Agonists of orally expressed TRP channels stimulate salivary secretion and modify the salivary proteome

Natural compounds that can stimulate salivary secretion are of interest in developing treatments for xerostomia, the perception of a dry mouth, that affects between 10 and 30% of the adult and elderly population. Chemesthetic transient receptor potential (TRP) channels are expressed in the surface of the oral mucosa. The TRPV1 agonists capsaicin and piperine have been shown to increase salivary flow when introduced into the oral cavity but the sialogogic properties of other TRP channel agonists have not been investigated. In this study we have determined the influence of different TRP channel agonists on the flow and protein composition of saliva. Mouth rinsing with the TRPV1 agonist nonivamide or menthol, a TRPM8 agonist, increased whole mouth saliva (WMS) flow and total protein secretion compared to unstimulated saliva, the vehicle control mouth rinse or cinnamaldehyde, a TRPA1 agonist. Nonivamide also increased the flow of labial minor gland saliva but parotid saliva flow rate was not increased. The influence of TRP channel agonists on the composition and function of the salivary proteome was investigated using a multi-batch quantitative mass spectrometry method novel to salivary proteomics. Inter-personal and inter-mouth rinse variation was observed in the secreted proteomes and, using a novel bioinformatics method, inter-day variation was identified with some of the mouth rinses. Significant changes in specific salivary proteins were identified after all mouth rinses. In the case of nonivamide, these changes were attributed to functional shifts in the WMS secreted, primarily the over representation of salivary and non-salivary cystatins which was confirmed by immunoassay. This study provides new evidence of the impact of TRP channel agonists on the salivary proteome and the stimulation of salivary secretion by a TRPM8 channel agonist, which suggests that TRP channel agonists are potential candidates for developing treatments for sufferers of xerostomia.

respond to a variety of somatosensory and endogenous stimuli. TRPV1, 3, 4, TRPA1 and TRPM8 are 67 expressed in the oral cavity that have thermo-and chemoreceptive functions. They are expressed on 68 mucosal and epithelial free afferent nerve endings of myelinated Aδ and non-myelinated C fibres (1), 69 oral epithelial cells (2-4), taste buds (5, 6), and keratinocytes (7). 70 71 Cinnamaldehyde is a TRPA1 agonist, which is produced synthetically and found in cinnamon, a spice 72 that comes from the bark of cinnamon trees (8). Cinnamaldehyde makes up 90% of the essential oil 73 extracted from cinnamon bark. Upon contact, cinnamaldehyde provokes a feeling of warmth (8) and  74 has potential anti-inflammatory (9-11) and anti-cancer (12)(13)(14)(15)(16)(17)(18) properties . Menthol is a TRPM8 agonist 75 that provokes a cooling sensation. It is found in mint leaves and produced synthetically (19). Nonivamide 76 is a capsaicinoid that elicits a burning sensation (20). It is structurally very similar to the more widely 77 studied TRPV1 agonist capsaicin and is naturally found in chilli peppers or produced synthetically. 78

79
The salivary response to basic tastants is well studied but the salivary response to TRP channel 80 agonists requires further investigation. Increased salivary flow rate and specific protein secretion have 81 been demonstrated in response to other tastants (21)(22)(23)(24) and there are studies demonstrating increases 82 in salivary flow rates and specific protein changes in response to the TRPV1 agonists (25-29) but there 83 has been limited study of agonists to other TRP channels, despite expression of these channels in the 84 oral cavity, nor has the mechanism of TRP channel agonist stimulated salivary secretion been 85 elucidated. 86 87 Studying compounds that can stimulate salivary flow is of interest to the development of treatments for 88 xerostomia, the perception of a dry mouth, that affects between 10 and 30% of the adult and elderly 89 populations (30,31). Acidic tastants that strongly stimulate salivary secretion erode enamel tissues, so 90 alternative molecules are sought (32). Although xerostomia is often associated with hyposalivation, 91 where the WMS flow rate is reduced by ~50% (33), this is not always the case (34). Xerostomia in the 92 absence of hyposalivation may be due to changes in the interaction of saliva with oral surfaces due to 93 the altered integrity of salivary proteins (35) or changes in saliva rheology (36). There is evidence that 94 TRP agonists modify the rheological properties of saliva but the mechanism by which these changes 95 occur remains to be elucidated. Taken together, identifying compounds that not only induce salivary 96 secretion but also modify the rheological properties of saliva is of interest to developing treatments for 97 xerostomia. 98 99 Specific protein changes in saliva in response to differing stimuli are possible due to the many sources 100 of proteins which are likely to respond differently to different nerve mediated stimuli. For example, the 101 submandibular and sublingual glands secrete in response to olfaction (37) whereas the parotid glands 102 do not (38). Conversely, the parotid glands are preferentially stimulated by chewing which results in a 103 higher amylase output (39). In these scenarios, proteins associated with specific glands, e.g. higher 104 amylase secretion by the parotid glands or mucin secretion by the submandibular and sublingual 105 glands, will have a relatively increased abundance when compared to unstimulated levels. 106

