	  	  	  	  	  I-­‐V	   curve	  with	  no	   rectification.	   	   	   	   The	  maxi-­‐anion	   channel	   is	   potently	   inhibited	  by	  Gd3+,	  but	  only	  partially	  suppressed	  by	  non-­‐selective	  anion	  channel	  blockers	  such	  as	  DIDS	   and	   NPPB	   (Sabirov	   et	   al.,	   2001;	   Alvarez-­‐Leefmans	   and	   Delpire,	   2009).	   	   The	  channel	   is	   also	   inhibited	   by	   arachidonic	   acid	   at	   the	   micromolar	   level	   (Liu	   et	   al.,	  2008a).	  	  	  	  Due	  to	  the	  large	  pore	  of	  the	  maxi-­‐anion	  channel,	  it	  is	  a	  candidate	  protein	  for	   regulating	   ATP	   release,	   as	   has	   been	   shown	   in	   cultured	   astrocytes	   (Liu	   et	   al.,	  2008b).	   	   In	   dorsal	   root	   ganglion	   neurons	   a	   channel	   with	   maxi-­‐anion	   like	  pharmacological	   properties	   was	   shown	   to	   mediate	   non-­‐vesicular	   ATP	   release	  following	  action	  potential	  mediated	  minute	  swelling	  of	  axons	  (Fields	  and	  Ni,	  2010),	  possibly	   involved	  in	  neuron-­‐glia	  signaling.	   	   	  The	  most	  commonly	  reported	  VAAC	  is	  the	  volume-­‐sensitive	  outwardly	  rectifying	  anion	  channel	  (VSOR),	  also	  referred	  to	  as	  the	   volume-­‐regulated	   anion	   channel	   (VRAC)	   and	   ICl,swell	   (Alvarez-­‐Leefmans	   and	  Delpire,	   2009).	   	  VSOR	  displays	  moderate	  outward	   rectification.	   	   In	   contrast	   to	   the	  maxi-­‐anion	  channel,	  VSOR	  has	  a	  strict	  requirement	  for	  cytosolic	  ATP	  and	  is	  inhibited	  by	  intracellular	  Mg2+	  (Oiki	  et	  al.,	  1994),	  possibly	  due	  to	  it’s	  ATP	  binding	  properties.	  	  VSOR	  is	  also	  inhibited	  by	  intracellular	  acidification	  (Sabirov	  et	  al.,	  2000).	  	  VSOR	  can	  be	   inhibited	   by	   a	   wide	   spectrum	   of	   non-­‐specific	   anion	   channel	   blockers,	   such	   as	  DIDS,	   NPPB	   and	   niflumic	   acid	   (Strange	   et	   al.,	   1996;	   Jentsch	   et	   al.,	   2002;	   Alvarez-­‐Leefmans	   and	   Delpire,	   2009).	   	   VSOR	   currents	   were	   reported	   to	   underlie	   the	   Cl-­‐	  influx	   required	   for	   excitotoxic	  neuronal	  necrosis	   in	   cell	   culture	   (Inoue	   and	  Okada,	  	   33	  2007),	  however	  these	  results	  contrast	  with	  a	  report	  from	  the	  same	  group	  suggesting	  that	  VSOR	  is	  required	  for	  neuronal	  volume	  recovery	  following	  swelling	  (Inoue	  et	  al.,	  2005).	   	   Additionally,	   it	   was	   observed	   by	   others	   that	   blocking	   VRAC	   with	   NPPB,	  inhibited	  neuronal	  volume	  recovery	  following	  excitotoxic	  insults	  such	  as	  veratridine	  exposure	  (Churchwell	  et	  al.,	  1996).	   	  These	  results,	  suggest	   that	  pyramidal	  neurons	  are	   endowed	   with	   volume	   regulatory	   machinery	   displaying	   pharmacological	  properties	   of	   VRAC.	   	   VRAC	   like	   currents	   have	   also	   been	   recorded	   from	   cultured	  astrocytes	   (Crepel	   et	   al.,	   1998)	   and	   may	   release	   excitatory	   amino	   acids	   when	  activated	  (Kimelberg	  et	  al.,	  2006).	  	  In	  vivo	  block	  of	  astrocyte	  but	  not	  neuronal	  VRACs	  with	   DCPIB	   (Zhang	   et	   al.,	   2011),	   resulted	   in	   decreased	   glutamate	   release	   and	  neuroprotection	  in	  a	  rat	  model	  of	  MCAO	  (Zhang	  et	  al.,	  2008).	  	  It	  is	  also	  possible	  that	  blocking	   VRACs	   could	   conceivably	   reduce	   cerebral	   edema,	   not	   solely	   by	   reducing	  neurotransmitter	  release	  or	  by	  blocking	  cell	  swelling	  (VRAC	  activation	  reduces	  cell	  volume),	  but	  by	  blocking	  the	  release	  of	  large	  osmolytes	  such	  as	  taurine,	  which	  may	  draw	   water	   into	   the	   brain	   across	   an	   intact	   BBB.	   	   The	   molecular	   identity	   of	   the	  classically	   described	   VRAC	   or	   VSOR	   channel,	   was	   recently	   proposed	   by	   two	  independent	  groups	  to	  be	  encoded	  by	  heteromers	  of	  the	  LRRC8	  gene	  family,	  and	  to	  be	   comprised	   of	   LRRC8A	   (Qiu	   et	   al.,	   2014;	   Voss	   et	   al.,	   2014)	   and	   at	   least	   one	   of	  several	   other	   LRRC8	   members	   (B-­‐D)	   (Voss	   et	   al.,	   2014).	   The	   identification	   of	  LRRC8A	   as	   a	   key	   component	   of	   the	   VRAC	   was	   based	   on	   pharmacological	   and	  electrophysiological	   properties	   and	   that	   the	   hypoosmotic	   activated	   current	   in	  human	  embryonic	  kidney	   (HEK)-­‐293	   cells	  was	  blocked	  by	   siRNA	  against	  LRRC8A,	  however,	  it	  is	  still	  possible	  that	  VRAC	  may	  be	  encoded	  by	  different	  genes	  in	  different	  	   34	  cell	   types.	   	   CLC	   channels	   (see	   section	   1.4.1)	   represent	   another	   possible	   VAAC,	  however,	  it	  is	  clear	  that	  these	  are	  distinct	  from	  VRAC	  due	  to	  smaller	  single	  channel	  conductance	   and	   strong	   inward	   rectification.	   CLC-­‐2	   is	   found	   on	   the	   plasma	  membrane,	  is	  osmosensitive	  (Grunder	  et	  al.,	  1992)	  and	  is	  expressed	  in	  both	  neurons	  and	  astrocytes.	  	  	  1.5.4	  Calcium-­‐Activated	  Anion	  Channels	  	  	  	   Calcium-­‐activated	  chloride	  channels	   (CaCCs)	  or	   ICl(Ca)	  were	   first	   identified	   in	  1982	   (Miledi,	  1982).	   	  Although	  several	   candidates	  exist	   for	  CaCCs,	   their	  molecular	  identity	   has	   not	   been	   conclusive,	   however,	   this	   may	   be	   due	   to	   different	   genes	  encoding	   different	   CaCCs	   in	   different	   cell	   types.	   	   	   CaCCs	   are	   described	  by	   a	   single	  channel	   conductance	   ~5ps,	   that	   is	   either	   directly	   activated	   by	   Ca2+	   or	   by	  Ca2+/calmodulin-­‐dependent	  kinase	   II	   	   (CaMKII)	   (Hartzell	   et	   al.,	   2005).	   	  A	   common	  protocol	   used	   to	   activate	   CaCCs	   is	   to	   trigger	   Ca2+	   entry	   via	   voltage	   gated	   calcium	  channels	   (VGCCs)	   with	   a	   depolarizing	   step,	   CaCCs	   are	   slowly	   inactivating	   and	  generally	  give	  rise	  to	  a	  prolonged	  tail	  current.	  	  CaCCs	  can	  also	  be	  activated	  by	  other	  sources	   of	   Ca2+	   entry	   such	   as	   TRP	   channels,	   cyclic	   nucleotide	   gated	   channels,	   and	  release	  of	  Ca2+	  from	  intracellular	  stores.	  	  	  It	  can	  additionally	  not	  be	  ruled	  out	  that	  the	  VSOR	   may	   in	   fact	   be	   a	   CaCC	   in	   some	   cell	   types.	   	   CaCCs	   often	   show	   similar	  conductance,	  outward	  rectifying	   I-­‐V	  (current-­‐voltage)	  curves	  and	  pharmacology	   to	  VSOR.	   	   This	   is	   in	   line	   with	   work	   in	   cultured	   astrocytes	   showing	   that	   hyposmotic	  activation	  of	  VRAC	  was	  secondary	  to	  ATP	  release,	  and	  that	  ATP	  itself	  could	  trigger	  	   35	  activation	   of	   VRAC	   like	   currents	   (Darby	   et	   al.,	   2003).	   	   TMEM16A,	   a	   recently	  identified	  CaCC	  can	  also	  be	  activated	  by	  cell	  swelling	  via	  secondary	  ATP	  release	  and	  contribute	   to	   regulatory	   volume	  decrease	   (Almaca	   et	   al.,	   2009).	   	  Of	   the	  molecular	  candidates	  for	  CaCC,	  TMEM16A	  most	  closely	  resembles	  the	  properties	  of	  the	  initially	  described	  ICl(Ca)	  (Schroeder	  et	  al.,	  2008).	   	  TMEM16B,	  also	  operates	  as	  an	  outwardly	  rectifying	  CaCC	  and	  was	  recently	  shown	  to	  be	   functionally	  expressed	   in	  pyramidal	  neurons	  of	   the	  hippocampus	   (Huang	  et	   al.,	   2012).	   	  Bestrophins	   represent	   another	  newly	   discovered	   family	   of	   proteins	   that	   exhibit	   Cl-­‐	   channel	   activity	   and	   also	  function	  as	  modulators	  of	  voltage	  gated	  calcium	  channels.	  Bestrophin-­‐1,	   a	   calcium	  and	   volume	   sensitive	   channel	   (Fischmeister	   and	   Hartzell,	   2005)	   was	   recently	  suggested	   as	   a	   pathway	   for	   non-­‐vesicular	   glutamate	   and	   GABA	   release	   from	  astrocytes	  (Park	  et	  al.,	  2009;	  Lee	  et	  al.,	  2010),	  and	  possibly	  underlies	  the	  molecular	  identity	  of	  the	  astrocyte	  VRAC.	  	   	  The	  tweety	  family	  of	  anion	  channels	  represent	  yet	  another	  possible	  candidate	  for	  a	  CaCC	  in	  neurons	  (Suzuki,	  2006).	   	  Tweety	  proteins	  show	   linear	   I-­‐V	   curves	   with	   large	   single	   channel	   conductance	   (260pS),	   and	   are	  osmosensitive,	   similar	   to	  maxi-­‐anion	   channels	   (Suzuki	   and	  Mizuno,	  2004).	   	   	  Given	  the	  diversity	  of	  proteins	  that	  function	  as	  CaCCs,	  it	  is	  most	  likely	  that	  different	  genes	  encode	  for	  different	  CaCCs	  in	  different	  cell	  types	  and	  tissues.	  	  1.5.5	  Cation-­‐Chloride	  Cotransporters	  	  	   The	  mammalian	  cation-­‐chloride	  cotransporters	  are	  encoded	  by	   the	  SLC12A	  gene	   family	   and	   consist	   of	   9	   distinct	   members,	   seven	   of	   which	   are	   known	   to	   be	  	   36	  expressed	   at	   the	   plasma	  membrane	   (Blaesse	   et	   al.,	   2009).	   	   Of	   the	   cation-­‐chloride	  cotransporters,	   the	  Na+,	  K+,	  2Cl-­‐	   transporter	   isoform	  1	   (NKCC1)	  and	  all	   four	  K+,	  Cl-­‐	  cotransporters	   (KCC1-­‐4)	   are	   expressed	   in	   the	   brain	   at	   some	   point	   in	   CNS	  development.	   	   KCC1	   is	   not	   expressed	   in	   neurons	   whereas	   KCC2	   shows	   exclusive	  expression	  in	  neurons,	  and	  has	  therefore	  been	  the	  most	  extensively	  studied	  KCC	  in	  the	  brain.	   	  Although	  KCC3	  has	  been	  detected	   in	  all	  adult	  CNS	  regions,	  not	  much	   is	  known	   in	   regards	   to	   its	   functional	   contribution	   towards	   neuronal	   Cl-­‐	   regulation.	  	  	  NKCC1	  and	  KCC2	  are	  commonly	  studied	  by	  antagonism	  with	  loop	  diuretics,	  such	  as	  furosemide	   and	   bumetanide,	   however	   these	   results	   are	   not	   always	   conclusive	   as	  they	  inhibit	  both	  NKCC1	  and	  KCC2	  in	  a	  concentration	  dependent	  manner	  (Blaesse	  et	  al.,	  2009).	  Expression	  of	  NKCC1	  and	  KCC2	  are	  thought	  to	  be	  the	  main	  transporters	  responsible	   for	   the	   switch	   in	   the	   polarity	   of	   GABAAR	   currents	   that	   occurs	   in	  development.	   	  Whereas	  immature	  neurons	  express	  NKCC1	  and	  low	  levels	  of	  KCC2,	  mature	   neurons	   are	   reported	   to	   express	   higher	   levels	   of	   KCC2,	   whereas	   NKCC1	  expression	  drops	  substantially	  (Ben-­‐Ari	  et	  al.,	  2007;	  Blaesse	  et	  al.,	  2009).	  As	  NKCC1	  and	   KCC2	   are	   electroneutral	   transporters	   equilibrium	   is	   defined	   solely	   by	   the	  concentration	   gradients	   of	   the	   transported	   ions.	   	   NKCC1	   polarity	   is	   defined	   by	  	  [Na+]o[K+]o[Cl-­‐]2o=[Na+]i[K+]i[Cl-­‐]2i.	   	  Given	   this	   stoichiometry	  one	  would	  expect	   that	  by	  increasing	  intracellular	  Na+	  and	  Cl-­‐	  the	  driving	  force	  for	  Cl-­‐	  entry	  by	  NKCC1	  would	  be	   reduced.	   	   	  As	  KCC2	   is	   the	  major	  neuronal	   chloride	  extrusion	  mechanism,	  KCC2	  may	   be	   important	   for	   the	   recovery	   of	   ECl-­‐	   following	   excitotoxic	   Cl-­‐	   increases.	  	  However,	  [K+]ext	  increases	  that	  occur	  will	  decrease	  and	  could	  possibly	  even	  reverse	  KCC2	  polarity,	  as	  KCC2	  functions	  close	  to	  equilibrium	  in	  mature	  pyramidal	  neurons.	  	  	   37	  In	   addition,	   it’s	   been	   reported	   that	   glutamate	   activation	   of	   NMDARs	   leads	   to	  phosphorylation	  and	  decreased	  expression	  of	  KCC2,	   leading	  to	  decreased	  recovery	  from	  excitotoxic	  Cl-­‐	  loads	  (Lee	  et	  al.,	  2011).	  	  However,	  the	  authors	  of	  this	  study	  were	  unable	   to	   conclude	   the	  entry	  pathway	  of	  Cl-­‐,	   but	  noted	   that	   it	  was	   independent	  of	  NKCC1.	  	  	  Consistent	  with	  these	  results,	  another	  study	  examining	  mechanisms	  for	  Cl-­‐	  influx	  during	  OGD	  in	  hippocampal	  brain	  slices	  saw	  no	  reduction	  in	  Cl-­‐	  increases	  by	  blocking	   cation-­‐chloride	   cotransporters	   with	   bumetanide/furosemide,	   or	   by	  blocking	   GABAARs	   with	   picrotoxin	   (Pond	   et	   al.,	   2006).	   	   Interestingly,	   the	   authors	  were	   able	   to	   block	   a	   secondary	   increase	   in	   Cl-­‐	   that	   occurred	   (>1	   hour)	   after	   re-­‐oxygenation,	  although	  the	  significance	  of	  this	  late	  Cl-­‐	  influx	  in	  the	  brain	  slice	  remains	  uncertain.	  	  	  1.5.6	  SLC4	  Family	  of	  Anion	  Exchangers	  and	  Transporters	  	  	   The	  mammalian	  SLC4	  family	  is	  made	  up	  of	  10	  functional	  members	  (SLC4A1	  to	  -­‐4A5	  and	  SLC4A7	  to	  -­‐4A11).	  	  All	  members	  of	  the	  SLC4	  family	  except	  for	  SLC4A11	  mediate	  the	  transport	  of	  HCO3-­‐.	  	  Of	  the	  SLC4	  family,	  -­‐A1,	  -­‐A2,	  -­‐A3,	  -­‐A8	  and	  -­‐A10	  have	  been	  reported	  to	  also	  transport	  Cl-­‐	  in	  exchange	  for	  HCO3-­‐	  (Pushkin	  and	  Kurtz,	  2006;	  Alper,	   2009;	   Romero	   et	   al.,	   2013).	   	   Another	   common	   feature	   to	   all	   SLC4	   family	  members	   is	   the	   reported	   inhibition	  by	  disulfonic	   stillbenes,	  DIDS	  and	  SITS	   (Alper,	  2009;	  Romero	  et	  al.,	  2013).	   	  The	  anion	  exchangers	  that	  mediate	  the	  electroneutral,	  sodium	  independent	  exchange	  of	  Cl-­‐	  for	  HCO3-­‐	  are	  the	  anion	  exchangers	  -­‐	  AE1,	  AE2	  and	   AE3	   encoded	   by	   the	   genes	   SLC4A1,	   SLC4A2	   and	   SLC4A3	   respectively	   (Alper,	  	   38	  2009).	  AE1	  to	  -­‐3	  prefer	  Cl-­‐	  and	  HCO3-­‐	  as	  substrates,	  however	  they	  can	  also	  transport	  OH-­‐,	   and	   AE1	   has	   been	   reported	   to	   cotransport	   SO42-­‐	   and	   H+	   in	   exchange	   for	   Cl-­‐,	  although	   at	   a	   very	   low	   transport	   rate	   compared	   to	   Cl-­‐/HCO3-­‐	   exchange	   (Jennings,	  1976).	   	  AE3	  is	  predominantly	  expressed	  in	  the	  brain	  and	  in	  the	  heart	  and	  is	  highly	  expressed	   in	  hippocampal	  and	  cortical	  pyramidal	  neurons	  (Hentschke	  et	  al.,	  2006;	  Svichar	  et	   al.,	   2009).	   	  AE1	  and	  AE2	  are	  also	  expressed	   in	   the	  brain,	  however,	  AE1	  was	   shown	   to	   be	   absent	   from	   pyramidal	   neurons	   and	   AE3	  mRNA	   expression	   has	  been	  reported	  to	  be	  5	  fold	  higher	  than	  AE2	  (Svichar	  et	  al.,	  2009).	  	  Additionally,	  the	  generation	  of	  SLC4A3	  knockout	  mice	  has	  revealed	  that	  AE3	  is	  the	  dominant	  sodium-­‐independent	  anion	  exchanger	  in	  the	  hippocampus,	  mediating	  acid	  loads	  in	  response	  to	  alkalosis	  (Hentschke	  et	  al.,	  2006;	  Svichar	  et	  al.,	  2009).	  	  Interestingly,	  AE3	  KO	  mice	  show	  reduced	  seizure	  threshold	  possibly	  due	  to	  impaired	  pH	  regulation	  (Hentschke	  et	  al.,	  2006),	  however	  it	  is	  also	  possible	  that	  the	  Cl-­‐	  transport	  by	  AE3	  may	  influence	  the	   reversal	  potential	   for	  GABAAR	  mediated	   inhibitory	   transmission,	   as	  previously	  reported	   in	   embryonic	  motorneurons	   (Gonzalez-­‐Islas	   et	   al.,	   2009).	   	   In	   addition	   to	  the	   sodium-­‐independent	   Cl-­‐/HCO3-­‐	  exchangers,	   the	   brain	   also	   shows	   expression	   of	  two	  distinct	  Na+	   coupled	  Cl-­‐/HCO3-­‐	   exchangers,	   encoded	  by	   the	  genes	  SLC4A8	  and	  SLC4A10.	   	   The	   Na+	   driven	   Cl-­‐/HCO3-­‐	  exchanger	   (NDCBE),	   SLC4A8,	   is	   expressed	   in	  both	  hippocampal	  and	  cortical	  neurons	  of	  rodent	  and	  human	  brain	  (Damkier	  et	  al.,	  2007;	   Chen	   et	   al.,	   2008;	   Sinning	   et	   al.,	   2011).	   	   Under	   resting	   conditions,	   SLC4A8	  mitigates	   against	   acid	   loads	   by	   exchanging	   extracellular	  HCO3-­‐	   for	   intracellular	   Cl-­‐	  driven	   by	   the	   large	   inward	   Na+	   gradient.	   	   However,	   it	   is	   possible	   for	   NDBCE	   to	  reverse	  depending	  on	   the	   intracellular	   and	   extracellular	   concentrations	  of	  Na+,	   Cl-­‐	  	   39	  and	  HCO3-­‐	  (Grichtchenko	  et	  al.,	  2001).	  	  Since	  NDCBE	  is	  an	  electroneutral	  exchanger	  transporting	  one	  Na+	  and	  two	  HCO3-­‐	  ions	  in	  exchange	  for	  one	  Cl-­‐	  ion,	  equilibrium	  for	  NDCBE	   can	   be	   described	   by	   [Na+]o[HCO3-­‐]o2[Cl-­‐]i=[Na+]i[HCO3-­‐]i2[Cl-­‐]o.	   	   Knockout	  mice	  for	  the	  SLC4A8	  gene	  were	  recently	  shown	  to	  exhibit	  diminished	  acid	  extrusion	  and	   decreased	   spontaneous	   glutamate	   release	   in	   vitro	   and	   increased	   seizure	  threshold	  in	  vivo	  (Sinning	  et	  al.,	  2011).	  	  Although	  SLC4A8	  was	  recently	  shown	  to	  be	  localized	  preferentially	  to	  presynaptic	  terminals	  (Sinning	  et	  al.,	  2011;	  Burette	  et	  al.,	  2012),	  the	  antibodies	  used	  in	  these	  studies	  were	  made	  against	  a	  site	  located	  at	  the	  C-­‐terminus	  of	   the	   full	   length	  protein,	   and	  which	   is	   spliced	  out	   in	  2	   of	   the	  4	   splice	  variants	   identified	   in	   human	   brain	   (Parker	   et	   al.,	   2008a).	   	   Other	   studies	   using	   an	  antibody	   targeting	   the	   N-­‐terminus	   of	   SLC4A8	   show	   a	   soma/dendritic	   staining	  pattern	   in	   pyramidal	   neurons	   (Damkier	   et	   al.,	   2007;	   Chen	   et	   al.,	   2008).	   	   SLC4A10	  represents	  another	  electroneutral	  Na+,	  HCO3-­‐	  cotransporter	  that	  has	  been	  reported	  to	   exhibit	   Cl-­‐	   exchange	   activity,	   although	   this	   remains	   controversial	   (Parker	   et	   al.,	  2008b;	   Damkier	   et	   al.,	   2010;	   Romero	   et	   al.,	   2013).	   	   Whereas	   rat	   and	   mouse	  expression	  of	  SLC4A10	  products	  have	  been	  shown	  to	  function	  as	  Na+	  driven	  HCO3-­‐/Cl-­‐	  exchangers	  similar	  to	  SLC4A8	  (Damkier	  et	  al.,	  2010),	  human	  SLC4A10	  expressed	  in	  Xenopus	  Oocytes	  showed	  no	  functional	  requirement	  for	  the	  presence	  of	  Cl-­‐,	  and	  no	  net	  change	  in	  Cl-­‐	  concentration	  when	  Cl-­‐	  was	  present	  (Parker	  et	  al.,	  2008b).	   	  These	  results	  suggest	  the	  difference	  in	  reported	  Cl-­‐	  exchange	  activity	  may	  be	  simply	  due	  to	  differences	   in	   rodent	   vs.	   human	   SLC4A10	   gene	   products,	   however,	   functional	  differences	  due	   to	  different	   expression	   systems	   cannot	  be	   ruled	  out.	   	   	   SLC4A10	   is	  expressed	   in	   the	  brain	   (Damkier	  et	  al.,	  2007),	  with	  high	  expression	  of	  SLC4A10	   in	  	   40	  CA3	  neurons	  of	  the	  hippocampus	  (Jacobs	  et	  al.,	  2008).	  	  Genetic	  deletion	  of	  SLC4A10	  has	  been	  shown	  to	  result	  in	  decreased	  recovery	  from	  acid	  loads	  in	  the	  CA3	  region	  of	  the	   hippocampus,	   increased	   seizure	   threshold	   and	   decreased	   brain	   ventricle	   size	  (Jacobs	  et	  al.,	  2008).	  	  	  1.5.7	  SLC26	  Family	  of	  Anion	  Exchangers	  and	  Transporters	  	  	   The	   solute	   linked	   carrier	   26	   isoforms,	   originally	   described	   as	   sulfate	  transporters,	   are	   made	   up	   of	   a	   family	   of	   anion	   transporters	   with	   10	   distinct	  members.	   	   The	   isoforms	   are	   made	   up	   of	   SLC26A1-­‐A11,	   with	   SLC26A10	   being	   a	  pseudo-­‐gene.	   	   It	   is	  now	  clear	  that	  most	  SLC26	  isoforms	  are	  actually	  quite	  versatile	  capable	   transporting	   a	   variety	   of	   different	   anions	   such	   as	   chloride,	   bicarbonate,	  sulfate,	  formate,	  iodide,	  oxalate	  and	  hydroxyl	  (Mount	  and	  Romero,	  2004;	  Romero	  et	  al.,	  2004;	  Alper	  and	  Sharma,	  2013).	  	  Various	  SLC26	  members	  have	  been	  shown	  to	  be	  highly	  expressed	  in	  the	  kidney	  and	  play	  critical	  roles	  in	  kidney	  salt	  adsorption,	  acid-­‐base	  balance,	   vascular	  volume	  homeostasis	   and	  blood	  pressure	   regulation	   (Mount	  and	  Romero,	  2004;	  Romero	  et	  al.,	  2004;	  Alper	  and	  Sharma,	  2013).	   	   	   	  SLC26A3,	  A4,	  A6,	   A9	   and	   A11	   have	   all	   been	   shown	   to	   mediate	   Cl-­‐/HCO3-­‐	   exchange	   (Soleimani,	  2013).	   	   Several	   SLC26	   members	   have	   also	   been	   shown	   to	   be	   expressed	   in	   brain	  tissue,	   including	   SLC26A11	   which	   was	   recently	   shown	   to	   be	   highly	   expressed	   in	  pyramidal	   neurons	   of	   the	   cortex	   and	   hippocampus	   (Rahmati	   et	   al.,	   2013),	   and	  named	  Kidney	   Brain	   Anion	   Transporter	   (KBAT).	   	   	   Of	   note,	   SLC26A11	   is	   the	  most	  divergent	  member	  of	  the	  SLC26	  family,	  more	  closely	  related	  to	  sulfate	  transporters	  	   41	  from	   yeast	   and	   plants	   than	   to	   other	   members	   of	   the	   mammalian	   SLC26	   family	  (Vincourt	   et	   al.,	   2003;	   Soleimani,	   2013).	   	   SLC26A11	  has	  been	   shown	   to	  be	   a	  DIDS	  sensitive	   transporter	   capable	   of	   functioning	   in	   several	   different	   modes;	   a	   sulfate	  transporter,	  an	  exchanger	  for	  Cl-­‐,	  SO42-­‐,	  HCO3-­‐	  or	  H+-­‐Cl-­‐	  or	  as	  a	  Cl-­‐	  channel,	  depending	  on	   the	   tissue	   type	  or	   the	  expression	  system	  (Vincourt	  et	  al.,	  2003;	  Xu	  et	  al.,	  2011;	  Lee	  et	  al.,	  2012a;	  Rahmati	  et	  al.,	  2013).	  	  The	  explanation	  for	  these	  differences	  in	  the	  reported	  properties	   of	   SLC26A11	   are	   unclear	   but	  may	   represent	   variations	   in	   the	  recombinant	  protein	   that	  was	  expressed,	  differences	   in	   the	  expression	   systems	  or	  physiological	   changes	   in	   the	   functional	   states	   of	   SLC26A11.	   The	   degree	   of	   DIDS	  sensitivity	  of	  SLC26A11	  has	  also	  been	  reported	  to	  vary	  depending	  on	  the	  expression	  system	   used	   to	   analyze	   the	   properties	   of	   recombinant	   SLC26A11.	   	   	   Recently	  SLC26A11	  was	   shown	   to	   be	   able	   to	   function	   as	   an	   electrogenic	   Cl-­‐	   transporter	   by	  measurements	  of	  36Cl	  flux	  and	  also	  as	  a	  Cl-­‐/HCO3-­‐	  exchanger	  by	  measurements	  of	  pH	  with	   the	   pH	   indicator	   2’,7’-­‐bis(2-­‐carboxymethyl),5(and-­‐6)carboxyfluorescein	  (BCECF)	   (Xu	   et	   al.,	   2011).	   	   SLC26A11	   has	   also	   been	   proposed	   to	   function	   as	   a	  constitutively	  active	  plasma	  membrane	  Cl-­‐	  channel	  independent	  of	  intracellular	  Ca2+	  or	   cAMP,	   based	   on	   voltage	   clamp	   recordings	   in	   HEK-­‐293	   cells	   overexpressing	  SLC26A11	  (Rahmati	  et	  al.,	  2013).	   	  SLC26A11	  was	  reported	   to	  be	  co-­‐localized	  with	  the	   vacuolar	   H+-­‐ATPase	   in	   the	   plasma	   membrane	   of	   Purkinje	   neurons	   and	   of	  intercalated	   kidney	   cells	   (Xu	   et	   al.,	   2011;	   Rahmati	   et	   al.,	   2013),	   where	   they	   may	  functionally	   cooperate.	   	   	   	   SLC26A11	   is	   trafficked	   to	   the	   plasma	   membrane	   in	  expression	   systems	   with	   the	   N-­‐terminus	   of	   the	   protein	   directed	   towards	   the	  extracellular	   side	   (Vincourt	   et	   al.,	   2003).	   	   