UBC Faculty Research and Publications

Relatively frequent switching of transcription start sites during cerebellar development Zhang, Peter; Dimont, Emmanuel; Ha, Thomas; Swanson, Douglas J; Hide, Winston; Goldowitz, Dan Jun 13, 2017

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

Item Metadata


52383-12864_2017_Article_3834.pdf [ 2.09MB ]
JSON: 52383-1.0361998.json
JSON-LD: 52383-1.0361998-ld.json
RDF/XML (Pretty): 52383-1.0361998-rdf.xml
RDF/JSON: 52383-1.0361998-rdf.json
Turtle: 52383-1.0361998-turtle.txt
N-Triples: 52383-1.0361998-rdf-ntriples.txt
Original Record: 52383-1.0361998-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessRelatively frequent switching oftranscription start sites during cerebellardevelopmentPeter Zhang1†, Emmanuel Dimont2,3†, Thomas Ha1, Douglas J. Swanson1, the FANTOM Consortium,Winston Hide2,3,4 and Dan Goldowitz1*AbstractBackground: Alternative transcription start site (TSS) usage plays important roles in transcriptional control ofmammalian gene expression. The growing interest in alternative TSSs and their role in genome diversificationspawned many single-gene studies on differential usages of tissue-specific or temporal-specific alternative TSSs.However, exploration of the switching usage of alternative TSS usage on a genomic level, especially in the centralnervous system, is largely lacking.Results: In this study, We have prepared a unique set of time-course data for the developing cerebellum, as partof the FANTOM5 consortium (http://fantom.gsc.riken.jp/5/) that uses their innovative capturing of 5′ ends of alltranscripts followed by Helicos next generation sequencing. We analyzed the usage of all transcription start sites(TSSs) at each time point during cerebellar development that provided information on multiple RNA isoformsthat emerged from the same gene. We developed a mathematical method that systematically compares theexpression of different TSSs of a gene to identify temporal crossover and non-crossover switching events. Weidentified 48,489 novel TSS switching events in 5433 genes during cerebellar development. This includes 9767crossover TSS switching events in 1511 genes, where the dominant TSS shifts over time.Conclusions: We observed a relatively high prevalence of TSS switching in cerebellar development where theresulting temporally-specific gene transcripts and protein products can play important regulatory and functional roles.Keywords: Cerebellum, Developmental biology, Promoter, Promoter switching, HeliScopeCAGE, Alternative promoters,Alternative splicing, Transcription start siteBackgroundAlternative splicing can provide a large reservoir of tran-scriptional variants from the ~22,000 genes identified bythe Human Genome Project [1]. The production of dif-ferent isoforms due to the usage of alternative transcrip-tion start sites (TSSs), which was once considered asuncommon, has now been found in the majority ofhuman genes [2, 3]. Alternative TSSs could be results ofa gene duplication event followed by the loss of func-tional exons in the upstream copy and diversification ofthe two duplicated promoters. Alternative TSS usage canaffect gene expression and generate diversity in a varietyof ways. On the transcriptional level, alternative TSScould result in tissue-specific expression, temporally reg-ulated expression, and the amplitude of expression. Onthe post-transcriptional level, alternative TSS can affectthe stability and translational efficiency of the mRNA.Furthermore, alternative TSS can result in protein iso-forms with a different amino terminus, which can leadto alterations in protein levels, functions, or subcellulardistribution. Therefore, the investigation of temporalswitching of TSSs can provide insights into the regula-tion of different protein isoforms, and presumably theirdifferences in function. One way to optimally identify* Correspondence: dang@cmmt.ubc.ca†Equal contributors1Centre for Molecular Medicine and Therapeutics, Child and Family ResearchInstitute, Department of Medical Genetics, University of British Columbia, 950West 28th Avenue, Vancouver, BC V5Z 4H4, CanadaFull list of author information is available at the end of the article© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Zhang et al. BMC Genomics  (2017) 18:461 DOI 10.1186/s12864-017-3834-zdifferential use of isoforms is to examine transcriptionalregulation over developmental time.One high-throughput technique to survey gene ex-pression at the transcriptome level is Cap Analysis GeneExpression (CAGE) which generates a genome-wide ex-pression profile based on sequences from the 5′ end ofthe mRNA [4, 5]. In the FANTOM project, CAGE hasbeen shown to identify different TSSs and the corre-sponding promoters for single genes [6–9]. With CAGEdata, one can infer the TSS usage through the numberof transcripts produced at that particular TSS. Whenmore than one TSS is used at a single time point froma single gene, the TSS with highest expression is con-sidered the “dominant” TSS. The understanding of howthe TSS usage changes during development can shedlight on how a single gene can function differently overdevelopmental stages through temporally regulated al-ternative mRNA and protein isoforms.The complexity of brain development requires intri-cately controlled expression of specific genes acrosstime. The cerebellum is often used as a model in ana-lyses of brain development due to its limited number ofmajor cell types. These cells are positioned in spatiallydefined territories of the developing cerebellum. Thecerebellum has also been the focus of two extensivegenome-wide gene expression profiling of the developingcerebellum [10, 11]. Detailed information on temporallyregulated promoter usage of developmentally importantgenes - which is still largely lacking - can providevaluable information on genome diversity. Moreover, dif-ferent isoforms of these genes may be translated intodistinct protein products that perform different tasks.Such analyses would give insight to the alterations madeto the form of the final transcript, localization for tran-scription factors motif prediction, utilization, and associ-ated regulatory network changes. Thus, in collaborationwith the FANTOM5 project [12], we generated a CAGEdataset for the developing cerebellum with 12 time pointsto study temporally-regulated gene expression and alter-native TSS usage during cerebellar development.TSS switching events across samples were systematicallyidentified by comparing differential promoter transcrip-tion levels between pairs of TSSs and pairs of develop-mental time points, and by applying the Silvapulle FQ test,a statistical method for constrained hypothesis testing thatwe specifically apply for the detection of crossover TSSswitching events [13]. The FQ test produces p-values to es-timate significance of a crossover switching event. Wehave applied the FQ test to our cerebellar time series toidentify novel TSS switching events during cerebellar de-velopment. Our hypothesis was that differential TSS usagecan result in significant regulatory changes that underliecellular events critical for cerebellar development andmorphogenesis. By taking advantage of the FANTOM5collaboration with our cerebellar developmental timecourse, we identified 48,489 novel TSS switching events,including 9767 