Accepted Manuscript

The role of saliva in aroma release and perception

Sarah Ployon, Martine Morzel, Francis Canon

PII: S0308-8146(17)30055-9
DOI: http://dx.doi.org/10.1016/j.foodchem.2017.01.055
Reference: FOCH 20447

To appear in: Food Chemistry

Received Date: 19 August 2016

Revised Date: 14 December 2016
Accepted Date: 12 January 2017

Please cite this article as: Ployon, S., Morzel, M., Canon, F., The role of saliva in aroma release and perception,
Food Chemistry (2017), doi: http://dx.doi.org/10.1016/j.foodchem.2017.01.055

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1 The role of saliva in aroma release and perception
2
3 Sarah Ployona , Martine Morzela,, Francis Canona*
4
5 a Centre des Sciences du Goût et de l'Alimentation, CNRS, INRA, Université de Bourgogne
6 Franche-Comté, F-21000 Dijon, France
7
8
9 Corresponding author: francis.canon@inra.fr ; tel: +33 (0) 3 80 69 35 29
10
11 Author email addresses: sarah.ployon@gmail.com; martine.morzel@inra.fr;
12 francis.canon@inra.fr

1
13 Abstract

14 Aroma perception is an important factor driving food acceptance. Volatile organic compounds

15 (VOCs) are released from the food matrix and then reach the receptors located in the nasal

16 cavity, leading to their perception. These steps are closely dependent on the physicochemical

17 properties of the volatile compounds and the food matrix, but also on human physiology. Among

18 the different physiological parameters involved, the literature reports that saliva has various

19 effects on VOCs and therefore appears as a major actor impacting the perception of aroma.

20 This article reviews how saliva takes part in aroma release, considering both in vitro and in vivo

21 approaches, and how it may affect perception. It describes the direct mechanisms (molecular

22 interactions, enzymatic conversion, salting-out effect, dilution) involving salivary components

23 (salts, proteins including enzymes, microbiota) that can modify the release of aroma

24 compounds. It also considers the indirect impact of saliva, such as changes of aroma diffusion

25 through modification of the physicochemical properties of the food matrix.

26

27 Keywords: Saliva, aroma release, interactions, enzymatic conversion, mucins, volatile organic

28 compounds, salivary proteins, flavour

29

2
30 1. Introduction

31 Retronasal olfaction is a part of flavor, together with taste and trigeminal sensation, and it is a

32 key factor driving food choices and acceptability. The flavor modality corresponds to the

33 perception of aromas after activation of olfactory receptors by the binding of aroma compounds.

34 Different factors impact aroma release from food and can therefore modulate the final

35 perception. Two main factors affecting aroma release are inherent to the food product, namely

36 the chemical properties of volatile compounds and the physicochemical properties of the food

37 matrix. For the first factor, the vapor pressure is an important characteristic involved in aroma

38 partitioning, which is defined as the ratio of concentration between a gas phase and a liquid

39 phase for one compound at equilibrium. It has also been reported that aroma partitioning can

40 vary with hydrophobicity (Buttery, Ling, & Guadagni, 1969) or other properties such as polarity

41 of the molecule (Tromelin, Merabtine, & Andriot, 2010). For the second factor, food matrices

42 exhibit a large range of physicochemical properties due to different physical states and

43 composition. These characteristics are important factors affecting aroma release (Boisard, et al.,

44 2014; Frank, et al., 2015).

45 In contrast to odorants, which reach the receptors by the orthonasal route, aroma compounds

46 are released in the oral cavity and then follow the retronasal route. Differences between

47 retronasal and orthonasal perception of the same odorant molecule have been reported,

48 indicating an influence of oral processing and oral physiology on aroma perception (Linforth,

49 Martin, Carey, Davidson, & Taylor, 2002). For example, opening of the soft palate (velum

50 palatinum) during mastication, which allows the transfer of volatiles to the nose cavity

51 (Buettner, Beer, Hannig, & Settles, 2001), influences the timing and extent of aroma release. It

52 was thus shown that the highest amount of aroma delivery occurs during the first expiration

53 after swallowing (Buettner & Schieberle, 2000; Linforth & Taylor, 2000). Total aroma release

54 from model cheeses was positively correlated with number of swallows (Boisard, et al., 2014),

55 demonstrating the importance of taking into account oral processing to understand aroma

56 perception.

3
57 Saliva also plays a significant role in aroma release. In vitro experiments have revealed

58 differences in aroma release from coffee or wine when water, human or artificial saliva was

59 added to the food matrix (Genovese, Caporaso, Civitella, & Sacchi, 2014; Munoz-Gonzalez, et al.,

60 2014). In an in vivo experiment aiming at determining the relative contribution of the main food

61 oral processing events in aroma release during cheese consumption, it was found that aroma

62 release was poorly linked to the composition of resting saliva but it was explained by the

63 characteristics of stimulated saliva (Feron, Ayed, Qannari, Courcoux, Labouré, & Guichard,

64 2014), emphasizing the importance of saliva composition and flow. Saliva is made of 98% water,

65 and contains salts and more than a thousand proteins (Denny, et al., 2008), among which

66 mucins, histatins, statherin, IgA, proline rich proteins and enzymes like α-amylase, carbonic

67 anhydrase IV and lysozyme. Therefore, different mechanisms, such as non-covalent or covalent

68 binding, enzymatic reactions or degradation of the food matrix should be taken into account

69 when considering the role of salivary proteins in aroma release, as illustrated in Figure 1. Saliva

70 secretion is characterized by both intra-individual and inter-individual variability. Indeed,

71 unstimulated flow rate is on average 0.3 mL/min but it can increase to 7 ml/min during

72 stimulation, especially during chewing (Dawes, 1969; Dawes & Dong, 1995). Inter-individual

73 variability of saliva impacts both the flow and the composition of saliva. For example, diet can

74 modify saliva composition with an impact on amylase activity (Mejean, et al., 2015). Other

75 factors impacting on saliva inter-individual variability are age, pathological conditions,

76 medications etc. (Humphrey & Williamson, 2001). These differences in terms of composition

77 and flow could be at the origin of inter-individual variation of aroma perception (Piombino, et

78 al., 2014; Repoux, Semon, Feron, Guichard, & Laboure, 2012b).

79 This review focuses specifically on the impact of saliva on aroma release, considering the

80 different mechanisms by which it may modify the amount, the nature or the release kinetic of

81 aromatic molecules in the air of the oral cavity. First, physicochemical interactions between

82 salivary proteins and aroma molecules and their effect on aroma release will be reviewed. Then,

83 we will consider in-mouth enzymatic reactions, both of host and microbial origin, leading to the

4
84 production of odorous volatiles from non-odorant precursors or to the conversion of aroma

85 molecules. A possible correlation between saliva characteristics and aroma perception will be

86 discussed. Finally, we will review how saliva can influence aroma release from food matrices

87 through dilution or modification of the matrix or food bolus properties.

88
89 1. Interactions between saliva components and aroma molecules

90 1.1. Binding of aroma molecules to salivary proteins

91 Proteins have demonstrated their ability to interact with aroma compounds, and consequently

92 to modify aroma release and perception (Guichard, 2006). This capacity is determined by their

93 conformations and physicochemical properties, themselves dependent on the proteins’ peptide

94 sequence, post-translational modifications and environment. For example, the structure of β-

95 lactoglobulin includes a central hydrophobic pocket, allowing the binding of hydrophobic

96 ligands such as aroma compounds (Tavel, Moreau, Bouhallab, Li-Chan, & Guichard, 2010). In this

97 case, interactions are due to hydrophobic effects, but other kinds of non-covalent interactions

98 can also occur between aroma compounds and proteins, such as electrostatic interactions

99 (including hydrogen bonds) or Van der Waals forces (Lubbers, Landy, & Voilley, 1998).

100 Proteomic analyses have revealed that saliva contains at least 1166 proteins (Denny, 2008), the

101 most abundant being mucins and α-amylase. Therefore, numerous studies have used artificial

102 salivas containing only these two proteins, alone or in combination, to investigate the effect of

103 saliva on aroma release.

