Inducible desensitization to capsaicin with repeated low-dose exposure in human volunteers

Abstract

Responses to capsaicin are reduced following repeated exposure, a phenomenon known as capsaicin desensitization. Heavy consumers of chilies consistently report reduced oral burn relative to infrequent consumers, presumably due to chronic desensitization. However, the mechanism(s) underlying capsaicin desensitization remain poorly understood. We hypothesized that reduced response to capsaicin due to repeated oral exposure may result from a change in the expression of the capsaicin receptor (TRPV1) gene. To test this, we conducted two longitudinal desensitization studies in healthy human volunteers. In Study 1, 51 adults completed a 17-day capsaicin desensitization protocol. The study consisted of three in-person visits where they were asked to sample stimuli, including 3, 6, and 9 ppm capsaicin, and rate intensity on a general labeled magnitude scale (gLMS). Between days 3 & 17, participants rinsed at home with 6 ppm capsaicin (n=31) or a control (n=20) solution (20 uM sucrose octaccetate; SOA) twice a day. Before and after the oral exposure protocol, a clinician collected fungiform papillae. Participants randomized to the capsaicin rinse showed a statistically significant reduction in oral burn ratings that was not observed in controls, indicating repeated low-dose exposure can systematically induce desensitization. TRPV1 expression was not associated with reported capsaicin burn, and there was no evidence of a decrease in TRPV1 expression following capsaicin exposure. In Study 2, participants (n=45) rinsed with 6 ppm capsaicin in a similar protocol, rating capsaicin, vanillyl butyl ether (VBE), cinnamaldehyde, ethanol, menthol, and sucrose on days 1, 3, & 17. Burn from capsaicin, VBE, cinnamaldehyde, and ethanol all showed a statistically significant change – capsaicin, VBE and cinnamaldehyde burn all dropped ~20%, and a larger reduction was seen for ethanol – while menthol cooling and sucrose sweetness did not change. Collectively, this suggests reductions in oral burn following chronic capsaicin exposure generalizes to other stimuli (i.e., cross desensitization) and this cannot be explained by a change in TRPV1 mRNA expression. More work is needed to elucidate the underlying mechanism for capsaicin desensitization in the oral cavity.

1. Introduction

Perceptual responses to capsaicin are reduced following repeated exposure, both orally [1–3] and extra-orally [4, 5]. This phenomenon has also been leveraged by treatments that can be used to manage pain [6–8]. Diminished burn from capsaicin is also evident in free-living individuals who frequently consume chili peppers [9–12]. Such observations have been replicated in controlled laboratory trials, where capsaicin exposure leads to decreased response to suprathreshold concentrations of capsaicin [1, 13–15]. However, the specific mechanism(s) behind the reduction in the burn from capsaicin following repeat exposure are less understood. Psychophysicists have classically described this phenomenon as capsaicin desensitization. However, it may be more accurate to describe diminished psychophysical responses to capsaicin as hypoalgesia, given that desensitization is more narrowly defined in other fields, implying certain physiological or pharmacological mechanisms. That said, we still use the conventional phrasing here to be consistent with prior literature, with a caveat that this usage is not meant to imply a specific mechanism of action.

While mechanism(s) responsible for capsaicin desensitization are not fully understood (e.g. [16–19]), evidence suggests the TRPV1 (transient receptor potential cation channel subfamily V member −1 ) receptor [20] is implicated. This receptor is a ligand-gated ion channel that allows passage of calcium and sodium into the cytoplasm. It opens when exposed to noxious heat (~43C) and it is sensitized by endogenous and exogenous ligands [21–23]. This homotetrameric channel consists of six transmembrane domains for each subunit, with a hydrophobic loop between domains 5 and 6 (see[24] for more detail). TRPV1-sensitive neurons can become desensitized or inactive following agonist binding, lasting only seconds or up to days or weeks. This short-term inactivity is thought to be regulated by intracellular Ca²⁺[18, 25]. Conversely, long-lasting desensitization (also known as channel defunctionalization) is independent of cellular calcium concentrations and may be due to neuronal defunctionalization (see [26]). It has been hypothesized that the influx of calcium and sodium leads to apoptosis and degradation of epidermal nerve fibers [18, 25, 27, 28]. Both of these stages of desensitization are reversible; however, at very high doses, irreversible neuronal toxicity has been observed in animal models, resulting in permanent nerve damage [25, 29, 30]. For more information regarding the pharmacological effects of capsaicin, see [27].

Here, we investigated one mechanism that might potentially explain how repeated low dose exposure to oral capsaicin reduces the burn from capsaicin. We hypothesized that TRPV1 mRNA expression in the oral cavity may be downregulated as a response to chronic exposure, with subsequent impact on the perception of oral burn. Fungiform papillae (FP) contain specialized taste receptor cells innervated by the chorda tympani nerve as well as free nerve endings of the trigeminal nerve, which express TRPV1. Critically, FP regrow when biopsied, meaning that it is possible to harvest these papillae in human volunteers. Previously, Lipchock and colleagues [31] showed that participants with greater mRNA expression of a functional haplotype for the bitter taste receptor TAS2R38 in FP reported more bitterness for 6-n-propylthiouracil compared to those with lower expression. This report inspired us to perform a cross-sectional pilot study where we biopsied human FP, quantified TRPV1 mRNA expression, and related it to ratings of capsaicin burn in the laboratory and self-reported intake of spicy foods. These cross-sectional data (Nolden and Hayes, unpublished) suggested there might be a relationship between TRPV1 mRNA expression and perceived burn, with greater expression associated with greater burn from sampled capsaicin; however, self-reported intake of chili peppers was not associated with TRPV1 expression. Given these conflicting results, we revisited this hypothesis in a longitudinal randomized controlled trial.