events in which the dominant TSS shiftsover time. These TSS switching events were predicted toproduce temporally-specific gene transcripts and proteinproducts that can play important regulatory and func-tional roles during cerebellar development.MethodsMouse colony maintenance and breedingThis research was performed with ethics approval fromthe Canadian Council on Animal Care and research con-ducted in accordance with protocol A12–0190. C57BL/6 J mice were used in all experiments and were importedfrom The Jackson Laboratory (Maine, US) and main-tained in our colony as an inbred line. To standardizethe time of conception, timed pregnancies were set up.Every weekday at 10:00 am, females were coupled withmale; at 3:00 pm, the females were checked for vaginalplugs and removed from their partners. The appearanceof a vaginal plug was recorded as the day of conception(i.e. embryonic day 0) and embryos were collected at10 am on embryonic day 11–18 (E11-E18) every day andpostnatal day 0–9 (P0-P9) every 3 days for a total of 12time points in our cerebellar time series.Tissue processingOn the day of embryo collection, the mothers weresacrificed and embryos were removed from the uterus inice-cold RNAse-free PBS. Cerebella were dissected fromthe head of the embryos, then pooled with littermates,and snap-frozen in liquid nitrogen. Three replicate poolsof whole cerebella samples were collected at each timepoint. The standard TRIzol RNA extraction protocol[14] was used for tissue homogenization and RNAextraction.Quality assessmentA Bioanalyzer (Agilent, Santa Clara, CA) was used toexamine RNA quality. All RNA samples used for thetime series achieved high RNA Integrity (RIN) scoresabove 9.0. The samples were sent to RIKEN Omics Centerat Yokohama, Japan, as part of Functional Annotation ofthe Mammalian Genome 5 (FANTOM5) collaboration forCAGE analysis.Transcriptome library generation by HeliScopeCAGECAGE is a technique that generates a genome-wide ex-pression profile based on sequences from the 5′ end ofthe mRNA. With CAGE, the first 27 bp from the 5′ endof RNAs were extracted and reverse-transcribed to DNA[4]. The short DNA fragments were then systematicallysequenced with the Helicos platform [15]. Each se-quenced tag was then mapped to the reference genomeZhang et al. BMC Genomics  (2017) 18:461 Page 2 of 14to identify the transcription start site (TSS) of the genefrom which it was transcribed. “Tag per million” (tpm)was used as a measure of the expression level of RNAsbased on concentration – an expression of “10tpm”means that out of each million total transcripts, 10 weretranscribed from the TSS in question. Alternative TSSs(illustrated in Fig. 1a) can be detected when multipleCAGE tags are mapped to the same gene locus in thereference genome. Mapped CAGE tags can be clusteredinto promoter regions after thresholding to determinebona fide promoter regions in the genome. For thisanalysis, we use the list of promoter regions publishedby the FANTOM5 Consortium [5].TSS switch detectionTSS switching events are detected by comparing the ex-pression of transcripts from two TSSs of a single gene attwo time points. The difference in expression level ofthe two TSSs is designated d1 and d2 at time point 1and time point 2, respectively. The null hypothesis isthat there is no switching for the two TSSs (d1 = d2, seeFig. 1b). The test of this hypothesis was performed usingthe standard t-test, with candidate switching eventsidentified at this preliminary stage if the adjusted p-valuewas <0.2. A non-crossover TSS event is detected if oneTSS is used more frequently at one time point comparedto the other, but the same TSS is used dominantly atboth time points (d1 > d2, or d1 < d2, both d1 and d2same sign, Fig. 1b). A crossover TSS switching event isdetected if one TSS is used more frequently at one timepoint compared to the other, and that the dominant TSSswitches at the two time points (d1 > 0 and d2 < 0 ord1 < 0 and d2 > 0, Fig. 1b). In order to reduce potentialconfounding of TSS switching events by differential ag-gregate promoter expression between time points, candi-date events were further limited to TSS pairs that do notchange in overall mean expression between developmen-tal stages being compared. The null hypothesis tested atthis stage is that the mean TSS expression at the twotime points is equal, and results were filtered out if thet-test adjusted p-value was <0.1.In addition to the differences in expression (d1,d2),the results of TSS switching are represented using theFQ statistic [12] which formally tests for the presence ofcrossover switching for each gene. The test of the nullhypothesis of no differential crossover promoter usagecorresponds to a test involving the FQ statistic, which isfunctionally similar to the ANOVA F-test. Exact p-valuesfor this test are obtained as described in Silvapulle [12].To our knowledge, the Silvapulle FQ test is the onlystatistical test available that was specifically developedfor testing hypotheses regarding qualitative interaction,and which we apply in the current study for testing thepresence of crossover switching in gene promoterusage.All P-values are adjusted for multiple comparisonsusing the Benjamini–Hochberg method to control thefalse discovery rate. The P-value of the FQ test was usedas an indicator of significance for choosing biologicalvalidation candidates.Fig. 1 A schematic diagram of alternative transcription start sites (TSSs) and the classes of TSS switching. a Alternative TSSs can generate differentsplicing variants that can be translated into different protein isoforms. *the functional domains may be affected by alternative TSSs which resultsin functional diversity. b Different outcomes comparing alternative TSS usage at two time points – no TSS switching, non-crossover TSS switchingor crossover TSS switching. Y-axis represents the quantitative measure of TSS usage measured by the expression level of its mRNA transcript. X-axisrepresents the two developmental time points used in the comparison (t1 vs. t2)Zhang et al. BMC Genomics  (2017) 18:461 Page 3 of 14Gene ontology analysis for gene with crossover switchingeventsTo identify cellular processes and molecular pathways ingenes with crossover TSS switching events, we usedDatabase for Annotation, Visualization and IntegratedDiscovery program (DAVID, https://david.ncifcrf.gov/[16]) to examine the gene ontology of genes with at leastone crossover event with p < 0.05 in FQ test. Top 20 GOterms were used for overall analysis in crossover TSSswitching genes during cerebellar development. Further-more, for temporal functional analysis of crossover TSSswitching events, top 20 GO terms were generated withDAVID for all events associated with three developmen-tal time points – E13, E15 and P0.In silico validation of gene expression with establisheddatabases and experimental validation with genestructure prediction and quantitative real-time PCRWe used online databases to examine the 20 genes withthe lowest p-values. First, we used in situ resources -Genepaint (http://genepaint.org [17]) and Allen BrainAtlas (http://www.brain-map.org [18]) to examine thegenes’ expression in the cerebellum. Second, we examinedthe predicted mRNA structures from the two TSSs withthe intron/exon database Aceview (http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/ [19]) as well as functionaldomains of their protein products from protein domaindatabase PhosphoSitePlus (http://www.phosphosite.org[20]) to determine the potential effect of TSS switchingevents on biological function.Three genes were chosen for further validation forTSS-specific quantitative real-time PCR for the valid-ation of alteration in TSS usage at E12, E15 and P9.Cerebellar RNA was extracted from C57BL/6 J mice atE12, E15 and P9 following the same procedure that wereused for HeliScopeCAGE RNA collection. cDNAs wereproduced with random hexamers using the High Capa-city cDNA Archive Kit (Applied Biosystems). cDNAproducts were diluted to 100 ng total RNA input.Sequences of the transcript of interest were loaded intoPrimer Express® software (Applied Biosystems). For eachgene, an isoform-specific forward primer was designedfor each of the long and short isoform, while the reverseprimer aligns to a common sequence that is shared byboth isoforms. Amplicon lengths were between 80 and120 bp. The qPCR was performed with the FAST SYBRGreen PCR Master Mix (Applied Biosystems) on an ABIStepOne Plus Sequence Detection System (AppliedBiosystems). All runs were normalized to the controlgene, Gapdh. Three biological replicates were preparedfor each gene target and three technical replicates wereperformed for each biological replicate. Gene expressionwas represented as relative quantity against the negativecontrol which used water as the template (noted as“Relative Quantity vs. H2O” in figures). The results ofReal-Time PCR were analyzed and graphed by ABIStepOne Plus Sequence Detection System (AppliedBiosystems). The expression data were compared withthe HeliScope-CAGE data.ResultsOverview of promoter switch events during cerebellardevelopmentOur cerebellar time series, which consisted of transcrip-tome data from 12 time points, yielded a total of183,903,557 CAGE tags that are mapped to 25,207 genesin the reference genome. We identified 48,489 TSSswitching events (Fig. 2a) in the cerebellar time seriesdata that occur in 5433 genes. These events are com-prised of 38,722 non-crossover switching events (Fig. 2b)that occur in 5293 genes, and 9767 crossover switchingevents (Fig. 2c) that occur in 1511 genes. One thousandthree hundred seventy-one out of 1511 genes (~91%)that have crossover TSS switching events also have atleast one non-crossover switching event. This indicatesthat crossover TSS switching events are rarer and occurin fewer genes when compared to the non-crossoverevents.When comparing the cerebellar TSS switching data tonine other tissues in the FANTOM5 dataset (see Table1; detailed descriptions about time series on these tissuescan be found in [12]), our cerebellar development timeseries has the 3rd highest total number of TSS switchingevents (48,489) behind “Epithelial to mesenchymal”(132,661 events) and “Adipocyte differentiation” (66,087events); and is the highest of the three samples derivedfrom ectoderm [“Human iPS to neuron (wt) 1” and“Trachea epithelia differentiation”]. While the cerebellardevelopment time series has less total events than“Epithelial to mesenchymal” and “Adipocyte differenti-ation” samples, it has a higher frequency of crossoverTSS switching events - 20.1% vs 17.6 and 12.5%, re-spectively. Interestingly, when compared to 48,489events found in the cerebellum, four out of the fiveremaining datasets had a higher percentage of crossoverevents but a much lower number of total switching events.In conclusion, cerebellar development showed a highfrequency in crossover TSS switching among datasetswith a high number of total switching events.Distribution of TSS switching events in cerebellartranscriptomeWhen we looked at the distribution of the 48,489 TSSswitching events over the 5433 genes, we found a ma-jority of genes with few events and a minority of geneswith many events. Thus, we found there are 1534 (28%of TSS switching gene) genes with one TSS switchingevent; and only two genes with more than 800 switchingZhang et al. BMC Genomics  (2017) 18:461 Page 4 of 14events (Fig. 3a). When we looked at the top 20 genes withthe most TSS events (listed in Table 2), we observed thatthese genes account for 13.5% for all TSS switchingevents, or a total of 6567 events. From Fig. 3 (as well asTable 2), we can see that there are two outlier genes thathave the largest number of TSS switching events for all 3groups (all TSS, non-crossover and crossover, indicated byarrows in Fig. 3a-c) - Frmd4a (FERM domain containing4A) with a total of 852 TSS switching events and Ank3(ankyrin 3) with a total of 801 TSS switching events (seeTable 2). These two genes have more than twice the num-ber of TSS switching events than the next closest gene,Abr (active BCR-related gene) with a total of 386 TSSswitching events. The numbers of TSS switching eventsare more evenly distributed across the rest of the 18 geneswith a higher frequency of switching (see Fig. 3) as the dif-ference between each rank is less than 10% of the numberof events in this group.When comparing the distribution of crossover and non-crossover events, we found that crossover switching eventsare clustered in fewer genes when compared with non-crossover events. Since the frequency of non-crossoverswitching is about four times the number of cross-over(38,722:9767 or 3.96:1), we would expect roughly a 4:1 ra-tio for non-crossover: crossover events for any given gene,assuming an even distribution of both categories. Indeed,we observed roughly a 4:1 ratio for Ablim1 (204 non-crossover events and 50 crossover events) and Dlg2Fig. 2 Overview of TSS switching events during cerebellar development. a Overview of 48,489 TSS switching events during cerebellardevelopment. These events significantly deviate from the no-switching line (indicated by d1 = d2) (p < 0.05). b Overview of 38,722 non-crossoverTSS switching events during cerebellar development. c Overview of 9767 crossover TSS switching events during cerebellar development. X-axisrepresents d1, which is the difference in expression between the two TSSs, measured in tags per million (tpm), at developmental time point 1(t1), see Fig. 1b for a graphic illustration. Y-axis represents d2, which is the difference in expression between the two TSSs, measured in tag permillion (tpm) at developmental time point 2 (t2), see Fig. 1b for a graphic illustrationTable 1 Comparison of TSS switching events during cerebellar development with other FANTOM5 datasetsTime Series Germ layer TP# Switching# Gene# % Non-Xover % Xover %Cerebellar development Ectoderm 12 48,489 5433 21.6 38,722 79.9 9767 20.1Human iPS to neuron (wt) 1 Ectoderm 4 45,069 6692 26.5 41,302 91.6 3767 8.4Trachea epithelia differentiation Endoderm 19 8389 2458 9.8 6112 72.9 2277 27.1Adipocyte differentiation Mesoderm 16 66,087 5996 23.8 57,857 87.5 8230 12.5Epithelial to mesenchymal Mesoderm 21 132,661 7004 27.8 109,252 82.4 23,409 17.6BMM TB activation IL13 Mesoderm 11 825 527 2.1 564 68.4 261 31.6AoSMC response to IL1b Mesoderm 10 192 159 0.6 129 67.2 63 32.8Macrophage response to LPS Mesoderm 23 32,234 4557 18.1 22,239 69.0 9995 31.0ES to cardiomyocyte Mesoderm 13 189 163 0.6 100 52.9 89 47.1Myoblast to myotube Mesoderm 9 21,912 4249 16.9 18,735 85.5 3177 14.5TP# number of time points in the time series, Switching # total