Regulation of SPRY3 by X chromosome and PAR2-linked promoters in an autism susceptibility region

Abstract

Sprouty proteins are regulators of cell growth and branching morphogenesis. Unlike mouse Spry3, which is X-linked, human SPRY3 maps to the pseudoautosomal region 2; however, the human Y-linked allele is not expressed due to epigenetic silencing by an unknown mechanism. SPRY3 maps adjacent to X-linked Trimethyllysine hydroxylase epsilon (TMLHE), recently identified as an autism susceptibility gene. We report that Spry3 is highly expressed in central and peripheral nervous system ganglion cells in mouse and human, including cerebellar Purkinje cells and retinal ganglion cells. Transient over-expression or knockdown of Spry3 in cultured mouse superior cervical ganglion cells inhibits and promotes, respectively, neurite growth and branching. A 0.7 kb gene fragment spanning the human SPRY3 transcriptional start site recapitulates the endogenous Spry3-expression pattern in LacZ reporter mice. In the human and mouse the SPRY3 promoter contains an AG-rich repeat and we found co-expression, and promoter binding and/or regulation of SPRY3 expression by transcription factors MAZ, EGR1, ZNF263 and PAX6. We identified eight alleles of the human SPRY3 promoter repeat in Caucasians, and similar allele frequencies in autism families. We characterized multiple SPRY3 transcripts originating at two CpG islands in the X-linked F8A3—TMLHE region, suggesting X chromosome regulation of SPRY3. These findings provide an explanation for differential regulation of X and Y-linked SPRY3 alleles. In addition, the presence of a SPRY3 transcript exon in a previously described X chromosome deletion associated with autism, and the cerebellar interlobular variation in Spry3 expression coincident with the reported pattern of Purkinje cell loss in autism, suggest SPRY3 as a candidate susceptibility locus for autism.

Introduction

The mammalian SPROUTY (SPRY) proteins are homologous to Drosophila Sprouty, a receptor tyrosine kinase signalling inhibitor involved in airway branching (1). There are four paralogs in mammals, SPRY1–4 (1–3), that underpin multiple roles during development through interactions with multiple proteins (4,5). All SPROUTY proteins and related SPRED (Sprouty-related EVH1 domain) proteins contain a carboxy-terminal cysteine-rich region, the SPR domain, that facilitates inhibition of receptor tyrosine kinase signalling by targeting the proteins to phosphatidylinositol 4,5-bisphosphate (6–11). SPROUTY proteins are frequently co-expressed with FGF or other growth factors and form integral components of feedback loops that regulate cell survival, growth and migration, tissue growth and morphogenesis and tumourigenesis (4,12–15).

Spry1, 2 and 4 are widely expressed during mouse organogenesis (16,17). Null mutants have confirmed their importance in mouse development. Spry1 is critical for kidney development (9). Spry2 null mice exhibit defects in the organ of Corti leading to impaired hearing (13) and enteric neuronal hyperplasia and oesophageal achalasia (18). Spry4 null mice exhibit mandibular defects with low penetrance, and growth retardation and polysyndactyly with high penetrance (14). A double null Spry1/Spry2 mutant exhibits a defect in formation of the temporomandibular joint (19), and a Spry2/Spry4 double null is embryonic lethal due to cardiovascular defects, with additional severe defects in limb, lung and craniofacial morphogenesis (14), suggesting redundancy between various Spry family members. Human SPRY genes have been predominantly studied in the context of disease. Altered expression of one or more of SPRY1, 2 and 4 has been detected in many tumours (4), in cardiovascular disease (20,21) and in psychiatric disease (22). Manipulation of SPRY expression may have therapeutic benefits in peripheral ischemic disease (23).

Relatively little is known of SPRY3 function, although recently it was shown to inhibit brain-derived neurotrophic factor (BDNF) signalling through TrkB, resulting in reduced axonal branching in Xenopus (24). Interestingly, SPRY4 is an inhibitor of nerve growth factor (NGF) signalling through TrkA (8), suggesting commonalities in modes of action of different SPRY proteins.

The mouse Spry3 locus is X-linked, whereas human SPRY3 maps to the pseudoautosomal region 2 (PAR2) between the X-linked Trimethyllysine hydroxylase epsilon (TMLHE) gene and the PAR2-linked Synaptobrevin-like 1 (SYBL1) gene, a linkage group conserved on the mouse X chromosome. The PAR2, which is not found in rodents and non-human primates, including the chimp, and may therefore be unique to the human, has several unusual genetic and epigenetic features. At 320 kb, it is far smaller than PAR1, exhibits a lower frequency of pairing and recombination in male meiosis and is not necessary for male fertility (25).

SPRY3 and SYBL1 undergo an unusual form of dosage compensation involving random X-inactivation in females and inactivation of the Y-linked locus in males, whereas the more distally located PAR2 genes, IL9R and CXYorf1, are expressed on the Y chromosome and escape X-inactivation in females (26,27). Therefore, notwithstanding PAR2-linkage, SPRY3 and SYBL1 mutant phenotypes are expected to exhibit X-linked genetic transmission. Silencing of SYBL1 on the inactive X and Y chromosomes is associated with methylation of its promoter-associated CpG island. The SPRY3 locus does not have a CpG island and its silencing is associated with modifications typical of repressive chromatin (28). However, the mechanisms that select the Y-linked copies of these genes for silencing is unknown.

In order to understand the role of SPRY3 in development, and to attempt to predict the phenotypic consequences of SPRY3 deregulation, we analysed human and mouse SPRY3 regulation, expression and function. Our results suggest that SPRY3 inhibits growth and branching of ganglion cells during postnatal development and in the adult. We describe complex transcriptional regulation of SPRY3 by X-linked and pseudoautosomal promoters, and we provide evidence that SPRY3 is a susceptibility gene for autism.

Results

Conservation of SPRY3 expression in mouse and human central and peripheral nervous system ganglion cells

We searched online databases for clues to the Spry3 expression pattern and found reported high expression in adult mouse cerebellar Purkinje cells (Allen Brain Atlas, www.brain-map.org; Gensat Brain Atlas, www.gensat.org). We used qRT-PCR to compare mRNA levels in dissected brain compartments from neonatal (P1), recently weaned (P21) and adult C57Bl/6 mice using intron-spanning primers to confirm this expression pattern. All examined brain compartments had relatively high expression, compared with non-neural tissues (data not shown), with ∼3- to 4-fold higher expression in postnatal and adult cerebellum compared with other brain areas (Fig. 1A). High expression in cerebellum was also observed in a qRT-PCR analysis of a panel of human brain regions (Fig. 1B). To further examine Spry3 protein in the cerebellum, we tested two anti-SPRY3 antibodies, and used Abcam antibody ab54231 for immunohistochemistry (IHC) of frozen sections of adult mouse brain. Consistent with data from Allen and Gensat Brain Atlases (Fig. 2A and B), intense staining of Purkinje cell bodies and projections was observed (Fig. 2C, E and F). Western blot analysis detected a predicted ∼30 kD band in cerebellum, but not in other brain regions, confirming relatively high Spry3 protein expression in cerebellum (Fig. 2H).

Figure 1.

High expression of SPRY3 in mouse and human cerebellum. (A) qRT-PCR of mouse brain tissues. P1, P21, one and twenty-one days post-natal, respectively. STR, striatum; CTX, cerebral cortex; MB, midbrain; BS, brain stem; CB, cerebellum; SC, spinal cord; FB, forebrain; HB, hindbrain. Means of three replicates ± SEM. (B) qRT-PCR of human brain regions, spinal cord and pituitary gland.

Figure 2.