In this report, we test (a) if repeated systematic exposure to low doses of capsaicin over two weeks causes reduced response to sampled capsaicin in free living adult volunteers, and (b) whether induction of such desensitization corresponds with decreased TRPV1 mRNA expression in fungiform papillae. We also test (c) the specificity of altered responses to other stimuli (i.e., cross desensitization). We chose to collect FP tissue to quantify mRNA levels for TRPV1 as it is minimally invasive and low risk for participants, and because TRPV1-positive trigeminal neurons innervate FP [32, 33]. If successful, our data would highlight one mechanism by which capsaicin desensitization occurs in humans.

2. Materials and Methods

2.1. Overview

We performed two separate studies in non-overlapping cohorts of adult volunteers. In each, participants visited the Penn State Clinical Research Center (Study 1) or the Penn State Sensory Evaluation Center (Study 2) three times over the course of 17 days. In the first visit (Day 1; typically a Monday or Tuesday), a researcher provided participants with a study overview, and written consent was obtained, followed by psychophysical ratings of various stimuli. Participants returned for a second visit (Day 3; typically, a Wednesday or Thursday), where they made psychophysical ratings of the same stimuli before being given a 14-day supply of ‘mouthwash’ to use at home. In the final visit (Day 17; typically, a Wednesday or Thursday), they made psychophysical ratings of the same stimuli. For Study 1 only, biopsies were performed on the second and third visits after stimuli were rated. All procedures were approved by the Penn State IRB; the study complies with the Declaration of Helsinki for Medical Research involving Human Subjects and all local regulations. All participants were given a token cash incentive for their time. Data were collected in 2016 or 2017 (i.e., before the COVID-19 pandemic), so no special precautions were needed to minimize SARS-CoV-2 exposure. All psychophysical data were collected using Compusense Cloud (Guelph ONT).

2.2. Participants

Individuals were pre-screened and recruited from the Penn State campus and surrounding community. Potentially interested individuals completed a brief online questionnaire before enrollment to see if they met the recruitment criteria. These included: between the ages of 18-45, Caucasian/White ethnicity, not pregnant or breastfeeding, non-smoker, no tongue, cheek or lip piercings, no difficulty swallowing or history of choking, no known problems of taste or smell, not taking prescription pain medication, not taking blood thinners, no known history of oral cancer or disease, no hyperactive thyroid, no history of chronic pain, and willing to taste solutions containing ethanol. Recruitment was limited to those who do not frequently consume chili peppers, to avoid individuals who may already have some baseline desensitization via dietary exposure (e.g., [9]). To screen for chili pepper use, individuals answered questions on and intake of chilies and preferred heat levels. Individuals who reported consuming chili peppers (including hot sauce, red pepper flakes or cayenne pepper, jalapenos, habanero or other hot peppers) less than once per week, and reported a heat preference of medium or less, were invited to participate.

In Study 1, 51 participants (31 women; 20 men) with a mean age of 27.3 (±0.7 std. err.) years completed the study. Three additional individuals enrolled but did not complete the study: two were removed during the first visit due to high blood pressure, and one participant was removed after missing their second visit. (See below for details on biopsy procedure and the exclusion for high blood pressure). For Study 1, participants were randomized using an unequal allocation ratio of 2 to 3 – upon study completion, the control group containing 20 participants while the treatment group had 31. For Study 2, 45 participants (35 women; 10 men) with a mean age of 30.6 (±1.2 std. err) years completed the study; all were assigned to the treatment group.

2.3. Psychophysical scaling and practice session

Psychophysical ratings for test stimuli were collected on a horizontal gLMS labeled with ‘’NS’ (no sensation) at 0 and ‘the strongest imaginable sensation of any kind’ at 100; additional labels of ‘BD’ (barely detectable), ‘weak’, ‘moderate’, ‘strong’ and ‘very strong’ were located at 1.4, 6, 17, 35, and 51, respectively. Participants were given oral and written instructions on use of a gLMS [34]. They were told they should not let whether or not they liked or disliked each stimulus influence their rating of intensity, and that they should feel free to click anywhere along the scale, and not just on the semantic labels. Next, participants rated 15 remembered or imagined sensations to familiarize them with the task (e.g., [35]). These included both food and non-food sensations to ensure participants were making ratings in context to all stimuli, not just oral sensations. This warm-up procedure was completed at each visit prior to rating sampled stimuli, to ensure participants were using the scale correctly. To confirm participants were using the scale as instructed, ratings were examined for each participant (see [36]); no participants were removed.

For Study 1, participants used a gLMS to rate sweetness, bitterness, and burning sensations for all stimuli. Scale order was randomized across individuals to avoid any first or and last position biases. Given the additional stimuli in Study 2 (detailed below), gLMS ratings were collected for the following sensations: burning, cooling, tingling/numbing, warming, bitterness, sweetness, minty, medicinal, cinnamon, and other. Ratings for the non-dominant sensation of a given stimulus (i.e., burning for sucrose, etc) were typically close to zero, so we focused solely on the predominant sensation of each stimulus in our analysis (e.g., cooling for menthol, burn for capsaicin, etc).

