OUP user menu

5-HT1A gene variants and psychiatric disorders: a review of current literature and selection of SNPs for future studies

Antonio Drago , Diana De Ronchi , Alessandro Serretti
DOI: http://dx.doi.org/10.1017/S1461145707008218 701-721 First published online: 1 August 2008

Abstract

5-HT1A receptors are key components of the serotonin system, acting both pre- and post- synaptically in different brain areas. There is a growing amount of evidence showing the importance of 5-HT1A in different psychiatric disorders, from mood to anxiety disorders, moving through suicidal behaviour and psychotic disorders. Findings in the literature are not consistent with any definite 5-HT1A influence in psychiatric disorders. 5-HT1A gene variants have been reported to play some role in mood disorders, anxiety disorders and psychotic disorders. Again, the literature findings are not unequivocal. Concerning response to treatment, the C(-1019)G variant seems to be of primary interest in antidepressant response: C allele carriers generally show a better response to treatment, especially in Caucasian samples. Together with the C(-1019)G (rs6295) variant, the Ile28Val (rs1799921), Arg219Leu (rs1800044) and Gly22Ser (rs1799920) variants have been investigated in possible associations with psychiatric disorders, also with no definitive results. This lack of consistency can be also due to an incomplete gene investigation. To make progress on this point, a list of validated single nucleotide polymorphisms (SNPs) covering the whole gene is proposed for further investigations.

Key words
  • Depression
  • mood disorders
  • psychiatric genetics
  • serotonin
  • serotonin receptor 1A

Introduction

Why this receptor is important: evidence

Serotonin (5-hydroxytryptamine or 5-HT) is an intrinsically fluorescent biogenic amine, which acts as a neurotransmitter by binding to a class of transmembrane receptors termed serotonin receptors (Chattopadhyay et al., 1996). These receptors represent one of the largest, evolutionarily ancient, and highly conserved families of G-protein-coupled receptors (GPCR) (Peroutka and Howell, 1994). Through specific receptors, 5-HT mediates a large variety of cognitive and behavioural functions including sleep, mood, pain, addiction, locomotion, sexual activity, depression, anxiety, alcohol abuse, aggression, and learning (Artigas et al., 1996; Ramboz et al., 1998); since all those functions are receptor-mediated, genetic variations of the receptor coding sequences are of primary interest for psychiatric research.

There is a number of known mammalian 5-HT receptor genes (5-HT1A/B/D/E/F, 5-HT2A/B/C, 5-HT3A/B and 5-HT3C/D/E, 5-HT4, 5-HT5A/B, 5-HT6, 5-HT7) some of which encode additional receptor variants (Albert and Lemonde, 2004); within this group, 5-HT1A is one of the most investigated; more than 1750 research papers up to January 2007 are dedicated to this protein.

5-HT1A is a G protein-associated receptor, preferentially coupled to Gi/o proteins to inhibit adenylyl cyclase. Many other second-messenger cascades have been associated with 5-HTR1A activation: MAPK pathways (which may be excitatory in post-synaptic cells or inhibitory in presynaptic neurons), c-Jun N-terminal kinase, GH/IGF-1 receptors and NMDA receptor channels (Adayev et al., 2003; Appert-Collin et al., 2005; Chen et al., 2002; Cowen et al., 1996, 2005; Millan et al., 2001; Stark et al., 2007; Sullivan et al., 2005; Turner et al., 2007; Yuen et al., 2005). Influence on inwardly rectified potassium channels (Andrade and Nicoll, 1987; Grunnet et al., 2004) and endocrine hormones secretion (Serres et al., 2000) have also been reported. Interestingly, membrane proved to play a relevant functional role: Evans and colleagues showed that the responsiveness of the 5-HT1A receptor can be enhanced or depressed as a consequence of activation of phospholipid-coupled receptor systems (Evans et al., 2001), and Pucadyil and colleagues showed that 5-HT1A activation is cholesterol-dependent since there is evidence of 5-HT1A agonists and antagonists inhibitory effects after membrane cholesterol oxidation (Pucadyil et al., 2005). This underlines the importance of the membrane environment where receptor activation takes place, and gives further details about the intracellular signal cascade, highlighting the high complexity of cellular mechanisms. On this topic see also Kalipatnapu and Chattopadhyay (2005), for a review about the membrane solubilization of 5-HT1A receptors. The aim of this study is to review the literature evidence about associations of 5-HT1A genetic variations and some of the most important psychiatric disorders, clinical presentation and treatment responses, attempting to highlight the prerogatives and opportunities of this line of research. This will follow a short review of receptor distribution in the brain and a 5-HT1A coding sequence description.

Where in the neurons?, where in the brain?

Brain 5-HT1A receptors are located both pre- and post-synaptically. Presynaptic 5-HT1A autoreceptors are located in the somatodendritic neuronal region, at the site of the serotonergic cell bodies in the dorsal and medial raphe nuclei, and they mediate a feedback inhibition of the 5-HT system, the so called ‘short feedback loop’, regulating the release of 5-HT by modulating neurotransmitter synthesis, terminal release and cell discharge rate (Sprouse and Aghajanian, 1988). In greater detail, the short negative feedback loop involves GABAergic cells which are stimulated by serotonergic neurons via HTR2A/C, and in turn stimulates the 5-HT1A receptors on the surface of the serotonergic neurons which first stimulated them, closing the auto-inhibition cycle. That is the reason why blockade of the 5-HT autoreceptor 1A (and also 1B) can enhance the increase of extracellular 5-HT levels achieved after selective serotonin reuptake inhibitor (SSRI) treatment (see Hjorth et al., 2000 for a review). Moreover, the presynaptic 5-HT1A down-regulation is believed to be associated with therapeutic effects of serotonergic antidepressant treatments and related to its time delay before onset (de Montigny et al., 1984). Further, it is suggested that multiple 5-HT1A subtypes govern different aspects of 5-HT function in the dorsal raphe nucleus (Stamford et al., 2000), and the 5-HT system flexibility is also enriched by action of opiate receptors and neurokinins (substance P and neurokinin B) stimulating 5-HT neurons via glutamatergic influences (see Guiard et al., 2005 and Liu et al., 2002 for interesting papers on this topic).

