This electronic thesis or dissertation has been downloaded from the King’s Research Portal at https://kclpure.kcl.ac.uk/portal/

The influence of salivary metabolite composition on taste and oral perception

Gardner, Alex

Awarding institution: King's College London

The copyright of this thesis rests with the author and no quotation from it or information derived from it may be published without proper acknowledgement.

END USER LICENCE AGREEMENT

Unless another licence is stated on the immediately following page this work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ You are free to copy, distribute and transmit the work Under the following conditions:  Attribution: You must attribute the work in the manner specified by the author (but not in any way that suggests that they endorse you or your use of the work).  Non Commercial: You may not use this work for commercial purposes.  No Derivative Works - You may not alter, transform, or build upon this work.

Any of these conditions can be waived if you receive permission from the author. Your fair dealings and other rights are in no way affected by the above.

Take down policy

If you believe that this document breaches copyright please contact [email protected] providing details, and we will remove access to the work immediately and investigate your claim.

Download date: 11. Oct. 2021 THE INFLUENCE OF SALIVARY METABOLITE COMPOSITION ON TASTE AND ORAL PERCEPTION

Thesis submitted for the degree of DOCTOR OF PHILOSOPHY

By ALEXANDER GARDNER

Centre for Host-Microbiome Interactions, Faculty of Dental, Oral & Craniofacial Sciences, King’s College London

December 2019 Supervisors: Professor Guy Carpenter and Dr. Po-Wah So

1

Summary

Metabolomics is the study of the small molecules that are present within a biological sample or system. Metabolomic analysis is yielding ever-increasing new information about the role of metabolites in health and disease. Much of the literature to date has focused on biofluids such as urine or plasma although saliva is increasingly being identified as an important fluid for metabolomic study. Functional roles for metabolites generated in the gut are being discovered, in relation to both health and disease, however little is known about the role(s) salivary metabolites play in health. Taste perception is an important physiological process occurring in the oral cavity that has wide reaching implications for health and it has been speculated that the metabolic composition of saliva may influence taste perception.

This thesis explores the metabolic composition of saliva with particular emphasis on relationships with taste perception and other oral sensations such as astringency. Other questions that are addressed include the role of the oral microbiome in shaping the metabolic composition of saliva and the association between the metabolic composition and physical properties of saliva. The main analytical platform for metabolomic profiling of saliva was proton nuclear magnetic resonance spectroscopy (1H-NMR).

In the first instance, validation of the analytical and sensory methodologies adopted throughout this thesis were conducted, to satisfy that subsequent data gathering was valid. It was then found, through a combination of in-vivo and in-vitro analyses, that the salivary metabolome is largely shaped by oral bacteria. The most concentrated salivary metabolites such as the short chain fatty acids (SCFAs) acetate, butyrate and propionate are of microbial origin whereas metabolites such as lactate, citrate and urea are of host origin. Furthermore, oral bacteria were shown in-vitro to catabolise salivary and endogenous metabolites, generating abundant SCFAs and other salivary metabolites. Modulating effects of gustatory reflexes on salivary metabolic composition were then investigated for (sweet), caffeine (bitter), menthol (cooling) and capsaicin (warming). Sucrose was found to be rapidly catabolised by oral bacteria, generating abundant metabolites such as pyruvate and lactate, whereas capsaicin stimulation increased citrate concentrations. Citrate was also found to relate to the extensional rheology of capsaicin-stimulated saliva.

Assessment of the relationship between taste perception and salivary composition was approached from two angles. Firstly, taste recognition and detection thresholds were

2 assessed in a taste panel. A positive association between salivary urea and sucrose detection threshold was discovered. As urea was found to be consumed by oral bacteria, inversely correlating with salivary bacterial load, it is speculated that differences in salivary urea between sensitive and insensitive detectors might reflect differences in the composition and/or function of the oral microbiome. Additional findings relating to the salivary metabolome that stemmed from this thesis include an apparent lack of any significant genetic contribution to the salivary metabolome and a significantly higher degree of intra- individual stability compared to inter-individual stability of salivary metabolite profile.