104 Salivary mucins exhibit high molecular weight. Their peptide chain is highly glycosylated and

105 has several hydrophobic domains. Mucins are largely responsible for the viscous and elastic

106 properties of saliva. These properties allow saliva to fulfil several functions, such as lubrication

107 of oral surfaces during the actions of eating or talking, and maintenance of a hydrated layer

108 covering the epithelium (Bansil & Turner, 2006). Different mucins have been tested to formulate

109 artificial saliva. The effect of pig gastric mucins and mucins from bovine submaxillary glands on

110 the release of four aroma compounds has been compared, and no significant difference was

5
111 observed (Friel & Taylor, 2001). Mucins are globally negatively charged proteins, due to the

112 sulfate groups on the oligosaccharide side chains, leading to repulsion of mucin molecules

113 between each other. The pH and the presence of salts, which can cause a charge-shielding effect,

114 can affect the charge of the protein and lead to mucin self-aggregation and modification of the

115 viscosity. Friel and Taylor (2001) investigated the effect of mucin on the release of fourteen

116 volatile compounds while taking into account the effect of salt and ionic strength. Volatiles were

117 classified into three groups according to their behaviors. The first group was not affected by the

118 presence of mucin in contrast to the second and third groups, suggesting a binding between

119 mucin and compounds of the 2nd and 3rd groups. The presence of salts did not modify the effect

120 of mucin for the second group, while differences were observed for the molecules classified in

121 the third group. The types of interactions between mucin and volatiles therefore probably

122 differed depending on the volatile structure.

123 The retention by mucin of aroma molecules belonging to ketones and esters families increases

124 as a function of the aliphatic chain length, suggesting the involvement of hydrophobic effects for

125 these molecules (Pages-Helary, Andriot, Guichard, & Canon, 2014). However, the same authors

126 did not exclude the involvement of other types of non-covalent interactions to fully explain their

127 results. For a group of molecules representing different chemical classes, Friel et al. (2001) did

128 not observe a correlation between molecule retention and their log P value (Friel, et al., 2001).

129 Moreover, addition of sugar to mucin solution affected differently the retention of the different

130 molecules by the protein. This could be due to the existence of different types of binding sites of

131 aroma compounds on mucins, which are differently affected by competition between aroma

132 molecules and sugars.

133 Establishment of covalent interactions between aroma compounds and mucins can also occur.

134 For example, release of aldehydes is strongly affected by the presence of mucins (Friel, et al.,

135 2001; van Ruth, Roozen, & Cozijnsen, 1995). Aldehydes and ketones may bind to protein

136 through formation of Schiff bases. However this mechanism was excluded on the basis that

6
137 compounds from different chemical classes acted similarly to aldehyde in the presence of

138 mucins (Friel, et al., 2001).

139 The second abundant protein widely used in aroma release studies is α-amylase. This globular

140 protein exhibits a well-defined three-dimensional structure (Kandra & Gyémánt, 2000) unlike

141 mucins. In their study, Pagès et al. showed that human α-amylase could decrease release of

142 ketones and esters (Pagès-Hélary, Andriot, Guichard, & Canon, 2014). Retention was negatively

143 correlated to hydrophobicity of the molecules, which suggests binding through hydrophobic

144 interactions between hydrophobic domains of the protein and aroma compounds. However,

145 there was no cumulative effect of mucin and α-amylase for saliva containing both proteins. The

146 authors proposed that protein‒protein interactions could preclude the access to a part of the

147 binding sites of the proteins, modifying the total number of binding sites available for aroma

148 molecules (Pagès-Hélary, et al., 2014).

149 Regarding differences between artificial and human saliva, Van Ruth et al. (2000) did not

150 observe any significant difference on aldehyde release from rehydrated diced bell peppers in

151 presence of whole human saliva or artificial saliva (Van Ruth & Roozen, 2000). In contrast Pagès

152 et al. (2014) demonstrated a stronger effect of human whole saliva than artificial saliva on ester

153 release. The difference observed between the two studies may be explained by variability in

154 human saliva composition and/or by the different chemical natures of the aroma analysed

155 (esters vs aldehydes), which may be more or less prone to interaction with proteins or

156 enzymatic conversion. This observation highlights the importance of other salivary proteins that

157 are not present in artificial saliva. Enzymes in particular may convert aroma compounds into

158 different molecules, as described in the next section.

159 Among other salivary proteins that could impact the release of aroma compounds, lipocalin 1

160 (also referred to as LCN1 or Von Ebner gland protein) deserves special attention. LCN1 belongs

161 to the superfamily of lipocalins. Lipocalins are small-secreted proteins characterized by their

162 ability to bind small hydrophobic molecules. Lipocalins family proteins exhibit a common

163 pattern structure with β-barrel forming a hydrophobic core formed by the side-chains of both

7
164 hydrophobic and polar amino acids (Flower, 1996). Although LCN1 concentration in saliva is

165 quite low, around 100 ng/mL (Mounayar, Septier, Chabanet, Feron, & Neyraud, 2013), and its

166 ability to bind aroma has not been reported, this protein could interact with aroma compounds,

167 since other proteins from the lipocalin family, namely β-lactoglobulin and odorant binding

168 proteins (OBPs), already exhibit this property. For example, β-lactoglobulin, present in dairy

169 products, has been extensively studied for its capacity to bind aroma compounds (Tavel, et al.,

170 2010). OBPs present in the nasal mucus of many mammalian species, also bind aroma

171 compounds and assist the transfer of odorants from the gas phase to the olfactory receptors.

172

173 1.2. Aroma enzymatic conversion by human saliva

174 Numerous salivary proteins are enzymes catalysing chemical reactions, which can potentially

175 involve aroma compounds depending on their chemical structure. Salivary aroma-converting

176 enzymes may originate from the salivary glands, oral tissues or even microbiota. Some studies

177 have reported enzymatic degradations of different aroma compounds in the presence of human

178 saliva. Pagès et al. (2014) reported a stronger effect of human saliva compared to artificial saliva

179 composed of mucin and alpha-amylase on the release of esters and the production of their acid

180 forms. These observations led to the conclusion that esters undergo hydrolysis in the presence

181 of saliva, which can be attributed to enzymatic activity of an esterase in whole saliva (Pagès-

182 Hélary, et al., 2014). Buettner et al. (2002) previously described degradation of esters by human

183 saliva (Buettner, 2002b). This degradation was inhibited after thermal denaturation of saliva,

184 confirming its enzymatic origin. The same authors reported a very low inter-individual

185 variability, but this result was obtained with only two subjects and needs to be confirmed. In

186 both articles, it was reported that the extent of enzymatic degradation varies according to the

187 chemical structure of esters. In particular, it increases with the aliphatic chain length up to 7

188 carbons (Buettner, 2002b; Pagès-Hélary, et al., 2014).

189 Regarding the identification of the enzyme(s) at the origin of this activity, it has been reported

190 that there is heterogeneity of esterases present in saliva. Three groups of esterase have been

8
191 identified in human saliva (Lindqvist & Augustinsson, 1980), all mainly secreted by the

192 submandibular glands (Lindqvist & Augustinsson, 1975). An aldehyde dehydrogenase is also

193 present in saliva: this enzyme exhibits esterolytic activity (8390 mIU/mg protein), as do other

194 aldehyde dehydrogenases (Sreerama, Hedge, & Sladek, 1995).

195 As indicated by its name, this enzyme also has enzymatic activity on aldehydes. Aldehyde

196 dehydrogenases catalyze the pyridine nucleotide-dependent oxidation of aldehydes to acids. The

197 hydrogen acceptor is usually NAD, although NADP can also be used depending on the enzyme.

198 The reaction is essentially irreversible. The main salivary aldehyde dehydrogenase probably

199 consists of the isozyme ALDH3A1, also called ALDH-3 (Giebultowicz, Wolinowska, Sztybor,

200 Pietrzak, Wroczynski, & Wierzchowski, 2009). However, another ALDH enzyme, known as

201 ALDH6 or ALDH1A3, has also been identified in salivary glands (Hsu, Chang, Hiraoka, & Hsieh,

202 1994), although its activity in saliva has not been confirmed. Interestingly, Giebułtowicz et al.

203 (2009) have compared the salivary ALDH activity toward a series of aromatic aldehydes with

204 that of the purified ALDH3A1 and reported no significant difference (Giebultowicz, et al., 2009).

