1. Introduction
TRP (Transient Receptor Potential) channels are evolutionarily conserved integral membrane proteins. They are structurally characterized by six transmembrane helices forming ion channels with variable cation selectivity [
1], expressed in almost all tissues and cell types (GTEx Release V8).
In mammals, 28 TRP channels have been identified, grouped into subfamilies based on amino acid sequence homology: TRPC (‘canonical’), TRPM (‘melastatin’), TRPV (‘vanilloid’), TRPA (‘ankyrin’), TRPML (‘mucolipin’) and TRPP (or PKD) (‘polycystin’) [
2]. These channels play essential roles in various physiological processes and participate in many sensory modalities [
3].
They are involved in environmental sensing, being able to respond to a wide range of stimuli such as temperature, pH, osmolarity, pheromones, and plant compounds. In the oral cavity and nose, by transmitting signals generated by compounds in food to the brain, they give rise to the chemesthetic sensations of irritation, hotness, coolness, and pungency [
4]. TRP channels play a key role in thermosensory perception and adaptation in several species [
5,
6].
TRPV (transient receptor potential vanilloid) channels were so named because the first identified member of this group, VR1, later named TRPV1, responded to the vanillylamide capsaicin, a chemesthetic compound produced by plants of the genus
Capsicum [
7]. Chemesthesis plays an important role in the body’s chemically activated defense mechanisms, the ‘chemofensor complex’ [
8], used to avoid (or reduce) contact, inhalation, and ingestion of potentially harmful compounds. Interestingly, the TRPV1 receptor is also involved in innate immunity, being activated by N-acyl homoserine lactones, quorum-sensing molecules produced by Gram-negative bacteria [
9]. This role is also demonstrated for taste receptors, in particular certain bitter receptors [
10].
The
TRPV1 gene is expressed in 54 human tissues (GTEx Release V8). Activation of oro-pharyngeal TRPV1 following consumption of food containing hot compounds is perceived as high temperature. Physiological responses, such as gustatory sweating, are activated to counteract this input. The TRPV1 activation threshold temperature is lowered by vanilloids and many other natural compounds, such as piperine in pepper and gingerols in ginger. pH values below 6, a level easily reached by tissue injury such as infection and inflammation, also show this activity [
11], giving TRPV1 an important role in the process of injury-related hyperalgesia, inflammation, and pain [
12]. Activation of TRPV1 by vanilloids is followed by rapid and sustained desensitization [
13,
14], resulting in a particular form of analgesia, thus making TRPV1 a potential pharmacological target in pain therapy [
15,
16,
17].
TRPV1 channels are also expressed in the central nervous system (CNS), where temperature and pH are strictly constant, suggesting the existence of endogenous brain agonists for this receptor, identified in endovanilloids such as anandamide and N-arachidonoyl-dopamine (NADA) [
18,
19,
20].
The systemic response resulting from the activation of TRPV1 (among other TRPs) plays a role in adipocyte thermogenesis, adipogenesis, adipose tissue inflammation, and obesity [
21,
22,
23], as well as water retention [
24]. However, the relationship between TRPV1 variants and body composition in healthy individuals is a genotype-phenotype relationship that has not been explored. In previous papers PROP (6-n-propylthiouracil) phenotype (depending mainly from the bitter taste receptor gene
TAS2R38 genotype) has been proposed as indicator of body mass and adiposity, with contradictory results from different research groups [
25,
26,
27,
28].
The use of chili peppers is widely applied in several folk medicines to prevent and treat diabetes and other metabolic disorders [
29,
30,
31], influencing, directly or indirectly, the energy balance and therefore body weight [
32]. The mechanism of action of capsaicin in glucose control, energy homeostasis, and obesity-related diseases has been explained by both TRPV1-dependent and TRPV1-independent mechanisms [
33].
Despite the involvement of TRPVs in important physiological and environmental signaling, little is known about their genetics, and very few papers have reported a systematic analysis of genomic data [
34].
