Distinct chromatin structures at the monoamine oxidase‐A promoter correlate with allele‐specific expression in SH‐SY5Y cells

Monoamine oxidase‐A (MAOA) metabolises monoamines and is implicated in the pathophysiology of psychiatric disorders. A polymorphic repetitive DNA domain, termed the uVNTR (upstream variable number tandem repeat), located at the promoter of the MAOA gene is a risk factor for many of these disorders. MAOA is on the X chromosome suggesting gender could play a role in regulation. We analysed MAOA regulation in the human female cell line, SH‐SY5Y, which is polymorphic for the uVNTR. This heterozygosity allowed us to correlate allele‐specific gene expression with allele‐specific transcription factor binding and epigenetic marks for MAOA. Gene regulation was analysed under basal conditions and in response to the mood stabiliser sodium valproate. Both alleles were transcriptionally active under basal growth conditions; however, the alleles showed distinct transcription factor binding and epigenetic marks at their respective promoters. Exposure of the cells to sodium valproate resulted in differential allelic expression which correlated with allele‐specific changes in distinct transcription factor binding and epigenetic marks at the region encompassing the uVNTR. Biochemically our model for MAOA promoter function has implications for gender differences in gene × environment responses in which the uVNTR has been implicated as a genetic risk.

Monoamine oxidase-A (MAOA) metabolises monoamines and is implicated in the pathophysiology of psychiatric disorders. A polymorphic repetitive DNA domain, termed the uVNTR (upstream variable number tandem repeat), located at the promoter of the MAOA gene is a risk factor for many of these disorders. MAOA is on the X chromosome suggesting gender could play a role in regulation. We analysed MAOA regulation in the human female cell line, SH-SY5Y, which is polymorphic for the uVNTR. This heterozygosity allowed us to correlate allele-specific gene expression with allele-specific transcription factor binding and epigenetic marks for MAOA. Gene regulation was analysed under basal conditions and in response to the mood stabiliser sodium valproate. Both alleles were transcriptionally active under basal growth conditions; however, the alleles showed distinct transcription factor binding and epigenetic marks at their respective promoters. Exposure of the cells to sodium valproate resulted in differential allelic expression which correlated with allele-specific changes in distinct transcription factor binding and epigenetic marks at the region encompassing the uVNTR. Biochemically our model for MAOA promoter function has implications for gender differences in gene × environment responses in which the uVNTR has been implicated as a genetic risk.  3 The uVNTR is a 30 base pair repeat element that can be present in 2, 3, 3.5, 4 or 5 copies. The 2, 3 and 5 repeat variants are defined as low expression (MAOA-L) alleles whilst the 3.5 and 4 repeat uVNTR are considered high expression variants (MAOA-H). 4 These low and high expression alleles can represent either a risk or a protective factor depending on the disorder studied. [4][5][6][7][8][9][10][11] It is thought that the uVNTR acts as a classical transcriptional regulator in the promoter of the MAOA gene and that the distinct variants support differential MAOA expression in a tissuespecific and stimulus inducible manner. This gene and the uVNTR polymorphism are amongst the most often cited example of a genotype by environment interaction (G × E) in which an individual's genotype moderates the effect of environmental experience to alter mental health outcomes. 6 Simplistically, the function of the uVNTR is modified by the complement of active transcription factors in the cell, altered in response to the environment, that are then able to recognise and bind to the uVNTR thus allowing the domain to act as a transcriptional regulator. However, bioinformatic data now indicates that there are at least 2 transcriptional start sites (TSSs) for the MAOA gene, a further isoform is produced from a more 5 0 TSS. This newly described TSS now places the uVNTR in the 5 0 UTR (5 0 untranslated region) of the MAOA gene, Figure 1, adding to the complexity of transcriptional regulation at this locus.
MAOA is located on the X chromosome and therefore genetic associations of the uVNTR with neuropsychiatric disorders in the male are simplified by having only one allele. However, associations of the uVNTR in females are complicated by not knowing the contribution of each allele to MAOA expression and the potential for one allele to exhibit X inactivation. This problem is made more difficult as the MAOA locus has been proposed as one of the loci that can escape X inactivation. [12][13][14] To address the mechanisms which might underpin allele expression in the female we took advantage of the human neuroblastoma-derived SH-SY5Y cell line which has a female karyotype and is heterozygous for the uVNTR. This heterozygosity of the uVNTR allows us to distinguish allelic-specific differences in epigenetic marks and transcription factor binding to the DNA encompassing the uVNTR in addition to identification of allele-specific expression directed by the most 5 0 TSS. In this communication we were able to show that both alleles, at least in the population of SH-SY5Y cells, can support mRNA expression. This phenomenon was also observed in cDNA prepared from saliva from females who were heterozygous for the uVNTR.
Using chromatin immunoprecipitation (ChIP) analysis in SH-SY5Y cells we show that each allele can support distinct transcription factor, histone-binding and methylation patterns. The pattern of factor binding and methylation over each allele was also correlated with a switch in allele-specific gene expression in response to sodium valproate in SH-SY5Y cells.