Although	   the	   abundant	   expression	   of	  	   42	  SLC26A11	   protein	   has	   been	   reported	   in	   neurons,	   the	   function	   of	   endogenous	  SLC26A11	  expressed	  in	  neurons	  remains	  untested.	  	   43	  Chapter	  2:	  Lipid	  Nanoparticle	  Delivery	  of	  siRNA	  to	  Silence	  Neuronal	  Gene	  Expression	  in	  the	  Brain.	  	  2.1	  Introduction	  	   	   Since	   the	   discovery	   that	   RNA	   interference	   (RNAi)	   is	   mediated	   by	   double-­‐stranded	  RNA	  (dsRNA)	  (Fire	  et	  al.,	  1998),	  the	  use	  of	  small	  interfering	  RNA	  (siRNA)	  to	   silence	   specific	   genes	   has	   become	   a	   powerful	   method	   for	   manipulating	   gene	  expression	   in	   vitro	   and,	   increasingly,	   in	   vivo.	   	   	   However,	   issues	   concerning	   the	  delivery	  of	  siRNA	  into	  neurons	  both	  in	  vitro	  and	  in	  vivo	  limit	  the	  widespread	  use	  of	  siRNA	   in	   neuroscience	   research	   in	  mammals.	   	   Viral	   delivery	   of	   short	   hairpin	  RNA	  (shRNA)	  has	  been	  used	  successfully	   in	  vivo	   (e.g.	   (Sun	  et	  al.,	  2009))	   to	  knock	  down	  selected	   targets,	   but	   the	   time	   and	   expense	   of	   packaging	   shRNA	   into	   high	   titre	  viruses,	   as	   well	   as	   the	   toxicological	   and	   immunological	   problems	   associated	  with	  viral	  vectors,	  must	  be	  considered.	  	  In	  cell	  culture	  the	  utility	  of	  siRNA	  approaches	  for	  silencing	   genes	   in	   neurons	   remains	   limited	   due	   to	   low	   transfection	   levels	   and	  toxicity	  with	  techniques	  such	  as	  lipofectamine.	  	  	  Transgenic	  approaches	  to	  modulate	  CNS	   gene	   expression	   are	   time	   consuming	   and	   costly.	   	   Antisense	   oligonucleotides	  (ASOs)	   can	  be	  effective	  when	  stabilized	   forms	  are	   injected	   into	   the	  brain	  but	   they	  require	   large	   quantities	   of	   ASOs	   to	   be	   injected	   for	   effective	   uptake.	   	   	   The	  development	   of	   alternative	   delivery	   methods	   to	   facilitate	   the	   use	   of	   siRNA	   to	  manipulate	   gene	   expression	   in	   the	   mammalian	   CNS	   would	   be	   of	   great	   value	   to	  	   44	  neuroscientists,	   and	   would	   accelerate	   progress	   in	   our	   understanding	   of	   brain	  function.	  	  	  Lipid	   nanoparticles	   (LNPs)	   are	   currently	   the	   leading	   delivery	   systems	   for	  enabling	  the	  therapeutic	  potential	  of	  siRNA	  in	  peripheral	  cells	  (Zimmermann	  et	  al.,	  2006;	   Davidson	   and	   McCray,	   2011).	   LNP	   siRNA	   systems	   containing	   optimized	  cationic	   lipids	   can	   silence	   therapeutically	   relevant	   genes	   in	   a	   variety	   of	   tissues	  (particularly	   liver)	   (Semple	   et	   al.,	   2010;	   Basha	   et	   al.,	   2011;	   Lee	   et	   al.,	   2012b)	  following	  intravenous	  injection	  in	  animal	  models.	  Positive	  clinical	  trial	  results	  using	  these	   LNPs	   have	   been	   reported	   for	   treatment	   of	   cardiovascular	   disease,	   certain	  forms	   of	   amyloidosis	   and	   other	   disorders	   (http://www.alnylam.com/Programs-­‐and-­‐Pipeline/Alnylam-­‐5x15/index.php).	   However	   the	   efficacy	   of	   LNP	   approaches	  for	  delivering	  siRNA	  to	  neurons	  in	  the	  CNS	  is	  unknown.	  	  Due	  to	  the	  inability	  of	  LNP	  systems	  to	  cross	  the	  blood-­‐brain	  barrier	  the	  potency	  of	  these	  systems	  for	  silencing	  genes	   in	   brain	   tissue	   has	   not	   been	   investigated.	   	   Here	   we	   report	   the	   conditions	  under	  which	  LNP	  delivery	  of	   siRNA	   is	   a	   remarkably	   efficient	  method	   for	   silencing	  neuronal	   gene	   expression	   in	   both	   primary	   neuronal	   culture	   and	   following	  intracranial	  injection	  in	  vivo.	  	  	   45	  2.2	  Materials	  and	  Methods	  	  2.2.1	  	  Synthesis	  of	  the	  Lipid,	  3-­‐(dimethylamino)propyl	  3,3-­‐bis(linoleyl)	  propionate,	  DMAP-­‐BLP,	  Compound	  8	  	  	  	  (12Z,15Z)	   3-­‐(dimethylamino)propyl	   3-­‐((9Z,12Z)-­‐heptadeca-­‐9,12-­‐dien-­‐1-­‐yl)icosa-­‐12,15-­‐dienoate	  	  The	   pKa	   of	   the	   amine	   headgroup	   is	   6.64	   and	   the	   ED50	   is	   0.009	   mg/kg	   for	   FVII	  knockdown.	  	  Scheme	  1	  	  	  	   46	  i)	  Ms-­‐Cl,	  TEA,	  DMAP,	  DCM;	  ii)	  NaCN,	  DMF,	  53-­‐72%;	  iii)	  DIBAL-­‐H,	  toluene,	  68-­‐69	  %;	  iv)	  NaBH4,	  THF/methanol,	  86%;	  v)	  oxone,	  DMF,	  69%;	  vi)	  EDAC,	  DMAP,	  DIEA,	  DCM,	  81%.	  	  	  Synthesis	  of	  compound	  2	  To	   a	   solution	   of	   compound	   1	   (50	   g,	   94.45	  mmol)	   in	   DCM	   (400	  mL)	   under	   argon	  atmosphere	  was	  added	  TEA	  (53	  mL,	  378	  mmol)	  and	  DMAP	  (1.2	  g,	  9.5	  mmol),	  and	  the	  solution	  was	  stirred	  at	  room	  temperature	  for	  5	  minutes.	  	  The	  reaction	  mass	  was	  cooled	   to	   -­‐5	   °C,	   and	   a	   solution	   of	  methyl	   sulfonyl	   chloride	   (15	  mL,	   190	  mmol)	   in	  DCM	   (100	  mL)	  was	   added	   slowly	   at	   temperature	   below	   -­‐5°C	   and	   then	   allowed	   to	  warm	  to	  room	  temperature.	  After	  30	  minutes,	   the	  reaction	  was	  quenched	  with	   ice	  cold	  water	  (20	  ml).	  	  The	  organic	  layer	  was	  separated,	  washed	  with	  1	  N	  HCl	  (30	  mL),	  water,	  and	  brine,	  dried	  over	  sodium	  sulfate,	  and	  evaporated	  at	  reduced	  pressure	  to	  obtain	  the	  mesylate	  as	  a	  pale	  yellow	  liquid	  (55	  g,	  95%).	  1H	  NMR	  (400	  MHz,	  CDCl3):	  d	  0.89	  (t,	  6H,	  J	  =	  6.8),	  1.2-­‐1.5	  (m,	  36H),	  1.67	  (m,	  4H),	  2.05	  (q,	  8H,	  J1	  =	  6.8,	  J2	  =	  6.8),	  2.77	   (t,	  4H,	   J	  =	  6.4),	  2.99	   (s,	  3H),	  4.71(m,	  1H)	  and	  5.36	   (m,	  8H).	  This	  product	  was	  used	  for	  the	  next	  step	  without	  any	  further	  purification.	  To	  a	  solution	  of	  the	  mesylate	  (55	   g,	   90.6	  mmol)	   in	   dimethylformamide	   (400	  mL)	  was	   added	   sodium	   cyanide	   at	  room	  temperature.	  Reaction	  mixture	  was	  heated	   to	  65	   °C	   for	  16	  hrs.	  The	  reaction	  was	   monitored	   by	   TLC	   (10%	   ether-­‐hexane).	   	   It	   was	   then	   cooled	   to	   room	  temperature	  and	  diluted	  with	  7	  volumes	  of	  water	  and	  extracted	  with	  ether	   (three	  times	  5	  volumes).	  The	  combined	  ether	  layers	  were	  washed	  with	  water	  (two	  times	  3	  volumes)	   and	   brine	   (two	   times	   3	   volumes),	   dried	   over	   sodium	   sulfate,	   and	  	   47	  evaporated	  at	  reduced	  pressure	  to	  obtain	  the	  crude	  product,	  which	  was	  purified	  by	  silica	  gel	  chromatography	  using	  a	  hexane	  as	  eluent	  yield	  compound	  2	  (26.1	  g,	  53%)	  as	  a	  pale	  yellow	  liquid.	   1H	  NMR	  (400	  MHz,	  CDCl3):	  d	  0.89	  (t,	  6	  H,	   J1	  =	  6	  Hz,	   J2	  =	  6.8	  Hz),	  1.29-­‐1.37	  (m,	  36	  H),	  1.41-­‐1.49	  (m,	  2	  H),	  1.60-­‐1.63	  (m,	  2	  H),	  2.05	  (q,	  8	  H,	  J1	  =	  6.8	  Hz,	  J2	  =	  13.6	  Hz),	  2.78	  (t,	  4	  H,	  J1	  =	  6.4	  Hz,	  J2	  =	  6	  Hz),	  5.32-­‐5.41	  (m,	  8	  H).	  	  	  Synthesis	  of	  compound	  3	  To	  a	  solution	  of	  compound	  2	  (29	  g,	  54	  mmol)	  in	  toluene	  (300	  mL)	  at	  -­‐60	  °C	  under	  argon	  atmosphere	  was	  added	  diisobutylaluminium	  hydride	  (108	  mL,	  108	  mmol,	  1	  M	  solution	  in	  toluene)	  gradually	  through	  a	  cannula.	  The	  solution	  was	  stirred	  for	  1	  hr	  at	  -­‐60	  °C.	  The	  reaction	  mixture	  was	  then	  quenched	  at	  -­‐60	  °C	  with	  saturated	  solution	  of	  sodium	   potassium	   tartrate,	   warmed	   to	   room	   temperature,	   and	   filtered	   through	   a	  celite	   bed.	   The	   filtrate	   was	   washed	   twice	   with	   500	   mL	   brine,	   dried	   over	   sodium	  sulfate,	  and	  evaporated	  at	  reduced	  pressure	  to	  obtain	  the	  crude	  product,	  which	  was	  purified	   by	   silica	   gel	   chromatography	  using	   3%	  ether/97%	  hexane	   to	   afford	   pure	  product	  3	  (19.8	  g,	  68%)	  as	  a	  pale	  yellow	  liquid.	  1H	  NMR	  (400	  MHz,	  CDCl3):	  	  d	  0.89	  (t,	  6	  H,	  J1	  =	  6.4	  Hz,	  J2	  =	  7.2	  Hz),	  1.25-­‐1.42	  (m,	  38	  H),	  1.56-­‐1.63	  (m,	  2	  H),	  2.02	  to	  2.06	  (m,	  8	  H),	  2.17	  to	  2.26	  (m,	  1	  H),	  2.78	  (t,	  5	  H,	  J	  =	  6.4	  Hz),	  5.30-­‐5.42	  (m,	  8	  H),	  9.54	  (d,	  1	  H,	  J	  =	  2.8	  Hz).	   	  13C	  NMR	  (100	  MHz,	  CDCl3):	  d	  14.0,	  22.6,	  25.6,	  27.1,	  27.2,	  28.9,	  29.2,	  29.3,	  29.4,	   29.43,	   29.6,	   29.7,	   31.5,	   52.0,	   127.9,	   128.0,	   130.1,	   128.0,	   130.2,	   205.6.	   MS	  calculated	  for	  C38H68O	  540.52,	  found	  541.54	  (M+H).	  	  	  Synthesis	  of	  compound	  4	  	   48	  To	  a	  solution	  of	  compound	  3	  (10	  g,	  18.4	  mmol)	  in	  THF/methanol	  (1:1,	  100	  mL)	  at	  0	  °C	   was	   added	   sodium	   borohydride	   (1.4	   g,	   37	   mmol).	   The	   solution	   was	   gradually	  warmed	  to	  room	  temperature	  and	  stirred	  for	  1	  hr.	  The	  reaction	  was	  quenched	  with	  ice-­‐cold	  water	  and	  extracted	  with	  three	  times	  with	  ether	  (100	  ml).	  The	  organic	  layer	  was	  washed	  once	  with	  brine	  (100	  mL),	  dried	  over	  sodium	  sulfate,	  and	  evaporated	  at	  reduced	   pressure	   to	   obtain	   the	   crude	   product,	   which	   was	   purified	   by	   silica	   gel	  chromatography	  using	  3%	  ether/97%	  hexane	  to	  afford	  pure	  product	  4	  (8.7	  g,	  86%)	  as	  pale	  yellow	  liquid.	  1H	  NMR	  (400	  MHz,	  CDCl3):	  d	  0.87	  (t,	  6	  H,	  J1	  =	  6.8	  Hz),	  1.25-­‐1.33	  (m,	  40	  H),	  1.4-­‐1.5	  (m,	  1	  H),	  2.05	  (m,	  8	  H),	  2.78	  (t,	  5	  H,	  J	  =	  6.4	  Hz),	  3.53	  (d,	  2	  H,	  J	  =	  5.2	  Hz),	  5.30-­‐5.42	   (m,	  8	  H).	   	   13C	  NMR	  (100	  MHz,	  CDCl3):	  d	  14.1,	  22.6,	  25.6,	  26.9,	  27.2,	  29.3	  29.6,	  29.6,	  29.7,	  30.1	  30.9,	  31.5,	  40.5,	  127.9,	  130.1.	  MS	  calculated	   for	  C38H70O	  542.54,	  found	  542.90	  (M+).	  	  	  Synthesis	  of	  compound	  5	  To	   a	   solution	   of	   compound	   4	   (215	   g,	   390	   mmol)	   in	   DCM	   (2	   L)	   under	   nitrogen	  atmosphere	  was	  added	  TEA	  (220	  mL,	  1.58	  mol)	  and	  DMAP	  (4.83	  g,	  39	  mmol).	  The	  solution	  was	  stirred	  at	   room	  temperature.	   	  The	  reaction	  mass	  was	  cooled	   to	   -­‐5	  °C	  and	   a	   solution	   of	   mesyl	   chloride	   in	   DCM	   was	   slowly	   added	   ensuring	   that	   the	  temperature	   remained	   below	   -­‐5	   °C.	   The	   solution	   was	   allowed	   to	   warm	   to	   room	  temperature	   and	   stirred	   for	   1	   hr.	   The	   reaction	   was	   then	   quenched	   with	   ice	   cold	  water.	  The	  organic	   layer	  was	  separated,	  washed	  successively	  with	  1	  N	  HCl,	  water,	  and	  brine,	  dried	  over	  sodium	  sulfate,	  and	  evaporated	  at	  reduced	  pressure	  to	  obtain	  the	  pure	  product	  as	  a	  yellow	  liquid	  (250	  g,	  96%).1HNMR	  (400	  MHz,	  CDCl3):	  d	  0.89	  (t,	  	   49	  6	  H,	  J	  =	  6.8	  Hz),	  1.2	  –	  1.4	  (m,	  40	  H),	  1.65	  –	  1.72	  (m,	  1	  H),	  2.05	  (dd,	  8	  H,	  J	  =	  6.8	  Hz,	  J	  =	  13.6Hz),	  2.78	  (t,	  4	  H,	  J	  =	  6.4	  Hz),	  3.00	  (s,	  3	  H),	  4.12	  (d,	  2	  H,	  J	  =	  5.6	  Hz),	  5.29	  –	  5.43	  (m,	  8	  H).	  13C	  NMR	  (100	  MHz,	  CDCl3):	  d	  14.0,	  22.5,	  25.6,	  26.5,	  27.1,	  27.2,	  29.2,	  29.3,	  29.5,	  29.6,	  29.8,	  30.6,	  31.5,	  37.1,	  37.7,	  72.4,	  127.85,	  127.9,	  130.0,	  130.1.	  To	  a	  solution	  of	  the	  mesylate	  in	  dimethylformamide	  (1.75	  L)	  was	  added	  sodium	  cyanide	  (59	  g,	  1.20	  mol)	  at	  room	  temperature	  under	  argon.	  The	  solution	  was	  heated	  to	  65	  °C	  for	  16	  hrs.	  The	  reaction	  was	  cooled	  to	  0	  °C	  and	  added	  to	  cold	  water	  (6L)	  with	  stirring,	  and	  the	  temperature	   was	   maintained	   below	   10	   °C.	   The	   mixture	   was	   extracted	   with	  diethylether	   (five	   times	   2	   L),	   and	   the	   combined	   organic	   layers	   were	   washed	  successively	   with	   water	   and	   brine,	   dried	   over	   sodium	   sulfate,	   and	   evaporated	   at	  reduced	   pressure	   to	   obtain	   the	   crude	   product,	   which	   was	   purified	   by	   silica	   gel	  chromatography	  using	  1%	  ether/99%	  hexane	  as	  eluent	   to	  yield	  product	  5	   (160	  g,	  72%)	  as	  a	  pale	  yellow	  liquid.	  	  1H	  NMR	  (400	  MHz,	  CDCl3):	  d	  0.89	  (t,	  6	  H,	  J	  =	  6.8	  Hz),	  1.2	  –	  1.46	  (m,	  40	  H),	  1.64	  –	  1.71	  (m,	  1	  H),	  2.05	  (dd,	  8	  H,	  J	  =	  6.8	  Hz,	  13.6	  Hz),	  2.32	  (d,	  2	  H,	  J	  =	  6	  Hz),	  2.78	  (t,	  4	  H,	  J	  =	  6.4	  Hz),	  5.29	  –	  5.43	  (m,	  8	  H).	  	  Synthesis	  of	  compound	  6	  To	  a	  solution	  of	  compound	  5	  (55	  g,	  100	  mmol)	  in	  toluene	  (500	  mL)	  at	  -­‐70	  °C	  under	  argon	  atmosphere	  was	  added	  diisobutylaluminium	  hydride	  (200	  mL,	  200	  mmol,	  1	  M	  solution	  in	  toluene)	  gradually	  through	  a	  cannula.	  The	  solution	  was	  stirred	  for	  1	  hr	  at	  -­‐70	  °C.	  The	  reaction	  mixture	  was	  then	  quenched	  at	  -­‐70	  °C	  with	  a	  saturated	  solution	  of	  sodium	  potassium	  tartrate,	  warmed	  to	  room	  temperature,	  and	  filtered	  through	  a	  celite	   bed.	   The	   filtrate	   was	   evaporated	   at	   reduced	   pressure	   to	   obtain	   the	   crude	  	   50	  product.	   The	   crude	   product	   was	   dissolved	   in	   THF	   (6	   volumes),	   and	   1	   N	   HCl	   (10	  volumes)	  was	  added.	  The	  solution	  was	  stirred	  for	  1	  hr	  at	  room	  temperature.	  Ether	  was	  added,	  and	  the	  mixture	  was	  transferred	  to	  a	  separatory	  funnel.	  The	  ether	  layer	  was	  washed	  four	  times	  with	  water	  (400	  mL),	  then	  with	  saturated	  NaHCO3	  solution	  (200	  ml),	  and	  finally	  with	  brine	  (200	  ml),	  dried	  over	  sodium	  sulfate,	  and	  evaporated	  at	   reduced	   pressure	   to	   obtain	   the	   crude	   product,	  which	  was	   purified	   using	   230	   x	  400	  mesh	  silica	  gel	  chromatography	  with	  2%	  ether/98%	  hexane	  as	  eluent	  to	  afford	  pure	  product	  6	  (38	  g,	  69%)	  as	  a	  pale	  yellow	  liquid.	  1H	  NMR	  (400	  MHz,	  CDCl3):	  d	  0.89	  (t,	  6	  H,	  J	  =	  6.8	  Hz),	  1.2	  –	  1.41	  (m,	  40	  H),	  1.89	  –	  1.99	  (m,	  1	  H),	  2.05	  (dd,	  8	  H,	  J	  =	  6.8	  Hz,	  J	  =	  13.6	  Hz),	  2.32	  (dd,	  2	  H,	  J	  =	  2.4	  Hz,	  J	  =	  6.4	  Hz),	  2.77	  (t,	  4	  H,	  J	  =	  6.4	  Hz),	  5.28	  –	  5.42	  (m,	  8	  H),	  9.76	  (t,	  1	  H,	  J	  =	  2.4	  Hz).	  	  Synthesis	  of	  compound	  7	  To	   a	   solution	   of	   compound	   6	   (21	   g,	   38	   mmol)	   in	   diethylether	   (85	   mL)	   and	  dimethylformamide	   (125	   mL)	   was	   added	   oxone	   (23.2	   g,	   38	   mmol)	   at	   room	  temperature.	  The	  solution	  was	  stirred	  at	  room	  temperature	  for	  5	  hrs.	  The	  reaction	  was	  quenched	  with	  cold	  water	   (500	  mL)	  and	  extracted	   five	   times	  with	  ether	   (150	  ml).	   The	   combined	   organic	   layer	   was	   washed	   successively	   with	   water	   and	   brine,	  dried	  over	  sodium	  sulfate,	  and	  evaporated	  at	  reduced	  pressure	  to	  obtain	  the	  crude	  product,	   which	   was	   purified	   by	   silica	   gel	   chromatography	   using	   5%	   ether/95%	  hexane	  as	  eluent	  to	  yield	  the	  product	  7	  as	  a	  pale	  yellow	  liquid	  (15	  g,	  69%).	  1H	  NMR	  (400	  MHz,	  CDCl3):	  δ	  0.89	  (t,	  6	  H,	  J	  =	  6.8	  Hz),	  1.27-­‐1.42	  (m,	  40	  H),	  1.8	  –	  1.9	  (m,	  1	  H),	  2.05	  (dd,	  8	  H,	  J	  =	  6.8,	  J	  =	  13.6	  Hz),	  2.27	  (d,	  2	  H,	  J	  =	  6.4	  Hz),	  2.77	  (t,	  4	  H,	  J	  =	  6.4	  Hz),	  	   51	  5.29-­‐5.42	   (m,	  8	  H).	   13C	  NMR	  (100	  MHz,	  CDCl3):	  d	  14.0,	  22.6,	  25.6,	  26.5,	  27.2,	  27.2,	  29.3,	  29.4,	  29.5,	  29.6,	  29.7,	  29.9,	  31.5,	  33.8,	  34.8,	  39.0,	  76.7,	  77.0,	  77.3,	  127.9,	  130.0,	  180.3.	  MS	  calculated	  for	  C39H70O2	  Cal	  570.54,	  found	  570.97	  (M+).	  	  Synthesis	  of	  compound	  8	  Carboxylic	  acid	  7	  (20.00	  g,	  35.02	  mmol)	  and	  the	  alcohol	  8	  (5.78	  g,	  56	  mmol)	  were	  dissolved	  in	  dichloromethane	  (100	  mL)	  under	  argon.	  To	  this	  mixture	  EDCI	  (10.02	  g,	  1.5	  eq.),	  DMAP	  (500	  mg,	  10	  mol%),	  and	  DIEA	  (18.25	  ml,	  3	  eq.)	  were	  added,	  and	  the	  mixture	  was	   stirred	   overnight.	   The	   reaction	   progress	  was	  monitored	   by	  TLC	   (5%	  methanol	   in	   dichloromethane).	   The	   reaction	   mixture	   was	   transferred	   to	   a	  separatory	   funnel,	  diluted	  with	  DCM,	  and	  washed	  with	   twice	  water	   (200	  mL).	  The	  organic	   layer	  was	  dried	  over	   sodium	  sulfate.	   Solvent	  was	   removed,	   and	   the	   crude	  product	  was	   purified	   by	   silica	   gel	   chromatography	  using	   hexane/EtOAc	   (30-­‐50%)	  containing	   triethylamine	   as	   eluent	   to	   isolate	   the	   product	   as	   a	   colorless,	   viscous	  liquid	  (18.36	  g,	  81%).	  1H	  NMR	  (400	  MHz,	  CDCl3)	  δ	  5.48	  –	  5.24	  (m,	  8	  H),	  4.11	  (t,	  J	  =	  6.5	  Hz,	  2H),	  2.77	  (t,	  J	  =	  6.4	  Hz,	  4	  H),	  2.32	  (t,	  J	  =	  6.5	  Hz,	  2	  H),	  2.26	  –	  2.17	  (m,	  8	  H),	  2.04	  (q,	  J	  =	  6.8	  Hz,	  8	  H),	  1.82-­‐1.77	  (m	  4	  H),	  1.37-­‐1.36	  (m,	  39	  H),	  0.89	  (t,	  J	  =	  6.9	  Hz,	  6	  H).	  13C	  NMR	   (101	   MHz,	   CDCl3)	   δ	   173.84,	   130.38,	   130.35,	   128.15,	   128.14,	   62.74,	   56.54,	  45.70,	   39.49,	   35.32,	   34.11,	   31.74,	   30.15,	   29.89,	   29.83,	   29.77,	   29.56,	   29.54,	   27.45,	  27.41,	   27.27,	   26.94,	   26.78,	   25.84,	   22.78,	   14.28.	   MS	   calculated	   for	   C44H82NO2	  656.6346	  (M+H);	  found	  656.6329	  (M+H).	  	  	  	   52	  2.2.2	  Lipid	  Nanoparticle	  Formulation	  and	  siRNA	  Encapsulation	  	  	   The	   ionizable	   cationic	   lipid	   DMAP-­‐BLP	   and	   polyethylene	   glycol	   (PEG)	   lipid	  PEG-­‐DMG	   were	   synthesized	   as	   described	   above.	   	   1,2-­‐distearoyl-­‐sn-­‐glycero-­‐3-­‐phosphocholine	  (DSPC)	  and	  cholesterol	  were	  obtained	  from	  Avanti	  (Alabaster,	  AL)	  and	  Sigma-­‐Aldrich	  Co.	  (St.	  Louis,	  MO)	  respectively.	  	  Lipophilic	  carbocyanine	  dyes	  to	  monitor	   LNP	   siRNA	   uptake	   3,3′-­‐dioctadecyloxacarbocyanine	   perchlorate	   (DiOC18)	  and	   1,1'-­‐dioctadecyl-­‐3,3,3',3'-­‐tetramethylindocarbocyanine	   perchlorate	   (DiIC18)	  were	  obtained	  from	  Invitrogen	  (Carlsbad,	  CA).	   	  All	   lipid	  stocks	  were	  dissolved	  and	  maintained	  in	  100%	  ethanol.	  	  Lipids	  were	  mixed	  together	  at	  a	  molar	  %	  ratio	  of	  50%	  cationic	   lipid,	   1.5%	   PEG-­‐DMG,	   37.5%	   cholesterol,	   10%	   DSPC	   and	   1%	   DiOC18	   or	  DiIC18.	  	  LNP	  were	  prepared	  by	  mixing	  appropriate	  volumes	  of	  lipid	  stock	  solutions	  in	  ethanol	  with	  an	  aqueous	  phase	  containing	  siRNA	  duplexes	  employing	  a	  microfluidic	  micro-­‐mixer	   (Lee	   et	   al.,	   2012b)	   provided	   by	   Precision	   NanoSystems	   (Vancouver,	  BC).	  	  For	  the	  encapsulation	  of	  siRNA,	  the	  desired	  amount	  of	  siRNA	  (0.056mg	  siRNA:	  µmole	   of	   lipid)	   was	   dissolved	   in	   formulation	   buffer	   (25mM	   sodium	   acetate,	   pH	  4.0).	  	  1	  X	  volume	  of	  the	  lipid	  in	  ethanol	  and	  3	  X	  volumes	  of	  the	  siRNA	  in	  formulation	  buffer	  were	   combined	   in	   the	  microfluidic	  micro-­‐mixer	   using	   a	   dual-­‐syringe	   pump	  (Model	  S200,	  KD	  Scientific,	  Holliston,	  MA)	  to	  drive	  the	  solutions	  through	  the	  micro-­‐mixer	   at	   a	   combined	   flow	   rate	   of	   2ml/min	   (0.5mL/minute	   for	   syringe	   with	   lipid	  mixture	  and	  1.5mL/minute	  for	  syringe	  with	  siRNA	  in	  formulation	  buffer).	  	  The	  LNP-­‐siRNA	   systems	   formed	   were	   then	   dialyzed	   for	   4	   hrs	   against	   50mM	   MES/50mM	  sodium	  citrate	  buffer	  pH	  6.7	  followed	  by	  an	  overnight	  dialysis	  against	  1	  X	  phosphate	  	   53	  buffered	  saline,	  pH	  7.4	  (GIBCO,	  Carlsbad,	  CA)	  using	  Spectro/Por	  dialysis	  membranes	  (molecular	   weight	   cutoff	   12000	   –	   14000Da,	   Spectrum	   Laboratories,	   Rancho	  Dominguez,	   CA).	  	   The	   mean	   diameter	   and	   polydispersity	   of	   the	   LNP	   are	   listed	   in	  Table	   2.1.	   	   LNP	   size	  was	   determined	  by	   dynamic	   light	   scattering	   (number	  mode;	  NICOMP	  370	  Submicron	  Particle	  Sizer,	  Santa	  Barbara,	  CA).	  Encapsulation	  efficiency	  was	   determined	   by	   quantifying	   siRNA	   by	   measuring	   absorbance	   at	   260nm	   in	  samples	   collected	   before	   and	   after	   dialysis	   following	   removal	   of	   free	   siRNA	   using	  VivaPureD	   MiniH	   columns	   (Sartorius	   Stedim	   Biotech,	   Aubagne,	   France).	   	   Lipid	  concentration	  was	   determined	   by	  measurement	   of	   cholesterol	   content	   by	   using	   a	  Cholesterol	   E	   enzymatic	   assay	   (Wako	   Chemicals	   USA,	   Richmond,	   VA).	   	   The	   final	  siRNA:lipid	  ratios	  (mg/μmol)	  are	  listed	  in	  Table	  2.1.	  	  	   54	  Table	  2.1:	  	  LNP-­‐siRNA	  properties	  	  	  siRNA	   	   Size	  (nm)	   PDI	  Measured	  siRNA/Lipid	  (mg/µmol)	  PTEN	   	   55.4	   0.041	   0.053	  Luc	   	   62.5	   0.034	   0.057	  GluN1-­‐1	   	   62.0	   0.045	   0.056	  GluN1-­‐2	   	   40.8	   0.108	   0.058	  GluN1-­‐3	   	   55.9	   0.058	   0.059	  	  	  	   55	  2.2.3	  Hippocampal	  Neuronal	  Cultures	  and	  Lipid	  Nanoparticle	  Treatments.	  	  	   Hippocampal	   cultures	  were	   prepared	   as	   described	   previously	  with	   a	   slight	  modification	   from	   (Craig	   et	   al.,	   1996).	   In	   brief,	   hippocampi	  were	  dissociated	   from	  18-­‐d-­‐old	   rat	   embryos	   by	   treating	   with	   trypsin	   then	   triturated	   with	   a	   constricted	  Pasteur	   pipette.	   Subsequently,	   the	   dissociated	   cells	   were	   plated	   on	   poly-­‐l-­‐lysine	  coated	  glass	  coverslip	  using	  minimum	  essential	  medium	  (MEM)	  supplemented	  with	  10%	  horse	  serum.	  Then	  the	  coverslips	  were	  inverted	  over	  a	  feeder	  layer	  of	  astroglia	  cells	   to	   facilitate	   communication	  between	  neurons	  and	   feeder	   layer	   cells.	  Neurons	  were	  maintained	   in	   Neurobasal	   medium	  with	   B-­‐27	   and	   L-­‐glutamine	   (Invitrogen).	  Cytosine	   arabinoside	   (5μM;	   Calbiochem,	   Darmstudt,	   Germany)	   was	   added	   after	   2	  days	  in	  vitro	  to	  inhibit	  the	  proliferation	  of	  glia.	  Neuronal	  cultures	  were	  used	  for	  the	  experiment	   between	   12	   and	   14	   DIV.	   	   Cells	   were	   treated	   with	   either	   LNP-­‐	  phosphatase	   and	   tensin	   homolog	   1	   (PTEN)	   siRNA,	   LNP-­‐luciferase	   (luc)	   siRNA	   or	  with	  non-­‐encapsulated	  PTEN-­‐siRNA	  as	  a	  control.	  	  	  2.2.4	  Intracranial	  and	  Intracerebroventricular	  	  Injections	  	   All	  experimental	  protocols	  were	  approved	  by	  the	  Committee	  on	  Animal	  Care,	  University	   of	   British	   Columbia,	   and	   conducted	   in	   compliance	   with	   guidelines	  provided	  by	  the	  Canadian	  Council	  of	  Animal	  Care.	   	  