205 Therefore, it is likely that the salivary ALDH activity is due to ALDH3A1. The ability of this

206 enzyme to oxidize aldehydes into carboxylic acids has been reported for numerous aldehydes

207 including saturated aliphatic aldehydes (Pappa, Estey, Manzer, Brown, & Vasiliou, 2003) and

208 aromatic aldehydes (Giebultowicz, et al., 2009). A large variability in the level of the salivary

209 ALDH activity has been reported, which could not be attributed to gender, ethnicity, alcohol

210 consumption, use of tobacco or consumption of animal products. However, it appeared to be

211 correlated with the consumption of coffee or vegetables such as broccoli (Sreerama, et al., 1995).

212 Buettner et al. (2002) also confirmed that aldehydes undergo enzymatic degradation in the

213 presence of saliva. They reported the reduction of aldehydes to alcohols but did not observe the

214 formation of the corresponding acids. A tendency to higher degradation was observed for the

215 less polar compounds. However, the exact mechanism at the origin of this reduction was not

216 determined (Buettner, 2002a).

9
217 Aldose reductase catalyzes the NADPH-dependent reduction of a wide variety of carbonyl-

218 containing compounds to their corresponding alcohols with a broad range of catalytic

219 efficiencies. Therefore, this enzyme and aldehyde reductase (Bohren, Bullock, Wermuth, &

220 Gabbay, 1989) are two potential candidates to catalyze the reduction of aldehydes into alcohols.

221 However, proteomic investigations did not report the presence of aldose reductase in human

222 saliva (Denny, et al., 2008).

223 Thiols have been also reported to be sensitive to degradation in the presence of non-denatured

224 human saliva, while they are not affected by heated saliva, suggesting that they are degraded by

225 salivary enzymatic activity (Buettner, 2002b). Peroxidases are enzymes able to catalyze

226 oxidation of a variety of substrates including thiols (Svensson, 1988). Lactoperoxidase is a

227 peroxidase that is secreted in saliva and milk in humans (Gothefors & Marklund, 1975). The

228 peroxidase activity of saliva shows high between-subject variability. It depends for example on a

229 subject’s physiological status and age (Buettner, 2002b). In samples of palm wine, esters,

230 aldehydes and thiols were also observed to be highly degraded in the presence of human saliva

231 while no noticeable effect was observed for pyrazines, pyrrolines and most alcohols (Lasekan,

232 2013). These observations are in accordance with those of Buettner (Buettner, 2002a, 2002b)

233 and demonstrate that degradation of aroma compounds by salivary enzymes can also be

234 observed in vitro in real food matrices.

235 Regarding the effect of enzymatic degradations on aroma perception, two effects have to be

236 considered: the decrease of the concentration of the degraded molecules, but also the formation

237 of new volatile odorant metabolites.

238 1.3. Impact of oral microbiota on aroma release

239 The oral microbial population presents a huge diversity. At the species level, 707 entries are

240 currently documented (Mark Welch, Rossetti, Rieken, Dewhirst, & Borisy, 2016). Comparing the

241 bacterial population of 9 mouth sites of more than 200 healthy individuals, it was shown that

242 most bacteria species are specialized in one oral habitat (Eren, Borisy, Huse, & Mark Welch,

243 2014). Analysis of the salivary microbiome has revealed a high diversity between individuals,

10
244 but relatively little variation between geographical locations around the globe (Nasidze, Li,

245 Quinque, Tang, & Stoneking, 2009). Saliva has a strong influence on the oral microbiota. Indeed,

246 saliva acts as buffer, thereby maintaining a pH favorable to bacterial growth. Saliva favors

247 clearance of fermentable substrates, limiting the growth of bacteria with a saccharolytic mode of

248 metabolism. Antimicrobial factors present in saliva (i.e. statherin, histatins, defensins, lysozyme)

249 also contribute to the regulation of the microbial populations (Marsh, Do, Beighton, & Devine,

250 2016).

251 Oral microbiota is one factor at the origin of oral malodor due to its ability to produce odorant

252 molecules (Krespi, Shrime, & Kacker, 2006). Oral malodor is mainly due to the putrefactive

253 degradation by Gram-negative anaerobic bacteria of peptides and proteins found in saliva,

254 gingival crevicular fluid, blood and residues of food found in the oral cavity. These degradations

255 lead to the production of volatile sulfur compounds (VSC), especially hydrogen sulfide and

256 methyl mercaptan from cysteine/cysteine and methionine, respectively (Codipilly & Kleinberg,

257 2008). The most commonly identified bacteria producing VSCs are Porphyromonas, Prevotella,

258 Actinobacillus, and Fusobacterium (Krespi, et al., 2006). Microbial metabolism produces a large

259 variety of metabolites, and many of them are odorants (Takahashi, 2015).

260 The production of odorant molecules by oral microbiota from non-odorant precursors has been

261 observed for molecules containing cysteine. Cysteine-S-conjugates from fruit are odourless

262 molecules, which are transformed into odorant VSC such as thiols by oral bacteria, especially

263 Fusarium nucleatum (Starkenmann, Le Calvé, Niclass, Cayeux, Beccucci, & Troccaz, 2008), a

264 species also identified as responsible for oral malodor (Krespi, et al., 2006). More recently, the

265 ability of human oral microbiota to produce odorant molecules from odorless grape glycosidic

266 aroma precursors was demonstrated in in vitro and ex vivo conditions, using cultures of

267 representative oral bacteria. The hydrolysis of odorless glycosidic precursors was highly

268 variable and produced different types of odorant molecules belonging to different chemical

269 families (terpenes, benzenic derivatives and C6-alcohols), depending on the bacteria species

11
270 (Muñoz-González, Cueva, Pozo-Bayón, & Moreno-Arribas, 2015). The conversion of ethanol into

271 acetaldehyde by oral streptococci has also been evidenced (Kurkivuori, et al., 2007).

272 Regarding the effect of the oral microbiota composition on aroma release, even though oral

273 bacterial composition differences were observed between two populations (obese vs normal

274 weight subjects), there was no link with differences of aroma release from wine between the

275 two groups (Piombino, et al., 2014).

276 Table 1 summarizes the effects of specific salivary components (salts, proteins including

277 enzymes, microbiota) on aroma release.

278 1.4. Potential sensory relevance

279 Usually, consumption of food and beverages leads to the release of a mixture of aroma

280 compounds and not to the release of a monomolecular aroma. The perception process of a

281 molecular aroma has been well characterized. Its quality and intensity are mainly linked to the

282 chemical structure (Stevens, 1960) and the concentration of the molecule (Chastrette, 1997),

283 respectively. However, the perception of mixtures of aroma compounds is far more complicated,

284 as exposed in a review dealing with odor perception (Thomas-Danguin, et al., 2014), whose

285 concepts can be extended to aroma perception. Briefly, the perception of mixtures of odorant

286 compounds can either be homogeneous, leading to the perception of a single odor, or

287 heterogeneous, when several odors are perceived. In the homogeneous case, the specific odors

288 of the different molecules are not individually perceived and blend into a new odor. The mixture

289 is called a blending mixture and the perception can be considered as configural or synthetic.

290 Configural perception is both specific to the molecules composing the blending mixture but also

291 to their respective proportions (Sinding, et al., 2013). There is another case that can be

292 considered as homogeneous, when one molecule presents a strong intensity and masks the

293 quality of the other molecules. When the perception of the mixture is heterogeneous, the specific

294 odor of at least some molecules in the mixture can be identified. In this case, the processing is

295 qualified as analytical or elemental. In that case, the odor quality of the mixture is linked to the

296 odor intensity of molecules.

12
297 Previous paragraphs of the present review show that different phenomena occurring in the

298 mouth affect the release of aroma compounds through two main mechanisms: binding of aroma

299 compounds to salivary proteins and enzymatic conversion of odorant or non-odorant molecules

300 into new odorant molecules. Binding of aroma compounds modifies the concentration of the

301 released molecules. Therefore, in the case of the perception of a monomolecular aroma, it will

302 only affect the intensity of the aroma as a function of the binding affinity of the molecules to its

303 salivary partner(s) and the concentration of such partner(s). In the case of mixtures, the

304 different molecules composing the mixture will also be differently affected as a function of their

305 binding affinity. In configural perception, changes in relative proportion of the blending mixture

306 can influence the mode of perception of the mixture and impact the specific odor quality of the

307 blending mixture (Sinding, et al., 2013). In contrast, in the case where one odorant molecule

308 masks the others, if the main molecule is not affected or poorly affected, the quality and intensity

309 of the odor will not be affected. When the mixture is heterogeneous, the quality of the mixture

310 could also be affected as it depends on the odor intensity of the components.