2.4. Stimuli preparation and sampling procedure for Days 1, 3, & 17

In Study 1, capsaicin solutions were presented at 3, 6, and 9 ppm. These were made with natural capsaicin (Sigma) and reverse osmosis (RO) water, along with ~4% ethanol (95% USP-grade ethanol) to help solubilize the capsaicin [37]. Natural capsaicin sold by Sigma is a mix of capsaicin and dihydrocapsaicin (~65%/~35%, respectively, with small variations between lots); given similar potency between capsaicin and dihydrocapsaicin, as well as the nominal branding, we refer to it as capsaicin here for simplicity. All capsaicin stimuli were made from a single stock in neat USP ethanol (95%; 190 proof). This stock was diluted with reverse osmosis (RO) water and ethanol to reach the desired capsaicin concentrations, with a constant concentration of ethanol that was nominally ~4% v/v (i.e., calculations were not adjusted for the use of 190 proof ethanol). While ~4% ethanol evokes taste and chemesthetic sensations, mean burn and bitterness rating fall between barely detectable and weak on a gLMS [36] so they can be assumed to be negligible. For Study 2, only 6-ppm capsaicin was used; it was prepared as above.

For Study 1, other stimuli included 0.5 M sucrose (Domino) and 20 uM sucrose octaacetate (SOA; Sigma-Aldrich), a food-grade bitterant with GRAS status. For consistency, SOA samples also contained ~4% ethanol (i.e., a constant vehicle). As an exploratory aim, participants also sampled and rated 40% (v/v) ethanol in water, which should have strong burn and moderate to strong bitterness on a gLMS [36]. For Study 2, other stimuli included 0.5 M sucrose, 0.5 ppm vanillyl butyl ether (VBE), 44 mM cinnamaldehyde, 10.5 mM menthol, and 32% (v/v) ethanol, a concentration intended to have mean burn and bitterness between moderate and strong on a gLMS [36].

Prior to tasting any stimuli, participants rinsed with room temperature RO water. All stimuli were presented at room temperature in 10 mL aliquots in plastic medicine cups labeled with a random 3-digit blinding code. Participants were instructed to swish the entire sample for 5 seconds, spit it out and wait 10 to 15 seconds before rating intensity. Presentation order of stimuli was randomized within visits, except where noted below.

In Study 1, a total of 10 samples were given within a session. Four of these, 3 ppm and 6 ppm capsaicin, 20 uM SOA, and 0.5 M sucrose, were presented in duplicate in a blocked counterbalanced Williams’ design, except the 6 ppm capsaicin stimuli, which was always presented in the 4th and 8th position. The remaining two samples – 40% ethanol and 9 ppm capsaicin – were presented in fixed positions (9th and 10^th, respectively) to limit potential carry-over effects or potential sensitization or desensitization within a test session. In Study 2, a total of 12 samples were given within a session. All six stimuli – 6ppm capsaicin, sucrose, VBE, cinnamaldehyde, menthol and 32% ethanol – were presented in duplicate in a counterbalanced Williams’ design.

In Study 1, a minimum 1-min wait time was enforced between all stimuli, with an additional minute after the 4^th and 8^th stimuli (6 ppm capsaicin) and after the 9^th stimuli (40% ethanol). During this break, participants rinsed with room temperature RO water until they no longer experienced any remaining sensations; they were encouraged to wait longer than the minimum wait time if they experienced any lingering sensation(s). For Study 1, due to logistical and space constraints in the Clinical Research Center, all samples were served at room temperature. For Study 2, similar procedures were used, with a few minor modifications. All stimuli were served at 25°C, via temperature-controlled water baths, wait times were fixed at 90-sec between samples, and presentation order was counterbalanced with a Williams Design.

2.5. Longitudinal Exposure Protocol for Studies 1 & 2

The oral rinse protocol was described to each participant in the first visit to ensure they understood fully and were willing to participate. Study 1 participants were randomized during their second visit (Day 3) to treatment (capsaicin) or control (SOA) groups with an allocation ratio of 3:2. In Study 2, there was no control group, (i.e., all participants got the capsaicin oral rinse). In both studies, participants were asked to swish with 15 mL of a room temperature “mouthwash” solution for 30 seconds, twice per day for 14 days. Participants were asked to refrain from eating or drinking for at least 1 minute after expectorating the mouthwash. Participants were given a total of 420 mLs of the oral rinse to take home at the end of visit 2. The oral rinse was provided in two 250 mL plastic (HDPE) light-block bottle (Bel-Art Products). All oral rinses were made with RO water and stored at room temperature. In Study 1, the treatment oral rinse consisted of 6 ppm capsaicin, while the control oral rinse consisted of 20 uM SOA; both contained ~4% ethanol. SOA is a potent bitterant with GRAS status; it was used to ensure we did not have differential compliance or dropout across the two rinse conditions. To estimate compliance with the mouthwash protocol in Study 1, participants were provided a paper calendar to record the time of each rinse. These were returned to research staff at Visit 3 (day 17): 38 participants (74%) completed all 28 rinses, 9 (~18%) missed 1 rinse, 2 (~4%) missed 2 rinses, and 1 missed 3 rinses. For Study 2, no attempt was made to assess compliance, based on the high compliance seen in Study 1.