Sprague-­‐Dawley	  rats	  (P22-­‐P26)	  were	   anaesthetized	  with	   isofluorane	   before	   and	   throughout	   the	   surgery.	   	   A	   small	  hole	  (diameter	  ~1mm)	  was	  drilled	  in	  the	  skull	  to	  allow	  access	  to	  the	  brain	  (-­‐2.0mm	  AP	  and	  ±3.0mm	  ML	   from	  bregma,	  0.8mm	  DV).	   	  A	  glass	  micropipette	   (tip	  diameter	  	   56	  ~40μm)	  was	  connected	  to	  a	  Hamilton	  syringe	  and	  LNP-­‐siRNAs	  were	  injected	  using	  an	   infusion	   pump	   (Harvard	   Apparatus)	   at	   a	   rate	   of	   50nL/min.	   	   The	   total	   volume	  injected	   was	   500nL	   of	   LNP	   siRNA	   (5mg	   siRNA/mL	   in	   sterile	   PBS).	   	   For	  intracerebroventricular	  (ICV)	  injections	  holes	  were	  drilled	  	  -­‐0.8mm	  AP	  and	  ±1.4mm	  ML	  from	  bregma	  and	  -­‐3.1mm	  DV	  and	  microdialysis	  silicon	  tubing	  was	  used	  to	  inject	  LNPs.	  	  The	  total	  volume	  injected	  was	  2	  μL	  	  bilaterally	  at	  a	  rate	  of	  200nL/min.	  	  2.2.5	  Brain	  Slice	  Preparation	  	   Five	  days	  after	  LNP	  siRNAs	  (GluN1,	  PTEN,	  Luciferase)	  were	  injected	  into	  the	  cortex,	   Sprague-­‐Dawley	   rats	   (postnatal	   day	   26-­‐30)	   were	   anaesthetized	   with	  halothane	   and	   decapitated	   according	   to	   protocols	   approved	   by	   the	   University	   of	  British	   Columbia	   Committee	   on	   Animal	   Care.	   	   Brains	   were	   rapidly	   extracted	   and	  placed	   into	   ice-­‐cold	   slicing	   solution	   containing	   (in	   mM):	   N-­‐methyl-­‐D-­‐glucamine	  (NMDG),	  120;	  KCl,	  2.5;	  NaHCO3,	  25;	  CaCl2,	  1;	  MgCl2,	  7;	  NaH2PO4,	  1.25;	  glucose,	  20;	  Na-­‐pyruvate,	  2.4;	  Na-­‐ascorbate,	  1.3;	  saturated	  with	  95%	  O2/5%	  CO2.	  Coronal	  hemi-­‐sections,	   300µm	   thick,	   were	   sliced	   using	   a	   vibrating	   tissue	   slicer	   (VT1200,	   Leica,	  Nussloch,	   Germany).	   	   For	   Western	   blot	   preparation	   tissue	   within	   1mm	   but	   not	  including	   the	   injection	   sites	   (needle	   tracts)	  was	   collected.	   	   	   For	   electrophysiology	  experiments	  slices	  were	   incubated	  at	  32°C	   in	  artificial	  cerebral	  spinal	   fluid	  (ACSF)	  containing	  (in	  mM):	  NaCl,	  126;	  KCl,	  2.5;	  NaHCO3,	  26;	  CaCl2,	  2.0;	  MgCl2,	  1.5;	  NaH2PO4,	  1.25;	  glucose,	  10;	  saturated	  with	  95%	  O2/5%	  CO2	  for	  45min.	  	  For	  experiments	  slices	  were	  at	  22–24°C	  and	  perfused	  at	  ~2ml/min.	  	  	  	  	   57	  2.2.6	  Electrophysiology	  	  Whole-­‐cell	   patch	   clamp	   recordings	   were	   made	   using	   electrodes	   (4–6	   MΩ	  resistance)	   filled	  with	   a	  pipette	   solution	   containing	   (in	  mM):	  Cs-­‐methanesulfonate	  ,108;	  Na-­‐Gluconate,	  8;	  Cs-­‐EGTA,	  1;	  TEA-­‐Cl,	  8;	  MgCl2,	  2;	  HEPES,	  10;	  K2ATP,	  4;	  Na3GTP,	  3;	  at	  pH	  7.2.	  Whole-­‐cell	  voltage-­‐clamp	  recordings	  were	  obtained	  from	  neurons	  of	  the	  somatosensory	   cortex	   (layer	   5)	   under	   microscope	   guidance	   using	   differential	  interference	  contrast.	  	  All	  recordings	  were	  filtered	  at	  2	  kHz,	  digitized	  at	  10	  kHz	  and	  acquired	  with	  Clampex	  (Axon	  Instruments,	  Foser	  City,	  CA).	  Membrane	  potential	  was	  clamped	   at	   −70mV.	   A	   monopolar	   stimulation	   electrode	   was	   positioned	   ~100µm	  from	   the	   soma	   of	   the	   recorded	   neuron.	   	   The	   extracellular	   solutions	   were	  supplemented	  with	  50µM	  picrotoxin	   (Sigma)	   and	  8mM	  of	   each	  MgCl2	  and	  CaCl2	  to	  block	   GABAA	   synaptic	   potentials,	   block	   epileptiform	   activity	   and	   minimize	  polysynaptic	  responses.	  Synaptic	  responses	  were	  evoked	  with	  monophasic	  voltage	  pulses	  every	  10	  s.	   	  Cells	  were	  allowed	  to	  dialyze	  for	  at	  least	  15	  min	  before	  starting	  recordings.	  Access	  resistance	  was	  continuously	  monitored	  during	  the	  experiments.	  	  The	   AMPAR	   excitatory	   postsynaptic	   current	   (EPSC)	   was	   recorded	   at	   Vhold=-­‐70mV	  followed	  by	  the	  NMDAR	  +	  AMPAR	  EPSC	  at	  Vhold=+40mV.	  	  Vhold	  was	  then	  returned	  to	  -­‐70mV	  at	   the	  end	  of	   the	  experiment	   to	  verify	   there	  was	  no	   change	   in	   the	  baseline.	  	  20-­‐50	   traces	   were	   averaged	   per	   recording.	   	   25µM	   D-­‐APV	   (Abcam	   Biochemichals,	  Cambridge,	  MA)	  was	  applied	  to	  some	  experiments	  to	  illustrate	  outward	  NMDAR	  and	  AMPA	  currents.	   Signals	  were	  amplified	  with	   the	  Multiclamp	  700B	  amplifier	   (Axon	  Instruments,	  Foser	  City,	  CA),	  low-­‐pass	  filtered	  at	  2kHz,	  and	  digitized	  at	  10kHz	  using	  the	  Digidata	  1322	   (Axon	   Instruments).	   	  Data	  were	   collected	   (pClamp,	   version	  9.2;	  	   58	  Axon	   Instruments)	   and	   stored	   on	   computer	   for	   offline	   analysis	   using	   clampfit	  software	  (Axon	  Instruments).	  	  	  	  2.2.7	  Imaging	  	   Live	   cell	   imaging	   (brain	   slice)	   was	   performed	   with	   a	   two-­‐photon	   laser-­‐scanning	  microscope	  (Zeiss	  LSM510-­‐Axioskop-­‐2;	  Zeiss,	  Oberkochen,	  Germany)	  with	  a	   40X-­‐W/1.0	   numerical	   aperture	   objective	   lens	   directly	   coupled	   to	   a	   Chameleon	  ultra2	  laser	  (Coherent,	  Santa	  Clara,	  CA).	  	  DiI	  and	  CoroNa	  were	  excited	  at	  760nm	  and	  the	  fluorescence	  from	  each	  fluorophore	  was	  split	  using	  a	  dichroic	  mirror	  at	  560nm,	  and	   the	   signals	  were	   each	  detected	  with	   a	  dedicated	  photo	  multiplier	   tube	   (PMT)	  after	  passing	  through	  an	  appropriate	  emission	  filter	  (DiI:	  605nm,	  55nm	  band-­‐pass;	  CoroNa:	  525nm,	  50nm	  band-­‐pass)	  Transmitted	   light	  was	   simultaneously	   collected	  using	  understage	   IR-­‐DIC	  optics	  and	  an	  additional	  PMT.	   	  For	  AM-­‐dye	   loading	  slices	  were	   incubated	   at	   32°C	   for	   45	  min	   at	   16.7μg/mL.	   	   Cell	   density	   was	  measured	   in	  10μm	  z-­‐stacks	  of	  200μm	  x	  200μm,	  and	  all	  cells	  including	  partial	  cells	  were	  counted,	  therefore	  resulting	  in	  an	  overestimate	  of	  cell	  density/mm3.	  	  2.2.8	  Immunohistochemistry	  	   Free-­‐floating	   sections	   (40µm	   transverse	   sections)	   were	   processed	   for	  immunostaining	   as	   described	   previously	   (Choi	   et	   al.,	   2012b).	   	   The	   primary	  antibodies	  used	  for	  immunostaining	  were	  as	  follows:	  Rabbit	  anti-­‐PTEN	  (Santa	  Cruz,	  1:300),	   mouse	   anti-­‐microtubule	   associated	   protein-­‐2	   (MAP-­‐2,	   Chemicon,	   1:2000).	  	  	   59	  Alexa	  Fluor	  633	  anti-­‐mouse	  or	  Alexa	  Fluor	  488	  anti-­‐rabbit	  IgG	  (1:1000)	  secondary	  antibodies	  (Invitrogen,	  Carlsbad,	  CA)	  were	  used	  for	  immunofluorescent	  staining.	  	  As	  a	   negative	   control	   experiment,	   the	   primary	   antibody	   was	   omitted	   during	   the	  immunostaining.	  	  	  	  2.2.9	  Immunocytochemistry	  	   Rat	   hippocampal	   neurons	   (DIV	   12-­‐14	   days)	   grown	   on	   poly-­‐L-­‐lysine-­‐coated	  glass	  coverslips	  were	   incubated	   in	  neurobasal	  media	  with	  LNP	  siRNAs	  (Luciferase	  siRNA	   or	   GluN1	   siRNA)	   for	   48	   hrs	   and	   then	   processed	   with	   immunostaining.	   	   In	  brief,	  cells	  were	  fixed	  in	  2%	  paraformaldehyde	  in	  0.1M	  PBS	  for	  10	  min,	  washed	  with	  PBS,	  and	  then	  permeabilized	  in	  0.05%	  Tween20	  in	  0.1M	  PBS	  for	  20	  min.	  	  Cells	  were	  then	   incubated	   in	   mouse	   anti-­‐GluN1	   (1:300	   dilution;	   Invitrogen)	   	   or	   rabbit	   anti-­‐MAP-­‐2	  at	  4°C	  for	  24	  hrs	  and	  followed	  by	  Alexa	  Fluor	  488	  anti-­‐mouse	  IgG	  secondary	  antibody	  (1:1000;	  Molecular	  Probes,	  Eugene,	  OR)	  or	  Alexa	  Fluor	  633	  anti-­‐rabbit	  IgG	  (1:1000)	   incubation	  at	   room	   temperature	   for	  1	  hr	   in	   the	  dark.	   	  After	  PBS	  washes,	  coverslips	  were	  then	   immersed	   in	  4′,6-­‐diamidino-­‐2-­‐phenylindole	  (DAPI;	  Molecular	  Probes)	   at	  1μg/ml	   in	  water	   to	  visualize	   cell	  nuclei.	   	  The	   coverslips	  were	  mounted	  onto	  glass	   slides	  using	  FluoroSave	   (Calbiochem)	  and	  examined	  under	  an	  Olympus	  confocal	  microscope	  (FluoView	  1000,	  Olympus,	  Center	  Valley,	  PA).	  	   60	  2.2.10	  Western	  Blotting	  	   Cultured	  rat	  hippocampal	  neurons	  and	  rat	  cortical	  brain	  slices	  were	  used	  for	  Western	   blotting.	   	   At	   5	   days	   post-­‐injection	   with	   LNP-­‐siRNAs	   (GluN1,	   PTEN,	  Luciferase)	   into	   the	   cortex,	   cortical	   brain	   slices	   were	   prepared	   as	   previously	  described	  (Choi	  et	  al.,	  2012b).	  	  Cells	  and	  brain	  slices	  were	  homogenized	  using	  lysis	  buffer	  containing	  (in	  mM):	  Tris	  pH	  7.0	  (100),	  EGTA	  (2),	  EDTA	  (5),	  NaF	  (30),	  sodium	  pyrophosphate	   (20),	   0.5%	  NP40	  with	  phosphatase	   and	  protease	   inhibitor	   cocktail	  (Roche,	  Basel,	  Switzerland).	  	  The	  homogenates	  were	  then	  centrifuged	  at	  13,000	  ×	  g	  (20	  min,	   4oC)	   to	   remove	   cellular	   debris,	   then	   protein	   concentrations	   of	   the	   crude	  lysates	  were	  determined	  by	  performing	  a	  Bradford	  assay	  with	  the	  DC	  Protein	  Assay	  dye	  (Bio-­‐Rad,	  Mississauga,	  ON,	  Canada).	  	  The	  protein	  samples	  were	  diluted	  with	  2	  X	  Laemmli	   sample	   buffer	   and	   boiled	   for	   5	  minutes.	   	   Following	   SDS/PAGE,	   proteins	  were	  transferred	  to	  PVDF	  membranes,	  blocked	  in	  5%	  milk	  overnight	  at	  4oC,	  rinsed	  with	   Tris	   buffered	   saline	  with	   0.1%	  Tween	   20	   (TBST)	   and	   incubated	  with	  mouse	  anti-­‐GluN1	  monoclonal	   antibody	   (1:300)	   or	   rabbit	   anti-­‐PTEN	   polyclonal	   antibody	  (1:300)	  overnight	  at	  4oC.	   	  Following	   four	  washes	  with	  TBST,	   the	  membranes	  were	  incubated	   with	   the	   anti-­‐mouse	   or	   anti-­‐rabbit	   secondary	   antibody	   conjugated	   to	  horseradish	   peroxidase	   (1:500)	   for	   1	   hr	   at	   room	   temperature.	   	   The	   membranes	  were	  then	  washed	  3-­‐4	  times	  (15	  min)	  with	  TBST,	  and	  bands	  were	  visualized	  using	  enhanced	  chemiluminescence	  	  (GE	  HealthCare,	  Cleveland,	  OH).	  	  	  Tissue	   analyzed	   from	   ICV	   injections	  was	   sampled	   as	   follows;	   entire	   dorsal	  hippocampal	   slices	   were	   dissected	   from	   -­‐2.0mm	   to	   -­‐4.2mm	   AP	   from	   bregma;	  	  	   61	  striatal	   slices	  were	   taken	  +1.0mm	  to	   -­‐0.5mm	  AP	   from	  bregma	  within	  1.5mm	  from	  the	  ventricle	  border.	  	  2.2.11	  Lactate	  dehydrogenase	  (LDH)	  Assay	  	   LDH	  assay	  kits	  (Biomedical	  Research	  Service	  Center,	  State	  Univerisity	  of	  New	  York	   at	   Buffalo)	   were	   used	   to	   examine	   cell	   death	   using	   cultures	   of	   hippocampal	  neurons	   with	   astrocytes	   on	   a	   separate	   feeder	   layer.	   Cells	   were	   treated	   with	   LNP	  siRNA	  (Luciferase	  siRNA)	   for	  72	  hrs	   then	  assessed	   for	  cell	  death	  using	  LDH	  assay.	  	  Media	   containing	  Triton	  X-­‐100	   (1%)	  was	  used	  as	  a	  positive	   control	   for	   cell	  death.	  	  Supernatants	   were	   collected	   at	   72	   hours	   after	   the	   treatment	   then	   cells	   on	   the	  coverslips	   were	   lysed	   using	   lysis	   buffer.	   	   The	   LDH	   level	   in	   the	   supernatant	  represents	   the	   cell	   death	  while	   the	   LDH	   level	   in	   lysed	   cells	   represents	   the	   viable	  cells.	   	  In	  brief,	  supernatants	  and	  cell	  lysates	  were	  centrifuged	  for	  3	  min	  at	  maximal	  speed	  (16,000g)	  at	  4oC.	  	  All	  samples	  were	  added	  into	  a	  96-­‐well	  plate	  with	  LDH	  assay	  solution	  and	  incubated	  for	  30	  min	  at	  37oC.	  	  The	  reaction	  was	  stopped	  with	  3%	  acetic	  acid.	   	   LDH	   reduces	   tetrazolium	   salt	   INT	   to	   formazan,	   which	   is	   water-­‐soluble	   and	  exhibits	  an	  absorption	  maximum	  at	  492nm.	   	  Absorbance	  at	  492nm	  was	  measured	  using	  a	  microplate	  reader.	  	  Cell	  death	  was	  calculated	  as	  percentage	  of	  released	  LDH	  compared	  to	  the	  sum	  of	  superfusate	  LDH	  and	  cell	  lysate	  LDH.	  	  	  	  	   62	  2.2.12	  DiI	  Uptake	  Assay	  	   Cortical	  neuronal	   cultures	   (3	   x	  104	   cell	   in	  12	  well	   plate)	  were	   treated	  with	  Luc	   siRNA-­‐LNP	   (at	   1.6mg/ml)	   conjugated	   with	   DiI	   for	   2	   hr	   in	   the	   absence	   and	  presence	  of	   recombinant	   apolipoprotein	  E	   (ApoE)	  4	   (Peprotech,	  Rocky	  Hill,	  NJ)	   at	  different	  concentrations	  (0.1,	  1,	  5,	  10	  mg/ml).	  Then	  cultures	  were	  washed	  with	  PBS	  5	  times	  to	  wash	  out	  unbound	  Luc	  siRNA-­‐LNP.	  	  Subsequently,	  400	  μl	  of	  filtered	  dH20	  was	  added	  to	  the	  wells	  to	  rupture	  cells.	  DiI	  fluorescence	  was	  measured	  (excitation	  at	  520nm,	  emission	  at	  578nm)	  using	  Gemini	   fluorescence	  microplate	   reader	  systems	  (Molecular	   Devices	   Corporation,	   Union	   city,	   CA).	   	   Fluorescence	   reading	   from	   the	  untreated	  group	  was	  subtracted	  from	  other	  groups.	  	  	  2.2.13	  Tumor	  necrosis	  factor-­‐α	  (TNF-­‐α)	  ELISA	  	   Enzyme-­‐linked	  immunosorbent	  assays	  (ELISA)	  were	  performed	  according	  to	  the	  manufacturer	   instructions	   (eBioscience,	   San	   Diego,	   CA).	   In	   brief,	   hippocampal	  brain	  slices	  (400µm)	  were	  incubated	  and	  treated	  in	  a	  homemade	  chamber	  (using	  6	  multi-­‐well	  plates)	  equipped	  with	  continuous	  aeration	  with	  95%	  O2/5%	  CO2.	  Slices	  were	  treated	  with	  Luc	  siRNA-­‐LNPs	  (3.3µg/ml)	   for	  5	  hrs.	  As	  a	  positive	  control,	  LPS	  (40µg/ml)	   was	   applied	   to	   the	   slices.	   Then	   the	   hippocampal	   brain	   slices	   were	  harvested	  and	  homogenized.	  Cell	  lysates	  were	  centrifuged	  for	  20	  minutes	  at	  4˚C	  and	  subsequent	  supernatants	  were	  used	   for	  protein	  assay	  and	  tumor	  necrosis	   factor-­‐α	  (TNF-­‐α)	  measurement.	  	  	  	   63	  2.2.14	  siRNA	  Sequences	  and	  Chemistry	  	   PTEN	   siRNA:	   The	   PTEN	   siRNA	   strands	   had	   the	   following	   sequence	   and	  chemistry:	  sense	  strand,	  5’-­‐GAuGAuGuuuGAAAcuAuudTsdT-­‐3’	  and	  antisense	  strand	  5’-­‐AAuAGUUUcAAAcAUcAUCdTsdT-­‐3’	   where	   the	   uppercase	   letters	   represent	  unmodified	   ribonucleotides,	   lowercase	   letters	   represent	   2’-­‐OMe	   modified	  nucleotides	  and	   “s”	   represents	  a	  phosphorothioate	   (P=S)	   linkage	  between	   the	   two	  dT	  at	  the	  3’-­‐end.	  The	  P=S	  linkage	  provided	  protection	  from	  exonuclease	  degradation	  and	  the	  2’-­‐OMe	  modifications	  reduced	  potential	  for	  immunostimulatory	  activity	  and	  provided	  stability	  towards	  endonucleolytic	  degradation.	  	  	  Luciferase	  siRNA	  had	  the	  following	  sequence	  as	  previously	  described	  (Addepalli	  et	  al.,	  2010):	  sense	  strand,	  5’-­‐	  cuuAcGcuGAGuAcuucGAdTsdT-­‐3’	  and	  antisense	  strand,	  5’-­‐UCGAAGuACUcAGCGuAAGdTsdT-­‐3’	   with	   the	   modification	   designation	   as	   listed	  above.	  	  	  GRIN1	   siRNA:	   	   The	   sequences	   of	   “stealth	   siRNA”	   (Invitrogen	   Life	   Technologies,	  Gaithesburg,	  MD)	  had	  the	  following	  sequences:	  	  GRIN1-­‐#1,	  sense	  strand,	  5’-­‐UGCAUGUCCCAUCACUCAUUGUGGG-­‐3’	  	  and	  antisense	  strand,	  5’-­‐CCCACAAUGAGUGAUGGGACAUGCA-­‐3’;	  	  GRIN1-­‐#2,	  sense	  strand,	  5’-­‐CUUCUGUGAAGCCUCAAACUCCAGC-­‐3’	  	  and	  antisense	  strand,	  5’-­‐GCUGGAGUUUGAGGCUUCACAGAAG-­‐3’;	  	  GRIN1-­‐#3,	  sense	  strand,	  5’-­‐UUGACGUACACGAAGGGCUCUUGGU-­‐3’	  	   64	  	  and	  antisense	  strand,	  5’-­‐ACCAAGAGCCCUUCGUGUACGUCAA-­‐3’.	  The	  reason	  for	  variations	  in	  the	  siRNA	  constructs	  is	  due	  to	  production	  from	  different	  suppliers.	  	  2.2.15	  Statistical	  Analysis	  	   Experimental	  values	  are	  presented	  as	  the	  mean	  ±	  SEM,	  expressed	  in	  percent	  from	   100%	   baseline.	   The	   “n”	   value	   represents	   the	   number	   of	   experiments	  conducted	   for	   analysis.	   Statistical	   analyses	   were	   performed	   using	   a	   two-­‐tailed	  Student’s	   t	   test	   or	   analysis	   of	   varience	   (ANOVA)	   followed	   by	   Newman-­‐Keuls	  Multiple	   Comparison	   Test.	   	   p	  <	   0.05	  was	   accepted	   as	   statistically	   significant	   (*p	  <	  0.05,	  **p	  <	  0.01,	  ***p<0.001).	  	  2.3	  Results	  	  2.3.1	  Encapsulation	  of	  siRNA	  in	  Lipid	  Nanoparticles	  	   LNPs	   were	   prepared	   by	   mixing	   appropriate	   volumes	   of	   lipid	   mixture	   in	  ethanol	  with	  an	  aqueous	  phase	  containing	  siRNA	  duplexes	  employing	  a	  microfluidic	  micro-­‐mixer	  (Figure	  2.1a)	  as	  described	  elsewhere	  (Belliveau	  et	  al.,	  2012).	  	  The	  lipid	  composition	   used	   was	   DMAP-­‐BLP	   (structure	   shown	   in	   Figure	   2.1b)	  /distearoylphosphatidlycholine	  (DSPC)/cholesterol/PEG-­‐DMG	  in	  the	  molar	  %	  ratios	  50/10/37.5/1.5.	  The	  LNPs	  also	  contained	  1	  mol%	  of	  the	  fluorescently	  labeled	  lipids	  DiOC18	  or	  DiIC18	  to	  monitor	  LNP	  uptake.	  Previous	  work	  has	  shown	  that	  LNP	  systems	  	   65	  with	   this	   lipid	   composition	   can	   silence	   target	   genes	   in	   hepatocytes	   following	  intravenous	   injection	   at	   dose	   levels	   as	   low	   as	   0.01mg	   siRNA/kg	   body	   weight	  (Belliveau	  et	  al.,	  2012;	  Jayaraman	  et	  al.,	  2012)	  in	  an	  ApoE	  dependent	  fashion	  (Akinc	  et	  al.,	  2010).	  The	  LNP	  PTEN-­‐siRNA	  systems	  produced	  had	  a	  diameter	  55±11nm	  (size	  distribution	  shown	   in	  Figure	   2.1c).	  The	  sizes	  of	   the	  other	  LNP	  siRNA	  systems	  are	  listed	  in	  Table	  2.1.	  	  	  No	  further	  optimization	  was	  done	  to	  increase	  the	  efficiency	  of	  LNP-­‐siRNA	  systems	  in	  this	  study.	  	  	  	  	  	   66	  	  	  Figure	   2.1:	   Schematic	   of	   LNP-­‐siRNA	   formulation	   process	   employing	   the	  staggered	  herringbone	  micromixer.	  	  	  	  a,	   The	   lipid	   mixture	   in	   ethanol	   and	   siRNA	   in	   aqueous	   solution	   are	   pumped	  separately	  into	  the	  two	  inlets	  of	  the	  microfluidic	  mixing	  device	  using	  a	  syringe	  pump	  with	  a	  total	  flow	  rate	  of	  2mL/min.	  Herringbone	  structures	  induce	  chaotic	  advection	  of	  the	  laminar	  streams	  causing	  rapid	  mixing	  of	  the	  ethanol	  and	  aqueous	  phases	  and	  correspondingly	  rapid	  increases	  in	  the	  polarity	  experienced	  by	  the	  lipid	  solution.	  At	  a	  critical	  polarity	  precipitates	  form	  as	  LNPs.	  Dimensions	  of	  the	  mixing	  channel	  were	  200μm	  x	  79μm,	   and	   the	  herringbone	   structures	  were	  31μm	  high	   and	  50μm	   thick.	  (modified	   from	  (Belliveau	  et	  al.,	  2012))	  b,	  Chemical	   structure	  of	   ionizable	  cationic	  lipid	   –	   3-­‐(dimethylamino)propyl	   (12Z,15Z)-­‐3-­‐[(9Z,12Z)-­‐octadeca-­‐9,12-­‐dien-­‐1-­‐yl]henicosa-­‐12,15-­‐dienoate	  (DMAP-­‐BLP).	  c,	  Representative	  size	  distribution	  of	  LNPs	  (LNP	   PTEN-­‐siRNA)	   analyzed	   in	   number	   mode	   using	   the	   NICOMP	   370	   Submicron	  Particle	  Sizer.	  	  	   67	  2.3.2	  Lipid	  Nanoparticle	  Mediated	  Neuronal	  Gene	  Silencing	  in	  vitro	  	   To	   test	   whether	   LNPs	   are	   taken	   up	   by	   neurons,	   cultured	   neurons	   were	  incubated	   with	   siRNA-­‐containing	   LNPs	   at	   a	   final	   concentration	   of	   246nM	   (3.3µg	  siRNA/mL).	   	   For	   LNPs	   of	   55	   nm	   diameter	   at	   an	   siRNA-­‐to-­‐lipid	   ratio	   of	   0.056	  (mg/µmol)	   this	   represents	   approximately	   5.6x1011	   LNPs/mL.	   	  We	   used	   a	   culture	  system	   containing	   pure	   hippocampal	   neurons	   on	   a	   coverslip	  with	   astrocytes	   on	   a	  separate	  feeder	  layer	  (Craig	  et	  al.,	  1996)	  (Figure	  2.2a).	  	  Surprisingly,	  we	  found	  that	  100%	  of	  neurons	  in	  the	  culture	  dish	  had	  taken	  up	  LNPs	  following	  incubation	  for	  24	  hours,	   as	   indicated	   by	   discrete	   punctate	   DiI	   fluorescence	  within	   neurons	   (Figure	  2.2b).	  	  DiI	  labeled	  intracellular	  puncta	  were	  observed	  in	  both	  live	  cultured	  neurons	  and	   in	   neurons	   after	   cultures	   were	   fixed.	   	   The	   punctate	   staining	   pattern	   of	   DiI	  indicates	  LNPs	  located	  in	  endosomes	  and	  lysosomes	  (Basha	  et	  al.,	  2011).	  	  LNPs	  were	  non-­‐toxic	   at	   the	   concentrations	   used,	  with	   no	   observable	   differences	   in	   cell	   death	  relative	  to	  controls	  measured	  either	  by	  comparing	  nuclear	  morphology	  or	  by	  lactate	  dehydrogenase	  (LDH)	  release	  (Figure	  2.2c).	   	  As	  a	  positive	  control	  cultures	  treated	  with	  1%	  Triton	  X-­‐100	  showed	  a	  large	  increase	  in	  LDH	  release	  (90.21±0.62%).	  	  	  	  We	   next	   tested	   whether	   the	   LNP	   mediated	   delivery	   of	   siRNA	   to	   neurons	  effectively	  silenced	  gene	  expression	  by	  using	  Western	  blots	   to	  examine	  changes	   in	  protein	  expression	  encoded	  by	  the	  corresponding	  target	  gene.	  	  We	  first	  tested	  LNP	  delivery	   of	   siRNA	   against	   PTEN,	   a	   protein	   highly	   expressed	   in	   pyramidal	   neurons	  (Kwon	   et	   al.,	   2006).	   Incubation	   of	   primary	   rat	   neuronal	   cultures	   with	   LNPs	  containing	  PTEN	  siRNA	  (246nM)	  for	  48	  hrs	  resulted	  in	  robust	  knockdown	  of	  PTEN	  	   68	  protein	   indicated	   by	   Western	   blot	   (PTEN/β-­‐actin	   reduced	   by	   80%	   compared	   to	  control,	  P<0.001,	  Figure	  2.2d).	  	  	  Importantly,	  in	  control	  experiments,	  incubation	  of	  cultures	  with	  LNPs	  containing	  siRNA	  against	   luciferase	  (luc),	  which	  is	  not	  found	  in	  the	   mammalian	   genome,	   had	   no	   effect	   on	   levels	   of	   PTEN	   protein	   (Figure	   2.2d).	  	  Additionally,	  neurons	  treated	  with	  non-­‐encapsulated	  PTEN	  siRNA	  (246nM)	  showed	  no	   significant	   change	   in	   PTEN/β-­‐actin	   compared	   to	   control	   (Figure	   2.2e).	   To	  determine	  the	  efficiency	  of	  these	  LNP-­‐siRNA	  systems	  we	  performed	  a	  dose	  response	  curve	  and	  found	  that	  even	  concentrations	  as	  low	  as	  0.7nM	  resulted	  in	  robust	  protein	  knockdown	  (PTEN/β-­‐actin	   reduced	  by	  59%	  compared	   to	   control,	  P<0.001,	  Figure	  2.2e).	  	  These	  results	  indicate	  that	  LNPs	  are	  much	  more	  efficient	  and	  less	  toxic	  than	  current	  methods	  used	  to	  deliver	  siRNA	  to	  cultured	  neurons	  such	  as	  electroporation,	  calcium	  phosphate	  or	  lipofection	  which	  usually	  result	  in	  transfection	  rates	  of	  1-­‐10%	  (Karra	   and	   Dahm,	   2010).	   	   