The prototypical member of the vanilloid-responsive-like subfamily of transient receptor potential (TRP) channels is TRPV1. TRPV1 mediates aspects of nociception and neurogenic inflammation; however, new roles are emerging in sensation of both luminal stretch and systemic tonicity. Although at least six nonsynonymous polymorphisms in the human TRPV1 gene have been identified, there has been no systematic investigation into their functional consequences. When heterologously expressed in HEK293 cells, all variants exhibited equivalent EC50 for the classic agonist capsaicin. This agonist elicited a greater maximal response in TRPV1I315M and TRPV1P91S variants (relative to TRPV1WT), as did a second agonist, anandamide. Expression of these two variants in whole-cell lysates and at the cell surface was markedly greater than that of wild-type TRPV1, whereas expression at the mRNA level was either unchanged (TRPV1P91S) or only very modestly increased (TRPV1I315M). Incorporation of multiple nonsynonymous SNPs, informed by the population-specific haplotype block structure of the TRPV1 gene, did not lead to variant channels with unique features vis-à-vis capsaicin responsiveness. Recently, polymorphisms/mutations were identified in two highly conserved TRPV1 residues in the nonobese diabetic (NOD) murine model. Incorporation of these changes into human TRPV1 gave rise to a channel with a normal EC50 for capsaicin, but with a markedly elevated Hill slope such that the variant channel was hyporesponsive to capsaicin at low doses (<10 nM) and hyperresponsive at high doses (>10 nM). In aggregate, these data underscore expression-level and functional differences among naturally occurring TRPV1 variants; the implications with respect to human physiology are considered.

  • capsaicin
  • single-nucleotide polymorphism
  • diabetes

the transient receptor potential (TRP) family of channels transmit a broad range of environmental information to the cell interior, generally via calcium entry. Specific TRP channels respond to high and low temperatures, anisotonicity, and an array of tastants, as well as numerous other stimuli (reviewed in Ref. 45). TRP channels are divided into six broad families based on sequence homology (49); the TRPV family is named for its founding member, the vanilloid receptor, or TRPV1.

TRPV1 is a nonselective cation channel (7, 37, 51) that was initially cloned and molecularly characterized as the capsaicin (i.e., “hot pepper”) or vanilloid receptor (14). TRPV1 mediates the ion fluxes accompanying exposure of neurons and mucosae to a variety of noxious stimuli. In vitro, TRPV1 is activated by voltage, vanilloid compounds, heat, protons, and a variety of lipid agonists including endogenous cannabinoids (89) and eicosanoids (27). TRPV1 is also activated by zingerone and piperone, compounds lending distinctive pungency to ginger and black pepper, respectively (36, 42). TRPV1 is predicted to have six transmembrane segments (14) and to assemble into a tetrameric structure (33). TRPV1, like other TRPV family members, participates in heteromerization (70).

Capsaicin, as mentioned above, is the prototypical activator of TRPV1; the compound is used therapeutically in desensitization of nociceptive pathways (78). Capsaicin likely traverses the plasma membrane to interact with intracellular site(s) on the channel (30). Initial data suggested involvement of intracellular residues R114 and E761 (31); data derived in part from domain-swapping experiments with the capsaicin-insensitive avian and lapine TRPV1 have implicated the second through fourth membrane-spanning helices and an intervening intracellular loop (21, 29). Interpretation is complicated by evidence for uncoupling of capsaicin binding and actual channel activation, consistent with an allosteric gating mechanism (33). Deletion of a 60-residue cassette from the NH2 terminus in a naturally occurring human TRPV1 splice variant renders the channel insensitive to capsaicin; the response to noxious heat, however, may be preserved (38, 83). Binding assays and single-channel electrophysiological studies suggest that interaction with at least two capsaicin molecules is required for full channel activation (reviewed in Ref. 61).

Consistent with its role in nociception, TRPV1 is primarily expressed in the dorsal root ganglia and peripheral sensory nerve endings and, to a much lesser extent, in the central nervous system (14, 43, 63). Increasingly, however, TRPV1 has been identified in both healthy and diseased epithelia. TRPV1 is expressed in urethra and bladder (1, 9, 53, 79), including the epithelial cells lining the bladder (9). Mice lacking TRPV1 (TRPV1−/− mice) exhibited impaired sensation of bladder filling and abnormal reflex voiding (10). TRPV1 is also expressed in nasal and airway mucosa (2, 60, 77, 79, 80) and the gastrointestinal tract (5, 22), where its dysregulation has been implicated in a number of disease states (11, 12, 28, 40, 85). TRPV1 is highly expressed in the kidney (14, 16, 62, 83), along with potentially unique splice variants (14, 16, 62, 81). Although its function remains obscure in this tissue, a pathological role in the uptake of the nephrotoxic antibiotic gentamicin has been suggested (48). TRPV1 also innervates the renal pelvis and mediates the renorenal reflex, wherein elevated intrapelvic pressure (as might accompany unilateral ureteral obstruction) promotes natriuresis via the contralateral kidney (88).

TRPV1, like the closely related TRPV4 (34, 35, 76), is suspected of playing a pivotal role in systemic osmoregulation (67). Hypertonicity activates a cation current in the osmotically sensitive ADH neurons of the supraoptic nucleus; however, corresponding neurons from TRPV1−/− mice were unresponsive to this physiological stimulus. In further experiments, a channel exhibiting electrophysiological properties consistent with TRPV1 was identified in these cells. Interestingly, only a portion of TRPV1 mRNA could be PCR amplified from RNA prepared from this tissue (67), and no osmotically responsive variant of TRPV1 has been described to date.