2.6. Other measures collected from Study 1 participants

Although participants had been pre-screened for chili pepper liking and intake during recruitment (to eliminate frequent consumers), we confirmed this during the first visit using 7 prompts followed by frequency categories. Exact wording was: ‘How often do you consume … [foods containing chili peppers; chili pepper containing foods that most people would consider to be medium to very spicy; hot sauce; chili peppers; habanero peppers; red pepper flakes; spice mix containing chilies]?’. We also assessed preferred level of heat when eating out at a restaurant – categorical responses were ‘No heat, I avoid eating spicy foods,’ ‘mild,’ ‘medium,’ ‘spicy,’ or ‘very spicy’. Frequency of alcohol consumption was also assessed, as ethanol activates the TRPV1 receptor [38]. Participants also completed a survey to rate liking/disliking for 28 foods (both spicy and non-spicy), and 9 non-food items. The alcohol intake and liking survey data were not used, but are noted here to provide a full accounting of tasks completed by participants (see supplemental materials). Last, to characterize the participants psychographically, we used the Variety Seeking Tendency Scale (VARSEEK) to estimate their willingness to try new or unusual foods [39]. They responded to each VARSEEK prompt using a 5-point Likert scale (completely disagree to completely agree). Responses were scored from 0 to 4 and summed to give a total score of 0 to 32 for each participant.

2.7. Biopsies to harvest Fungiform Papillae

For Study 1, all fungiform papillae (FP) biopsies 1 were conducted in the Clinical Research Center at Penn State by two licensed clinicians who had been previously trained on this procedure. Participants underwent FP biopsies after sampling all stimuli on visits 2 and 3 (Day 3 and 17). Collection of FP followed published methods [31, 40]. Prior to all biopsies, the clinicians checked the oral cavity for signs of inflammation and disease, and confirmed that the participants were not taking blood thinners, before measuring blood pressure. Participants with a blood pressure 143/91 mmHg or above were removed from the study by the clinician to minimize excessive bleeding, or other complications during the biopsy [31, 40]; two participants were removed for elevated blood pressure and none were removed due to oral health, inflammation, or medication use. Surgical microscissors (McPherson-Vannas; Roboz) were used to remove 6 to 10 FP (equal numbers from the left and right side of the tongue). The tissues were transferred to RNAlater (Invitrogen, Life Technologies Corp) using forceps. During collection, a researcher marked the location for each harvested papillae (as done previously [31]). The tissues were stored temporarily (no more than 1 week) at 4°C until transferred to −80°C for storage. All scissors and forceps were cleaned and sterilized in the Clinical Research Center using appropriate methods.

2.8. Isolation and quantification of mRNA using real-time RT-PCR

Tissues and RNAlater were brought to room temperature prior to homogenization. RNA was isolated using Rneasy Mini Kit (Qiagen) per the manufacturer’s protocol. ABI High Capacity RT Kit was used to generate complementary DNA. The control gene PPIA (peptidylproplyl isomerase A; Hs04194521_s1) and the target gene TRPV1 (Hs00218912_m1) were acquired from Invitrogen, Life Technologies Corp. Fungiform papillae may vary in size and quantity of tissue for each participant; therefore it is required to have a housekeeping gene (PPIA), which serves as a control, which makes it possible to measure the quantity of TRPV1 relative to PPIA [41]. The quantity of mRNA was determined using TaqMan quantitative real-time polymerase chain reaction. Assays were repeated in duplicate.

Three different approaches were used to compare relative expression. First, the relative expression for samples harvested during visit 2 (baseline) was normalized to the average of all samples (see below). Second, for samples harvested at the follow-up visit, where participants were assigned to either treatment or control group (i.e., samples from visit 3), individual ΔCt (cycle threshold) values were subtracted from the average ΔCt of their respective group (treatment and control). Lastly, the follow-up relative expression was subtracted from baseline expression values to measure a change in expression. However, to be consistent with calculating both baseline and follow-up values, the baseline measurement of relative expression was reanalyzed, following the equation using normalization to their respective groups [31, 42].

$$\frac{\left(\text{Expression}\text{of}TRPV1\right)}{\left(\text{Expression}\text{of}PPIA\right)}=2^{-\left(C{t}_{TRPV1}-C{t}_{PPIA}\right)-\left(\text{Avg}.C{t}_{TRPV1}-\text{Avg}.C{t}_{PPIA}\right)}$$

Alternate methods for estimating expression were considered. Individual’s ΔCt values were not subtracted from the group average ΔCt for follow-up samples (as done in [31]), as there were two different groups for the second tissue analysis. Similarly, the method used to analyze the follow-up samples was not employed for the first collection, as at the time of collection of the first tissue sample, participants had not yet been randomized into treatment or control groups.