311 Enzymatic conversion will both generate new aroma molecules and decrease the amount of the

312 precursor, which can be an odorant. The release of new molecules will consequently affect the

313 quality of the aroma, except in the case of a homogenous mixture when a molecule has a strong

314 odor and covers the odor of the other molecules.

315 It is also important to take into account the time dimension of perception and release. The

316 majority of the in vitro experiments are performed after exposure to saliva from a few minutes

317 to one hour and generally at thermodynamic equilibrium, which does not represent the

318 condition during food consumption (Pages-Helary, et al., 2014; Van Ruth, Grossmann, Geary, &

319 Delahunty, 2001; Van Ruth, et al., 2000; Van Ruth, et al., 1995). While these studies suggest an

320 effect of saliva on aroma release, using a dynamic mouth model system and monitoring aroma

321 release in the few seconds after saliva addition, Rabe et al. (2004) did not observe any effect of

322 artificial saliva on initial flavour release from solutions (Rabe, Krings, & Berger, 2004). In

323 general, the highest release of aroma molecules occurs a few seconds after swallowing, within

13
324 the first breath (Hodgson, Linforth, & Taylor, 2003) as a consequence of the velum opening. It is

325 possible that the effect of saliva is limited during this initial step. However in vivo measurements

326 of aroma compounds’ concentrations in the nasal cavity have revealed temporal differences in

327 aroma release depending on the molecule, following aroma solutions (Buffo, Rapp, Krick, &

328 Reineccius, 2005; Linforth, et al., 2000) or wine consumption (Buettner, 2004), with release

329 time ranging overall from 0 to 240 s. This longer release of aroma compounds is thought to be at

330 the origin of a phenomenon named aroma persistence, whose mechanisms remain unknown. An

331 in vivo study following aroma release in the nasal cavity of several subjects reported a

332 correlation between the release kinetic of ethyl hexanoate and the saliva composition of the

333 subjects. Moreover, in-mouth persistence of ethyl hexanoate was significantly lower after mouth

334 rinsing. This observation did not occur for ethyl propanoate, which is a more hydrophilic

335 compound. Authors suggested that reversible binding of the molecule to salivary proteins could

336 be responsible for ethyl hexanoate persistence (Canon, Pages-Helary, & Guichard, 2015). Indeed,

337 a residual salivary film with thickness from 10 µm to 70 µm persists on the different oral

338 surfaces, after swallowing (DiSabato-Mordarski & Kleinberg, 1996). Furthermore, some salivary

339 proteins, especially salivary mucins, are anchored onto epithelial cells and form a thin structure

340 named the mucosal pellicle. Consequently, a mixture of aroma molecules and residual saliva

341 coats the oral surfaces even after swallowing, allowing the establishment of slower processes.

342 From a sensory point of view, the desorption of aroma compounds from salivary proteins or oral

343 mucosa is a slow process that could participate in aroma persistence (Hussein, Kachikian, &

344 Pidel, 1983; Rabe, et al., 2004), even though to date, the role of saliva in the phenomenon is not

345 clearly established. In in vitro conditions, initial flavour release from solutions was not

346 influenced by artificial saliva or the presence of oral mucosa, but the authors did not exclude

347 that adsorption/desorption onto the mucosa could be involved in the persistence phenomenon

348 (Rabe, et al., 2004).

349 Enzymatic conversion of molecules may lead to the production of new odorant molecules.

350 Taking into account that the longer the molecule is absorbed onto the mucosa, the higher the

14
351 probability that the precursor is converted, the relative proportion of the different molecules

352 could strongly change over time. Therefore, the odorant quality of “aftertaste” could be quite

353 different than those of the initial breath. For example, ester concentration was affected by

354 degradation by whole saliva during 180 s (Pagès-Hélary, et al., 2014), while the corresponding

355 acid was produced over the same period.

356 To date, no direct links have been established between salivary composition and aroma

357 perception. However, adsorption/desorption mechanism of aroma compounds onto/from

358 salivary components and production of odorant molecules from odorless precursor or by

359 conversion of aroma molecules are mechanisms likely to influence aroma perception, in

360 particular after swallowing.

361
362 2. Effect of saliva on aroma release from food matrix

363 To deeply understand the effect of saliva on aroma perception, it is important to consider the

364 food systems in which aroma molecules are initially present. Indeed, aroma compounds are

365 released from food matrices, which show a huge diversity of composition, structure, texture and

366 physicochemical properties. These characteristics impact the release of aroma compounds, but

367 also mastication and salivation. During consumption and mastication, dissolution in saliva and

368 changes in temperature can lead to transition in the physical states of food, changing from a

369 semisolid or solid to a liquid state. As a result, the release of aroma compounds in the mouth will

370 be impacted. Mastication, which corresponds to the breakdown of food, increases the

371 surface/volume ratio of food particles, and thereby improves the transfer of flavour molecules

372 to saliva and to air of the oral cavity according to Fick’s first law (Salles, et al., 2011; Van Ruth, et

373 al., 2000). It is also important to consider diffusion in the system. The diffusion rate depends on

374 the type of diffusion in the system. In a static fluid system, mass transport occurs by molecular

375 diffusion, which is mathematically described by the Ficks’s laws of diffusion. In this case the

376 diffusion and release of aroma compounds is affected by the viscosity (Guichard, Issanchou,

377 Descourvieres, & Etievant, 1991) but also by molecular interactions occurring between aroma

15
378 compounds and the molecules composing the food matrix (Bylaite, Adler-Nissen, & Meyer,

379 2005). However during mastication, mass transport takes place by eddy diffusion, leading to

380 higher release rates. In this case, aroma release is less affected by the viscosity of the matrix

381 (Hollowood, Linforth, & Taylor, 2002).

382 Saliva is incorporated into food to form the food bolus during oral processing. The amount of

383 saliva incorporated depends on the food matrix and can represent up to 96% of the bolus weight

384 (Repoux, et al., 2012a). This incorporation is an essential step of bolus formation (Salles, et al.,

385 2011). It affects the release of aroma compounds through direct and indirect effects. Regarding

386 the mechanisms involved in the direct effect, the first to consider is the solubilisation of the

387 compounds into saliva. Concentrations of molecules in the different phases, namely food, saliva

388 and air, are characterized by two inter-related food‒saliva and saliva‒air equilibria (Taylor,

389 2002) Consequently, the dilution of aroma molecules in saliva directly affects their release to the

390 oral cavity. Even if no significant differences were observed in total aroma release from coffee

391 with or without artificial saliva addition (Genovese, et al., 2014), differences were observed

392 when compounds were considered individually, highlighting the influence of the molecule

393 chemical nature. Different behaviors were also observed in vitro in polyphasic products. When

394 artificial saliva was added to water/oil emulsions, hydrophilic compounds were preferentially

395 diluted in saliva, which is essentially made of water, resulting in a lower release of these

396 compounds in the air phase. In contrast, hydrophobic compounds were mostly solubilized in the

397 oil phase, thus saliva addition did not impact on their partitioning with air (Odake, Roozen, &

398 Burger, 1998).

399 Effect of bite size can affect aroma release, likely through different dilution rates, but Genovese

400 et al (2015) also noticed during in vitro experiments that for the same ratio wine/saliva, sip

401 volume could affect aroma release as a function of chemical classes (Genovese, Moio, Sacchi, &

402 Piombino, 2015).

403 Another direct effect of saliva on the release of aroma compounds is due to noncovalent

404 interactions occurring between food matrix molecules and aroma compounds. The retention of

16
405 aroma molecules by salivary proteins has been frequently reported, but phenomena occurring in

406 real conditions are much more complex and the literature contains contradictory observations.