107
The regulation of specific proteins separate from preferential gland stimulation has also been reported. 108 Annexin A1 and calgranulin A are upregulated in WMS through an inflammatory-like response after 109 mouth rinsing with bitter, umami and sour tastants (40). Bader et al. demonstrated the upregulation of 110 lysozyme in saliva stimulated by citric acid rinse (41). The TRPV1 agonist 6-gingerol upregulated 111 salivary sulfhydryl oxidase 1 resulting in reduced 2-furfurylthiol levels in exhaled breath and thus 112 reduction in the perceived sulphur-like after-smell (42). However, the mechanism of these specific 113 protein upregulations has not been elucidated. 114

115
The present study is formed of two parts. A bottom-up quantitative proteomics study of the salivas 116 secreted by two participants in response to menthol, cinnamaldehyde, nonivamide and propylene glycol 117 (PG) that were compared to unstimulated saliva using mass spectrometry. In addition, data on WMS 118 flow rates and protein output were also collected. In order to improve the identification of lower 119 abundance salivary proteins, a method novel to salivary proteomics was used. Secondly, studies were 120 conducted to confirm the specific protein changes of the proteomes of salivas identified in the 121 proteomics study and to consider the mechanism by which the compounds exert their effects on the 122 salivary proteomes. 123 124

Experimental Procedures 125 Experimental Design and Statistical Rationale 126
For the proteomics study, the proteome of 60 WMS samples, obtained from two male volunteers of 127 ages 24 and 27, were analysed by TMT quantitative mass spectrometry. Forty eight experimental 128 samples consisting of WMS produced after mouth rinsing were split randomly across six TMT10plex 129 batches with each batch containing two controls consisting of pooled unstimulated saliva from each 130 participant. The 48 WMS samples were collected from two participants after being exposed to eight 131 conditions each with three experimental repeats. In a further study of the effects of agonists on WMS 132 secretion, 25 participants were recruited (the demographic information of each participant group is 133 shown in 134 Table 1) six of these subjects also participated with further participants in the following studies. For the 135 parotid saliva study, eight volunteers were recruited (38.7 ± 5.3 years, male n = 4, female n = 4). For 136 the lower labial gland saliva study, ten volunteers were recruited (29.4 ± 4.7 years, male n = 5, female 137 n = 5). For all studies, volunteers were healthy individuals recruited by internal advertisement with the 138 following exclusion criteria: on prescription medication, age > 65years or < 18years, suffering from oral 139 discomfort. The controls and statistical tests used for each analysis are described below. 140

141
Proteomics study of TRP agonist stimulation on two subjects. 142 Forty eight saliva collections were made in total, each collection including an unstimulated saliva 143 sample, followed by a mouth rinse and then two post-mouth rinse saliva samples (Table 2). Eight 144 different mouth rinse solutions were tested in triplicate: nonivamide, cinnamaldehyde, menthol and PG 145 (Symrise AG) ( Table 2). The solutions were prepared in pre-weighed universal tubes and the total 146 weight recorded. The compounds were diluted in water (Buxton) on the day of collection and were 147 stored at room temperature. Participants were asked not to consume food, water or smoke in the 1 hour 148 prior to collection. The following guidance was given to each participant prior to each collection: tilt your 149 head slightly forward to allow saliva to pool underneath the tongue; do not move your mouth unless it 150 is to spit out collected saliva; spit out whenever it is comfortable; do not swallow. For each collection, 151 the following protocol was adhered to: One minute of unstimulated WMS was collected in a pre-weighed 152 universal tube; 10 mL of mouth rinse was then taken into the mouth for 30 seconds and spit back into 153 a pre-weighed universal tube; two, one minute collections of post-mouth rinse WMS in pre-weighed 154 universal tubes. Immediately after collection, participants were asked, "How would you rate the intensity 155 of the mouth rinse" and were asked to give a rating from 0 -10 on a visual analogue scale alongside 156 an oral description of their perception of the mouth rinse. One collection was carried out per day at 2pm 157 and the order of mouth rinses were randomised for each participant. All samples were weighed in the 158 universal tube straight after collection. Saliva was then processed for storage prior to mass 159 spectrometry analysis: samples were transferred to ice cooled 1.5 mL microtube for centrifugation (13 160 500 rpm, 5 minutes, 4 °C). Supernatants were removed, frozen at -20 °C and finally moved to -80 °C 161 storage; the pellets were discarded. 162