2.9. Statistical analysis

All analyses were conducted using SAS 9.2 or SAS 9.4 (Cary, NC). Sex was coded as 0 for women and 1 for men. Sample size for Study 1 was based on observed differences in the pilot study mentioned previously and power calculations; specifically, we calculated a medium effect size (f =0.25) at alpha = 5% would require ~54 participants to observe a within-between interaction. T-tests were conducted to determine the statistical significance between the treatment and control group for continuous variables (psychophysical ratings, expression, VARSEEK scores). Similarly, t-tests were conducted to test for a significant relationship between sex on age and VARSEEK scores as well as group effects and VARSEEK scores. A chi-square test was utilized in order to test compliance with mouthwashes with group (treatment and control). Regression analyses were used to explore relationships between continuous variables, specifically VARSEEK scores, and relative TRPV1 mRNA expression with psychophysical ratings. Multiple regression analysis was used to determine the relationship between relative TRPV1 expression and burn response across capsaicin concentrations. To determine effects of the repeated mouthwash exposure on psychophysical ratings, a change score was calculated for each participant. After averaging duplicate ratings, visit 3 (follow-up) ratings were subtracted from visit 2 (baseline) ratings to create a change score for each stimulus. This analysis approach was selected from the recommended analysis strategies for determining differences between groups in randomized control trials [43, 44]. For all means, the standard error of the mean (SEM) is reported.

3. Results

3.1. Study 1 - Participant characteristics

As shown in Table 1, participants were randomized to treatment and control groups stratifying by sex. The VARSEEK scale was used to estimate each participant’s food adventurousness [39], with a mean score of 29.3 (±0.75) across all participants: no differences were observed for sex [t(29)=−0.1; p=0.8], age [F(1,49)=0.2; p=0.6], or treatment group [t(49)=0.4; p=0.6]. During screening, we recruited infrequent capsaicin consumers to maximize potential effect sizes. Based on intake measures collected after enrollment (Supplemental Table 1), one participant reported consuming chilies 1 to 2 times per week. Because their preferred spice level was mild, we did not remove them from analysis, as we assumed their total dietary exposure to capsaicin was likely to be relatively low.

Table 1:

Participant characteristics of control and treatment groups following randomization

	Control (n=20)	Treatment (n=31)
Females/Males	11/9	19/10
Age(±SEM)	28.4±1.3	26.5±0.9
VARSEEK	29.7±1.0	29.0±1.0

3.2. Study 1 – responses to sampled capsaicin in visits 1 and 2

For Visits 1 and 2, the mean burn for 3 ppm capsaicin was 20.3 (±1.5) and 17.3 (±1.3), respectively. As expected [45, 46], participants reported mean bitterness near weak for both visits (5.0±1.0 and 5.0±0.07, respectively). For the 6-ppm capsaicin, mean burn was 28.9 (±2.1) 1 and 27.7 (±2.2), with mean bitterness (6.3±1.4 and 5.9±1.4) for Visits 1 and 2, respectfully. For the 9-ppm capsaicin, mean burn was 28.0 (±2.6) and 28.3 (±2.6), and mean bitterness remained weak: 5.6±1.1 and 5.2±1.5. For sucrose (0.5 M), sweetness on Visits 1 and 2 was rated above moderate: 21.1 (±2.1) and 18.9 (±1.5), respectively. For SOA (20 uM), mean bitterness was near or below moderate: 17.1 (±1.9) and 14.6 (±2.1) for visits 1 and 2. As expected, 40% ethanol was predominantlys burning, with means of 21.6±3.2 and 27.3±3.6 for Visits 1 and 2; mean bitterness was 16.7±2.7 and 16.3±3.0.

Responses to individual VARSEEK prompts were summed to create individual scores for each participant. Two groups were created using a median split of individual scores (see [12]): 30 individuals fell into the low group (low food adventurousness) and 21 fell in the high group (high food adventurousness). Independent t-tests revealed these groups were not related to intensity ratings for sampled stimulus for any sensations (p’s>0.1).

3.3. Study 1 - evidence of desensitization following repeated capsaicin exposure

Participants rinsed with treatment (n=31) or control (n=20) stimuli between visits 2 and 3 (days 3 and 17) for a total of ~28 oral rinses. There were no differences in compliance between treatment and control groups (X²(3)=2.9; p=0.3). During visit 3, participants repeated the same rating tasks as in visits 1 and 2. Reported mean intensity ratings for 3, 6, and 9 ppm capsaicin are shown in Figure 1. To determine whether there was a statistically significant reduction in oral burn following capsaicin exposure, we calculated individual change scores for each participant, as recommended [43, 44], with separate scores calculated for each sensation for all stimuli (i.e., rating from Visit 3 – rating from Visit 2). These change scores were then compared between groups to determine if the oral burn was statistically reduced following capsaicin rinses. Visit 1 ratings were intended as a warm-up and were not included in the change score calculation. As shown in Table 2, the treatment group reported a statistically significant reduction in oral burn of ~ 5 to 8 points on a gLMS for all three capsaicin stimuli (3, 6, and 9 ppm), while no significant change was observed for the SOA control.

Figure 1:

Intensity of oral burn (mean ±SEM) for 3, 6 and 9 ppm capsaicin samples rated on a gLMS (‘BD’ is barely detectable) over the 17-day study. Ratings are grouped by treatment (right; n=31) and control (left; n=20) group. during warm-up (visit 1), baseline (visit 2), and follow-up (visit 3) of the 17-day study. The change in burn was statistically significant for the capsaicin exposure group (right) but not the control group (left). Warm-up ratings on Day 1 were not included in the analysis, as individuals had not yet been randomized to a condition. See text and Table 2 for additional details.

Table 2:

Mean change scores (±SEM) and independent sample t-tests comparing change scores between treatment (capsaicin) and control (SOA) groups.