407 For example, although hydrophobicity was observed to be a significant parameter impacting

408 volatiles release from artificial and human saliva (Pagès-Hélary, et al., 2014; Van Ruth, et al.,

409 2001), correlation between hydrophobicity and release from food products showed discordance

410 between studies (Boland, 2004; Doyennette, de Loubens, Deleris, Souchon, & Trelea, 2011;

411 Flores & Olivares, 2008), and the rate of release of aroma compounds from dry fermented

412 sausages was not related with the compound’s hydrophobicity (Flores, et al., 2008). To explain

413 such results, it is necessary to consider the food systems used, since other phenomena due to

414 different physicochemical properties could take place simultaneously to non-covalent binding

415 between aroma compounds and food matrix molecules. As a result, saliva may have a different

416 effect according to the matrix studied. Regarding semisolid products, in gelatin and pectin gels,

417 an increase of the gel water content by artificial saliva addition increased hydrophobic volatiles

418 partitioning by enhancing the hydrophilic nature of the system and enhancing the diffusion of

419 flavor through the gel, whereas partitioning of more hydrophilic compounds decreased (Boland,

420 2004). By contrast, saliva addition increased the release of a hydrophilic compound (ethyl

421 propanoate) from a model cheese bolus (Doyennette, et al., 2011). Therefore, both the

422 physicochemical properties of aroma compounds and those of the non-volatile matrix

423 components have to be taken into account.

424 Saliva may also have indirect effect on aroma release. Indeed, saliva incorporation contributes to

425 bolus moistening and modifies the rheological properties of the food bolus, especially viscosity

426 and spreadability. It was observed in vitro, adding artificial saliva to dressings, that a higher

427 viscosity of the bolus increases aroma release to the gas phase (Odake, et al., 1998) by enhancing

428 spreadability of the matrix onto the measurement flask, thus increasing the surface exchange

429 with the gas phase. This finding was confirmed in vivo, where aroma release could be explained

430 partly by bolus spreadability after model cheese consumption. A higher spreadability enhanced

431 exposure to the gas phase and thus release of volatiles to the gas phase (Feron, et al., 2014).

17
432 Moreover, saliva contains enzymes which can modify the food matrix and bolus properties. For

433 example, addition of α-amylase (in the concentration range found in human saliva) to starch-

434 based systems increased the volatiles’ release through modification of the matrix viscosity

435 (Ferry, Hort, Mitchell, Lagarrigue, & Pamies, 2004).

436 Some salivary proteins are also able to interact with food matrix components resulting in a

437 modification of aroma release. This phenomenon was well studied in wines. Red wines differ

438 from white wines for example by their composition in tannins, which are able to bind salivary

439 proteins, especially proline-rich proteins (PRPs) and mucins, to form soluble and non-soluble

440 complexes (Canon, et al., 2013b). Mitropoulou et al. (2011) noticed that artificial saliva addition

441 to reconstituted red wines decreased the release of the more hydrophilic molecules but

442 enhanced that of hydrophobic volatiles. Authors proposed that the protein‒tannin complexes

443 could expel hydrophobic compounds from the matrix (Mitropoulou, Hatzidimitriou, &

444 Paraskevopoulou, 2011). Two other studies used both human and artificial saliva. A first study

445 demonstrated that aroma release from white wines was more affected by saliva addition than

446 for red wines. This lower effect of saliva on aroma release was attributed to the fact that

447 tannins‒salivary proteins complexes inhibited binding of aroma molecules to salivary mucins

448 (Genovese, Piombino, Gambuti, & Moio, 2009). In a more recent study, it was observed in

449 general a reduction of aroma release by saliva addition, but red wines were more affected than

450 white wines. In this case, authors suggested that tannin‒salivary protein complexes could retain

451 aroma molecules in hydrophobic cavities and decrease their release in the headspace, (Munoz-

452 Gonzalez, et al., 2014). In both studies, human saliva had a stronger effect than artificial saliva,

453 possibly due to specific interactions between tannins and proteins in real saliva, as these

454 proteins are not included in artificial saliva. Indeed, proline-rich-proteins have demonstrated a

455 high affinity for tannins, due to their extended conformation; the presence of several proline

456 clusters in their sequence leading to high protein‒tannins stoichiometries (Canon, et al., 2013a;

457 Canon, et al., 2013b). An important phenomenon to consider is the formation of large

18
458 PRP‒tannin aggregates (Canon, et al., 2013b), which could interact with aroma compounds

459 differently to isolated proteins and tannins.

460 Salivary proteins, especially mucins, can also interact with the sucrose contained in food

461 matrices. Mucins have a limited number of binding sites which are preferentially taken by

462 sucrose (Friel, et al., 2001). Thus competition between sucrose and aroma molecules can lead to

463 an increase of some aroma compounds’ release from sugar-containing solutions (Friel, et al.,

464 2001). Surprisingly, this did not happen in sugar containing gels (Boland, 2004), maybe due to

465 an effect of the viscosity.

466 The pH of some food matrices, like wine, is quite different from the pH of saliva. Despite saliva’s

467 strong buffering capacity, beverages can therefore affect pH in the mouth during consumption.

468 The structure and activity of proteins are pH-sensitive, in particular enzymes, but other salivary

469 proteins such as mucins can also be strongly affected by pH. Mucins show a conformational

470 transition at pH 4, from a random coil to a stiff worm-like chain, leading to gelation at acidic pH,

471 due to the exposure of mucin hydrophobic domains(Bansil, et al., 2006). Therefore, interaction

472 of aroma compounds with salivary proteins can also be affected by pH. Ionic strength is another

473 parameter that can strongly affect the structure of proteins and the interaction between

474 proteins. Indeed, proteins like mucins are strongly and globally negatively charged. Salts in

475 solution can cause a charge-shielding effect. The neutralisation of the coulomb repulsions

476 between units can lead to interactions between mucin units and modify the number of binding

477 sites available for aroma binding (Friel, et al., 2001) . During food consumption, modification of

478 the pH and ionic strength conditions can affect the conversion and binding mechanisms

479 previously presented.

480 This highlights once again the complexity of aroma release from the food matrix, which cannot

481 be explained by a single effect, but rather by combined effects of dilution, retention by non-

482 covalent interaction and modification of the physicochemical properties of the food matrix such

483 as viscosity.

484

19
485 Table 2 summarizes the main effects of saliva on aroma release from the food matrix.

486

487 Conclusion

488 Saliva is a complex biological fluid with numerous functions. The role of saliva in aroma release

489 is complex and cannot be explained by a single effect. To fully understand the role of saliva in

490 aroma release, it is necessary to take into account combined effects of dilution, retention by

491 salivary proteins, enzymatic conversion by salivary proteins or microbiota and modification of

492 the physicochemical properties of the matrix by saliva. Moreover, aroma compounds show a

493 huge diversity of chemical structures, resulting in a large diversity of physicochemical

494 properties. These properties modulate the effects of saliva. For example, volatiles partitioning

495 between oral gas phase, saliva and matrix are closely dependent on their chemical nature and

496 thus their affinity for each phase.

497 To our knowledge, the impact of saliva on aroma perception has never been directly

498 investigated. Extrapolation between in-nose aroma concentration measurements to aroma

499 sensory perception must be cautiously done since the amount of volatiles present in the nasal

500 cavity does not necessarily correlate well with sensory perception. However, the rates of volatile

501 release seem to correlate with sensory data (Baek, Linforth, Blake, & Taylor, 1999). With regards

502 to the potential effect of saliva on aroma perception, it appears crucial in the future to both

503 deeply understand these phenomena and to link them with aroma perception. Studies

504 combining sensory analysis, in-nose aroma release monitoring and saliva biochemical analysis

505 would be useful to better understand the role of saliva in aroma perception.

506

507 Acknowledgements

508 This study was funded by the French National Research Agency (Grant ANR-10-ALIA-

509 001MUFFIN). The French Ministry of Higher Education and Research provided the PhD

510 fellowship for Sarah Ployon.