Mutual influences between deep serotonergic nuclei and upper brain systems is the biological basis for the ‘long feedback loop’ which is mainly mediated by post-synaptic 5-HT1A receptors and involves cortico-limbic areas probably influencing the activity of GABAergic and glutamatergic neurons (Davis et al., 2002). In fact, post-synaptic 5-HT1A receptors are found at high density in limbic regions and in the frontal, medial prefrontal and enthorhinal cortices (Grunschlag et al., 1997; Pazos et al., 1985; Schmitz et al., 1998), and the hippocampal regions and medial prefrontal cortex (mPFC) seem to be of primary interest for 5-HT1A investigation. In fact, the long feedback loop has been proposed to be of primary interest in the antidepressant effect of drug treatments through the down-regulation of post-synaptic 5-HT1A receptors (Artigas et al., 1996). Focusing on the limbic system, there is evidence that CA1 and CA3 hippocampal pyramidal cells are mainly inhibited by 5-HT1A activation (Andrade and Nicoll, 1987; Okuhara and Beck, 1994; Segal et al., 1989; Zgombick et al., 1989). Indeed, a neuronal activation can also be achieved after 5-HT1A stimulation, via 5-HT1A-mediated inhibition of putative inhibitory interneurons (Schmitz et al., 1995; Segal, 1990). This is quite interesting in view of the recent great interest in hippocampal functions associated with psychiatric disorders, especially for the possible influence of 5-HT tone on hippocampal neurogenesis associated with depressive episodes or disorders. This is consistent with the animal study by Miquel and colleagues (1994), where 5-HT1A was found to be highly expressed in the cortex and hippocampus of adult rats (for a review see also Elder et al., 2006). Indeed, there is evidence that long-term antidepressant treatment (SSRI, tricyclics, ECT), is associated with increased 5-HT1A-mediated inhibitory tone on CA3 hippocampal pyramidal cells (Haddjeri et al., 1998), this being more evident after short-term lithium augmentation (Haddjeri et al., 2000). There are also conflicting findings: Knobelman and colleagues suggested that 5-HT1A plays a larger role in regulating 5-HT release in the striatum and possibly other brain regions innervated by the dorsal raphe nucleus (e.g. cortex) than in the hippocampus, suggesting a major role for 5-HT1B in that region (Knobelman et al., 2001). Accordingly, Parsons and colleagues described a more significant disinhibition of 5-HT release in the frontal cortex than in the ventral hippocampus in 5-HT1A knockout rats after exposure to SSRIs (Parsons et al., 2001). Conflicting with this, a recent interesting report showed 5-HT1A reduced density in the ventral hippocampus, which is primarily implicated in emotional processing in the offspring of stressed female rats (Van den Hove et al., 2006).

An important role in brain functioning, reasonably related to psychiatric disorders, is played by the mPFC: pyramidal neurons in the PFC integrate cortical, thalamic, hypothalamic, limbic inputs and project to many widespread neuronal structures. In turn, the mPFC receives a large amount of afferences from the same neuronal net. On this basis, the PFC regulates a large number of cognitive and associative functions and is involved in the planning and execution of complex tasks, which are commonly impaired in some of the most important psychiatric disorders such as schizophrenia or severe depression. There is evidence that 5-HT1A is highly expressed in the mPFC (Read et al., 1994) and 5-HT1A is localized at an axon hillock in the pyramidal neurons providing key negative regulation of these neurons. Particularly in this area, highly complex interactions with 5-HT2A and 5-HT1A occur at the level of individual neurons (Amargos-Bosch et al., 2004), thus leading to a difficult interpretation of findings. For example, it has recently been reported that higher 5-HT1A density in the PFC in rats is associated with increased aggressive behaviour (Caramaschi et al., 2007), probably through a diminished 5-HT tone. This conflicts with a previous imaging study in humans which found that a lower density of 5-HT1A was associated with a history of aggressive behaviours in psychiatric patients and healthy controls (Parsey et al., 2002). More research efforts in this direction are needed.

Finally, 5-HT1A is expressed in other brain structures such as the cerebellar Purkinje cells (Darrow et al., 1990), the midbrain periaqueductal grey (PAG, a region known to be involved in pain modulation and fear responses), the amygdala (a brain structure highly associated with anxiety) (Behbehani et al., 1993; Davis et al., 2002), sensory neurons of dorsal root ganglia (Scroggs and Anderson, 1990), nucleus prepositus hypoglossi (Bobker and Williams, 1990, 1995), ventromedial hypothalamus (Newberry, 1992), lateral septum (Joels et al., 1987; Van den Hooff and Galvan, 1992) and laterodorsal tegmental nucleus (Thakkar et al., 1998) where 5-HT1A interaction was found to be associated to modified REM sleep patterns.