Previous studies have shown that the genetic variability at
TRPV1 includes splicing variants, sometimes with tissue-specific expression, with greater or lesser sensitivity to a specific category of stimuli (e.g., vanilloid agonists, [
35]). Although variability in the sensory perception of hotness between and within populations is known, no study to date has been able to identify a relationship between this phenotypic trait and
TRPV1 variants [
36]. Furthermore, the role of TRPV1 in systemic responses implies a relationship with other traits, such as pain perception [
37] and inflammatory disease risk [
38,
39].
In this study, we addressed some fundamental questions about the role of TRPV1 channels in recent human evolution. How is the variability at the TRPV1 gene distributed among human populations? Is this variability related to variations in environmental conditions? Is it related to a phenotype with adaptive relevance, like body composition?
To answer these questions, we first explored human variability upstream and downstream of the TRPV1 gene through database searching and in silico analyses. Then, we collected both genotypic and phenotypic data from 46 volunteers of sub-Saharan origin and 45 volunteers of Italian origin to associate TRPV1 variants with body composition and sensory perception.
2. Materials and Methods
2.1. Linkage Blocks Detection
A survey of genomes deposited at the International Genome Sample Resource (IGSR) and the Estonian Genome Centre Database (EGDP 483 high-coverage genomes, 148 populations worldwide) was performed. Within the IGRS, three datasets were considered: the 1000 Genome Project Phase 3 (KGP3, 2504 whole genomes, 26 populations worldwide), the Human Genome Diversity Project (HGDP, 828 individuals, 54 populations), and the Simons Genome Diversity Project (SGDP, 276 individuals, 129 populations). SNP frequencies were obtained from the ORF of the human
TRPV1 gene (transcript NM_080704.4) and a surrounding region of −5000 bp/+10,000 bp (GRCh38.p14 coordinate 17:3560446-3619411,
Figure 1).
The analytical software PLINK version 1.9 [
40] was used to calculate the minor allele frequency (MAF, --freq function) and genetic divergence between groups (Wright–Malécot Fst values, --fst function) for each SNP. Fst values above 0.15 were considered moderately high, and values over 0.25 were considered as high according to Frankham et al. [
41]. PLINK v. 1.9 was also used to identify haplotype blocks in pairwise linkage disequilibrium (D’ and R
2 > 0.75, --r2 and --ld-snp functions), and DNAsp [
42] was used to detect natural selection signals (Tajima’s neutrality test).
2.2. Sample Composition
A group of 46 healthy sub-Saharan African donors (SSA) of both sexes was recruited and compared with an Italian (ITA) sample of 46 healthy volunteers (
Table 1 and
Supporting Information: Table S1). Exclusion criteria were: subjects with pulmonary, severe cardiovascular or uncontrolled metabolic diseases, electrolyte abnormalities, cancer, inflammatory conditions or using implanted electrical devices were excluded from the study. The mean age of the subjects was 32 ± 10.9 and 26 ± 4.9 years for the SSA and ITA groups, respectively.
The ancestry of the volunteers was verified by the residence on the identity document and by a self-declaration regarding the country of origin of the 2 parents.
By signing a consent form, volunteers agreed to complete a questionnaire on dietary habits, perform a sensory perception test, measure their body composition, and donate a saliva sample for genetic analysis.
2.3. Dietary Habits
A questionnaire was administered to reconstruct individual food consumption, as well as personal data and health conditions, including metabolic disorders (
Supporting Information: Table S2 and Supporting Information: Figure S1). Food consumption frequency was assessed for seven hot foods (containing TRPV1 agonists) and three cooling foods (containing TRPM8 agonists). Questionnaire responses were organized into three categories: null (never), moderate (at least 2 times per month), and frequent (at least 2 times per week).
Also, questions were asked about hot, cold, and pain tolerance, categorized as low, medium, and high.
2.4. Perception Tests
A test was performed to assess the capsaicin perception threshold in each individual. Ten solutions of pure capsaicin were prepared by consecutive 1:2 dilutions in ethanol 99.3%, ranging from 2.243 μg/mL to 0.004 μg/mL. Then, 20 μL of each solution was aliquoted onto cotton swabs numbered 1 to 10, and ethanol allowed for drying out completely. Individuals were not informed of the substance they were about to taste. The swabs were tested sequentially, starting from the lowest concentration (swab 1), proceeding to the next higher concentration until the subject could feel a sensation. Volunteers were asked to hold each swab on the tongue for at least 5 s and to rinse their mouths with room temperature water before each new swab. The corresponding cotton swab number was recorded, and the intensity of the stimulus was indicated on a Labelled Magnitude Scale (LMS, [
43]), with a scale of 0 to 100.