Effects of TRP agonists on WMS flow rates 165
Cinnamaldehyde, menthol and nonivamide were obtained from Symrise AG and prepared in PG. 300 166 ppm cinnamaldehyde, 500 ppm menthol, 1ppm nonivamide and 3 x 10 4 ppm PG were prepared by 167 diluting in water (Buxton) in pre-weighed universal tubes and the total weights were recorded. The 168 concentration of PG in the nonivamide, menthol and cinnamaldehyde mouth rinses was 3 x 10 3 , 1 x 10 4 169 and 3 x 10 4 ppm respectively. The solutions were kept at room temperature (20 °C). Participants were 170 asked not to consume food, water or smoke in the 1 hour prior to collection. Prior to collection each 171 participant was asked to tilt their head slightly forward to allow saliva to pool underneath the tongue, to 172 not move their mouth unless it was to spit out collected saliva, to spit out whenever it is comfortable and 173 to not swallow. Five minutes of unstimulated WMS was collected in a pre-weighed universal tube as a 174 control. Ten mL of a control mouth rinse containing either the equivalent concentration of PG as in the 175 TRP agonist containing mouth rinse or water was then taken into the mouth for 30 seconds and spat 176 back into a pre-weighed universal tube, this was followed by five one minute collections of WMS into 177 pre-weighed universal tubes. This was repeated with the experimental mouth rinse. All samples were 178 weighed in the universal tube immediately after collection. Samples were kept on ice after collection. 179 The neat saliva samples were aliquoted into 2 mL microtubes and then centrifuged (13 500 rpm, 4 °C, 180 5 minutes). The supernatant was removed, aliquoted and stored at -20 °C. 181 182 Parotid Saliva Collection 183 Five 10 mL solutions were prepared: water (Buxton); propylene glycol (3.0 x 10 4 ppm), menthol (100 184 ppm), cinnamaldehyde (60 ppm), nonivamide (1 ppm). These solutions were prepared in pre-weighed 185 universal tubes and the total weights recorded. The solutions were kept at room temperature (20 °C). 186 Lashley cups were fitted over the exit of the Stenson's ducts, secured and correct fitting was tested by 187 the administration of a few drops of 2% citric acid onto the tongue to stimulate parotid secretion. Time 188 was allowed so that the collection tubes of the Lashley tubes were filled with parotid saliva. Prior to 189 collection each participant was asked to not swish any solution around in their mouth in order to prevent 190 Lashley cups being dislodged. The volunteer was given 10 mL water to practice holding the solution in 191 the mouth and spitting it out. Unstimulated parotid saliva was collected in a pre-weighed universal tube 192 for 5 minutes. Ten mL of water (Buxton) was then taken into the mouth and held for 5 minutes. During 193 this time parotid saliva was collected in a pre-weighed universal tube. This was repeated with the control 194 and TRP agonist solutions in the following order: propylene glycol, menthol, cinnamaldehyde, and 195 nonivamide. A two minute break was taken between each solution. Saliva samples were kept on ice 196 after collection. The neat saliva samples were aliquoted into 2 mL microtubes and then centrifuged (13 197 500 rpm, 4 °C, 5 minutes). The supernatant was removed, aliquoted and stored at -20 °C. 198 199 Lower labial gland saliva collection 200 A cotton roll was placed over each Stenson duct's papilla and under the tongue to absorb major gland 201 saliva. The inferior labial surface was dried, and unstimulated lower labial saliva was allowed to bead 202 on the surface of the inferior labium for 2 minutes. A 2 cm x 1 cm piece of pre-weighed Whatman's 203 (General Electric) filter paper was then placed on the lower labial surface with one of the 1 cm edges 204 halfway down the mid-point of the inferior labium to collect the beads of saliva. The saliva-soaked filter 205 paper was placed in a pre-weighed 1.5 mL microtube, weighed and the flow rate calculated by 206 subtraction of the pre-weighed paper and pre-weighed microtube weights and divided by the time of 207 collection in minutes. To allow for slight variations in the size of the filter paper, flow rates were scaled 208 according to the mass of the dried filter paper. This process was repeated but with a 30 second mouth 209 rinse of either 3.0 x 10 4 ppm PG, 300 ppm cinnamaldehyde, 500 ppm menthol or 1 ppm nonivamide 210 being administered prior to the drying of the inferior labium. The following guidance was given to each 211 participant prior to collection: ensure the mouth rinse baths the surface of your lower lip; do not swallow 212 the mouth rinse. A three minute break, or until the perception of the previous mouth rinse had 213 diminished, was taken between each solution. Saliva infused filter paper samples were kept on ice after 214 collection. 215 216 Saliva infused filter paper was placed into 0.5 mL microtubes that had 4 needle-sized holes pierced into 217 their underside. Each 0.5 mL microtube was then placed into a 1.5 mL microtube and centrifuged (13 218 000 rpm, 4 °C, 5 minutes). The saliva collected in the 1.5 mL microtube was immediately processed for 219 SDS PAGE (see below) with the following modification: the entire volume of the collected saliva (~1 µL) 220 was treated with 10 µL lithium dodecyl sulphate (LDS) sample buffer and 1 µL dithiothreitol (DTT) prior 221 to heating and electrophoresis. 222