SPRY3 and TF expression in CNS and PNS. (A) Spry3 mRNA expression in adult mouse cerebellum detected by in situ hybridization. (Allen Brain Atlas; www.brain-map.org). (B) IHC of BAC-transgenic EGFP expression driven by Spry3 genomic region in adult mouse cerebellum. (GENSAT Brain Atlas; www.gensat.org). (C) IHC using anti-SPRY3 antibody of adult mouse cerebellum. (D) Transgenic human SPRY3/357-LacZ reporter expression in adult mouse cerebellum. Arrowheads indicate lobular regions of high (red) and low (green) expression. (E–G) Low and high power images of IHC using anti-SPRY3 antibody in adult mouse (E and F) and human (G) cerebellum. Black arrowheads indicate expression in Purkinje cell dendrites extending into molecular layer (ML). PCL, Purkinjie cell layer; GCL, granule cell layer. (H) Western blot analysis of male and female adult mouse cerebellum (CB), cerebral cortex (CTX) and striatum (STR) using the anti-SPRY3 antibody. (I–K) IHC using anti-SPRY3 antibody of SCG (I), DRG (J), retina (K) of adult mouse. Black arrowheads indicate staining in GCL and INL. (M–O) Human SPRY3/357-LacZ reporter expression in wholemout of SCG (M), DRG (N); and retina (O) of adult transgenic mouse. Black arrowheads indicate staining in GCL and INL. (L) and (P) IHC using anti-MAZ and anti-ZNF263 antibodies, respectively, of adult mouse cerebellum. Purkinje cell nucleus is arrowed. Scale bars: A–D, 500 μm; E–G, I–L and O, 50 μm; M and N, 100 μm; P, 25 μm.

In a wider screen of the nervous system, we observed Spry3 expression in all examined central and peripheral nervous system ganglion cells in adult mice. Robust staining was also observed in the superior cervical ganglion (SCG) and dorsal root ganglion (DRG) dissected from neonatal (P1) and adult mice, respectively (Fig. 2I and J). In the retina there was intense staining of the retinal ganglion cell layer (GCL) and the inner nuclear layer (INL) (Fig. 2K). In analyses of human cerebellum post-mortem samples from three individuals, we observed staining of cerebellar Purkinje cells (Fig. 2G), indicating conservation of expression pattern in mouse and human.

Spry3 regulates neurite growth and branching of mouse SCG in vitro

We determined the effects of over- and under-expression of Spry3 in a model of postnatal ganglion cell growth. Spry3 function was analysed by transient over-expression and shRNA-mediated knockdown of Spry3 mRNA in cultured primary mouse SCG cells (29), which express Spry3 protein as determined by IHC (Fig. 2I).

For over-expression experiments, the Spry3 open reading frame (ORF) was cloned into the NcoI and PmlI sites of pQE-Tri System vector and the insert and flanking cloning sites were sequenced. The pQE-Spry3 expression vector was transfected into Hela cells and Spry3 protein expression was confirmed by western blot probed with Abcam anti-SPRY3 antibody (Supplementary Material, Fig. S1A).

Spry3 knockdown in SGC primary cell cultures was achieved using pSicoR-Spry3 shRNA vectors, which were designed and made as described in the Jacks laboratory protocol (http://web.mit.edu/jacks-lab/protocols/pSico.html). The negative control was a shRNA vector targeting the unrelated Psg22 gene (Spry3 and Psg22 target sequences used in shRNA vectors are given in Materials and Methods). Inserted sequences and flanking cloning sites of all vectors were sequenced. The shRNA vectors were tested by co-transfection with pQE-Spry3 expression vector into human JAR cells. Cell viability was assessed by MTT assay 24 and 48 h post-tranfection, and Spry3 knockdown was confirmed using Spry3-specific primers and qRT-PCR of cDNA made from mRNA extracted at 24 and 48 h post-transfection (Supplementary Material, Fig. S1B–E).

For each experiment, twenty SCGs from ten P1 C57Bl/6 mice were dissected using low power microscopy. Neurons were dissociated with trypsin/EDTA and seeded into four Greiner Cellstar dishes (Sigma, UK) in DMEM supplemented with 10% FBS and 10 ng/ml NGF. Co-transfection of pQE-Spry3 and pCX-EGFP vector expressing EGFP, and transfection of individual pSicoR vectors, was carried out using the Neon Microporation kit (Invitrogen, USA). Twenty-four hours post-transfection, images of EGFP-labelled neurons were acquired using a Leica DMI3000 microscope. Neuritic arbours were traced using MATLAB software (The MathWorks) and total neurite length and number of branch points were calculated and Sholl analysis was performed as previously described (30). In three replicated experiments, Spry3 over-expression significantly reduced neurite branching, arbour length and neurite complexity (Fig. 3A), whereas shRNA-mediated knockdown of endogenous Spry3 increased neurite branching, arbour length and neurite complexity (Fig. 3B).

Figure 3.

SPRY3 regulates neurite growth and branching. Effect of (A) transient over-expression and, (B) shRNA-mediated under-expression, of SPRY3 on cultured SCG cell growth and arbourization. Means ± SEM of 150 neurons per treatment. **P < 0.01; ***P < 0.001. Scale bar: 50 μm.

Mouse and human SPRY3 gene promoters contain polymorphic AG-rich repeats

As a step to understanding SPRY3 transcriptional regulation, epigenetic silencing and tissue-specific expression, we analysed DNA sequences at the transcriptional start site (TSS) of the human and mouse genes. Analysis of public databases indicated that the human SPRY3 TSS is associated with a complex AG-rich repeat (hereafter ‘AG repeat’) spanning ∼370 bp immediately upstream of the TSS (Fig. 4A). A GT-rich repeat (hereafter ‘GT repeat’) comprising seven copies of the pentamer ‘GTTTT’ begins 90 bp downstream of the TSS in the 5′ UTR (Supplementary Material, Fig. S7).

Figure 4.

Human and mouse SPRY3 promoters have AG-rich repeats. (A) Multiple alignments of eight alleles of AG-rich repeat in human SPRY3 promoter region. (B) Multiple alignments of AG-rich repeat in mouse Spry3 promoter region. (C) Predicted TF binding sites in human and mouse SPRY3 promoters determined by Genomatix software. Promoter allele sequences are given in Supplementary Material, Figure S7.

To determine whether SPRY3 promoter repeats are polymorphic, we used PCR primers spanning a ∼700 bp region including both AG and GT repeats, and we cloned and sequenced PCR products from thirty-nine individuals from ten families from the Autism Genetic Resource Exchange (AGRE) autism panel. We analysed a further eighty-five AGRE autism families for AG and GT repeat genotypes using an ABI Genetic Analyser. To confirm the integrity of our genotyping, ninety-one individual AGRE autism family samples were re-analysed by MRC geneservice (www.geneservice.co.uk). In total, we identified eight alleles of the AG repeat (Fig. 4A; Table 1), the shorter alleles being similar in size to four chimpanzee sequences analysed, which were non-polymorphic (data not shown). Although transmission of allelic variants in AGRE autism family pedigrees suggested that these variants are not artefacts of lymphoblastoid cell culture, we obtained additional confirmation of SPRY3 promoter polymorphism by genotyping cheek swab-derived DNA from twenty-eight normal children (data not shown). We named the AG repeat alleles from the size of their respective PCR product produced with PCR primer pair hSPRY3/1F&1R (Supplementary Material, Fig. S2). Using these primers, each AG repeat allele was cloned and sequenced from at least two individuals (Supplementary Material, Fig. S3). The sequence denoted as ENSG00000168939 in the Ensembl Genome Browser corresponds to allele 339 in our nomenclature. A variant of the GT repeat (ΔGT allele) has a 7 bp deletion and is found in ∼10% of Caucasians (Supplementary Material, Fig. S7).

Table 1.