number of TSS switching events found in the dataset, Gene # total number of genes with at leastone TSS switching event, Column 6: % TSS switching genes over all 25,207 genes, Non-Xover total number of non-crossover TSS switching events found in thedataset, Column 8: % percentage of non-crossover events over all switching events, Xover total number of crossover TSS switching events found in the dataset,Column 10: % percentage of crossover events over all switching eventsZhang et al. BMC Genomics  (2017) 18:461 Page 5 of 14(223 non-crossover events and 43 crossover events,Table 2). However, for the majority of the 20 genes withthe greatest number of switching events, the frequencyof crossover events is much higher than one fourth of thenon-crossover counterpart, such as the two outlier genesmentioned above - Frmd4a (509 non-crossover events vs343 crossover events) and Ank3 (464 non-crossoverevents vs 337 crossover events, Table 2). This un-even dis-tribution of crossover events is also reflected by the lowerabundance of genes with a low number of switchingevents – 3052 genes have less than 3 non-crossover events(Fig. 3b) and only 944 genes have less than 3 crossoverevents (Fig. 3c). In conclusion, we found that crossoverevents tend to cluster in a fewer number of genes whencompared to the non-crossover counterpart.Gradual increment in the number of crossover TSSswitching events over developmental timeNext, we focused in the temporal distribution of cross-over TSS switching. When we look at a day-to-daychange in promoter usage (E12 vs E11, E13 vs E12 etc.,underlined in Table 3), TSS switching occurs evenlyacross cerebellar development from 13 events to 39events - with the exception of the E13-E12 comparison(Table 3). There are 93 TSS switching events betweenE12 and E13 indicating a major shift in promoter usageat this developmental stage.To examine the general pattern of TSS switching duringcerebellar development, we counted promoter switchevents by developmental time points (Table 3). Among the12 data points in our time course, a total of 66 comparisonsbetween two data points have been carried out to searchfor the switching of alternative TSSs (Table 3). Over thetime series, there is a general incremental number of cross-over switching events that are detected between two sam-ples that are temporally distant. This most likely reflects thegradual shift of cerebellar transcriptome and TSS usageduring development. There are rare exceptions to this pat-tern, for example, there are more switching events betweenE11 and E17 samples than found between E11 and E18samples.Fig. 3 Distribution TSS switching events in different genes during cerebellar development. a Distribution 48,489 TSS switching events in genesduring cerebellar development. Arrow points to the two genes with more than 800 switching events. b Overview of 38,722 non-crossover TSSswitching events in 5293 genes during cerebellar development. c Overview of 9767 crossover TSS switching events in 1511 genes during cerebellardevelopment. x-axis – number of TSS events occurs within one gene (log2 scaled). y-axis – number of genes that have the number of TSS eventsindicated on the x-axisTable 2 Top 20 genes with highest numbers of TSS switchingeventsGene ID All events Non-crossovereventsCrossover events1 Frmd4a 852 509 3432 Ank3 801 464 3373 Abr 386 275 1114 Ednrb 356 211 1455 Iqsec1 348 206 1426 Bcat1 329 221 1087 Pde4d 308 176 1328 Ldb1 304 167 1379 Sorbs2 297 175 12210 Cnpy1 273 158 11511 Dlg2 266 223 4312 Ebf1 262 160 10213 Ablim1 254 204 5014 Zeb2 246 218 2815 Trim2 233 168 6516 Celf2 227 162 6517 Map2 226 170 5618 Itgb8 208 126 8219 Ank2 197 126 7120 Ptprg 194 111 83Zhang et al. BMC Genomics  (2017) 18:461 Page 6 of 14Gene ontology analysis for genes with the mostsignificant crossover TSS switching eventsTo functionally annotate the genes that undergo sig-nificant crossover TSS switching, we used the Data-base for Annotation, Visualization and IntegratedDiscovery program (DAVID, https://david.ncifcrf.gov/[16]) to examine the biological process and terms as-sociated with crossover TSS switching genes. From1509 genes with 9767 crossover TSS switching eventsat p < 0.05, we analyzed 20 gene ontology (GO) termswith the lowest p-value from the DAVID analysis (seeFig. 4a). Terms associated with neuronal development,such as “neuron development”, “neuron projection”and “synapse” also showed up at high significancelevels from DAVID analyses (Fig. 4a).We have found that the largest alteration in gene ex-pression occurs at E13, E15 and P0 (manuscript in prep-aration) and were interested to determine the extentthat crossover TSS switching plays a role in transcrip-tome diversity. When comparing crossover events at E13with all other time points we find 1440 significant(p < .05) events in 584 genes. When comparing cross-over events at E15 with other time points we find 1355significant (p < 0.05) events in 582 genes. Finally, whencomparing crossover events at P0 with all other timepoints we find 1152 significant (p < .05) events in 506genes. We used these gene lists as input to DAVID andthe top 20 terms were selected for these temporal com-parisons among the three time points (Fig. 4b). Wefound that 7 terms (phosphoprotein, alternative splicing,splice variant, cytoplasm, neuron projection, cytoskeletalprotein binding and cytoskeleton) were shared amongeach of the three time points. These 7 GO terms werealso found among the 8 most significant terms in theanalysis with all genes discussed previously. We also ob-served that comparisons between shorter time spansyield more common GO terms –e.g., there are 5 termsshared between genes with crossover TSS events at E13and E15, 1 term between E15 and P0 and no terms werecommon between E13 and P0. Lastly, the majority ofGO terms unique to a given time point shared a com-mon theme that may reflect active biological process oc-curring at the given time – e.g., four out of eight E13terms were associated with cell motion and cytoskeleton;five out of seven E15 terms were associated with ionbinding and six out of twelve P0 terms were associatedwith regulation of intracellular organization.Validation of promoter switching eventsTo further investigate the genes with the 20 most signifi-cant TSS switching events, we used the in situ hybridizationexpression database Genepaint (http://www.genepaint.org/)to examine their expression pattern in the cerebellum(summarized in Table 4). Three of these genes showed ro-bust cerebellar expression (Gpc6, Anp32a and Cntnap2)and were chosen to demonstrate the potential biologicalroles of the TSS switching events during cerebellar develop-ment. First, their mRNA structures were obtained from theintron/exon database Aceview (http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/); then their protein structurefor each isoform was obtained from protein domain data-base PhosphoSitePlus (http://www.phosphosite.org); finally,the TSS switching events for these three genes were vali-dated with quantitative real-time PCR with promoter-specific primers.When we investigate the role of the most significantTSS switching events, we found that some of the mostsignificant events do not seem to affect protein sequenceand may play roles in transcriptional or post-transcriptionalregulation. One example we