223
Quantitative tandem mass spectrometry 224 The first minute and second minute post-mouth rinse samples from each collection were pooled. The 225 24 unstimulated samples from each of the two participants (48 in total) were pooled into two 226 unstimulated pools, one for each participant. Five µL of each pooled sample was added to 95 µL 227 phosphate buffered saline (137mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4) for 228 protein quantification using a Bradford assay (Thermo Scientific, USA). Absorbance of each sample 229 was read by spectrophotometer at 595 nm and compared to a standard curve of bovine serum albumin 230 of known protein concentration. Fifty µg of protein was extracted from each sample and frozen at -80°C. 231 Frozen samples were freeze dried and reconstituted in 70 µL 100 mM triethylammonium bicarbonate 232 (TEAB) and 0.1% sodium dodecyl sulphate (SDS). 10 µL 8mM tris (2-carboxyethyl) phosphine (TCEP) 233 in 100 mM TEAB, 0.1% SDS was added to each sample and incubated at 55°C for one hour. 10 µL 375 234 mM iodoacetamide (IAA) in 100 mM TEAB, 0.1% SDS was added to each sample and incubated at 235 room temperature for 30 minutes. 4 µL of 0.25 µg/µL trypsin (Roche, sequencing grade) was added to 236 each sample and left overnight at 37 °C. 237 238 Forty one µL of TMT reagent was added to each of the 48 post mouth rinse samples and the twelve 239 unstimulated pool samples (see Table 3 for details) and incubated at room temperature for one hour. 240 Eight µL of 5% hydroxylamine was added to each sample and left at room temperature for 15 minutes. 241 Samples from each 10plex batch were pooled into six 10plex sample pools and stored at -80 °C prior 242 to freeze drying until completion. 243 244 IEF fractionation was carried out using the Agilent 3100 OFFGEL system (Agilent Technologies Inc, 245 Germany) and was carried out according to the manufacturers protocol. 1.8 mL OFFGEL buffer stock 246 added to each sample for reconstitution. Six OFFGEL strips with a linear pH gradient ranging from 3 to 247 10, one for each 10plex sample pool, were hydrated in 50 µL OFFEGL rehydration solution for 15 248 minutes. 12-fraction frames were fitted to each of the strips and 150 µL of reconstituted sample loaded 249 into each fraction well. IEF was carried out under the following conditions: 20 kVh (100 hours, V: 500-250 5400 V, max. I: 50 µA. Upon completion, each fraction was removed and frozen at -80 °C. Fractions 251 were thawed on ice and pooled into six fraction pools (Fraction 1 with 7, 2 with 8, 3 with 9, 4 with 10, 5 252 with 11 and 6 with 12). Ten µL of elution buffer (50% acetonitrile (ACN), 0.1% formic acid) was added 253 to each sample. Zip-Tips were hydrated twice in 10 µL hydration solution (50% ACN, trifluoroacetic acid 254 (TFA)) and then washed in 1 µL of wash solution (0.1% TFA). S10 µL samples was washed through 255 the Zip-Tip 10 times before eluting with elution solution (0.1% TFA). The elute was frozen at -80 °C 256 prior to freeze drying until completion. Fractions were reconstituted in 10 µL 50mM ammonium 257 bicarbonate. The peptides from each fraction were resolved using reverse-phase chromatography on a 258 75 µM C18 EASY column using a 3-step gradient of 5-40% ACN and a 95% ACN wash in 0.1% formic 259 acid at a rate of 300 µL/min over 220 minutes (EASY-NanoLC, ThermoScientific, USA). Nano-ESI was 260 performed directly from the column and ions were analysed by using an LTQ Orbitrap Velos Pro 261 (ThermoScientific, USA). Ions were analysed using a Top-10 data-dependent switching mode with the 262 10 most intense ions selected for HCD for peptide identification and reporter ion fragmentation in the 263 Orbitrap. Automatic gain control targets were 30,000 for the iontrap and 1,000,000 for the orbitrap 264 265 Quantitative MS Data analysis 266 Tandem mass spectra were extracted from the Xcalibur data system (version 2.2, ThermoScientific, 267 USA) and searched through Mascot (v. 2.6.0) using Proteome Discoverer software (version 1.4.0.288, 268 ThermoScientific, USA) to determine specific peptides and proteins. The parameters included: 20 ppm 269 peptide precursor mass tolerance; 0.5 Da for the fragment mass tolerance; 2 missed cleavages, trypsin 270 enzyme; TMT-6plex (N-terminus and K), carbamidomethyl (C) and oxidation (M) dynamic modifications; 271 database: UniProt_HUMAN (release-2018_02, 20 366 entries). False discovery rate was set at 0.05 272 and 0.01 for relaxed and strict parameters respectively, with validation based on q-Value. The data were analysed using KNIME and embedded R scripts (KNIME analytics platform, Germany). Peptides 274 were excluded from analysis if they were unassigned or had missing TMT channel intensity data; the 275 primary accession number was taken for each peptide and proteins were grouped by this accession 276 number with the geomean of individual peptide intensities given as the protein intensity value; TMT 277 intensities were normalised using a sum scaling method and to the geomean of the two standard values 278 for each peptide. Batches were then concatenated, batch corrected using ComBat (43)  The total protein concentration of collected saliva samples were determined by bicinchoninic acid assay 291 (Thermo Scientific). Frozen saliva samples were defrosted on ice and then diluted 1:10 in ddH20 in 292 duplicate alongside a serial dilution of bovine serum albumin standard (2 mg/mL -0.03125 mg/mL). 293 Samples and standards were incubated with bicinchonic acid for 30 minutes prior to measuring 294 absorbance as 540 nm using an iMark microplate absorbance reader (BioRad). Polyacrylamide gels were placed in 0.2% Coomassie Brilliant Blue R250 in 25% methanol and 10% 308 acetic acid at room temperature for 90 minutes, followed by overnight de-staining in 10% acetic acid. 309 Periodic acid Schiff's (PAS) staining: 60 minute fixing in 25% methanol and 10% acetic acid, incubation 310 with 1% periodic acid followed by water rinsing and Schiff's reagent staining. Gels were imaged using 311 the ChemiDoc MP Imaging System (BioRad). 312 313 Immunoblotting 314 Separated proteins were electroblotted to nitrocellulose membranes for 60 minutes at 190 mA and 30 315 V (constant). Blots were blocked in 5% semi skimmed milk (Fluka) and probed with either an affinity-316 purified antibody fraction of mouse antiserum to a synthetic peptide of human cystatin-s corresponding 317 to amino acid residues 21-141 (AF1296, R&D Systems) or an affinity-purified goat antibody raised 318 against a peptide mapping at the C-terminus of human amylase (sc-12821, Santa Cruz). Binding was 319 detected using a horseradish-peroxidase-labelled, affinity purified goat-ant-rabbit IgG (P0160, Agilent 320 Dako) or rabbit-anti-mouse IgG (P0161, Agilent Dako) followed by Clarity Western ECL substrate 321 detection system. Chemiluminescence was detected by ChemiDoc MP Imaging System (BioRad Data were tested for normality using the Shapiro-Wilks normality test. 1-way ANOVA were used for 331 determining statistically significant differences within the lower labial gland flow rates, parotid gland flow 332 rates, protein output, cystatin S abundance datasets and, in the in-depth analysis, grouped WMS flow 333 rate and protein output datasets. A 2-way ANOVA was used for determining statistically significant 334 differences within the WMS flow rate datasets and, in the in-depth analysis, in the subject separated 335 WMS flow rate and protein output datasets. The above analyses were carried out using Prism 6 336 software (GraphPad). The following were used to denote statistically significant differences in the 337 figures: **** = P ≤ 0.0001, *** = P ≤ 0.001, ** = P ≤ 0.01, * = P ≤ 0.05.