Subsequently, suprathreshold taste perception of sucrose, aspartame, caffeine, black tea and oleic acid (sweet, sweet, bitter, astringent, fatty; respectively) and salivary composition was assessed in a twin study. Generally, twins with discordant tastes as well as unrelated individuals with different tastes has similar differences in their saliva composition. Twins with similar levels of taste perception did not display these salivary differences. Invariably, these relationships were inverse i.e. higher metabolite concentration was associated with reduced taste sensitivity. Salivary composition therefore appeared to be an important local environmental factor in determining sensitivity to certain taste stimuli. Controlling for salivary flow rate also emerged as important, as higher flow rate was associated with lower metabolite concentrations and improved sensitivity. This suggested that the degree to which salivary fluid dilutes microbial-produced metabolites in the mouth may impact on taste perception. There was no evidence of a general inhibitory metabolite across the different tastants although taurine was associated with both sucrose and oleic acid perception.

Collectively, this work found that the salivary metabolome represents a complex balance between host and microbial metabolic activity. It can be concluded that there is evidence of associations between salivary metabolite composition and taste perception. Certain metabolites of microbial origin were inversely associated with taste perception, and an endogenous metabolite known to be consumed by oral bacteria (urea) was positively associated with sucrose detection. The findings of this work support the emerging role of metabolomics as a functional measure of complex microbial communities in relation to host physiology. While the findings of this work cannot be used to infer a causal relationship between specific metabolites and taste, target salivary metabolites for future study in association with taste perception are identified.

3

List of Publications and Presentations

Publications:

Gardner, A., Parkes, H.G., Carpenter, G.H. and So, P.W., 2018. Developing and Standardizing a Protocol for Quantitative Proton Nuclear Magnetic Resonance (1H NMR) Spectroscopy of Saliva. Journal of proteome research, 17(4), pp.1521-1531.

Gardner, A., Parkes, H.G., So, P.W. and Carpenter, G.H., 2019. Determining bacterial and host contributions to the human salivary metabolome. Journal of Oral Microbiology, 11:1, DOI: 10.1080/20002297.2019.1617014.

Gardner, A. & Carpenter, G.H., 2019. Anatomical stability of human fungiform papillae and relationship with oral perception measured by salivary response and intensity rating. Scientific Reports, 9(1).

Gardner, A., So, P-W, Carpenter, G. 2019. Endogenous salivary citrate is associated with enhanced rheological properties following oral capsaicin-stimulation. Experimental Physiology, 1-12.

Presentations:

Year Conference Title 2019 British Society for Oral and Oral: “Sweet Tooth” is Associated with Dental Research, Leeds, UK Altered Intra-oral Sucrose 2018 International Association for Poster: Influence of Oral Bacterial Dental Research, London UK Metabolism on Sucrose Taste Perception 2018 Food Oral Processing, Short oral and poster: The Influence of Nottingham, UK Salivary Metabolites on Oral Perception 2018 International Society of Clinical Poster: The Origin of Salivary Metabolites: Spectroscopy, Glasgow, UK 1H-NMR Spectroscopy of Multiple Biofluids 2017 European Saliva Symposium, Oral: Correlation between fungiform Egmond aan Zee, Netherlands papillae density and tastant-stimulated salivary compositional change 2017 London Oral Biology Club, Oral: Mechanisms for the oral perception of London, UK astringency 2017 Royal Society of Chemistry Oral and poster: Developing and NMR Discussion Group Standardising a Protocol for Quantitative Postgraduate Meeting, Proton Nuclear Magnetic Resonance (1H- Glasgow, UK NMR) Spectroscopy of Saliva

4

Acknowledgements

Completion of this work was only possible thanks to the input of an extensive list of individuals. Firstly, sincere thanks must go to my supervisors, Prof. Guy Carpenter and Dr. Po- Wah So for their constant support and encouragement. I have gotten far more out of this PhD experience than I could have ever imagined. I also express thanks to my industrial supervisor, Dr. Martin Walker, and his team at Diageo Woodside not only for their continued enthusiasm and ideas but also for making my brief time at their facility a productive and enjoyable experience. I must also thank Diageo plc. and the Biotechnology and Biological Sciences Research Council (BBSRC) for their financial support of the work.

I wish to thank all my friends and colleagues, past and present, on floor 17 of Guy’s Tower for readily sharing their knowledge and skills. In particular, I thank Steve Gilbert and Mukesh Mistry whose laboratory upkeep allowed the smooth completion of this work. Elsewhere at King’s, I thank Dr. Andrew Atkinson and Dr. Adrien Le Guennec at the NMR spectroscopy facility for their help in spectral acquisition, analysis and interpretation. I must also thank all the staff from TwinsUK at St. Thomas’s clinical research facility, in particular Dr. Amrita Vijay and Dr. Claire Steves for their help in conducting the work with their twin volunteers and advice on the subsequent data analysis.