Since the correlation of PROP phenotype with body mass and adiposity were reported with divergent results from different groups [
25,
26,
27,
28], we decided to also perform a PROP perception test as reported in Risso et al. [
44]. A 50 mM PROP solution (8.51 mg/mL) was prepared following the protocol with filter paper discs described in Zhao et al. [
45]. At the end of the capsaicin test, a single swab of PROP was offered to be held in the mouth for 10 s. Subjects were asked to record taste intensity on an appropriate LMS.
Ten categories were identified for capsaicin perception threshold, corresponding to the lowest concentration perceived. Two categories were reported for PROP perception, following Drayna’s classification into “tasters” (score ≥ 50) and “non-tasters” (score < 50) phenotypes [
46]. Each category was treated as a distinct phenotype in the genotype–phenotype and phenotype–phenotype correlation analysis (see below).
2.5. Anthropometry and Body Composition
The sampling phase took place in Italy during the temperate seasons, between late March and early June and in late September, to ensure that water loss levels were not affected by body acclimation. Individuals were asked to observe a complete fast of at least three hours (no food, no water) before the measurement.
Anthropometric measurements (height, weight, upper arm, waist, and circumferences) were taken according to standard international criteria [
47]. Body mass index (BMI) was calculated as weight/height
2 (kg/m
2).
Bioelectrical values of resistance (R, ohm) and reactance (Xc, ohm) were measured through a portable Vitality AnalyzerTM bioimpedance device (IPGDX, LLC, Littleton, CO, USA), which applies a current of 0.6 microA at 50 kHz and using the standard positions for the outer and inner electrodes on the right hand and foot.
To evaluate body composition, specific Bioelectrical Impedance Vector Analysis (specific BIVA; [
48]) was applied. BIVA allows the graphical and statistical analysis of bioelectrical vectors, as defined by their module, that is, impedivity (Z = (R
2 + Xc
2)
0.5, ohm), and their inclination, that is, phase angle, (PhA = arctangent Xc/R * 180/π, degrees).
Following the specific BIVA approach, R and Xc were standardized by a correction factor A/L, where A represents an estimate of the transverse area of the body (0.45 arm area + 0.10 waist area + 0.45 calf area) and L the distance between electrodes (height*1.1) [
48]. According to Ohm’s law, this correction reduces the influence of body size and shape on bioelectrical variables (specific resistance, Rsp, ohm*cm; specific reactance, Xcsp, ohm*cm), which are therefore mainly related to body composition variability (data are shown in the
Supporting Information: Table S3).
As in the classic vectorial approach [
49], two types of graphs represent the output of specific BIVA on the plane defined by specific resistance on the
x-axis and specific reactance on the
y-axis: the tolerance ellipses and the confidence ellipses. Tolerance ellipses represent the bioelectrical variability of the reference population; individual or mean vectors can be plotted on the graph, and their body composition can be evaluated depending on their position. According to specific BIVA [
48], the major axis of specific tolerance ellipses, mainly due to variations in vector length and Rsp, is related to variations in relative fat mass content (FM%), with higher values towards the upper pole. The minor axis, mainly due to variations in PhA and Xcsp, is related to body cell mass (higher values on the left) and extracellular-to-intracellular water ratio (ECW/ICW) (higher values on the right) and is considered a proxy for muscle mass and quality. Confidence ellipses represent the 95 percent confidence interval around the sample mean and allow the graphical comparison among samples, with no overlapping ellipses indicating significant differences [
49]. The statistical difference between confidence ellipses can be evaluated by the Hotelling T
2 test [
50].
Samples were grouped by sex, as muscle and fat mass and distribution are significantly different between males and females [
51], with the latter showing longer impedance vectors and lower phase angle [
48]. When focusing on a phenotypic trait, regardless of the sex, we considered males and females together after standardization (Z scores).