Stimulus	Exposure group
	Control	Treatment	p-value
3 ppm capsaicin	2.41±1.7	−4.85±1.4	0.0019
6 ppm capsaicin	3.32±1.9	−7.87±2.8	0.0001
9 ppm capsaicin	2.29±2.4	−7.29±2.8	0.019
40% ethanol	19.10±5.4	7.24±4.4	0.09
0.5 M sucrose	−2.86±1.4	0.29±1.3	0.11
20 uM SOA	−3.23±2.4	−1.20±1.0	0.39

Following recommended best practices, change scores were calculated for the 6 stimuli by subtracting the visit 2 rating (baseline) from the visit 3 rating (postexposure). When duplicate ratings were available (see text), values were averaged prior to calculating the change score. Here, change in burn are compared for 3, 6 and 9 ppm capsaicin, and 40% ethanol. For SOA, values are bitterness change scores, and for sucrose, sweetness change scores. P-values <0.05 are shown in bold.

This study included two control stimuli, sucrose and SOA, and were examined for changes in intensity ratings for sweetness and bitterness, respectfully between the treatment and control groups (see Figure 2). Change scores for these control stimuli (SOA and sucrose) were calculated as above, and t-tests revealed no statistical differences between groups for the sweetness of sucrose and the bitterness of SOA. Mean change scores for each group are reported in Table 2. Finally, as an exploratory aim, 40% (v/v) ethanol was tested for cross-desensitization. Mean change scores between the treatment and the control group were not statistically different when tested using independent sample t-tests (p = 0.09; Table 2). Yet, the mean change scores between the treatment and control group appeared to differ (19.1±5.4 and 7.2±4.4, respectfully).

Figure 2:

Intensity (mean ±SEM) of sweetness from sucrose (gray lines) and bitterness from SOA (black lines) rated on a gLMS (‘BD’ is barely detectable) over the 17-day study. Ratings are grouped by treatment (boxes; n=31) and control (circles; n=20) group. There was no evidence of a significant change in sweetness or bitterness before and after ~28 exposures. Warm-up ratings on Day 1 were not included in the analysis, as individuals had not yet been randomized to a condition. See text and Table 2 for additional details.

3.4. Study 1 – Relative TRPV1 mRNA expression at baseline was not associated with burn ratings

Fungiform papillae were harvested following Visits 2 and 3, with collection at Visit 2 serving as a baseline measure of TRPV1 expression. Two participants were removed from all the expression analysis as the samples did not meet quality standards for RT-PCR (i.e., abundance). This resulted in 49 participants total in the expression analysis, with 20 in the control group and 29 in the treatment group. The overall mean logged normalized relative TRPV1 expression was 0.0155±0.044. Multiple regression with an interaction term was used to assess the relationship between the initial relative expression of TRPV1 mRNA and rated burn intensity of capsaicin across concentrations (model: burn intensity rating ~ log relative TRPV1 expression x concentration x [log relative TRPV1 expression x concentration]). The overall model was statistically significant [F(3,143)=5.11; p=0.0022], explaining 9.7% of the variability in the reported oral burn. Concentration was a significant predictor of burn intensity of sampled capsaicin (p=0.0004), while TRPV1 expression (p=0.68) and the interaction term (p = 0.26) were not significant predictors. Figure 3 shows the relationship between TRPV1 relative expression and reported burn intensity for 3, 6, and 9 ppm capsaicin. For the 40% ethanol sample, linear regression was used to determine whether relative TRPV1 expression is associated with the perceived burn of ethanol in Visit 2; we failed to observe any evidence of an effect [R² =0.001; p=0.8].

Figure 3:

No evidence of a relationship between TRPV1 expression and burn ratings of 3, 6 and 9 ppm capsaicin at baseline (visit 2) prior to ~28 exposures.

3.5. Study 1 – Relative TRPV1 expression did not decrease as a result of capsaicin exposure

A second FP biopsy was performed on Visit 3 after 14 days of exposure to either 6 ppm capsaicin (n=29) or 20 uM SOA mouthwash (n=20). The fold change for relative TRPV1 expression post-rinse was 0.91(±0.11). Figure 4 visualizes the change in relative TRPV1 expression for both the control (left) and treatment (right) groups. To determine if treatment affected relative TRPV1 expression, a t-test was conducted on relative expression values generated from post-exposure samples normalized to the group’s average (either treatment or control). We failed to observe any evidence of a difference [t(47)=0.88; p=0.38], which suggests there was no relationship between repeated oral capsaicin exposure and TRPV1 expression.

Figure 4:

No evidence of a systematic change in relative TRPV1 expression from Visit 2 (baseline) to Visit 3 (post-exposure; 14 days later). Each line represents an individual participant. On the left (grey) is the SOA control group, while the right (red) shows the capsaicin treatment group. Group means are shown as black circles; error means on the means are standard errors.

To further explore possible relationships between repeated capsaicin exposure and TRPV1 expression in FP, a change score was generated by subtracting expression from samples collected following exposure (Visit 3) from baseline relative expression levels (Visit 2). Unlike the prior analysis, this approach takes an individual’s baseline TRPV1 expression into account. Each participant’s expression values (baseline and post-exposure) were normalized to their respective group average from each collection time point. The mean change in expression for the control group was −0.014 (±0.10), and the treatment group was 0.002(±0.05). When a t-test was used to determine if exposure affected the change in mRNA expression values, we still failed to see evidence of an effect [t(47)=−0.16; p=0.8], again, suggesting variability in change in mRNA expression was not associated with oral exposure.