Human SPRY3 PAR2-linked promoter AG-rich repeat alleles in AGRE autism families and Caucasian lymphoblastoid DNA collections

Allele	Autism father	Autism mother	Proband male	Proband female	HD200CAUC male	HD200CAUC female
306	3 (1.7%)	4 (2.1%)	4 (2.2%)	1 (10.0%)	3 (1.5%)	7 (3.5%)
309	0 (0.0%)	1 (0.5%)	0 (0.0%)	0 (0.0%)	1 (0.5%)	3 (1.5%)
321	7 (3.9%)	20 (10.5%)	15 (8.4%)	0 (0.0%)	7 (3.5%)	10 (5.0%)
324	38 (21.3%)	43 (22.6%)	45 (25.0%)	2 (20.0%)	50 (25.0%)	52 (26.0%)
339	1 (0.6%)	3 (1.6%)	1 (0.6%)	0 (0.0%)	2 (1.0%)	0 (0.0%)
357	102 (57.3%)	94 (49.5%)	96 (53.3%)	5 (50.0%)	99 (49.5%)	98 (49.0%)
360	10 (5.6%)	11 (5.8%)	6 (3.3%)	2 (20.0%)	25 (12.5%)	18 (9.0%)
363	17 (9.6%)	14 (7.4%)	13 (7.2%)	0 (0.0%)	13 (6.5%)	12 (6.0%)

We compared allele frequencies in ninety-five AGRE autism families and in two hundred individuals from the Coriell HD200CAUC Caucasian collection to determine whether specific AG or GT repeat alleles are associated with autism. No significant differences were observed between autism affected and unaffected individuals. The 324 and 357 alleles comprised ∼25% and 50% of the total, respectively, with the residual comprised of the other six alleles (Table 1).

To determine whether the mouse Spry3 promoter contains polymorphic repeats, we PCR amplified a 352 bp PCR product from the Spry3 promoter region using PCR primer pair mSpry3/1F&1R (Supplementary Material, Fig. S2) from a range of laboratory and wild mice: laboratory (C57BL/6J, WSB), M. m. domesticus poschiavinus (Zalende), M. m. musculus (CZECHII), M. m. molossinus (MOLC, MOLF, MOLG), M. m. castaneus, M. spretus, M. caroli and Peromyscus maniculatus. No detectable size differences were observed on a 2% agarose gel. We then cloned and sequenced PCR products from a subset of mouse species. Similar to the human, the mouse Spry3 putative promoter has an AG-rich repeat, and while no major allele length variants were found, we identified more subtle sequence changes between species (Fig. 4B; Supplementary Material, Fig. S7).

Notwithstanding the highly polymorphic human SPRY3 AG repeat, the presence of an AG repeat in the mouse, and the conservation of predicted transcription factor (TF) binding sites between mouse and human (see below), suggested that this region does indeed represent the promoter of SPRY3. We PCR amplified from human genomic DNA the SPRY3 AG repeat allele 357 using primers hSPRY3/2F&2R (Supplementary Material, Fig. S2), and cloned a 713 bp fragment into the NotI and SpeI sites of a pBluescript KS-derived reporter vector containing the β-Globin minimal promoter and the LacZ gene encoding β-Galactosidase—pGZ40 (31), hereafter SPRY3/357-LacZ vector. Transgenic mice were produced by oocyte microinjection and ten transgenic founder lines were established following diagnostic PCR of tail DNA. Three male and three female progeny from each line were examined for LacZ expression in multiple organs (brain, eye, spinal cord, SCG, DRG, lung, heart, thymus, liver, spleen, kidney). Only one of the transgenic lines expressed the transgene; however, the staining observed in brain, retina, spinal cord, SCG and DRG, recapitulated endogenous Spry3 expression. Notably, the cerebellar Purkinje cell and retinal GCLs were strongly stained, and whole-mount analysis of SCG and DRG also indicated intense staining (Fig. 2D and M–O).

SPRY3 transcription is regulated by multiple TFs

Genomatix software (https://www.genomatix.de/) was used to identify putative TF binding sites in the human and mouse SPRY3 promoters and indicated conservation of binding sites (Fig. 4C). We note that all eight human alleles and the mouse (C57Bl/6, M. spretus) AG repeats contain multiple predicted G-quadruplexes, which were not analysed further (Supplementary Material, Fig. S4). The four most highly ranked TFs (ZNF263, MAZ, PURA, EGR1) were analysed further. PAX6 was also analysed because mouse Spry3 promoter sequence variation suggested the occurrence of evolutionary selection on the number of binding sites (see below). ChIP-seq data from the UCSC browser (https://genome-euro.ucsc.edu/) supported binding of ZNF263, EGR1 and MAZ to the human SPRY3 promoter region in cell lines. PURA, EGR1 and PAX6 expression has previously been reported in cerebellar Purkinje cells (32–34), and EGR1 and PAX6 expression occurs in retinal ganglion cells (35–37). We used IHC on histological sections of adult mouse brain to confirm localization of MAZ and ZNF263 in Purkinje cell nuclei (Fig. 2L and P).

In the absence of human cell lines expressing significant levels of endogenous SPRY3, we used a reporter assay involving co-transfection of the SPRY3/357-LacZ construct and various TF expression vectors into JAR cells to determine whether TF over-expression alters LacZ expression. EGR1 and ZNF263 over-expression significantly increased, whereas MAZ and PURA over-expression significantly reduced, LacZ expression after 48 h (Fig. 5A, C, E and F). However, PAX6 had no effect (Fig. 5G). EGR1 over-expression resulted in robust upregulation of LacZ expression in the reporter assay; therefore, we used this system to probe TF dosage and SPRY3 allele effects on LacZ expression. In a dosage experiment, LacZ expression increased proportionately with transfection of increasing amounts of EGR1 expression vector DNA, peaking at 2 μg per well (Fig. 5B). The 2 μg dose was used to compare LacZ expression in a series of EGR1 transfection experiments in which the SPRY3/357-LacZ vector was replaced by SPRY3/306-LacZ and SPRY3/324-LacZ, containing SPRY3 306 and 324 promoter alleles, respectively. Alleles 306, 324 and 357 all have eight predicted EGR1 binding sites (Fig. 4C). Modest, but statistically significant different, levels of induction of LacZ was observed for the three promoter alleles (Fig. 5D).

Figure 5.

Regulation of SPRY3 promoter alleles by TFs. Effect of transient over-expression of TFs on human SPRY3/357-LacZ promoter reporter expression in human JAR cells: (A) EGR1; (C) ZNF263; (E) MAZ; (F) PURA; (G) PAX6. (B) Dose-response of EGR1 expression vector DNA on reporter expression. (D) Expression of SPRY3/306-LacZ, SPRY3/324-LacZ, and SPRY3/357-LacZ promoter allele reporter vectors to 2 μg EGR1 expression vector. Vertical axis denotes reporter expression level. Means ± SEM of three replicates. *P < 0.05; **P < 0.01; ***P < 0.001.

Compared with the extensive length variation among human SPRY3 promoter alleles, relatively minor differences were detected between laboratory and wild mice. We noted that three of the six sequence polymorphisms between C57Bl/6 and M. Spretus altered predicted PAX6 binding sites (Supplementary Material, Fig. S5A), and an analysis involving simulated mutagenesis of mouse Spry3 promoter sequences suggested that this was unlikely to be due to chance (Supplementary Material, Fig. S5B and C). We therefore analysed PAX6 regulation of SPRY3 expression. As noted above, PAX6 had no effect in the LacZ reporter assay. To examine possible Pax6/Spry3 interactions in vivo, we performed chromatin immunoprecipitation (ChIP) on cerebellum samples from male P21 C57Bl/6 mice using an anti-Pax6 antibody (sc7750, Santa Cruz Biotechnology). Compared with ‘no antibody’ and ‘IgG’ controls in three replicated experiments, quantitative PCR (qPCR) of mouse Spry3 promoter sequence using primers mSpry3/2F&2R (Supplementary Material, Fig. S2) indicated binding of Pax6 to the Spry3 promoter region in vivo (Supplementary Material, Fig. S5D). To determine whether Pax6 over- or under-expression in vivo led to altered Spry3 expression, we analysed Spry3 expression in Pax6 YAC transgenic (38) and Sey mutant (39) mice, respectively. Pax6 and Spry3 expression was analysed by qRT-PCR in cerebellum samples from three male and three female wild-type adult control CD1, YAC transgenic and Sey mice. Pax6 mRNA expression was increased in Pax6 YAC transgenics, but not in Sey or control mice as expected (Sey mice express Pax6 mRNA, but do not make PAX6 protein). However, no significant difference in Spry3 expression was found between the three groups (Supplementary Material, Fig. S5E).