With regard to brain layers, 5-HT1A can be found on both pyramidal and principal cells, throughout the neocortex (Azmitia et al., 1996; Miquel et al., 1994), and it may have a role in the control of motor neuronal excitability. Moreover, diminished serotonergic activity might increase the release of neurotoxic levels of glutamate, or play a direct neurotrophic role (Turner et al., 2005). Finally, 5-HT1A can also be found in calbindin- and parvalbumin-containing neurons, which generally define two different subtypes of interneurons (Aznar et al., 2003).

Those data taken together suggest a complex and highly integrated 5-HT1A action in the brain: it is conceivable that this receptor influences key regulation activities in different parts of the brain reasonably associated with psychiatric disorders. Even if the physiological fine mechanisms of these actions are still far from being understood, there is enough genetic association evidence to encourage an effort towards an evidence-based individual genetic approach with diagnostic and therapeutic benefits.

The gene and its variants

The 5-HT1A receptor is encoded by an intronless gene located on human chromosome 5q11.2–q13 and it spans about 1200 bp (http://www.ncbi.nlm.nih.gov/). It was first cloned by Fargin and Albert (Albert et al., 1990; Fargin et al., 1989), and it encodes a single receptor isoform, a protein 422 aa long. Fifty HTR1A single nucleotide polymorphisms (SNPs) are known so far, and 23 have been validated (http://snpper.chip.org/bio/snpper-enter).

The most important linkage study results were found for association between the long arm on chromosome 5 and schizophrenia (Bassett, 1992; Crowe and Vieland, 1998, 1999; Gorwood et al., 1991; Owen et al., 1990), even though there are many conflicting reports of linkage analysis (Chavarria-Siles et al., 2007; Choudhury et al., 2007; Christoforou et al., 2007; DeLisi et al., 2002; Jang et al., 2007; Le Hellard et al., 2007; Liu et al., 2006; Ni et al., 2007; Schosser et al., 2007), but many other reports can be found in the literature.

Regarding bipolar disorder, there is some evidence coming from linkage association studies for an involvement of chromosome 5 (Kerner et al., 2007), even though it should be considered that the dopamine transporter gene is within 5p (Homer et al., 1997), and in the distal 5q there are the genes for the glucocorticoid receptor (GRL) and the β2 adrenergic receptor (ADRB2), also putatively involved in bipolar disorder (Mirow et al., 1994). Conflicting with this, other linkage association studies did not mention chromosome 5 (Barden et al., 2006; Faraone et al., 2006; Schumacher et al., 2005; Underwood et al., 2006). There is also evidence of linkage study exclusion of association between bipolar disorder and the whole chromosome 5 (Detera-Wadleigh et al., 1992), although using a less dense marker coverage. No article reporting linkage association analysis and supporting involvement of chromosome 5 with unipolar depressive disorders, or with suicidal behaviour, was found. Generally, even though SNP association analysis give some interesting results, linkage analysis studies investigating chromosome 5 in association with psychiatric disorders, showed almost negative results.

The most investigated HTR1A variants are: C(-1019)G (rs6295); Ile28Val (rs1799921); Arg219Leu (rs1800044) and Gly22Ser (rs1799920). C(-1019)G can be found in the promoter zone of the gene and this variant was found to be quite prevalent in the normal population (Wu and Comings, 1999). Ile28Val is characterized by a base-pair substitution (A→G) at the first position of codon 28 in the putative amino-terminal domain of the receptor (Bruss et al., 1995); the Arg219Leu variant can be found in the third intracellular loop of the 5-HT1A receptor (Bruss et al., 2005) while the Gly22Ser variant is located in the putative amino-terminal domain of the receptor (Rotondo et al., 1997). C(-1019)G (rs6295) has a balanced allele frequency in Caucasians, this makes investigations useful since there is a good chance of finding patients with all the allelic re6295 variants. On the other hand, Arg219Leu (rs1800044) and Gly22Ser (rs1799920) have no frequency data in the general population, and the Ile28Val (rs1799921) G allele has a small prevalence in Caucasians (0.8%). No prevalence data is available for the other populations in international databases (for all prevalence distribution, see Table 1). A small probability of finding a carrier of a mutant allele impairs the clinical impact of association studies based on cited variations. This issue will be further discussed.

View this table:

Expression and functions

Table 2 gives a list of expression studies in association with 5-HT1A genetic variants. The main results and significance are reported.

View this table:

The 5-HT1A G(-1019) allele in the 5-HT1A upstream regulatory region fails to bind identified repressors Deaf-1 and Hes5, abolishing Deaf-1 action and impairing Hes5 action. This leads to the up-regulation of receptor expression (Albert and Lemonde, 2004; Lemonde et al., 2003). In fact, an increase in 5-HT1A receptor expression in individuals with the G/G genotype has been observed (Parsey et al., 2006b). Moreover, Deaf-1 has been shown to have a different function in presynaptic and post-synaptic cells or non-serotonergic neurons: it does not repress but actually enhances the 5-HT1A expression in post-synaptic cells (Czesak et al., 2006). This might help explain some of the conflicting findings reported in the literature: observations of cortical region-specific decreases in the 5-HT1A receptor in PET studies of psychiatric patients could be mediated by the disruption of Deaf-1 signalling leading to down-regulation of 5-HT1A expression in post-synaptic neurons. Another interesting evidence and proposed explanation, is highlighted by a PET study by David et al. (2005) : following the evidence of a decreased expression in 5-HT1A receptors in 5-HT transporter knock-out rats, especially in females (Li et al., 2000), David and colleagues studied the mutual relationship between two different gene variations: the 5-HT transporter and 5-HT1A. These authors reported that a minor availability of 5-HT1A was detected only in subjects (healthy volunteers) with 5-HTTLPR short (SS or SL) genotypes with no influence of 5-HT1A variants. A possible explanation proposed was that the lower transcriptional efficiency associated with the S allele of 5-HTTLPR may lead to decreased 5-HTT function, which in turn could lead to a lifelong increase in 5-HT tone, which may then desensitize and down-regulate 5-HT1A receptors. This is consistent with animal experiments: SERT knockout rats show a decreased 5-HT1A density (Li et al., 2004). Moreover, the absence of increased 5-HT1A density in normal subjects with the G/G genotype might also suggest that healthy subjects can better compensate for their unfavourable genetic make up than depressed patients.