Higher	   transfection	   rates	   have	   been	   reported	   upon	  optimization,	  but	  generally	  at	  the	  expense	  of	  cell	  toxicity	  (Karra	  and	  Dahm,	  2010).	  	   69	  	  Figure	  2.2:	   	  LNP-­‐siRNA	  systems	  mediate	  knockdown	  of	  target	  gene	  in	  neuron	  cultures	  	  	  a,	   Pure	   neuron	   cultures	   on	   a	   coverslip	   in	   a	   petri	   dish	   with	   a	   separate	   astrocyte	  feeder	  layer.	  	  LNP-­‐siRNA	  was	  added	  directly	  to	  the	  media.	  	  b,	  	  DiI	  fluorescence	  (red)	  shows	   that	  LNPs	  are	   found	   in	   	   the	  cytoplasm	  of	  neurons.	   	  Bottom:	  Cross-­‐sectional	  analysis	  of	  an	  image	  stack	  of	  fluorescence	  and	  transmitted	  IR	  images	  revealed	  that	  that	   DiI	   puncta	   were	   found	   within	   the	   boundaries	   of	   the	   cell	   membrane.	   	   scale:	  15μm	   c,	   Quantification	   of	   LDH	   release	   revealed	   that	   LNP	   were	   not	   toxic	   at	  concentrations	   used.	   d,	   Western	   blots	   reveal	   LNP-­‐PTEN	   siRNA	   resulted	   in	  knockdown	  of	  PTEN	  protein	   compared	   to	  Luc	   siRNA-­‐LNP	  control	   and	  non-­‐treated	  cultures.	   e,	   Dose	   response	   of	   LNP-­‐siRNA	   concentration	   versus	   PTEN	   knockdown,	  last	   column	   shows	   that	   non	   encapsulated	   siRNA	   did	   not	   result	   in	   protein	  knockdown.	  In	  all	  figures,	  experimental	  values	  are	  the	  mean	  and	  SEM.	  LDH,	  lactate	  dehydrogenase;	  LNP,	  lipid	  nanoparticle;	  ns,	  not	  significant;	  siRNA,	  small	  interfering	  RNA.	  	  	   	  	   70	  2.3.3	  Uptake	  of	  Lipid	  Nanoparticles	  by	  Neurons	  is	  Apolipoprotein	  E	  (ApoE)-­‐Dependent.	  	   In	  hepatocytes	  in	  vivo,	  LNP	  uptake	  is	  facilitated	  by	  adsorption	  of	  ApoE	  to	  the	  LNPs	  (Akinc	  et	  al.,	  2010),	  which	  can	  then	  be	  recognized	  by	  scavenging	  receptors	  and	  low	  density	  lipoprotein	  receptors	  (LDLR)	  on	  the	  hepatocyte	  surface.	  	  Because	  ApoE	  is	   the	   dominant	   lipoprotein	   in	   the	   brain	   we	   tested	   whether	   uptake	   of	   LNPs	   by	  neurons	   in	   cell	   culture	   was	   also	   ApoE	   dependent.	   	   ApoE	   is	   produced	   mainly	   by	  astrocytes	  and	  delivers	  cholesterol	  and	  other	  essential	  lipids	  to	  neurons	  via	  binding	  to	   members	   of	   the	   LDLR	   family	   followed	   by	   endocytosis	   (Pitas	   et	   al.,	   1987;	   Bu,	  2009).	  	  	  To	  directly	  test	  the	  ApoE	  dependence	  we	  transferred	  the	  neurons	  to	  media	  that	   had	   not	   been	   in	   contact	   with	   astrocytes	   and	   applied	   LNPs	   (246nM)	   with	   or	  without	  exogenous	  ApoE4	  (10µg/mL)	  (Figure	  2.3a).	  	  	  After	  1	  hour,	  we	  noticed	  rapid	  uptake	  of	  DiI	  labeled	  LNPs	  in	  the	  presence	  of	  exogenous	  ApoE,	  compared	  to	  neurons	  incubated	   without	   ApoE4	   (Figure	   2.3a,	   b).	   	   Furthermore,	   we	   tested	   the	   dose	  dependence	  of	  ApoE	  vs.	   LNP	  uptake	  by	  measuring	   the	   total	  DiI	   fluorescence	   from	  ruptured	  cells	  1	  hour	  after	  treatment.	  	  We	  found	  that	  LNP	  uptake	  reached	  saturation	  at	   around	   5µg	   ApoE/mL	   (Figure	   2.3c),	   in	   line	   with	   measurements	   from	   human	  cerebrospinal	  fluid	  (CSF)	   in	  vivo	  of	  9.09µg/mL	  (Wahrle	  et	  al.,	  2007).	   	  These	  results	  suggest	   that	   the	   efficient	  LNP	  uptake	  by	  neurons	   is	   facilitated	  by	   association	  with	  ApoE	  and	  subsequent	  endocytosis	  into	  neurons	  via	  an	  ApoE	  receptor.	  	   71	  	  Figure	  2.3:	  LNPs	  are	  taken	  up	  by	  neurons	  in	  an	  ApoE	  dependent	  manner.	  	  a,b,	   	  In	  the	  absence	  of	  astrocytes,	  addition	  of	  ApoE	  facilitated	  the	  uptake	  of	  LNPs	  by	  neurons	  shown	  by	  an	  increase	  in	  DiI	  fluorescence.	  	  Treatment:	  1	  hour	  LNP-­‐siRNA	  ±	  ApoE.	  	   	  c,	  Dose	  dependence	  of	  LNP	  uptake	  vs.	  ApoE	  concentration	  measured	  as	  DiI	  fluorescence.	  In	  all	  figures,	  experimental	  values	  are	  the	  mean	  and	  SEM.	  ***P<0.001;	  DAPI,	   4′,6-­‐diamidino-­‐2-­‐phenylindole;	   ns,	   not	   significant;	   siRNA,	   small	   interfering	  RNA.	  	  DAPI DiI (LNP) merge+APOE-APOEb100 μma+ ApoE+ 1hrNeurons on cover slip - ApoE+ 1hr0 0.1 1 5 100100200300400500APOE concentration (µg/ml)Fluorescence (a.u.)nsns******c	   72	  2.3.4	  Lipid	  Nanoparticle	  Mediated	  Neuronal	  Gene	  Silencing	  in	  vivo	  	  	  The	  gene	  silencing	  potency	  of	  LNP	  siRNA	  formulations	  was	  tested	  in	  vivo	  by	  direct	   injection	   into	   the	   cortex.	   	   In	   these	   experiments	   a	   single	   injection	   of	   LNPs	  (500nL	   at	   5mg/mL	   siRNA/10mins)	   was	   administered	   directly	   into	   the	  somatosensory	  cortex	  (Figure	  2.4a)	  and	  the	  distribution	  of	  DiI	  labeled	  LNPs	  and	  the	  impact	   on	   gene	   expression	  monitored	   subsequently.	   	   These	   LNPs	   had	   an	   average	  diameter	   of	   50-­‐60nm	   (Figure	   2.1c),	   small	   enough	   to	   diffuse	   through	   the	  extracellular	  space	  of	  the	  brain	  (Thorne	  and	  Nicholson,	  2006).	  	  To	  first	  test	  whether	  neurons	   in	   vivo	   accumulated	   LNPs,	   acute	   cortical	   slices	   were	  made	   from	   injected	  rats	  5	  days	  following	  injection,	  and	  monitored	  for	  DiI	  positive	  neurons.	   	  Astrocytes	  in	   vivo	   produce	   and	   secrete	   ApoE	   (Bu,	   2009);	   therefore	   no	   additional	   ApoE	   was	  added	   to	   injected	   LNPs.	   	   We	   consistently	   found	   robust	   DiI	   staining	   localized	   to	  neurons	   within	   a	   radius	   of	   ~800µm	   from	   the	   injection	   site	   (Figure	   2.4b).	   	   	   The	  neurons	  were	  visualized	  using	  a	  live	  cell	  fluorescent	  assay	  by	  loading	  the	  cells	  with	  an	  –AM	  form	  of	  a	  fluorescent	  Na+	  dye,	  CoroNa-­‐AM	  that	  is	  only	  retained	  in	  live	  cells	  (Tsien,	  1981).	  Consistent	  with	   the	   in	  vitro	   cell	   culture	   results,	   a	   single	   injection	  of	  LNP	  PTEN	  siRNA	  resulted	  in	  knockdown	  of	  PTEN	  protein	  measured	  using	  Western	  blots	  (PTEN	  to	  β-­‐actin	  ratios	  reduced	  by	  72%	  and	  69%	  compared	  to	  non-­‐injected	  or	  luc	   siRNA	   injected	   controls	   respectively,	   P<0.001,	   Figure	   2.4	   c,d).	   	   Luc	   siRNA	  injected	   tissue	   was	   not	   significantly	   different	   from	   non-­‐injected	   controls	   (Figure	  2.4c,	   d).	   	  Additionally,	   to	   test	   for	   toxicity	  of	  LNPs	   in	  vivo,	  we	   loaded	  cells	  with	   the	  vital	   dye	   calcein-­‐AM,	   which	   is	   only	   fluorescent	   when	   cleaved	   by	   endogenous	  	   73	  esterases	  present	   in	   live	   cells,	   and	   compared	   cell	  density	   in	  LNP	   injected	   rats	   and	  non	   injected	   rats.	   	   In	   Luc	   siRNA-­‐LNP	   injected	   rats	   we	   observed	   no	   significant	  difference	   in	   cell	   density	   200mm-­‐500mm	   from	   the	   injection	   tract	   compared	   to	  control	  (Figure	  4e).	  	  This	  is	  a	  region	  with	  robust	  LNP	  uptake	  and	  gene	  knockdown	  but	  not	  damaged	   from	  the	   injection	  needle	   itself,	   suggesting	   that	  LNPs	   themselves	  are	  non-­‐toxic	   in	  vivo.	   	   	  These	  results	  demonstrate	   that	  when	  administered	  directly	  into	   the	   brain,	   LNPs	   are	   capable	   of	   diffusing	   through	   the	   extracellular	   milieu	   to	  deliver	  siRNA	  to	  neurons	  and	  induce	  gene	  silencing	  at	  sites	  distant	  from	  the	  site	  of	  injection.	  	  	  Although	  clinical	  studies	  suggest	  that	  intravenous	  delivery	  of	  LNP-­‐siRNAs	  are	  rarely	  immunogenic,	  the	  immunostimulatory	  effects	  of	  LNPs	  on	  brain	  tissue	  has	  not	  yet	   been	   examined.	   	   To	   test	  whether	   LNP-­‐siRNAs	   caused	   any	   immunostimulatory	  effects	  in	  the	  brain,	  we	  directly	  incubated	  acute	  cortical/hippocampal	  brain	  slices	  in	  LNP-­‐siRNA	  and	  measured	   immune	  responses	  using	  tumor	  necrosis	   factor-­‐α	  	  (TNF-­‐	  α)	   ELISA.	   TNF-­‐α	   	  is	   a	   pro-­‐inflammatory	   cytokine	   that	   is	   released	   when	   immune	  responses	   occur	   in	   the	   brain.	   The	   results	   suggest	   that	   there	   was	   no	  immunostimulatory	   effect	   when	   Luc	   siRNA-­‐LNPs	   (246nM)	   were	   applied	   to	   brain	  slices.	   TNF-­‐	   α	   levels	   in	   the	   Luc	   siRNA-­‐LNP-­‐treated	   group	   were	   not	   significantly	  different	   from	   the	   control	   group	   (Figure	   4f,	   P>0.05).	   As	   a	   positive	   control,	   LPS	  (40mg/ml)	   was	   applied	   to	   the	   slices	   and	   induced	   a	   significant	   increase	   of	   TNF-­‐a	  compared	   to	  control	  and	  Luc	  siRNA-­‐LNPs	  groups	  (Figure	   4f,	  P<0.05).	   In	  Addition,	  TNF-­‐a	  levels	  were	  not	  significantly	  increased	  in	  siRNA-­‐LNP	  positive	  tissue	  following	  	   74	  direct	   injection	   in	   vivo	   (Figure	   4g,	   P>0.05).	   	   	   These	   data	   suggest	   that	   LNPs	  themselves	  do	  not	  cause	  immunostimulatory	  responses	  in	  the	  brain.	  	   75	  	  	  a bControlPTEN siRNA Luc siRNA0.00.20.40.60.81.013 9 6PTEN/β-actinc dNeurons: live cell dyeLNPs:  DiImergePTENβ-actinPTEN siRNAPTEN siRNALuc siRNALuc siRNAInjectedhemispherenon-injectedhemisphere (control)*** ***DiI labeledLNP siRNA+ 5 daysR at cerebral cortexCorpus callosumNeuronscontrolLNP injected0.00.51.01.52.02.58 1 0nsestimated cell density *105  /mm3eControlsiRNA-LNPs LPS0204060TNF- (pg/mL)*4 55fnscontrol hemispheresiRNA-LNP Injected010203044gbrain slices in vivonsTNF- (pg/mL)Figure 4 Rungta et al., 2013	   76	  Figure	  2.4:	  	  LNP-­‐siRNA	  systems	  mediate	  knockdown	  of	  target	  gene	  in	  vivo.	  	  a,	   	   LNP-­‐siRNA	   was	   injected	   directly	   into	   the	   somatosensory	   cortex	   using	   a	   glass	  micropipette.	  	  b,	   	  Imaging	  of	  acute	  brain	  slices	  5	  days	  following	  a	  single	  injection	  of	  LNP-­‐siRNA	   revealed	   that	   live	   neurons	   (AM-­‐dye)	   had	   taken	   up	   fluorescent	   LNPs	  (DiI).	  Scale:	  10 μm.	  c,	  d,	   	  Western	  blots	  revealed	  that	   injection	  of	  LNP-­‐PTEN	  siRNA	  resulted	   in	  knockdown	  of	  PTEN	  protein	  compared	   to	   tissue	   from	   the	  non-­‐injected	  hemisphere	   and	   Luc	   siRNA-­‐LNP	   injected	   rats.	   Tissue	   was	   dissected	   within	   1mm	  from	  the	  site	  of	  injection	  after	  5	  days.	  	  e,	  Analysis	  of	  the	  density	  of	  live	  cells	  stained	  with	  the	  vital	  dye	  calcein-­‐AM	  showed	  there	  was	  no	  indication	  of	  toxicity	  or	  cell	  loss	  with	  LNP	  uptake	  in	  vivo	  (200-­‐500µm	  from	  injection).	  	  f,	  TNF-­‐α	  ELISA	  measurements	  from	   acute	   brain	   slices	   treated	   with	   Luc	   siRNA-­‐LNPs	   shows	   lack	   of	   an	  immunostimulatory	   response	   to	   LNPs	   in	   brain	   tissue.	   g,	   	   TNF-­‐α	   ELISA	  measurements	  taken	  from	  LNP	  positive	  tissue	  (~100-­‐500um	  from	  injection),	  were	  not	   significantly	   different	   from	   the	   LNP	   negative	   non-­‐injected	   hemisphere.	  Measurements	  were	  normalized	  to	  0.5mg	  of	  protein.	  	  *P	  <	  0.05;	  ***P	  <	  0.001.	  ELISA,	  enzyme-­‐linked	   immunosorbent	   assay;	   LNP,	   lipid	   nanoparticle;	   LPS,	  lipopolysaccharide;	  NS,	  not	  significant;	  siRNA,	  small	  interfering	  RNA;	  TNF-­‐α,	  tumor	  necrosis	  factor-­‐α.	  	  	   77	  2.3.5	  Time	  Course	  and	  Distance	  Analysis	  of	  PTEN	  Knockdown	  in	  vivo	  	   We	   next	   determined	   how	   far	   from	   the	   injection	   site	   was	   gene	   silencing	  effectively	   induced,	   as	   well	   as	   the	   time	   course	   of	   the	   LNP	   mediated	   PTEN	  knockdown.	   	   To	   characterize	   the	   distance	   profile	   of	   the	   knockdown,	   we	   used	   a	  combination	   of	   immunohistochemistry	   as	  well	   as	  Western	   blot	   analysis	   on	   tissue	  taken	  at	  different	  distances	  away	  from	  the	  injection	  tract.	  	  Immunostaining	  of	  fixed	  tissue	   taken	   from	   rats	   5	   days	   following	   a	   single	   injection	   of	   PTEN	   siRNA	   LNPs	  revealed	   clear	   loss	   of	   PTEN	   staining	   in	  neurons	   less	   than	  1mm	   from	   the	   injection	  site,	  compared	  to	  higher	  levels	  of	  PTEN	  staining	  at	  distances	  further	  away	  (Figure	  2.5a).	   The	   montage	   in	   Figure	   2.5a	   shows	   a	   sectioning	   plane	   where	   the	   asterisk	  indicates	   the	   region	   400µm	   from	   the	   injection	   tract.	   The	   pattern	   of	   decreased	  immunofluorescence	   showed	   a	   good	   correlation	   with	   the	   Western	   blot	   analysis,	  which	   revealed	   that	   PTEN	   protein	   levels	   were	   significantly	   reduced	   at	   distances	  within	  1mm	  of	  the	  injection	  tract	  but	  not	  1-­‐3mm	  away	  (Figure	  2.5b).	   	   	  To	  test	  the	  time	   course	   of	   PTEN	   knockdown	   following	   single	   injections	   of	   LNP	   siRNA,	   the	  amount	   of	   PTEN	   protein	   expressed	   within	   0.5mm	   of	   the	   injection	   sites	   was	  measured	   using	   Western	   blots	   in	   a	   series	   of	   animals	   at	   different	   time	   points	  following	  injection.	  	  PTEN	  suppression	  following	  a	  single	  injection	  was	  sustained	  for	  at	   least	   15	   days	   (PTEN/β-­‐actin	   reduced	   by	   91%	   compared	   to	   control,	   P<0.001,	  Figure	  2.5c),	  the	  longest	  time	  point	  tested.	  	  	  	  	   78	  	   PTEN DiI (LNPs)control 5days 10days15days0.00.20.40.60.81.0time  af te r P T E Ns iR NA  injectio n4387PTEN/ab cPTENβ-actincontrol0-0.5 0.5-1 1-2 2-3distance  (mm)the plane of this cortical section is ~400 μm posterior from the injection tractPTENβ-actin5 days 10 days 15 dayscontrolcontrol 0-0.5 0.5-1 1-2 2-30.00.20.40.60.81.0d is tance f ro minjectio n tract (mm)6 4 4 4 4PTEN/β-actin** **********ns*mm 0.25 0.5 1.0 1.5β-actin	   79	  Figure	  2.5:	  Distance	  profile	  and	  time	  course	  of	  protein	  knockdown	  in	  vivo.	  	  	  (a)	   Immunofluorescence	   for	  PTEN	  protein	   is	  decreased	   in	  neurons	   less	   than	  1 mm	  from	  the	  site	  of	  LNP-­‐siRNA	  injection.	  The	  illustrated	  montage	  was	  obtained	  from	  a	  section	  plane	  posterior	  to	  the	  injection	  tract	  and	  the	  asterisk	  indicates	  the	  point	  that	  was	   400	   µm	   from	   the	   injection.	   (b)	   An	   example	   Western	   blot	   (inset)	   shows	  decreased	  PTEN	  protein	  in	  tissue	  obtained	  less	  than	  1 mm	  from	  the	  injection	  tract.	  Bar	  graph	  showing	  the	  summarized	  data	  from	  Western	  blots	  on	  tissue	  dissected	  at	  different	   distances	   adjacent	   to	   LNP-­‐siRNA	   injection.	   (c)	   Western	   blot	   revealed	  sustained	  knockdown	  of	  PTEN	  at	  5,	  10,	  and	  15	  days	  following	  a	  single	   intracranial	  injection	  of	  LNP-­‐siRNA.	  *P	  <	  0.05;	  **P	  <	  0.01;	  ***P	  <	  0.001.	  LNP,	  lipid	  nanoparticle;	  NS,	  not	  significant;	  siRNA,	  small	  interfering	  RNA.	  	  	  	   80	  2.3.6	  Lipid	  Nanoparticles	  are	  Capable	  of	  Widespread	  Distribution	  and	  Knockdown	  After	  ICV	  Administration	  	   Although	   local	   knockdown	   can	   be	   quite	   useful	   when	   attempting	   to	   target	  specific	  brain	   regions,	   sometimes	  more	  widespread	  knockdown	   is	  preferable.	   	  We	  therefore	   tested	  whether	   intracerebroventricular	   (ICV)	   injections	   could	   overcome	  the	  distance	  limitations	  imposed	  by	  small	  volume	  local	  injections.	  	  	  For	  the	  purpose	  of	   this	   study	  we	   focused	  on	   two	  brain	   regions	  close	   to	   the	  ventricular	   system,	   the	  dorsal	   hippocampus	   and	   the	   striatum.	   	   	   Following	   a	   single	   injection	   of	   2µL	   PTEN	  siRNA-­‐LNPs	  bilaterally	  into	  the	  lateral	  ventricles	  we	  observed	  robust	  LNP	  uptake	  in	  neurons	   as	   shown	   by	   DiI	   uptake	   in	   the	   CA1	   cell	   body	   layer	   of	   the	   hippocampus	  (Figure	  2.6a)	  at	  a	  distance	  over	  4mm	  from	  the	  injection	  site	  itself.	  	  We	  verified	  that	  the	  siRNA-­‐LNP	  was	  effective	  by	  measuring	  PTEN	  levels	  in	  both	  the	  striatum	  and	  the	  dorsal	  hippocampus	  of	  rats	  injected	  with	  PTEN	  siRNA-­‐LNPs	  (PTEN/β-­‐actin	  reduced	  55.8%	  in	  hippocampus	  and	  51.2%	  in	  striatum	  compared	  to	  control	  siRNA	  injected	  rats	  Figure	  2.6b-­‐d).	  	  	   81	  	   control siRNAPTEN siRNA0.00.20.40.60.81.01.2PTEN/-actin45control siRNAPTEN siRNA0.00.20.40.60.81.0PTEN/-actin45Hippocampus Striatumspsrso25µmaHippocampusHippocampusStriatumStriatumcontrolsiRNAPTENsiRNAbPTENβ-actinc d** *	   82	  Figure	  2.6:	  Lipid	  nanoparticles	  (LNPs)	  are	  capable	  of	  widespread	  distribution	  and	  knockdown	  after	  intracerebroventricular	  (ICV)	  administration.	  	  	  (a)	  DiI	  fluorescence	  in	  CA1	  region	  of	  the	  hippocampus	  shows	  robust	  uptake	  of	  LNPs	  by	  neurons	  in	  the	  cell	  body	  layer	  (sp).	  Scale:	  25	  μm.	  (b–d)	  Western	  blots	  show	  ICV-­‐injected	  siRNA-­‐LNPs	  results	  in	  knockdown	  of	  the	  target	  protein	  (PTEN)	  in	  different	  brain	  regions	  (dorsal	  hippocampus	  and	  striatum).	  *P	  <	  0.05;	  **P	  <	  0.01.	  siRNA,	  small	  interfering	   RNA;	   so,	   stratum	   oriens;	   sp,	   stratum	   pyramidale	   (cell	   body	   layer);	   sr,	  stratum	  radiatum.	  	   83	  2.3.7	  Lipid	  Nanoparticle	  Mediated	  Knockdown	  of	  GluN1	  in	  Cell	  Culture	  	   The	  ability	  of	  LNP-­‐siRNA	  technology	  to	  modify	  the	  synaptic	  function	  of	  nerve	  cells	  was	  tested.	  	  siRNA	  against	  GRIN1,	  the	  gene	  encoding	  the	  GluN1	  subunit	  of	  the	  NMDA	  receptor	  (NMDAR),	  was	  used	  to	  determine	  if	  a	  LNP	  siRNA	  delivery	  approach	  could	   effectively	   interfere	   with	   NMDAR	   function.	   	   The	   NMDAR	   ion	   channel	   is	  heteromeric,	  composed	  of	  two	  obligatory	  GluN1	  subunits	  plus	  two	  GluN2	  or	  GluN3	  subunits.	  	  Expression	  of	  GluN1	  is	  necessary	  for	  the	  formation	  of	  a	  functional	  NMDAR	  channel	   in	   the	  plasma	  membrane	   (Traynelis	   et	   al.,	   2010).	   	   Three	  different	   siRNAs	  directed	   against	   different	   regions	   of	   GRIN1	   were	   tested	   using	   LNP	   delivery	   in	  cultured	  neurons	  to	  determine	  their	  efficacy.	  	  All	  three	  siRNA	  sequences	  resulted	  in	  significant	  knockdown	  of	  GluN1	  expression	  (Figure	  2.7a,b),	  however	  sequence	  #1	  resulted	  in	  the	  most	  robust	  knockdown,	  and	  was	  therefore	  used	  for	  the	  subsequent	  experiments	  in	  vivo.	  	  	  Additionally,	  as	  a	  control	  for	  selectivity	  against	  the	  target	  gene,	  PTEN	  expression	  was	  not	  reduced	  by	  GluN1	  siRNA	  (Figure	  2.7a,c).	  NMDARs	  are	  typically	  arranged	  in	  clusters	  at	  synaptic	  and	  extrasynaptic	  sites	  (Rao	   and	   Craig,	   1997;	   Washbourne	   et	   al.,	   2002).	   	   To	   test	   the	   effect	   of	   GluN1	  knockdown	   on	   the	   presence	   of	   GluN1	   clusters,	   cultures	   of	   cortical	   neurons	   were	  treated	  with	  LNPs	  containing	  either	  GluN1	  siRNA	  or	  a	  control	   luc	  siRNA.	   	  Neurons	  exposed	   to	   LNP	   GluN1	   siRNA	   but	   not	   Luc	   siRNA-­‐LNP	   resulted	   in	   a	   significant	  decrease	   in	   the	   density	   of	   GluN1	   clusters	   (GluN1	   clusters/μm	   reduced	   by	   41%	  compared	  to	  control,	  P<0.001,	  Figure	  2.7d,	  e).	  	  	  	  	   84	  	  Figure	  2.7:	  LNP-­‐GluN1	  siRNA	  results	   in	  knockdown	  of	   the	  NMDAR	  obligatory	  subunit	  GluN1.	  	  	  (a,b)	  Western	  blots	  show	  that	  LNP	  encapsulation	  of	  siRNAs	  targeted	  against	   three	  distinct	   regions	   of	   GluN1	   RNA	   all	   resulted	   in	   knockdown	   of	   GluN1	   protein	  (compared	  with	  nontreated	   control	   and	   luc	   siRNA).	  Mean	  knockdown	  with	  GluN1	  siRNA	   no.	   1	   was	   greatest	   and	   was	   used	   for	   subsequent	   experiments.	   (c)	   Control	  shows	  LNP-­‐GluN1	  no.	  1	  siRNA	  did	  not	  affect	  PTEN	  expression.	  (d)	  Decreased	  GluN1	  immunofluorescence	   was	   observed	   in	   LNP-­‐GluN1	   siRNA–treated	   cultures	   (right	  panels)	   as	   compared	  with	   nontreated	   (controls)	   or	   cultures	   treated	  with	   LNP-­‐luc	  siRNA.	  Bottom:	  shows	  higher	  magnification	  of	  dashed	  box	  indicating	  GluN1	  clusters	  along	  dendrite.	  Scale:	  20	  μm.	   (e)	  Summary	  data	  of	  GluN1	  clusters/µm	  of	  dendrite.	  ***P	   <	   0.001.	   LNP,	   lipid	   nanoparticle;	   NS,	   not	   significant;	   siRNA,	   small	   interfering	  RNA.	  	  control GluN1 siRNAluc siRNAGluN1β-actinβ-actinPTENcontrolluc siRNA#1 #2 #3GluN1 siRNAa b cd eGluN1MAP2GluN1ControlLuc siRNA #1 #2 #30.00.20.40.6GluN1  s iRNAGluN1/β -actinControlLuc siRNA GluN1 #10.00.10.20.30.40.5PTEN/β -actinControlGluN1 siRNALu c siRNA0.00.51.01.52.02.51 5 2 21 8GluN1 clusters/μ m***ns******ns4 9 9 7 8 7 9 5	   85	  2.3.8	  Functional	  Knockdown	  of	  NMDAR	  Currents	  in	  vivo	  	   After	   establishing	   that	   LNP	   encapsulated	   siRNA	   against	   GRIN1	   can	   reduce	  GluN1	  and	  therefore	  NMDAR	  expression	  in	  vitro,	  we	  proceeded	  to	  test	  the	  ability	  of	  this	   approach	   to	   knock	   down	  GluN1	   expression	   and	   synaptic	   function	   in	  vivo.	   	   	   A	  single	  injection	  of	  LNPs	  containing	  GRIN1	  siRNA	  resulted	  in	  a	  significant	  decrease	  in	  GluN1	  protein	  expression	  compared	   to	  controls	   (no	   injection	  or	   rats	   injected	  with	  LNPs	   containing	   luc	   siRNA)	   when	   tested	   5	   days	   later	   (GluN1/β-­‐actin	   reduced	   by	  51%,	   and	   54%	   compared	   to	   non-­‐injected	   and	   Luc	   siRNA-­‐LNP	   injected	   controls	  respectively,	   P<0.01	   and	   P<0.05,	  Figure	   2.8a,	   b).	   	   	  We	   then	   tested	   the	   functional	  impact	   of	   disrupting	  GluN1	  expression	  by	  measuring	   the	   ratio	   of	  NMDAR/AMPAR	  excitatory	  postsynaptic	  currents	  (EPSCs)	  in	  voltage	  clamped	  neurons.	  	  Glutamate	  is	  the	   major	   excitatory	   neurotransmitter	   in	   the	   brain	   and	   acts	   on	   both	   AMPA	   and	  NMDA	  receptors	  at	  synapses.	   	  However,	  the	  NMDAR	  is	  normally	  blocked	  at	  resting	  membrane	   potential,	   and	   requires	   the	   simultaneous	   binding	   of	   glutamate	   plus	  depolarization	  to	  remove	  the	  voltage-­‐dependent	  open	  channel	  block	  by	  extracellular	  Mg2+.	   	  Therefore,	  a	  common	  protocol	  to	  test	  the	  NMDAR/AMPAR	  ratio	  in	  response	  to	  the	  synaptic	  release	  of	  glutamate	  is	  to	  measure	  the	  ratio	  of	  the	  evoked	  EPSC	  when	  the	   NMDAR	   is	   blocked	   by	   holding	   the	   cell	   at	   -­‐70mV	   to	   obtain	   a	   pure	   AMPAR	  component	   versus	   when	   the	   cell	   is	   held	   at	   +40mV	   to	   get	   a	   mixed	   AMPAR	   plus	  NMDAR	  component	  (Sah	  and	  Nicoll,	  1991;	  Mameli	  et	  al.,	  2011)(Figure	  2.8d).	  	  	  	  	  