To evaluate sample distribution patterns and identify outliers, the vectors of each group were plotted on the specific tolerance ellipse representing the best approximation of the source population. ITA vectors were plotted on an Italo-Spanish reference sample (213 males, 227 females, aged 18–30 years, [
52]). SSA vectors were plotted on an African American reference sample (181 males, 175 females, aged 18–49 years), representing a subsample of the NHANES dataset analyzed by Buffa et al. [
48], as body composition reference data from the African continent are still lacking in the literature.
2.6. DNA Sampling and Genotyping
A volume of 2 mL of saliva was collected with the Oragene™ DNA Self-Collection Kit (DNA Genotek, Ottawa, ON, Canada) and stored at room temperature for several weeks before extraction. Whole DNA extraction was performed with the prepIT©L2P Laboratory (DNA Genotek Inc., Canada) protocol according to the manufacturer’s instructions.
Two SNP-rich segments identified in the
TRPV1 gene region (herein referred to as R1 and R2,
Figure 1) were selected for PCR amplification and Sanger sequencing. To ensure the absence of polymorphisms within the primer sequences, the region was screened to identify all SNPs, using the USCS (genome.ucsc.edu) and ENSEMBL (ensemble.org) genome browsers. Candidate primers were generated with Primer3Plus software (v. 3.3.0) [
53]. Melting temperature and absence of dimerization were assessed with AutoDimer (implemented in STRBase 2.0, [
54]). Finally, the ability of primers to match sequences of other species was tested with NCBI BLAST [
55]. The total amplicon length was set to 699 bp and 508 bp for R1 and R2 segments, respectively, taking care to allow a distance of 20–30 bp from the outermost polymorphic locus (
Supporting Information: Table S4).
DNA was sequenced by the Cycle Sequencing method using the BigDye™ Direct Cycle Sequencing Kit (ThermoFisher Scientific, Waltham, MA, USA) and an ABI 3730xl DNA Analyzer system (phred: 20–1100 bp).
Genotype calling was performed manually by aligning FASTA files to the GenBank reference sequence (GRCh38) using BioEdit 7.7 software [
56]. For R2 genotype calling, the forward strand was considered, whereas R1 required the sequencing of both strands, forward and reverse, due to the presence of a tetranucleotide STR (nsv1874533, 17:3596296-3596321).
2.7. Data Analysis
Subjects were divided into groups according to ancestry (ITA and SSA) and sex. SSA was subsequently grouped according to the diplotype. Questionnaire responses organized into frequency or intensity categories were compared to capsaicin perception to assess the environmental component of diet on taste perception. Capsaicin perception threshold distribution among groups was also tested for association to sex, ancestry, and diplotype (for SSA only).
The specific BIVA approach was used to assess body composition. To assess the difference between subgroups, the Hotelling T
2 test was applied to confidence ellipses, considering the center of each ellipse as the bivariate mean of Rsp and Xcsp, and setting the significance threshold at
p = 0.05. The Mahalanobis distance index (D
2, [
57]) and Fisher F indices were also used to estimate the distance between the two data distributions. When calculating D
2, the critical values of variables Rsp and Xcsp correspond to the perimeter of the 95% confidence ellipses.
The representativeness of the sample was tested by comparing allele frequencies and haplotype occurrence with the KGP3 group of Tuscan individuals (TSI) and the AFR(-ASW) group (all African subgroups except ASW, which include individuals with African American ancestry). Minor allele frequency (MAF, --freq) and linkage disequilibrium (LD, --r2 and --ld-snp) at each locus were calculated using PLINK v. 1.9 software [
40].
For SSA only, the correlation between each diplotype–phenotype pair was assessed by performing ANOVA and linear regression tests (--linear --covar). Ten phenotypes were distinguished for capsaicin perception, corresponding to the 10 perception thresholds. In addition, the PROP perception intensity distribution was calculated and correlated to BMI values.
To test the correlation between genetic variability and body composition, the four bioelectric variables (Rsp, Xcsp, Zsp, PhA) were considered as four phenotypes and tested independently against each diplotype. To avoid sex bias on body composition, sex was considered as a covariant in the regression model.