We hypothesized that nonsynonymous polymorphisms of human TRPV1 may confer novel channel properties. We compared known nonsynonymous single-nucleotide polymorphisms (SNPs) to wild-type TRPV1 in vitro through a variety of approaches. Although EC50 for capsaicin was similar among the variants, at least two, TRPV1P91S and TRPV1I315M, exhibited increased expression at the protein level. Gene expression level, in contrast to aberrant function, is increasingly recognized as a major manifestation of human genetic variation. In addition, a potentially diabetogenic mutant of human TRPV1 modeled after a diabetogenic murine variant was shown to exhibit a unique capsaicin dose-response relationship.


Nonsynonymous polymorphisms in the human TRPV1 gene.

We screened SNPs in the TRPV1 gene appearing in the database at the NCBI website (http://www.ncbi.nlm.nih.gov/projects/SNP/). We chose to focus on six nonsynonymous SNPs in the human TRPV1 coding region (Fig. 1A). Two SNPs in exon 1 were located in the NH2 terminus of the protein: TRPV1K2N (SNP ID rs9894618) and TRPV1P91S(rs222749). TRPV1I315M (rs222747) affected exon 5 and localized to the region of the ankyrin repeat domains, which are postulated to mediate protein-protein interactions. TRPV1T469I (rs224534), TRPV1T505A(rs17633288), and TRPV1I585V (rs8065080), encoded by exons 8, 9, and 11, respectively, affect the transmembrane domain of the channel. TRPV1T469I is predicted to reside on an extracellular loop between membrane-spanning helices one and two, whereas TRPV1T505A affects an intracellular loop between membrane-spanning helices two and three. The TRPV1T505A site is adjacent to Ser-502, which undergoes regulatory phosphorylation by protein kinase C, protein kinase A, and calmodulin-dependent protein kinase II (reviewed in Ref. 82). A seventh nonsynonymous SNP, rs224496, resulted in a conservative substitution (TRPV1D421E) and exhibited extremely low minor allele frequency (= 0.00 in HapMap; http://www.hapmap.org/); functional studies with this SNP were not pursued. None of these SNPs map to the 60-residue NH2-terminal cassette absent from the human TRPV1b splice variant, and none (with the possible exception of TRPV1T505A) directly impact regions of the channel implicated in capsaicin binding or response (21, 2931, 38, 83).

Fig. 1.

Nonsynonymous polymorphisms in the human TRPV1 gene. A: diagram of the human TRPV1 protein secondary sequence depicting functional motifs, the location of transmembrane helices 1–6 (gray boxes), and corresponding exons in the human TRPV1 gene (staggered gray line). The site of the nonsynonymous human single-nucleotide polymorphism (SNP) is represented by the resultant amino acid substitution (dashed lines), and the corresponding SNP designation (“rs” format) is depicted above the cartoon. B: representative tracing of the time-dependent increment in fura 2 emission ratio (Em510; in the presence of alternating excitation at 340 and 380 nm) in HEK cells transiently transfected with a human TRPV1 cDNA mutagenized to incorporate a nonsynonymous SNP (TRPV1T505A, in this example), in response to various doses of capsaicin from 0 to 1,000 nM. Time of agonist application is denoted by the arrowhead. Error bars reflect SE for 12 wells at each capsaicin dose. A single experiment (i.e., n = 1) is depicted. C: graphic representation of the 4 parameters comprising the nonlinear regression fit to a sigmoidal dose-response curve (in this example, derived from the data in B). When the Hill slope was fixed at unity (i.e., no functional interaction among binding sites), a curve (black tracing) was generated with EC50 for capsaicin of 7.2 nM and a top of 0.67 (vertical and horizontal dashed lines, respectively, where shaded regions denote 95% confidence intervals). When the variable Hill slope model was applied, and although the calculated EC50 value remained nearly unchanged (i.e., 7.6 nM), the curve (pink tracing) was a better fit for the data. See text for further details. EC50 (with confidence interval in parentheses) is shown for each regression model. Data for 0 capsaicin were set at 2 logs below the lowest tested dose, in accordance with convention.

Nonsynonymous TRPV1 SNPs associated with diabetes development in the murine nonobese diabetic (NOD) model (58) were generated on the background of the human TRPV1 cDNA. The two residues affected in the NOD mouse TRPV1 gene are conserved between mouse and human; we designated the human equivalent hTRPV1P322A/D734E. Neither mutant residue affects the transmembrane/ion channel motif in TRPV1. For haplotype analysis, genotyping data reflecting 100 kb, 500 kb, and 1 Mb of chromosome 17 centered on the human TRPV1 gene were downloaded from the International HapMap Project (http://www.hapmap.org/) and analyzed within Haploview (http://www.broad.mit.edu/mpg/haploview/) (6). For the CEU population, data were converted to the D′ parameter for linkage disequilibrium for clarity in a grayscale depiction (i.e., see Fig. 8).

Fig. 2.

Calculated EC50 for capsaicin in HEK cells transiently transfected with wild-type human TRPV1 (TRPV1WT) or nonsynonymous polymorphic variants of TRPV1. EC50 (in nM capsaicin) was calculated using both the fixed [Hill slope (HS) = 1] and variable (HS = var) Hill slope models. Because of a potential concern regarding nonspecific effects at 1,000 nM capsaicin, data including and excluding this dose were reduced in parallel. Values are means ± SE. In terms of EC50, none of variants differed significantly from that of wild-type in either model, in the presence or absence of the capsaicin 1 μM data (n = 8 for TRPV1WT, 6 for TRPV1T505A, 7 for TRPV1K2N, 5 for TRPV1T469I, 5 for TRPV1P91A, and 5 for TRPV1I315M).

Fig. 3.