Rats	  were	   injected	  with	  LNPs	  containing	  GluN1	  siRNA	  or	   luc	  siRNA	  and	  the	  NMDAR/AMPAR	   ratio	  was	  measured	   in	   cortical	   brain	   slices	  4-­‐6	  days	   later,	   a	   time	  	   86	  that	   corresponded	   to	   decreased	   expression	   of	   GluN1	   expression	   as	   shown	   in	  Western	  blots.	   	   	  Neurons	  that	  were	  chosen	  for	  experiments	  were	  within	  500µm	  of	  the	   injection	   site,	   and	   were	   verified	   to	   have	   taken	   up	   LNPs	   based	   on	   punctate	  intracellular	  DiI	   fluorescence	   (Figure	   2.8c),	   and	   the	   ratio	   of	  NMDAR/AMPAR	  was	  measured.	  	  To	  get	  the	  NMDA	  component	  we	  measured	  the	  amplitude	  of	  the	  outward	  current	   at	   Vh	   +40mV	   50ms	   after	   the	   stimulation	   artifact,	   a	   time	   point	   when	   the	  AMPAR	  current	  had	  returned	  to	  baseline	  (This	  was	  verified	  by	  subtracting	  the	  trace	  in	   the	  presence	   of	   the	  NMDAR	  antagonist,	   APV	   to	   show	   the	   outward	   INMDAR	  (blue)	  and	   IAMPAR	  (red)	   components	   separately).	   The	  AMPA	   component	  was	  measured	   as	  the	  inward	  current	  when	  the	  cell	  was	  held	  at	  -­‐70mV	  (Figure	  2.8d).	  	  	  Consistent	  with	  the	   observed	   decrease	   in	   GluN1	   expression,	   we	   found	   that	   neurons	   treated	   with	  LNPs	   that	   contained	   GluN1	   siRNA	   had	   significantly	   decreased	   NMDAR/AMPAR	  ratios	  compared	  to	  neurons	  with	  LNPs	  from	  luc	  siRNA	  injected	  rats	  (Figure	  2.8d-­‐f).	  	  	  Therefore	  the	  NMDAR	  component	  of	  the	  synaptic	  response	  was	  selectively	  reduced	  by	   LNP-­‐delivered	  GluN1	   siRNA.	   	   These	   results	   further	   validate	   the	  use	   of	   LNPs	   to	  manipulate	  gene	  expression	  in	  the	  brain,	  showing	  that	  both	  the	  total	  protein	  levels	  as	  well	  as	  the	  function	  of	  the	  target	  protein	  were	  successfully	  disrupted.	  	   87	  	  Figure	  2.8:	  Selective	  knockdown	  of	  GluN1	  protein	  in	  vivo	  results	  in	  functional	  disruption	  of	  NMDAR	  synaptic	  currents.	  	  (a,b)	  Western	  blot	  shows	  intracranial	  injection	  of	  LNP-­‐GluN1	  siRNA	  resulted	  in	  knockdown	  of	  GluN1	  protein.	  Luc	  siRNA-­‐LNP	  control	  injection	  showed	  no	  effect.	  (c)	  DiI	  staining	  was	  observed	  in	  a	  voltage-­‐clamped	  neuron	  in	  a	  brain	  slice	  obtained	  5	  days	  after	  in	  vivo	  LNP	  injection	  indicating	  LNP	  uptake	  in	  this	  cell.	  (d)	  NMDA	  receptor	  (NMDAR)/AMPA	  receptor	  (AMPAR)	  ratio	  was	  calculated	  using	  the	  ratio	  of	  the	  NMDAR	  current	  at	  Vhold	  =	  +40	  mV	  (upper	  traces)	  at	  point	  2	  when	  the	  AMPAR	  component	  (in	  the	  lower	  trace)	  returned	  to	  baseline	  (50 ms)	  divided	  by	  the	  peak	  current	  at	  Vhold	  =	  −70	  mV	  (lower	  traces)	  at	  point	  1	  which	  was	  the	  peak	  AMPAR	  current.	  Red	  trace:	  in	  the	  presence	  of	  the	  NMDAR	  antagonist,	  APV,	  the	  NMDAR-­‐dependent	  current	  was	  blocked	  revealing	  the	  outward	  AMPAR	  current.	  Blue	  trace:	  total	  current−red	  (APV)	  current	  =	  NMDAR	  component.	  (e,f)	  Voltage	  clamp	  revealed	  a	  selective	  decrease	  in	  NMDAR/AMPAR	  currents	  in	  neurons	  from	  rats	  injected	  with	  LNP-­‐GluN1	  siRNA	  compared	  with	  a	  luc	  siRNA-­‐LNP	  control.	  *P	  <	  0.05;	  **P	  <	  0.01.	  LNP,	  lipid	  nanoparticle;	  NS,	  not	  significant;	  siRNA,	  small	  interfering	  RNA.	  	  50ms50pA+APVV hold = +40mV= -70mVholdVI NMDAR100pA50ms= -70mVholdVV hold = +40mVluciferase siRNA GluN1 siRNAp=0.0038I AMPARtransmitted DiI  ( LNP) mergeGluN1ß-actincontrol GluN1 siRNAluc siRNAI NMDARI AMPARI NMDARluc siRNAGluN1 siRNA0.00.20.40.60.81.01.21.41.6NMDAR/AMPAR1 2 21a b cd e fcontrolG luN1 siRNAL uc siRNA0.00.20.40.60.81.01.21.41.61.8GluN1/ -actinns** *6 6 3Figure 8 Rungta et al., 2013	   88	  2.4	  Discussion	  	  	   The	   results	   presented	   here	   show	   that	   LNP	   siRNA	   systems	   can	   exhibit	  efficient	  gene	  silencing	  properties	   in	  neurons	  both	   in	  vitro	   and	   in	  vivo	   that	   lead	   to	  functional	  consequences	  without	   inducing	  significant	   toxicity.	   In	  cell	  culture	  100%	  of	  neurons	  were	  observed	  to	  have	  taken	  up	  the	  LNPs	  as	  indicated	  by	  the	  appearance	  of	  DiI	  within	  the	  cytoplasm	  with	  no	  apparent	  toxicity	  or	  indications	  of	  neuron	  loss	  over	  several	  days	  to	  a	  week.	   	   In	  addition	  the	  reduction	  of	  protein	  expression	  from	  siRNA	  was	   extensive	   enough	   to	   detect	   by	  Western	   blots	   using	   total	   protein	   from	  cultures	   or	   immunocytochemistry	   of	   individual	   neurons.	   	   	   In	   experiments	   in	   vivo	  localized	  injection	  of	  siRNA-­‐LNPs	  within	  the	  cortex	  or	  ICV	  injections	  to	  cause	  more	  diffuse	  and	  widespread	  LNP	  uptake	   led	  to	  consistent	  decreases	   in	  both	  mRNA	  and	  protein	  expression	  again	  with	  no	  indication	  of	  cell	   loss	  or	  damage	  from	  the	  uptake	  of	  LNPs.	  	  Here	  we	  discuss	  the	  mechanism	  of	  action	  of	  LNP	  siRNA	  delivery;	  the	  utility	  of	  these	  LNP	  siRNA	  systems	  for	  gene	  knockdown	  both	  in	  vitro	  and	  in	  vivo	  in	  the	  brain;	  and	  potential	   therapeutic	   applications	  with	   regard	   to	  neurological	   and	  psychiatric	  diseases.	   With	   respect	   to	   the	   mechanism	   of	   action,	   the	   LNP	   delivery	   systems	  employed	  here	  contain	   the	   ionizable	  cationic	   lipid	  DMAP-­‐BLP,	  a	   lipid	   that	  exhibits	  optimized	   bilayer	   destabilizing	   and	   pKa	   properties	   leading	   to	   highly	   potent	   gene	  silencing	   in	  hepatocytes	   following	   intravenous	  administration	   (see	  Supplementary	  Information)	   that	   is	   similar	   to	   “gold	   standard”	   lipids	   such	   as	   DLinMC3-­‐DMA8.	  Detailed	   studies	   have	   shown	   that	   the	   ability	   of	   LNP	   siRNA	   systems	   containing	  	   89	  related	   ionizable	   cationic	   lipids	   to	   induce	   hepatocyte	   gene	   knockdown	   in	   vivo	   is	  ApoE	   dependent	   (Akinc	   et	   al.,	   2010).	   In	   ApoE	   knockout	   mice	   LNP	   siRNA	   gene	  silencing	  activity	  is	  inhibited	  in	  hepatocytes	  and	  can	  be	  re-­‐established	  by	  incubating	  the	  LNPs	  with	  ApoE	  prior	  to	  IV	  administration.	  In	  the	  brain	  and	  in	  neuronal	  cultures	  ApoE	  is	  synthesized	  in	  astrocytes(Pitas	  et	  al.,	  1987;	  Bu,	  2009).	  	  The	  astrocytes	  in	  the	  neuronal	  culture	  system	  are	  separated	  into	  a	  feeder	  layer	  that	  allowed	  us	  to	  test	  the	  ApoE	  dependence	  of	  LNP	  uptake	  by	   removing	   the	   separate	  astrocyte	   feeder	   layer.	  	  With	  no	  added	  ApoE	  there	  was	  no	  evidence	  of	  neuronal	  LNP	  uptake	  whereas	  graded	  uptake	  was	  observed	  when	  ApoE	  concentrations	  were	  increased	  above	  0.1µg/ml	  to	  1	  and	  5µg/ml	  ApoE.	   	  The	  concentrations	  of	  ApoE	  found	   in	  the	  brain	  are	  similar	   to	  the	  concentrations	  we	  found	  to	  be	  effective	  in	  cell	  culture	  (Wahrle	  et	  al.,	  2007).	  	  This	  suggests	  that	  the	  natural	  affinity	  of	  these	  LNP	  systems	  for	  endogenous	  ApoE	  in	  the	  brain	  provides	  a	  highly	  efficient	  method	  for	  LNP	  delivery	  into	  neurons	  in	  vivo.	  	  The	  utility	  of	  the	  LNP	  siRNA	  systems	  for	  silencing	  target	  genes	  in	  neurons	  in	  cell	   culture	   is	   indicated	   by	   our	   observations	   that	   transfection	   efficiencies	  approached	   100%,	   without	   overt	   signs	   of	   toxicity.	   	   In	   contrast	   current	   non-­‐viral	  vectors	  exhibit	   transfection	  efficiencies	  of	  10%	  or	   less,	  and	  are	  often	  accompanied	  by	  cell	  damage	  (Karra	  and	  Dahm,	  2010).	  	  	  LNPs	  appear	  to	  be	  a	  superior	  alternative	  to	  presently	  used	  techniques	  to	  transfect	  neurons	  in	  cell	  culture	  based	  not	  only	  on	  the	   high	   rate	   of	   uptake	   but	   also	   on	   the	   benign	   nature	   of	   the	   LNPs.	   Another	   key	  finding	  in	  our	  study	  is	  that	  the	  LNP-­‐siRNA	  approach	  we	  describe	  here	  provides	  an	  effective	   alternative	   to	   other	   in	   vivo	   transfection	   vectors	   presently	   used.	   	   Viral	  delivery	   is	   the	   most	   effective	   available	   way	   to	   transfect	   neurons	   (Hommel	   et	   al.,	  	   90	  2003).	  	  The	  disadvantage	  of	  viral	  delivery	  is	  the	  requirement	  for	  constructing	  virus	  vectors	  that	  take	  added	  time,	  potentially	  cause	  immune	  responses	  and	  raise	  safety	  concerns.	   	  The	  use	  of	  LNPs	  can	  circumvent	  these	  issues,	  and	  provides	  a	  method	  to	  rapidly	   test	   various	   siRNA	   constructs	   against	   the	   expression	   of	   different	   target	  proteins,	   and	   then	   examine	   the	   protein’s	   function.	   	   Injection	   of	   antisense	  oligonucleotides	   (ASOs)	   into	   the	   CSF	   represents	   another	   promising	   method	   to	  silence	  neuronal	   gene	   expression	   in	  vivo	   (Karra	   and	  Dahm,	  2010;	  Kordasiewicz	   et	  al.,	  2012).	  However,	   two	  major	  problems	  associated	  with	  ASOs	  are	   the	  high	  doses	  required	   to	   achieve	   gene	   silencing	   effects	   and	   immunostimulatory	   effects	  (Zalachoras	  et	  al.,	  2011;	  Robinson,	  2012).	  	  With	  regard	  to	  dose	  levels,	  ICV	  infusion	  of	  free	   antisense	   oligonucleotides	   against	   superoxide	   dismutase	   (SOD)	   by	   osmotic	  pump	   at	   100µg/day	   per	   rat	   over	   28	   days	   (total	   of	   2.8mg	   of	   antisense	  oligonucleotide)	  resulted	   in	  approximately	  50%	  SOD	  mRNA	  silencing	  (Smith	  et	  al.,	  2006).	   	   	  Here	  we	  demonstrate	   that	  a	  single	  LNP	  siRNA	   injection	  of	  2.5µg	  of	  siRNA	  per	   rat	   by	   intracortical	   administration	   results	   in	   pronounced	   (up	   to	   91%)	   gene	  silencing	  that	  is	  sustained	  for	  at	  least	  15	  days	  post	  injection.	  Clinical	  studies	  suggest	  that	   the	   immunostimulatory	   effects	   of	   LNP	   siRNA	   injections	   are	   infrequent	   and	  readily	   managed,	   see	   http://www.alnylam.com/capella/wp-­‐content/uploads/2012/07/Alnylam-­‐ALN-­‐TTR02-­‐PhaseI-­‐Results-­‐120716.pdf.	  Consistently,	   we	   found	   no	   evidence	   that	   LNP-­‐siRNAs	   caused	   immunostimulatory	  effects	   in	   brain	   tissue	   as	   we	   found	   no	   increased	   synthesis	   of	   TNF-­‐α	   triggered	   by	  LNPs	  (Figure	  2.4f).	   	  In	  addition	  there	  are	  now	  well-­‐established	  methods	  to	  reduce	  	   91	  or	   eliminate	   siRNA	   immunogenicity	   by	   adjusting	   siRNA	   chemistry	   (Bramsen	   and	  Kjems,	  2012).	  	  The	  LNP	  formulation	  employed	  here	  used	  PEG-­‐DMG	  which	  is	  a	  type	  of	  PEG-­‐lipid	  with	  short	  acyl	  chains	  that	  allow	  it	  to	  dissociate	  rapidly	  (Holland	  et	  al.,	  1996)	  from	  LNPs	  upon	  dilution,	  allowing	  association	  of	  proteins	  such	  as	  ApoE.	  	  It	  would	  be	  expected	   that	   larger	   radii	   of	   distribution	   could	   be	   achieved	   for	   LNPs	   containing	   a	  PEG-­‐lipid	   with	   longer	   acyl	   chains,	   which	   do	   not	   dissociate	   so	   rapidly.	   Enhanced	  tissue	  penetration	  would	  also	  be	  expected	  for	  smaller	  LNPs	  that	  can	  be	  achieved	  at	  higher	  PEG-­‐lipid	   levels	  (Belliveau	  et	  al.,	  2012).	  The	  LNP	  platform	  is	  highly	   flexible,	  potentially	   allowing	   delivery	   of	   other	   materials	   to	   neurons	   such	   as	   drugs,	  fluorescent	  dyes	  and	  plasmids.	  	  	  	  	  The	  results	  presented	  here	  demonstrate	  the	  utility	  of	  LNP	  siRNA	  systems	  for	  silencing	   target	  genes	   in	  neurons	  both	   in	  vitro	   and	   in	  vivo,	  with	  obvious	  utility	   for	  rapidly	  advancing	  functional	  genomics	  studies.	  The	  ability	  to	  silence	  multiple	  genes,	  even	   five	   or	   more	   genes	   at	   once	   (Love	   et	   al.,	   2010),	   may	   also	   prove	   useful.	   The	  therapeutic	  utility	  of	  direct	  intra-­‐cranial	  administration	  of	  LNP	  siRNA	  for	  treatment	  of	   neurological	   disorders	   remains	   to	   be	   established,	   however	   there	   are	   now	   a	  number	   of	   clinically	   accepted	   therapies	   relying	   on	  direct	   pumping	   of	   therapeutics	  into	  the	  CSF	  (North,	  1997;	  Dash	  and	  Cudworth,	  1998).	  A	  lipid-­‐based	  formulation	  of	  cytarabine	   has	   gained	   clinical	   acceptance	   for	   treatment	   of	   brain	   cancer	   following	  direct	   administration	   into	   CSF	   (Phuphanich	   et	   al.,	   2007).	   Further,	   a	   number	   of	  clinical	   trials	  are	  now	   in	  progress	   for	   treatments	   for	   serious	  neurological	  diseases	  	   92	  that	  rely	  on	  continuous	  infusion	  of	  ASOs	  into	  the	  CSF	  (Smith	  et	  al.,	  2006;	  Robinson,	  2012).	  	   93	  Chapter	  3:	  Mechanisms	  of	  neuronal	  chloride	  loading	  underlying	  cytotoxic	  edema.	  	  3.1	  Introduction	  	   Brain	   edema,	   the	   pathological	   hallmark	   of	   excitotoxic	   injury	   and	   traumatic	  brain	   injury	   (Klatzo,	   1987;	  Marmarou	   et	   al.,	   2006;	   Rosenblum,	   2007;	   Donkin	   and	  Vink,	   2010)	   was	   first	   characterized	   by	   Klatzo	   (1967)	   as	   either	   vasogenic	   or	  cytotoxic.	  	  Cytotoxic	  brain	  edema	  is	  caused	  by	  water	  movement	  into	  the	  intracellular	  compartment	   of	   neurons	   and/or	   astrocytes	   leading	   to	   brain	   swelling,	   while	  vasogenic	  edema	  is	  due	  to	  water	  entry	  into	  the	  brain	  from	  the	  vasculature	  (Klatzo,	  1967).	  Excitotoxic	   swelling	  of	   cultured	  neurons	   is	  known	   to	   involve	   influx	  of	  both	  Na+	  and	  Cl-­‐	  although	  the	  influx	  pathway(s)	  for	  Cl-­‐	  remain	  obscure	  (Rothman,	  1985;	  Choi,	   1987;	   Hasbani	   et	   al.,	   1998).	   	   The	   low	   resting	   Cl-­‐	   permeability	   in	   neurons	  suggests	  that	  a	  Cl-­‐	  channel	  or	  exchange	  mechanism	  must	  be	  activated	  for	  Cl-­‐	  entry	  to	  occur	   at	   sufficient	   levels	   to	   increase	   cell	   volume	   and	   cause	   cytotoxic	   edema.	   In	  mature	  pyramidal	  neurons	  of	  the	  cortex	  and	  hippocampus,	  the	  equilibrium	  potential	  for	  Cl-­‐	   (ECl-­‐)	   is	  set	  more	  hyperpolarized	  to	  the	  resting	  membrane	  potential	  (Em)	  by	  KCC2-­‐mediated	   active	   transport	   of	   Cl-­‐	   out	   of	   the	   cell	   against	   it’s	   electrochemical	  concentration	  gradient	  (Blaesse	  et	  al.,	  2009).	  As	  such,	  the	  Cl-­‐	  influx	  that	  is	  required	  for	   cytotoxic	   neuronal	   edema	   occurs	   as	   a	   result	   of	   either	   the	   activation	   of	   a	   Cl-­‐	  channel	   that	   is	   not	   open	   at	   rest,	   or	   activation	   of	   a	   Cl-­‐	   transporter.	   	   Putative	  candidates	   for	   Cl-­‐	   loading	   leading	   to	   swelling	   are	   the	   volume	   regulated	   anion	  	   94	  channel	   (VRAC),	   the	   Na+-­‐K+-­‐Cl+	   cotransporter	   1	   (NKCC1)	   and	   GABA	   activated	   Cl-­‐	  channels	   (Hasbani	   et	   al.,	   1998;	   Allen	   et	   al.,	   2004;	   Inoue	   et	   al.,	   2005;	   Pond	   et	   al.,	  2006).	  	  In	  addition,	  there	  are	  several	  newly	  described	  Cl-­‐	  channels	  and	  transporters	  that	   could	   also	   be	   important	   contributors	   to	   neuronal	   edema.	   	   Our	   experiments	  were	   designed	   to	   examine	   the	   interrelationship	   between	   neuronal	   volume,	  intracellular	  Na+	  ([Na+]i)	  and	  intracellular	  Cl-­‐	  ([Cl-­‐]i)	  in	  order	  to	  investigate	  the	  roles	  for	  Cl-­‐	  entry	  pathways	  that	  contribute	  to	  neuronal	  swelling.	  	  	  	  	  	  Neuronal	  swelling	  occurs	  as	  a	  result	  of	  multiple	  triggers	  that	  increase	  [Na+]i	  including	   excessive	   glutamate	   receptor	   activation,	   intense	   neuronal	   spiking,	  activation	  of	  non-­‐selective	  cation	  channels	  and	  inhibition	  of	  Na+,	  K+,	  ATPase	  (Liang	  et	  al.,	  2007).	  We	  tested	  the	   impact	  of	   increasing	  [Na+]i	  via	   ligand-­‐	  or	  voltage-­‐gated	  ion	   channels	  on	  neuronal	   swelling	   to	   test	   the	  hypothesis	   that	   extensive	  Na+	   influx	  itself,	   independent	   of	   the	   route	   of	   entry,	   leads	   to	   swelling	   by	   triggering	   Cl-­‐	   influx.	  Two-­‐photon	   imaging	   of	   cell	   morphology	   and	   fluorescence	   lifetime	  measurements	  (FLIM)	  of	  [Na+]i	  and	  [Cl-­‐]i	   in	  hippocampal	  and	  cortical	  neurons	  in	  acutely	  prepared	  brain	   slices	   were	   combined	   to	   specifically	   examine	   the	   relationship	   between	  increased	   [Na+]i,	   subsequent	   [Cl-­‐]i	   changes	   and	   neuronal	   swelling.	   	   The	   cytotoxic	  nature	   of	   this	   swelling	  was	  measured	   by	   lactate	   dehydrogenase	   (LDH)	   efflux	   (e.g.	  (Kajta	   et	   al.,	   2005)).	   	   Pharmacological	   blockers	   of	   known	   Cl-­‐	   channels	   and	  exchangers	  were	   further	  examined	   in	  order	   to	  determine	  the	  relative	  contribution	  of	  different	  Cl-­‐	  loading	  pathways	  to	  neuronal	  swelling.	  	  Finally,	  a	  novel	  nanoparticle	  (LNP)	   strategy	   to	   introduce	   siRNA	   into	   neurons	   in	   vivo	   (described	   in	   chapter	   2)	  (Rungta	  et	  al.,	  2013)	  was	  employed	  to	  determine	  the	  exact	  Cl-­‐	  pathway	  critical	  and	  	   95	  required	   for	   the	  majority	  of	  neuronal	  swelling.	   	  These	  results	   indicate	   that	  a	   large	  proportion	  of	  neuronal	  swelling	  	  and	  subsequent	  cell	  death	  requires	  SLC26A11;	  a	  Cl-­‐,	   HCO3-­‐,	   SO4-­‐	   exchanger	   recently	   reported	   to	   be	   highly	   expressed	   in	   cortical	   and	  hippocampal	   neurons	   (Rahmati	   et	   al.,	   2013).	   	   The	   identification	   of	   the	   principal	  pathway	  required	  for	  Cl-­‐	  entry	  could	  potentially	  lead	  to	  novel	  targets	  and	  therapies	  for	  treating	  cytotoxic	  brain	  edema.	  	  3.2	  Materials	  and	  Methods	  	  3.2.1	  Slice	  Preparation.	  	   Rats	   were	   anesthetized	   with	   halothane	   and	   decapitated	   according	   to	  protocols	   approved	   by	   the	   University	   of	   British	   Columbia	   Committee	   on	   Animal	  Care.	   Brains	   were	   rapidly	   extracted	   and	   placed	   into	   ice-­‐cold	   slicing	   solution	  containing	   (in	   mmol/l):	   NMDG,	   120;	   KCl,	   2.5;	   NaHCO3,	   25;	   CaCl2,	   1;	   MgCl2,	   7;	  NaH2PO4,	  1.25;	  glucose,	  20;	  Na-­‐pyruvate,	  2.4;	  and	  Na-­‐ascorbate,	  1.3;	  saturated	  with	  95%	  O2/5%	  CO2.	   Coronal	   hemisections	   or	   transverse	   hippocampal	   slices,	   400	   µm	  thick,	  were	  sliced	  using	  a	  vibrating	  tissue	  slicer	  (VT1200,	  Leica,	  Nussloch,	  Germany).	  	  Slices	  were	   incubated	   at	   32	   °C	   in	   artificial	   CSF	   containing	   (in	  mmol/l):	  NaCl,	   126;	  KCl,	   2.5;	   NaHCO3,	   26;	   CaCl2,	   2.0;	   MgCl2,	   1.5;	   NaH2PO4,	   1.25;	   and	   glucose,	   10;	  saturated	  with	  95%	  O2/5%	  CO2	  for	  30	  minutes.	  For	  experiments,	  slices	  were	  at	  22–24	  °C	  and	  perfused	  at	  ˜2 ml/minute.	  	  	   96	  3.2.2	  Imaging	  	   Live	   cell	   imaging	   (brain	   slice)	   was	   performed	   with	   a	   two-­‐photon	   laser-­‐scanning	  microscope	  (Zeiss	  LSM510-­‐Axioskop-­‐2;	  Zeiss,	  Oberkochen,	  Germany)	  with	  a	   40X-­‐W/1.0	   numerical	   aperture	   objective	   lens	   directly	   coupled	   to	   a	   Chameleon	  ultra2	   laser	   (Coherent,	   Santa	   Clara,	   CA).	   CoroNa,	   SR101	   and	   DiI	   were	   excited	   at	  770 nm,	   and	   N-(ethoxycarbonylmethyl)-6-methoxyquinolinium bromide	   (MQAE)	   was	  excited	   at	   760	   nm.	   The	   fluorescence	   from	   each	   fluorophore	   was	   split	   using	   a	  dichroic	   mirror	   at	   560 nm,	   and	   the	   signals	   were	   each	   detected	   with	   a	   dedicated	  photo	   multiplier	   tube	   after	   passing	   through	   an	   appropriate	   emission	   filter	   (DiI,	  SR101:	   605 nm,	   55 nm	   band	   pass;	   CoroNa,	   MQAE:	   525 nm,	   50 nm	   band	   pass).	  Transmitted	   light	   was	   simultaneously	   collected	   using	   understage	   infrared	  differential	  interference	  contrast	  optics	  and	  an	  additional	  photo	  multiplier	  tube.	  	  3.3.3	  Fluorescence	  Lifetime	  Imaging	  (FLIM)	  	  	  	   Fluorescence	  lifetime	  images	  were	  acquired	  using	  a	  Becker	  &	  Hickl	  SPC-­‐150	  module.	   Photon	   emission	  was	   detected	   using	   a	   high	   speed	   hybrid	   detector,	   HPM-­‐100-­‐40	   (Hamamatsu).	   	   Images	  were	   acquired	   128	   by	   128	   pixels	   in	   fast	   xy	   raster	  scanning	   mode.	   	   Photons	   were	   collected	   over	   20	   seconds	   before	   calculating	   and	  extracting	   lifetimes	   at	   each	   pixel	   using	   SPCImage	   software	   (Becker	   &	   Hickl).	  	  Instrument	  response	  function	  (IRF)	  was	  calculated	  using	  a	  10nm	  gold	  nanoparticle	  suspension	   (Sigma-­‐Aldrich)	   to	   generate	   a	   second-­‐harmonic	   signal.	   	   The	   IRF	   had	   a	  full	  width	  at	  half	  the	  maximum	  amplitude	  of	  121ps.	  	  The	  lifetimes	  presented	  in	  the	  	   97	  figures	   were	   the	   average	   of	   all	   lifetimes	   within	   a	   region	   of	   interest	   from	   the	  cytoplasm	  of	  the	  soma.	  	  The	  mean	  lifetime	  from	  all	  cells	  in	  a	  given	  experiment	  were	  combined	  and	  represented	  as	  an	  n=1.	  MQAE:	  The	  Cl-­‐	  dependence	  of	  MQAE	  lifetime	  is	  described	   by	   the	   Stern-­‐Volmer	   relation	   (τ0/τ	   =	   1	   +	   Ksv	   [Cl-­‐]i),	   where	   τ0	   is	   the	  fluorescence	  lifetime	  in	  0	  mM	  Cl-­‐,	  and	  Ksv	  (the	  Stern-­‐Volmer	  constant)	  is	  a	  measure	  of	  the	  Cl-­‐	  sensitivity	  of	  MQAE.	  Ksv	  varies	  greatly	  between	  studies	  due	  to	  differences	  in	  cell	   types,	  preparation	  and	  calibration	  methods	   (Kaneko	  et	  al.,	  2001;	  Kaneko	  et	  al.,	   2004;	   Hille	   et	   al.,	   2009;	   Doyon	   et	   al.,	   2011),	   making	   it	   difficult	   to	   accurately	  estimate	   absolute	   [Cl-­‐]i	   in	   situ.	   	   CoroNa:	   A	   biexponential	   decay	   was	   used	   to	   fit	  CoroNa	   lifetimes	   due	   to	   poor	   fit	   with	   a	   single	   exponential	   decay,	   suggesting	  fluorescence	   from	  multiple	   components.	  For	  calibration,	  neurons	  were	  continually	  voltage-­‐clamped	  at	  -­‐70	  mV	  and	  dialyzed	  for	  >10	  min	  before	  image	  acquisition.	  The	  intracellular	   solution	  contained	   (in	  mM)	  potassium	  gluconate,	  108;	  KCl,	  8;	   sodium	  gluconate,	   8;	   MgCl2,	   2;	   HEPES,	   10;	   potassium	   EGTA,	   1;	   potassium,	   ATP,	   4;	   and	  sodium	   GTP,	   0.3;	   pH	   7.2	   with	   KOH.	   	   	   