TRP agonists stimulate salivary secretion 347
Significantly greater relative WMS flow rates were observed in response to the TRP agonist containing 348 mouth rinses when compared to the UWMS flow rate (Figure 1a). Furthermore, 1 ppm nonivamide and 349 500 ppm menthol mouth rinsing significantly increased relative mean WMS flow rates compared to PG 350 mouth rinsing, which itself significantly increased WMS flow rates compared to UWMS. The 351 reproducibility of WMS flow rates in response to menthol and nonivamide mouth rinsing was 352 demonstrated by repeating measurements with two of the participants (Figure 2a Our study identified 288 unique human proteins absent from both datasets and so, to the best of our 379 knowledge, are novel findings for the salivary proteome (Figure 3b). Greater confidence can be 380 assigned to the 134 proteins that have a SwissProt annotation score of 5, relating to strong evidence 381 of their existence in vivo, and of these, 12 were identified with at least one unique peptide across the 382 batches, of which 9 had a relative abundance of less than 0.2%. suggesting that the cinnamaldehyde mouth rinses were not causing additional variation in the WMS 400 proteome than was already induced by the PG in the mouth rinse. However, post-nonivamide and 401 menthol coordinates were separated from the PG coordinates suggesting these compounds were 402 inducing proteome changes independently of PG (note the lower concentrations of PG in nonivamide 403 and menthol mouth rinses compared to cinnamaldehyde (Table 2) rinses. The peptides assigned to each of these proteins (13, 10 and 17 to S, SA and SN respectively) 420 were unique. Additionally, cystatin D was upregulated at both concentrations of nonivamide and cystatin 421 C was upregulated after 1 ppm nonivamide mouth rinsing. Menthol at 500 ppm caused upregulation in 422 salivary cystatins to a greater extent than PG. Although salivary cystatins were upregulated after 423 cinnamaldehyde mouth rinsing, it was less than with PG mouth rinses despite the same concentration 424 of PG being present in 1.8 x 10 4 ppm and 3.0 x 10 4 ppm PG to 180 ppm and 300 ppm cinnamaldehyde 425 respectively. The finding that salivary cystatins are upregulated by 1 ppm nonivamide mouth rinsing 426 was supported by qualitative immunoprobing (Figure 5). Statistically significant greater cystatin S was 427 observed in WMS after 1 ppm nonivamide mouth rinsing (Figure 5c). 428 429 Two other proteins were upregulated in the dataset, prolactin-inducible protein was upregulated after 430 both PG and cinnamaldehyde mouth rinsing whilst neutrophil defensin 1 (α-defensin) was upregulated 431 in response to PG (Table 4). Cinnamaldehyde (180 ppm) resulted in the down-regulation of IgG-3 chain 432 C region, caspase recruitment domain-containing protein 10 (CARD10) (also downregulated in 300 ppm 433 cinnamaldehyde) and phosphoglycerate kinase 1 (PGK1). IgG-3 chain C region was also 434 downregulated in response to nonivamide.

