Introduction

Serotonin (5-hydroxytryptamine, 5-HT) and its receptors are widely distributed in the central nervous system (CNS)1 and influence several behaviours such as food intake, sexual behaviour, activity rhythms, and emotional states,2 as well as higher cognitive functions such as learning and memory processes.3, 4 In particular, 5-HT2A receptors are located mainly in the prefrontal cortex and hippocampus,5, 6 which are key brain structures for memory.7

To date, associations of genetic polymorphisms with brain morphology have been widely reported.8, 9, 10, 11, 12 Recently, a functional single nucleotide polymorphism (SNP) in the 5-HT2A receptor gene leading to an amino-acid substitution at residue 452 (His452Tyr) has been found. The common variant is the C allele (His452His) and the rare is the 452Tyr allele (His452Tyr), with a frequency of 9.3% in Caucasian populations.13 Recent findings have demonstrated a functional influence of this SNP on human memory, 452Tyr carriers showing 21% poorer memory performance than noncarriers.14 In this study, healthy young volunteers viewed six sets of semantically unrelated words and performed an incidental episodic memory task with immediate and delayed recall after 5 min and 24 h. In both conditions, 452Tyr carriers showed lower memory performance compared to noncarriers.

Moreover, it has been shown that this effect is modified by age in that it can be detected in young to middle-aged, but not in older persons.15

Aim of this study was to investigate whether the 452Tyr allele is associated with morphological changes in the normal brain, particularly in the temporal structures typically involved in memory processes, such as the hippocampus and parahippocampal grey and white matter (GM and WM). We addressed this question by studying high-resolution structural magnetic resonance (MR) images with region-of-interest (ROI) and voxel-based morphometry (VBM) volumetric analyses in carriers and noncarriers of the allele. VBM is a recently developed technique allowing to study the effect of different factors on brain structure with exquisite detail.16 Whole brain maps can be drawn that indicate where atrophic changes occur with a precision of a few millimetres.

Materials and methods

Subjects

A total of 76 healthy subjects (28 males and 48 females) aged 40 years and older, living in Northern Italy, were enrolled in a study on normal brain structure with MR as described in detail elsewhere.17 Briefly, subjects were picked among those undergoing brain MR scan at the Neuroradiology Unit of the ‘Città di Brescia’ Hospital from March 2001 to March 2003 for reasons other than memory disturbance, cognitive impairment, degenerative diseases, or head trauma and whose MR scan was negative. The reasons for MR prescription were generally migraine and headache, auditory (hypoacusia, dizziness, tinnitus) or visual concerns (diplopia), sensory disturbances (paresthesias), suspected cerebrovascular disease, and miscellaneous (dyslexia, orbit study, lipotimic episodes, etc). Exclusion criteria were based on information before MR findings. Clinical exclusion criteria were: (1) MR scan for memory problems or cognitive impairment, (2) MR scan for clinical suspicion of neuro-degenerative diseases (Parkinson's disease, progressive supranuclear palsy, Huntington's disease, multiple system atrophy, etc), (3) patient undergoing MR for suspected stroke, (4) history of TIA or stroke, head trauma, alcohol and substance abuse, cortico-steroid therapy, and loss of weight greater than 5 kg in the last 6 months, and (5) cognitive impairment on neuropsychological testing. Radiological exclusion criteria included: (1) brain mass, (2) WM hyperintensities in a subject undergoing MR for suspected multiple sclerosis, (3) aneurysm larger than 10 mm, (4) arteriovenous malformation (except for small developmental venous anomalies), and (5) malformations of the CNS.

Subjects were intercepted in the waiting room of the neuroradiology unit, were explained aim and methods, and were asked to take part in the study. All subjects underwent multidimensional assessment including clinical, neurological and neuropsychological evaluations, and drawing of a blood sample. The study was approved by the local ethics committee.

5-HT2A His452Tyr polymorphism

Genomic DNA was isolated from peripheral white blood cell samples by Puregene kit (Gentra System), according to the manufacturer's protocol.

The 5-HT2A His452Tyr polymorphism changes a cleavage site for a specific restriction endonuclease and, after PCR amplification, was analysed through restriction enzyme BseXI (Fermentas) as described by Erdmann et al.18

MR imaging

Three-dimensional (3D) high-resolution T1-weighted MR images were acquired on a 1.0 Tesla Philips Gyroscan (TR=20 ms, TE=5.0 ms, flip angle=30°, field of view=220 mm, acquisition matrix 256 × 256, slice thickness 1.3 mm).

MR images were processed with SPM99 (http://www.fil.ion.ucl.ac.uk/spm) following an optimized protocol including: (a) generation of a whole brain customized template, (b) generation of customized prior probability maps, and (c) main VBM steps based on the previous customized images.

Customized template

This was obtained by normalizing MR images to the Montreal Neurological Institute (MNI) template19 of SPM99 using a 12-parameter affine transformation, smoothing the normalized images with an 8 mm isotropic Gaussian kernel and averaging the smoothed images. The anterior commissure (AC) was manually set as the origin of the spatial coordinates for the normalization algorithm:20 AC coordinates were detected through the ‘Display’ window by an investigator blinded to gene status, and were used to modify the origin using the ‘HDR edit’ button. The normalization uses a bilinear interpolation algorithm to reslice images to voxel size of 2 × 2 × 2 mm3. This voxel size will be used in the following processing and analysis. All steps were visually checked with the ‘Check reg’ routine.

Customized prior probability maps

These were computed by segmenting the normalized images into GM, WM, and cerebrospinal fluid (CSF), then smoothing with an 8 mm Gaussian filter, and finally averaging the segmented images, thus obtaining the customized prior probability maps specific for GM, WM, and CSF.21

Main VBM steps

Original MR images were normalized to the whole brain customized template through affine and nonlinear transformations with 7 × 8 × 7 basis functions, medium regularization, and reslicing 2 × 2 × 2 mm3. The normalized images were segmented into GM, WM, and CSF using the customized prior probability maps. The ‘X-brain’ routine, based on erosions and dilatations, was used to remove voxels of non-brain tissue from the segmented images, thus obtaining a brain image for each subject. This was used to clean GM, WM, and CSF images. In the modulation step, voxel values of the cleaned GM images were multiplied by the measure of relative volumes of warped and unwarped structures derived from the nonlinear step of spatial normalization (Jacobian determinant).22 The modulated images were smoothed with an 12 mm isotropic Gaussian kernel.

ROI analysis

In order to compute global and lobar volumes, a customized program (http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind02&L=spm&P=R176348&I=-1) was applied to GM, WM, and CSF modulated images. Modulated images are 3D matrices where the intensity of each voxel is proportional to GM, WM, and CSF volume within each voxel. The program calculates volumes by summing the voxels of the modulated images times voxel volumes. The total intracranial volume (TIV) was computed as the sum of GM, WM, and CSF volumes.

ROI volumes were computed by applying a binary mask to the modulated images. The mask was traced with MRIcro along the boundaries of the temporal lobe on the mean of normalized MR images,23, 24 and volumes were calculated as described above and normalized for TIV.

VBM analysis

SPM99 was used to carry out an ANCOVA, to detect atrophic regions of 452Tyr carriers compared to noncarriers. Proportional scaling was set at 0.8, and age, sex, and intracranial volume were included as nuisance covariates. A small volume correction procedure was used through a binary lobar mask outlining the temporal lobes. P was set at 0.05 with small volume correction. An uncorrected confirmatory full brain analysis was performed with significance threshold at P=0.001, in order to check that significant clusters were confined to the temporal ROIs.

Statistics

Hardy–Weinberg equilibrium was tested by the Arlequin software version 2.000.25

Despite the low number of His452Tyr carriers, parametric tests (T test) were used for brain measurement as values were normally distributed, as shown with Shapiro–Wilk test.

Results

Genotype frequencies were in Hardy–Weinberg equilibrium (P=0.59). The frequency of the 452Tyr allele was 0.105, similar to the value reported for Caucasian populations13 (χ2=0.125, d.f.=2, P=0.72), genotypic frequencies of His452His, His452Tyr, and Tyr452Tyr being 0.803 (61/76), 0.184 (14/76), and 0.013 (1/76), respectively.

452Tyr carriers and noncarriers had similar age, sex, and education (Table 1). The clinical reasons leading to the prescription of the MR exam – migraine and headache (28%), auditory (38%) or visual (7%) symptoms, paresthesia (5%), suspected cerebrovascular disease (6%), and miscellaneous reasons (16%) – were homogeneously distributed between carriers and noncarriers (χ2=5.455, d.f.=5, P=0.409).

Table 1 Socio-demographics and clinical features of carriers and noncarriers of the His452Tyr polymorphism are described in detail
Full size table

The normality of the brain variables has been assessed with Shapiro–Wilk test (left temporal GM: d.f.=15, P=0.226; right temporal GM: d.f.=15, P=0.529; left temporal WM: d.f.=15, P=0.397; right temporal GM: d.f.=15, P=0.971).

ROI volume analysis indicated a significant reduction of the WM in the left temporal lobe. It should be highlighted that, although not reaching the threshold for statistical significance, the point estimates of lobar volumes of the right WM and bilateral GM indicated smaller volumes in carriers compared to noncarriers (Table 2).

Table 2 Region of interest (ROI) volumetric analysis: temporal GM and WM volume in His452Tyr polymorphism carriers compared to noncarriers
Full size table

VBM analysis showed that 452Tyr carriers had reduced GM volume in the left hippocampus and left inferior temporal gyrus and, bilaterally, in the middle and superior temporal gyrus (Figure 1 and Table 3). The WM was reduced in the left temporal regions (superior temporal sulcus and gyrus) in the 452Tyr carriers, contiguous to the clusters of GM reduction.

Figure 1
figure 1

VBM analysis at P=0.05 corrected with small volume correction showing volume reduction in carriers vs noncarriers of the 452Tyr allele. L=left; R=right. Red spots indicate GM and green spots indicate WM regions of atrophy in subjects carrying the 452Tyr allele.

Full size image
Table 3 VBM analysis: atrophic regions in carriers versus noncarriers of the 452Tyr allele of the 5HT-2A receptor gene (P=0.05 with small volume correction)
Full size table

Confirmatory whole-brain VBM analysis showed that volume differences were confined exclusively to the temporal regions (Figure 2). The opposite comparison did not detect any clusters of atrophy in the regions of interest.

Figure 2
figure 2

Whole-brain VBM analysis confirming that volume reductions of carriers vs noncarriers of the 452Tyr allele were confined to the temporal regions. Maximal gray (red) and white (green) matter's Z-scores have been projected onto a brain surface (P=0.001 uncorrected).

Full size image

Discussion

This is the first study investigating the relationship between 5-HT2A receptor polymorphism and brain structure. The VBM findings show an effect of the 452Tyr allele on temporal brain structures, carriers showing a reduction of the GM volume in the left hippocampus and inferior temporal gyrus (Brodmann area 20), and in the middle and superior temporal gyri bilaterally (Brodmann areas 21 and 22). A significant reduction of the WM is also evident in the left temporal lobe, in a region adjacent to the hippocampal formation. The ROI findings show a significant reduction of the WM in the left temporal lobe and smaller volumes of the right WM and bilateral GM in carriers compared to noncarriers of the His452Tyr polymorphism. These findings are obtained with two methods that provide complementary information about a possible loss or reduction of volume, ROI analysis showing a widespread and VBM showing more localized changes. Indeed, the results show widespread reduction of the WM in the left temporal lobe, detected by ROI, and a more localized reduction of the GM, detected by VBM analysis.

These findings are consistent with the poorer cognitive performance of 452Tyr carriers showed by recent studies.14, 15 Such reduction of performance, observed for memory function, has actually been described for subjects younger than those in our sample. Our neuropsychological battery was not conceived to investigate in great detail memory function; therefore, a possible difference in performance might have not been detected in our study. Alternatively, older subjects might not display such difference at all. If this is the case, the smaller volume of brain tissue in the medial temporal regions might be associated to a lower potential capability for memory performance that might be compensated by alternative cognitive strategies in older people.

An effect on synaptic plasticity might explain the influence of the 452Tyr allele on temporal lobe structures, as suggested by studies showing a relationship of the 5-HT2A receptor with physiological processes closely related to synaptic plasticity such as, phospholipase C and G proteins activation,26 intracellular Ca2+ mobilization and release.27

The synaptic plasticity hypothesis is further supported by data indicating a role of 5-HT2A receptors in adult neurogenesis and synaptogenesis. Adult mammals cells proliferate in subgranular layer and tend to migrate to the granule cell layer of the dentate gyrus.28, 29 These ‘migrating’ cells mature, grow dendrites, and establish synaptic contact within the hippocampus.30 Serotonin is one of the factors mostly involved in this process.31 Using systemic administration of 5-HT2A receptor antagonists in a mouse model of neurogenesis, a 63% reduction of the number of proliferating cells in the subgranular layer of the dentate gyrus has been shown,32 which consequently might determine a reduction in the cells maturation, in the growth of dendrites, and in the establishment of synaptic contact.