examined is Glypican-6(Gpc6)- a member of Glypican family that is found on the cell sur-face and plays important roles in cellular growth controland differentiation. The two TSS sites are 32 bp apart inthe genome and mRNA that originate from the two TSSsites differ in the first exon in the 5’UTR region (Fig. 5a).The two forms of mRNA were predicted to be translatedinto the same protein isoform that contains 565 aminoacids. The single glypican domain that makes up the major-ity of the peptide is not effected by the TSS switching event(Fig. 5a). Therefore, the usage of alternative TSSs in Gpc6,which is expressed in the NE, NTZ and EGL in the cerebel-lum (Fig. 5b), could play a regulatory role, such as tempor-ally regulated expression, amplitude of expression, mRNAstability and mRNA translational efficiency. Our qRT-PCRdata confirmed the TSS switching prediction (Fig. 5c) andshowed that it undergoes a non-crossover TSS switchingbetween E15 (TSS2 is the dominant form and has >2 foldusage compared with TSS1) and P9 (TSS2 has slightlyTable 3 Distribution of crossover TSS switching events acrosstime in cerebellar development (N = 9767)E12 E13 E14 E15 E16 E17 E18 N0 N3 N6 N9E11 31 180 290 340 320 333 236 291 354 279 429E12 93 159 200 209 238 162 228 274 225 381E13 21 59 99 97 76 118 190 180 327E14 34 55 86 69 114 203 198 303E15 35 30 29 56 129 143 301E16 29 23 53 103 113 226E17 13 39 58 91 204E18 20 42 76 123N0 39 60 134N3 25 76N6 17Number of crossover TSS switching events that are found in adjacent timepoints are in BoldFor example, 93 in column 3, row 3 represents 93 crossover events foundbetween E12 and E13Zhang et al. BMC Genomics  (2017) 18:461 Page 7 of 14higher usage than TSS1, but remains as the dominant form,see Fig. 5d).Some of the most significant TSS switching eventsoccur between two TSSs that could produce protein iso-forms with different N-termini, which may or may notaffect the function of the protein isoforms. An exampleof this would be Acidic (leucine-rich) nuclear phospho-protein 32 family member A (Anp32a) - a member ofacidic nuclear phosphoprotein 32 kDa (Anp32) family.The two TSS sites are 328 bp apart in the genome andmRNA that originate from the two TSS sites differs inthe first exon in the 5′UTR region as well as the N-terminus of protein products. The first 12 amino acidsof the long isoform were absent on the short isoform.Functional domains were not affected by the TSS switch-ing event - both isoforms retained two LRR4 domainsand a single NOP14 domain (Fig. 6a). The difference atthe N-terminus can lead to alterations in Anp32a’s pro-tein level, subcellular distribution or function in the EGLwhere it is strongly expressed (Fig. 6b). As predicted(Fig. 6c) and validated with our qRT-PCR data, Anp32aundergoes a crossover TSS switching between E12(TSS9 as dominant form) and P9 (TSS 4 as dominantform, see Fig. 6d).Lastly, among the genes with the most significant TSSswitching events, we have discovered a crossover TSSswitching event where protein function is highly affectedin the Contactin-associated protein-like 2 (Cntnap2) – agene encodes a member of the neurexin family whichfunctions as cell adhesion molecules and receptors inneurons. The two TSS sites, that lead to the transcrip-tion of two NCBI-validated mRNA refseqs, are morethan 2 million bp apart in the genome. mRNAs that ori-ginate from the two TSS sites differ by more than6000 bp and consist of the first 20 exons of the longmRNA – only 4 exons at the 3′ end of the long formmRNA (NCBI Locus: NM_001004357.2) are present inthe short form (NCBI Locus: NM_025771.3, see Fig. 7a).The Cntnap2 protein, in its long isoform (NCBI Locus:NM_025771.3), contains 1400 amino acids and manyfunctional domains including one F5/8 type C domain,two epidermal growth factor repeats domains, fourlaminin G domains and a TM domain. The short proteinisoform of Cntnap2 (NCBI Locus: NP_080047.1), whichhas 190 amino acids has only two of the eight functionaldomains remaining, the last laminin G domain and theTM domain (Fig. 7a). In the Genepaint database, a probespecific to the long isoform of Cntnap2 was used, and itFig. 4 GO Analysis for genes significant (p < 0.05) for crossover switching at all time points (left) and at three selected time points (right). a Top20 terms from GO analysis of all 9767 crossover TSS switching events in 1509 genes For column heading: “Term” is the GO term, “Count” is thenumber of genes associated with the GO term and “%” is the fraction of the number of genes associated with the GO term divided by the totalinput of 1509 genes, “PValue” and “Bonferroni” represent the significance of the GO term. b A Venn diagram comparing the top 20 GO termsfrom crossover TSS switching events between all samples and either E13, E15 or P0 samplesZhang et al. BMC Genomics  (2017) 18:461 Page 8 of 14is indicated that the long isoform is primarily expressedin the rhombic lip of the cerebellum at E14.5 (Fig. 7b).According to our prediction (Fig. 7c) and qRT-PCRresults, Cntnap2 undergoes a crossover TSS switchingbetween E15 (TSS4 as dominant form) and P9 (TSS3 asdominant form, see Fig. 7d). The highly differentiatedprotein isoforms of Cntnap2 suggest the gene’s temporalshift in protein functions during cerebellar developmentwhere a truncated form is made specifically in the duringearly embryonic stages.DiscussionHigh prevalence of alternative TSSs in mammaliangenomesIn this study, we have identified 5293 genes (~21% of atotal of 25,207 genes) that exhibit differential TSS usageduring cerebellar development. These findings are in linewith previous studies and indicate that TSS switchingevents are common and can play an important role inthe diversity of the cerebellar transcriptome during de-velopment [21–23]. Furthermore, we have identified9767 crossover TSS switching events which suggests analteration in the dominant TSS over time. Since thealternative mRNA isoforms could be translated intofunctionally different products, a crossover switchingevent suggests that one gene can play different roles atdifferent time points in development.Alternative usage of multiple TSSs of one gene is com-mon in mammalian genomes. It is a key mechanism toincrease mRNA and protein diversity since multiplemRNAs from a single gene can encode distinct proteinisoforms with different functions (reviewed in [24]).Recent studies suggest that about half of the mousegenes have multiple alternative promoters [25, 26]. Forexample, alternative promoters have been identified in>20% of genes in ENCODE (http://genome.ucsc.edu/ENCODE/) regions [6]. Other genomic studies alsofound more than a quarter of human genes having mul-tiple active promoters [27–29]. The complex transcrip-tional regulation of alternative promoter usage hasbeen identified in several genes [24]. Furthermore, insome genes, such as tumor protein p53 (TP53) andTable 4 Cerebellar expression patterns of genes with most significant switching events at E14.5 from the in situ database,GenepaintGene Full name GenepaintDLG3 discs, large homolog 3 N/ESLC12A5 solute carrier family 12, member 5 N/EPDE4D phosphodiesterase 4D NE, interior cerebellumIQSEC1 IQ motif and Sec7 domain 1 N/ECNTNAP2 contactin associated protein-like 2 RL specificCNPY1 canopy 1 homolog N/AMAPK8IP1 mitogen activated protein kinase 8 interacting protein 