511

20
512 Conflict of interest statement

513 The authors declare no conflict of interest.

21
514 References:

515 - Baek, I., Linforth, R. S. T., Blake, A., & Taylor, A. J. (1999). Sensory perception is related to
516 the rate of change of volatile concentration in-nose during eating of model gels. Chemical
517 Senses, 24(2), 155-160.
518 - Bansil, R., & Turner, B. S. (2006). Mucin structure, aggregation, physiological functions
519 and biomedical applications. Current Opinion in Colloid & Interface Science, 11(2–3),
520 164-170.
521 - Bohren, K. M., Bullock, B., Wermuth, B., & Gabbay, K. H. (1989). The aldo-keto reductase
522 superfamily - cDNA and deduced amino-acid sequences of human aldehyde and aldose
523 reductases. Journal of Biological Chemistry, 264(16), 9547-9551.
524 - Boisard, L., Andriot, I., Martin, C., Septier, C., Boissard, V., Salles, C., & Guichard, E. (2014).
525 The salt and lipid composition of model cheeses modifies in-mouth flavour release and
526 perception related to the free sodium ion content. Food Chemistry, 145, 437-444.
527 - Boland, A. (2004). Influence of gelatin, starch, pectin and artificial saliva on the release of
528 11 flavour compounds from model gel systems. Food Chemistry, 86(3), 401-411.
529 - Buettner, A. (2002a). Influence of Human Saliva on Odorant Concentrations. 2.
530 Aldehydes, Alcohols, 3-Alkyl-2-methoxypyrazines, Methoxyphenols, and 3-Hydroxy-4,5-
531 dimethyl-2(5H)-furanone. Journal of Agricultural and Food Chemistry, 50(24), 7105-
532 7110.
533 - Buettner, A. (2002b). Influence of Human Salivary Enzymes on Odorant Concentration
534 Changes Occurring in Vivo. 1. Esters and Thiols. Journal of Agricultural and Food
535 Chemistry, 50(11), 3283-3289.
536 - Buettner, A. (2004). Investigation of potent odorants and afterodor development in two
537 Chardonnay wines using the buccal odor screening system (BOSS). Journal of
538 Agricultural and Food Chemistry, 52, 2339−2346.
539 - Buettner, A., Beer, A., Hannig, C., & Settles, M. (2001). Observation of the swallowing
540 process by application of videofluoroscopy and real-time magnetic resonance imaging-
541 consequences for retronasal aroma stimulation. Chemical Senses, 26(9), 1211-1219.
542 - Buettner, A., & Schieberle, P. (2000). Changes in the Concentrations of Key Fruit
543 Odorants Induced by Mastication. In Flavor Release, vol. 763 (pp. 87-98): American
544 Chemical Society.
545 - Buffo, R. A., Rapp, J. A., Krick, T., & Reineccius, G. A. (2005). Persistence of aroma
546 compounds in human breath after consuming an aqueous model aroma mixture. Food
547 Chemistry, 89(1), 103-108.
548 - Buttery, R. G., Ling, L. C., & Guadagni, D. G. (1969). Volatilities of aldehydes, ketones, and
549 esters in dilute water solution. Journal of Agricultural and Food Chemistry, 17(2), 385-
550 389.
551 - Bylaite, E., Adler-Nissen, J., & Meyer, A. S. (2005). Effect of xanthan on flavor release from
552 thickened viscous food model systems. Journal of Agricultural and Food Chemistry, 53(9),
553 3577-3583.
554 - Canon, F., Milosavljević, A. R., van der Rest, G., Réfrégiers, M., Nahon, L., Sarni-Manchado,
555 P., Cheynier, V., & Giuliani, A. (2013a). Photodissociation and dissociative
556 photoionization mass spectrometry of proteins and noncovalent protein-ligand
557 complexes. Angewandte Chemie (International ed. in English), 52(32), 8377-8381.
558 - Canon, F., Pages-Helary, S., & Guichard, E. (2015). In nose concentration of aroma
559 compounds is modified by salivary protein composition and saliva stimulation. Chemical
560 Senses, 40(3), 237.
561 - Canon, F., Paté, F., Cheynier, V., Sarni-Manchado, P., Giuliani, A., Pérez, J., Durand, D., Li, J.,
562 & Cabane, B. (2013b). Aggregation of the Salivary Proline-Rich Protein IB5 in the
563 Presence of the Tannin EgCG. LANGMUIR, 29(6), 1926-1937.
564 - Chastrette, M. (1997). Trends in structure-odor relationships. SAR and QSAR in
565 Environmental Research, 6(3-4), 215-254.

22
566 - Codipilly, D., & Kleinberg, I. (2008). Generation of indole/skatole during malodor
567 formation in the salivary sediment model system and initial examination of the oral
568 bacteria involved. J Breath Res, 2(1), 017017.
569 - Dawes, C. (1969). The effects of flow rate and duration of stimulation on the
570 concentrations of protein and the main electrolytes in human parotid saliva. Archives of
571 Oral Biology, 14(3), 277-294.
572 - Dawes, C., & Dong, C. (1995). The flow rate and electrolyte composition of whole saliva
573 elicited by the use of sucrose containing and sugar free chewing gum. Archives of Oral
574 Biology, 40(8), 699-705.
575 - Denny, P., Hagen, F. K., Hardt, M., Liao, L., Yan, W., Arellanno, M., Bassilian, S., Bedi, G. S.,
576 Boontheung, P., Cociorva, D., Delahunty, C. M., Denny, T., Dunsmore, J., Faull, K. F.,
577 Gilligan, J., Gonzalez-Begne, M., Halgand, F., Hall, S. C., Han, X., Henson, B., Hewel, J., Hu, S.,
578 Jeffrey, S., Jiang, J., Loo, J. A., Loo, R. R. O., Malamud, D., Melvin, J. E., Miroshnychenko, O.,
579 Navazesh, M., Niles, R., Park, S. K., Prakobphol, A., Ramachandran, P., Richert, M.,
580 Robinson, S., Sondej, M., Souda, P., Sullivan, M. A., Takashima, J., Than, S., Wang, J.,
581 Whitelegge, J. P., Witkowska, H. E., Wolinsky, L., Xie, Y., Xu, T., Yu, W., Ytterberg, J., Wong,
582 D. T., Yates, J. R., III, & Fisher, S. J. (2008). The proteomes of human parotid and
583 submandibular/sublingual gland salivas collected as the ductal secretions. Journal of
584 Proteome Research, 7(5), 1994-2006.
585 - DiSabato-Mordarski, T., & Kleinberg, I. (1996). Measurement and comparison of the
586 residual saliva on various oral mucosal and dentition surfaces in humans. Archives of
587 Oral Biology, 41(7), 655-665.
588 - Doyennette, M., de Loubens, C., Deleris, I., Souchon, I., & Trelea, I. C. (2011). Mechanisms
589 explaining the role of viscosity and post-deglutitive pharyngeal residue on in vivo aroma
590 release: A combined experimental and modeling study. Food Chemistry, 128(2), 380-390.
591 - Eren, A. M., Borisy, G. G., Huse, S. M., & Mark Welch, J. L. (2014). Oligotyping analysis of
592 the human oral microbiome. Proc Natl Acad Sci U S A, 111(28), E2875-2884.
593 - Feron, G., Ayed, C., Qannari, E. M., Courcoux, P., Labouré, H., & Guichard, E. (2014).
594 Understanding aroma release from model cheeses by a statistical multiblock approach
595 on oral processing. PLoS ONE, 9(4), e93113 (93115 p.).
596 - Ferry, A. L., Hort, J., Mitchell, J. R., Lagarrigue, S., & Pamies, B. V. (2004). Effect of amylase
597 activity on starch paste viscosity and its implications for flavor perception. Journal of
598 Texture Studies, 35(5), 511-524.
599 - Flores, M., & Olivares, A. (2008). Release of aroma compounds from dry-fermented
600 sausages as affected by antioxidant and saliva addition. European Food Research And
601 Technology, 228(2), 283-290.
602 - Flower, D. R. (1996). The lipocalin protein family: Structure and function. Biochemical
603 Journal, 318, 1-14.
604 - Frank, D., Eyres, G. T., Piyasiri, U., Cochet-Broch, M., Delahunty, C. M., Lundin, L., &
605 Appelqvist, I. M. (2015). Effects of Agar Gel Strength and Fat on Oral Breakdown, Volatile
606 Release, and Sensory Perception Using in Vivo and in Vitro Systems. Journal of
607 Agricultural and Food Chemistry, 63(41), 9093-9102.
608 - Friel, E. N., & Taylor, A. J. (2001). Effect of Salivary Components on Volatile Partitioning
609 from Solutions. Journal of Agricultural and Food Chemistry, 49(8), 3898-3905.
610 - Genovese, A., Caporaso, N., Civitella, A., & Sacchi, R. (2014). Effect of human saliva and sip
611 volume of coffee brews on the release of key volatile compounds by a retronasal aroma
612 simulator. Food Research International, 61, 100-111.
613 - Genovese, A., Moio, L., Sacchi, R., & Piombino, P. (2015). Sip volume affects oral release of
614 wine volatiles. Food Research International, 77, Part 3, 426-431.
615 - Genovese, A., Piombino, P., Gambuti, A., & Moio, L. (2009). Simulation of retronasal
616 aroma of white and red wine in a model mouth system. Investigating the influence of
617 saliva on volatile compound concentrations. Food Chemistry, 114(1), 100-107.
618 - Giebultowicz, J., Wolinowska, R., Sztybor, A., Pietrzak, M., Wroczynski, P., &
619 Wierzchowski, J. (2009). Salivary Aldehyde Dehydrogenase: Activity towards Aromatic