Interestingly, in the same year Arias and colleagues found that a combined genetic effect of HTR1A and 5-HTTLPR variants, and not the 5-HTTLPR variant alone, could influence the antidepressant response in a sample of 130 depressed patients, being the S/S G/G the at risk haplotype for failed remission achievement (Arias et al., 2005). It is really difficult to discuss the partially different findings in these papers: the first one, by David et al., is based on healthy subjects and the second one, by Arias et al., investigates the antidepressant (citalopram) response in a sample of depressed subjects. Moreover, sample sizes are different: 35 in the study by David and colleagues, 130 in the study by Arias et al. Stratification factors such as different sociodemographic variables, comorbidity, pre-treatment issues such as response, and type and duration of previous therapies in the depressed sample make it difficult to draw definitive conclusions.

Finally, a third hypothesis was proposed in 2002 by Gross and colleagues (Gross et al., 2002) and in 2004 by Lesch and colleagues (Lesch and Gutknecht, 2004): the influence of the 5-HT1A coding sequence variations should not be expected to reveal itself through expression on adult neuronal surfaces, but by the complex neuronal mutual net influences involved in brain formation. Thus, the genetically determined 5-HT1A enrichment would influence the neuronal net formation eventually predisposing to psychiatric disorders.

Evidence shows that 5-HT1A knockout rats show an anxiety-like behaviour (Groenink et al., 2003; Heisler et al., 1998; Olivier et al., 2001; Parks et al., 1998; Parsons et al., 2001; Ramboz et al., 1998; Toth, 2003): future animal research should be able to selectively shut down pre- or post-synaptic 5-HT1A receptors, and this would provide unique evidence for understanding the detection of the predictive utility of genetic variations.

The Arg219Leu variant was found to be expressed in virtually identical densities in wild and variant transfected HEK293 cells (Bruss et al., 2005), suggesting that this variant has no influence over the stability or efficiency of transductional and translational systems of the 5-HT1A coding sequence, even though it was found that the ability of G-coupling activity after receptor activation, and the inhibition of forskolin-stimulated accumulation, was decreased by 60–90% in variant type. In view of that, this variant apparently does not change the binding properties of 5-HT1A, but it is associated with a drastic impairment of signal transduction: it seems reasonable to expect that patients carrying the variant will show different drug treatment responses.

A reduced receptor activity together with a lower density on cell surfaces was also reported for the Gly22Ser variant but only after 24 h treatment with a 5-HT1A agonist, whereas the binding affinity and pharmacological profile were found to be similar in wild an variant genotype-expressing cells (Gothert et al., 1998; Rotondo et al., 1997). In the same study the Ile28Val variant was found not to influence affinity, pharmacologically promoted or agonist-promoted down-regulation issues. Most part of those investigations are over-expression studies in non-neuronal cells: further validation in neuronal models or in vivo would be advisable. Given that, a genetic approach to patients will offer the opportunity to highlight a different and predictable, i.e. genetically determined, receptor drug treatment determined down-regulation, which is the supposed therapeutic mechanism of antidepressant drugs. Finally, the Ile28Val variant was found to be equally expressed in psychiatric and healthy subjects (Erdmann et al., 1995), and to share the same pharmacological profile with the wild variant (Bruss et al., 1995).

Depression

Table 3 gives a list of studies dealing with major depressive disorder (MDD) and antidepressant response in association with 5-HT1A genetic variants. The main results and significance are reported.

View this table:

The 5-HT system is implicated in major depression and suicide and is negatively regulated by somatodendritic 5-HT1A autoreceptors (Albert and Lemonde, 2004); moreover, desensitization of 5-HT1A autoreceptors is believed to be related to the therapeutic effects of antidepressant drugs and it is probably implicated in the 2- to 3-wk latency for antidepressant treatments (Blier et al., 1998). A significant decrease in 5-HT1A mRNA was found in the dorsolateral PFC and hippocampus in chronically stressed animal models and in subjects with MDD (Lopez et al., 1998; Lopez-Figueroa et al., 2004).

This evidence makes the 5-HT1A genetic variants good candidates in the pathophysiology and treatment of depression. The G allele of the C(-1019)G polymorphism has been reported to be associated with enriched expression of the gene and there is some evidence suggesting that G allele carriers have higher risks for a depressive episode and of being more resistant to antidepressant therapy (Arias et al., 2005; Hong et al., 2006; Lemonde et al., 2003, 2004; Parsey et al., 2006b; Yu et al., 2006), even though conflicting evidence can also be found (Mandelli et al., 2006; Peters et al., 2004; Serretti et al., 2004). On this topic, our group reported some results in 2004: the sample under investigation was composed of 151 major depressed and 111 bipolar patients, treated with fluvoxamine for a period of 6 wk. In this sample, the depressed patients responded to treatment independently from 5-HT1A variants, whereas, in the bipolar sample, HTR1A C/C genotype carriers showed a better response to fluvoxamine.