1 specific cerebellar nuclei, spinal cordDLGAP4 discs, large homolog-associated protein 4 widespread cerebellumANK3 ankyrin 3, epithelial interior cerebellumCACNB4 calcium channel, voltage-dependent, beta 4 subunit N/EANP32a acidic (leucine-rich) nuclear phosphoprotein 32 family strong, EGL & NE specific stainingTMX3 thioredoxin-related transmembrane protein 3 N/AAPBB3 amyloid beta (A4) precursor protein-binding, family B, member 3 N/EPRMT8 protein arginine N-methyltransferase 8 widespread cerebellumEDNRB Mus musculus endothelin receptor type B strong NE specific stainingSEMA4G sema domain 4G widespread cerebellumFBLN5 fibulin 5 N/EZRANB1 zinc finger, RAN-binding domain containing 1 N/EZBTB38 zinc finger and BTB domain containing 38 N/AIBTK inhibitor of Bruton agammaglobulinemia tyrosine kinase N/EGPC6 glypican 6 Strong NE, NTZ specific stainingHSPH1 heat shock 105 kDa/110 kDa protein 1 N/AZFP451 Mus musculus zinc finger protein 451 moderate EGL stainingGRAMD1B GRAM domain containing 1B N/EN/E not expressed or ineffective probe, NE neuroepithelium, RL Rhombic lip, EGL external granular layer, NTZ nuclear transitory zone, N/A data not availableZhang et al. BMC Genomics  (2017) 18:461 Page 9 of 14Fig. 5 Alternative TSSs in glypican 6 (Gpc6) and experimental validation of its non-crossover switching events with Real-time PCR. a SchematicDNA structure of Gpc6, alternative mRNA variants and un-altered protein structure. b in situ expression of Gpc6 in mouse cerebellum at E14.5(from GenePaint). c HeliscopeCAGE expression data for the two alternative TSSs during cerebellar development. X-axis: time, from embryonic day 11 (E11)to postnatal day 9 (P9). Y-axis: expression level measured in tpm (tags per million). d qRT-PCR expression data demonstrating a non-crossover TSS switchingevent between E15 and P9. X-axis: time at E12, E15 and P9. Y-axis: expression level measured in RQ (relative quantity against H2O as negative control)Fig. 6 Alternative TSSs in Acidic nuclear phosphoprotein 32 family, member A (Anp32a) and experimental validation of its crossover switchingevents with Real-time PCR. a Schematic DNA structure of Anp32a, alternative mRNA variants and altered protein structure at the N-terminus.b in situ expression of Anp32a in mouse cerebellum at E14.5 (from GenePaint). c HeliscopeCAGE expression data for the two alternative TSSsduring cerebellar development. X-axis: time, from embryonic day 11 (E11) to postnatal day 9 (P9). Y-axis: expression level measured in tpm(tags per million). d qRT-PCR expression data demonstrating a crossover TSS switching events between E12 and P9. X-axis: time at E12, E15and P9 Y-axis: expression level measured in RQ (relative quantity against H2O as negative control)Zhang et al. BMC Genomics  (2017) 18:461 Page 10 of 14guanine nucleotide binding protein (GNAS), alternativepromoters were shown to be activated or silenced [29].However, the focus of previous studies has been thetissue-specific transcriptional regulation of alternativepromoters; the temporal aspect of alternative promoterusage during cerebellar development has been over-looked. Our analyses focused on the switching usage ofalternative promoter in the mouse cerebellum, and thisis the first systematic study of alternative promoterusage in the development of the mouse cerebellum.Temporal regulation of alternative TSS associated withdevelopmental processes in the cerebellumAlternative TSSs reflect different promoter regionsthat can be used for tissue-specific and/or temporal-specific expression. For example, albumin in hepato-cytes has several cis-acting elements that recruit differentsets of trans-acting factors, which enable spatial, temporaland dynamics regulation of the transcription of albu-min mRNA [30]. In this study, we have identified 9767crossover TSS switching events in 1511 genes. Thus, in~20% of genes there is more than one promoter that isused dominantly during cerebellar development. Func-tional annotation analysis for these genes revealed GOterms that are expected to be associated with alterna-tive promoter usage, such as “alternative splicing” and“splicing variants”, as well GO terms that point toprocesses where promoter switching might play a roleduring development, such as “phosphoprotein”,“cytoskeleton organization” and “neuron projection”.Phosphoproteins are involved in the post-translationalregulatory process phosphorylation, in which a phos-phate group is added to a peptide. The physicalbinding of phosphoproteins, such as Fas-activatedserine/threonine phosphoprotein (FAST), to regulatorsof alternative splicing has been evidenced by yeasttwo-hybrid screening and biochemical analyses [31].Furthermore, the sensory, motor, integrative, and adap-tive functions of neuron projections are associatedwith the development of a growth cone, which is com-posed primarily of an actin-based cytoskeleton [32].One of the cytoskeleton remodeling genes, Disabled-1(Dab1), has multiple isoforms, as a result of alternative spli-cing [33], that are activated by tyrosine-phosphorylationand play important roles in neuronal positioning byrecruiting a wide range of SH2 domain-containingproteins and activates downstream protein cascadesthrough the Reelin signalling pathway [34]. Deficiencyin Dab1 pathway resulted in a delay in the develop-ment of Purkinje cell dendrites and dysregulation ofthe synaptic markers of parallel fiber and climbingfiber in the cerebellum [35].Fig. 7 Alternative TSSs in contactin associated protein-like 2 (Cntnap2) and experimental validation of its crossover switching events with Real-timePCR. a Schematic DNA structure of Cntnap2, alternative mRNA variants and truncated protein structure of the short isoform. b in situ expression ofCntnap2 in mouse cerebellum at E14.5 (from GenePaint). c HeliscopeCAGE expression data for the two alternative TSSs during cerebellar development.X-axis: time, from embryonic day 11 (E11) to postnatal day 9 (P9). Y-axis: expression level measured in tpm (tags per million). d qRT-PCR expression datademonstrating a crossover TSS switching events between E12 (as well as E15) and P9. X-axis: time at E12, E15 and P9. Y-axis: expression level measuredin RQ (relative quantity against H2O as negative control)Zhang et al. BMC Genomics  (2017) 18:461 Page 11 of 14The dominant TSS usually switches gradually over timeso that only 3.7% of crossover TSS switching are detectedat adjacent time points (357 of 9767 events). However,more than a quarter of the changes at adjacent time pointsoccur between E12-E13 (93 out of 357). This time periodcoincides with key developmental events such as cell spe-cification, cell proliferation of granule cell precursors inthe rhombic lip, as well as the initiation of cells migratingtoward the anterior end of the cerebellum [36].Alternative TSS as post-transcriptional control duringcerebellar developmentAlternative TSSs can produce distinct mRNA isoformsthat have different RNA stability and translational effi-ciency of the mRNA isoforms. For example, VascularEndothelial Growth Factor A (VEGF-A) mRNA stabilityis regulated through alternative initiation codons thatare generated through usage of alternative promoters[37]. We found that two alternative forms of Anp32a aredominantly expressed at different developmental stages inthe cerebellum. The long form has 12 additional aminoacids on the N-terminus compared to the short form. Thisdifference could alter ANP32A protein stability and distri-bution. The role of Anp32a during cerebellar developmentis not known, but it is found to be involved in a variety ofcellular processes in both nucleus and cytoplasm, includingsignaling, apoptosis, protein degradation, and mor-phogenesis [38]. Moreover, Anp32a is known to be akey component of the inhibitor of acetyltransferase(INHAT) complex in the nucleus, involved in regulatingchromatin remodeling or transcription initiation [39].There are suggestions that Anp32a may play importantroles in the brain as the level of Anp32a is increased inAlzheimer’s disease and may be involved in the regulatorymechanism of affecting Tau phosphorylation and impairingthe microtubule network and neurite outgrowth [40].Alternative TSSs can also be a means of producingmRNA isoforms with various mRNA stability and transla-tion efficiency. In the case of Gpc6, we found that its twoforms only differ in mRNA sequence that could affect itsmRNA stability and translation efficiency. Gpc6 is mostabundantly expressed in the ovary, liver, and kidney, withlow level expression in the nervous system [41]. In mice,Gpc6 is critical to modulating the response of the growthplate to thyroid hormones [42]; while in human, muta-tions in the region where Gpc6 resides on Chromosome13 are associated with defects in endochondral ossificationand cause recessive omodysplasia [43].Functional importance of alternative TSS duringcerebellar developmentAlternative TSSs can produce protein isoforms with dis-tinct N-termini; this in turn would lead to alterations inprotein function. An example would be the secreted andmembrane-bound isoforms of mammalian Fos-responsivegene, Fit-1, that are generated and regulated by a pair ofalternative promoters [44]. We found that during cerebel-lar development, the short form of Cntnap2 loses most ofthe functional domains present in the long form – withonly the last laminin G domain retained. Cntnap2 hasbeen found to play a role in the local differentiation of theaxon into distinct functional subdomains [45]. The func-tion of Cntnap2 short form during cerebellar developmentis still to be investigated, but the lack of most functionaldomains suggest its role as a transcriptional suppressor –through mechanism such as non-sense mediated decay[46]; or a functional competitor for the same domainbinding region [47], for Cntnap2 long form counterpartduring early development. During postnatal development,the short form of Cntnap2 ceases to be expressed and thelong (and presumably fully functional) form is maintainedat a steady level. Cntnap2 is strongly associated withautism spectrum disorders, shown in previous studies[48–50]. A knockout mouse for Cntnap2 targeted thegene’s first exon and completely eliminated the expressionof the long form [51], which caused abnormalities in bodysize, neuronal migration and activity, and behaviour. Thusthe knockout has been used as an animal model forautism [52, 53]. However, the short form of Cntnap2should be present in the knockout, and no attention hasbeen directed to the expression of the short form in theknockout. A mutation targeted to the C-terminus wouldbe required to reveal Cntnap2’s overall function in con-sidering both its long and short protein isoforms.ConclusionWe analyzed the cerebellar developmental time coursedata from the FANTOM5 project and identified 9767TSS switching events with temporally specific dominantpromoters. This is the first study to investigate theprevalence of alternative TSS usage during cerebellar de-velopment and their potential roles in transcriptional,post-transcriptional and functional regulation.AcknowledgementWe thank J. Yeung, J. Cairns, S. Tremblay, A. Poon, J. Wilking for support andsuggestions on experimental design and manuscript preparation. We thankF. Lucero Villegas for animal management. We thank M. Larouche, D. Rainsand J. Boyle for technical support. We thank Dora Pak and Anita Sham formanagement support and Miroslav Hatas for systems support. We would liketo thank all members of the FANTOM5 consortium for contributing togeneration of samples and analysis of the data-set and thank GeNAS for dataproduction. We thank GenomeBC, National Institutes of Health, Natural Sciencesand Engineering Research Council of Canada, NeuroDevNet, FANTOM OMICSGroup and University of British Columbia for funding support.FundingThe efforts of PZ, TH, DS and DG was supported by GenomeBC and NationalInstitutes of Health, Natural Sciences and Engineering Research Council ofCanada. FANTOM5 was supported by a Research Grant for RIKEN OmicsScience Center from the Japanese Ministry of Education, Culture, Sports,Science and Technology.Zhang et al. BMC Genomics  (2017) 18:461 Page 12 of 14Availability of data and materialsThe datasets generated and/or analysed during the current study areavailable in the FANTOM5 repository, http://fantom.gsc.riken.jp/zenbu/.Authors’ contributionsPZ, TH, DS and DG generated samples for the time series. The FANTOMconsortium performed HeliScopeCAGE and data processing. ED and PZperformed data analysis. PZ performed biological validation experiments.PZ, DG, ED and WH wrote the manuscript. The authors read and approvedthe final manuscript.Competing interestsThe authors declare that they have no competing interests.Ethics approval and consent to participateThis research was performed with ethics approval from the Canadian Councilon Animal Care and research conducted in accordance with protocol A12–0190.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Author details1Centre for Molecular Medicine and Therapeutics, Child and Family ResearchInstitute, Department of Medical Genetics, University of British Columbia, 950West 28th Avenue, Vancouver, BC V5Z 4H4, Canada. 2Department ofBiostatistics, Harvard School of Public Health, 655 Huntington Ave, Boston,MA 02115, USA. 3Harvard Stem Cell Institute, 1350 Massachusetts Ave,Cambridge, MA 02138, USA. 4Department of Neuroscience, Sheffield Instituteof Translational Neuroscience, University of Sheffield, Room B37 385aGlossop Road, Sheffield, South Yorkshire S10 2HQ, UK.Received: 11 January 2017 Accepted: 31 May 2017References1. Schmutz J, et al. Quality assessment of the human genome sequence.Nature. 2004;429(6990):365–8.2. Davuluri RV, et al. The functional consequences of alternative promoter usein mammalian genomes. Trends Genet. 2008;24(4):167–77.3. Consortium EP. An integrated encyclopedia of DNA elements in the humangenome. Nature. 2012;489(7414):57–74.4. Shiraki T, et al. Cap analysis gene expression for high-throughput analysis oftranscriptional starting point and identification of promoter usage. Proc NatlAcad Sci. 2003;100(26):15776–81.5. Consortium TF. A promoter-level mammalian expression atlas. Nature. 2014;507(7493):462–70.6. Birney E, et al. Identification and analysis of functional elements in 1% ofthe human genome by the ENCODE pilot project. Nature. 2007;447(7146):799–816.7. Carninci P, et al. The transcriptional landscape of the mammalian genome.Science. 2005;309(5740):1559–63.8. Tsuchihara K, et al. Massive transcriptional start site analysis of human genesin hypoxia cells. Nucleic Acids Res. 2009;37(7):2249–63.9. Okazaki Y, et al. Analysis of the mouse transcriptome based on functionalannotation of 60,770 full-length cDNAs. Nature. 