Sodium	   concentration	   was	   altered	   by	  replacement	   of	   potassium	   gluconate	  with	   sodium	   gluconate.	   A	   linear	   fit	   of	   τfast	   vs	  [Na+]	  was	  used	  to	  then	  estimate	  experimental	  values	  of	  [Na+]i.	  	  	  3.2.4	  Dye	  Loading	  Protocols	  	   Slices	  were	  incubated	  in	  ACSF	  plus	  SR101	  (1	  μM)	  at	  room	  temperature	  for	  30	  minutes.	  	  For	  CoroNa	  and	  Calcein	  Red-­‐AM	  loading,	  slices	  were	  preincubated	  in	  3	  mL	  aCSF	  and	  8	  μL	  Cremophor	  EL	  solution	  (0.5%	  in	  DMSO)	  at	  32	  °C	  for	  5	  min.	  AM	  dye	  	   98	  (50	  μg)	  mixed	  with	  8	  μL	  DMSO	  and	  2	  μL	  pluronic	  F-­‐127	   solution	   (10%	   in	  DMSO)	  was	  then	  added,	  and	  slices	  were	  allowed	  to	  incubate	  for	  an	  additional	  45	  min.	  	  For	  MQAE	  loading,	  slices	  were	  bulk	  loaded	  with	  the	  dye	  MQAE	  (6	  mM)	  for	  15	  minutes	  at	  34	  °C.	  	  3.2.5	  LDH	  Assay	  	   LDH	  assay	  kits	  (Biomedical	  Research	  Service	  Center,	  State	  University	  of	  New	  York	   at	   Buffalo)	  were	   used	   to	   investigate	   cell	   death	   using	   rat	   hippocampal	   slices.	  Hippocampal	   brain	   slices	   were	   prepared	   as	   described	   in	   brain	   slice	   preparation	  section	   above.	  Hippocampal	   slices	  were	   pre-­‐treated	   for	   30	  min.	  with	   a	   cocktail	   of	  ligand	  gated	   and	  voltage	   gated	   ion	   channel	   inhibitors	   (100	  μM	  picrotoxin	   ,	   20	  μM	  CNQX,	  1	  μM	  TTX)	  when	  100	  μM	  NMDA	  was	  applied	  or	  (100	  μM	  picrotoxin	  ,	  20	  μM	  CNQX,	   30	   μM	   Cadmium,	   100	   μM	   d-­‐APV)	   	   when	   50	   μM	   veratridine	   was	   applied.	  	  NMDA	  experiments	  were	  done	  in	  0	  mM	  Ca2+,	  2	  mM	  ethylene	  glycol	  tetraacetic	  acid	  (EGTA).	   	   NMDA	   or	   Veratridine	  was	   applied	   to	   slices	   for	   15	  min	   in	   a	   6	  well	   plate	  aerated	  with	   95%	  O2/5%	  CO2	   on	   an	   insert	   for	   organotypic	   culture	   (Millipore)	   for	  better	   aeration.	   Subsequently	   slices	   were	   transferred	   to	   incubation	   chamber	   and	  further	  incubated	  for	  90	  min.	  	  Supernatants	  were	  collected	  at	  90	  min	  and	  then	  slices	  were	   lysed	  using	   lysis	  buffer.	  The	  LDH	  level	   in	   the	  supernatant	  represents	   the	  cell	  death,	   whereas	   the	   LDH	   level	   in	   lysed	   cells	   represents	   the	   viable	   cells.	   In	   brief,	  supernatants	  and	  cell	  lysates	  were	  centrifuged	  for	  3	  min	  at	  maximal	  speed	  (16,000	  g)	   at	   4	   °C.	   Samples	  were	   added	   into	   a	   96-­‐well	   plate	  with	   LDH	   assay	   solution	   and	  	   99	  incubated	  for	  30	  min	  at	  37	  °C.	  Acetic	  acid	  (3%)	  was	  added	  to	  stop	  the	  reaction.	  LDH	  reduces	   tetrazolium	   salt	   INT	   to	   formazan,	  which	   is	  water-­‐soluble	   and	   exhibits	   an	  absorption	   maximum	   at	   492 nm.	   Absorbance	   was	   measured	   at	   492 nm	   using	   a	  microplate	  reader.	  Cell	  death	  is	  presented	  as	  the	  percentage	  of	  LDH	  released	  (LDH	  in	  supernatant/cell	  lysate	  LDH)	  *100.	  	  3.2.6	  Intracranial	  Injections	  	   All	  experimental	  protocols	  were	  approved	  by	  the	  Committee	  on	  Animal	  Care,	  University	   of	   British	   Columbia	   and	   conducted	   in	   compliance	   with	   guidelines	  provided	  by	  the	  Canadian	  Council	  of	  Animal	  Care.	  Sprague–Dawley	  rats	  (P22-­‐P26)	  were	  anesthetized	  with	  isofluorane	  before	  and	  throughout	  the	  surgery.	  A	  small	  hole	  (diameter	   ˜1 mm)	   was	   drilled	   in	   the	   skull	   to	   allow	   access	   to	   the	   brain	   (−2.0 mm	  anterior/posterior	  (AP)	  and	  ±3.0 mm	  medial/lateral	  (ML)	  from	  bregma	  and	  0.8 mm	  dorsal/ventral	  (DV)).	  A	  glass	  micropipette	  (tip	  diameter	  ~40	  µm)	  was	  connected	  to	  a	  Hamilton	  syringe	  and	  LNP-­‐siRNAs	   in	  sterile	  PBS	  were	   injected	  using	  an	   infusion	  pump	   (Harvard	   Apparatus,	   Holliston,	   MA)	   at	   a	   rate	   =	   ~50	   nl/minute.	   The	   total	  volume	  injected	  was	  500	  nl	  of	  LNP-­‐siRNA	  (5 mg	  siRNA/ml	  in	  sterile	  PBS).	  	  	  3.2.7	  Lipid	  Nanoparticle	  Encapsulation	  of	  siRNA	  	  The	   ionizable	   cationic	   lipid	  3-­‐(dimethylamino)propyl(12Z,15Z)-­‐3-­‐[(9Z,12Z)-­‐octadeca-­‐9,12-­‐dien-­‐1-­‐yl]henicosa-­‐12,15-­‐dienoate	   (DMAP-­‐BLP)	   and	   PEG	   lipid	   PEG-­‐	   100	  DMG	   were	   provided	   by	   Alnylam	   Pharmaceuticals	   and	   have	   been	   previously	  described	  (Rungta	  et	  al.,	  2013).	  	  1,2-­‐distearoyl-­‐sn-­‐glycero-­‐3-­‐phosphocholine	  (DSPC)	  and	  cholesterol	  were	  obtained	  from	  Avanti	  (Alabaster,	  AL)	  and	  Sigma-­‐Aldrich	  Co.	  (St.	  Louis,	   MO)	   respectively.	   	   The	   lipophilic	   carbocyanine	   dye	   to	   monitor	   LNP	   siRNA	  uptake,	  1,1'-­‐dioctadecyl-­‐3,3,3',3'-­‐tetramethylindocarbocyanine	  perchlorate	  (DiIC18),	  was	   purchased	   from	   Invitrogen	   (Carlsbad,	   CA).	   	   The	   lipid	   composition	   of	   all	   lipid	  nanoparticles	   containing	   siRNA	   (LNP-­‐siRNA)	   was	   DMAP-­‐BLP/DSPC/cholesterol/PEG-­‐DMG/DiIC18	   (50/10/37.5/1.5/1;	   mol%)	   LNP-­‐siRNA	  were	  prepared	  employing	  a	  microfluidic	  mixing	  apparatus	  as	  previously	  described	  (Rungta	   et	   al.,	   2013).	   Physical	   parameters	   characterizing	   the	   LNP	   siRNA	   systems	  and	  the	  siRNA;lipid	  ratio	  are	  listed	  in	  Table	  3.1.	  	  	  	  	  	  	  	   101	  Table	  3.1:	  Characterization	  of	  LNP-­‐siRNA	  systems	  siRNA	   	   Size	  (nm)	   PDI	  siRNA/Lipid	  ratio	  (mg/µmol)	  Luc	   	   48.9	   0.039	   0.050	  SLC4A3	   	   50.6	   0.134	   0.064	  SLC4A8	   	   61.5	   0.075	   0.067	  SLC4A10	   	   59.3	   0.080	   0.062	  SLC26A11-­‐1	   	   51.4	   0.200	   0.052	  SLC26A11-­‐2	   	   49.3	   0.048	   0.051	  	   	   	   	   	  PDI,	  polydispersity	  index;	  siRNA	  small	  interfering	  RNA.	  	  	   102	  3.2.8	  Quantitative	  PCR	  (qPCR)	  	  	   Total	   RNA	   was	   Purified	   using	   Life	   Technologies	   MagMaxTM-­‐96	   Microarray	  Total	  RNA	  Isolation	  Kit	  (AM1839),	  and	  cDNA	  created	  using	  the	  Applied	  Biosystems	  High	   Capacity	   cDNA	   Reverse	   Transcription	   Kit	   (4368814).	   Gene	   specific	   qPCR	  reactions	  were	  set	  up	  using	  the	  KAPA	  Probe	  Fast	  Universal	  qPCR	  Kit	  (KK4702)	  using	  TaqManTM	  probes	   from	  Life	  Technologies	  and	   Integrated	  DNA	  Technologies	   (IDT).	  Specific	   TaqManTM	   probes	   used	   in	   this	   study:-­‐	   	   rSLC4A1	   (Rn00561909_m1),	  rSLC4A2	   (Rn00566910_m1),	   rSLC4A3	   (Rn00436642_m1),	   rSLC4A4	  (Rn00584747_m1),	   rSLC4A5	   (Rn01420902_m1),	   rSLC4A7	   (Rn00589539_m1),	  rSLC4A8	   (Rn01532883_m1),	   rSLC4A9	   (Rn00596175_m1),	   rSLC4A10	  (Rn00710136_m1),	   rSLC4A11	   (Rn01515154_m1),	   rACTB	   (4352340E),	   rGAPD	  (4352338E)	  from	  Life	  Technologies,	  and	  rSLC26A1	  (Rn.PT.53a.10186844),	  SLC26A2	  (Rn.PT.53a.38316256),	   rSLC26A3	   (Rn.PT.53a.13331783),	   rSLC26A4	  (Rn.PT.53a.37046344),	   rSLC26A5	   (Rn.PT.53a.14115266),	   rSLC26A6	  (Rn.PT.53a.12307129gs),	   rSLC26A7	   (Rn.PT.53a.10816391),	   rSLC26A9	  (Rn.PT.53a.6784539),	   rSLC26A10	   (Rn.PT.53a.5866317),	   rSLC26A11-­‐1	  (Rn.PT.53a.36735939),	   rSLC26A11-­‐2	   (Rn.PT.53a.13211186),	   rGAPDH	  (Rn.PT.56a.35727291)	  all	  from	  IDT.	  Quantitative	  PCR	  reactions	  were	  performed	  on	  a	  Life	  Technologies	  7500	  Real-­‐Time	  PCR	  system	  or	  Bio-­‐Rad	  CFX	  Real-­‐Time	  Systems,	  using	  cycling	  conditions	  of	  95oC	  for	  3	  minutes	  then	  95oC	  for	  15	  seconds	  followed	  by	  60oC	   for	   45	   seconds	   for	   40	   cycles.	   The	   rGAPDH	   or	   rACTB	   probes	   were	   used	   for	  	   103	  normalization	   of	   controls	   in	   relative	   quantification	   of	   other	   gene	   expression	  measurements.	  	  	  3.2.9	  Gene	  Knock-­‐Down	  Dicer-­‐Substrate	  RNAs	  (DsiRNAs)	  	  	   Chemically	  synthesized	  siRNA	  27mer	  duplexes	  were	  obtained	  from	  IDT	  and	  screened	   for	   knockdown	   potency	   by	   qPCR	   in	   culture	   using	   plasmid	   expression	  knockdown	  of	  the	  cloned	  rSLC	  targets	  expressed	  in	  HEK293	  cells	  and	  of	  cultured	  rat	  cortical	   neurons.	   The	   most	   potent	   duplexes	   were	   then	   re-­‐synthesised	   with	   2’-­‐O-­‐methyl	  (m)	  patterning	  for	  in	  vivo	  stabilization,	  packaged	  into	  LNPs	  and	  re-­‐tested	  for	  potency	  of	  gene	  expression	  knockdown	  in	  cultured	  rat	  cortical	  neurons	  by	  qPCR.	  rSLC4A3	   (Sense-­‐	   rGrGrArUrUrArCrUrCrUrArUrCrArCrArGrArCrArCrCrUAC,	  Antisense-­‐rGrUrArGrGrUrGrUrCrUmGrUmGrArUrArGrArGrUmArAmUrCmCmAmC)	  rSLC4A8	   (Sense-­‐	   rArCrArGrCrGrGrUrCrUrUrArArArGrUrUrUrArUrCrCrCAA,	  Antisense-­‐rUrUrGrGrGrArUrArArAmCrUmUrUrArArGrArCrCmGrCmUrGmUmCmA)	  rSLC4A10	   (Sense-­‐	   rUrGrCrUrUrArUrArArArGrCrUrArArArGrArCrCrGrCrAAT,	  Antisense-­‐rArUrUrGrCrGrGrUrCrUmUrUmArGrCrUrUrUrArUmArAmGrCmAmAmC)	  	  rSLC26A11-­‐1	   (Sense-­‐	   rGrCrArUrGrUrCrArGrCrArArUrArUrArGrArCrUrArCrACC,	  Antisense-­‐rGrGrUrGrUrArGrUrCrUmArUmArUrUrGrCrUrGrAmCrAmUrGmCmGmU)	  rSLC26A11-­‐2	  	  	  	  	  (Sense-­‐mGmGrAmGrAmUrCrCrArArUmArCmGrGmCrAmUrCrCrUrGrGmCA,	  Antisense-­‐rUrGmCrCrArGrGrAmUrGmCrCmGrUrArUrUrGrGrArUrCmUrCmCmCmA)	  	  	   104	  3.2.10	  Drugs	  	  	  Drugs	   were	   purchased	   from	   the	   following	   suppliers;	   Veratridine,	   d-­‐APV,	  TTX,	  CNQX	  (abcam);	  DIDS,	  picrotoxin,	  niflumic	  acid,	  carbenoxelone,	  acetazolamide,	  bumetinide	  (sigma);	  NPPB	  (tocris).	  Targets	  are	  as	  follows	  (Supplementary	  Table	  1);	  NPPB	  (200	  μM),	  volume	  regulated	  anion	  channel	  (VRAC,	  VSOR)	  (Inoue	  et	  al.,	  2005;	  Inoue	   and	   Okada,	   2007);	   niflumic	   acid	   (NFA)	   (200	   μM),	   Ca2+	   activated	   Cl-­‐	  conductance	  (CaCC,	  TMEM16B)	  (White	  and	  Aylwin,	  1990;	  Huang	  et	  al.,	  2012);	  Gd3+	  (100	  uM),	  Maxi-­‐anion	  channel(Sabirov	  et	  al.,	  2001;	  Sabirov	  and	  Okada,	  2009;	  Fields	  and	  Ni,	  2010);	  Zinc	  (300	  μM),	  CLC-­‐2(Staley	  et	  al.,	  1996);	  carbenoxelone	  (CBX)	  (100	  μM),	   pannexins/connexins(Bruzzone	   et	   al.,	   2005;	   Thompson	   et	   al.,	   2008);	  bumetanide	  (100	  μM),	  cation	  chloride	  cotransporters	  (NKCC1	  and	  KCC2)	  (Payne	  et	  al.,	   2003;	   Glykys	   et	   al.,	   2014);	   DIDS	   (250	   μM),	   SLC4	   and	   SLC26	   family	   Cl-­‐/HCO3-­‐	  exchangers	   (Grichtchenko	   et	   al.,	   2001;	   Vincourt	   et	   al.,	   2003;	   Parker	   et	   al.,	   2008b;	  Svichar	   et	   al.,	   2009;	   Xu	   et	   al.,	   2011;	   Romero	   et	   al.,	   2013).	   Veratridine	   and	  NMDA	  application;	  a	  glass	  micropette	   (tip	  diameter	  ~2	  μm)	  was	  positioned	  10	  μm	  above	  the	  slice	  and	  25	  μm	  lateral	  to	  the	  centre	  of	  the	  imaging	  frame.	  	  The	  pipette	  was	  filled	  with	   the	   perfusion	   solution	   plus	   either	   veratridine	   or	   NMDA.	   	   A	  monometer	   was	  used	  to	  standardize	  the	  rate	  of	  drug	  application.	  	  	  	   105	  3.2.11	  Data	  Collection,	  Analysis	  and	  Statistics	  	  	   Translational	  movement	  was	  removed	  using	  Image	  J	  software.	  	  Fluorescence	  signals	  were	  defined	  as	  delta	  F/F	  (dF/F)	  =	  [((F1	  –	  B1)-­‐(F0	  –	  B0))/(F0	  –	  B0)],	  where	  F1	  and	  F0	  are	  fluorescence	  at	  a	  given	  time	  and	  the	  control	  period	  mean,	  respectively.	  B1	  and	   B0	   are	   the	   corresponding	   background	   fluorescence	   signals.	   	   Swelling	   of	  individual	  neurons	  in	  cortical	  slices	  was	  analyzed	  as	  (%)	  increase	  in	  cross	  sectional	  area	  relative	  to	  a	  mean	  baseline	  period.	  	  	  The	  cross	  sectional	  area	  of	  the	  neuron	  was	  calculated	   using	   the	   fluorescence	   boundary	   of	   the	   neuron	   soma	   stained	   with	  CoroNa.	   To	   estimate	   the	   tissue	   volume	   from	   the	   2	   dimensional	   images	   of	  hippocampal	  slices	  a	  line	  was	  drawn	  to	  measure	  the	  diameter	  and	  the	  volume	  was	  estimated	  based	  on	  the	  equation	  for	  volume	  of	  sphere:	  (4/3)πr2.	  	  Experimental	   values	   are	   the	   mean	   ±	   SEM;	   baseline	   equals	   100%;	   n	   is	   the	  number	   of	   experiments	   conducted,	   data	   from	   ≥3	   cells/experiment	   was	   combined	  and	  averaged	  to	  make	  1	  n	  so	   that	  equal	  weight	  was	  given	  to	  each	  experiment	  and	  not	  affected	  by	  the	  number	  of	  cells	  imaged/experiment.	  Statistical	  tests	  were	  either	  a	   two-­‐tailed	  Student's	   t	   test	  or	  an	  ANOVA	  with	  a	  Neumann-­‐Keuls	  post-­‐hoc	   test	   for	  comparison	   between	   multiple	   groups.	   P	   <	   0.05	   was	   accepted	   as	   statistically	  significant	  (*P	  <	  0.05,	  **P	  <	  0.01,	  ***P	  <	  0.001)	  	  	   106	  3.3	  Results	  	  3.3.1	  	  Neuronal	  Swelling	  is	  Caused	  by	  Prolonged	  Increases	  in	  Intracellular	  Na+	  and	  is	  Independent	  of	  Ca2+.	  	   Several	   pathways	   that	   increase	   intracellular	   Na+,	   many	   of	   which	   occur	   in	  parallel,	  can	  trigger	  cytotoxic	  brain	  edema.	  	  However,	  it	  has	  been	  known	  for	  almost	  30	   years	   that	   in	   addition	   to	   Na+	   entry,	   excitotoxic	   neuronal	   swelling	   ultimately	  requires	  Cl-­‐	  entry	  (Rothman,	  1985).	   	  We	  examined	  the	  basis	  for	  Cl-­‐	  entry	  following	  prolonged	   Na+	   influx	   via	   two	   different	   pathways	   to	   determine	   if	   Na+	   influx	   itself,	  independent	  of	  the	  route	  of	  entry	  would	  activate	  a	  common	  Cl-­‐	  entry	  pathway	  that	  could	  conceivably	  be	  targeted	  therapeutically.	  We	  first	  investigated	  whether	  increasing	  [Na+]i	  is	  itself	  capable	  of	  triggering	  a	  cascade	   leading	   to	   an	   increase	   in	   cell	   volume	   and	   secondly,	  whether	   this	   cascade	  also	  leads	  to	  rapid	  cell	  death.	  	  Two	  parallel	  and	  independent	  approaches	  were	  taken	  to	   increase	   [Na+]i	   by	   either	   applying	   veratridine,	   which	   removes	   inactivation	   of	  voltage-­‐gated	   sodium	   channels	   (VGSCs)	   (Strichartz	   et	   al.,	   1987)	   prolonging	   Na+	  entry,	   or	   by	   applying	   NMDA	   to	   activate	   NMDA	   receptors	   (NMDARs).	   	   NMDA	  activates	  a	  non-­‐selective	  cation	  conductance	   leading	   to	  entry	  of	  Na+	  and	  also	  Ca2+.	  	  Neuronal	  Na+	  entry	  was	  induced	  under	  conditions	  in	  which	  other	  voltage-­‐gated	  ion	  channels	  and	  ligand-­‐gated	  transmitter	  receptors	  were	  blocked	  by	  a	  combination	  of	  Cd2+	  (30	  μM),	  CNQX	  (20	  μM)	  and	  picrotoxin	  (100	  μM).	  	  Either	  veratridine	  or	  NMDA	  was	   rapidly	   applied	  by	  pressure	   ejection	   from	  a	  pipette	  positioned	  directly	   above	  the	   region	   of	   the	   brain	   slice	   that	   was	   imaged.	   To	   ensure	   the	   selectivity	   of	   either	  	   107	  approach	   veratridine	   was	   applied	   with	   d-­‐APV	   (100	   μM)	   to	   block	   NMDARs	   and	  NMDA	   was	   applied	   with	   TTX	   (1	   μM)	   to	   block	   VGSCs.	   	   Changes	   in	   [Na+]i	   were	  monitored	   using	   the	   fluorescent	   Na+	   indicator	   CoroNa-­‐Green	   (Meier	   et	   al.,	   2006)	  which	  preferentially	  stains	  hippocampal	  and	  cortical	  neurons	  in	  brain	  slices	  (Figure	  3.1	   a).	   	   Astrocytes	  which	   did	   not	   show	   any	   obvious	   volume	   changes	   under	   these	  experimental	  manipulations	  were	  visualized	  using	  SR101	  (Nimmerjahn	  et	  al.,	  2004)	  to	  provide	  landmarks	  to	  track	  during	  swelling	  of	  the	  tissue.	  	  Of	  note,	  the	  activation	  of	  either	  VGSCs	  by	  veratridine	  or	  NMDARs	  by	  NMDA	  consistently	   led	   to	  a	   significant	  increase	  in	  [Na+]i,	   followed	  after	  a	  delay	  of	  seconds,	  by	  an	  increase	  in	  neuronal	  cell	  volume	  (Figure	  3.1	  b-­‐d,	   j,	  k,	  Figure	  3.2a,	  b).	   	  We	  further	  compared	  the	  impact	  of	  Ca2+	   versus	   Na+	   entry	   through	   NMDARs	   on	   swelling	   by	   repeating	   experiments	   in	  Ca2+	  or	  Na+	   free	  extracellular	   solutions.	   	  The	   increase	   in	  cell	  volume	   from	  NMDAR	  activation	  was	  still	  observed	  in	  extracellular	  Ca2+	  free	  solution	  (cross	  sectional	  area	  increased	   to	   161.60	   ±	   10.55	   %	   of	   baseline).	   However, in the presence of low extracellular Na+ [Na+]ext and normal (2mM) [Ca2+]ext, swelling was completely absent and NMDAR activation actually resulted in a decrease in neuronal volume	  (Figure	  3.1j	  and	   Figures	   3.2c	   and	   	   d).	   	   Control	   experiments	   showed	   that	   neuronal	   [Na+]i	  	  increases	  and	  swelling	  induced	  by	  veratridine	  were	  blocked	  by	  the	  VGSC	  antagonist,	  TTX	   (Figure	   3.1j,	   k;	   p<	   0.001,	   two-­‐tailed	   student's	   t	   test)	   and	   those	   induced	   by	  NMDAR	  were	  blocked	  by	   the	  NMDAR	  antagonist,	  d-­‐APV	  (Figure	   3.1j,	   k;	  p<	  0.001,	  analysis	  of	  variance	  (ANOVA)).	  	   108	  	  	   109	  Figure	   3.1:	   	   Neuronal	   swelling	   is	   caused	   by	   prolonged	   increases	   in	  intracellular	  Na+	  and	  is	  independent	  of	  Ca2+.	  	  	  	  (a)	  CoroNa	  Green	   (Na+	  indicator)	   loaded	  neurons	  versus	  SR101	   stained	  astrocytes	  (red)	   in	   a	   hippocampal	   brain	   slice	   imaged	   using	   two-­‐photon	   laser	   scanning	  microscopy.	  (b-­‐d)	  Cortical	  neurons	  treated	  with	  veratridine	  (50	  µM)	  show	  increase	  in	   [Na+]i	  followed	   by	   swelling	   (increase	   in	   cross	   sectional	   area),	   astrocytes	   do	   not	  swell.	  (e,	  f)	  CoroNa	  FLIM	  measurements	  of	  [Na+]i	  as	  neurons	  swell	  reveals	  true	  time	  course	   and	   magnitude	   of	   Na+	   signals	   that	   are	   independent	   of	   dye	   concentration	  (n=4).	   (g-­‐i)	  Calibration	  of	   FLIM	  measurements	  of	  neuronal	   [Na+]i	  with	  CoroNa.	   (g)	  Decay	  of	  CoroNa	  fluorescence	  changes	  in	  salt	  solutions	  with	  varying	  [Na+].	  (h)	  Dual	  (simultaneous)	  whole	  cell	  patch	  clamping	  of	  2	  neurons	  dialyzed	  with	  high	  (109	  mM)	  and	   low	   (9	   mM)	   [Na+]	   show	   distinct	   separation	   of	   lifetimes.	   	   (i)	   Calibration	   of	  CoroNa	  lifetimes	  measured	  in	  soma	  of	  neurons	  dialyzed	  with	  different	  [Na+]	  shows	  that	  the	  [Na+]i	  	  can	  be	  predicted	  from	  τfast.	  (j	  and	  k)	  	  Quantified	  data	  shows	  neuronal	  swelling	  is	  triggered	  by	  sodium	  influx	  via	  independent	  pathways.	  	  NMDAR	  swelling	  was	  dependent	  on	  Na+	   influx	  and	   independent	  of	  Ca2+.	  Control	  confirms	  Na+	  signal	  and	   swelling	   caused	   by	   veratridine	   and	   NMDA	   was	   via	   VGSCs	   and	   NMDARs	  respectively,	  as	  they	  were	  blocked	  by	  antagonists,	  TTX	  (1	  µM)	  and	  d-­‐APV	  (100	  µM).	  All	   experiments	  were	  done	   in	   the	   presence	   of	   30	  mM	  Cd2+,	   20	   µM	  CNQX,	   100	  µM	  picrotoxin.	   	   Additionally,	   neurons	   were	   pretreated	   with	   100	   µM	   d-­‐APV	   (NMDAR	  antagonist)	  for	  veratridine	  experiments	  and	  1	  µM	  TTX	  (VGSC	  antagonist)	  for	  NMDA	  experiments	   to	  confirm	  pathways	  were	   independent.	   	  Scale	   in	   (b),	  20	  µm;	  scale	   in	  (h),	  15	  µm.	  VER,	  veratridine;	  x-­‐sectional,	  cross	  sectional;	  VGSC,	  voltage	  gated	  sodium	  channel;	  SR101,	   sulforhodamine	  101.	  Control	  values	   in	  panel	   (j)	  and	  panel	   (k)	  are	  also	   re-­‐plotted	   in	   Fig.	   3.5	   and	   Fig.	   3.6	   Error	   bars	   and	   shaded	   region	   above	   and	  below	  the	  mean	  represent	  SEM.	  	  	  	   110	  	  Figure 3.2: NMDAR activation triggers neuronal swelling that requires Na+ influx, but that is independent of Ca2+ influx.   (a and b) Na+ influx triggers an increase in neuronal volume, measured as the cross sectional area in the absence of extracellular Ca2+ (0 mM Ca2+, 2 mM EGTA) (n=5).  (c and d) Iso-osmotic replacement of extracellular Na+ with NMDG (from 152 mM to 26 mM), to reduce Na+ entry through NMDARs prevents neurons from swelling and causes them to shrink (86.7% of baseline, p < 0.05) (n=4). Scale bars, 15 µm (b and d). Shaded area above and below mean represent SEM.  	   111	  Although an increase in Na+ preceding swelling was consistently observed, the magnitude and duration of CoroNa fluorescence signals were distorted during cellular swelling due to dye dilution which also reduced fluorescence intensity of the inert dye, Calcein red-AM (Figure 3.2). In order to define the true magnitude and time course of the [Na+]i increases, we developed a method to record real-time calibrated measurements of [Na+]i using two photon fluorescence lifetime imaging (FLIM) which was independent of changes in dye concentrations.  When lifetime measurements of CoroNa were first tested in iso-osmotic salt solutions the time constant of decay (τ) increased with increasing [Na+] (Figure 3.1g).  However, as the local environment can affect lifetime measurements of dyes (Berezin and Achilefu, 2010) calibrations of CoroNa lifetimes were obtained within the cytoplasm of neurons by whole cell voltage-clamping of neurons and dialysis with different [Na+] concentrations.  CoroNa lifetimes were best fit using a biexponential decay (Figure 3.3) with a short lifetime (τ fast) predictive of [Na+]i (Figure 3.1h and 3.1i).  FLIM of CoroNa loaded neurons revealed that [Na+]i increased to approximately 94.46 ± 2.14 mM (calibrated value) throughout veratridine application and gradually recovered after washout (Figure 1E, 1F) 	   112	  	  Figure 3.3: Dye dilution results in decreases fluorescence intensity as neurons swell. Left: example images of neurons loaded with CoroNa and Calcein red that were exposed to veratridine.  Right: time course of CoroNa and Calcein Red signals in neurons exposed to veratridine.  As neurons begin to swell Calcein Red signal decreases. Scale bar, 10 µm.      	   113	   Figure 3.4: Biexponetial decay of CoroNa fluorescence indicates multiple fluorescence components. Example of single and biexponential fits from a pixel within the cytosol of a cortical neuron loaded with CoroNa-AM. Large deviations in χ2 from unity (i.e. χ2 >1.5) typically indicates multiexponential decay resulting from multiple fluorescence components (Berezin and Achilefu, 2010).	  	  	   114	  3.3.2	  Na+	   Influx	   is	  Correlated	  with	  a	  Secondary	  Cl-­‐	   Influx	  that	   is	  Required	   for	  Neuronal	  Swelling.	  	   Since cytoplasmic impermeant anions make up the bulk of the intracellular anionic milieu, changes in  [Cl-]i  must be met by an accompanying influx of water, possibly via transporters (Zeuthen, 2010), in attempt to achieve Gibbs-Donnan equilibrium (Glykys et al., 2014). We therefore examined whether prolonged [Na+]i increases were associated with a secondary influx of Cl-, and further whether Cl- entry was ultimately required for neuronal swelling.  