In a interesting recent study, Kraus and colleagues reported that interferon therapy associated with depressive episode was found to be related to the G-allele homozygosity of the C(-1019)G variant (Kraus et al., 2007), whereas 5-HT promoter (SLC6A4 or 5-HTTLPR) or tryptophan hydroxylase 2 (TPH2) gene variations were not associated with such an important iatrogenic effect.

Finally, in 40 remitted mood disorders patients, C/C genotype carriers had lower scores at TCI (temperament and character inventory) subscales of total transcendence, transpersonal identification and spiritual acceptance (items, or item clusters) (Lorenzi et al., 2005), this finding agrees with a converging imaging study (Borg et al., 2003).

There are other gene variations that have been investigated as relevant for the antidepressant response: in 2004 Suzuki and colleagues (Suzuki et al., 2004) found that the Gly272Asp variant could influence fluvoxamine response in a sample of 65 depressed patients being Asp allele-associated with higher Hamilton Depression Rating Scale (HAMD) score reductions after 2, 6 and 12 wk of treatment. Conflicting with this, lack of association between fluvoxamine response and the Gly272Asp variant was found in a sample of 222 Chinese major depressive patients (Yu et al., 2006), even though the C(-1019)G variant, which was reported as in strong linkage disequilibrium (LD) with Gly272Asp, showed a significant association in the female group; C carriers having better clinic results after fluvoxamine treatment. This observation raises the issue of LD calculation, in fact in the paper the authors define strong LD based on a high D', while the r2 low value (less sensitive to low frequencies and small sample sizes) indicated a marginal LD. Therefore it is not surprising to observe discrepant findings between variants.

As far as Caucasian samples are considered, the positive association studies by Parsey and colleagues are based on small sample size (cases n=22 and n=28) (Parsey et al., 2006a,b): this is an important limitation. Moreover, the significant result by Parsey et al. (2006a) was found in genotype distribution within remitters and the control group (p=0.005), while no significant difference was found between remitters and non-remitters (p=0.073). Analysis on allele distribution reported non-significant results (p=0.058). Moreover, the positive study by Arias and colleagues (Arias et al., 2005) considered a sufficient sample size (cases n=130) but revealed significant association between 5-HT1A genetic variations and antidepressant response only after considering the genetic variation of the HTR1A together with the functional polymorphism in the promoter of the serotonin transporter (HTTLPR), prompting the hypothesis that the statistical significance is mostly based on the latter. This is consistent with the negative results our group reported about Caucasian samples (Mandelli et al., 2006; Serretti et al., 2004) together with other negative findings obtained by other groups (Peters et al., 2004). The positive findings reported in Asian samples (Hong et al., 2006; Suzuki et al., 2004; Yu et al., 2006) could be explained by the different prevalence of the investigated polymorphisms between different ethnicities. Another possible reason to explain conflicting results is that complex phenotypes can not be explained by single gene mutations: considering haplotypes involving different genes within the 5-HT system and beyond it, could be a good strategy. The previously mentioned paper by Arias et al. (2005) stressed this line of research. Finally, a more complete investigation of the all the single gene variants is suggested.

Suicide

Table 4 gives a list of studies dealing with suicidal behaviour in association with 5-HT1A genetic variants. The main results and significance are reported.

View this table:

The C(-1019)G variant has been associated with suicidal behaviour, even though there is also conflicting evidence and generally this variant together with the structural Gly272Asp and Pro16Leu variants do not seem to be strongly associated with suicide behaviour or attempt. In a recent study we performed (Serretti et al., 2006), no significant association was found between 5-HT1A variants and suicide. Wassermann and colleagues (2006) had similar results in a sample of 272 suicide attempter families: the HTR1A promoter variant seemed to be not associated with suicide history except for a trend of G allele over transmission in a subgroup of suicide attempters characterized by high levels of traumatic or stressful life events prior to the suicide attempt. Interestingly, stressful life events were not found to be associated with HTR1A variants and mood disorders in a large Italian sample (n=686) recently investigated by our group (Mandelli et al., 2006). Those lines of evidence taken together might suggest that suicide could be associated with the HTR1A promoter polymorphism and stressful life events independently from mood disorders. A stronger boundary with the impulsiveness cluster, might be a possible hypothesis.

Other negative association results can be found in some studies considering the promoter or the structural polymorphisms (Huang et al., 2004; Nishiguchi et al., 2002; Ohtani et al., 2004; Videtic et al., 2006). Positive association findings are also available (Lemonde et al., 2003). Again, a possible explanation of non unequivocal evidence could be addressed to incomplete gene polymorphism analysis, to the unknown and probably multifactorial pathogenesis of psychiatric disorders, and to the small impact of a single gene in such complex phenotypes (Kendler, 2005). According to the last point, an interesting study performed by Hsiung and colleagues (2003) reported that at least two independent second-messenger cascades after 5-HT1A activation in the brain of suicide victims were differently active when compared with controls, this suggests that future studies should consider complex haplotypes, choosing them on the basis of evidence and rational study of internal cell mechanisms.

