- Research article
- Open Access
Relatively frequent switching of transcription start sites during cerebellar development
- Peter Zhang†1,
- Emmanuel Dimont†2, 3,
- Thomas Ha1,
- Douglas J. Swanson1,
- the FANTOM Consortium,
- Winston Hide2, 3, 4 and
- Dan Goldowitz1Email author
© The Author(s). 2017
- Received: 11 January 2017
- Accepted: 31 May 2017
- Published: 13 June 2017
The Correction to this article has been published in BMC Genomics 2018 19:39
Alternative transcription start site (TSS) usage plays important roles in transcriptional control of mammalian gene expression. The growing interest in alternative TSSs and their role in genome diversification spawned 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 central nervous system, is largely lacking.
In this study, We have prepared a unique set of time-course data for the developing cerebellum, as part of the FANTOM5 consortium (http://fantom.gsc.riken.jp/5/) that uses their innovative capturing of 5′ ends of all transcripts 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 isoforms that emerged from the same gene. We developed a mathematical method that systematically compares the expression of different TSSs of a gene to identify temporal crossover and non-crossover switching events. We identified 48,489 novel TSS switching events in 5433 genes during cerebellar development. This includes 9767 crossover TSS switching events in 1511 genes, where the dominant TSS shifts over time.
We observed a relatively high prevalence of TSS switching in cerebellar development where the resulting temporally-specific gene transcripts and protein products can play important regulatory and functional roles.
- Developmental biology
- Promoter switching
- Alternative promoters
- Alternative splicing
- Transcription start site
Alternative splicing can provide a large reservoir of transcriptional variants from the ~22,000 genes identified by the Human Genome Project . The production of different isoforms due to the usage of alternative transcription start sites (TSSs), which was once considered as uncommon, has now been found in the majority of human genes [2, 3]. Alternative TSSs could be results of a gene duplication event followed by the loss of functional exons in the upstream copy and diversification of the two duplicated promoters. Alternative TSS usage can affect gene expression and generate diversity in a variety of ways. On the transcriptional level, alternative TSS could result in tissue-specific expression, temporally regulated expression, and the amplitude of expression. On the post-transcriptional level, alternative TSS can affect the stability and translational efficiency of the mRNA. Furthermore, alternative TSS can result in protein isoforms with a different amino terminus, which can lead to alterations in protein levels, functions, or subcellular distribution. Therefore, the investigation of temporal switching of TSSs can provide insights into the regulation of different protein isoforms, and presumably their differences in function. One way to optimally identify differential use of isoforms is to examine transcriptional regulation over developmental time.
One high-throughput technique to survey gene expression at the transcriptome level is Cap Analysis Gene Expression (CAGE) which generates a genome-wide expression profile based on sequences from the 5′ end of the mRNA [4, 5]. In the FANTOM project, CAGE has been shown to identify different TSSs and the corresponding promoters for single genes [6–9]. With CAGE data, one can infer the TSS usage through the number of transcripts produced at that particular TSS. When more than one TSS is used at a single time point from a single gene, the TSS with highest expression is considered the “dominant” TSS. The understanding of how the TSS usage changes during development can shed light on how a single gene can function differently over developmental stages through temporally regulated alternative mRNA and protein isoforms.
The complexity of brain development requires intricately controlled expression of specific genes across time. The cerebellum is often used as a model in analyses of brain development due to its limited number of major cell types. These cells are positioned in spatially defined territories of the developing cerebellum. The cerebellum has also been the focus of two extensive genome-wide gene expression profiling of the developing cerebellum [10, 11]. Detailed information on temporally regulated promoter usage of developmentally important genes - which is still largely lacking - can provide valuable information on genome diversity. Moreover, different isoforms of these genes may be translated into distinct protein products that perform different tasks. Such analyses would give insight to the alterations made to the form of the final transcript, localization for transcription factors motif prediction, utilization, and associated regulatory network changes. Thus, in collaboration with the FANTOM5 project , we generated a CAGE dataset for the developing cerebellum with 12 time points to study temporally-regulated gene expression and alternative TSS usage during cerebellar development.