Discussion 436
In this study we have found that mouth rinsing with menthol or nonivamide increases WMS flow rate 437 ( Figure 1 & Figure 2). These observations expand on the current reports in the literature that TRPV1 438 agonists, such as piperine, nonivamide, capsaicin and 6-gingerol can stimulate salivary secretion since 439 stimulation of salivary secretion by menthol has not previously been described. We have further found 440 that nonivamide can stimulate minor gland secretion. Cinnamaldehyde mouth rinse did not evoke a 441 salivary response even though it was perceived to be as intense or more intense than the menthol or 442 nonivamide mouth rinses (Supplementary data c), which indicates that salivary responses are TRP 443 agonist specific. The effect of a cinnamaldehyde mouth rinse was no greater than the vehicle PG but 444 both were greater than unstimulated WMS (Figure 1a). Nonivamide, menthol and PG increased outputs 445 of total protein in saliva suggesting that the protein composition and properties of saliva might be 446 altered. Cinnamaldehyde decreased protein secretion compared to the PG vehicle. This is likely due 447 to cinnamaldehyde diminishing the sialogogic properties of PG through a reaction between the 448 compounds rather than inhibiting the nerve mediated reflex PG induces as no inhibitory neurones exist 449 (45). The source of increased protein secretion is presumably salivary gland exocytosis of protein 450 storage granules but it may be that there are other contributions from within the oral cavity. In order to 451 investigate further, quantitative changes in salivary protein composition we implemented a bottom-up 452 mass spectrometry pipeline new to salivary proteomics, which led to the identification of novel whole 453 WMS proteome changes and specific protein changes in response to the TRP channel agonists studied. 454 From PCA we identified that the largest source of variation in the salivary proteome was between 455 subjects but that changes in the proteome were also caused by different mouth rinses (Figure 4). 456 Repeat analyses on subjects demonstrated that there was variation from day to day in response to 457 some of the mouth rinses. 458