| Total RNA preparation and cDNA synthesis
Total RNA was extracted from SH-SY5Y cells using Trizol reagent (Invitrogen, Paisley, UK) and quantified using a NanoDrop 8000 Spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA).
Three micrograms of total RNA was reverse-transcribed to singlestranded cDNA using GoScript Reverse Transcription System (Promega, Southampton, UK) and random primers according to the manufacturers' instructions.

| Polymerase chain reaction
Amplification of the MAOA uVNTR was performed using the primers previously described, 3

| Chromatin immunoprecipitation
Human SH-SY5Y cells were grown to 80% confluence in a T175 flask and treated for 1 hour under the following conditions, basal FIGURE 1 Graphic representation of MAOA promoter region based on The University of California Santa Cruz (UCSC) genome web browser, Hg38. From top to bottom: reference gene sequence; previous version of UCSC genome using Hg19; upstream tandem repeat: uVNTR; CpG islands; Encyclopedia of DNA Elements (ENCODE) transcription factor data (data version: ENCODE Mar 2012 Freeze) indicating those factors shown to bind to this region in specific cell lines, in short the darker the line the more confidence in the association; ENCODE data for H3K27Ac mark (often found near active regulatory elements) on 7 cell lines (untreated), 2 μM sodium valproate and vehicle alone. Samples were processed using the methods described by Murgatroyd et al. 16 The immunoprecipitations were performed using the following antibodies:  Attempts to use qPCR to measure allele-specific expression over heterozygous VNTRs were problematic perhaps due to preferential amplification of a specific allele in early rounds of PCR and consistent, FIGURE 2 Expression of 2 MAOA isoforms: 1.5% agarose gel of MAOA uVNTR PCR amplification. Amplicon size 324 bp for 3R uVNTR, 354 bp for 4R uVNTR. A, SH-SY5Y cell line cDNA: lanes from left to right: 100 bp ladder marker; basal condition; vehicle control: H 2 O 1 hour; sodium valproate treatment 1 hour; PCR negative control. B, Human female cDNA samples: lanes from left to right: 100 bp ladder marker; 2× homozygous 3R/3R uVNTR; 3× heterozygous 3R/4R uVNTR reproducible data was difficult to obtain. However, using the heterozygous cell line SH-SY5Y we could utilise the size variation of the distinct uVNTRs to compare intensity between and within experiments for the same primer set; this showed that in SH-SY5Y cells in response to exposure to sodium valproate, a mood stabiliser, a switch in the levels of MAOA allelic expression was observed. Specifically, the ratio of L-to S-allele expression was reproducibly reversed, Figure 2A. This switch in relative levels of expression cannot be explained by PCR allele preference. As the MAOA gene is on the X chromosome the expression of both alleles in the SH-SY5Y cells might require this region to escape X inactivation as has been previously reported for the MAOA locus. [12][13][14] Alternatively, the expression of both alleles could be construed as a mosaic effect of random silencing of one allele in any particular cell and thus the population would appear to express both alleles.
To explore the allelic expression further we analysed by ChIP transcriptional and epigenetic variation at the uVNTR domain in SH-SY5Y cells in response to sodium valproate. The methylation status of both alleles was addressed by their ability to bind Methyl-CpG-binding domain (MBD). This showed, by addressing the ratio of each uVNTR allele, a clear specificity in that the 4 copy uVNTR promoter was predominantly methylated under basal growth conditions whereas the 3 copy uVNTR promoter was hypomethylated, Figure 3. The methylation pattern was altered when the cells were exposed to sodium valproate. Specifically, variation was mostly clearly observed in the methylated fractions, in which methylation could now be easily observed over the 3 copy allele; the ratio of methylated 4 copy allele to 3 copy allele was quite distinct from basal conditions, Figure 3. We therefore explored binding of RNA polymerase II and the transcription factors CTCF and SP1 identified from ENCODE data as binding to the uVNTR, Figure 1, to gain insight into potential transcriptional mechanisms operating at this locus.
RNA polymerase II was showed to be present on both alleles under basal conditions, consistent with expression from both alleles, Figure 4A, lane 4. This was in stark contrast to Histone 3 (His3) under basal conditions, Figure 4B, lane 4 (His3), which was found predominantly on the shorter, 3 copy repeat, uVNTR allele. Similar to the observations obtained for His3, under basal growth conditions, or exposure to H 2 0 as the vehicle control, CTCF and SP1 transcription factor binding was predominantly observed on the short, 3 copy uVNTR allele, Figure 4C, top and middle panel, lanes 7 and 8. Upon exposure to sodium valproate, Figure 4C, bottom panel, lanes 8, there was a significant increase for SP1 binding to the short 3 copy allele which is correlated with the switch to increased expression of the longer 5 0 TSS containing the 3 copy uVNTR in the 5 0 UTR observed in Figure 2A.
The distinct promoter architecture is suggestive of a distinct/specific mechanism to regulate each MAOA-allelic promoter based in part on uVNTR repeat copy number. In the literature, CTCF and SP1 both bind double stranded DNA but recognise the same sequences as the single stranded DNA-binding proteins, hnRNPK and CNBP. 18,19 These single stranded DNA-binding proteins were also bound predominantly to the short 3 copy allele under basal conditions, Figure 4C top and centre panel, lanes 4 and 5, but in response to valproate they now also bound the 4 copy allele, Figure 4C, bottom panel, lanes 4 and 5. This was in stark contrast to SP1 binding which was dramatically increased only on the 3 copy allele. In contrast the binding of CTCF was not significantly changed in response to valproate with any change in distribution, it remained bound predominantly to the 3 copy allele. CTCF function on transcription is varied, ranging from repression to transcriptional pausing and transactivation which in part is dependent on interacting partners, one of which is nucleolin which can mediate intra-and interchromosomal interactions. 20 Analysis of nucleolin by ChIP indicated it binding weakly to the 3 copy allele under basal conditions with a vast increase in binding after exposure of the cells to valproate, Figure 4C, bottom panel, lane 6. It should be noted that nucleolin can also interact with SP1 and both these factors were dramatically increased in their binding to the short allele after exposure to valproate.