TSS switching events across samples were systematically identified by comparing differential promoter transcription levels between pairs of TSSs and pairs of developmental time points, and by applying the Silvapulle FQ test, a statistical method for constrained hypothesis testing that we specifically apply for the detection of crossover TSS switching events . The FQ test produces p-values to estimate significance of a crossover switching event. We have applied the FQ test to our cerebellar time series to identify novel TSS switching events during cerebellar development. Our hypothesis was that differential TSS usage can result in significant regulatory changes that underlie cellular events critical for cerebellar development and morphogenesis. By taking advantage of the FANTOM5 collaboration with our cerebellar developmental time course, we identified 48,489 novel TSS switching events, including 9767 events in which the dominant TSS shifts over time. These TSS switching events were predicted to produce temporally-specific gene transcripts and protein products that can play important regulatory and functional roles during cerebellar development.
Mouse colony maintenance and breeding
This research was performed with ethics approval from the Canadian Council on Animal Care and research conducted in accordance with protocol A12–0190. C57BL/6 J mice were used in all experiments and were imported from The Jackson Laboratory (Maine, US) and maintained in our colony as an inbred line. To standardize the time of conception, timed pregnancies were set up. Every weekday at 10:00 am, females were coupled with male; at 3:00 pm, the females were checked for vaginal plugs and removed from their partners. The appearance of a vaginal plug was recorded as the day of conception (i.e. embryonic day 0) and embryos were collected at 10 am on embryonic day 11–18 (E11-E18) every day and postnatal day 0–9 (P0-P9) every 3 days for a total of 12 time points in our cerebellar time series.
On the day of embryo collection, the mothers were sacrificed and embryos were removed from the uterus in ice-cold RNAse-free PBS. Cerebella were dissected from the head of the embryos, then pooled with littermates, and snap-frozen in liquid nitrogen. Three replicate pools of whole cerebella samples were collected at each time point. The standard TRIzol RNA extraction protocol  was used for tissue homogenization and RNA extraction.
A Bioanalyzer (Agilent, Santa Clara, CA) was used to examine RNA quality. All RNA samples used for the time series achieved high RNA Integrity (RIN) scores above 9.0. The samples were sent to RIKEN Omics Center at Yokohama, Japan, as part of Functional Annotation of the Mammalian Genome 5 (FANTOM5) collaboration for CAGE analysis.
Transcriptome library generation by HeliScopeCAGE
TSS switch detection
TSS switching events are detected by comparing the expression of transcripts from two TSSs of a single gene at two time points. The difference in expression level of the two TSSs is designated d1 and d2 at time point 1 and time point 2, respectively. The null hypothesis is that there is no switching for the two TSSs (d1 = d2, see Fig. 1b). The test of this hypothesis was performed using the standard t-test, with candidate switching events identified at this preliminary stage if the adjusted p-value was <0.2. A non-crossover TSS event is detected if one TSS is used more frequently at one time point compared to the other, but the same TSS is used dominantly at both time points (d1 > d2, or d1 < d2, both d1 and d2 same sign, Fig. 1b). A crossover TSS switching event is detected if one TSS is used more frequently at one time point compared to the other, and that the dominant TSS switches at the two time points (d1 > 0 and d2 < 0 or d1 < 0 and d2 > 0, Fig. 1b). In order to reduce potential confounding of TSS switching events by differential aggregate promoter expression between time points, candidate events were further limited to TSS pairs that do not change in overall mean expression between developmental stages being compared. The null hypothesis tested at this stage is that the mean TSS expression at the two time points is equal, and results were filtered out if the t-test adjusted p-value was <0.1.
In addition to the differences in expression (d1,d2), the results of TSS switching are represented using the FQ statistic  which formally tests for the presence of crossover switching for each gene. The test of the null hypothesis of no differential crossover promoter usage corresponds to a test involving the FQ statistic, which is functionally similar to the ANOVA F-test. Exact p-values for this test are obtained as described in Silvapulle . To our knowledge, the Silvapulle FQ test is the only statistical test available that was specifically developed for testing hypotheses regarding qualitative interaction, and which we apply in the current study for testing the presence of crossover switching in gene promoter usage.
All P-values are adjusted for multiple comparisons using the Benjamini–Hochberg method to control the false discovery rate. The P-value of the FQ test was used as an indicator of significance for choosing biological validation candidates.
Gene ontology analysis for gene with crossover switching events
To identify cellular processes and molecular pathways in genes with crossover TSS switching events, we used Database for Annotation, Visualization and Integrated Discovery program (DAVID, https://david.ncifcrf.gov/ ) to examine the gene ontology of genes with at least one crossover event with p < 0.05 in FQ test. Top 20 GO terms were used for overall analysis in crossover TSS switching genes during cerebellar development. Furthermore, for temporal functional analysis of crossover TSS switching events, top 20 GO terms were generated with DAVID for all events associated with three developmental time points – E13, E15 and P0.