459
The mass spectrometry pipeline applied in this study produced results that contribute to the salivary 460 proteome literature, since it identified proteins in saliva that have not previously been reported 461 (Supplementary Table). This may be due to the novel application of IEF using OFFGEL electrophoresis 462 with TMT labelled quantitative tandem mass spectrometry LC-MS/MS to salivary proteomics but may 463 also be the result of searching against updated databases or inter-personal differences in salivary 464 composition, which has previously been observed to have a larger coefficient of variation than intra-465 personal variation (46). Three previous studies of WMS have used IEF in tandem mass spectrometry 466 (47-49), and a further study coupled it with mTRAQ quantification methodology (50). However, these 467 studies did not couple IEF with isobaric labelling such as TMT. It could be that the novel methodology 468 contributes to better identification of lower abundance proteins, or this could be a result of the 469 experimental stochasticity in bottom-up mass spectrometry approaches, the use of updated protein 470 sequence database or differences in raw data analysis software. Despite being in lower abundance, 471 the novel proteins are of sufficient length (median amino acid length being 897 and ranging from 97 to 472 7570) to produce detectable tryptic peptides. This suggests that the method is not just identifying small 473 proteins with a high abundance but proteins of a range of sizes with relative abundances ranging from 474 3.2% of total peptides to < 0.005% (Supplementary table). A bottom-up approach was implemented 475 with the intention to maximise the quantification of the salivary proteome. With 459 proteins quantified, 476 the coverage was limited when compared to other TMT quantification studies with more state of the art 477 equipment. Furthermore, good proteome coverage that also represents the variety of gene products 478 has been achieved in top-down and data independent acquisition proteomic studies and could be used 479 to further investigate the diversity of the salivary proteome (51, 52). 480