Calculated top parameter of the fit sigmoidal dose-response curve for capsaicin in HEK cells transiently transfected with wild-type human TRPV1 (TRPV1WT) or nonsynonymous polymorphic variants of TRPV1. The top parameter of the fit curve for each polymorphic variant was computed using models as in Fig. 2. The top parameter for the TRPV1I315M variant exceeded that of TRPV1WT in all models, whereas that of TRPV1P19S exceeded wild-type only when all data points were included. The false discovery rate procedure was applied for multiple comparisons (see methods). *Statistical significance with respect to TRPV1WT.

Fig. 4.

Effect of TRPV1 agonists on calcium signaling in vector- or TRPV1WT-transfected HEK cells. HEK cells were transfected with TRPV1WT or with empty vector and treated with the indicated agonists (100 nM capsaicin, 100 nM EGF, pH 5.4, or 10 μM anandamide) for ∼120 s. For pretreatment experiments (i.e., Forsk+Cap and PMA+Cap), cells received PMA (100 nM) or forskolin (10 nM) for 30 min before medium supplementation with capsaicin. Maximal increment in fura 2 ratio is depicted (n = 3 for each condition). Forskolin or PMA alone exerted no effect (data not shown). Forskolin significantly potentiated the effect of capsaicin, whereas PMA abrogated it.

Fig. 5.

Effect of anandamide treatment on TRPV1WT and selected TRPV1 SNPs exhibiting increased capsaicin responsiveness. HEK cells were transfected with empty vector control, TRPV1WT, or with either of the SNP-containing variants exhibiting increased responsiveness to capsaicin in Fig. 3 (TRPV1P91S or TRPV1I315M) and then treated with anandamide (10 μM) for ∼120 s. To minimize interexperimental variability in the magnitude of the anandamide response (n = 3), data were normalized to the anandamide effect on TRPV1WT. The effect of this agonist on both TRPV1P91S and TRPV1I315M exceeded that of wild-type in each of the 3 experiments and in the aggregated data shown. *Statistical significance with respect to TRPV1WT in the presence of anandamide treatment.

Fig. 6.

Expression at the protein level for TRPV1WT and SNP-containing variants. Expression of TRPV1 or variant TRPV1 was determined by anti-TRPV1 immunoblotting of either cell surface-expressed proteins [in avidin-affinity precipitates prepared from surface-biotinylated TRPV1 transient transfectants top blot] or of whole-cell lysates (bottom blot). Data from 4–5 experiments, depending on experimental condition, are quantified (bottom) and normalized to TRPV1WT. For statistical analysis, data are expressed as raw densitometry values (to avoid excluding the variability in TRPV1WT). Student's t-test was performed with correlated samples and the false discovery procedure for multiple comparisons. *,†: Statistical significance with respect to TRPV1WT in biotinylated avidin-affinity precipitates and in whole-cell lysates, respectively.

Fig. 7.

Expression at the mRNA level for TRPV1WT and SNP-containing variants. Expression of TRPV1 or variant TRPV1 was determined by RNase protection assay. The protected fragment for a representative experiment is depicted (blot) above quantified data from 4 independent experiments. When data were expressed as raw densitometry values and Student's t-test was performed with correlated samples, none of the polymorphic variants differed from TRPV1WT. When data were normalized within each experiment to TRPV1WT, abundance of TRPV1K2N and TRPV1I315M exceeded that of TRPV1WT by a statistically significant but a very modest degree (8 ± 4 and 17 ± 8%, respectively). *Statistical significance with respect to TRPV1WT.

Fig. 8.

Relation of nonsynonymous SNPs and haplotype blocks within the human TRPV1 gene. The region in the figure spans nucleotides 3,410,000–3,470,000 of human chromosome 17; the human TRPV1 gene (so designated, gray line) is punctuated by 17 exons (vertical black bars), a subset of which gives rise to the TRPV1 coding sequence (CDS; open bar). Exons to which the TRPV1 nonsynonymous variants map are depicted. Haplotype blocks deduced from the HapMap data sets for Utah residents with ancestry from Northern and Western Europe (CEU), Han Chinese in Beijing (CHB), Japanese in Tokyo (JPT), and Yoruba in Ibadan, Nigeria (YRI) are shown above the gene map; some blocks extend beyond the chromosomal region depicted (blocks ending in same-colored arrowheads). It can be inferred that in the CEU (Caucasian) population, TRPV1I315M and TRPV1T469I reside within the same haplotype block and are likely to be coinherited (see text). A plot of linkage disequilibrium (LD; expressed here as D′) over this interval was generated using the HapMap data set for the CEU population. High linkage disequilibrium is evident throughout the potential regulatory regions “upstream” of the TRPV1 gene and over smaller intervals consistent with the haplotype block structure shown above.

Cell culture and transient transfection.

HEK293 cells were maintained in DMEM/F12 medium (JRH) supplemented with 10% fetal bovine serum (JRH). All reagents were obtained from Sigma (St. Louis, MO) unless otherwise specified. Agonists and inhibitors were applied as follows: 1–1,000 nM capsaicin, 100 nM epidermal growth factor; 10 nM forskolin, 100 nM phorbol 12-myristate 13-acetate, and 100 nM anandamide. Pretreatment with these agents, where indicated, was applied ∼10 min before capsaicin treatment and remained present for the duration of treatment. Cells were transiently transfected with Lipofectamine PLUS (Life Technologies) in accordance with the manufacturer's directions using 15 μl of PLUS reagent, 30 μl of Lipofectamine, and 6 μg of plasmid DNA reagent/100-mm dish of cells. All transfections were performed in parallel into HEK drawn from the same pool of trypsinized cells.

Determination of intracellular calcium.