23
620 - Aldehydes and Comparison with Recombinant ALDH3A1. Molecules, 14, 2363-2372.
621 - Gothefors, L., & Marklund, S. (1975). Lactoperoxidase activity in human milk and in
622 saliva of newborn infants. Infection and Immunity, 11(6), 1210-1215.
623 - Guichard, E. (2006). Flavour retention and release from protein solutions. Biotechnology
624 Advances, 24(2), 226-229.
625 - Guichard, E., Issanchou, S., Descourvieres, A., & Etievant, P. (1991). Pectin concentration,
626 molecular-weight and degree of esterification - influence on volatile composition and
627 sensory characteristics of strawberry jam. Journal of Food Science, 56(6), 1621-1627.
628 - Hodgson, M., Linforth, R. S., & Taylor, A. J. (2003). Simultaneous real-time measurements
629 of mastication, swallowing, nasal airflow, and aroma release. Journal of Agricultural and
630 Food Chemistry, 51(17), 5052-5057.
631 - Hollowood, T. A., Linforth, R. S., & Taylor, A. J. (2002). The effect of viscosity on the
632 perception of flavour. Chemical Senses, 27(7), 583-591.
633 - Hsu, L. C., Chang, W. C., Hiraoka, L., & Hsieh, C. L. (1994). Molecular-cloning, genomic
634 organization, and chromosomal localization of an additional human aldehyde
635 dehydrogenase gene, ALDH6. Genomics, 24(2), 333-341.
636 - Humphrey, S. P., & Williamson, R. T. (2001). A review of saliva: normal composition, flow,
637 and function. The Journal of Prosthetic Dentistry, 85(2), 162-169.
638 - Hussein, M. M., Kachikian, R., & Pidel, A. R. (1983). Analysis for Flavor Residuals in the
639 Mouth by Gas Chromatography. Journal of Food Science, 48(6), 1884-1885.
640 - Kandra, L., & Gyémánt, G. (2000). Examination of the active sites of human salivary α-
641 amylase (HSA). Carbohydrate Research, 329(3), 579-585.
642 - Krespi, Y. P., Shrime, M. G., & Kacker, A. (2006). The relationship between oral malodor
643 and volatile sulfur compound-producing bacteria. Otolaryngol Head Neck Surg, 135(5),
644 671-676.
645 - Kurkivuori, J., Salaspuro, V., Kaihovaara, P., Kari, K., Rautemaa, R., Gronroos, L., Meurman,
646 J. H., & Salaspuro, M. (2007). Acetaldehyde production from ethanol by oral streptococci.
647 Oral Oncol, 43(2), 181-186.
648 - Lasekan, O. (2013). A comparative analysis of the influence of human salivary enzymes
649 on odorant concentration in three palm wines. Molecules, 18(10), 11809-11823.
650 - Lindqvist, L., & Augustinsson, K. B. (1975). Esterases in human saliva. Enzyme, 20(5),
651 277-291.
652 - Lindqvist, L., & Augustinsson, K. B. (1980). Esterase patterns in human-saliva established
653 by chromatography on con-a-sepharose and by isoelectric-focusing. Archives of Oral
654 Biology, 25(6), 363-368.
655 - Linforth, R., Martin, F., Carey, M., Davidson, J., & Taylor, A. J. (2002). Retronasal Transport
656 of Aroma Compounds. Journal of Agricultural and Food Chemistry, 50(5), 1111-1117.
657 - Linforth, R., & Taylor, A. J. (2000). Persistence of volatile compounds in the breath after
658 their consumption in aqueous solutions. Journal of Agricultural and Food Chemistry,
659 48(11), 5419-5423.
660 - Lubbers, S., Landy, P., & Voilley, A. (1998). Retention and release of aroma compounds in
661 foods containing proteins. Food Technology, 52(5), 68.
662 - Mark Welch, J. L., Rossetti, B. J., Rieken, C. W., Dewhirst, F. E., & Borisy, G. G. (2016).
663 Biogeography of a human oral microbiome at the micron scale. Proceedings of the
664 National Academy of Sciences, 113(6), E791-E800.
665 - Marsh, P. D., Do, T., Beighton, D., & Devine, D. A. (2016). Influence of saliva on the oral
666 microbiota. Periodontology 2000, 70, 80-92.
667 - Mejean, C., Morzel, M., Neyraud, E., Issanchou, S., Martin, C., Bozonnet, S., Urbano, C.,
668 Schlich, P., Hercberg, S., Peneau, S., & Feron, G. (2015). Salivary Composition Is
669 Associated with Liking and Usual Nutrient Intake. PLoS ONE, 10(9), e0137473.
670 - Mitropoulou, A., Hatzidimitriou, E., & Paraskevopoulou, A. (2011). Aroma release of a
671 model wine solution as influenced by the presence of non-volatile components. Effect of
672 commercial tannin extracts, polysaccharides and artificial saliva. Food Research
673 International, 44(5), 1561-1570.

24
674 - Mounayar, R., Septier, C., Chabanet, C., Feron, G., & Neyraud, E. (2013). Oral Fat
675 Sensitivity in Humans: Links to Saliva Composition Before and After Stimulation by Oleic
676 Acid. Chemosensory Perception, 6(3), 118-126.
677 - Muñoz-González, C., Cueva, C., Pozo-Bayón, M. Á., & Moreno-Arribas, M. V. (2015). Ability
678 of human oral microbiota to produce wine odorant aglycones from odourless grape
679 glycosidic aroma precursors. Food Chemistry, 187(15), 112-119.
680 - Munoz-Gonzalez, C., Feron, G., Guichard, E., Rodriguez-Bencomo, J. J., Martin-Alvarez, P. J.,
681 Moreno-Arribas, M. V., & Pozo-Bayon, M. A. (2014). Understanding the Role of Saliva in
682 Aroma Release from Wine by Using Static and Dynamic Headspace Conditions. Journal of
683 Agricultural and Food Chemistry.
684 - Nasidze, I., Li, J., Quinque, D., Tang, K., & Stoneking, M. (2009). Global diversity in the
685 human salivary microbiome. Genome Research, 19(4), 636-643.
686 - Odake, S., Roozen, J. P., & Burger, J. J. (1998). Effect of saliva dilution on the release of
687 diacetyl and 2-heptanone from cream style dressings. Nahrung, 42(6), 385-391.
688 - Pages-Helary, S., Andriot, I., Guichard, E., & Canon, F. (2014). Retention effect of human
689 saliva on aroma release and respective contribution of salivary mucin and alpha-
690 amylase. Food Research International, 64, 424-431.
691 - Pagès-Hélary, S., Andriot, I., Guichard, E., & Canon, F. (2014). Retention effect of human
692 saliva on aroma release and respective contribution of salivary mucin and α-amylase.
693 Food Research International, 64, 424-431.
694 - Pappa, A., Estey, T., Manzer, R., Brown, D., & Vasiliou, V. (2003). Human aldehyde
695 dehydrogenase 3A1 (ALDH3A1): biochemical characterization and
696 immunohistochemical localization in the cornea. Biochemical Journal, 376(Pt 3), 615-
697 623.
698 - Piombino, P., Genovese, A., Esposito, S., Moio, L., Cutolo, P. P., Chambery, A., Severino, V.,
699 Moneta, E., Smith, D. P., Owens, S. M., Gilbert, J. A., & Ercolini, D. (2014). Saliva from obese
700 individuals suppresses the release of aroma compounds from wine. PLoS ONE, 9(1),
701 e85611.
702 - Rabe, S., Krings, U., & Berger, R. G. (2004). In vitro Study of the Influence of Physiological
703 Parameters on Dynamic In-mouth Flavour Release from Liquids. Chemical Senses, 29(2),
704 153-162.
705 - Repoux, M., Labouré, H., Courcoux, P., Andriot, I., Sémon, É., Yven, C., Feron, G., &
706 Guichard, E. (2012a). Combined effect of cheese characteristics and food oral processing
707 on in vivo aroma release. Flavour and Fragrance Journal, 27(6), 414-423.
708 - Repoux, M., Semon, E., Feron, G., Guichard, E., & Laboure, H. (2012b). Inter-individual
709 variability in aroma release during sweet mint consumption. Flavour and Fragrance
710 Journal, 27(1), 40-46.
711 - Salles, C., Chagnon, M. C., Feron, G., Guichard, E., Laboure, H., Morzel, M., Semon, E.,
712 Tarrega, A., & Yven, C. (2011). In-Mouth Mechanisms Leading to Flavor Release and
713 Perception. Critical Reviews in Food Science and Nutrition, 51(1), 67-90.
714 - Sinding, C., Thomas-Danguin, T., Chambault, A., Beno, N., Dosne, T., Chabanet, C., Schaal,
715 B., & Coureaud, G. (2013). Rabbit neonates and human adults perceive a blending 6-
716 component odor mixture in a comparable manner. PLoS ONE, 8(1), e53534.
717 - Sreerama, L., Hedge, M. W., & Sladek, N. E. (1995). Identification of a class 3 aldehyde
718 dehydrogenase in human saliva and increased levels of this enzyme, glutathione S-
719 transferases, and DT-diaphorase in the saliva of subjects who continually ingest large
720 quantities of coffee or broccoli. Clin Cancer Res, 1(10), 1153-1163.
721 - Starkenmann, C., Le Calvé, B., Niclass, Y., Cayeux, I., Beccucci, S., & Troccaz, M. (2008).
722 Olfactory Perception of Cysteine−S-Conjugates from Fruits and Vegetables. Journal of
723 Agricultural and Food Chemistry, 56(20), 9575-9580.
724 - Stevens, S. S. (1960). The psychophysics of sensory function. American Scientist, 48(2),
725 226-253.
726 - Svensson, B. E. (1988). Abilities of peroxidases to catalyse peroxidase-oxidase oxidation
727 of thiols. Biochemistry Journal, 256, 757-762.