481
The presence of some lower abundance proteins appeared to be influenced by mouth rinsing, for 482 example CARD10 and phosphoglycerate kinase 1 (PGK1), which were 0.3 and 0.2% of total identified 483 peptides respectively (Table 4). This is the first time CARD10 has been identified in WMS. Both 484 CARD10 and PGK1 were downregulated specifically in response to cinnamaldehyde mouth rinsing:. 485 Despite there being no previous reports of association between CARD10 and cinnamaldehyde, there 486 have been previous reports of cinnamaldehyde inhibiting other caspase recruitment domain proteins in 487 mice and subsequent anti-inflammatory effects (10). Similarly, there have been no previous reports of 488 an association between cinnamaldehyde and PGK1. However, anti-angiogenesis properties of 489 cinnamaldehyde and cinnamon extract have been previously reported (12)(13)(14). The observation of 490 down-regulation of CARD10 and PGK1 could be preliminary evidence that the anti-inflammatory and 491 bactericidal effects of cinnamaldehyde extend to short term mouth rinsing in the oral cavity. 492 493 Upregulation of cystatin S in the WMS secreted in response to nonivamide was detected by mass 494 spectrometry and western blotting ( Figure 5). Despite significant sequence homology between the 495 salivary cystatins, the peptides assigned to S, SN and SA were unique to each protein. Furthermore, 496 the antibody used in western blotting had a reasonable specificity for cystatin S, with 30% and 5% cross 497 reactivity to cystatins SN/SA or D/C respectively. To further increase the confidence in specificity, a top 498 down approach could be used as demonstrated in the literature (53). Greater quantities of cystatin S in 499 saliva could result in an improvement in mucosal adhesion, a property of saliva important in mouthfeel 500 and xerostomia. Cystatin S has been shown to interact with oral mucosal surfaces and play a role in 501 the formation of protein pellicles in vitro on hydrophobic surfaces that mimic the mucosa (54). Coupled 502 with previous observations that the rheological properties of saliva are modified by nonivamide (29, 55), 503 mouth rinsing with nonivamide as a treatment for xerostomia warrants further study. Increased cystatin 504 S expression may have other potential benefits for oral health. due to inhibition of cysteine protease 505 activity, as indicated by significant enrichment of the "negative regulation of cysteine-type 506 endopeptidase activity" GO. The upregulation of the GO for cysteine protease inhibition mirrors the 507 western blotting findings and work in the literature (56, 57). Cystatin S has been shown to inhibit 508 proteolytic activity in the culture supernatant of P. gingivalis (58), a Gram negative bacterial species 509 that produces the gingipain class of cysteine proteases which are implicated in periodontal disease 510 (59). Additionally, cystatin S, as well as prolactin-inducible protein, upregulation could improve 511 acceptance of bitter taste as indicated by the GO enrichment "detection of chemical stimulus involved 512 in sensory perception of bitter taste" (60). This suggests that TRPV1 agonists could be used to promote 513 the consumption of bitter foods, the reduced consumption of which has been implicated in the health, 514 dietary intake and weight of "super tasters" (61). 515 516 This study is the first to demonstrate an acute salivary cystatin S response to TRPV1 agonists in 517 humans ( Figure 5). A cystatin S-like protein response to capsaicin has been demonstrated in rats fed 518 on a capsaicin-adulterated diet; the presence of a new protein in rat saliva was demonstrated and the 519 protein found to have cystatin S-like properties such as inhibition of cysteine protease activity (57). In 520 the rat increased cystatin S-like protein levels enhanced consumption of a capsaicin rich diet and it was 521 hypothesised that this response may be triggered by irritation of the oral mucosa (56). Although these 522 studies, along with the current study, both show increases in cystatin S and cystatin S-like proteins in 523 saliva, the time scales over which the phenomenon occurs are significantly different. The current study 524 shows the reversible increase within two minutes of nonivamide mouth rinsing whilst in the studies in 525 rat the increase was observed after three days of capsaicin-adulterated diet, suggesting different 526 mechanisms are responsible. The increase in cystatin S levels in WMS in the current study must be 527 due to the release of preformed protein as it takes 30 minutes for newly synthesised protein containing 528 vesicles to pass from the rough endoplasmic reticulum to the condensing vacuoles in secretory cells 529 (62). 530

531
The identification of proteins regulated across all mouth rinses alongside proteins only regulated in 532 response to one mouth rinse suggests, in agreement with the total protein secretion data, that there are 533 different mechanisms responsible for the regulation of proteins in WMS. Furthermore, some of the 534 proteins are known to be produced by the salivary glands whereas others are non-salivary proteins. 535 The upregulation of salivary cystatins (S, SN and SA) may reflect a preferential stimulation of the 536 submandibular/sublingual glands, the primary producers of salivary cystatins (63) agonists. There is a precedence in sensory science for responders/non-responders, such as in the case 550 of the detection of the bitter compound PROP which is associated with the expression of the TAS2R28 551 bitter receptor gene (67). Although the comparison seems to be limited by the fact that participants in 552 the current study do have a sensory perception of the TRP agonists, the mechanism for salivary 553 secretion in response to TRP agonist detection is yet to be elucidated and unknown genetic factors 554 could be responsible for the prevalence of salivary non-responders to TRP agonists despite a sensory 555 perception. A breakdown of the dataset shown in Figure 1a reveals that only 2 of the 19 participants 556 given a TRP containing mouth rinse did not exhibit an increase in WMS flow rate (as defined by a flow 557 rate 150% that of unstimulated flow rate). This suggests that the prevalence of non-responders in the 558 population is lower than the 50% suggested in the proteomics study. 559

560
In summary this study provides the first evidence for stimulation of salivary secretion by a non-TRPV1 561 TRP channel agonist. Increased minor gland secretion may be a direct action of the TRP agonists on 562 submucosal salivary glands alongside nerve-mediated mechanisms. Furthermore, novel changes in the 563 proteome of the saliva secreted in response to the TRPV1 agonist nonivamide were identified by mass 564 spectrometry and supported by western blotting. These findings suggest that TRP channel agonists 565 can be explored as potential candidates for altering salivary secretion, particularly in subjects with 566 xerostomia and reduced levels of saliva. 567 568 Acknowledgements 569 The authors would like to acknowledge a BBSRC PhD Case Studentship subsided by Symrise AG as 570 the source of funding for the work.   c) Intensity of the cystatin S band on a western blot, relative to the amylase western blot band intensity, in WMS collected after a 30 second TRP agonist mouth rinse normalised to unstimulated saliva (Mean±SEM; n = 6).