X-linked promoters of SPRY3 in the F8A3—TMLHE locus region

We hypothesized that silencing of SPRY3 on the Y-linked PAR2 might be a default state, with expression of the X-linked copy requiring a signal from the X-chromosome. We identified four EST clones (BI752892, BQ072670, DB209851, DA102958) on the UCSC browser that are transcribed in the X-linked F8A3—TMLHE region in an antisense direction to F8A3 and TMLHE and towards PAR2. DA102958, which originates in the first intron of TMLHE, splices onto exon 1 of SPRY3 and contains part of SPRY3 exon 2. BQ072670 and DB209851 both originate in the region between F8A3 and TMLHE, contain four non-overlapping exons each, and terminate in introns 6 and 2 of TMLHE, respectively. BI752892 initiates within the F8A3 gene sequence and has two exons (Fig. 6A). These and related transcripts were characterized by RT-PCR and 5′-RACE-PCR of cDNA samples from human adult cerebellum and Hela cells, using forward primers hSPRY3/3F&4F, which are in the first exons of DB209851 and DA102958, respectively, and a reverse primer, hSPRY3/3R in the second exon of SPRY3, followed by cloning and sequencing of products. 5′-RACE-PCR primers were placed in the third exon of DB209851 (Fig. 6; Supplementary Material, Fig. S2). Agarose gel electrophoresis, and cloning and sequencing of PCR products, from transcripts initiating in the F8A3—TMLHE region revealed a complicated pattern of alternative splicing of SPRY3 upstream transcripts (Fig. 6A and C). Transcripts initiating at the TMLHE intron 1 region did not contain additional exons beyond those in DA102958. We named the SPRY3 transcript transcribed from the PAR2-linked promoter as SPRY3 variant 1 (SPRY3_V1); the SPRY3 transcript initiating in intron 1 of TMLHE as SPRY3 variant 2 (SPRY3_V2); and all transcripts initiating near the F8A3 region as SPRY3 variant 3 (SPRY3_V3).

Figure 6.

Schematic overview of SPRY3 transcripts. (A) Human F8A3—SPRY3 genomic region spanning X/PAR2 boundary. Exons denoted as filled boxes, with transcriptional direction given by arrowed edge. Four EST clones from UCSC browser and seven RT-PCR or RACE-PCR-confirmed sequenced transcripts are shown as exons connected by horizontal dashed lines. A vertical line between exons indicates that the exon is found in several transcripts. TSS, transcriptional start sites on X chromosome defined by 5′-RACE PCR in this study, or on PAR2 from UCSC browser (SPRY3 PAR2 promoter transcript), denoted by arrows. (V1), (V2), (V3), associated with TSS denote SPRY3_V1, SPRY3_V2 and SPRY3_V3 transcripts initiating at each TSS, as described in main text. A polyadenylation signal identified by 3′-RACE is also denoted by an arrow. Rectangular boxes denote two CpG islands and a deletion described by Celestino-Soper et al. (2012). (B) Mouse Tmlhe—Spry3 X-linked region and transcripts. In (A) and (B), exons and introns are shown on different scales: 1 intron length unit = 20 exon length units. Diagonal double slashes indicate three exons and one intron not shown to scale. Transcript sequences are given in Supplementary Material, Figure S7. (C) and (D) RT-PCR of Hela and human cerebellum using primers hSPRY3/3F and hSPRY3/3R, and primers hSPRY3/4F and hSPRY3/3R. M, male; F, female; A, Alzheimer's disease sample; C, control (unaffected) sample; CB, cerebellum. Sequences of transcripts T1–T9 are listed in Supplementary Material, Figure S7.

5′-RACE PCR was used to identify the TSS of transcripts SPRY3_V2 & V3 in Hela cells and in human cerebellum. TSS was identified between F8A3 and TMLHE exon 1 (Fig. 6A; Supplementary Material, Fig. S6). Two TSS at chrX:154841915 and chrX:154841933, close to exon 1 of TMLHE, were found in Hela cells and cerebellum, respectively, and a TSS at chrX:154687410 in F8A3 was found in Hela cells. A further putative TSS at chrX:154722922 in the sixth intron of TMLHE was found in cerebellum; however, this sequence appears to be an extension of the fourth exon of BQ072670 which splices to the third exon of DB209851 (Fig. 6A; Supplementary Material, Fig. S6).

We did not detect the fourth exon of DB209851 in qRT-PCR experiments using reverse primers in SPRY3 exons, suggesting that it may not be included in transcripts terminating at SPRY3 exon 2. Inspection of sequence downstream of this exon identified a polyadenylation signal, suggesting that, unlike other F8A3 —TMLHE region exons, the inclusion of this exon would lead to transcriptional termination rather than splicing of X-linked exons to SPRY3 exons. 3′-RACE was used to confirm that the polyadenylation signal in the fourth exon of DB209851 leads to termination and polyadenylation of the transcript. Using primers hSPRY3 3′RACE/1F&2F in the fourth exon of DB209851, we identified the transcript termination site at chrX:154775004, 13 bp downstream of the polyadenylation signal (Fig. 6A; Supplementary Material, Fig. S6). This exon, in addition to TMLHE exon 2, is within a deletion previously identified in a case of autism (40).

Expression of SPRY3_V1 (PAR2-linked promoter) and SPRY3_V2 & V3 (X-linked promoter) transcripts is correlated

We developed a qRT-PCR strategy to determine the relative contribution of SPRY3_V1 and SPRY3_V2 & V3 transcripts to total SPRY3 expression. SPRY3_V3 transcripts initiating in the F8A3 region have a complex splicing pattern with variable inclusion of multiple alternative exons, and the further complication that an unknown proportion of transcripts may terminate at exon 4 of DB209851. This complexity constrained our ability to quantify relative abundance of these transcripts. However, the SPRY3_V2 transcript initiating in intron 1 of TMLHE does not have splice variants (Fig. 6A). Therefore, we focussed on this transcript as potentially representative of SPRY3 transcripts initiating on the X chromosome. A reverse primer in the 150 bp region between the SPRY3 PAR2-linked promoter TSS and the SPRY3 exon 1 splice acceptor site (which is included in the SPRY3_V1 transcript, but excluded from SPRY3_V2 & V3 transcripts), allowed us to separately amplify and quantify SPRY3_V1 and SPRY3_V2 transcripts. This strategy allowed us to measure (i) total SPRY3 expression using primers hSPRY3/5F&5R; (ii) SPRY3_V1 expression using primers hSPRY3/6F&6R; (iii) SPRY3_V2 expression using primers hSPRY3/7F&7R (Supplementary Material, Fig. S2). Subtraction of the sum of ii and iii from i, gives the proportion of SPRY3_V3 transcripts initiating at the F8A3-associated TSS. A vector containing subcloned copies of two SPRY3 transcripts allowed normalization of the relative efficiencies of different PCR primer pairs.

Using this approach, we measured total SPRY3 expression and relative contribution of X-linked and PAR2-linked promoter transcripts. In cell lines, which express very low levels of SPRY3 compared with cerebellum, the SPRY3_V1 transcript comprises ∼10% of total expression, SPRY3_V2 comprises ∼50% and, by subtraction, SPRY3_V3 comprises ∼40% of total SPRY3 (data not shown). In most adult human cerebellum samples, the opposite pattern was observed with the majority (∼74%–87%) of SPRY3 expression from SPRY3_V1 (Table 2). In a further study, using a human brain cDNA panel, we measured total SPRY3 and SPRY3_V1 expression in a variety of brain regions and in spinal cord (Table 3). Consistent with our mouse studies, human SPRY3 expression varies greatly in different brain regions, with highest expression in cerebellum and pituitary gland; however, the relative proportions of SPRY3_V1 and SPRY3_V2 & V3 transcripts contributing to total SPRY3 expression remained relatively constant in different tissues (Fig. 1B; Table 3).

Table 2.

Proportions of human SPRY3 transcripts in human post-mortem cerebellum

	SPRY3_V1	SPRY3_V2	SPRY3_V3
Male unaffected	83.01%	9.11%	7.89%
Male Alzheimer's	87.35%	11.79%	0.85%
Female unaffected	86.34%	10.24%	3.42%
Female unaffected	73.94%	9.21%	16.85%
Female Alzheimer's	46.89%	9.98%	43.13%

Table 3.

Proportions of SPRY3_V1 (PAR2-linked promoter) transcript relative to SPRY3_V2 & V3 (X-linked) transcripts in neural tissues

Tissue	SPRY3_V1	SPRY3_V2 + V3
Paracentral gyrus	68.36%	31.64%
Postcentral gyrus	61.33%	38.67%
Olfactory bulb	69.14%	30.86%
Hypothalamus	45.94%	54.06%
Hippocampus	27.26%	72.74%
Substantia Nigra	78.08%	21.92%
Pituitary gland	76.40%	23.60%
Cerebellum grey	69.13%	30.87%
Cerebellum white	80.89%	19.11%
Cerebellum vermis	97.29%	2.71%
Nucleus accumbens	88.10%	11.90%
Pons	60.03%	39.97%
Spinal cord	78.05%	21.95%
Choroid plexus	89.17%	10.83%

Refer to Figure 1B for total SPRY3 expression in these samples.

The first exon of DA102958 in TMLHE intron 1 partly overlaps a CpG island associated with the TMLHE promoter and first exon; therefore, expression of the SPRY3_V2 transcript may be regulated by CpG methylation. We compared methylation of this region in males and females using pyrosequencing of bisulphite treated DNA from B lymphoblastoid cell lines, cheek swabs and human cerebellum samples. We observed consistently higher methylation levels in females compared with males, consistent with the occurrence of X-inactivation at this locus in females (Table 4).

Table 4.

DNA methylation at nine CpG sites of TMLHE exon 1 associated CpG island

CpG:	1	2	3	4	5	6	7	8	9
Male
Cerebellum	1.30%	4.10%	3.70%	6.20%	0.00%	0.00%	4.60%	0.00%	4.60%
Cheek swab	2.30%	2.60%	3.90%	6.20%	0.00%	0.00%	5.50%	0.00%	5.00%
B-lymphoblastoid cells	1.40%	3.30%	3.00%	5.80%	0.00%	1.80%	8.80%	3.90%	5.70%
Female
Cerebellum	11.90%	30.60%	12.40%	45.80%	21.50%	10.60%	14.30%	33.80%	21.20%
Cheek swab	17.30%	45.10%	21.20%	55.20%	41.50%	16.20%	26.60%	35.70%	35.00%
B-lymphoblastoid cells	29.10%	27.60%	19.90%	46.50%	12.20%	4.90%	28.90%	21.20%	34.70%

CpG site positions are given in Supplementary Material, Figure S7.

Scrutiny of the mouse TMLHE intron 1 region indicated similarity to the homologous human region, including a splice donor site, suggesting the presence of an upstream TSS in the mouse. RT-PCR using primers mSpry3/3F&3R (Supplementary Material, Fig. S2), and sequencing of cloned products from cDNA of adult C57Bl/6 cerebellum confirmed the presence of an upstream transcript (Fig. 6B). 5′-RACE PCR was used to map the mouse Spry3 TSS in adult cerebellum using primers listed in Supplementary Material, Figure S2. Two TSS were identified at chrX_GL456233_random:110 973 and chrX_GL456233_random:160 310, corresponding to regions homologous to the human SPRY3 promoter/exon 1, and TMLHE intron 1 TSS regions, respectively (Fig. 6B; Supplementary Material, Fig. S6).

Discussion

We characterized a set of transcripts that initiate in the Xq28-linked F8A3—TMLHE locus and exhibit a complex splicing pattern involving variable inclusion of multiple alternative exons. The majority of these transcripts ultimately splice onto exon 1 of the PAR2-linked SPRY3 gene and presumably terminate at the SPRY3 exon 2 termination signal. This observation has several implications for understanding SPRY3 regulation and its deregulation in disease. First, regulation of SPRY3 expression by X-linked promoters may explain the imprinted expression of SPRY3 in males, in which the Y-associated PAR2-linked (hereafter Y/PAR2) copy is silenced epigenetically, but the X-associated PAR2 copy (hereafter X/PAR2) remains active (27). The adjacent SYBL1 gene undergoes similarly imprinted regulation associated with methylation of the promoter CpG island of the inactive Y/PAR2 copy (28). The SPRY3 PAR2-linked promoter does not have a CpG island, although distinct chromatin modifications have been reported on the inactive SPRY3 in males (28). However, the initiating signal that determines the parentally imprinted nature of SPRY3 expression in males is unknown. We speculate that silencing of the Y/PAR2 SPRY3 may be a default state with X/PAR2 SPRY3 expression maintained by transcription from the Xq28 region through the PAR2 boundary to the SPRY3 locus. Transcription through this region could maintain an open chromatin configuration facilitating escape from epigenetic silencing, as reported at other loci (41).

Since the first exon of multiple X/PAR2 transcripts are close to TMLHE and F8A3-associated CpG islands, expression of these transcripts and, ultimately, activity and X-inactivation status of X/PAR2 SPRY3 may be regulated by CpG methylation. In support of this, we found that the CpG island associated with exon 1 of TMLHE and the 5′-RACE PCR defined boundary of exon 1 of DA102958 is largely unmethylated in males and relatively highly methylated in females, consistent with X-inactivation in females. We did not examine the F8A3-associated CpG island because this analysis would be confounded by virtually identical sequences at F8A1 and F8A2.

The expression of SPRY3_V2 & V3 transcripts and total SPRY3 mRNA is positively correlated because the relative proportion of total SPRY3 expression from X-linked and PAR2-linked promoters is similar in tissues with low (e.g. paracentral gyrus, olfactory bulb) and high (e.g. pituitary gland, cerebellum) total expression. Across different brain areas in which total SPRY3 expression varies over two orders of magnitude, ∼30% of expression is from SPRY3_V2 & V3 transcripts in several of the tissues analysed. This suggests either that there may be global coordinated regulation of expression across the entire F8A3—SPRY3 locus, or that the level of SPRY3_V2 & V3 transcript expression determines PAR2-linked promoter activity and SPRY3_V1 transcript expression.

Evidence from this and previous studies indicates that SPRY3 and TMLHE are both strongly expressed in cerebellar Purkinje cells, suggesting co-regulation (40). Transcription initiating near the TMLHE promoter CpG island, which comprises a significant proportion of SPRY3 expression, and which is correlated with total SPRY3 expression, suggests a mechanism for coordinated regulation of these linked genes. Moreover, the fact that there is a conserved transcript linking Tmlhe and Spry3 in the mouse suggests that this mechanism pre-dates the evolution of human PAR2. TMLHE is the first enzyme in the carnitine biosynthesis pathway. The reasons for expression of TMLHE in the brain are unclear, but it has been suggested that there may be a specific requirement for de novo production of carnitine in neural tissues, although evidence is lacking (40). For completeness, we note that TMLHE expression in brain might be required solely for maintenance of transcription of SPRY3_V2 & V3 transcripts that splice to SPRY3 exon 1, with neural expression of TMLHE merely a side-effect of its role in regulating SPRY3. However, identification of mutations in the ORF of TMLHE associated with autism do not support this view (see below).

Beaudet and colleagues have reported a range of deletions encompassing exon 2 of TMLHE which may be low penetrance risk factors for autism (40,42). A role for TMLHE deficiency in autism is further supported by the reduced levels of carnitine measured in blood of autism cohorts, and subsequent reports of other deletions (43), and nonsense and missense mutations in the TMLHE ORF (44). We noted that the deletion described by Beaudet et al. (42), in addition to deleting TMLHE exon 2, also deletes the last exon of EST DB209851. This exon is part of a transcript that initiates at the F8A3 promoter CpG island region and does not splice to SPRY3 exon 1 but, rather, terminates at a polyadenylation signal within the region defined by the deletion. Therefore, the inclusion or exclusion of this exon could influence the proportion of SPRY3_V3 transcripts contributing to total SPRY3 expression. The relative contribution of these transcripts varied from 0.85–43.13% of total SPRY3 transcripts in a collection of post-mortem cerebellum samples from unaffected and Alzheimer's subjects, and was ∼40% of total SPRY3 expression in lymphoblastoid B-cell and other cell lines which, however, express very low levels of SPRY3 compared with brain tissues (data not shown). If these results are representative, they suggest that SPRY3_V3 transcripts may contribute substantially to total SPRY3 expression in certain tissues and contexts. Moreover (unlike the SPRY3_V2 transcript, which appears to stably contribute ∼10% of total SPRY3 in all brain samples analysed), the variable contributions of SPRY3_V3 transcripts to total SPRY3 expression in different samples suggests that SPRY3_V3 transcripts may be regulated. A possible mechanism is inclusion or exclusion of the last exon of EST DB209851, which may influence the proportion of SPRY3_V3 transcripts that splice to SPRY3 exon 1, thereby influencing total SPRY3 expression. Similarly, the deletion of the fourth exon of EST DB209851 in the patient described by Beaudet and collaegues (42) could potentially alter SPRY3 expression, thereby suggesting SPRY3 as a candidate autism susceptibility locus.