Using two-photon FLIM of the Cl- sensitive dye MQAE (Verkman et al., 1989; Ferrini et al., 2013) we observed that [Cl-]i increased in neurons (indicated by a decrease in the fluorescence lifetime) when Na+ influx was triggered by veratridine application (Figures 3.5a and 3.5b).  This Cl- influx was independent of entry via GABAARs as all experiments were performed in the presence of the ligand-gated Cl- channel antagonist, picrotoxin (100 µM).   Whether neuronal Na+ and subsequent Cl- influx were sufficient to increase tissue volume were next investigated by imaging hippocampal/cortical brain slices at low magnification. Application of veratridine triggered dramatic swelling of brain slices that was reduced but still substantial even when the number of Na+, Ca2+ and Cl- entry pathways were reduced by blockade of glutamate gated AMPARs and NMDARs, voltage gated Ca2+ channels (VGCCs), and GABA activated Cl- channels with a cocktail of blockers (20 µM CNQX, 100 µM d-APV, 30 µM Cd2+, and 100 µM picrotoxin) (Figures 3.5c, 3.5d). In contrast, blocking all Cl- influx pathways by reducing [Cl-]ext with iso-osmotic replacement of NaCl for Na-gluconate in the extracellular solution dramatically reduced the magnitude of the volume increase of brain slices (Figure 3.5d; p < 0.001, 	   115	  ANOVA). These results suggest that even when fast ionotropic glutamate and GABA activated receptors are blocked, increased neuronal [Na+]i leads to cytotoxic edema of brain tissue that is dependent on Cl- influx.   We next tested whether reducing the concentration of extracellular Cl- ([Cl-]ext) also prevented Na+ induced swelling of individual neurons. Indeed, reducing [Cl-]ext reduced the swelling of neurons visualized with CoroNa fluorescence (Figures 3.5e, 3.5f; p < 0.001, ANOVA), without affecting the [Na+]i  signal (Figure 3.5h; p > 0.05, two-tailed student's t test).  As it has been previously reported that GABAAR mediated Cl- influx can contribute to neuronal swelling in cell culture (Hasbani et al., 1998), and to swelling following oxygen glucose deprivation in situ (Allen et al., 2004), the contribution of GABAAR Cl- influx to neuronal swelling in our experimental conditions was examined.. Consistent with previous reports, pre-application of the GABAAR antagonist picrotoxin slightly but significantly reduced the magnitude of neuronal swelling (from 161.7% to 146.9%; Figure 3.5f; p < 0.05, ANOVA), however, the majority of the volume increase persisted in picrotoxin suggesting that the cause of swelling was dominated by Cl- influx via an as yet unidentified mechanism.  NMDA-induced swelling was also blocked by low [Cl-]ext (iso-osmotic replacement of NaCl for Na-isethionate) (Figure 3.5j; p < 0.05, two-tailed student's t test).  Together, these data indicate that neuronal swelling requires Cl- influx through a mechanism that is triggered by an increase in [Na+]i and that Na+ entry alone is not sufficient to swell neurons.   	  	   116	  	  	   117	  Figure	  3.5:	  Na+	  influx	  is	  correlated	  with	  a	  secondary	  Cl-­‐	  influx	  that	  is	  required	  for	  neuronal	  swelling	  and	  causes	  cell	  death.	  	  (a	   and	   b)	   FLIM	  of	   Cl-­‐	  sensitive	   dye,	  MQAE,	   shows	   that	   Cl-­‐	  influx	   is	   correlated	  with	  increases	   in	   intracellular	   [Na+]	   (n=5).	   (c	   and	   d)	   Swelling	   of	   neurons	   causes	   an	  increase	  in	  brain	  tissue	  volume	  shown	  by	  changes	  in	  volume	  of	  a	  hippocampal	  brain	  slice.	   (d)	   Cocktail	   of	   fast	   glutamate	   receptor,	   GABA	   receptor	   and	   VGCC	   blockers	  slightly	   reduce	   tissue	   swelling	   (p<0.01)	   but	   significant	   Cl-­‐	  dependent	   swelling	   still	  occurs	  (p	  <	  0.01)	  indicating	  that	  swelling	  is	  dominated	  by	  other	  mechanisms.	  (d-­‐g)	  Neuronal	  swelling	   is	  prevented	  by	  reducing	  extracellular	  Cl-­‐	  (10.5	  mM)	  and	  is	  only	  partially	  inhibited	  by	  blocking	  GABAARs.	  (h)	  Positive	  control	  shows	  veratridine	  and	  NMDAR	  Na+	  signals	  were	  unaffected	  by	   low	  Cl-­‐	  solution.	  (i)	  Neuronal	  Na+	   influx	  via	  VGSCs	   causes	   cell	   death	   that	   is	   Cl-­‐	   dependent	   as	   measured	   by	   LDH	   release	   (iso-­‐osmotic	   replacement	   of	   Cl-­‐	  with	   gluconate).	   (j)	   Neuronal	   Na+	   influx	   via	   NMDARs	  causes	   cell	   death	   that	   is	   Cl-­‐	   dependent	   and	   Ca2+	   independent	   (iso-­‐osmotic	  replacement	  of	  Cl-­‐	  with	  isethionate).	  Scale	  bars:	  (a),	  10	  µm;	  (c),	  1.0	  mm;	  (e),	  15	  µm.	  VER,	  veratridine;	  	  VGCC,	  voltage	  gated	  calcium	  channel;	  VGSC,	  voltage	  gated	  sodium	  channel.	  For	  experiments	   in	  panels	  (a,b,e	  and	  g-­‐j)	  solutions	  contained	  blockers:	  30	  µM	  Cd2+,	  20	  µM	  CNQX,	  100	  µM	  picrotoxin,	  plus	  either	  100	  µM	  d-­‐APV	  for	  veratridine	  experiments	   or	   1	   µM	  TTX	   for	  NMDA	   experiments.	   n	   values	   in	   (f);	   blockers	   (n=5),	  +picrotoxin	   (n=13),	   low	  Cl-­‐	   (n=5).	   Error	  bars	   and	   shaded	   region	   above	   and	  below	  the	  mean	  represent	  SEM.	  	  	   118	  3.3.3	  Na+	  and	  Cl-­‐	  Dependent	  Neuronal	  Swelling	  Causes	  Death	  	   Aberrant	   calcium	   influx	   via	   NMDARs	   can	   lead	   to	   mitochondrial	  depolarization	   and	   cell	   death,	   however,	   Cl-­‐	   removal	   also	   reduces	   ischemia	   and	  glutamate	  evoked	  early	  neuronal	  death	  in	  cell	  culture	  (Rothman,	  1985;	  Choi,	  1987;	  Goldberg	   and	   Choi,	   1993),	   suggesting	   the	   existence	   of	   two	   independent	   pathways	  ultimately	   leading	   to	   cell	   death.	   The	   impact	   of	   the	   [Na+]i-­‐triggered	   Cl-­‐	   entry	   and	  neuronal	   swelling	   on	   cell	   viability	   was	   further	   investigated	   using	   LDH	   release	   as	  measure	  of	   cell	  death	   (e.g.(Kajta	   et	   al.,	   2005)).	   	  Even	   in	   the	   combined	  presence	  of	  CNQX,	   picrotoxin	   and	   Cd2+	   to	   block	   fast	   AMPA/KA	   receptors,	   GABA-­‐activated	   Cl-­‐	  channels	  and	  VGCCs	  respectively,	  application	  (15	  min)	  of	  either	  veratridine	  (50	  μM)	  or	  NMDA	  (100	  μM,	  in	  artificial	  cerebrospinal	  fluid	  (ACSF)	  containing	  0mM	  Ca2+	  and	  2	   mM	   EGTA)	   caused	   a	   rapid	   and	   significant	   increase	   in	   LDH	   release	   indicating	  neurons	  were	  dying	  after	  90	  min	  (Figure	  3.5i	  and	  3.5j;	  p	  <	  0.01,	  ANOVA).	  Both	  the	  NMDA-­‐induced	   and	   veratridine-­‐induced	   neuronal	   death,	   as	   indicated	   by	   LDH	  release,	  were	  abolished	  by	  reducing	  [Cl-­‐]ext	  (Figures	  3.5i	  and	  3.5j;	  p	  <	  0.01,	  ANOVA).	  	  This	   suggests	   that	   Na+-­‐induced	   Cl-­‐	  influx	   and	   subsequent	   swelling	   results	   in	   Ca2+-­‐independent	  cell	  death.	  	  	  	   119	  3.3.4	   Neuronal	   Swelling	   and	   Death	   Show	   the	   Pharmacological	   Profile	   of	   a	  HCO3-­‐	  /	  Cl-­‐	  Exchanger	  	   There	  are	  several	  candidates	  for	  the	  transmembrane	  influx	  of	  Cl-­‐	  in	  neurons	  that	  can	  be	  distinguished	  based	  on	  their	  sensitivity	  to	  different	  antagonists	  (Jentsch	  et	   al.,	   2002;	   Alvarez-­‐Leefmans	   and	   Delpire,	   2009;	   Verkman	   and	   Galietta,	   2009)	  (Table	   3.2).	   	  We	   hypothesized	   that	   by	   identifying	   and	   blocking	   the	   source	   of	   Cl-­‐	  entry	  that	  was	  triggered	  by	  Na+	  entry	  both	  the	  Na+	   induced	  neuronal	  swelling	  and	  corresponding	   cell	   death	   could	   be	   prevented.	   	   As	   a	   first	   step,	   pharmacological	  analyses	   using	   the	   imaging	   assay	   of	   swelling	   of	   neurons	   in	   brain	   slices	   were	  undertaken	  in	  order	  to	  screen	  for	  the	  possible	  involvement	  of	  different	  Cl-­‐	  channels	  and	   transporters.	   	   In	   separate	   experiments	   the	   following	   blockers	  were	   tested	   as	  described	   in	   Table	   3.2;	   NPPB	   (200	   μM)	   to	   block	   the	   volume-­‐regulated	   anion	  channel	   (VRAC,	   VSOR),	   zinc	   (300	   μM)	   to	   block	   CLC-­‐2,	   Gd3+	  (100	   μM)	   to	   block	   the	  Maxi-­‐anion	   channel,	   niflumic	   acid	   (NFA)	   (200	   μM)	   to	   block	   the	   Ca2+	   activated	   Cl-­‐	  conductance	   (CaCC,	   bestrophin),	   carbenoxelone	   (CBX)	   (100	   μM)	   to	   block	  pannexins/connexins,	  bumetanide	  (100	  μM)	  to	  block	  cation	  chloride	  cotransporters	  (NKCC1	   and	   KCC2),	   and	   DIDS	   (250	   μM)	   to	   block	   Cl-­‐/HCO3-­‐	   exchangers.	   All	  antagonists	  were	  both	  bath	  applied	  and	  present	  in	  the	  puffing	  pipette	  used	  to	  apply	  either	   NMDA	   or	   veratridine.	   	   Of	   note,	   of	   the	   various	   Cl-­‐	   channel	   and	   transporter	  blockers	   examined	   only	   DIDS	   reduced	   the	   swelling	   induced	   by	   increased	   [Na+]i	  (Figure	  3.6a;	  p	  <	  0.05	  compared	  to	  all	  other	  antagonists,	  ANOVA).	  The	  small	  volume	  change	  in	  the	  presence	  of	  DIDS	  was	  not	  significantly	  different	  from	  those	  observed	  in	  low	  Cl-­‐	  extracellular	  solution	  (Figure	  3.6a;	  p	  >	  0.05,	  ANOVA).	  A	  substantial	  [Na+]i	  	   120	  increase	   was	   still	   observed	   in	   DIDS	   indicating	   that	   Na+	   entry	   was	   not	   affected	  (Figure	   3.6b).	   	  This	  pattern	  of	  block	  by	  DIDS	  but	  no	  effect	  of	   the	  numerous	  other	  blockers	   suggested	   that	   a	   Cl-­‐/HCO3-­‐	   exchanger	   was	   the	   most	   likely	   source	   of	   Cl-­‐	  entry.	  	  Although	  DIDS	  also	  blocks	  VRAC,	  which	  has	  been	  implicated	  to	  play	  a	  role	  in	  excitotoxic	   cell	  death	   in	  neuronal	   cell	   culture	   (Inoue	  and	  Okada,	  2007),	  under	  our	  conditions	   we	   observed	   no	   protection	   of	   either	   cell	   volume	   or	   cell	   death	   in	   the	  presence	   of	   the	   potent	   VRAC	   blocker,	   NPPB.	   	   DIDS	   also	   blocked	   NMDA-­‐evoked	  neuronal	  swelling	  in	  a	  dose-­‐dependent	  manner	  (Figure	  3.6c).	  	  	  	   121	  Table 3.2: Pharmacology of antagonists that inhibit chloride channels and chloride transporters  	  VRAC, volume regulated anion channel; CaCC, calcium activated chloride channel; NPPB, 5-nitro-2-(3-phenylpropylamino) benzoic acid; NFA, niflumic acid; CBX, carbenoxolone; DIDS, 4,4'-Diisothiocyanato-2,2'-stilbenedisulfonic acid. *Voltage dependent block References: 1.(Inoue and Okada, 2007) 2.(Inoue et al., 2005) 3.(Qiu et al., 2014) 4.(Huang et al., 2012) 5. (White and Aylwin, 1990) 6.(Sabirov et al., 2001) 7. (Sabirov and Okada, 2009) 8. (Fields and Ni, 2010) 9. (Bruzzone et al., 2005) 10. (Thompson et al., 2008) 11. (Staley et al., 1996) 12. (Payne et al., 2003) 13. (Romero et al., 2013) 14.(Svichar et al., 2009) 15.(Grichtchenko et al., 2001) 16.(Parker et al., 2008b) 17.(Xu et al., 2011) 18.(Vincourt et al., 2003) 	  	  	  	   	  	   	  	   Antagonist	  and	  concentration	   	  	   	  	  Channel/Transporter	   NPPB	  	  (200	  μM)	   NFA	  	  (200	  	  μM)	   Gd3+	  	  (100	  	  μM)	   CBX	  	  (100	  	  μM)	   Zinc	  	  (300	  	  μM)	   Bumetanide	  	  (100	  	  μM)	   DIDS	  	  (250	  	  μM)	  VRAC	   40-­‐100	  μM1,2	   	  	   	  	   	  	   	  	   	  	   100-­‐400	  μM*2,3	  CaCC	   100	  μM4	   50-­‐100	  μM4,5	   	  	   	  	   	  	   	  	   	  	  Maxi-­‐Anion	  Channel	   100	  μM6-­‐8	   	  	   30-­‐50	  μM6-­‐8	   	  	   	  	   	  	   100	  μM7	  Pannexins,	  connexins	   	  	   	  	   	  	   100	  	  μM9-­‐10	   	  	   	  	   	  	  CLC-­‐2	   	  	   	  	   	  	   	  	   100	  	  μM11	   	  	   	  	  NKCC1	   	  	   	  	   	  	   	  	   	  	   Ki	  ~	  0.1	  uM12	   	  	  KCC2	   	  	   	  	   	  	   	  	   	  	   Ki	  ~	  25-­‐50	  uM12	   	  	  SLC4-­‐A3,	  -­‐A8,	  -­‐A10	   	  	   	  	   	  	   	  	   	  	   	  	   0.1-­‐0.5	  mM13-­‐16	  SLC26A11	   	  	   	  	   	  	   	  	   	  	   	  	   0.5-­‐1	  mM17,18	  	   122	   	   123	  Figure	  3.6:	  Neuronal	  swelling	  shows	   the	  pharmacological	  profile	  of	  a	  HCO3-­‐	  /	  Cl-­‐	  exchanger.	  	  (a)	  Veratridine	  induced	  neuronal	  swelling	  was	  blocked	  by	  the	  HCO3-­‐	  /Cl-­‐	  exchanger	  inhibitor,	   DIDS	   (250	   µM)	   but	   not	   by	   blockers	   of	   several	   other	   Cl-­‐	   channels	   or	  transporters	   (see	   Supplementary	   Table	   1).	   (b)	   Positive	   control	   shows	   veratridine	  and	  NMDA	  induced	  Na+	  signal	  in	  the	  presence	  of	  DIDS.	  	  (c)	  	  NMDA	  induced	  neuronal	  swelling	  was	  blocked	  by	  DIDS	  in	  a	  dose	  dependent	  manner;	  control	  (n=5),	  250	  µM	  (n=4),	  500	  µM(n=5),	  1mM	  (n=5).	  	  	  	  (d)	  Reduction	  in	  intracellular	  HCO3-­‐,	  (HCO3-­‐	  free	  extracellular	  solution	  +	  100	  µM	  AZM)	  reduced	  the	  magnitude	  of	  neuronal	  swelling,	  p	  <	  0.001,	   two-­‐tailed	  student's	   t	   test.	   (e)	  Positive	  control	   shows	  veratridine	   induced	  Na+	  signals	  in	  (HCO3-­‐	  free	  +	  AZM).	  All	  solutions	  contained	  blockers:	  30	  mM	  Cd2+,	  20	  µM	  CNQX,	  100	  µM	  picrotoxin,	  plus	  either	  100	  µM	  d-­‐APV	  for	  veratridine	  experiments	  or	   1	   µM	  TTX	   for	  NMDA	  experiments.	   VER,	   veratridine;	  AZM,	   acetazolamide.	   Error	  bars	  and	  shaded	  region	  above	  and	  below	  the	  mean	  represent	  SEM.	  	  	   124	  As	   it	   was	   observed	   that	   extracellular	   Cl-­‐	   was	   required	   for	   both	   neuronal	  swelling	  and	  the	  subsequent	  cell	  death	  and	  that	  DIDS	  prevented	  neuronal	  swelling,	  we	  predicted	  that	  DIDS	  would	  block	  the	  Cl-­‐	  dependent	  cell	  death	  pathway	  without	  affecting	   the	  Ca2+-­‐dependent	  death.	   	  DIDS	  was	   initially tested for its effectiveness	   in	  preventing	  the	  swelling	  induced,	  Cl-­‐	  dependent	  cell	  death	  as	  measured	  by	  LDH	  efflux	  in	   brain	   slices	   exposed	   to	   veratridine.	   	   Indeed,	   DIDS	   prevented	   cell	   death	   from	  veratridine	   induced	   Na+	   influx	   and	   swelling	   (Figure	   3.7a;	   p	   <	   0.005,	   ANOVA),	  whereas	   the	   VRAC	   blocker	   NPPB	   had	   no	   effect.	   DIDS	  was further examined	   on	   the	  NMDA	  Cl-­‐-­‐dependent,	  Ca2+-­‐independent	  cell	  death	  pathway	  and	  on	  the	  NMDA	  Ca2+-­‐dependent	  cell	  death	  pathway.	  	  As	  predicted,	  DIDS	  blocked	  the	  cell	  death	  by	  NMDA	  in	   Ca2+	   free	   extracellular	   solution	   (Figure	   3.7b;	   p	   <	   0.005,	   ANOVA).	   	   If	   however,	  NMDA	  was	   applied	   in	   the	   presence	   of	   extracellular	   Ca2+	   but	   reduced	   extracellular	  Na+,	  cell	  death	  still	  occurred	  (Figure	  3.7c;	  p	  <	  0.005,	  ANOVA)	  but	  was	  not	  blocked	  by	   DIDS	   (Figure	   3.7c;	   p	   >	   0.05,	   ANOVA).	   	   These	   results	   suggest	   that	   two	  independent	  cell	  death	  pathways	  co-­‐exist	   that	  can	  be	  distinguished	  based	  on	  their	  ionic	   basis;	   one	   that	   involves	   swelling,	   requires	   Na+	   and	   Cl-­‐	   influx,	   is	   Ca2+-­‐independent	  and	  is	  blocked	  by	  DIDS,	  and	  one	  that	  is	  triggered	  by	  Ca2+	  influx,	  but	  that	  is	  not	  blocked	  by	  DIDS.	  	   125	     	  Figure	  3.7:	  	  DIDS	  blocks	  Na+	  and	  Cl-­‐	  dependent,	  Ca2+	  independent	  cell	  death.	  	  	  (a)	  LDH	  release	  measurements	  show	  Na+	  and	  Cl-­‐	  dependent	  cell	  death	  triggered	  by	  veratridine	  was	  blocked	  by	  the	  HCO3-­‐/Cl-­‐	  exchanger	  antagonist,	  DIDS	  but	  not	  by	  the	  VRAC	  blocker	  NPPB.	   	   	   (b)	   	  NMDAR	  Na+	  influx	   triggers	   cell	   death	   in	   the	  absence	  of	  extracellular	  Ca2+	  that	  is	  blocked	  by	  DIDS	  but	  not	  NPPB.	  	  (c)	  NMDAR	  Ca2+	  influx	  also	  triggers	  cell	  death	  that	  is	  not	  blocked	  by	  DIDS,	  indicating	  separate	  pathways.	  	  	   126	  3.3.5	   Identification	   of	   SLC26A11	   as	   the	   Predominant	   Cl-­‐	   Influx	   Pathway	  Underlying	  Na+	  Dependent	  Cytotoxic	  Neuronal	  Swelling	  	  Our	  data	   indicates	   that	  Na+	  entry	   into	  neurons	   is	   linked	  to	  a	  DIDS-­‐sensitive	  Cl-­‐	   influx	   pathway	   that	   is	   required	   for	   neuronal	   swelling	   and	  mediates	   cell	   death.	  	  Several	   DIDS-­‐sensitive	   candidates	   are	   expressed	   in	   central	   nervous	   system	   (CNS)	  neurons	  of	  which	  several	  act	  as	  Cl-­‐/HCO3-­‐	  exchangers	  and	  include	  the	  SLC4	  family	  of	  exchangers	   that	   are	   Cl-­‐,	   HCO3-­‐	   dependent	   (Alvarez-­‐Leefmans	   and	   Delpire,	   2009;	  Boron	  et	  al.,	  2009;	  Romero	  et	  al.,	  2013).	   	  Removing	  extracellular	  HCO3-­‐	  and	  adding	  the	   carbonic	   anhydrase	   inhibitor,	   acetazolamide,	   to	   reduce	   intracellular	   HCO3-­‐	  generation	   resulted	   in	   reduced	   swelling	   (Figure	   3.6d;	   p	   <	   0.001,	   two-­‐tailed	  student's	   t	   test)	   supporting	   the	   possibility	   that	   one	   or	   several	   of	   the	   Cl-­‐/HCO3-­‐	  exchangers	   contribute	   to	   Cl-­‐	   loading	   and	   swelling.	   	   The	   DIDS-­‐sensitive	   Cl-­‐,	   HCO3-­‐	  exchangers	   that	   are	   known	   to	   be	   expressed	   in	   the	   cortex	   and	   hippocampus	   are	  SLC4A3,	  SLC4A8	  and	  SLC4A10	  (Boron	  et	  al.,	  2009;	  Romero	  et	  al.,	  2013).	  In	  addition,	  SLC26A11	   was	   recently	   shown	   to	   be	   highly	   expressed	   in	   CNS	   cortical	   neurons	  (Rahmati	  et	  al.,	  2013).	  	  	  SLC26A11	  is	  a	  member	  of	  the	  sulfate	  transporter	  family	  that	  in	   different	   expression	   systems	   has	   been	   reported	   to	   act	   variously	   as	   a	   DIDS-­‐sensitive	  sulfate	  transporter,	  a	  DIDS-­‐sensitive	  exchanger	  for	  Cl-­‐,	  SO42-­‐,	  HCO3-­‐	  or	  H+-­‐Cl-­‐	  or	  as	  a	  Cl-­‐	  channel	  (Vincourt	  et	  al.,	  2003;	  Xu	  et	  al.,	  2011;	  Lee	  et	  al.,	  2012a;	  Rahmati	  et	  al.,	  2013).	  	  Utilizing	   qRT-­‐PCR,	   the	   expression	   of	   SLC4	   and	   SLC26	   family	  members	  was	  confirmed	   in	   both	   cortical	   and	   hippocampal	   brain	   tissue	   (Figure	   3.8).	   	   Based	   on	  	   127	  their	   combined	  pharmacological	   profile	   and	  expression	  profiles,	   SLC4-­‐A3,-­‐A8,-­‐A10	  and	   SLC26A11	   appeared	   to	   be	   the	   most	   promising	   candidates	   for	   the	   Cl-­‐	   entry	  pathway	   that	  causes	  neuronal	   swelling.	   	  We	  recently	   reported	   the	  development	  of	  an	  efficient	  lipid	  nanoparticle	  (LNP)-­‐mediated	  delivery	  system	  to	  introduce	  siRNAs	  against	  specific	  molecular	  targets	  into	  CNS	  neurons	  both	  in	  vivo	  and	  in	  vitro	  (Rungta	  et	   al.,	   2013)	   (described	   in	   chapter	   2).	   	   Individual	   siRNAs	   targeted	   against	   the	  different	   SLC	   candidate	   genes	   were	   encapsulated	   in	   LNPs	   and	   initially	   tested	   for	  their	  ability	  to	  attenuate	  expression	   in	  both	  primary	  neuron	  cultures	  and	  HEK	  cell	  expression	  system	  (Figure	  3.8).	  These	   in	  vitro-­‐validated	  siRNA	  LNPs	  against	   the	  4	  different	  SLC	  candidates	  or	  a	  control	  (luciferase)	  siRNA	  were	  subsequently	  injected	  intracranially	   into	   the	   rat	   somatosensory	   cortex.	   	   After	   allowing	   for	   5-­‐6	   days	   of	  recovery,	   neurons	   that	   had	   taken	   up	   DiI	   labeled	   LNPs	   were	   examined	   for	   Na+	  induced	  Cl-­‐-­‐dependent	  swelling	  in	  cortical	  slices.	  Knockdown	  of	  SLC4A-­‐3,	  -­‐8	  or	  -­‐10	  either	   separately	   (Figure	   3.9)	   or	   together	   had	   no	   effect	   on	   the	   magnitude	   of	  veratridine	   induced	   neuronal	   swelling	   compared	   to	   the	   control	   luciferase	   siRNA	  injected	  animals	  (Figures	  3.10a	  and	  3.10c;	  p	  >	  0.05,	  ANOVA).	  	  In	  striking	  contrast,	  targeted	  knockdown	  of	   SLC26A11	  with	   two	  different	   siRNAs	   significantly	   reduced	  the	   magnitude	   of	   the	   swelling	   in	   neurons	   as	   measured	   by	   increases	   in	   cell	   size	  (Figures	   3.10b,	   d;	   p	   <	   0.05,	   ANOVA	   was	   performed	   comparing	   results	   from	   all	  siRNA	  groups	  (luciferase,	  A3,	  A8,	  A10,	  A3+A8+A10,	  A11	  No.1	  and	  A11	  No.2)).	  	  These	  results	  indicate	  that	  the	  Cl-­‐	  influx	  that	  is	  required	  for	  neuronal	  swelling	  is	  mediated	  by	  SLC26A11.	  	   128	  	  	   129	  Figure 3.8: SLC26A and SLC4A gene families, siRNA-mediated knockdown and expression profiles.  (a) Protein sequence similarity tree of the SLC26A and SLC4A family members from mouse, rat and humans. (CaV = voltage-gated calcium channels CaV1.2, CaV2.1 and CaV3.1 from human). (b) Testing of LNP-packaged modified Dicer siRNA knockdown duplexes (for SLC4A3, SLC4A8, SLC4A10 and SLC26A11) in vitro. The data represent quantification using qPCR 72 hrs post-treatment of cultured rat cortical neurons except for unmodified SLC4A10 Dicer siRNA which was tested in HEK293 cells expressing the cloned rSLC4A10 target. Data are normalized to internal rGAPDH mRNA levels. (c) Expression profiles of the SLC4A and SLC26A family members in rat cortex and hippocampus as determined using qPCR. Data are normalized to internal β-actin mRNA levels. 	   130	    Figure 3.9:  siRNA-mediated knockdown of individual SLC4A family members does not alter the magnitude of neuronal swelling.  Transfection of neurons in vivo by intracranial injection of LNPs encapsulated with siRNA against SLC4A3, SLC4A8 or SLC4A10 does not significantly decrease neuronal swelling in brain slices following veratridine treatment compared to control (luciferase siRNA transfected neurons).  Error bars represent SEM. 	  	   131	  	  	   132	  Figure	   3.10:	   Cl-­‐	   influx	   via	   SLC26A11	   causes	   cytotoxic	   neuronal	   edema	  following	  increased	  [Na+]i.	  	  	  	  (a)	  In	  vivo	  knockdown	  of	  SLC4A3,	  A8,	  A10	  with	  LNP-­‐siRNAs	  results	  in	  no	  significant	  difference	  in	  the	  magnitude	  of	  neuronal	  swelling	  compared	  to	  a	  control	  (luciferase	  siRNA)	   in	   cortical	   brain	   slices	   imaged	   5	   days	   following	   the	   injection	   (p	   >	   0.05,	  ANOVA).	  (b)	  Two	  different	  siRNA	  constructs	  against	  SLC26A11	  result	  in	  a	  significant	  reduction	   in	   the	  magnitude	  of	   veratridine	   induced	  neuronal	   swelling	   compared	   to	  luciferase	  siRNA	  (p	  <	  0.05,	  ANOVA).	   (c	  and	  d)	  Example	   images	  of	  cortical	  neurons	  transfected	   with	   siRNA	   using	   lipid	   nanoparticle	   delivery	   shows	   SLC26A11	  knockdown	   results	   in	   protection	   from	  veratridine	   triggered	   swelling	   compared	   to	  neurons	   transfected	  with	   SLC4A8	   siRNA.	   	   DiI	   staining	   (red)	   shows	   cell	   uptake	   of	  LNP-­‐siRNA.	  	  Scale	  bar	  in	  (c)	  matches	  scale	  in	  (d).	  	  Luciferase	  controls	  are	  combined	  and	  plotted	   in	  both	  panel	   (a),	   (b)	  and	   in	  Figure	  3.9.	  For	  statistics	  on	  magnitude	  of	  swelling,	   ANOVA	   was	   performed	   comparing	   results	   from	   all	   siRNA	   groups	  (luciferase,	   SLC4A3,	   -­‐A8,	   -­‐A10,	   -­‐A3+A8+A10,	   SLC26A11	   No.1	   and	   No.2).	   	   Only	  SLC26A11	   No.1	   and	   No.	   2	   were	   significantly	   different	   from	   luciferase	   (control)	  siRNA,	  p	  <	  0.05.	  