To investigate the effects of climate differences, SSA was divided into West Africans (22 subjects) and East Africans (24 subjects), and association was measured with haplotype status.
4. Discussion
Despite the potential involvement of the TRPV1 gene in adaptive processes, associations between its variants and phenotypes have not been studied across human populations to date. In this research, two highly polymorphic LD blocks (named H1 and H2) were identified in silico in the ORF and the 5′UTR upstream region of the TRPV1 gene, showing patterns compatible with a process of balanced selection. Their evolutionary significance was explored by measuring the association between diplotype states, body composition, sensory perception, and dietary habits.
At the level of blocks H1 and H2, we identified haplotypes showing high MAF (43–49%) and highly positive Tajima’s D values (D > +4.5) only in the sample with sub-Saharan ancestry. A positive D reflects episodes of population subdivision/contraction or balanced selection [
58], and it is not trivial to discern between the two alternatives. In this case, however, the influence of demography can reasonably be considered negligible, as the excess of pairwise mutation differences over the number of segregating sites has been observed on a continental scale and localized to a short nucleotide sequence, with no impact on the genome as a whole.
The fact that, both in the databases and in the sampled groups, no major deviations from Hardy–Weinberg expectations were observed can be interpreted as the effect of natural selection operating over long periods of time. Presumably, the
TRPV1 region has escaped previous genomic scans [
59,
60,
61,
62,
63] due to the small size of the regions involved (2.7% of the total gene length), the difficulty in simulating different types of stabilizing selection, and the low power of statistical approaches, which ignore intragenic recombination [
64].
Our results suggest that the selective agent is to be sought in a different history of human–environment interaction in Africans and non-Africans.
Dietary habit was the candidate agent we analyzed first. The TRPV1 agonist capsaicin is involved in weight loss by decreasing appetite and by increasing fat mobilization and insulin/leptin resistance [
33]. The process is coupled to an increase in brown adipose cells and a decrease in white adipogenesis [
65]. Nonetheless, no significant association of diplotypic states was observed with sensitivity to capsaicin (or to PROP), BMI, or food consumption. Genetic influence may be masked by polygeny or by exogenous factors such as frequency of consumption or the synergistic action exerted by the gut microbiota ([
36], Vinerbi et al. in preparation). Further research on this regard is in progress.
Comparative studies examining animal species adapted to different thermal environments demonstrated how changes in TRPV1 heat responses (but not to capsaicin or acids, thereby maintaining its function as a detector of chemical cues) arise from just one amino acid difference in the orthologous genes [
66]. Therefore, physiological response to climatic conditions, which is known to strongly influence body composition and shape, was the second candidate agent analyzed, and it has to be intended here as the complex of processes that, in the long term, guarantee osmotic and energetic homeostasis. As we have observed experimentally through anthropometric and impedance measurements, the SSA and ITA groups are clearly distinct in terms of body composition. Sub-Saharan Africans recorded higher Zsp and PhA values, indicative of a higher percentage of fat mass (FM%) and a lower extracellular/intracellular water ratio (ECW/ICW). This could be due to a different lifestyle, but it is also consistent with an evolutionary adaptation to long periods of drought and famine. Indeed, in an arid environment, heat loss and transpiration are maximized by increased surface area/volume ratio and fat accumulation [
67]. Fat deposition/mobilization and water retention are controlled by sexual hormones, so their regulation is different in the two sexes [
68,
69]. This explains why the sampled women showed higher values of FM% and ECW/ICW, regardless of ancestry. Furthermore, SSA women show higher FM% and lower ECW/ICW than ITA women. Interestingly, we found that in SSA individuals, the BIVA values were correlated with
TRPV1 H1-b and H2-b haplotypes, in both homozygous and heterozygous states. These haplotypes are absent in non-African populations as the relevant variants are almost monomorphic and not in linkage.
Regarding the ECW/ICW ratio, it is known that a state of hyperosmolarity of the circulating blood corresponds to an activation of TRPV1 in the central nervous system, which leads to an increase in water retention at the systemic level [
24]. During episodes of hyperthermia, when the sympathetic nervous system (SNS) and the hypothalamic–pituitary–adrenal (HPA) axis trigger processes that favor water loss (sweating, tachypnoea, skin vasodilation, salivation), TRPV1 induces the same compensatory reactions [
70,
71].