| DISCUSSION
The MAOA uVNTR is one of the best characterised G × E interactions in the literature and the accepted hypothesis is that the different repeat copy numbers of the uVNTR direct the differential expression of MAOA in response to the environment. This is a relatively straightforward explanation for the male gender which has only one allele for MAOA. However, the situation is more complex in females for several reasons including (1) gene dosage, (2) the potential for MAOA to escape X-inactivation and (3) heterozygosity of the uVNTR whose genotype might direct distinct MAOA gene expression patterns.   Figure 4A. The methylation of one allele which apparently did not repress MAOA expression, Figure 3, would be consistent with recent work on X chromosome gene expression patterns which indicated that methylation did not necessarily correlate with gene repression but rather functioned as a parameter affecting a region's ability to escape X inactivation. 13 To further investigate the regulation of MAOA gene expression by the uVNTR we analysed binding for 5 transcription factors: CTCF, SP1, hnRNP K, CNBP and nucleolin. Our data provided evidence that these proteins recognised the proximal MAOA promoter encompassing the uVNTR and more importantly a distinct pattern of binding was observed over each individual allele which was modulated both by genotype of the uVNTR and exposure of the cells to sodium valproate, Figure 4C. This data was consistent with a role for gender in regulation of MAOA expression via epigenetic mechanisms, for example, the gender-specific changes of methylation in a longitudinal study of twins 21 and hypomethylation in the pathogenesis of panic disorder particularly in female patients. 22 Our model could in part explain the differences in uVNTR genotype association for specific mental health issues when male and females are compared, such examples include: MAOA uVNTR moderating the relationship between childhood maltreatment and dysthymia only in females, 23 gender differences including pronounced effects on Attention-deficit hyperactivity disorder (ADHD) and anxiety in females whereas the opposite is true for autism, bipolar disorder and aggressive behaviour in males. 4,24 Indeed many studies of MAOA G × E report only on males due to compounding affects including female heterozygosity for the uVNTR. 6,8,11 This regulation could be further complicated in that the MAOA gene is one of the approximately 15% of X chromosome genes estimated to escape X inactivation. [12][13][14] However, there is still much debate about X inactivation in vivo and there may be no reason why X-inactivation of the second allele in females itself is not dependent on G × E interactions. Consistent with this, the gene termed a master regulator of X chromosome inactivation, X Inactive Specific Transcript gene (XIST), is over expressed in females with major affective disorders. 25 Thus, it is likely that both sex-based and disorder-based differences exist and our study highlights one potential mechanism to explain such differences.