Error	  bars	  represent	  SEM.	  	  	   133	  3.4	  Discussion	  	  	   Our	   results	   demonstrate	   that	   prolonged	   sodium	   entry	   via	   either	   of	   two	  independent	  pathways	  (either	  VGSCs	  or	  NMDARs)	  converge	  to	  activate	  a	  Cl-­‐	  	  influx	  pathway	   via	   SLC26A11	   that	   is	   ultimately	   required	   for	   neuronal	   swelling	   and	  subsequent	  cell	  death.	  When	  extracellular	  Cl-­‐	  is	  removed	  both	  neuronal	  swelling	  and	  cell	   death	   are	   prevented.	   	   	   Other	   potential	   Cl-­‐	   channels	   and	   exchangers	   did	   not	  appear	   to	   contribute	   significantly	  under	   these	  conditions	  as	   their	  pharmacological	  inhibitors	   did	   not	   reduce	   neuronal	   swelling.	   Both	   pharmacological	   blockade	   with	  DIDS	  and	  removal	  of	  HCO3-­‐	   resulted	   in	  reduced	  neuronal	  swelling,	   indicating	  a	  Cl-­‐,	  HCO3-­‐	  exchanger(s)	  was	   likely	   involved.	   	  Utilizing	  LNP-­‐siRNA-­‐mediated	  depression	  of	  expression	  of	  several	  candidates	  showed	  that	  only	  reducing	  SLC26A11,	  recently	  described	   to	   be	   highly	   expressed	   in	   CNS	   neurons,	   was	   successful	   in	   reducing	  swelling.	  	  	   In	   mature	   pyramidal	   neurons	   of	   the	   cortex	   and	   hippocampus,	   resting	  membrane	  potential	   (Em)	   is	  set	  positive	  compared	   to	   the	  equilibrium	  potential	   for	  Cl-­‐	  (ECl-­‐)	  suggesting	  that	  Cl-­‐	  is	  not	  passively	  distributed	  across	  the	  plasma	  membrane	  (Alvarez-­‐Leefmans	  and	  Delpire,	  2009).	  	  Changing	  membrane	  potential	  also	  has	  little	  effect	   on	   intracellular	   chloride	   concentrations	   indicating	   that	   there	   is	   little	   Cl-­‐	  membrane	  permeability	  at	  rest	  (Thompson	  et	  al.,	  1988).	  	  As	  such,	  in	  order	  for	  [Cl-­‐]i	  	  to	  rapidly	  increase	  in	  neurons	  either	  a	  chloride	  transporter	  has	  to	  be	  activated	  or	  a	  transmembrane	  Cl-­‐	  channel	  must	  be	  opened.	   	  Membrane	  depolarization	  could	  also	  further	  contribute	  to	  Cl-­‐	  influx	  by	  increasing	  the	  driving	  force	  for	  Cl-­‐	  entry.	  	  	   134	  Using	  an	  siRNA	  knockdown	  approach,	  we	  were	  able	  to	  identify	  the	  molecular	  nature	  of	  the	  predominant	  Cl-­‐	  influx	  pathway	  that	  is	  activated	  following	  increases	  in	  [Na+]i	   and	   causes	   neuronal	   cytotoxic	   edema.	   	   Our	   study	   is	   the	   first	   report	   of	  SLC26A11	  as	  a	  functional	  Cl-­‐	  influx	  pathway	  in	  neurons.	  	  A	  recent	  study	  showed	  that	  SLC26A11	   protein	   is	   expressed	   in	   neurons	   throughout	   the	   brain	   and	   we	   would	  predict	   that	   similar	   mechanisms	   of	   swelling	   and	   neuronal	   death	   likely	   occur.	  	  SLC26A11,	  originally	  identified	  as	  a	  sulfate	  transporter	  has	  been	  shown	  to	  operate	  in	   several	  modes,	   including	   an	   exchanger	   for	   Cl-­‐,	   SO42-­‐,	   HCO3-­‐	   or	  H+-­‐Cl-­‐	   or	   as	   a	   Cl-­‐	  channel,	   depending	   on	   the	   tissue	   type	   and	   the	   expression	   system	   (Vincourt	   et	   al.,	  2003;	   Xu	   et	   al.,	   2011;	   Rahmati	   et	   al.,	   2013).	   	   The	   precise	  mechanism	   linking	   Na+	  influx	   and	  SLC26A11	  mediated	  Cl-­‐	  influx	   remains	   to	  be	  determined.	   	   	   If	   SLC26A11	  functions	  as	  a	  constitutively	  active	  Cl-­‐	  channel	  as	  proposed	  (Rahmati	  et	  al.,	  2013),	  it	  would	  be	  constantly	  loading	  the	  cell	  with	  Cl-­‐	  	  at	  Em	  due	  to	  the	  increased	  driving	  force	  for	   Cl-­‐	  entry.	   	   Therefore,	   there	   must	   either	   be	   a	   change	   in	   the	   activation	   state	   of	  SLC26A11	   or	   alternatively	   SLC26A11	  may	   be	   trafficked	   to	   the	   plasma	  membrane	  following	  increases	  in	  Na+.	  	  	  Electrogenic	  transport	  of	  Cl-­‐	  remains	  a	  viable	  possibility,	  however	   depolarization	   of	   cortical	   neurons	   with	   high	   K+	   solution	   (40mM),	   is	   not	  sufficient	  to	  cause	  neuronal	  swelling	  unless	  spreading	  depression	  triggers	  glutamate	  release	  which	  causes	  Na+	  influx	  (Zhou	  et	  al.,	  2010;	  Zhou	  et	  al.,	  2013).	  	  	  Several	  questions	  arise	  as	  to	  the	  specific	  conditions	  and	  times	  that	  SLC26A11	  may	  modulate	  local	  and	  global	  Cl-­‐	  concentration.	  	  	  Aberrant,	  chloride	  homeostasis	  is	  central	   to	   several	   neurological	   diseases,	   and	   it	   would	   therefore	   be	   interesting	   to	  examine	   whether	   SLC26A11	   expression	   or	   localization	   changes	   under	   such	  	   135	  conditions.	   	   	   Epileptic	   seizures	   are	   commonly	   observed	   in	   patients	   following	  ischemia	  and	  TBI	  with	  these	  patients	  responding	  poorly	  to	  anticonvulsants	  that	  act	  on	  GABAA	  receptors	  (Young	  et	  al.,	  1990),	  presumably	  due	  to	  increased	  [Cl-­‐]i	  leading	  to	   a	   depolarizing	   shift	   in	  EGABA	  (Cohen	   et	   al.,	   2002;	  Miles	   et	   al.,	   2012).	   	   If	   blocking	  SLC26A11	   reduces	   the	   increases	   in	   Cl-­‐	   that	   occur	   during	   pathologies	   that	   are	  associated	   with	   cytotoxic	   edema,	   it	   may	   be	   possible	   to	   maintain	   the	   direction	   of	  hyperpolarizing	   GABAAR	   currents,	   and	   reduce	   the	   generation	   of	   post	   traumatic	  seizures.	  	  	  While	  we	  find	  that	  SLC26A11	  represents	  the	  predominant	  Cl-­‐	  influx	  pathway	  causing	   neuronal	   swelling	   following	   increased	   [Na+]i,	   pyramidal	   neurons	   are	  endowed	  with	  multiple	  Cl-­‐	  influx	  pathways	  that	  may	  further	  contribute	  to	  Cl-­‐	  influx	  and	   cause	   swelling	   under	   other	   conditions.	   	   Following	   excitotoxic	   insults	   such	   as	  ischemia,	   the	   extracellular	   GABA	   concentration	   increases	   due	   to	   a	   combination	   of	  vesicular	   release	   and	   reversal	   of	   GABA	   transporters	   (Allen	   et	   al.,	   2004),	   possibly	  causing	  Cl-­‐	  influx.	  	  Our	  data	  suggest	  that	  although	  GABAARs	  do	  in	  fact	  contribute	  to	  a	  small	   portion	   of	   neuronal	   swelling	   observed	   in	   our	   experiments	   (Figure	  3.5f),	   the	  majority	  of	   the	  Cl-­‐	  influx	   is	  mediated	  by	   another	  mechanism.	   	  These	   results	   are	   in	  agreement	   with	   other	   studies	   showing	   a	   contribution	   of	   GABAARs	   to	   neuronal	  swelling	  in	  cell	  culture	  (Hasbani	  et	  al.,	  1998).	  	  In	  brain	  slices	  it	  has	  also	  been	  shown	  that	   blocking	  GABAARs	   leads	   to	   a	   small	   reduction	   in	   the	   initial	   rapid	   increase	   (<1	  min)	   of	   the	   intrinsic	   optical	   signal	   (a	   marker	   of	   cell	   swelling)	   following	   OGD.	  	  However,	   blocking	  GABAARs	  has	  no	   effect	   on	   final	   intrinsic	   optical	   signal	   increase	  (>2	  min)	  (Allen	  et	  al.,	  2004).	  	  Additionally,	  in	  this	  study	  it	  was	  observed	  that	  calcium	  	   136	  influx	   caused	  GABAARs	   to	   become	   completely	   inactivated	  within	   2	  minutes	   of	   the	  anoxic	  depolarization.	  	  These	  results	  suggest	  that	  although	  GABAARs	  can	  contribute	  to	   neuronal	   swelling	   via	   a	   transient	   chloride	   influx,	   the	   majority	   of	   the	   Cl-­‐	   influx	  observed	   during	   ischemia	   and	   other	   excitotoxic	   insults	   occurs	   via	   a	   distinct	  mechanism.	  	  In	  another	  study,	  Cl-­‐	  imaging	  in	  pyramidal	  neurons	  with	  the	  genetically	  encoded	   Cl-­‐	   indicator	   Clomeleon	   showed	   no	   effect	   of	   blocking	   GABAARs	   on	   the	  magnitude	   of	   the	   chloride	   increase	   during	   ischemia	   (Pond	   et	   al.,	   2006),	   again	  suggesting	   that	   blocking	   GABAARs	   does	   not	   affect	   the	   ultimate	   extent	   of	   [Cl-­‐]i	  increase	  that	   occurs	   during	   ischemia.	   	   	   	   The	   authors	   of	   this	   study	  were	   unable	   to	  identify	  the	  mechanism	  underlying	  the	  pathway	  for	  Cl-­‐	   influx	  that	  occurred	  during	  OGD,	  however	  they	  were	  able	  to	  block	  a	  secondary	  increase	  in	  Cl-­‐	  that	  occurred	  (>1	  hour)	   after	   re-­‐oxygenation	   by	   blocking	   NKCC1	   with	   bumetanide,	   although	   the	  significance	  of	  this	  late	  Cl-­‐	  influx	  remains	  uncertain.	  	  	  In	  addition	  to	  the	  Cl-­‐	  loading	  that	  occurs	  during	  excitotoxic	  insults,	  Cl-­‐	  efflux	  may	   also	   be	   compromised.	   	   As	   KCC2	   directional	   transport	   is	   dependent	   on	   the	  potassium	  gradient,	  small	  changes	  in	  extracellular	  K+	  can	  have	  substantial	  effects	  on	  KCC2	   mediated	   Cl-­‐	   clearance.	   	   Additionally,	   a	   recent	   study	   demonstrated	   that	  glutamate	   activation	   of	   NMDARs	   leads	   to	   phosphorylation	   and	   thereby	   decreased	  expression	  of	  KCC2,	  leading	  to	  decreased	  recovery	  of	  excitotoxic	  Cl-­‐	  loads	  (Lee	  et	  al.,	  2011).	  	  In	  this	  study	  the	  authors	  were	  unable	  to	  identify	  the	  source	  of	  Cl-­‐	  influx,	  but	  showed	   that	   it	  was	   independent	  of	  NKCC1.	   	   	   	   If	  KCC2	  mediated	  Cl-­‐	  efflux	   is	   indeed	  compromised	   following	   cytotoxic	   edema,	   in	   addition	   to	   blocking	   the	   influx	   of	   Cl-­‐	  	   137	  perhaps	   enhancing	   extrusion	   of	   Cl-­‐	   (Gagnon	   et	   al.,	   2013)	   would	   be	   additionally	  beneficial.	  	  	   The	   identification	   to	   SLC26A11	   as	   a	   significant	   Cl-­‐	   entry	   pathway	   during	  pathological	  swelling	  triggered	  after	  Na+	  entry	  suggests	  that	  new	  strategies	  could	  be	  developed	  to	  reduce	  brain	  edema.	  	  There	  are	  numerous	  different	  pathways	  for	  Na+	  entry	   that	   are	   activated	   during	   conditions	   such	   as	   hypoxia,	   stroke	   and	   traumatic	  brain	  injury.	  Our	  observations	  that	  cell	  death	   is	  significantly	  reduced	  when	  overall	  Cl-­‐	   entry	   is	   prevented	   suggests	   that	   therapeutic	   strategies	   to	   inhibit	   SLC26A11	  dependent	  Cl-­‐	  entry	  may	  have	  widespread	  benefit	   towards	   treating	   these	  different	  conditions.	  	   138	  Chapter	  4:	  Conclusions	  	  4.1	  Summary	  of	  Research	  Findings	  	  4.1.1	  Lipid	  Nanoparticle-­‐siRNA	  Delivery	  Mediates	  Targeted	  Knockdown	  of	  Neuronal	  Gene	  Expression	  in	  vitro	  and	  in	  vivo.	  	   In	  chapter	  2,	  we	  presented	  a	  novel	  method	  for	  transfection	  of	  neurons	  both	  in	  vitro	   and	   in	  vivo.	   	   In	  primary	  neuronal	   cultures,	  we	  observed	   transfection	   rates	  approaching	  100%	  with	  no	  signs	  of	  toxicity.	  	  By	  contrast,	  current	  non-­‐viral	  methods	  exhibit	   transfection	   efficiencies	   of	   10%	  or	   less	   and	   are	   often	   accompanied	   by	   cell	  damage	  (Karra	  and	  Dahm,	  2010).	  	  Therefore,	  our	  results	  suggest	  that	  LNPs	  currently	  represent	   the	   best	   available	  method	   for	   transfecting	   neurons	   in	   cell	   culture.	   	   	  We	  conclude	   that	   the	   highly	   efficient	   uptake	   of	   LNPs	   by	   neurons,	   is	   mediated	   via	   an	  endogenous	   ApoE	   dependent	   uptake	   pathway	   similar	   to	   LNP	   uptake	  mechanisms	  utilized	   by	   hepatocytes	   in	   vivo	   (Akinc	   et	   al.,	   2010).	   	   	   As	   astrocytes	   produce	   and	  secrete	  the	  majority	  of	  the	  ApoE	  found	  in	  the	  brain	  (Pitas	  et	  al.,	  1987),	  we	  were	  able	  to	  test	  the	  ApoE	  dependence	  in	  vitro,	  by	  removing	  neurons	  from	  the	  astrocytes	  that	  are	  normally	   separated	  on	   a	   feeder	   layer.	   	  Addition	  of	   exogenous	  ApoE	  proved	   to	  facilitate	  the	  neuronal	  uptake	  of	  LNPs	  in	  a	  dose	  dependent	  manner,	  with	  saturation	  at	  ApoE	  concentrations	  previously	  reported	  in	  vivo	  (Wahrle	  et	  al.,	  2007).	  	  	  	  	  LNP-­‐siRNA	  systems	  used	  in	  this	  study	  had	  a	  mean	  diameter	  of	  around	  50	  nm,	  allowing	   them	   to	   freely	   diffuse	   through	   the	   extracellular	   space	   of	   the	   brain.	   	   We	  	   139	  demonstrated	   that	   a	   single	   intra-­‐cortical	   LNP-­‐siRNA	   injection	  of	  2.5 μg	   siRNA	  per	  rat	  resulted	  in	  robust	  (up	  to	  91%)	  gene	  silencing	  that	  was	  sustained	  for	  at	  least	  15	  days.	   	  We	  proved	  applicability	  of	  LNP-­‐siRNA	  systems	  for	  use	  in	  functional	  genomic	  studies,	   by	   targeted	   functional	   loss	   of	   a	   synaptic	   plasma-­‐membrane	   protein,	   NR1.	  Viral	   delivery	   is	   currently	   the	   most	   effective	   available	   method	   for	   neuronal	  transfection	  in	  vivo,	  however,	  disadvantages	  include	  time	  associated	  with	  packaging	  shRNA	   into	   high-­‐titer	   viruses,	   immunological	   and	   safety	   concerns.	   LNP-­‐siRNA	  systems	  can	  circumvent	  these	  issues,	  and	  provide	  a	  method	  to	  rapidly	  test	  various	  siRNA	   constructs	   against	   the	   expression	   of	   different	   target	   proteins.	   Additionally,	  LNP-­‐siRNA	  systems	  can	  be	  used	  to	  simultaneously	  silence	  multiple	  genes	  at	  once,	  as	  is	  often	  needed	  to	  rule	  out	  compensatory	  and/or	  complementary	  pathways	  (Love	  et	  al.,	  2010).	  	  In	  chapter	  3	  we	  demonstrate	  the	  usefulness	  of	  the	  LNP-­‐siRNA	  approach	  for	  rapidly	  evaluating	  the	  function	  of	  different	  target	  proteins.	   	   	   	  We	  demonstrated	  the	  ability	  of	  knocking	  down	  different	  target	  genes	  using	  LNP-­‐siRNA	  systems	  in	  vivo	  either	   separately	   or	   together	   to	   determine	   the	   molecular	   identity	   of	   a	   novel	  neuronal	   chloride	   entry	   pathway	   that	   caused	   cytotoxic	   neuronal	   edema	   and	   cell	  death.	   	  4.1.2	  Fluorescence	  Lifetime	  Imaging	  of	  CoroNa	  Reports	  Intracellular	  Na+	  Concentration,	  Independent	  of	  CoroNa	  Dye	  Concentration.	  	   Whereas	   conventional	   intensity	   based	   fluorescence	   measurements	   are	  dependent	  on	  the	  total	  number	  of	  photons	  emitted,	  and	  are	  therefore	  reliant	  on	  dye	  concentration,	   fluorescence	   lifetime	   imaging	   (FLIM)	   measures	   the	   time	   that	   the	  	   140	  fluorophore	   spends	   in	   the	   excited	   state,	   and	   FLIM	   measurements	   are	   therefore	  independent	   of	   dye	   concentration.	   	   	   This	   is	   of	   particular	   utility	   when	   dye	  concentrations	  change	  such	  as	  during	  cytotoxic	  edema	  that	  occurs	  during	  ischemia,	  spreading	   depression,	   traumatic	   brain	   injury	   and	   epilepsy.	   	   	   Dye	   dilution	   from	  cellular	  swelling	  will	  alter	  the	  intensity	  of	  the	  fluorescence	  signal	  solely	  as	  a	  result	  of	  the	   decrease	   in	   the	   number	   of	   dye	   molecules	   within	   an	   equivalent	   volume.	  	  	  Additionally,	   FLIM	   provides	   the	   capability	   to	   calibrate	   lifetimes	   with	   known	  concentrations	   of	   substrate,	   and	   report	   absolute	   concentrations.	   	   Although	  calibration	   based	   measurements	   can	   also	   be	   performed	   with	   ratiometric	   Na+	  indicators	   such	  as	   sodium-­‐binding	  benzofuran	   isophthalate	   (SBFI),	   the	  broad	   two-­‐photon	  excitation	   spectrums	  of	   these	   fluorophores	  make	   this	   approach	  difficult	   in	  situ.	   	   In	   chapter	   3	   we	   report	   and	   characterize	   a	   novel	   method	   for	   quantifying	  intracellular	  Na+	  using	  FLIM	  with	  the	  Na+	  indicator	  CoroNa-­‐Green.	  	  Surprisingly,	  we	  found	  that	  the	  dye	  CoroNa-­‐Green	  selectively	  loads	  neurons	  and	  not	  glial	  cells	  when	  bulk	   loaded	   in	   brain	   slices,	   making	   it	   an	   excellent	   indicator	   for	   neuronal	   Na+.	  	  Compared	   to	   other	   sodium	   indicators	   (SBFI	   and	   Sodium	   Green),	   CoroNa-­‐Green	  exhibits	   larger	   fluorescence	   changes	   to	   physiological	   changes	   in	   Na+	   and	   the	   AM-­‐form	  of	  the	  dye	  has	  been	  reported	  by	  the	  manufacturer	  to	  load	  cells	  more	  efficiently	  than	  SBFI	  or	   Sodium	  Green.	   	   CoroNa-­‐Green	   staining	  of	  neurons	  after	   loading	  with	  the	  AM-­‐form	  was	  observed	  to	  gradually	  decrease	  due	  to	  efflux	  out	  of	  neurons.	  	  This	  led	  to	  a	  decrease	  in	  intracellular	  dye	  concentration	  that	  made	  it	  difficult	  to	  monitor	  changes	   in	  Na+	  over	   long	   time	  periods	   (e.g.	  >30	  min)	   (Meier	  et	  al.,	  2006).	   	  FLIM	   is	  	   141	  able	   to	   circumvent	   the	   error	   associated	   with	   these	   gradual	   changes	   in	   dye	  concentration	  as	  we	  observed	  stable	  lifetime	  measurements	  of	  CoroNa	  for	  >	  1	  hour.	  	  	  	  The	   FLIM	   signal	   from	   CoroNa	   excitation	  was	   best	   fit	   using	   a	   biexponential	  function,	   with	   the	   short	   lifetime	   (τfast)	   representative	   of	   the	   intracellular	   Na+	  concentration,	   whereas	   the	   long	   lifetime	   (τslow)	   did	   not	   appear	   to	   change	   with	  changing	  Na+	  concentration.	   	  The	  molecular	  changes	  that	  cause	  τfast	  to	  change	  with	  increasing	   sodium	   concentration	   are	   unknown.	   	   In	   the	   case	   of	   other	   dyes	   that	  increase	   in	   fluorescence	  upon	  binding	  substrate	  such	  as	  the	  Ca2+	  indicator	  Oregon-­‐Green-­‐BAPTA-­‐1	   (OGB-­‐1),	   the	   fluorescence	   lifetime	   is	   predicted	   by	   the	   relative	  amplitudes	   of	   a	   short	   lifetime	   that	   is	   given	   from	   unbound	   dye	   molecules	   and	   a	  longer	  lifetime	  from	  the	  dye	  bound	  to	  substrate	  (Wilms	  et	  al.,	  2006;	  Kuchibhotla	  et	  al.,	   2009),	   therefore	   leading	   to	   an	   increase	   in	   the	   average	   lifetime	   as	   the	  concentration	  of	   substrate	   increases.	   	   It	   is	  possible	   that	  τfast	   similarly	   represents	  a	  mixture	  of	  bound	  and	  unbound	  CoroNa	  molecules,	  with	  unbound	  CoroNa	  exhibiting	  a	  very	  short	  lifetime	  close	  to	  the	  instrument	  response	  function	  (IRF)	  of	  our	  imaging	  system	  (121ps).	  	  In	  situ	  τfast	  showed	  a	  linear	  increase	  over	  the	  concentrations	  of	  Na+	  tested	  (9-­‐109	  mM),	  consistent	  with	  observations	  by	  (Meier	  et	  al.,	  2006)	  showing	  a	  linear	   increase	   in	   fluorescence	   intensity	  with	   increasing	  Na+	   concentrations	   in	   the	  physiological	  range.	   	   	  The	  inwardly	  directed	  Na+	  gradient	  is	  a	  critical	  factor	  for	  the	  homeostasis	   of	   numerous	   ions	   and	   transmitters,	   and	   changes	   in	   intracellular	   Na+	  contribute	   to	   the	   cellular	   damage	   during	   various	   brain	   pathologies.	   	   Currently,	  imaging	  of	  neuronal	  activity	  is	  primarily	  performed	  with	  Ca2+	  dyes,	  whereas	  Na+	  is	  the	   major	   charge	   carrier	   during	   action	   potentials	   and	   excitatory	   postsynaptic	  	   142	  currents.	  	  	  	  We	  conclude	  that	  FLIM	  of	  CoroNa-­‐Green	  can	  be	  used	  to	  quantify	  changes	  in	   intracellular	  Na+,	  and	  will	   facilitate	  the	  study	  physiological	  and	  pathological	  Na+	  homeostasis	  in	  neurons	  and	  possibly	  other	  cell	  types.	  	  4.1.3	  Neuronal	  Swelling	  is	  Dependent	  on	  Na+	  and	  Cl-­‐	  Influx	  but	  Independent	  of	  Ca2+	   	  We	  demonstrated	  in	  chapter	  3	  that	  prolonged	  sodium	  entry	  via	  either	  of	  two	  independent	   pathways	   (VGSCs	   or	   NMDARs)	   causes	   neuronal	   swelling	   and	  subsequent	   cell	   death.	   	  The	  neuronal	   swelling	  ultimately	   required	   the	   influx	  of	  Cl-­‐	  but	   was	   completely	   independent	   of	   Ca2+	   entry.	   	   These	   results	   build	   on	   previous	  studies	  done	  in	  the	  1980s	  that	  identified	  Na+	  and	  Cl-­‐	  entry	  as	  a	  key	  requirement	  for	  excitotoxic	   cell	   death	   of	   neurons	   in	   cell	   culture	   (Rothman,	   1985;	   Choi,	   1987).	   	   	   In	  fact,	  it	  was	  convincingly	  demonstrated	  that	  excitotoxic	  insults	  lead	  to	  two	  separate	  and	   distinguishable	   pathways	   both	   causing	   cell	   death	   (Choi,	   1987).	   One	   pathway	  was	   characterized	   by	   neuronal	   swelling,	   occurred	   early,	   and	   was	   dependent	   on	  extracellular	   Na+	   and	   Cl-­‐,	   but	   was	   independent	   of	   extracellular	   Ca2+.	   	   The	   other	  pathway	   was	   marked	   by	   gradual	   neuronal	   disintegration,	   occurred	   late,	   was	  dependent	  on	  extracellular	  Ca2+,	  and	  could	  be	  mimicked	  by	  a	  Ca2+	   ionophore.	   	  The	  fast	   Na+	   and	   Cl-­‐	  dependent	   cell	   death	   resembled	   the	   morphological	   phenotype	   of	  necrosis,	  whereas	   the	  slower	  Ca2+	  dependent	  cell	  death	  resembled	  apoptosis.	   	  Our	  results	  show	  that	  in	  brain	  slices,	  Na+	  entry	  into	  neurons	  results	  in	  swelling	  and	  rapid	  cell	   death	   that	   occurs	   within	   1	   hour	   and	   that	   can	   be	   completely	   prevented	   by	  removal	  of	  extracellular	  chloride.	  	  	  These	  results	  suggest	  that	  neuronal	  swelling	  and	  	   143	  the	  subsequent	  necrotic	  cell	  death	  can	  indeed	  be	  prevented,	  and	  that	  identification	  and	  blocking	  the	  chloride	  influx	  pathway	  could	  conceivably	  be	  used	  as	  a	  therapeutic	  treatment	  against	  cytotoxic	  edema	  and	  necrotic	  cell	  death.	  	  	  	  4.1.4	  	  Identification	  of	  SLC26A11	  as	  the	  Predominant	  Neuronal	  Cl-­‐	  Influx	  Pathway	  that	  Causes	  Cytotoxic	  Edema	  	   In	  order	  to	  identify	  the	  Cl-­‐	  entry	  pathway	  that	  causes	  neuronal	  swelling	  and	  subsequent	   cell	   death	   we	   used	   a	   two-­‐step	   approach.	   	   First,	   we	   performed	   a	  pharmacological	   survey	   to	   identify	   a	   short	   list	   of	   candidate	   proteins.	   	   Second,	  we	  used	  LNP-­‐siRNA	  delivery	  in	  vivo	  (described	  in	  chapter	  2)	  to	  knockdown	  different	  Cl-­‐	  influx	  candidate	  proteins	  and	  tested	  for	  their	  involvement	  using	  our	  swelling	  assay	  in	  brain	  slices.	  	  	  Our	  pharmacological	  screen	  revealed	  that	  the	  neuronal	  swelling	  was	  completely	   blocked	   by	   a	   non-­‐specific	   anion	   transporter	   blocker,	   DIDS,	   but	   was	  insensitive	   to	   all	   other	   Cl-­‐	   channel	   and	   transporter	   blockers	   tested.	   	   	   This	  pharmacological	   profile	   suggests	   a	   Cl-­‐/HCO3-­‐	  exchanger	   as	   the	   protein	   responsible	  for	   the	  Cl-­‐	  influx	   that	  causes	  neuronal	   swelling	  and	  cell	  death	  (Jentsch	  et	  al.,	  2002;	  Alvarez-­‐Leefmans	  and	  Delpire,	  2009;	  Romero	  et	  al.,	  2013)	   (Table	  3.2).	   	  Consistent	  with	   the	   pharmacology,