I also thank my family for their inherent support throughout my time in London. I am grateful for their restraint in enquiring about my work, allowing a separation between professional and family life.

Finally, and most importantly, I thank every individual who willingly gave up their time, information, saliva and, in some cases, blood to participate in this work. It is not an exaggeration or a cliché to say this work would not have been possible otherwise.

5

Abbreviations

3-AFC ………………………………………………………………………………… 3-alternative forced-choice test AC …………………………………………………………………………………………………………… Adenylate cyclase Ach ………………………………………………………………………………………………………………… Acetylcholine AMP ………………………………………….………………………………………………… Adenosine monophosphate ANOVA ………………………………………………………………………………………………… Analysis of variance ATP …………………………………………………………………………………………………… BCA ………………………………………………………………………………………………………… Bicinchoninic acid BMI ……………………………………………………………………………………………………………… Body mass index CaBER …………………………………………………………………… Capillary breakup extensional rheometer CA VI ………………………………………………………………………………………………… Carbonic anhydrase 6 CBB …………………………………………………………………………………………………… Coomassie brilliant blue CPMG …………………………………………………………………… Carr-Purcell-Meiboom-Gill pulse sequence COSY ……………………………………………………………………………………………… Correlation spectroscopy CE-TOF-MS ………………………………… Capillary electrophoresis time-of-flight mass spectrometry CFU ………………………………………………………………………………………………………… Colony forming unit CVP ……………………………………………………………………………………………………… Circumvallate papillae

D2O ……………………………………………………………………………………………………………… Deuterium oxide DMFT …………………………………………………………………………………… Decayed, missing, or filled teeth DTT …………………………………………………………………………………………………………………… Dithiothreitol DZ ……………………………………………………………………………………………………………………………… Dizygotic ED ……………………………………………………………………………………………………………… External diameter ENaC ……………………………………………………………………………………………… Epithelial sodium channel FAA ………………………………………………………………………………………………… Fastidious anaerobe agar FCF …………………………………………………………………………………………………………… For colouring food FP ……………………………………………………………………………………………………………… Fungiform papillae FPD ………………………………………………………………………………………………. Fungiform papillae density FUPLC-MS …………………… Faster ultra-performance liquid chromatography mass spectrometry GABA …………………………………………………………………………………………… Gamma aminobutyric acid GCF …………………………………………………………………………………………………… Gingival crevicular fluid GC-MS …………………………………………………………………… Gas chromatography‐mass spectrometry gLMS …………………………………………………………………………… Generalised labelled magnitude scale gVAS …………………………………………………………………………………… Generalised visual analogue scale glVAS ……………………………………………………………………. Generalised labelled visual analogue scale gwVAS ……………………………………………………………. Generalised words-only visual analogue scale 6 glPRP ………………………………………………………………………………………… Glycated proline-rich GPCR ……………………………………………………………………………………………. G-protein coupled receptor HPLC ……………………………………………………………………… High performance liquid chromatography LDS ………………………………………………………………………………………..………… Lithium dodecyl sulphate MS ……………………………………………………………………………………………………………. Mass spectrometry MSG …………………………………………………………………………………………………. Monosodium glutamate MZ ……………………………………………………………………………………………………………………… Monozygotic NOESY …………………………………………… Nuclear Overhauser effect spectroscopy pulse sequence 1H-NMR ……………………………………………………. Proton nuclear magnetic resonance spectroscopy

IP3 …………………………………………………………………………………………………………. Inositol triphosphate kDA ……………………………………………………………………………………………………………………….. Kilo Dalton MDA …………………………………………………………………………………………………………………… Mega Dalton MUC5B ………………………………………………………………………………………………………………….. Mucin 5B MUC7 ………………………………………………………………………………………………………………………... Mucin 7 NMR …………………………………………………………………… Nuclear magnetic resonance spectroscopy PAS …………………………………………………………………………………………………………… Periodic acid-Schiff PBS ………………………………………………………………………………………………… Phosphate buffered saline PCA …………………………………………………………………………………………… Principal component analysis PKA………………………………………………………………………………………………………………… Protein Kinase A PLC………………………………………………………………………………………………………………… Phospholipase C Ppm …………………………………………………………………………………………………………… Parts per million* PROP ………………………………………………………………………………………………………. 6-n-propylthiouracil PRP …………………………………………………………………………………………………………… Proline-rich protein PTC ……………………………………………………………………………………………………… Phenylthiocarbamide PS ……………………………………………………………………………………………………………………… Parotid saliva SEM ……………………………………………………………………………………………… Standard error of the mean SCFA ……………………………………………………………………………………………………… Short chain fatty acid SDS-PAGE ………………………………… Sodium dodecyl sulphate polyacrylamide gel electrophoresis SLC ……………………………………………………………………………………………… Solute carrier protein family SNP ……………………………………………………………………………………… Single nucleotide polymorphism T1R1 (or Tas1R1) ………………………………………………………………… Taste receptor type 1 member 1 T1R2 (or Tas1R2) ………………………………………………………………… Taste receptor type 1 member 2 T1R3 (or Tas1R3) ………………………………………………………………… Taste receptor type 1 member 3 T2R38 (or Tas2R38) …………………………………………………………… Taste receptor type 2 member 38 TRC …………………………………………………………………………………………………………….. Taste receptor TRPM ……………………………………………………………… Transient receptor potential channel family M 7