25
728 - Takahashi, N. (2015). Oral Microbiome Metabolism: From “Who Are They?” to “What Are
729 They Doing?”. Journal of Dental Research, 94(12), 1628-1637.
730 - Tavel, L., Moreau, C., Bouhallab, S., Li-Chan, E. C. Y., & Guichard, E. (2010). Interactions
731 between aroma compounds and [beta]-lactoglobulin in the heat-induced molten globule
732 state. Food Chemistry, 119(4), 1550-1556.
733 - Taylor, A. J. (2002). Release and Transport of Flavors In Vivo: Physicochemical,
734 Physiological, and Perceptual Considerations. Comprehensive Reviews in Food Science and
735 Food Safety, 1(2), 45-57.
736 - Thomas-Danguin, T., Sinding, C., Romagny, S., El Mountassir, F., Atanasova, B., Le Berre,
737 E., Le Bon, A. M., & Coureaud, G. (2014). The perception of odor objects in everyday life: a
738 review on the processing of odor mixtures. Front Psychol, 5, 504.
739 - Tromelin, A., Merabtine, Y., & Andriot, I. (2010). Retention–release equilibrium of aroma
740 compounds in polysaccharide gels: study by quantitative structure–activity/property
741 relationships approach. Flavour and Fragrance Journal, 25(6), 431-442.
742 - van Ruth, S. M., Grossmann, I., Geary, M., & Delahunty, C. M. (2001). Interactions between
743 Artificial Saliva and 20 Aroma Compounds in Water and Oil Model Systems. Journal of
744 Agricultural and Food Chemistry, 49(5), 2409-2413.
745 - van Ruth, S. M., & Roozen, J. P. (2000). Influence of mastication and saliva on aroma
746 release in a model mouth system. Food Chemistry, 71(3), 339-345.
747 - van Ruth, S. M., Roozen, J. P., & Cozijnsen, J. L. (1995). Changes in flavour release from
748 rehydrated diced bell peppers (Capsicum annuum) by artificial saliva components in
749 three mouth model systems. Journal of the Science of Food and Agriculture, 67(2), 189-
750 196.
751

752

753

754 Figure Captions

755
756 Fig. 1. Salivary mechanisms involved in aroma release.
757
758
759
760

26
 Type of Salivary Odorous volatiles Effect and proposed mechanisms
References
saliva component studied when available
Retention by mucins for some
14 compounds from compounds.
Friel et al.
different physicochemical No correlation with hydrophobicity and (2001)
classes different behaviours suggest different
binding sites.
Mucins
12 compounds from
different physicochemical Van Ruth et al.
classes (from rehydrated Retention by mucins by hydrophobic (1995)
diced peppers) effects
Linear series of esters & Pages-Helary
methylketones et al. (2014)
Artificial Alpha- Linear series of esters & Retention by alpha-amylase by
Saliva amylase methylketones hydrophobic effects
Pages-Helary
No cumulative effect of mucins and
Mucins + et al. (2014)
Linear series of esters & amylase
Alpha-
methylketones Modification of binding sites by protein-
amylase
protein interactions
Effect of salts only In presence of
20 compounds from
salivary proteins: salts increase Van Ruth et al.
different physicochemical
hydrophilic compounds and reduce (2001)
classes
hydrophobic compounds release
Salts Salts affect release of some compounds.
14 compounds from No correlation with hydrophobicity.
Friel et al.
different physicochemical Modification of mucin binding sites by (2001)
classes ionic strength: charge-shielding effect &
self aggregation
Linear series of Retention of aroma by salivary proteins Pages-Helary
Proteins
methylketones by hydrophobic effects et al. (2014)

Hydrolysis of esters increasing with Buettner

chain length. (2002b)
Esters
Involvement of esterases or aldehyde Pages-Helary
dehydrogenases. et al. (2014)
Enzymes
Human Oxidation of thiols. Buettner
saliva Thiols
Involvement of peroxidases. (2002b)

Reduction of aldehydes to alcohols. Buettner

Aldehydes
aldehyde reductases (2002a)
Release of aroma (terpenes, benzenic
Munoz-
Oral derivatives and lipid derivatives) from
- Gonzales et al.
microbiota odorless glycosidic precursors (2015)
Involvement of microbial enzymes
761
762
763
764 Table 1. Effect of specific salivary components on aroma release
765

27
766
767
In
Food vivo Observed impact of Mechanisms of action of
Saliva effect References
matrix / in saliva saliva
vitro
Lower release of
Water oil In hydrophilic compounds. Odake et al.
emulsion vitro No effect on Modification of volatiles (1998)
Dilution hydrophobic compounds partition into the
different phases.
In Different behaviours Genovese et al.
Coffee (2014)
vitro depending on aroma
Enhancement of the
Gelatin and In Increased release of Boland et al.
hydrophilic nature of the (2004)
pectin gels vitro hydrophobic compounds
system
In Increased release of Enhancement of matrix Odake et al.
Dressings (1998)
vitro compounds spreadability and
Modification
Model In Increased release of increase of exchange Feron et al.
of food
cheese vivo compounds surfaces (2014)
matrix
Starch-
properties In Increased release of Ferry et al.
based Starch degradation due to (2004)
vitro compounds
system amylase activity, leading
Increased release of to modification of matrix
Rehydrated In Van Ruth et al.
highly volatile viscosity (2000)
bell pepper vitro
compounds
Expulsion of hydrophobic
compounds by complexes
In Increased release of Mitropoulou
formed by saliva et al. (2009)
vitro hydrophobic compounds
proteins, tannins and
carbohydrates
Retention of hydrophobic
Interaction compounds by
Munoz-
of food Red wine In Decreased release of hydrophobic cavities of
Gonzales et al.
matrix vitro hydrophobic compounds complexes formed by (2014)
constituents saliva proteins and
with tannins
salivary Modification of salivary
proteins In Higher aroma release proteins structure and Genovese et al.
vitro than from white wines binding sites by (2009)
polyphenols
Competition between
Sucrose In Increased release of sugar and aromas for Friel et al.
solution vitro some molecules binding sites on salivary (2001)
proteins
768
769 Table 2: Saliva effects on aroma release from food matrix
770
771
772
773
774
775
776
777
778
779
780
781

28
782
783
784
785
786
787 Highlights
788
789
790  Interaction of aroma compounds with saliva proteins modifies their release
791  Saliva modifies aroma release through the modification of matrix structure and
792 properties
793  Saliva may impact aroma perception
794

29