For intracellular calcium assays, cells were passaged on the second day after transfection at 1:1 into poly-d-lysine-coated black-wall, flat-bottom 96-well plates (BioCoat; Becton-Dickinson Labware) at 3–5 × 104 cells/well. The following day, cells were loaded with fura 2 [fura 2-AM, 2 μM, plus 0.01% Pluronic-127 in 130 mM NaCl, 4.7 mM KCl, 1.26 mM CaCl2, 1.18 mM MgSO4, 5.6 mM glucose, 20 mM HEPES (pH 7.5); HBSS] for 20 min at 37°C. Cells were washed twice with HBSS and placed in 100 μl/well of HBSS before the assay. Compounds were added in 100-μl volume after baseline readings were established using a FlexStation II (Molecular Devices). Data were collected using an emission wavelength of 510 nm and alternating excitation at 340 and 380 nm at 3.9-s intervals for a total of 2 min. Twelve replicates were performed for each capsaicin dose (1 “row” on a 96-well plate), and eight doses were tested simultaneously (0, 1, 3, 10, 30, 100, 300, and 1,000 nM); this was considered a single experiment for a single TRPV1 variant. A total of five to eight experiments were performed for each variant. Raw data were exported and reduced in Excel (Microsoft) and graphed using Prism software (GraphPad Software). For each concentration of capsaicin, the response (Δfura 2 ratio) was calculated as (rmax − rinitial), where rinitial represents the mean of the first two data points, and rmax is the maximal fura 2 ratio achieved after treatment. The initial downward deflection in the fura 2 ratio (e.g., in Fig. 1B) is an artifact of the agonist injection. A logarithmic dose-response relationship was plotted for each experiment with each TRPV1 variant, and a sigmoidal four-parameter (top, bottom, EC50, and Hill slope) Hill equation dose-response curve was fit using the nonlinear regression feature of Prism (GraphPad Software). Curves were fit with both a fixed Hill slope of 1 and with a variable Hill slope, and EC50 for capsaicin was reported (47).

Cell-surface biotinylation and immunoblotting.

Cell-surface biotinylation was performed 48 h after transient transfection. HEK293 monolayers were washed three times with ice-cold PBS, incubated with 0.5 mg/ml Sulfo-NHS-LC-Biotin (Pierce Biotechnology, Rockford, IL) for 30 min at 4°C, quenched by incubation with 100 mM glycine in ice-cold PBS for 30 min at 4°C, and then washed three times with ice-cold PBS. Monolayers were lysed in lysis buffer [125 mM NaCl, 50 mM Tris (pH 7.5), 0.1% SDS, 0.5% sodium deoxycholate, 1% Nonidet P-40, 1 mM sodium orthovanadate, 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 1 μg/ml leupeptin, 1 μg/ml aprotinin, 1 μg/ml pepstatin A, 25 mM β-glycerophosphate, 2 mM sodium pyrophosphate] for 30 min at 4°C. The protein concentrations were determined by the Bradford method (Bio-Rad). Twenty micrograms of whole-cell lysate was set aside for blotting in parallel. To the remaining lysates, ImmunoPure Streptavidin Beads (40 μl; Pierce Biotechnology) were added to ∼3 mg of biotinylated protein and the mixture was incubated at 4°C for 4 h. The beads were washed five times with ice-cold PBS and eluted with 1× SDS sample buffer. The eluted proteins were resolved via denaturing SDS/PAGE, transferred to a nylon membrane and blotted with anti-TRPV1 antibody (catalog no. P-19; Santa Cruz Biotechnology) and horseradish peroxidase-conjugated donkey anti-goat IgG secondary antibody. Visualization was via Chemiluminescence Plus reagent (PerkinElmer Life Science).

RNase protection assay.

For RNase protection assay, cells were transfected as above (with 6 μg RNA/transfection reaction) and RNA was harvested 48 h later via TRIzol in accordance with the manufacturer's directions (Life Technologies). Solution hybridization was performed with biotinylated antisense riboprobe directed against human TRPV1 (PCR-amplified from full-length human TRPV1 cDNA using the following primer pair: hTRPV1-2261-5′-tcaacaagatcgcacaggag and hTRPV1-2556-3′-gcctgaaactctgcttgacc, where numbers refer to position in NM_080705) and vector backbone [PCR-amplified from plasmid DNA using the following primer pairs: pcDNA3.1-V5HisTOPO-4779-5′ ttgccgggaagctagagtaa and pcDNA3.1-V5HisTOPO-5003-3′ gataacactgcggccaactt; pcDNA3.1-V5HisTOPO-1437-5′ agcgtgaccgctacacttg and pcDNA3.1-V5HisTOPO-1656-3′ aagggcgaaaaaccgtctat, where numbers refer to nucleotide position in pcDNA3.1-V5/HisTOPO as reported by the manufacturer (Invitrogen)]. Detection was achieved via autoradiography.

Image processing and statistical analysis.

For quantitation of autoradiograms, exposed films were scanned (Canon LiDE80) and data were reduced using ImageJ (http://rsb.info.nih.gov/ij/; National Institutes of Health) and Excel (Microsoft). For all depicted scans of enhanced chemiluminescence exposures of immunoblots and RNase protection assays, contrast was improved by decreasing the maximum input level from 255 to ∼175 (Adobe PhotoShop CS) to mimic the true appearance of the exposed film. For each figure, all depicted lanes are from the same exposure of the same autoradiogram and have been treated identically. All experiments were performed a minimum of three times (see figure legends). Data are expressed as means ± SE (Excel, Microsoft). Statistical significance was ascribed using Student's t-test [for correlated samples using raw data, or for independent samples using normalized data (VassarStats; http://faculty.vassar.edu/lowry/VassarStats.html)]. Where multiple comparisons were performed, the false discovery rate procedure was used (17, 18).