TRPV ………………………………………………………………… Transient receptor potential channel family V TYC ……………………………………………………………………………………… Tryptone, Yeast extract, Cysteine

2 TSP ……………………………………………………………………………… trimethylsilyl-[2,2,3,3- H4]-propionate UHPLC-MS/MS ……… Ultra-high-performance liquid chromatography and tandem mass spectrometry UPLC-QTOFMS ……… Ultraperformance liquid chromatography coupled with quadrupole/time- of-flight mass spectrometry UWMS ……………………………………………………………………………… Unstimulated whole-mouth saliva VEG …………………………………………………………………………………………………………… Von Ebner’s glands VIP………………………………………………………………………………………………Vasoactive intestinal peptide WMS ………………………………………………………………………………………………………. Whole-mouth saliva

* = In this thesis, ppm is used to refer to both the relative chemical shift of NMR spectral peaks as well as the concentration of low concentration tastants.

8

Table of Contents

Summary ...... 2 List of Publications and Presentations ...... 4 Acknowledgements ...... 5 Abbreviations ...... 6 Table of Figures ...... 14 Table of Tables ...... 17 Chapter 1: Introduction ...... 19 1.1 Salivary secretion ...... 19 1.1.1 Macroscopic anatomy of human salivary glands and ducts ...... 19 1.1.2 Histology of salivary glands ...... 19 1.1.3 Salivary secretion ...... 20 1.1.4 Neural regulation of salivary secretion ...... 22 1.2 Salivary composition and function ...... 22 1.2.1 Major salivary ions and their function ...... 22 1.2.2 Major salivary proteins and their function ...... 23 1.2.3 Lower abundance salivary components ...... 25 1.2.4 Non-exocrine components of saliva ...... 25 1.2.5 The oral microbiome ...... 26 1.3 Metabolic composition of saliva ...... 27 1.3.1 Historical context and advent of metabolomics ...... 27 1.3.2 Current knowledge of salivary metabolomics ...... 28 1.3.3 Metabolomics technologies ...... 33 1.3.4 Metabolomics and the oral microbiome ...... 34 1.4 Taste and oral perception ...... 37 1.4.1 Overview ...... 37 1.4.2 Anatomy of taste perception ...... 37 1.4.3 Transduction and neural processing of basic tastes ...... 38 1.4.4 Central perception of taste ...... 41 1.4.5 Additional oral sensations ...... 41 1.4.6 Sensory assessment of taste ...... 43 1.4.7 Biological measurement of taste perception ...... 46 1.5 The relationship between saliva and taste ...... 47 1.5.1 Tastant-stimulated salivary changes ...... 47 1.5.2 Tastant mediated salivary compositional changes independent of gustatory reflexes ...... 49 9