Data in the literature indicate that the 452Tyr allele might determine a reduced ability of 5-HT2A receptors to activate both phospholipase C and G proteins,26 which are known to have a role in long-term potentiation (LTP).33, 34, 35 LTP is regarded as one of the most popular and widely researched model involved in synaptic plasticity,36, 37, 38, 39, 40 which is also able to yield structural changes41 in hippocampal formation.

Moreover, it has been shown that 5-HT2A receptor development is associated with an increased Ca2+ availability in neurons,42 that in 452Tyr carriers has been shown to be significantly reduced.24 This physiological process, acting on the internal cytoskeleton of the cell, is pivotal to cell proliferation and synaptogenesis,42 which are physiological processes interested in synaptic plasticity.43, 44, 45, 46

Also, the 5-HT2A receptor has been shown to be involved in the regulation of BDNF mRNA in the hippocampus and neocortex.47 The administration of a 5-HT2A receptor agonist, the hallucinogenic phenylalkylamine 4-iodo-2,5-dimethoxyphenylisopropylamine (DOI), showed a downregulation of BDNF mRNA levels in the frontal, parietal, and temporal cortex, particularly in the granule cell layer of the dentate gyrus. Indeed, it has been shown that BDNF mRNA expression in the hippocampal formation is influenced by stress,48 and animal studies have demonstrated a role of 5-HT2A receptors in mediating BDNF mRNA expression under stressful conditions49, 50 supporting the involvement of 5-HT2A receptors in BDNF mRNA expression. It has been shown that a lower level of BDNF mRNA yields a decrease in neural plasticity.51, 52 Subjects carrying the rare 452Tyr allele variant of the 5-HT2A receptor gene have a smaller peak amplitude in Ca2+ mobilization and release,27 which is related to a lower expression of BDNF mRNA53 in hippocampal neurons, and consequently it might lead to a decrease in neural plasticity.

Although the above-cited evidence converges in supporting the hypothesis that the different brain morphology observed in subjects with the 452Tyr allele might be due to an effect on neural plasticity, alternatives should not be disregarded. It might be hypothesized that persons with the 452Tyr allele have smaller brain volumes to start with. To date, very few studies have assessed the expression of 5-HT2A receptors in the embryo and their effect on neural development. One such a study has shown that 5-HT2A receptors affect morphogenesis of a variety of embryonic tissues such as the neuroepithelia of the brain and spinal cord, although specific studies on medial temporal structures are lacking.54 The neurodevelopmental hypothesis is particularly attractive to explain the effect of the 452Tyr allele on the WM where the absence of synapses makes the synaptic plasticity hypothesis less tenable.

Considering the small number of subjects in our sample, these findings should be considered preliminary and needing further confirmation with a larger number of subjects. Healthy volunteers selected randomly from the general population would be preferable than subjects intercepted in a neuroradiological department. The strict exclusion criteria we have employed should have allowed us to rule out conditions possibly related to both predictors (genotype) and outcome (regional volume changes). Indeed, although our experimental groups have a high prevalence of conditions, such as migraine and headache, these are not differentially distributed in carriers and noncarriers of the His452Tyr polymorphism and thus cannot be reported as confounders. The suggested hypotheses, of synaptic plasticity and neurodevelopment, might be better analysed studying the possible role of the His452Tyr polymorphism in the period of life between childhood and adulthood, in order to better understand its mechanism of action. These brain structural differences should also be confirmed with different brain measurement tools. The use of different scanners with a greater Tesla value (1.5/3.0) may improve the spatial resolution, in fact the higher is the Tesla value of the scanner, the better is the spatial, temporal, and spectral resolution and this may result in an improved accuracy. But it has also been noted a greater noise and also a higher susceptibility for artefacts, leading to geometric distortion and signal loss, particularly near the air-filled sinuses.55 Greater insight into the mechanisms through which genes influence brain structure and function certainly deserves deeper investigation.