3.6. Study 1 – Relative TRPV1 expression following exposure was not associated with the burn intensity from capsaicin

We also tested potential relationships between relative TRPV1 expression and intensity from sampled capsaicin following exposure in Visit 3. In multiple regression, we tested if TRPV1 expression could explain variability in oral burn across 3, 6, and 9 ppm capsaicin (Figure 5). The two groups (treatment and control) were tested in separate models, as individual relative TRPV1 expression was normalized to the mean of their respective groups (see methods). The overall model for the SOA control group was not statistically significant [F(2,57)=2.35; p=0.10] (Figure 5, left), whereas the parallel model for the treatment group (Figure 5, right) was statistically significant [t(2,57)=4.35; p=0.01]. For the capsaicin treatment group, the model explained 9.3% of the variability in reported burn intensity: capsaicin concentration was a significant predictor (p=0.01), but as with the biopsy data above, relative TRPV1 expression was not statistically significant predictor of the burn ratings (p=0.11).

Figure 5:

Lack of a systematic relationship between TRPV1 expression and burn ratings of 3, 6, and 9 ppm capsaicin at follow-up (visit 3), after exposure. The left panel is the control group who rinsed with SOA ~28 times while the right panel is the treatment group who rinsed with capsaicin ~28 times. Relative TRPV1 expression for an individual was determined by normalizing to the average of the groups’ expression (control and capsaicin, respectively). See text for additional details.

3.7. Study 2 – Evidence of self desensitization following repeated capsaicin exposure

In Study 2, 45 participants rinsed with 15 mL of 6ppm capsaicin twice daily for 14 days, in a protocol very similar to Study 1. In each visit, they tasted and rated 12 samples, and no biopsies were taken. As anticipated, the predominant quality for VBE, capsaicin, ethanol, and cinnamaldehyde was burning, along with sweetness for sucrose, and cooling for menthol (not shown). In repeated measures mixed model ANOVA predicting the intensity rating of predominant quality for each stimulus (stimulus by day), we saw statistically significant main effects of day [F(1,44) = 25.6; p < 0.0001] and stimulus [F(5,240) = 43.87; p <0.0001]. However, the interaction was not significant [F5,240] = 1.25; p=0.29]. As shown in Figure 5, the burn of capsaicin dropped ~24% (p = 0.037) after ~28 exposures.

3.8. Study 2 – Evidence of cross desensitization for burn but not sweetness or cooling following repeated capsaicin exposure

In Study 2, in addition to ~24% drop in capsaicin burn, we also saw evidence of a ~19% drop in burn for cinnamaldehyde, along with an even bigger drop (~44%) for ethanol. For VBE, the apparent magnitude of the drop (~21%) was similar to the drops seen for capsaicin and cinnamaldehyde, but the evidence was less convincing (p=0.06). Collectively, these data suggest repeated oral capsaicin exposure resulted in a reduction in burn that generalized across multiple chemesthetic stimuli that elicit burning sensations. Conversely, sweetness from sucrose and cooling from menthol were not affected by long term capsaicin exposure, suggesting that any effect was specific to burn and not merely a shift in scale usage.

4. Discussion

Here, we explored effects of repeatedly and systematically rinsing with a 6-ppm capsaicin oral rinse twice daily for two weeks in two separate cohorts of participants. We find that rinsing with capsaicin ~28 times resulted in a statistically significant drop in perceived burn of sampled capsaicin within participants, while a similar drop was not observed in participants rinsing with a control oral rinse of SOA. In these same participants, we also failed to observe any statistical differences for ratings for SOA, sucrose, or menthol between treatment groups, suggesting any differences observed are specific to burn. Notably, these perceptual changes were not explained by TRPV1 mRNA expression in biopsied FP. In a second cohort, we found additional evidence of cross desensitization for other chemesthetic stimuli, and confirmed that this phenomenon was specific to burn. Collectively, these data suggest systematic exposure to low doses of capsaicin was sufficient to induce perceptual desensitization for burn, and these shifts are not merely contextual shifts in how participants use the scale, nor can they be explained by changes in TRVP1 mRNA expression. This implies a unknown mechanism, not investigated here, seems to be responsible for broad changes in oral burn response following repeated capsaicin exposure.

Numerous prior psychophysical studies indicate chronic exposure to capsaicin results in reduced perceptual responses to capsaicin (i.e., a drop in burn ratings), a phenomenon that is widely referred to as capsaicin desensitization [15, 47]. However, other disciplines, such as neuroscience and related fields, uses the same term, desensitization, to describe a pharmacological mechanism [48, 49], rather than a psychophysical outcome, implying hypoalgesia might be a more accurate descriptor. Nonetheless, to be consistent with prior literature, we refer to this reduced perceptual response as desensitization. The mechanism(s) behind this reduction in response are poorly understood, but they do not seem to be mere habituation (see discussion in [12]).

Here, we tested one potential explanation. Specifically, the hypothesis that chronic capsaicin exposure reduces the expression of TRPV1 gene in fungiform papillae. A decrease in mRNA may represent a reduction in the protein available for binding, which in turn would reduce the burn evoked by capsaicin. This hypothesis was based on two different lines of prior research. First, repeated capsaicin exposure in vitro causes TRPV1 to become desensitized, and can result in defunctionalization [25, 28]. Second, tissues harvested from individuals experiencing chronic inflammation and/or cancer patients have shown upregulation of TRPV1 (i.e., increased TRPV1 mRNA expression) in oral lingual tissue relative to healthy controls [50–52]. Collectively, these previous reports led us to speculate that that perceptual desensitization might result from changes in TRPV1 expression.

