Presence of Multiple Functional Polyadenylation Signals and a Single Nucleotide Polymorphism in the 3′ Untranslated Region of the Human Serotonin Transporter Gene

Abstract

Abstract: The human serotonin transporter (hSERT) gene is a candidate for involvement in the aetiology of affective disorders. In humans, multiple transcripts of the gene have been detected by northern blot analysis of brain and other tissues. We performed 3′ rapid amplification of cDNA ends to identify the common sites of polyadenylation in hSERT mRNA from human JAR cells and whole blood. Two major polyadenylation sites were identified: one 567 bp downstream of the stop codon, consistent with the usage of the polyadenylation signal AATGAA, and a second site 690 bp downstream of the stop codon. The putative polyadenylation signal upstream of this site contained a single nucleotide polymorphism (AG/TTAAC). However, allelic variation at this site did not influence polyadenylation site usage, and there were no significant differences in the abundance of the two alleles of this polymorphism between 329 control subjects, 158 individuals with major depression, and 130 individuals with bipolar affective disorder. This single nucleotide polymorphism in the 3′ untranslated region of the hSERT gene should provide a useful genetic marker in the evaluation of hSERT as a candidate gene influencing susceptibility to mood disorders.

Several lines of evidence suggest that alterations in the function of the human serotonin transporter (hSERT) may be implicated in the pathogenesis of affective disorders. hSERT terminates neurotransmission by the reuptake of serotonin into the presynaptic terminal and is the site of action of the selective serotonin reuptake inhibitor group of antidepressants (Owens and Nemeroff, 1994). Mood disorders are generally thought to be polygenic diseases, and the hSERT gene has been identified as a possible candidate among genetic factors that might confer susceptibility to depression (Harmar et al., 1996). Two variable number tandem repeat polymorphisms have been described within the gene, one in intron 2 (Lesch et al., 1994; Battersby et al., 1996) and a second (5-HTTLPR) close to the promoter region ∼1 kb upstream of the transcription start site (Heils et al., 1996).

Association studies comparing the allelic frequencies of the two polymorphisms in control subjects and patients with affective disorder have given conflicting results. Two studies (Collier et al., 1996b; Furlong et al., 1998) have demonstrated a weak association between alleles of the 5-HTTLPR and affective disorder. In addition, several studies have reported significant associations between alleles of the intron 2 polymorphism and bipolar (Collier et al., 1996a; Kunugi et al., 1997; Rees et al., 1997) and unipolar (Battersby et al., 1996) depression. However, other investigators have been unable to demonstrate associations between these polymorphisms and mood disorder (Esterling et al., 1998; Hoehe et al., 1998). Definitive studies of the role of hSERT in affective disorder will require the identification of further polymorphic markers within and flanking the gene.

There are large differences in hSERT mRNA sizes seen in different tissues in vivo, possibly resulting from the use of alternative polyadenylation sites (Bradley and Blakely, 1997). We report here the use of the 3′ rapid amplification of cDNA ends (RACE) technique to identify two commonly used sites of polyadenylation in the hSERT gene and describe a polymorphism within a putative polyadenylation signal for one of these sites.

MATERIALS AND METHODS

DNA isolation

Genomic DNA was isolated from whole venous blood as described previously (Smith et al., 1992). One hundred microlitres of whole blood was washed three times in TE buffer (10 mM Tris-HCl, pH 8, 1 mM EDTA), and peripheral leukocytes were harvested and resuspended in 100 μl of lysis buffer (50 mM KCl, pH 8.3, 2.5 mM MgCl₂, 0.45% Nonidet P-40, 0.45% Tween 20) containing 200 mg/ml proteinase K. Lysis was completed by incubation for 20 min at 55°C. Lysates were diluted with an equal volume of sterile distilled water and heated to 96°C for 10 min to inactivate the proteinase. In some cases, DNA was isolated using the Nucleon Genomic DNA extraction kit according to the manufacturer’s instructions.