Nonsynonymous polymorphisms in the human TRPV1 gene.

We sought nonsynonymous polymorphisms in the human TRPV1 gene using a public domain SNP database (i.e., dbSNP; www.ncbi.nlm.nih.gov/projects/SNP/) and publication references. We identified five nonsynonymous polymorphisms expected to result in nonconservative amino acid substitution; these are diagrammed in Fig. 1A. A sixth site, TRPV1I585V, results in a conservative substitution and was not the subject of initial interest; however, a recent publication addressed potential functional significance of this SNP (32), and it was later included in our analysis. Of note, an additional nonsynonymous SNP designated rs17856604 similarly resulted in a conservative amino acid substitution (TRPV1D421E) and was not investigated; however, it is no longer indexed by this designation. Two of these SNPs affect the intracellular NH2 terminus of TRPV1: one is in the ankyrin repeat-containing region, and three are in the transmembrane domain. The ankyrin repeat domain is predicted to play a role in mediating protein-protein interactions and perhaps homotetramerization of the channel. TRPV1T469I is expressed on a predicted extracellular loop between membrane-spanning helices one and two, and TRPV1I585V is predicted to reside within membrane-spanning helix five. TRPV1T505A is expressed on an intracellular loop between helices two and three. This general region of TRPV1 confers responsiveness to capsaicin (29). The minor allele frequencies for each of these polymorphisms in the HapMap populations (where data were available) are shown in Table 1.

View this table:
Table 1.

Alleles of the human TRPV1 gene investigated

TRPV1WT and variant forms exhibit similar EC50 for capsaicin.

HEK293 cells were transiently transfected with a wild-type human TRPV1 cDNA, or one that had undergone site-directed mutagenesis to introduce a polymorphism. In transfected cells loaded with fura 2-AM, capsaicin produced a prompt and dose-dependent increment in intracellular calcium (Fig. 1B). Sham treatment (vehicle; 0 nM capsaicin) elicited a negligible effect. [There was no effect of capsaicin in vector-transfected or untransfected cells, and the capsaicin effect was completely dependent on the presence of calcium in the medium (not shown).] The magnitude of the capsaicin response, the increment in fura 2 ratio (Δfura 2 ratio), was quantified for each capsaicin dose as (rmax − rinitial), where rinitial represents the mean of the first two data points, and rmax was the maximal ratio achieved. A dose-response relationship was plotted for each TRPV1 variant in each experiment, and a sigmoidal dose-response curve was fit (see methods) (Fig. 2C). Curves were fit with both a fixed Hill slope of 1 and with a variable Hill slope (generally <1), and EC50 for capsaicin was computed. In many experiments with TRPV1WT or TRPV1 variants, the data for 1 μM capsaicin deviated substantially from the fit curve; this suggested either a nonspecific effect at this high concentration or the superimposition of an additional mechanism of action. Therefore, curves were fit separately with the data from 1 μM capsaicin either included or excluded. We found no statistically significant differences between the EC50 for capsaicin for TRPV1WT and any of the variant forms (Fig. 2). In most cases, the EC50 approximated 10 nM. An effect of 1 nM capsaicin was readily detectable in our assay (i.e., Fig. 1C and data not shown); however, there was no effect of capsaicin at 0.1 nM (n = 2; data not shown). This was lower than was observed by Smart et al. (69) at standard pH (7.4) using a similar platform; this group also reported Hill coefficients in excess of 2 for capsaicin under similar conditions. Data from these authors suggested nonspecific effects of capsaicin at concentrations in excess of 1 μM (69).

Unlike EC50, the curve top parameter varied significantly among the expressed TRPV1 polymorphic variants (Fig. 3). Irrespective of constraint of the Hill slope to unity, or inclusion of the capsaicin 1 μM data, the TRPV1I315M variant exhibited a more robust maximal effect of capsaicin. Similar findings were noted for the TRPV1P91S variant, but only when all data points were included.

Cells transfected with TRPV1P91S and TRPV1I315M exhibit enhanced responsiveness to anandamide.

We sought a second TRPV1 agonist to confirm this relationship. We tested vector- and TRPV1-transfected HEK293 cells with a panel of known TRPV1 agonists. EGF was used as an activator of phospholipase Cγ to alleviate the tonic phosphoinositol-4,5-bisphosphate-dependent inhibition of TRPV1 channel function (55). The activator of protein kinase A, forskolin, was predicted to potentiate the capsaicin response (8, 20, 44, 57), as was the activator of protein kinase C, PMA (3, 41, 54). In our model, forskolin modestly potentiated the effect of capsaicin (Fig. 4). Unexpectedly, pretreatment with the activator of protein kinase C, PMA, inhibited the capsaicin response. This may be a consequence of our longer PMA pretreatment interval (i.e., 10 min), or the use of confluent monolayers; an analogous context-dependent bidirectional regulatory effect on TRPV1 has recently been shown for phosphatidylinositol (4,5)-bisphosphate (39), reconciling prior conflicting observations (55, 72). EGF, low pH (5.4), and anandamide, the endogenous cannabinoid, all modestly activated calcium transients in TRPV1-transfected cells. Of note, the effect of EGF was also observed in the vector-transfected control cells and was therefore independent of TRPV1; the pH and anandamide effects, in contrast, required TRPV1 expression. The effect of anandamide on TRPV1P91S and TRPV1I315M was then compared with TRPV1WT. The anandamide effect was more robust in the TRPV1P91S and TRPV1I315M variants than in TRPV1WT (Fig. 5). Vehicle treatment alone produced no response in vector or TRPV1- or variant TRPV1-transfected cells.

TRPV1P91S and TRPV1I315M proteins are expressed at higher levels following transfection.