After confirming that systematic exposure to capsaicin resulted in ~26% to 31% drop in perceived burn (Study 1), we compared the levels of mRNA for TRPV1 before and after exposure. We also explored the potential between mRNA levels and burn cross sectionally at baseline and follow up. Contrary to our original hypothesis, we saw no evidence of a change in mRNA levels following repeated capsaicin exposure, or of any association between perceived burn intensity ratings and mRNA levels. Accordingly, we conclude that perceptual desensitization cannot be explained via simple changes in mRNA levels. Additional research is needed to test other mechanisms that regulate TRPV1 channel activity following repeated exposure to capsaicin.

In light of this negative result, two key design decisions must be noted. First is our choice to measure mRNA versus other approaches, as mRNA quantification can be achieved through RT-PCR; we assumed a correlation exists between mRNA expression levels and protein production. Although differences in TRPV1 mRNA have previously been associated with diseased tissues in humans, a more sensitive measurement may have been needed to precisely measure changes in proteins located on the surface of the cell. This could be accomplished via immunohistochemistry staining on harvested FP. However, to obtain enough tissue, a tongue punch biopsy may be required; as this is substantially more invasive than the FP biopsy used here, we deemed it impractical. Separately, another alternative mechanism to explain perceptual desensitization following chronic exposure would be the degeneration of neurons in the FP, analogous to phenomenon observed on the skin with high doses of capsaicin [53, 54]. We do not think this is likely given the much lower doses used here, but we cannot completely rule this out without using more invasive biopsy techniques.

Second, we must acknowledge that in contrast to prior work linking bitterness perception with expression of TAS2R38 mRNA among a cohort of participants that are heterozygotes for the polymorphisms in TAS2R38 (i.e., PAV/AVI heterozygotes[31], the FP biopsy can only collect the free nerve endings of the trigeminal fibers, and not the cell body of the axon. With current techniques, it is challenging to overcome this concern, as we can think of no ethical way to collect the cell body of these neurons in healthy volunteers.

Collectively, while present data failed to show an effect of capsaicin exposure on the expression of TRPV1 mRNA, other methods of quantifying changes in protein regulation may be needed to fully examine this hypothesis. Given this, it is possible that desensitization occurs independent of the translation of the protein, or modification in the transcription regulators of TRPV1. Alternatively, the cross-desensitization observed in Study 2 may imply the reduction in oral burn is not TRPV1 specific. For example, we observed a reduction in the burn from cinnamaldehyde, a known TRPA1 agonist, implying that the mechanism behind the desensitization observed here might not be regulated by the expression of TRPV1.

5. Conclusions

This study provides the first direct evidence that repeated exposure to low doses of capsaicin in a protocol designed to mimic dietary exposure robustly results in perceptual desensitization. It has long been assumed that widely documented differences in perceived burn from capsaicin in cross sectional studies comparing those who frequently consuming capsaicin containing foods to those who do not (e.g. [11, 12, 55] are a result of desensitization. Here, we confirm that exposure to ecologically relevant doses can cause a statistically significant reduction in burn rating following systematic repeated exposure. To explore potential mechanisms responsible for perceptual capsaicin desensitization, we investigated whether mRNA expression of TRPV1 decreases following repeated capsaicin exposure. In spite of achieving ~25 to 30% reduction in oral burn following repeated exposure, we did not observe evidence of any meaningful changes in TRPV1 expression. This suggests that other methods, such as immunohistochemistry may be required to observe potential changes in the cellular location of TRPV1 as a result of repeated capsaicin exposure.

Additionally, TRPV1 expression was not correlated with burn ratings of sampled capsaicin. More research is needed to determine whether chronic capsaicin exposure results in internalization of the TRPV1 channel, or potentially in neuronal withdrawal (i.e., degeneration of nerve fibers), although this would require invasive biopsies, which were not used in the current study. Finally, we find that long-term exposure results in generalized reductions in burn, including the burn from known TRPA1 agonists, which potentially challenges the assumption the desensitization observed here can be explained at the receptor level. Future studies are needed to identify the mechanism(s) through which chronic capsaicin desensitization is achieved, to inform work on therapeutic use and/or dietary behavior.

Figure 6:

Intensity (mean ±SEM) of burn from cinnamaldehyde (CIN), vanillyl butyl ether (VBE), capsaicin (CAP) and ethanol (EtOH), along with sucrose sweetness (SUC) and menthol cooling (MENTH) rated on a gLMS before and after repeated exposure to low doses of capsaicin ~28 times. In mixed model ANOVA (see text), there was evidence of a significant drop in burn for the four stimuli on the left side of the plot; conversely, there was no evidence of a change in sweetness or cooling over the same time period.

Highlights.

• Participants rinsed with capsaicin ^~28 times over 2 weeks

• Capsaicin burn was reduced in the treatment group

• Cross desensitization was observed for other burning stimuli

• No changes were observed for sucrose, menthol, or SOA

• Papillae biopsies were performed

• TRPV1 expression did not predict changes in burn