Genotyping of polymorphism in 3′ untranslated region

Target DNA (3 μl of lysate or 100 ng of purified DNA) was amplified by PCR using oligonucleotide primers: HTT.PCR1: 5′-CCGCTTGAATGCTGTGTAACACAC; and HTT.PCR3: 5′-GTACCCTTCCAATAATAACCTCC. PCR was undertaken using 1.5 U of Taq polymerase (Promega), 100 ng of each primer, 200 μM each of dATP, dGTP, dCTP, and dTTP, and 1.5 mM MgCl₂ in 50 μl of PCR buffer. PCR was carried out in a Hybaid Omnigene thermal cycler, using 30 s of denaturation at 94°C, 30 s of primer annealing at 56°C, and 45 s of polymerization at 72°C for 35 cycles. A final polymerization step of 2 min was carried out to complete elongation of all strands.

Genotyping of a single nucleotide polymorphism within the region of the hSERT gene amplified by primers HTT.PCR1 and HTT.PCR3 was performed by digestion with the restriction enzyme Tru9 1 (NBL). Fifteen microlitres of each PCR fragment was incubated overnight at 65°C in 3 U of Tru9 1 in a total volume of 50 μl of 1× restriction buffer. In addition, in some cases, PCR products were cut overnight at 55°C in 2 U of MaeIII restriction endonuclease (Boehringer) in a total volume of 50 μl of 1× restriction buffer. Digests were resolved on 2% agarose gels, and bands were visualised by ethidium bromide staining and UV transillumination.

DNA sequencing

PCR products were resolved on 1% agarose gels, and bands were excised and purified by the Wizard PCR DNA purification system (Promega). DNA was concentrated using Microcon 100 columns (Amicon). For characterisation of the polymorphism, 30 samples were sequenced using the Prism Dye Terminator Cycle Sequencing kit with Amplitaq DNA polymerase, FS (Perkin-Elmer) in a Hybaid Omnigene thermal cycler. Thirty cycles of sequencing consisted of 30 s at 96°C, 15 s at 50°C, and 4 min at 60°C. PCR fragments were sequenced with primers HTT.PCR1 and HTT.PCR3 together with primer 3′ untranslated region forward (UTRF): 5′-GTTCATGAATACGTAAACTGCG. 3′ RACE products were sequenced as above using primer 3′UTRF. Unincorporated nucleotides were removed by precipitation with 95% ethanol. Electrophoresis was carried out on an Applied Biosystems model 373 Stretch DNA Sequencer at a constant power of 30 W for 12 h using a 4.75% denaturing polyacrylamide gel.

RNA extraction

RNA was extracted from 23 samples of whole blood and from human placental JAR cells by the method of Chomczynski and Sacchi (1987) using the RNAzol LS and RNAzol reagents (Biogenesis), according to the manufacturer’s instructions. In brief, 2 ml of frozen whole blood was diluted 1:3 in diethyl pyrocarbonate (DEPC)-treated water and thawed in 9 ml of RNAzol LS. The samples were chloroform-treated and centrifuged, and the aqueous layer precipitated overnight in isopropanol. Samples were ethanol-washed and resuspended in DEPC-treated water. A 75-mm² flask of JAR cells was homogenised in 2 ml of RNAzol and choroform-treated, and the RNA precipitated as above.

3′ RACE

3′ RACE reactions were performed using a 3′ RACE kit from Life Technologies according to the manufacturer’s instructions. In brief, 5 μg of RNA was incubated for 10 min at 70°C with 1 μl of 10 μM adapter primer in 12 μl of DEPC-treated water. Two and five-tenths microlitres of 10 mM dNTPs, 5 μl of 0.1 M dithiothreitol, 5 μl of 25 mM MgCl₂, 7 μl of water, and 10 μl of 10× PCR buffer were added and heated to 42°C. One microlitre (200 U) of Superscript II reverse transcriptase was added, and samples were incubated for 50 min at 50°C. The reverse transcriptase was inactivated by heating for 15 min at 70°C, and the RACE reactions were treated with 2 U of RNase H for 20 min at 37°C to destroy the RNA. For the PCR stages of the 3′ RACE, 5 μl of each RACE reaction was amplified in the presence of 2 mM MgCl₂, 200 μM dNTPs, 100 ng of an abridged universal amplification primer (AUAP), and 100 ng of primer HTT.PCR1 in 50 μl of 1× PCR buffer. DNA was amplified with 35 cycles of denaturation at 94°C for 30 s, annealing at 56°C for 30 s, and elongation at 72°C for 1 min. A second round of nested PCR was performed using the same buffer conditions as above with 1 μl of the amplification product from the first round and the primers AUAP and 3′UTRF, under the same conditions but with elongation for 30 s at 72°C.

Genotyping of other hSERT polymorphisms

Genotyping for the intron 2 polymorphism of hSERT was performed as described previously (Battersby et al., 1996). For the 5-HTTLPR polymorphism, amplification was carried out using 100 ng each of forward primer 5′-CACCTAACCCCTAATGTCCCTACT and reverse primer 5′-GGACTGAGCTGGACAACCAC in a reaction volume of 50 μl containing 1×Pfu buffer, 200 μM each of dATP, dCTP, and dTTP together with 100 μM dGTP and 100 μM 7-deaza-GTP. The reaction was heated to 98°C for 5 min, and 2.5 U of Pfu Exo-minus polymerase (Stratagene) was added. Amplification consisted of 40 cycles of 98°C for 45 s, 65°C for 45 s, and 72°C for 90 s. Products were resolved on 2-3% agarose gels, and bands were visualised by ethidium bromide staining under UV transillumination.

Subjects

The design of the study was approved by the relevant committee for Medical Ethics. For the polymorphism in the 3′ untranslated region of hSERT, 158 individuals with single or recurrent major depressive episodes and 130 individuals with bipolar disorder were compared with a group of 329 controls. Patients with major affective disorder were recruited from the in-patient and out-patient populations of the Royal Edinburgh Hospital. All patients met DSM-III-R criteria for major depressive disorder or bipolar disorder and also the probable Research Diagnostic Criteria according to the Schedule for Affective Disorders and Schizophrenia (Lifetime version) (SADS-LA) (Endicott and Spitzer, 1978). One hundred ten control samples were obtained from volunteers who were screened by brief interview to exclude past psychiatric illness, as well as 219 anonymous donors from the Scottish Blood Transfusion service.

A detailed retrospective case note analysis was undertaken on 100 of the sample (52 bipolar and 48 unipolar) selected randomly so as to be matched for genotypes, to focus on gaining an impression of clinical severity. Of these, the mean age of onset was 29.2 years, the mean number of psychiatric hospital admissions 4.5, and the mean number of days in hospital 320.8. Of these patients, 39% had been treated with electroconvulsive therapy, having had a mean number of 5.9 treatments. Forty-six percent had experienced mood congruent psychotic symptoms, and 41% had been so ill as to have required involuntary hospital detention under the mental health act at some point.

Statistical methods

Analysis of allele and genotype distribution was carried out on the raw frequencies by the χ² test using the Statistical Package for the Social Sciences (SPSS Apple Macintosh version 4.0). Haplotype analysis was performed using the Arlequin software package (Schneider et al., 1997).

RESULTS

3′ RACE

3′ RACE was performed using nested PCR, with primer pairs AUAP and HTT.PCR1 for the first round and pairs AUAP and 3′UTRF for the second round. Two major RACE products of ∼340 and 220 bp were identified from mRNA extracted from JAR choriocarcinoma cells or whole human blood (Fig. 1A). In addition, a weaker band ∼290 bp in size, together with other weakly positive bands, was seen both on gels (Fig. 1A) and by Southern hybridisation (data not shown). Sequencing showed that the products of 340 and 220 bp corresponded to usage of two polyadenylation sites 567 and 690 bp downstream of the stop codon, as illustrated in Fig. 2. The RACE products of intermediate size were consistent with usage of the proximal polyadenylation site, but with size differences corresponding to large variation in the length of the poly(A) tail. The two major 3′ RACE amplification products were not seen in every human RNA sample examined. In some cases, the longest RACE product was the only one seen, whereas in others only the shortest fragment was present (Fig. 1A).

Figure 1. A: 3′ RACE products for the hSERT gene; RNA extracted from human placental JAR cells (lane 1) and whole human blood (lanes 2-11). Bands of 340 and 220 bp correspond to usage of two major polyadenylation sites. Intermediate bands are consistent with usage of the proximal site with variable length of poly(A) tail. B: Restriction digests of PCR products from the 3′ untranslated region of hSERT gene using Tru9 1 (A) and MaeIII (B). Three genotypes are shown from six individuals. Subjects 1 and 2 (lanes 1-4), ATTAAC/ATTAAC. Subjects 3 and 4 (lanes 5-8), ATTAAC/AGTAAC. Subjects 5 and 6 (lanes 9-12), AGTAAC/AGTAAC.

Figure 2. Sequence of 741-bp PCR product from exon 14 of hSERT gene. End of coding sequence is indicated. Two major polyadenylation sites are indicated in bold type, and putative polyadenylation signals are underlined. G/T polymorphism is present at base 689.

FIG. 1.

FIG. 2.

Characterisation of the polymorphism

Sequencing of the PCR products from the 3′ untranslated region of the hSERT gene revealed a fragment 741 bp in length (Fig. 2). DNA fragments from 30 samples of human genomic DNA were sequenced, and a single nucleotide polymorphism was identified at base 689, where either a T or G nucleotide was present, the sequence being either ATTAAC or AGTAAC (Fig. 2). In 11 cases, electropherograms showed both T and G peaks. To confirm the presence of a polymorphism at base 689, the PCR products were digested with the restriction endonuclease Tru9 1, which cuts the sequence TTAA, but not GTAA (Fig. 1B). Both alleles from samples that sequence data indicated were homozygous for the T allele were cut by Tru9 1, giving bands of 52 and 689 bp, whereas those that were homozygous for G were not affected by restriction digestion. Samples in which both bases were present on sequencing had bands of 52, 689, and 741 bp on restriction digestion (Fig. 1B). To verify further the presence of a polymorphism, the PCR products were digested with the enzyme MaeIII, which cuts the sequence GTNAC. There were three sites for this enzyme in those samples with a G at base 689. Samples with two G alleles had bands of 15, 215, 458, and 53 bp. Those with two T alleles had bands of 15, 215, and 511 bp. Heterozygotes showed bands of 15, 53, 215, 458, and 511 bp. There was no consistent pattern of polyadenylation in RNA from human blood in relation to genotyping of the polymorphism in the putative polyadenylation signal.

Association study

DNA samples from control subjects and patients with affective disorder were genotyped by amplification of the 3′ untranslated region of the hSERT gene followed by Tru9 1 digestion and resolution of products on 2% agarose gels. Table 1 illustrates the distribution of allele frequencies and genotypes in control and patient samples. The control genotype distribution was close to Hardy-Weinberg equilibrium. There was no significant difference in allele or genotype frequencies between the control and affective disorder groups.

Table 1. Allele and genotype frequencies of polymorphism in 3′ untranslated region of hSERT in control and patient groups

		Genotypes	Alleles
	n	T/G	T/T	G/G	Total	T	G
Values in parentheses represent percent frequencies.
Control	329	163 (50)	97 (29)	69 (21)	658	357 (54)	301 (46)
All affective disorder	288	147 (51)	95 (33)	46 (16)	576	337 (59)	239 (41)
Unipolar	158	76 (48)	54 (34)	28 (18)	316	184 (58)	132 (42)
Bipolar	130	71 (55)	41 (31)	18 (14)	260	153 (59)	107 (41)

TABLE 1.

Haplotype analysis

Haplotype analysis was performed for three polymorphisms in the hSERT gene. Two alleles were identified for the 5-HTTLPR polymorphism (long and short), whereas three alleles were present for the intron 2 polymorphism (nine, 10, or 12 repetitive elements). Haplotype analysis showed linkage disequilibrium between the three polymorphisms, with the long/10/T and short/12/G haplotypes occurring most frequently (Table 2). The hypothesis of random distribution of the 12 haplotypes among the control and bipolar and unipolar depression populations was tested using the population pairwise differentiation test (Raymond and Rousset, 1995). This test provided evidence that the distribution of haplotypes differed between the control and bipolar affective disorder populations (p < 0.05). Differences between the population frequencies for long/10/G (p = 0.063) and short/12/G (p = 0.062) haplotypes (Table 2) provide the most probable source for this effect.

Table 2. Distribution of haplotype frequencies for three polymorphisms in three populations for the hSERT gene

	Group
Haplotype	Control	Bipolar	Unipolar
Estimated haplotype frequencies for control subjects, bipolar disorder, and unipolar disorder are presented. The values in parentheses represent the standard deviations of the estimates. p > 0.1 for all comparisons except as noted. ^aT statistic = -1.867, p = 0.063, control vs. bipolar. ^bT statistic = 1.875, p = 0.062, control vs. bipolar. ^cT statistic = 2.133, p = 0.034, control vs. unipolar. ^dT statistic = -1.679, p = 0.094, control vs. unipolar.
Long/10/G	0.038 (0.012)	0.011 (0.008)^a	0.026 (0.015)
Long/10/T	0.245 (0.023)	0.306 (0.044)	0.253 (0.032)
Long/12/G	0.137 (0.018)	0.116 (0.030)	0.153 (0.027)
Long/12/T	0.121 (0.017)	0.177 (0.030)	0.141 (0.029)
Long/9/G	0.000	0.000	0.005 (0.005)
Long/9/T	0.012 (0.006)	0.004 (0.006)	0.012 (0.013)
Short/10/G	0.010 (0.009)	0.005 (0.007)	0.001 (0.006)
Short/10/T	0.078 (0.015)	0.077 (0.024)	0.059 (0.017)
Short/12/G	0.300 (0.024)	0.226 (0.032)^b	0.218 (0.030)^c
Short/12/T	0.058 (0.016)	0.071 (0.022)	0.106 (0.024)^d
Short/9/G	0.000	0.000	0.000
Short/9/T	0.000	0.008 (0.007)	0.015 (0.009)

TABLE 2.

DISCUSSION

The classical hexanucleotide sequence AATAAA, which is highly conserved among mammalian polyadenylation signals (Wickens, 1990) and is functional in the dopamine transporter (Kawarai et al., 1997), was not found in the 3′ untranslated region of the hSERT gene. Instead, we found evidence of two polyadenylation sites, consistent with the use of two of the three noncanonical polyadenylation signals suggested by Bradley and Blakely (1997). The more distal of the two polyadenylation signals that we detected (ATTAAC) was also identified by Heils et al. (1995) and is similar in sequence and location to functional polyadenylation signals in the rat and mouse serotonin transporter genes (ATTAAAC) (Gregor et al., 1993; Heils et al., 1995; Chang et al., 1996). However, we cannot exclude the possibility that polyadenylation signals downstream of the region of the hSERT gene that we have sequenced may give rise to longer transcripts.

The functional significance of the polymorphism that we have identified in the ATTAAC polyadenylation signal remains to be established. Although there was no evidence that polymorphic variation at this locus influenced polyadenylation site usage, the 3′ RACE technique used was not quantitative.

No significant association was seen between the polymorphism in the 3′ untranslated region of the hSERT gene and susceptibility to affective disorder, although our initial analysis has suggested some differences in haplotype distribution between the control group and patients with mood disorder. As suggested by Collier (1998), a definitive assessment of the contribution of genetic variation in the serotonin transporter gene to susceptibility to mood disorders will require the analysis of haplotypes of many markers within and around the gene, probably using large family trio or sib-pair samples. The polymorphism described here should provide a useful tool in such studies.

Acknowledgements

We thank Lilly Industries plc for financial support, Margaret van Beck for help in data management, and Norma Brearley for her assistance in preparation of the manuscript.