Regarding FM%, it is known that TRPV1 and lipids mutually interact to regulate their expression [
22,
65,
72]). Various types of lipids, such as phospholipids, triglycerides, and steroids (including estrogen and oxytocin), influence the gating activity and/or expression of
TRPV1 [
73,
74,
75,
76,
77]. On the other hand, TRPV1 channels influence lipid metabolism in a complex manner that has so far yielded contradictory results. Decreased
TRPV1 expression in mice has been found to protect against diet-induced obesity [
78] and promotes oxygen consumption, fat oxidation, and locomotor activity [
79]. In contrast, loss of the
TRPV1 gene in Western-fed mice causes hyperlipidemia, and animals exhibit reduced locomotor activity with a more pronounced effect in males than in females [
80,
81]. TRPV1 agonists, such as vanilloids, capsaicin, and oxytocin, influence lipid metabolism, mainly by reducing lipid deposition. By activating TRPV1 channels, they cause an increase in intracellular free Ca
2+ levels, thus triggering the desensitization of nociceptive neurons [
82] and the suppression of visceral fat accumulation through the upregulation of UCP1 (Mitochondrial Uncoupling Protein 1) [
83]. Lipid metabolism has also been found to be regulated by the interaction of
TRPV1 with transcription factors involved in energy homeostasis, such as peroxisome proliferation-activated receptors (PPARs, [
84,
85]) and sterol-responsive element-binding proteins (SREBPs, [
86,
87]).
Our results suggest a role for the H1 and H2 haplotypes as sequences harboring binding sites that regulate the activation/suppression of TRPV1 expression by different mechanisms, acting independently or synergistically to enhance fat storage efficiency. They have not yet been associated with an altered expression level of TRPV1, and no SNPs of the two blocks are present in the GTEx or RegulomeDB browsers. However, the overlap of H1 with an enhancer, an open chromatin region, and a transcription repressor (CTCF) suggests that these mechanisms are involved in the regulation of TRPV1 channel expression in peripheral tissues. Furthermore, the overlap with TRPV3 and SHK gene regions may suggest that H1 and H2 play a regulatory role in these genes in addition to or instead of TRPV1.
Balanced SNP/haplotype frequencies and neutrality tests suggest that the most likely cause of the association between
TRPV1 haplotypes and body composition is balancing selection. A prerequisite for this selective process is the greater fitness of heterozygous than homozygous diplotypes or the frequency-dependent fluctuation of alleles around an average value. The frequency pattern of
TRPV1 alleles in sub-Saharan populations does not faithfully obey the model of balancing selection based on heterozygous advantage. The framework found is more compatible with a sex-specific directional selection, based on the higher fertility (fitness) of females, who more efficiently accumulate subcutaneous fat, and with an enhancing effect of sexual selection, which promotes the male selection of females with pronounced gynoid forms (low waist-to-hip ratio), according to local beauty standards [
88] (Yu & Shepard, 1998).
Women show a greater capacity than men to store excess free fatty acids, obtained during periods of energy surplus, in the form of subcutaneous adipose tissue (SAT) [
69,
89]. This excess fat mass is highly mobile, and lipids can be easily recovered to get through pregnancy and lactation under conditions of prolonged nutrient deprivation. The SAT is rich in brown fat, which is highly supplied by blood vessels and thus serves as an easily exploitable water reserve. Furthermore, unlike visceral adipose tissue (VAT), SAT has low lipase activity and protects against diet-dependent ectopic fat formation, thus providing protection against cardiovascular disease and type 2 diabetes [
90,
91,
92].
An extreme case of this phenomenon is the steatopygia of African hunter-gatherers, such as the Khoi-khoi of southern Africa or the pygmy groups of West Africa [
93]. A sex-dependent role of
TRPV1 in SAT metabolism is also suggested by the fact that estrogen regulates
TRPV1 in the endometrium of immature rats [
94] and in the arterial and bladder smooth muscle of post-pubertal female rats [
95].