Further links between SPRY3 and factors associated with autism are evident from the literature. There may be a regulatory interaction between SPRY3 and carnitine metabolism as indicated by the greater than 10-fold upregulation of SPRY3 following chronic administration of high levels of dietary carnitine in the pig (45). Moreover, co-regulation of Xenopus Spry3 and BDNF-TrkB signalling (24), potentially implicates SPRY3 function in neuropsychiatric disorders, including autism, associated with changes in BDNF expression (46,47), assuming conservation of Xenopus and mammalian SPRY3 function.

Purkinje cell abnormalities observed in autism are consistent with a role for SPRY3 deregulation because of relatively high expression of SPRY3 in Purkinje cells and evidence from this and previous studies implicating SPRY3 in neuronal growth and branching (24). We note, in particular, the apparently coincident pattern of relatively low SPRY3 expression described in this study with the pattern of Purkinje cell loss reported in a recent study of eight autism post-mortem brains (48). In that study, reduced Purkinje cell density was predominantly observed in lobule VIIa in both genders, and in lobule X exclusively in males (48). In these lobules, relatively low Spry3 expression is consistently observed in Allen and Gensat atlas data, in our immunohistochemical analysis of mouse cerebellum and in our human SPRY3/357-LacZ reporter transgenic mouse line (see our Fig. 2A–D). The latter observation suggests that the inter-lobular variation in Spry3 expression observed in mice may be conserved in the human, and it implicates the human SPRY3 PAR2-linked promoter in regulating inter-lobular variation in SPRY3 expression.

The variability of the SPRY3 PAR2-linked promoter due to the presence of a polymorphic AG repeat appears to be unique to the human, as an AG repeat in the mouse Spry3 promoter is highly conserved across multiple mouse species, and strains of M. domesticus. We also did not detect repeat length polymorphisms in four individual chimpanzees (P. troglodytes). This suggests that human SPRY3 repeat length polymorphisms arose after the evolution of PAR2. The shortest human repeat (our 306 allele) is similar to chimpanzee, gorilla and orangutan sequences; therefore, the other eight human alleles are expansions of the ancestral great ape sequence, with a 57 bp maximum length difference between human alleles. Using Genomatix analysis, we found that mouse and human AG repeat sequences have multiple predicted conserved TF binding sites, the number of which varies between the eight human alleles and, less dramatically, among mouse species. Online database searches and IHC of adult mouse brain, showed that the transcription factors ZNF263, MAZ, PURA and EGR1, whose predicted binding sites are highly represented in the SPRY3 promoter, bind in the region and are co-expressed with SPRY3 in cerebellar Purkinje cells or retinal ganglion cells. All four TFs altered SPRY3 promoter allele 357 reporter expression, either increasing or reducing it. Whether the directionality of these effects recapitulates their effects on endogenous SPRY3 is unknown, but these results provide additional evidence supporting regulation of SPRY3 by these TFs. While different numbers of TF binding sites (e.g. ZNF263, MAZ) in different promoter alleles might be expected to influence SPRY3 expression, the number of predicted PURA and EGR1 binding sites is identical in all eight alleles. We compared the effect of EGR1 over-expression on reporter vectors with the 306, 324 or 357 alleles, to determine whether repeat expansions per se may affect expression. We observed modest but statistically significant effects on reporter gene expression between the 306 allele and the more common 324 and 357 alleles, suggesting that different promoter alleles may influence expression irrespective of differences in the number TF binding sites. This may be due to altered distances between different TF binding sites, including G-quadruplexes, and the TSS, with resultant differences in chromatin structure and promoter activity. Perhaps unsurprisingly, as SPRY proteins are inhibitors of growth factor signalling, we were unable to identify human cell lines with robust SPRY3 expression; therefore, we were unable to analyse regulation of endogenous SPRY3 in this study.

We analysed regulation of mouse Spry3 by Pax6 because we noted that, notwithstanding high conservation of M. domesticus (C57Bl/6) and M. spretus AG repeat sequences, three of the six sequence differences altered a predicted Pax6 binding site. A simulation involving random mutagenesis of mouse AG repeat sequences in silico suggested that this was unlikely to be due to chance and might reflect the outcome of selection, implying a significant role for Pax6 in Spry3 regulation. ChIP analysis of adult mouse cerebellum using an anti-PAX6 antibody pulled down Spry3 promoter sequences, supporting this hypothesis. However, analysis of adult cerebellum from mouse mutants that under- or over-express Pax6 did not result in detectable alterations in Spry3 expression. These results may indicate that Pax6 has a subtle, but evolutionarily significant, effect on Spry3 expression, which may have been undetected due to redundancy of Pax6 and other TFs. Alternatively, as our analysis was confined to adult cerebellum, we may have missed more pronounced effects on Spry3 expression in other tissues. Further work, involving more extensive analysis of multiple Spry3-expressing tissues, is required to resolve these issues.

The region upstream of the human SPRY3 PAR2-linked promoter is a recombination hotspot reportedly centred on a polymorphic (SNP rs700442) predicted PRDM9 binding site ∼400 bp upstream of the SPRY3 TSS. This hotspot has an unusually high male inter-individual variability (∼35-fold) in the ratio of non-crossover to crossover events (NCO:CO ratio) compared with autosomal hotspots, which cannot be fully accounted for by the PRDM9 binding site or PRDM9 locus genotypes, or combinations thereof (49). We note that the AG repeat, which starts 120 bp downstream of the polymorphic PRDM9 binding site, contains multiple motifs that may satisfy PRDM9 binding criteria, just upstream of the polymorphic region. We speculate that the presence of extensive polymorphism adjacent to PRDM9 binding sites may explain the residual inter-individual NCO:CO ratio variability previously unaccounted for (49).

The function of SPRY3 in the human is unknown in the absence of genetic associations with disease phenotypes. Nevertheless, our observations of conserved regulatory elements and expression patterns between the mouse and human, in addition to work reported in this study and previously, indicating a role in neuronal branching morphogenesis (24), suggest that SPRY3 may inhibit growth factor signalling in neuronal development. In our expression analysis in neural tissues, we noted a striking pattern of high expression in ganglion cells postnatally including cerebellar Purkinje cells, which have been found to be abnormal in number or structure in post-mortem studies of autism and ataxia (48,50–52), and in mouse models of autism and ataxia (53,54). A Spry2 null mutant exhibits enteric neuronal hyperplasia through a GDNF-mediated mechanism (18), and may be indicative of the type of pathology that would be expected in a Spry3 null mutant, with ganglion cell hyperplasia in the cerebellum, retina and other SPRY3 expressing tissues. Conversely, deregulation of SPRY3 leading to increased or inappropriate expression may result in unpredictable gain-of-function phenotypes, or cell death due to inappropriate inhibition of growth factor signalling.

Based on previous work (42,48) and our observations on SPRY3 expression and regulation, we conclude that SPRY3 is a candidate susceptibility locus for autism and related or over-lapping disorders such as ataxia. We did not detect association of SPRY3 promoter AG repeat alleles with autism, but note that a more extensive study of genetic and epigenetic variation at this locus is warranted, including exome sequencing and analysis of genome variation in the F8A3 region in the normal population and in patients.