Based on the more robust response of TRPV1I315M and perhaps TRPV1P91S to capsaicin in the absence of an effect on EC50 for the agonist, and based on the differing response to anandamide, we hypothesized that channel abundance might account for the differences among variants. We quantified total expression of each variant following transient transfection and also determined the relative expression at the plasma membrane using cell-surface biotinylation. Relative to TRPV1WT, the variants TRPV1I315M and TRPV1P91S were expressed at higher levels in total lysates (by 82 ± 25 and 89 ± 18%, respectively) and in avidin-affinity precipitates derived from biotinylated transfectants (by 110 ± 40 and 140 ± 40%, respectively) (Fig. 6).

TRPV1I315M exhibits greater expression at the mRNA level following transfection.

Because nonsynonymous SNPs can influence mRNA transcription and stability in vivo, we sought to determine whether a component of the increased expression of TRPV1I315M variant could be accounted for by increased expression at the mRNA level. HEK293 cells were transiently transfected with TRPV1WT or variant forms as in our other studies. RNA was harvested and subjected to RNase protection assay with a labeled antisense riboprobe directed against human TRPV1. Expression of the TRPV1I315M and TRPV1K2N variants exceeded that of control by a modest degree (i.e., by 17 and 8%, respectively) (Fig. 7).

TRPV1 haplotype structure can inform investigations into functional consequences of SNPs.

With an increasing understanding of the haplotype structure of the human genome comes an appreciation of the interdependent nature of SNPs. SNPs are likely to be coinherited along with adjacent SNPs that reside within the same haplotype block; that is, these often lengthy stretches of genomic sequence are only infrequently subject to recombination. Therefore, we sought to determine which TRPV1 polymorphisms were likely to operate in concert on expression of TRPV1. Figure 8 depicts the haplotype blocks for each of the four diverse human populations represented in the HapMap data set, projected on to the genomic structure of human TRPV1. In the Caucasian study population (CEU), the TRPV1I315M and TRPV1T469I polymorphisms may cosegregate; they map to the same haplotype block. In this population, the minor allele frequencies for TRPV1I315M and TRPV1T469I are similar at 0.18 and 0.27, respectively (see Table 1). Also, the linkage disequilibrium between these two SNPs in this population is high (D′ = 0.87, where D′ is an index of genetic linkage, spanning 0 to 1, that is normalized to allele frequency). For the haplotype block including these SNPs, five relatively abundant haplotypes constitute 97.5% of the all possible genomic variation at this locus (data not shown). The TRPV1I315M and TRPV1T469I SNPs coexist in only one of these five possible blocks, which accounted for 15.8% of sequenced alleles from the CEU (Caucasian) HapMap population. Therefore, we tested the effect on TRPV1 function of a combination of these two polymorphisms. We found that the EC50 of the TRPV1I315M and TRPV1T469I variants, and of TRPV1WT, did not differ from that of TRPV1I315M/T469I, the double-mutant (n = 3; data not shown). In addition, although we again found that the top of the fit curve for TRPV1I315M exceeded that of TRPV1WT (n = 3; P ≃ 0.01; data not shown), the effect of the double-mutant TRPV1I315M/T469I was intermediate and not statistically different from either (n = 3; data not shown).

In similar fashion, although not residing on the same haplotype block, the TRPV1T469I SNP exhibits relatively strong linkage disequilibrium with a sixth nonsynonymous polymorphism, TRPV1I585V, among the Han Chinese population in the HapMap data set. Owing to the very conservative nature of this latter polymorphism (i.e., isoleucine to valine), we initially did not select it for detailed investigation; however, when haplotype analysis indicated that it was potentially coinherited with the TRPV1T469I SNP in a Chinese population (linkage disequilibrium, D′, was 0.77), we tested the combined effect of these two SNPs on TRPV1 function. (Interestingly, the “intervening” nonsynonymous SNP on the diagram, corresponding to TRPV1T505A, exhibited a minor allele frequency of 0 in all non-CEU HapMap study populations; therefore, this SNP was not studied in combination with either TRPV1T469I or TRPV1I585V). Of note, although there is generally a high degree of concordance between the Japanese and Han Chinese allele frequency in the HapMap data set, there was only very limited linkage disequilibrium (D′ = 0.21) between the TRPV1T469I and TRPV1I585V SNPs in the Japanese population. With respect to the TRPV1I585V SNP, the disparity in minor allele frequency among the diverse HapMap study populations is so great that the minor allele in the Caucasian (CEU) population (i.e., Val585) actually represents the major allele in the Han Chinese study population (see Table 1). However, TRPV1T469I, TRPV1I585V, and TRPV1T469I/I585V did not differ from TRPV1WT in terms of either EC50 for capsaicin or other fit parameters of the capsaicin sigmoidal dose-response curve (n = 3; data not shown).

TRPV1 polymorphisms potentially conferring type 1 diabetes in a murine model markedly impact TRPV1 function.

A pair of nonsynonymous SNPs in the murine TRPV1 gene (giving rise to murine TRPV1P322A/D734E) was recently described in the NOD mouse model of type 1 diabetes, where TRPV1 may represent the diabetes risk gene in this naturally arising mutant (58). Because these polymorphisms affect amino acids that are conserved across all mammalian and avian species investigated, we tested their effect in the context of the human TRPV1 cDNA. Interestingly, when four-parameter sigmoidal dose-response curves were generated for hTRPV1P322A/D734E and hTRPV1WT, the EC50 for capsaicin was not significantly affected (14.3 ± 2.1 vs. 10.9 ± 0.4; P = 0.09); the top and bottom of the fit curves were also not impacted (Fig. 9A). However, the Hill slope (or Hill coefficient), the steepness of the dose-response curve, was dramatically impacted by the double mutation and was nearly twice that of the wild-type (1.3 ± 0.1 vs. 0.7 ± 0.4; P = 0.005) (Fig. 9B). Thus a NOD-like human TRPV1 variant was hyporesponsive to capsaicin at low doses (below the EC50) and hyperresponsive at higher doses (Fig. 9B). To put this anomaly into perspective, the Hill slope for the capsaicin dose-response was unaffected by incorporation of any of the human polymorphisms we had investigated (Fig. 9C), and in none of these individual assays (n = 42 separate experiments) did the Hill slope exceed unity (individual data not shown).