In silico validation of gene expression with established databases and experimental validation with gene structure prediction and quantitative real-time PCR
We used online databases to examine the 20 genes with the lowest p-values. First, we used in situ resources - Genepaint (http://genepaint.org ) and Allen Brain Atlas (http://www.brain-map.org ) to examine the genes’ expression in the cerebellum. Second, we examined the predicted mRNA structures from the two TSSs with the intron/exon database Aceview (http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/ ) as well as functional domains of their protein products from protein domain database PhosphoSitePlus (http://www.phosphosite.org ) to determine the potential effect of TSS switching events on biological function.
Three genes were chosen for further validation for TSS-specific quantitative real-time PCR for the validation of alteration in TSS usage at E12, E15 and P9. Cerebellar RNA was extracted from C57BL/6 J mice at E12, E15 and P9 following the same procedure that were used for HeliScopeCAGE RNA collection. cDNAs were produced with random hexamers using the High Capacity cDNA Archive Kit (Applied Biosystems). cDNA products were diluted to 100 ng total RNA input. Sequences of the transcript of interest were loaded into Primer Express® software (Applied Biosystems). For each gene, an isoform-specific forward primer was designed for each of the long and short isoform, while the reverse primer aligns to a common sequence that is shared by both isoforms. Amplicon lengths were between 80 and 120 bp. The qPCR was performed with the FAST SYBR Green PCR Master Mix (Applied Biosystems) on an ABI StepOne Plus Sequence Detection System (Applied Biosystems). All runs were normalized to the control gene, Gapdh. Three biological replicates were prepared for each gene target and three technical replicates were performed for each biological replicate. Gene expression was represented as relative quantity against the negative control which used water as the template (noted as “Relative Quantity vs. H2O” in figures). The results of Real-Time PCR were analyzed and graphed by ABI StepOne Plus Sequence Detection System (Applied Biosystems). The expression data were compared with the HeliScope-CAGE data.
Overview of promoter switch events during cerebellar development
Comparison of TSS switching events during cerebellar development with other FANTOM5 datasets
Human iPS to neuron (wt) 1
Trachea epithelia differentiation
Epithelial to mesenchymal
BMM TB activation IL13
AoSMC response to IL1b
Macrophage response to LPS
ES to cardiomyocyte
Myoblast to myotube
In conclusion, cerebellar development showed a high frequency in crossover TSS switching among datasets with a high number of total switching events.
Distribution of TSS switching events in cerebellar transcriptome
Top 20 genes with highest numbers of TSS switching events
When comparing the distribution of crossover and non-crossover events, we found that crossover switching events are clustered in fewer genes when compared with non-crossover events. Since the frequency of non-crossover switching is about four times the number of cross-over (38,722:9767 or 3.96:1), we would expect roughly a 4:1 ratio 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 Dlg2 (223 non-crossover events and 43 crossover events, Table 2). However, for the majority of the 20 genes with the greatest number of switching events, the frequency of crossover events is much higher than one fourth of the non-crossover counterpart, such as the two outlier genes mentioned above - Frmd4a (509 non-crossover events vs 343 crossover events) and Ank3 (464 non-crossover events vs 337 crossover events, Table 2). This un-even distribution of crossover events is also reflected by the lower abundance of genes with a low number of switching events – 3052 genes have less than 3 non-crossover events (Fig. 3b) and only 944 genes have less than 3 crossover events (Fig. 3c). In conclusion, we found that crossover events tend to cluster in a fewer number of genes when compared to the non-crossover counterpart.
Gradual increment in the number of crossover TSS switching events over developmental time
Distribution of crossover TSS switching events across time in cerebellar development (N = 9767)
To examine the general pattern of TSS switching during cerebellar development, we counted promoter switch events by developmental time points (Table 3). Among the 12 data points in our time course, a total of 66 comparisons between two data points have been carried out to search for the switching of alternative TSSs (Table 3). Over the time series, there is a general incremental number of crossover switching events that are detected between two samples that are temporally distant. This most likely reflects the gradual shift of cerebellar transcriptome and TSS usage during development. There are rare exceptions to this pattern, for example, there are more switching events between E11 and E17 samples than found between E11 and E18 samples.
Gene ontology analysis for genes with the most significant crossover TSS switching events
We have found that the largest alteration in gene expression occurs at E13, E15 and P0 (manuscript in preparation) and were interested to determine the extent that crossover TSS switching plays a role in transcriptome diversity. When comparing crossover events at E13 with all other time points we find 1440 significant (p < .05) events in 584 genes. When comparing crossover events at E15 with other time points we find 1355 significant (p < 0.05) events in 582 genes. Finally, when comparing crossover events at P0 with all other time points we find 1152 significant (p < .05) events in 506 genes. We used these gene lists as input to DAVID and the top 20 terms were selected for these temporal comparisons among the three time points (Fig. 4b). We found that 7 terms (phosphoprotein, alternative splicing, splice variant, cytoplasm, neuron projection, cytoskeletal protein binding and cytoskeleton) were shared among each of the three time points. These 7 GO terms were also found among the 8 most significant terms in the analysis with all genes discussed previously. We also observed that comparisons between shorter time spans yield more common GO terms –e.g., there are 5 terms shared between genes with crossover TSS events at E13 and E15, 1 term between E15 and P0 and no terms were common between E13 and P0. Lastly, the majority of GO terms unique to a given time point shared a common theme that may reflect active biological process occurring at the given time – e.g., four out of eight E13 terms were associated with cell motion and cytoskeleton; five out of seven E15 terms were associated with ion binding and six out of twelve P0 terms were associated with regulation of intracellular organization.
Validation of promoter switching events
Cerebellar expression patterns of genes with most significant switching events at E14.5 from the in situ database, Genepaint
discs, large homolog 3
solute carrier family 12, member 5
NE, interior cerebellum
IQ motif and Sec7 domain 1
contactin associated protein-like 2
canopy 1 homolog
mitogen activated protein kinase 8 interacting protein 1
specific cerebellar nuclei, spinal cord
discs, large homolog-associated protein 4
ankyrin 3, epithelial
calcium channel, voltage-dependent, beta 4 subunit
acidic (leucine-rich) nuclear phosphoprotein 32 family
strong, EGL & NE specific staining
thioredoxin-related transmembrane protein 3
amyloid beta (A4) precursor protein-binding, family B, member 3
protein arginine N-methyltransferase 8
Mus musculus endothelin receptor type B
strong NE specific staining
sema domain 4G
zinc finger, RAN-binding domain containing 1
zinc finger and BTB domain containing 38
inhibitor of Bruton agammaglobulinemia tyrosine kinase
Strong NE, NTZ specific staining
heat shock 105 kDa/110 kDa protein 1
Mus musculus zinc finger protein 451
moderate EGL staining
GRAM domain containing 1B
High prevalence of alternative TSSs in mammalian genomes
In this study, we have identified 5293 genes (~21% of a total of 25,207 genes) that exhibit differential TSS usage during cerebellar development. These findings are in line with previous studies and indicate that TSS switching events are common and can play an important role in the diversity of the cerebellar transcriptome during development [21–23]. Furthermore, we have identified 9767 crossover TSS switching events which suggests an alteration in the dominant TSS over time. Since the alternative mRNA isoforms could be translated into functionally different products, a crossover switching event suggests that one gene can play different roles at different time points in development.
Alternative usage of multiple TSSs of one gene is common in mammalian genomes. It is a key mechanism to increase mRNA and protein diversity since multiple mRNAs from a single gene can encode distinct protein isoforms with different functions (reviewed in ). Recent studies suggest that about half of the mouse genes have multiple alternative promoters [25, 26]. For example, alternative promoters have been identified in >20% of genes in ENCODE (http://genome.ucsc.edu/ENCODE/) regions . Other genomic studies also found more than a quarter of human genes having multiple active promoters [27–29]. The complex transcriptional regulation of alternative promoter usage has been identified in several genes . Furthermore, in some genes, such as tumor protein p53 (TP53) and guanine nucleotide binding protein (GNAS), alternative promoters were shown to be activated or silenced . However, the focus of previous studies has been the tissue-specific transcriptional regulation of alternative promoters; the temporal aspect of alternative promoter usage during cerebellar development has been overlooked. Our analyses focused on the switching usage of alternative promoter in the mouse cerebellum, and this is the first systematic study of alternative promoter usage in the development of the mouse cerebellum.
Temporal regulation of alternative TSS associated with developmental processes in the cerebellum
Alternative TSSs reflect different promoter regions that can be used for tissue-specific and/or temporal-specific expression. For example, albumin in hepatocytes has several cis-acting elements that recruit different sets of trans-acting factors, which enable spatial, temporal and dynamics regulation of the transcription of albumin mRNA . In this study, we have identified 9767 crossover TSS switching events in 1511 genes. Thus, in ~20% of genes there is more than one promoter that is used dominantly during cerebellar development. Functional annotation analysis for these genes revealed GO terms that are expected to be associated with alternative promoter usage, such as “alternative splicing” and “splicing variants”, as well GO terms that point to processes where promoter switching might play a role during development, such as “phosphoprotein”, “cytoskeleton organization” and “neuron projection”. Phosphoproteins are involved in the post-translational regulatory process phosphorylation, in which a phosphate group is added to a peptide. The physical binding of phosphoproteins, such as Fas-activated serine/threonine phosphoprotein (FAST), to regulators of alternative splicing has been evidenced by yeast two-hybrid screening and biochemical analyses . Furthermore, the sensory, motor, integrative, and adaptive functions of neuron projections are associated with the development of a growth cone, which is composed primarily of an actin-based cytoskeleton . One of the cytoskeleton remodeling genes, Disabled-1 (Dab1), has multiple isoforms, as a result of alternative splicing , that are activated by tyrosine-phosphorylation and play important roles in neuronal positioning by recruiting a wide range of SH2 domain-containing proteins and activates downstream protein cascades through the Reelin signalling pathway . Deficiency in Dab1 pathway resulted in a delay in the development of Purkinje cell dendrites and dysregulation of the synaptic markers of parallel fiber and climbing fiber in the cerebellum .
The dominant TSS usually switches gradually over time so that only 3.7% of crossover TSS switching are detected at adjacent time points (357 of 9767 events). However, more than a quarter of the changes at adjacent time points occur between E12-E13 (93 out of 357). This time period coincides with key developmental events such as cell specification, cell proliferation of granule cell precursors in the rhombic lip, as well as the initiation of cells migrating toward the anterior end of the cerebellum .
Alternative TSS as post-transcriptional control during cerebellar development
Alternative TSSs can produce distinct mRNA isoforms that have different RNA stability and translational efficiency of the mRNA isoforms. For example, Vascular Endothelial Growth Factor A (VEGF-A) mRNA stability is regulated through alternative initiation codons that are generated through usage of alternative promoters . We found that two alternative forms of Anp32a are dominantly expressed at different developmental stages in the cerebellum. The long form has 12 additional amino acids on the N-terminus compared to the short form. This difference could alter ANP32A protein stability and distribution. The role of Anp32a during cerebellar development is not known, but it is found to be involved in a variety of cellular processes in both nucleus and cytoplasm, including signaling, apoptosis, protein degradation, and morphogenesis . Moreover, Anp32a is known to be a key component of the inhibitor of acetyltransferase (INHAT) complex in the nucleus, involved in regulating chromatin remodeling or transcription initiation . There are suggestions that Anp32a may play important roles in the brain as the level of Anp32a is increased in Alzheimer’s disease and may be involved in the regulatory mechanism of affecting Tau phosphorylation and impairing the microtubule network and neurite outgrowth .
Alternative TSSs can also be a means of producing mRNA isoforms with various mRNA stability and translation efficiency. In the case of Gpc6, we found that its two forms only differ in mRNA sequence that could affect its mRNA stability and translation efficiency. Gpc6 is most abundantly expressed in the ovary, liver, and kidney, with low level expression in the nervous system . In mice, Gpc6 is critical to modulating the response of the growth plate to thyroid hormones ; while in human, mutations in the region where Gpc6 resides on Chromosome 13 are associated with defects in endochondral ossification and cause recessive omodysplasia .
Functional importance of alternative TSS during cerebellar development
Alternative TSSs can produce protein isoforms with distinct N-termini; this in turn would lead to alterations in protein function. An example would be the secreted and membrane-bound isoforms of mammalian Fos-responsive gene, Fit-1, that are generated and regulated by a pair of alternative promoters . We found that during cerebellar development, the short form of Cntnap2 loses most of the functional domains present in the long form – with only the last laminin G domain retained. Cntnap2 has been found to play a role in the local differentiation of the axon into distinct functional subdomains . The function of Cntnap2 short form during cerebellar development is still to be investigated, but the lack of most functional domains suggest its role as a transcriptional suppressor – through mechanism such as non-sense mediated decay ; or a functional competitor for the same domain binding region , for Cntnap2 long form counterpart during early development. During postnatal development, the short form of Cntnap2 ceases to be expressed and the long (and presumably fully functional) form is maintained at a steady level. Cntnap2 is strongly associated with autism spectrum disorders, shown in previous studies [48–50]. A knockout mouse for Cntnap2 targeted the gene’s first exon and completely eliminated the expression of the long form , which caused abnormalities in body size, neuronal migration and activity, and behaviour. Thus the knockout has been used as an animal model for autism [52, 53]. However, the short form of Cntnap2 should be present in the knockout, and no attention has been directed to the expression of the short form in the knockout. A mutation targeted to the C-terminus would be required to reveal Cntnap2’s overall function in considering both its long and short protein isoforms.
We analyzed the cerebellar developmental time course data from the FANTOM5 project and identified 9767 TSS switching events with temporally specific dominant promoters. This is the first study to investigate the prevalence of alternative TSS usage during cerebellar development and their potential roles in transcriptional, post-transcriptional and functional regulation.
We thank J. Yeung, J. Cairns, S. Tremblay, A. Poon, J. Wilking for support and suggestions on experimental design and manuscript preparation. We thank F. Lucero Villegas for animal management. We thank M. Larouche, D. Rains and J. Boyle for technical support. We thank Dora Pak and Anita Sham for management support and Miroslav Hatas for systems support. We would like to thank all members of the FANTOM5 consortium for contributing to generation of samples and analysis of the data-set and thank GeNAS for data production. We thank GenomeBC, National Institutes of Health, Natural Sciences and Engineering Research Council of Canada, NeuroDevNet, FANTOM OMICS Group and University of British Columbia for funding support.
The efforts of PZ, TH, DS and DG was supported by GenomeBC and National Institutes of Health, Natural Sciences and Engineering Research Council of Canada. FANTOM5 was supported by a Research Grant for RIKEN Omics Science Center from the Japanese Ministry of Education, Culture, Sports, Science and Technology.
Availability of data and materials
The datasets generated and/or analysed during the current study are available in the FANTOM5 repository, http://fantom.gsc.riken.jp/zenbu/.
PZ, TH, DS and DG generated samples for the time series. The FANTOM consortium performed HeliScopeCAGE and data processing. ED and PZ performed data analysis. PZ performed biological validation experiments. PZ, DG, ED and WH wrote the manuscript. The authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Ethics approval and consent to participate
This research was performed with ethics approval from the Canadian Council on Animal Care and research conducted in accordance with protocol A12–0190.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the 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.
- Schmutz J, et al. Quality assessment of the human genome sequence. Nature. 2004;429(6990):365–8.View ArticlePubMedGoogle Scholar
- Davuluri RV, et al. The functional consequences of alternative promoter use in mammalian genomes. Trends Genet. 2008;24(4):167–77.View ArticlePubMedGoogle Scholar
- Consortium EP. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489(7414):57–74.View ArticleGoogle Scholar
- Shiraki T, et al. Cap analysis gene expression for high-throughput analysis of transcriptional starting point and identification of promoter usage. Proc Natl Acad Sci. 2003;100(26):15776–81.View ArticlePubMedPubMed CentralGoogle Scholar
- Consortium TF. A promoter-level mammalian expression atlas. Nature. 2014;507(7493):462–70.View ArticleGoogle Scholar
- Birney E, et al. Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature. 2007;447(7146):799–816.View ArticlePubMedGoogle Scholar
- Carninci P, et al. The transcriptional landscape of the mammalian genome. Science. 2005;309(5740):1559–63.View ArticlePubMedGoogle Scholar
- Tsuchihara K, et al. Massive transcriptional start site analysis of human genes in hypoxia cells. Nucleic Acids Res. 2009;37(7):2249–63.View ArticlePubMedPubMed CentralGoogle Scholar
- Okazaki Y, et al. Analysis of the mouse transcriptome based on functional annotation of 60,770 full-length cDNAs. Nature. 2002;420(6915):563–73.View ArticlePubMedGoogle Scholar
- Ha T, et al. CbGRiTS: Cerebellar gene regulation in time and space. Dev Biol. 2015;397(1):18–30.View ArticlePubMedGoogle Scholar
- Sato A, et al. Cerebellar development transcriptome database (CDT-DB): profiling of spatio-temporal gene expression during the postnatal development of mouse cerebellum. Neural Netw. 2008;21(8):1056–69.View ArticlePubMedGoogle Scholar
- Arner E, et al. Transcribed enhancers lead waves of coordinated transcription in transitioning mammalian cells. Science. 2015;347(6225):1010–4.View ArticlePubMedPubMed CentralGoogle Scholar
- Silvapulle MJ. Tests against qualitative interaction: exact critical values and robust tests. Biometrics. 2001:57(4)1157–65.Google Scholar
- Rio DC, et al. Purification of RNA using TRIzol (TRI reagent). Cold Spring Harb Protoc. 2010;2010(6):pdb. prot5439.View ArticlePubMedGoogle Scholar
- Goren A, et al. Chromatin profiling by directly sequencing small quantities of immunoprecipitated DNA. Nat Methods. 2010;7(1):47–9.View ArticlePubMedGoogle Scholar
- Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4(1):44–57.View ArticleGoogle Scholar
- Visel A, Thaller C, Eichele G. GenePaint. org: an atlas of gene expression patterns in the mouse embryo. Nucleic Acids Res. 2004;32(suppl 1):D552–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Hawrylycz MJ, et al. An anatomically comprehensive atlas of the adult human brain transcriptome. Nature. 2012;489(7416):391–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Thierry-Mieg D, Thierry-Mieg J. AceView: a comprehensive cDNA-supported gene and transcripts annotation. Genome Biol. 2006;7(1):1.View ArticlePubMedGoogle Scholar
- Hornbeck PV, et al. PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res. 2015;43(D1):D512–20.View ArticlePubMedGoogle Scholar
- Baek D, et al. Characterization and predictive discovery of evolutionarily conserved mammalian alternative promoters. Genome Res. 2007;17(2):145–55.View ArticlePubMedPubMed CentralGoogle Scholar
- Sun H, et al. MPromDb: an integrated resource for annotation and visualization of mammalian gene promoters and ChIP-chip experimental data. Nucleic Acids Res. 2006;34(suppl 1):D98–D103.View ArticlePubMedGoogle Scholar
- Takeda, J-I, et al. H-DBAS: alternative splicing database of completely sequenced and manually annotated full-length cDNAs based on H-Invitational. Nucleic acids research, 2007;35(suppl 1): p. D104-D109.Google Scholar
- Landry J-R, Mager DL, Wilhelm BT. Complex controls: the role of alternative promoters in mammalian genomes. Trends Genet. 2003;19(11):640–8.View ArticlePubMedGoogle Scholar
- Kimura K, et al. Diversification of transcriptional modulation: large-scale identification and characterization of putative alternative promoters of human genes. Genome Res. 2006;16(1):55–65.View ArticlePubMedPubMed CentralGoogle Scholar
- Cooper SJ, et al. Comprehensive analysis of transcriptional promoter structure and function in 1% of the human genome. Genome Res. 2006;16(1):1–10.View ArticlePubMedPubMed CentralGoogle Scholar
- Tan JS, Mohandas N, Conboy JG. High frequency of alternative first exons in erythroid genes suggests a critical role in regulating gene function. Blood. 2006;107(6):2557–61.View ArticlePubMedPubMed CentralGoogle Scholar
- Kim TH, et al. A high-resolution map of active promoters in the human genome. Nature. 2005;436(7052):876–80.View ArticlePubMedPubMed CentralGoogle Scholar
- Murray-Zmijewski F, Lane D, Bourdon J. p53/p63/p73 isoforms: an orchestra of isoforms to harmonise cell differentiation and response to stress. Cell Death Differ. 2006;13(6):962–72.View ArticlePubMedGoogle Scholar
- Hu J-M, et al. Functional analyses of albumin expression in a series of hepatocyte cell lines and in primary hepatocytes. Cell Growth Differ. 1992;3:577.PubMedGoogle Scholar
- Simarro M, et al. Fas-activated serine/threonine phosphoprotein (FAST) is a regulator of alternative splicing. Proc Natl Acad Sci. 2007;104(27):11370–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Schaefer AW, Kabir N, Forscher P. Filopodia and actin arcs guide the assembly and transport of two populations of microtubules with unique dynamic parameters in neuronal growth cones. J Cell Biol. 2002;158(1):139–52.View ArticlePubMedPubMed CentralGoogle Scholar
- Katyal S, Godbout R. Alternative splicing modulates Disabled-1 (Dab1) function in the developing chick retina. EMBO J. 2004;23(8):1878–88.View ArticlePubMedPubMed CentralGoogle Scholar
- Rice DS, Curran T. Role of the reelin signaling pathway in central nervous system development. Annu Rev Neurosci. 2001;24(1):1005–39.View ArticlePubMedGoogle Scholar
- Qiao S, et al. Dab2IP GTPase activating protein regulates dendrite development and synapse number in cerebellum. PLoS One. 2013;8(1):e53635.View ArticlePubMedPubMed CentralGoogle Scholar
- Ben-Arie N, et al. Math1 is essential for genesis of cerebellar granule neurons. Nature. 1997;390(6656):169–71.View ArticlePubMedGoogle Scholar
- Arcondéguy T, et al. VEGF-A mRNA processing, stability and translation: a paradigm for intricate regulation of gene expression at the post-transcriptional level. Nucleic Acids Res. 2013;41(17):7997–8010.View ArticlePubMedPubMed CentralGoogle Scholar
- Wang S, et al. The expression and distributions of ANP32A in the developing brain. Biomed Res Int. 2015;2015:207347. doi:10.1155/2015/207347.
- Kadota S, Nagata K. pp32, an INHAT component, is a transcription machinery recruiter for maximal induction of IFN-stimulated genes. J Cell Sci. 2011;124(6):892–9.View ArticlePubMedGoogle Scholar
- Chen S, et al. I PP2A 1 affects Tau phosphorylation via association with the catalytic subunit of protein phosphatase 2A. J Biol Chem. 2008;283(16):10513–21.View ArticlePubMedPubMed CentralGoogle Scholar
- Veugelers M, et al. Glypican-6, a new member of the glypican family of cell surface heparan sulfate proteoglycans. J Biol Chem. 1999;274(38):26968–77.View ArticlePubMedGoogle Scholar
- Bassett J, et al. Thyroid hormone regulates heparan sulfate proteoglycan expression in the growth plate. Endocrinology. 2006;147(1):295–305.View ArticlePubMedGoogle Scholar
- Campos-Xavier AB, et al. Mutations in the heparan-sulfate proteoglycan glypican 6 (GPC6) impair endochondral ossification and cause recessive omodysplasia. Am J Hum Genet. 2009;84(6):760–70.View ArticlePubMedPubMed CentralGoogle Scholar
- Bergers G, et al. Alternative promoter usage of the Fos-responsive gene Fit-1 generates mRNA isoforms coding for either secreted or membrane-bound proteins related to the IL-1 receptor. EMBO J. 1994;13(5):1176.PubMedPubMed CentralGoogle Scholar
- Poliak S, et al. Caspr2, a new member of the neurexin superfamily, is localized at the juxtaparanodes of myelinated axons and associates with K+ channels. Neuron. 1999;24(4):1037–47.View ArticlePubMedGoogle Scholar
- Anczuków O, et al. Does the nonsense-mediated mRNA decay mechanism prevent the synthesis of truncated BRCA1, CHK2, and p53 proteins? Hum Mutat. 2008;29(1):65–73.View ArticlePubMedGoogle Scholar
- Darieva Z, et al. A competitive transcription factor binding mechanism determines the timing of late cell cycle-dependent gene expression. Mol Cell. 2010;38(1):29–40.View ArticlePubMedPubMed CentralGoogle Scholar
- Alarcón M, et al. Linkage, association, and gene-expression analyses identify CNTNAP2 as an autism-susceptibility gene. Am J Hum Genet. 2008;82(1):150–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Arking DE, et al. A common genetic variant in the neurexin superfamily member CNTNAP2 increases familial risk of autism. Am J Hum Genet. 2008;82(1):160–4.View ArticlePubMedPubMed CentralGoogle Scholar
- Bakkaloglu B, et al. Molecular cytogenetic analysis and resequencing of contactin associated protein-like 2 in autism spectrum disorders. Am J Hum Genet. 2008;82(1):165–73.View ArticlePubMedPubMed CentralGoogle Scholar
- Poliak S, et al. Juxtaparanodal clustering of shaker-like K+ channels in myelinated axons depends on Caspr2 and TAG-1. J Cell Biol. 2003;162(6):1149–60.View ArticlePubMedPubMed CentralGoogle Scholar
- Ellegood J, et al. Clustering autism: using neuroanatomical differences in 26 mouse models to gain insight into the heterogeneity. Mol Psychiatry. 2015;20(1):118–25.View ArticlePubMedGoogle Scholar
- Kloth AD, et al. Cerebellar associative sensory learning defects in five mouse autism models. elife. 2015;4:e06085.View ArticlePubMedPubMed CentralGoogle Scholar