Materials and Methods

Materials, DNA and tissues

Chemicals were purchased from Sigma Aldrich, UK, unless otherwise stated. Restriction enzymes were purchased from NEB, USA. PCR and sequencing primers were purchased from MWG Eurofins, Germany. C57Bl/6 mice were purchased from Biological Services Unit, UCC. Pax6 YAC and Sey mouse cerebellum samples were provided by Dr D-J Kleinjan. WSB, M. m. domesticus poschiavinus (Zalende), M. m. musculus (CZECHII), M. m. molossinus (MOLC, MOLF, MOLG), M. m. castaneus, M. spretus, M. caroli and Peromyscus maniculatus genomic DNA samples were purchased from the Jackson Laboratory, USA. All animal experiments were conducted under permissions obtained following animal ethics and welfare review by UCC committees, under national and European legislation. Human brain cDNA panel (cat. HBRT101) was purchased from Origene, USA. Human cerebellum samples were obtained from the Netherlands Brain Bank. Ethical approval and written informed consent was obtained in all cases. The work of the NBB abides by the ethical code of conduct approved by the ethics committee. Autism DNA and B lymphoblastoid cell lines were purchased from AGRE, USA. Causasian DNA panel was purchased from Coriell Institute for Medical Research, USA. Cheek swab DNA samples were collected from children following ethical review by the Clinical Research Ethics Committee of the Cork Teaching Hospitals.

PCR and qRT-PCR

PCR amplification was performed in a 50 μl reaction containing 10 pmol each of forward and reverse primers, 0.2 mm dNTP, 1.25 units Go Taq DNA Polymerase in 1x Green GoTaq® reaction buffer and 20–50 ng DNA. In general, PCR was performed with a MJ Research PTC-200 thermocycler as follows: 95°C for 2 min, followed by 35 cycles of 95°C for 30 s, 55°C for 45 s and 72°C for 60 s/Kb; and a final extension step at 72°C for 10 min. Reverse transcription was performed using the high capacity cDNA reverse transcription kits (Applied Biosystems, USA) on total RNA which was harvested from tissues or cells using the TRI reagent. Quantitative PCR reaction was performed as follows: 10 µl of 2x SYBR green master mix, 10 pmol each of forward and reverse primers, 1 µl of cDNA template and ddH2O up to 20 µl. qPCR was performed in an ABI 7900HT thermocycler as follows: 50°C for 2 min, 94°C for 10 min, followed by 40 cycles of 94°C for 30 s, 60°C for 60 s and a final dissociation step.

DNA vectors

The empty β-Globin/LacZ reporter vector—pGZ40 (31,55), which contains the β-Globin/LacZ reporter sequence cloned into the SmaI site of pBluescript KS was provided by Professor J. Quinn (31,55). The human SPRY3 promoter alleles 306, 324 and 357 were PCR amplified from genomic DNA of individual AGRE samples H12328, H12886 and HI2394, using primers hSPRY3/8F&8R. PCR products were cloned between NotI and SpeI sites of the empty β-Globin/LacZ reporter vector. The mouse Spry3 expression vector was made by cloning the Spry3 ORF between the NcoI and PmlI sites of the pQE vector using primers mSpry3/4F&4R. The shRNA vector targeting mouse Spry3 or Psg22 mRNA was made in the pSicoR vector (56) with target sequence: 5′-GATGCCACAGTGATAGATG-3′ and 5′-GAAGAGAGATATTGTTCAT-3′, according to the Jacks laboratory protocol (http://web.mit.edu/jacks-lab/protocols/pSico.html). Inserted sequences and flanking cloning sites of all vectors were sequenced. For quantification of relative expression of human SPRY3 transcripts, the PCR products from the primer pairs hSPRY3/9F&9R and hSPRY3/10F&10R were cloned into the pSTBlue-1 vector, using the STBlue-1 perfectly blunt cloning kit to generate two vectors for standardization of relative PCR primer pair efficiencies. Expression vectors for transcription factors PAX6, MAZ and ZNF263 were purchased from Origene, USA. The EGR1 expression vector (57) and pCX-EGFP vector expressing EGFP were obtained from Addgene, USA.

Western blotting, cryostat sectioning and immunostaining

Mouse brain tissues were lysed using the M-PER mammalian protein extraction reagent (Thermo Fisher Scientific, Ireland). Protein sample concentration was determined by Bradford reagent (Sigma, UK), as per manufacturer's instructions. Anti-SPRY3 antibody (Abcam, UK) was used at 1:500 dilution in western blotting. Western blotting method was previous described (58). For IHC, tissues from C57Bl/6 mice comprising whole brain, retina, SC and DRG were dissected and fixed in 4% Paraformaldehyde-Phosphate buffered saline (PFA-PBS) (pH 7.5) overnight and stored in 30% sucrose before sectioning. Cryostat sectioning was carried out on a Leica CM1950 cryostat. Brain was sectioned at 5 micron intervals and retina, SCG and DRG at 7 microns. Sections were mounted on superfrost slides (VWR International, USA) and stored at −80°C. For immunostaining, slides were placed in 10 mm sodium citrate buffer and microwaved for 15 min for antigen retrieval, followed by cooling in running water. Slides were then equilibrated in TBS (50 mm Tris-Cl, 150 mm NaCl, pH 7.5) and incubated with 3% H₂O₂-TBS (1:10 dilution of 30% H₂O₂ in TBS) to quench endogenous peroxidase. After three washes with TBS, the slides were blocked with 1.5% goat serum in TBS-Tx (TBS plus 0.1% Tx100) for 1 h at room temperature (RT). Slides were then incubated with primary antibody (anti-SPRY3 antibody 1:100 dilution; anti-MAZ antibody 1:200 dilution; anti-ZNF263 antibody 1:200 dilution) overnight at 4°C. Slides were then washed three times in TBS-Tx, and incubated with secondary antibody (1:200 dilution) for 1 h at RT. After 3 × 10 min washes with TBS-Tx, ABC complex (Vector Laboratories, UK) was applied to slides for 45 min. Slides were washed twice in TBS-Tx and once in TBS. 200 µl DAB staining buffer was placed on each slide for 2 min at RT. Slides were then washed with deionized water for 10 min. Slides were counterstained with haematoxylin, dehydrated through an alcohol series: 50%, 75%, 90%, 100% ethanol and a final step in Xylene and mounted under glass coverslips.

Cell culture and transfection

Human JAR cell line (ATCC, USA) was grown in T75 flasks at 37°C, 5% CO₂ with RPMI-1640 medium supplemented with 10% FBS, 2 mmol/L L-glutamine, 50 IU/mL penicillin and 50 μg/mL streptomycin. Twenty-four hours before transfection, cells were seeded at 4 × 10⁵ cells/well into 6-well tissue culture plate. For TF analysis, 2 μg SPRY3 promoter/β-Globin/LacZ reporter construct was transfected into JAR cells using lipofectamine 2000 (Life Technologies, UK), as per manufacturer's instructions. Three hours post incubation, the transfection medium was removed, JAR cells were re-transfected with transcriptional factor expression vectors (ZNF263, MAZ, PURA, EGR1 or PAX6) and the change in LacZ expression level was measured. Transfection of reporter and expression vectors was done separately to avoid plasmid interference (59). Forty-eight hours post-transfection, cells were harvested and LacZ activity was measured using the Pierce β-Galactosidase Assay Kit (Thermo Scientific, USA), as per manufacturer's instructions. To test the Spry3 shRNA vectors, 2 μg pQE-Spry3 expression vector and pSicoR-Spry3 vector were transfected into JAR cells using lipofectamine 2000. Twenty-four and forty-eight hours post-transfection, cell viability was measured by MTT assay (60). Total RNA was extracted using the TRI reagent and qRT-PCR was used to determine the Spry3 mRNA level. Human Hela cell line (ATCC, USA) was cultured in T75 flasks at 37°C, 5% CO_2, with DMEM supplemented with 10% FBS, 2 mmol/L L-glutamine, 50 IU/mL penicillin and 50 μg/mL streptomycin. Hela cell transfection was carried out using the Amaxa Nucleofector, as per manufacturer's instructions. Superior cervical ganglion neurons were cultured and transfected as described (61,62). Briefly, twenty SCGs from ten C57Bl/6 P1 mice were dissected using low power microscopy, pooled and treated with trypsin/EDTA (Sigma,UK) to dissociate the neurons. Co-transfection of pQE-Spry3 and pCX-EGFP vector expressing EGFP, and transfection of individual pSicoR-Spry3 or pSicoR-Psg22 vector, was carried out using the Neon Microporation kit (Invitrogen, USA), as per manufacturer's instructions. Neurons were seeded into four Greiner Cellstar dishes (Sigma, UK) in DMEM supplemented with 10% FBS and 10 ng/ml NGF (Sigma,UK). Twenty-four hours post-transfection, images of EGFP-labelled neurons were acquired using a Leica DMI 3000 microscope. Neuritic arbours were traced using MATLAB software (The MathWorks) and total neurite length and number of branch points were calculated and Sholl analysis was performed as previously described (30).