Fig. 9.

Effect on TRPV1 function of 2 linked nonsynonymous polymorphisms potentially causal for a type 1 diabetes-like syndrome in the nonobese diabetic (NOD) mouse model. A recent report identified a mutation/polymorphism in each of 2 highly conserved amino acid residues in the murine TRPV1 gene in the NOD mouse model (58); neurons derived from dorsal root ganglia of mice of this strain were hyporesponsive to capsaicin. The 2 polymorphisms were engineered into the human TRPV1 cDNA, generating TRPV1P322A/D734E. A: results of 4-parameter curve fit for a nonlinear regression (with unconstrained Hill slope) of the capsaicin dose-response for HEK cells transfected with either TRPV1WT or the polymorphic variant, TRPV1P322A/D734E (n = 3 separate experiments). EC50 likely did not differ (14 ± 2 vs. 11 ± 0.4; P = 0.09); however, the Hill slope for the variant was dramatically elevated relative to wild-type (1.3 ± 0.1 vs. 0.7 ± 0.4; P = 0.005). Note that some parameters were scaled by a factor of 10 (top, Hill slope) or 100 (bottom) to share the same y-axis. *Statistical significance with respect to TRPV1WT. B: representative capsaicin dose-response curves for HEK cells transfected with TRPV1WT or with the polymorphic variant, TRPV1P322A/D734E, showing the “steeper” nature of the dose-response relationship in the TRPV1P322A/D734E-transfected cells. C: calculated Hill slope values for all capsaicin dose-response experiments (excluding those depicted in A) with TRPV1WT and with the human SNP-containing variants for comparison; unlike TRPV1P332A/D734E, all are <<1, and no significant differences were seen.


We sought SNPs that would impact function of TRPV1 in vitro and, potentially, in vivo. Monogenic human diseases associated with point mutations have generally resulted in functional disruption or aberrant trafficking of the gene product, and we initially considered nonsynonymous SNPs as an extension of this paradigm. Clinically significant human polymorphisms, however, need not be defined so narrowly. Accruing evidence suggests that genetic polymorphisms may exert major effects simply by influencing the level of expression of the gene product, independent of its function. To this end, polymorphisms may alter transcription, mRNA stability, or protein half-life. Copy-number polymorphisms, where the number of alleles varies among individuals, affect at least 12% of the genome (59) and account for nearly one-fifth of all detected genetic variations (74); in such cases, the disease state is mediated through altered expression of a normal protein (e.g., Ref. 4). Individual variation in gene expression has long been recognized, and the expression level of individual gene products is hereditary (15, 46, 65, 71, 84). Therefore, it is reasonable to conclude that much of intersubject phenotypic variation will be encoded by polymorphisms that influence levels of gene expression. We postulate that TRPVI315M and perhaps TRPV1P91S are examples of this type. These two polymorphisms resulted in markedly increased abundance of the variant TRPV1 protein at the level of whole-cell expression, and at the level of expression at the cell surface, in our heterologous expression model. Although there was also a very modest increment in TRPV1I315M mRNA level in this model (∼17%), it is doubtful that it is sufficient to account for the marked change in protein expression (>100%). Definitive demonstration will require independent testing of allele-specific expression in human subjects. Thus far, there are no data examining the level of TRPV1 expression in any human tissue as a function of allele although this would be of interest. TRPV1 was not among the genes exhibiting differential expression in lymphoblastoid cells derived from subjects of various ethnicities (71); however, it is unclear whether TRPV1 is expressed in this tissue.

Because SNPs are not inherited in isolation, haplotype structure can be used to further inform investigations into the functional consequences of polymorphic variants. Using utilities and data in the public domain (www.hapmap.org), we show that the TRPV1I315M and TRPV1T469I alleles reside on the same ∼7-kb haplotype block in Utah residents with ancestry from Northern and Western Europe (i.e., HapMap CEU population). In genomic DNA, they are separated by ∼6.5 kb, exhibit strong linkage disequilibrium, and are likely coinherited in this population. Although the allelic frequencies of both of these SNP variants are much higher in Asian populations [e.g., Japanese in Tokyo, Japan (HapMap JPT population) and Han Chinese in Beijing, China (HapMap CHB population)], they are not predicted to reside on the same haplotype blocks in these populations and are therefore much less likely to be coinherited. When both SNPs were introduced into human TRPV1, giving rise to the TRPV1T469I/I315M variant, response of the channel to capsaicin was indistinguishable from TRPV1WT or from either single mutant (data not shown).

A second SNP combination was also tested. In an earlier analysis using HapMap-supplied data (c. 2006), the rs224534 (TRPV1T469I) and rs8065080 (TRPV1I585V) polymorphisms were predicted to reside on the same haplotype block in the Japanese HapMap population (data not shown); however, more recent and comprehensive analysis did not support this view (i.e., Fig. 8). Nonetheless, there was a high degree of linkage disequilibrium between the TRPV1T469I and TRPV1I585V SNPs in the Japanese population (D′ = 0.77), so these were tested in combination. (Of note, the intervening nonsynonymous SNP, giving rise to TRPV1T505A, is extremely uncommon and was not tested in combination.) In terms of capsaicin responsiveness, we found no novel properties attributable to the combination of both polymorphisms. Interestingly, the major alleles of these polymorphisms, TRPV1T469I and TRPV1I585V, in the HapMap Caucasian population (CEU) are actually the minor alleles in the Chinese and Japanese populations. That is, the tested TRPV1T469I/I585V variant is predicted to represent the most common genotype in both Asian populations.