Anxiety disorders

Table 5 gives a list of studies dealing with anxiety disorders in association with 5-HT1A genetic variants. The main results and significance are reported. Several lines of evidence show a significant relationship between 5-HT1A and anxiety symptoms.

View this table:

Animal experiments show anxiety-like behaviour in 5-HT1A knockout rats (Groenink et al., 2003; Heisler et al., 1998; Olivier et al., 2001; Parks et al., 1998; Parsons et al., 2001; Ramboz et al., 1998; Toth, 2003), probably because of a perturbation in the prefrontal GABA/glutamate system (Bruening et al., 2006). Moreover, severe acute stress as well as stress of 2 wk duration has been found to be associated with down-regulation of 5-HT1A mRNA synthesis in the rat hippocampus (Lopez et al., 1998, 1999). Neumeister and colleagues consistently reported lower 5-HT1A density in the anterior cingulate, posterior cingulate and raphe of patients suffering from panic disorders with or without depressive comorbidity (Neumeister et al., 2004). Anti-anxiety properties of 5-HT1A active drugs are well known in everyday clinical practice. Consistent with this, panic disorder was recently found to be associated with HTR1A and COMT variants (Freitag et al., 2006), and when HTR1A is considered alone, a significant relationship is still found between the promoter polymorphism and panic (Huang et al., 2004; Rothe et al., 2004). Those findings seem to be more significant when panic with agoraphobia is considered, prompting the idea that a more in-depth dissection of clinical phenotypic subtypes would be a good strategy for future investigations. Post-traumatic stress disorder (PTSD) does not seem to be associated with 5-HT1A density variants (Bonne et al., 2005). There is a lack of evidence regarding other anxiety disorders.

Bipolar disorder

Imaging studies investigating the hypothesis of a different density or affinity of 5-HT1A receptors in hippocampal and cortex regions reported negative findings (Dean et al., 2003; Gray et al., 2006), even though an interesting dissociation between the number of receptors and the activity of G protein-associated second-messenger cascade was reported in the study performed by Gray and colleagues: this could suggest a kind of segregation of 5-HT1A receptors leading to a normal number of receptors associated with a reduced activity in bipolar disorder. In the same study a gender difference was reported, i.e. 5-HT1A receptors were expressed less in female brains (Gray et al., 2006). A mRNA in-situ hybridization study confirmed this report (Lopez-Figueroa et al., 2004).

Conflicting with this, some animal studies investigated the possible role of 5-HT1A receptors in the pharmacodynamics of mood stabilizers, reporting positive association results. Lithium was found to enhance the sensitivity of post-synaptic 5-HT1A after short-term administration (Blier et al., 1987). This finding was replicated by Haddjeri and colleagues (2000): lithium augmentation of long-term antidepressant treatment in rats was found to be associated with an enhanced disinhibition of dorsal hippocampal CA3 pyramidal neurons, an action which is known to be mediated by post-synaptic 5-HT1A receptors. More recently, McQuade and colleagues reported an association with chronic lithium treatment and reduced 5-HT1A mRNA in the hippocampus, whereas no different levels of 5-HT1A mRNA could be found in the dorsal raphe nucleus where 5-HT1A receptors are present as autoreceptors (McQuade et al., 2004). Post-synaptic 5-HT1A receptors were found to be influenced also by lamotrigine, another mood stabilizer (Bourin et al., 2005), whereas negative findings were reported for valproate (Delva et al., 2002; Khaitan et al., 1994). As mentioned previously, in our 2004 study, we reported that HTR1A C/C bipolar genotype carriers showed a better response to fluvoxamine in regard to depressive episodes (Serretti et al., 2004). Those lines of evidence could suggest a possible pharmacological impact of 5-HT1A gene variants, but more studies and a larger gene investigation are needed in this field of research.

Psychosis

Table 6 gives a list of studies dealing with psychotic disorders in association with 5-HT1A genetic variants. The main results and significance are reported.

View this table:

5-HT1A seems to be highly involved in the pathological neurotransmitter disequilibrium associated with schizophrenia. Several post-mortem brain studies reported HTR1A altered expression in schizophrenia patients (see Frankle et al., 2006 for a review), and, consistent with this, two PET studies (Kasper et al., 2002; Tauscher et al., 2002) reported increased 5-HT1A density in prefrontal and temporal cortices, whereas a diminished density in amygdala or no different density compared to controls were found in other PET studies (Frankle et al., 2006; Yasuno et al., 2004). This diversity could be due to the different resolution of post- mortem and PET studies: only the former can investigate the 5-HT1A density in different layers of brain tissues.

Besides conflicting imaging studies, there is evidence that 5-HT1A is involved in antipsychotic response. Many atypical antipsychotics act on 5-HT1A (clozapine, serizotan, bifeprunox, nomonapride, ziprasidone, aripiprazole, tiospirone) and even though other atypicals such as olanzapine or risperidone are not directly active on 5-HT1A, it has been suggested that the improvement of depressive symptomatology, which is characteristic of atypical antipsychotics, could be associated with a dopamine-enhanced release in the mPFC, this being dependent also on 5-HT1A activity (Ichikawa et al., 2001; Newman-Tancredi et al., 2005). The role of 5-HT1A receptors in antipsychotic treatment, especially as far as negative symptomatology is concerned, is suggested by Richtand and colleagues who recently reported that the strongest correlation with clinically effective drugs and receptor activity profile is the HTR2C/D2 ratio for the typical antipsychotics, and HTR2A/5-HT1A and HTR2C/5-HT1A for the atypical antipsychotics: the higher the 5-HT1A expression or function, the less effective the antipsychotic treatment (Richtand et al., 2007). Moreover, Cosi and colleagues reported that atypical antipsychotics that are active on 5-HT1A, but not haloperidol which is devoid of 5-HT1A, were associated with a reduced kainite-induced brain damage in rats (Cosi et al., 2005). This is quite interesting because of the possible neurodegenerative aetiopathogenesis of schizophrenia, and the suggestion of a possible use of 5-HT1A agonists as protectors against excitotoxic injuries.

In regard of the benefit over negative symptomatology, Sato and colleagues recently reported that risperidone can increase acetylcholine release (which is assumed to be associated with cognitive performance), in the PFC of rats through a complex mechanism enhanced by 5-HT1A activation: if its activation is missing, the acetylcholine levels in prefrontal brain structures are impaired after risperidone treatment, whereas the simple 5-HT1A stimulation does not alter acetylcholine levels in the brain (Sato et al., 2007). The same result was obtained by Chung and colleagues in 2004, using clozapine (Chung et al., 2004). This is consistent with other papers which reported the association of 5-HT1A with cognitive functions (Sumiyoshi et al., 2000; Sumiyoshi and Meltzer, 2004; Tamminga, 2006). The benefit on antipsychotic-induced motor side-effects could be also partially associated with 5-HT1A function when atypical antipsychotics are used: it has been proposed that the serotonergic modulation over dopaminergic system is principally inhibitory, thus, the stimulation of presynaptic 5-HT1A should lead to an inhibitory effect over HTR2C which in dopaminergic neurons will lead to a final effect of diminished inhibition and a benefit for movement dyscontrol (Haleem, 2006).

Other evidence of 5-HT1A modified function in schizophrenia comes from a study by Lee and Meltzer (2001) reporting that 5-HT1A functionality is impaired in schizophrenic females: the plasma cortisol concentration increased after ipsapirone administration was found to be blunted in schizophrenic female patients. Other parameters such as hypotermina, nausea, dizziness, irritability, and feeling less well did not differ between schizophrenic and healthy subjects. It must be underlined that the blunted plasma cortisol concentration stands for a hypofunction of 5-HT1A post-synaptic receptors (Lee and Meltzer, 2001).

Even though all these lines of evidence should encourage investigations on HTR1A variants, there are not many studies investigating 5-HT1A genetic variants and the study of clinical characteristics of schizophrenia or schizophrenia-like disorders.

In 2004 Huang and colleagues (Huang et al., 2004) reported that the C(-1019)G variant was associated with schizophrenia together with substance use disorder and panic attacks, this result was replicated 2 yr later by Reynold and colleagues (2006) who reported that C/C carriers had better scores at PANSS (Positive and Negative Syndrome Scale), but not for positive symptoms, and at the Calgary Depression Scale than G/G carriers after 3 months of a standard treatment for acute psychosis. Interestingly, with the Calgary Depression Scale G/G carriers showed worse conditions at the end of treatment than at the beginning. This is consistent with the literature findings discussed above about HTR1A variant and depression.

In 2006 Lane and colleagues investigated the hypothesis that HTR1A variants, together with another 15 genetic variants across 10 candidate genes [5-HT1A, 5-HT2A, 5-HT2C, 5-HT6, D1, D2, D3, α1-adrenergic receptors, brain-derived neurotrophic factor (BDNF), and cytochrome P450 2D6], could influence the risperidone-associated weight gain side-effect. A sample of 122 Han Chinese schizophrenia in-patients was selected: the HTR1A Gly272Asp variant showed no evidence of association (Lane et al., 2006). Finally, our group investigated the Ile28Val polymorphism of the 5-HT1A coding sequence in a sample of 84 in-patients affected by mood disorders (72 bipolar and 12 major depressive disorder) verifying the hypothesis that 5-HT1A could be involved in psychotic symptoms of mood disorders. No association result was found (Serretti et al., 2000). The different ethnicity of the samples investigated [Caucasians (Erdmann et al., 1995; Huang et al., 2004; Reynolds et al., 2006; Serretti et al., 2000) and Asians (Kawanishi et al., 1998; Lane et al., 2006)], could explain the inconsistency of results. As far as the study on the bipolar sample is concerned (Serretti et al., 2000), the different psychiatric disorders at the basis of the investigation, is probably the most important cause of discrepancy.

The Ile28Val variant was also investigated in a large case (n=352) control (n=210) study with no evidence of association with schizophrenia. The sample also included other diagnoses such as bipolar disorder and Tourette's syndrome (Erdmann et al., 1995). Lack of association within HTR1A variants and schizophrenia was also reported by Kawanishi and colleagues (1998): in this work a list of variants was considered [Gly272Asp, Pro16Leu, A(294)G, T(549C), C(51T), G(-152)C, C(-321)G, (-480)delA, and A(-581)C]. So far, studies investigating the C(-1019)G variant tended to find positive association results in Caucasian samples. The other considered variants tended to give negative association results. Study designs involving all the known and validated genetic variants of 5-HT1A are missing.

Other findings

Anxiety, as measured by the NEO Personality Inventory, showed a negative association with 5-HT1A density in a PET study performed in a sample of 19 healthy volunteers (Tauscher et al., 2001). In another PET study, the 5-HT1A density was found to be negatively correlated with spiritual acceptance, a sub-item of the TPQ (Tridimensional Personality Questionnaire) in 15 healthy subjects (Borg et al., 2003), and in 2005 our group investigated this field from a genetic point of view, finding that C/C carriers at the C(-1019)G variant were more likely to have significantly lower scores at the total self-transcendence and at the subscales of transpersonal identification and spiritual acceptance (Lorenzi et al., 2005). Finally, an association study considering personality traits and the C(-1019)G HTR1A promoter variant in 284 healthy students, reported an association with the G allele and higher scores on the neuroticism scale on the NEO Personality Inventory and more harm avoidance on the TPQ (Strobel et al., 2003).

SNP suggestion

As previously seen, not all the validated SNPs of the HTR1A sequence have been investigated over the reviewed studies. This is an important methodological bias. A complete and commonly accepted investigation of the gene is advised for the following reasons:

  1. to facilitate the interpretation of results throughout different studies;

  2. to cover important functional areas such as 5′-UTR, 3′-UTR, promoter, enhancers, silencers;

  3. to permit an inductive research approach to the genetic sequence, identifying relevant mutations still to be hypothesized;

  4. to permit the identification of mutations interfering with important functional mechanisms such as: DNA access, reading and transduction, mRNA stability and translation, mRNA primary structure-based functions. In all those mechanisms, intronic sequences are as important as exonic ones.

Actual genome-wide techniques make it possible to investigate several polymorphisms (up to 600 000) in a single investigation scan, and the costs have started to decrease substantially, not least because of competition between the leading commercial platforms, Illumina and Affymetrix. This offers a unique opportunity to investigate haplotypes in a complete way. As far as 5-HT1A is concerned, its possible involvement in psychiatric disorders, following evidence of animal and imaging studies, should be investigated at least through all the validated variants. A list of validated SNPs covering all of the gene is presented in Table 1; data are collected from international databases (http://www.ncbi.nlm.nih.gov/). To select the SNPs two filters have been applied: complete coverage of the gene and validation status. The first point has been discussed previously. Validation status is advised because it would be clinically useful and statistically advisable, to concentrate research on genetic variants that have a sufficient chance to be detected in the general population. Table 1 gives the list of selected variants with allele prevalence information.

Conclusions

Research suggesting that association studies may have success in finding complex trait genes (De La Vega et al., 2005; Risch and Merikangas, 1996) and the discovery and validation of large numbers of SNPs in the human genome (HapMap, 2003), have elicited a growing interest in the performance of large-scale genetic association studies using SNPs at the candidate-gene, whole-chromosome, and eventually whole-genome level (De La Vega et al., 2005).

Genetic association studies require a large sample size and strict investigation methodologies (Risch and Merikangas, 1996) given that the reliably observed gene variants odd ratios range around 1.2–1.5 (Kendler, 2005; Lasky-Su et al., 2005). The most part of published work does not reach the theoretical methodology threshold. It has recently been demonstrated that going genome-wide makes the sample size a predominant determinant of the feasibility of the studies. A number of 1.000 subjects per arm is advised for common causal alleles with relatively large effect sizes [genotype relative risk (GRR) 1.7] (Nannya et al., 2007). Further, the definition of GRR depends also on the disease model. If f0, f1, f2 are the probabilities of being affected for individuals with 0, 1, or 2 copies of the risk allele, then GRR can be: multiplicative (GRR=f1/f0=f2/f1), additive (GRR=f1/f0), dominant (GRR=f1/0=f2/f0), recessive (GRR=f2/f0=f2/f1) (http://csg.sph.umich.edu/). Even though most SNPs are believed to have a co-dominant, additive effect, other models have been also proposed (Greenberg et al., 1999; Hranilovic et al., 2004; Lesch, 2001).

However, as a practical example, using an additive model and considering the prevalence of rs6295 (C(-1019)G), a sample size of 600 cases and 600 controls is needed to obtain a power of 80% (one-stage design) with a quite liberal p of 0.01 to detect and odds ratio of 1.5. For the rarest Ile28Val mutant allele, a total sample size of over 50 000 subjects is needed! (Cohen, 1988). This simulation is driven on the basis of the investigation of a single gene, which is the most common path in the reviewed papers. Consideration of haplotypes further complicates the interpretation of results given the risk of false positives and non-independent observation (Purcell et al., 2003; Purcell and Sham, 2001).

None of the studies reviewed in the present paper reached the theorically requested sample size for the investigated mutations. Moreover, many stratification factors such as ethnicity, sociodemographic and clinical issues, different treatment, different period of treatment, different study design, multicentre vs. monocentre investigations, different age of onset, previous treatments, comorbidity and other important variables generally influence the validity of results. A large part of this methodological issue has been recently summarized and critically reviewed by our group regarding pharmacogenetics (Serretti et al., in press) and may be responsible for inconclusive findings even in very large sample sizes (Lewis et al., 2003; Segurado et al., 2003; Wellcome Trust, 2007). However, beyond clinical and endophenotipic dissections, one important and easy-to-fix bias is the complete gene variant coverage. In fact, 5-HT1A gene variants have been reported to play some role in association with psychiatric disorders such as mood disorders, anxiety disorders and psychotic disorders, but the literature findings are not unequivocal. The C(-1019)G variant seems to be of primary interest in antidepressant response, and preliminary studies seem to indicate a role for 5-HT1A gene variants in the psychotic spectrum but the evidence remains weak and difficult to decipher. We suggest that genetic association research would be reasonably more efficient if all the validated SNP variants were included, this point is easy to accomplish and technically feasible.

Acknowledgements

None.

Statement of Interest

None.

References

View Abstract