Transgenic mouse production

SPRY3 promoter/β-Globin/LacZ reporter vector was digested with NotI and SalI to remove the bacterial DNA vector backbone. The digestion product was purified using a Qiagen gel extraction kit. Transgenic mice were produced by microinjection of linearized vector DNA into the pronucleus of fertilized mouse embryos. (C57Bl/6xCBA)F1 female mice were used as embryo donors and treated with 5 IU pregnant mare's serum gonadotropin (PMSG) (Intervet, UK) in 0.1 ml PBS by intraperitoneal (IP) injection, and 46–50 h later received 5 IU human chorionic gonadotropin (hCG) in 0.1 ml PBS by IP injection. Embryo donors were mated with (C57Bl/6xCBA) F1 stud males immediately after hCG treatment. The following day embryos were collected in M2 medium (Sigma, UK) and injected with 1.5 ng/µl DNA as described (63,64). Embryo transfer into pseudopregnant CD1 female mice was carried out as described (64). Tail DNA of progeny was screened by PCR with primers hSPRY3/8F&8R. Founders were used to establish transgenic lines and F1 progeny were genotyped to identify transmission of the transgene. At least three male and three female F1 progeny were killed and their organs (brain, eye, DRG, SCG, spinal cord, lung, heart, thymus, liver, spleen and kidney) were harvested for X-gal whole-mount staining.

Lacz staining of mouse tissues

Mouse tissues were dissected and fixed in 4% PFA-PBS overnight. Tissues were washed three times in rinse buffer (100 mm sodium phosphate, 2 mm MgCl₂, 0.01% sodium deoxycholate, 0.02% NP-40) for 30 min. Tissues were then stained in staining buffer (rinse buffer plus 5 mm potassium ferricyanide, 5 mm potassium ferrocyanide and 1 mg/ml X-gal) for 48 h on a rocker at RT and post-fixed with 4% PFA-PBS overnight. Stained tissues were preserved in 30% sucrose for cryostat sectioning, or in 70% ethanol for long-term storage.

5′-RACE and 3′-RACE PCR

Human SPRY3 5′-RACE was performed on Hela cell and cerebellum RNA. Mouse Spry3 5′-RACE was performed on C57bl/6 cerebellum RNA. First strand cDNA synthesis was carried out using M-MLV reverse transcriptase (Life technologies, USA) and three SPRY3-specific reverse transcription primers (Supplementary Material, Fig. S2). Single-strand DNA was purified and a poly(A) tail was added to the 3′ end of the cDNA using terminal transferase (NEB, USA). PCR amplification was performed on poly(A)-tailed cDNA using gene-specific 5′-RACE PCR primers and a Linker primer (Supplementary Material, Fig. S2). The first PCR products were diluted 1000 times and nested PCR was performed using a gene-specific primer and Linker primer. The second PCR product was cloned into the pSTBlue-1 vector using a perfectly blunt cloning kit (Novagen, Germany), and sequenced. Human SPRY3 3′-RACE was performed similarly to 5′-RACE, except that the oligo(dT) primer was used in first strand cDNA synthesis and without the poly(A) addition step.

Fragment analysis of SPRY3 AG and GT repeat polymorphisms

The SPRY3 AG and GT repeats were amplified using 5′FAM labelled sense strand primer and an unlabelled reverse primer. PCR primer sequences for amplifying the GT and AG repeat regions were hSPRY3/11F&11R and hSPRY3/12F&12R, respectively. PCRs were performed in 15 μl reaction containing 50 ng DNA, 0.25 mm dNTPs, 1.5 mm MgCl₂, 0.25 units of AmpliTaq Gold enzyme, 1x BufferII and 10 pmol of each primer. A total of 2% DMSO was added to the AG repeat PCR reaction mix. PCR cycles were: 10 min at 95°C; 30 cycles of 30 s at 94°C, 45 s at 55°C, 1 min at 72°C; followed by a final extension step of 10 min at 72°C in an MJ Research PTC-200 thermocycler, with heated lid. After amplification, 1ul of sample was added to 12 μl deionized formamide and 0.5 μl fluorescence-labelled molecular weight marker (GS500 TAMRA) and denatured at 95°C for 5 min. Electrophoresis was carried out in an ABI 3130 DNA sequencer (Applied Biosystems, USA). Fragment sizes were calculated using GeneMapper v.3.7 software (Applied Biosystems, USA).

DNA methylation analysis

A total of 1 μg DNA from human cerebellum, cheek swab or lymphoblastoid cell lines (NA17209, NA17211 from Coriell HD200CAUC Caucasian collection) was bisulphite treated using EZ DNA Methylation-lightning kit (ZYMO Research, USA). Primer pairs hSPRY3/13F&13R and hSPRY3/14F&14R were used to generate PCR products, followed by pyrosequencing assays using the primers hSPRY3/13S or hSPRY3/14S, respectively. Single-stranded biotinylated PCR products were prepared for sequencing using the pyrosequencing vacuum prep tool (Biotage AB, Sweden). Pyrosequencing was performed according to manufacturer's instructions, using the PyroGold 96 SNP reagent kit (Qiagen, Germany) on the PSQ 96MA pyrosequencer (Biotage AB, Sweden). Degree of methylation at CpG sites was determined using AQ software (Biotage AB, Sweden).

Pax6 ChIP

Pax6 ChIP was carried out on P21 C57Bl/6 mouse cerebellum, using the ChIP-IT express kit (Active Motif, USA). Briefly, mouse cerebellum was dissected and fixed in 1% PFA-PBS for 10 min. Fixed cerebellum was lysed with lysis buffer and homogenized by dounce homogenizer with 10 strokes. Cell nuclei were pelleted by centrifugation at 8000 rpm at 4°C for 10 min. Chromatin was sheared by sonicator which yielded DNA fragments of 200–1000 bp. Goat anti-Pax6 antibody (Santa Cruz Biotechnology, USA) was used for pull-down. Goat anti-rabbit IgG antibody (Abcam, UK) was used as negative control. All reagents were supplemented with PMSF and proteinase inhibitor cocktail provided by the ChIP-IT express kit. The pull-down DNA was purified using the Qiagen PCR purification kit. qPCR was performed on purified DNA.

Supplementary Material

Supplementary Material is available at HMG online.

Funding

This work was supported by a Science Foundation Ireland (www.sfi.ie) Research Frontiers Program award to T.M. and a Higher Education Authority (www.hea.ie) PRTLI3 award under the National Development Plan to T.M.

Acknowledgements

The authors would like to thank J. Zhang (UCC) for technical help; M. Gill (TCD), L. Gallagher (TCD), D. Skuse (UCL) and an anonymous reviewer for discussion. We gratefully acknowledge the resources provided by the Autism Genetic Resource Exchange (AGRE) Consortium and the participating AGRE families. We also thank the Coriell Institute and other individuals who provided DNA, tissue samples and cell lines. The Autism Genetic Resource Exchange is a program of Autism Q2 Speaks and is supported, in part, by grant 1U24MH081810 from the National Institute of Mental Health to Clara M. Lajonchere (PI).

Conflict of Interest statement. None declared.