Two mutations or polymorphisms in the murine TRPV1 gene were linked to the development of the diabetic phenotype in the NOD mouse model of type 1 diabetes (58). There is remarkable conservation between the human and rodent TRPV1 at the amino acid level (i.e., 87% identity, mouse NP_001001445 vs. human NP_542436, with most sequence divergence residing at the extreme NH2 and COOH termini of the 839-residue proteins), and the two affected residues are conserved across all mammalian species investigated (58). Razavi et al. (58) tested intact NOD mice and their tissues for capsaicin sensitivity. The NOD mice were less responsive than control mice to foot pad injection of an irritating dose of capsaicin, based on observed biting and licking of the affected extremity (58). Neurons of the dorsal root ganglia isolated from NOD mice were also less responsive to capsaicin than were corresponding neurons isolated from the control mice; however, these neurons retained an intact response to KCl, suggesting specificity of the capsaicin effect (58). We tested the effect of a human variant corresponding to the NOD mutant by generating hTRPV1P322A/D734E. Although three of four parameters defining the sigmoidal dose-response relationship for capsaicin were unaffected (including EC50 for capsaicin), there was a striking alteration in the “steepness” of this dose-response; the Hill slope for the “diabetic” variant was nearly twice that of wild-type TRPV1 (Fig. 9B) or any of the other tested polymorphic variants (Fig. 9C). This relationship conferred diminished capsaicin sensitivity at low doses (i.e., 1 and 3 nM capsaicin) and an augmented capsaicin response above the EC50. In contrast, Razavi et al. (58) described reduced capsaicin responsiveness of the NOD dorsal root ganglion neurons across the full dose range tested (i.e., 500 nM-100 μM capsaicin). This discrepancy may reflect differences in cell type (dorsal root ganglion neurons vs. HEK cells), model system (native vs. heterologous expression), assay design (single-cell vs. monolayer), or clone background (human vs. murine). Of note, neither of the mutated residues is predicted to impact capsaicin binding or response based on its location (21, 2931, 38, 83). Thus far, polymorphisms in these residues have not been described in human populations; however, they might have substantial implications for the phenotype of pain sensitivity or neurogenic inflammation.

The relationship between TRPV1 and diabetes warrants comment. Razavi and colleagues (58) further showed that TRPV1−/− mice exhibit increased insulin sensitivity relative to TRPV1+/+ mice and that genetic reconstitution of wild-type TRPV1 in the NOD mouse model restored insulin sensitivity. Gram et al. (23) made a complementary observation in a second diabetic model; this group observed that pharmacological destruction of capsaicin-sensitive (TRPV1-positive) sensory afferent nerve fibers blocked development of the type 2 diabetic-like state in the Zucker diabetic fatty rat. Whereas TRPV1 may impact diabetes, other studies suggest the converse: that the diabetic state may alter the morphology and/or physiology of TRPV1-containing sensory afferents (e.g., Refs. 19, 25, 52, 56).

A role for TRPV1 in the predisposition to human diabetes has not yet been demonstrated. In five recent large-scale genome-wide association studies, no polymorphisms in the human TRPV1 gene, or in adjacent genes on human chromosome 17, were linked to type 2 diabetes (Refs. 64, 66, 68, 73, 87 and online supporting materials); the presence of a modest effect or a subset-specific effect cannot be excluded, however, nor can a role for TRPV1 in type 1 diabetes.

Little has been written about physiological effects of polymorphisms in the human TRPV1 gene. Kim and colleagues (32) observed that a number of variables influence sensitivity to experimentally-induced pain. American female subjects of European descent with the TRPV1I585V allele exhibited shorter cold withdrawal times (i.e., increased sensitivity to cold-induced pain) than did those with the wild-type allele (32). For these studies, subjects were instructed to submerse their hands in ice water up to the wrists with repeated clenching and unclenching until the pain reached an “unbearable level” (maximum 180 s). There were no allele-specific differences in the subjects' ability to tolerate elevated temperatures in similarly designed studies. We initially did not investigate this polymorphism in detail because of the highly conservative nature of the amino acid substitution and because data from Hayes and colleagues (24) previously suggested that the channel encoded by the TRPV1I585V allele exhibited a normal functional response to agonists in vitro. We did, however, incorporate it into our investigation of the linked SNP, TRPV1T469I.

Both the TRPV1I315M and TRV1T469I alleles are overrepresented in South Asian populations (i.e., Han Chinese and Japanese subjects from Tokyo in the HapMap cohort), and TRPV1I315M is associated with increased expression of the channel at both the mRNA and protein levels in the present study. TRPV1 likely plays a major role in integrating nociceptive signals, particularly in the contexts of inflammatory and neuropathic pain (reviewed in Refs. 13 and 78). A body of literature underscores racial differences in the perception and/or response to painful stimuli (reviewed in Ref. 50); for example, two clinical studies found that Asian patients may require less postoperative analgesia than European patients (26, 75). This phenomenon may be attributed to purely psychological or cultural factors and has not been replicated in all studies (86); nonetheless, it raises the intriguing possibility that genetic polymorphisms, such as those investigated here, may account for or contribute to a racial difference in pain threshold or tolerance.


This work was supported by the American Heart Association, the Department of Veterans Affairs, and the National Institute of Diabetes and Digestive and Kidney Diseases.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract