In this issue of Gut, Yeo and colleagues¹ report on the association between a functional polymorphism in the serotonin transporter gene and diarrhoea predominant irritable bowel syndrome (D-IBS) in women (see page 1452). In a study of 194 North American female participants with D-IBS in a clinical trial programme and 448 female controls, there was an association between the homozygous short genotype of SERT-P (serotonin reuptake transporter gene) and the D-IBS phenotype, with an odds ratio of 2.25 (95% confidence interval 1.51–3.31). The fact that the confidence interval does not cross the value of 1 suggests that the association is statistically significant, and the authors suggest that the SERT-P may be a candidate gene for D-IBS in women.

To place these interesting observations in perspective, the reader may wish to address the following. What is SERT? What is the theoretical consequence of a polymorphism of the SERT-P gene? What happens to animals when the SERT gene is knocked out? Have other studies addressed the association of this polymorphism and IBS? Are there any pitfalls in the interpretation of such association studies?

WHAT IS SERT?

Serotonin (5-hydroxtryptamine, 5-HT) is secreted in copious amounts from gut enteroendocrine cells and serves as a critical messenger for gastrointestinal fluid secretion and gut motility.^2,3 There are seven subclasses of serotonergic receptors, differentiated on the basis of structure, molecular mechanism, and function.⁴ In contrast with the remarkable diversity of 5-HT receptors, only a single protein, the 5-HT transporter (or SERT), mediates reuptake of 5-HT from the synaptic cleft and thus termination of its action. The approved gene symbol for SERT is SLC6A4 (solute carrier family 6 member 4); this abbreviation is used in other papers and this may be confusing to readers. Human SERT is encoded by a single gene on chromosome 17q11 and is composed of 14–15 exons. Transcriptional activity of the SERT gene, SERT availability, and ultimately 5-HT reuptake is modulated by a polymorphic repetitive element unique to humans and simian primates, the 5-HTT gene linked polymorphic region (5-HTTLPR) upstream of the transcription start site. Additional variations have been described in the 5′ untranslated region (5′UTR), in intron 2, and 3′UTR (reviewed by Reif and Lesch⁵).

Neurotransmitter transporters are channel-like proteins that are involved in chemical signalling in the brain and periphery; in fact they are considered to do the heavy lifting in neurotransmitter inactivation.^5,6 SERT in the gut is similar to that in the brain of the same species.⁷ To control 5-HT actions in the gut and limit 5-HT receptor desensitisation, both neurones and crypt epithelial cells synthesise SERT proteins.^8,9

WHAT IS THE THEORETICAL CONSEQUENCE OF A POLYMORPHISM OF THE SERT-P GENE?

In elegant functional studies, Lesch et al showed that, compared with the homozygous long genotype, polymorphic homozygous short and heterozygous SERT-P were associated with reduced function of the transporter protein in a lymphoblastoid cell line.¹⁰ Figure 1 illustrates the consequence of a less effective reuptake process for 5-HT, as might occur in a patient with an s/s homozygous or l/s heterozygous polymorphism.

WHAT HAPPENS TO ANIMALS WHEN THE SERT GENE IS KNOCKED OUT OR IS DECREASED BY DISEASE?

SERT deficient mice display increased anxiety related behaviours based on increased serotonergic neurotransmission resulting in desensitised and downregulated 5-HT_1A¹¹ and 5-HT_2A or 5-HT_2C receptors.¹² Gastrointestinal motility is also abnormal in SERT knockout mice.¹³ Adaptive changes occur in the subunit composition of enteric 5-HT₃ receptors in these knockout mice. Such changes are reflected in altered 5-HT₃ receptor affinity and desensitisation in the response of the receptor to 5-HT released from enteroendocrine cells.¹⁴

An intriguing recent observation suggests that experimental colitis alters 5-HT signalling by increasing 5-HT availability while decreasing 5-HT reuptake.¹⁵ The authors speculated that altered 5-HT availability might contribute to the dysmotility of inflammatory bowel disease, possibly due to desensitisation of 5-HT receptors.

HAVE OTHER STUDIES ADDRESSED THE ASSOCIATION OF THIS POLYMORPHISM AND IBS?

There have been five studies that explored the association of SERT polymorphisms and IBS; results from the three genetic association studies are summarised in fig 2.

The first report was a study from Turkey in 54 IBS patients (18 with D-IBS) and 91 healthy controls. Pata et al examined possible associations between SERT polymorphisms and the different clinical patterns of IBS. They observed that the l/s genotype was present in 18 D-IBS patients with a frequency of 88% whereas none of the patients had the l/l genotype. Overall, the distribution of SERT polymorphisms were similar in healthy subjects and the whole group of IBS patients.¹⁶

The second report by Kim et al at the Mayo Clinic, which studied people residing in the Mid-West in the USA, was recently published in Gut.¹⁷ The study explored the distributions of genetic polymorphisms for SERT-P, α_2C Del 322–325, and α_2A−1291 (C→G) polymorphisms in 274 patients with lower functional gastrointestinal disorders (FGID) and 120 controls. The distribution of SERT polymorphisms was not significantly different between the lower FGID patients versus controls (an observation that replicated the result in the whole IBS group studied in Turkey) or between the subset with the phenotype of D-IBS (n = 128) versus controls. In the final analysis of 90 C-IBS patients reported by Kim et al, the odds ratio for SERT-P l/s or s/s polymorphisms was not significant (0.7; 95% confidence interval 0.4, 1.2). This study included a post hoc analysis to determine whether clinically meaningful associations of the three candidate genotypes could have been detected with at least 80% power. Thus it was reported that the sample size of the study could detect a difference in prevalence of wild type (l/l) versus polymorphic (l/s or s/s) genotypes of 12% for all lower FGID, and 19–20% for IBS-C and IBS-D. On the other hand, the study identified that combinations of polymorphisms were associated with high somatic symptom scores: α_2C Del 322–325 with SERT-P (odds ratio 5.0 (95% confidence interval 1.11, 22.22)).¹⁷ This suggests that the SERT-P genotype may predispose to somatisation or other complex behavioural traits (including fibromyalgia, anxiety, and depression, as discussed by Yeo and colleagues¹) and that studies of interactions between candidate genes that modify motor, sensory, or behavioural functions would be of interest.

A third report from Korea¹⁸ showed no differences in genotype distributions between IBS, IBS subgroups, and healthy controls.

The fourth report from Italy was centred around the function of platelet SERT. SERT is found in peripheral sites, including platelets.¹⁹ As the SERT protein displays the same molecular properties at all known cellular locations⁷ and, in accordance with the general receptor theory,²⁰ Bellini et al proposed that it is conceivable that similar alterations in 5-HT uptake efficiency may also occur at the intestinal level. SERT was found to be expressed on platelet membranes of 12 D-IBS female patients at a low density (decreased Bmax) as well as to display a low degree of affinity (increased Kd) at its ligand binding site compared with 12 healthy female volunteers.²¹ They also suggest a possible correlation between the reduced capacity of serotonin reuptake and the severity of D-IBS symptoms.²¹ However, the association with genotype was not formally explored.

The fifth report is the paper in the current issue of Gut. Yeo and colleagues¹ observed a significant odds ratio of having the s/s genotype in D-IBS patients relative to controls in North America. This observation contrasts with the significant association with the l/s genotype in Turkey; however, from the function studies performed in vitro,¹⁰ both l/s and s/s genotypes would be predicted to produce SERT molecules with reduced function, and therefore result in higher synaptic levels of 5-HT. On the other hand, the odds ratio provides statistical evidence of an association although it does not prove a disorder in the function of the SERT or that it is causatively related to the development of D-IBS. In fact, the proportion of the l/l, l/s, and s/s genotypes in D-IBS were virtually equivalent in the 194 patients from North America.¹

In summary, these data show that the distribution of the l/s genotype seems very heterogeneous in different populations. A polymorphism which is associated with a disorder in one population may not be in another ethnic group. This is a problem that is increasingly being appreciated in behavioural genetics. This may explain, at least in part, a number of apparently contradictory findings gathered on specific genetic variations in different studies.

INTERPRETATION OF GENE-ASSOCIATION STUDIES

There are several potential pitfalls that need to be considered in the interpretation of these studies.

Ethnic differences

Difficulties in interpretation of population based association studies arise due to ethnic differences in SERT-P allele frequencies which may also explain some conflicting results. The frequency of the l/l genotype was 6% in Japanese subjects,^22,23 5% in Koreans,¹⁸ 34% in European-Americans,¹⁰ and 24% in Turks.¹⁶ This variation in background prevalence of a genotype clearly influences the statistical power to detect a genotype related difference.

Racial homogeneity of control and disease groups:

In the current paper by Yeo and colleagues,¹ there were no differences in the proportion with the l/l genotype in D-IBS patients and controls. In contrast, they observed differences in the distribution of s/s and l/s between controls (17.2% and 47.8%, respectively) and D-IBS (31.4% and 31.9%, respectively). Regrettably, the authors did not provide critically important data on the racial derivation of patients and controls participating in this study. They included a larger number of controls to attempt to suppress confounding effects such as population stratification and admixture. The proportions in predominantly European-Americans of the s/s and l/s genotypes of 19% and 49%, respectively,¹⁰ and 20% and 49%,¹⁷ suggests that the controls in the paper by Yeo and colleagues¹ were similar to two other independent control cohorts. This is reassuring as the 448 control DNA samples used by Yeo et al were obtained from three different commercial sources. In contrast, the study by Kim et al which did not show an association between SERT-P polymorphisms and D- IBS (n = 128) or other IBS phenotype, drew patients (n = 276) and controls (n = 120) from the same geographical region in the Mid-West of the USA and provided detailed information about the sex and race of participants. Thus female participants predominated in both lower functional gastrointestinal disorder patients (82%) and controls (79%), and European Americans were 89% of controls and 97% of patients; there were 6% Asian and 3% Hispanic among healthy participants. Control for ethnic differences is critically important in appraising disease associations between genotype and phenotype.

Interpretation of data from surrogate measurements

Although the structure and biochemistry of SERT molecules at different sites are similar within the same species, it is unclear whether the results obtained in binding studies of SERT in platelets really reflect the function of enteric SERT. For example, binding studies of SERT in brainstem nuclei may be normal²⁴ in diseases such as major depression whereas other studies suggest that platelet SERT function is abnormal in depression.²⁵

Genetic dissections of complex diseases are extremely complex

Investigations of gene-gene and gene-environment interactions in humans and animals are revealing genes that may underlie behavioural variations. However, studies also lead to the notion expressed by Reif and Lesch that a probabilistic rather than deterministic impact of the genes can be identified on the phenotypic expression of these behavioural or functional disorders.⁵ To understand the probabilistic impact or contribution of each genotypic variation in a semi quantitative manner requires studies with large sample sizes. Such studies are laborious and difficult, particularly in IBS where there is no diagnostic test.

CONCLUSION

A hypothesis proposed by Blakely²⁶ is not that IBS is equivalent to SERT dysfunction but that SERT dysfunction may present with a gastrointestinal phenotype or as a behavioural disorder. Anxiety disorders are commonly observed to be comorbid with IBS.^27,28 SERT dysfunction may contribute to behavioural and functional gut disorders; however, as previously noted by others, the influence of a single polymorphism on continuously distributed traits is likely to be small in humans. Much work is needed to assess the influence of genetic variation of SERT in the manifestations and response to therapy²⁹ of irritable bowel syndrome before we can be certain of the role of SERT genetics in D-IBS. Acquired changes in SERT mRNA in animal models of colitis³⁰ and in patients with ulcerative colitis and IBS³¹ suggest that there specific molecular alterations deserve further study.