2002;420(6915):563–73.10. Ha T, et al. CbGRiTS: Cerebellar gene regulation in time and space. Dev Biol.2015;397(1):18–30.11. Sato A, et al. Cerebellar development transcriptome database (CDT-DB):profiling of spatio-temporal gene expression during the postnataldevelopment of mouse cerebellum. Neural Netw. 2008;21(8):1056–69.12. Arner E, et al. Transcribed enhancers lead waves of coordinated transcription intransitioning mammalian cells. Science. 2015;347(6225):1010–4.13. Silvapulle MJ. Tests against qualitative interaction: exact critical values androbust tests. Biometrics. 2001:57(4)1157–65.14. Rio DC, et al. Purification of RNA using TRIzol (TRI reagent). Cold SpringHarb Protoc. 2010;2010(6):pdb. prot5439.15. Goren A, et al. Chromatin profiling by directly sequencing small quantitiesof immunoprecipitated DNA. Nat Methods. 2010;7(1):47–9.16. Huang DW, Sherman BT, Lempicki RA. Systematic and integrativeanalysis of large gene lists using DAVID bioinformatics resources. NatProtoc. 2009;4(1):44–57.17. Visel A, Thaller C, Eichele G. GenePaint. org: an atlas of gene expressionpatterns in the mouse embryo. Nucleic Acids Res. 2004;32(suppl 1):D552–6.18. Hawrylycz MJ, et al. An anatomically comprehensive atlas of the adulthuman brain transcriptome. Nature. 2012;489(7416):391–9.19. Thierry-Mieg D, Thierry-Mieg J. AceView: a comprehensive cDNA-supportedgene and transcripts annotation. Genome Biol. 2006;7(1):1.20. Hornbeck PV, et al. PhosphoSitePlus, 2014: mutations, PTMs andrecalibrations. Nucleic Acids Res. 2015;43(D1):D512–20.21. Baek D, et al. Characterization and predictive discovery of evolutionarilyconserved mammalian alternative promoters. Genome Res. 2007;17(2):145–55.22. Sun H, et al. MPromDb: an integrated resource for annotation andvisualization of mammalian gene promoters and ChIP-chip experimentaldata. Nucleic Acids Res. 2006;34(suppl 1):D98–D103.23. Takeda, J-I, et al. H-DBAS: alternative splicing database of completelysequenced and manually annotated full-length cDNAs based on H-Invitational. Nucleic acids research, 2007;35(suppl 1): p. D104-D109.24. Landry J-R, Mager DL, Wilhelm BT. Complex controls: the role of alternativepromoters in mammalian genomes. Trends Genet. 2003;19(11):640–8.25. Kimura K, et al. Diversification of transcriptional modulation: large-scaleidentification and characterization of putative alternative promoters ofhuman genes. Genome Res. 2006;16(1):55–65.26. Cooper SJ, et al. Comprehensive analysis of transcriptional promoterstructure and function in 1% of the human genome. Genome Res. 2006;16(1):1–10.27. Tan JS, Mohandas N, Conboy JG. High frequency of alternative first exons inerythroid genes suggests a critical role in regulating gene function. Blood.2006;107(6):2557–61.28. Kim TH, et al. A high-resolution map of active promoters in the humangenome. Nature. 2005;436(7052):876–80.29. Murray-Zmijewski F, Lane D, Bourdon J. p53/p63/p73 isoforms: an orchestraof isoforms to harmonise cell differentiation and response to stress. CellDeath Differ. 2006;13(6):962–72.30. Hu J-M, et al. Functional analyses of albumin expression in a series ofhepatocyte cell lines and in primary hepatocytes. Cell Growth Differ. 1992;3:577.31. Simarro M, et al. Fas-activated serine/threonine phosphoprotein (FAST) is aregulator of alternative splicing. Proc Natl Acad Sci. 2007;104(27):11370–5.32. Schaefer AW, Kabir N, Forscher P. Filopodia and actin arcs guide theassembly and transport of two populations of microtubules with uniquedynamic parameters in neuronal growth cones. J Cell Biol. 2002;158(1):139–52.33. Katyal S, Godbout R. Alternative splicing modulates Disabled-1 (Dab1)function in the developing chick retina. EMBO J. 2004;23(8):1878–88.34. Rice DS, Curran T. Role of the reelin signaling pathway in central nervoussystem development. Annu Rev Neurosci. 2001;24(1):1005–39.35. Qiao S, et al. Dab2IP GTPase activating protein regulates dendritedevelopment and synapse number in cerebellum. PLoS One. 2013;8(1):e53635.36. Ben-Arie N, et al. Math1 is essential for genesis of cerebellar granuleneurons. Nature. 1997;390(6656):169–71.37. Arcondéguy T, et al. VEGF-A mRNA processing, stability and translation: aparadigm for intricate regulation of gene expression at the post-transcriptional level. Nucleic Acids Res. 2013;41(17):7997–8010.38. Wang S, et al. The expression and distributions of ANP32A in the developingbrain. Biomed Res Int. 2015;2015:207347. doi:10.1155/2015/207347.39. Kadota S, Nagata K. pp32, an INHAT component, is a transcriptionmachinery recruiter for maximal induction of IFN-stimulated genes. J CellSci. 2011;124(6):892–9.40. Chen S, et al. I PP2A 1 affects Tau phosphorylation via association with thecatalytic subunit of protein phosphatase 2A. J Biol Chem. 2008;283(16):10513–21.41. Veugelers M, et al. Glypican-6, a new member of the glypican family of cellsurface heparan sulfate proteoglycans. J Biol Chem. 1999;274(38):26968–77.42. Bassett J, et al. Thyroid hormone regulates heparan sulfate proteoglycanexpression in the growth plate. Endocrinology. 2006;147(1):295–305.43. Campos-Xavier AB, et al. Mutations in the heparan-sulfate proteoglycanglypican 6 (GPC6) impair endochondral ossification and cause recessiveomodysplasia. Am J Hum Genet. 2009;84(6):760–70.44. Bergers G, et al. Alternative promoter usage of the Fos-responsive gene Fit-1generates mRNA isoforms coding for either secreted or membrane-boundproteins related to the IL-1 receptor. EMBO J. 1994;13(5):1176.Zhang et al. BMC Genomics  (2017) 18:461 Page 13 of 1445. Poliak S, et al. Caspr2, a new member of the neurexin superfamily, is localizedat the juxtaparanodes of myelinated axons and associates with K+ channels.Neuron. 1999;24(4):1037–47.46. Anczuków O, et al. Does the nonsense-mediated mRNA decay mechanismprevent the synthesis of truncated BRCA1, CHK2, and p53 proteins? HumMutat. 2008;29(1):65–73.47. Darieva Z, et al. A competitive transcription factor binding mechanismdetermines the timing of late cell cycle-dependent gene expression. Mol Cell.2010;38(1):29–40.48. Alarcón M, et al. Linkage, association, and gene-expression analysesidentify CNTNAP2 as an autism-susceptibility gene. Am J Hum Genet.2008;82(1):150–9.49. Arking DE, et al. A common genetic variant in the neurexin superfamilymember CNTNAP2 increases familial risk of autism. Am J Hum Genet. 2008;82(1):160–4.50. Bakkaloglu B, et al. Molecular cytogenetic analysis and resequencing ofcontactin associated protein-like 2 in autism spectrum disorders. Am J HumGenet. 2008;82(1):165–73.51. Poliak S, et al. Juxtaparanodal clustering of shaker-like K+ channels inmyelinated axons depends on Caspr2 and TAG-1. J Cell Biol. 2003;162(6):1149–60.52. Ellegood J, et al. Clustering autism: using neuroanatomical differences in 26mouse models to gain insight into the heterogeneity. Mol Psychiatry. 2015;20(1):118–25.53. Kloth AD, et al. Cerebellar associative sensory learning defects in five mouseautism models. elife. 2015;4:e06085.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submitSubmit your next manuscript to BioMed Central and we will help you at every step:Zhang et al. BMC Genomics  (2017) 18:461 Page 14 of 14


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items