1.5.3 The influence of saliva composition on taste perception ...... 50 1.6 Summary ...... 54 1.7 Aims and Objectives ...... 54 Chapter 2: Materials and Methods ...... 57 2.1 Ethical approval and recruitment ...... 57 2.2 Sample collection, processing and storage ...... 57 2.2.1 Whole mouth saliva ...... 57 2.2.2 Parotid saliva ...... 58 2.2.3 Submandibular/sublingual saliva ...... 58 2.2.4 Plasma ...... 58 2.2.5 Tongue biofilm ...... 58 2.2.6 Buccal epithelial cells ...... 59 2.2.7 Gingival-crevicular fluid ...... 59 2.3 Salivary analyses ...... 59 2.3.1 Flow rate ...... 59 2.3.2 Cellular and microbial composition ...... 59 2.3.3 Protein composition ...... 60 2.3.4 Metabolite composition - 1H-NMR spectroscopy ...... 64 2.3.5 Extensional Rheology ...... 66 2.4 Sensory analyses ...... 67 2.4.1 Tastant preparation ...... 67 2.4.2 Tastant administration and intensity rating ...... 68 2.4.3 Fungiform papillae density measurement ...... 68 2.5 Statistical analyses ...... 69 Chapter 3: Methodological validation ...... 70 3.1 Introduction ...... 70 3.2 Aims and Objectives ...... 71 3.3 Materials and methods ...... 72 3.3.1 Assessment of centrifugation and freezing on salivary bacterial viability ...... 72 3.3.2 Assessment of multivariate analyses of digitised SDS-PAGE salivary protein profile72 3.3.3 Assessment of suprathreshold intensity scale and tastant concentrations ...... 72 3.3.4 Assessment of protocol to return salivary metabolites to baseline levels ...... 74 3.4 Results ...... 76 3.4.1 Developing and Standardizing a Protocol for Quantitative Proton Nuclear Magnetic Resonance (1H-NMR) Spectroscopy of Saliva ...... 76 3.4.2 Assessment of centrifugation and freezing on salivary bacterial viability ...... 102

10

3.4.3 Assessment of multivariate analyses of digitised SDS-PAGE salivary protein profile ...... 102 3.4.4. Assessment of suprathreshold intensity scale and tastant concentrations ...... 107 3.4.5 Assessment of protocol to return salivary metabolites to baseline levels ...... 108 3.5 Discussion ...... 109 Chapter 4: Determining host and microbial contributions to the salivary metabolome ..... 112 4.1 Introduction ...... 112 4.2 Aims and Objectives ...... 113 4.3. Materials and methods ...... 114 4.3.1 Additional analyses of published data ...... 114 4.3.2 Assessment of potential salivary taurine sources ...... 114 4.3.3 In-vitro saliva inoculation study ...... 115 4.3.4 Sample analyses ...... 116 4.4 Results ...... 119 4.4.1 Determining bacterial and host contributions to the human saliavary metabolome ...... 119 4.4.2 Assessment of potential salivary taurine sources ...... 131 4.4.3 In-vitro saliva inoculation study ...... 134 4.5 Discussion ...... 149 Chapter 5: Analysis of the effects of taste stimulation on salivary composition and extensional rheology ...... 152 5.1 Introduction ...... 152 5.2 Aim ...... 154 5.3 Materials and methods ...... 154 5.4 Results ...... 156 5.4.1 Intra-oral and post expectoration flow rates ...... 156 5.4.2 Salivary metabolite concentrations and protein abundance...... 157 5.4.3 Extensional rheology ...... 158 5.4.4 Endogenous salivary citrate is associated with enhanced rheological proerties following capsaicin stimulation ...... 166 5.5 Discussion ...... 189 Chapter 6: Determining the intra-individual variability of salivary metabolite and major protein profiles and the relationship with detection and recognition thresholds of basic tastes...... 193 6.1 Introduction: ...... 193 6.2 Aims and Objectives ...... 195 6.3 Methods ...... 195 6.3.1 Participants ...... 195

11

6.3.2 Study design ...... 196 6.3.3 Sensory methodology ...... 196 6.3.4 Saliva collection, storage and analysis ...... 197 6.3.5 Statistical analyses ...... 198 6.4 Results ...... 200 6.4.1 Intra-individual salivary flow rate is consistent between days ...... 200 6.4.2 Salivary protein composition is significantly more stable within than between individuals ...... 202 6.4.3 Salivary metabolite composition shows comparable or greater relative stability than salivary major protein composition ...... 205 6.4.4 Taste thresholds of trained and untrained panellists ...... 210 6.4.5 Relationship between taste thresholds and salivary composition ...... 213 6.4.6 Comparison of salivary composition between trained panellists, untrained panellists and age and gender matched controls ...... 216 6.5 Discussion ...... 218 6.5.1 Stability of salivary protein composition, salivary metabolite composition and taste thresholds ...... 218 6.5.2 Associations between taste thresholds and salivary composition ...... 219 6.5.3 Comparisons of taste thresholds and saliva composition of trained and untrained panellists ...... 221 6.5.6 Summary ...... 222 Chapter 7: Investigation of salivary composition as an oral environmental factor in relation to suprathreshold oral perception ...... 224 7.1 Introduction ...... 224 7.2 Aims and Objectives ...... 228 7.3 Methodology ...... 228 7.3.1 Participant recruitment ...... 228 7.3.2 Tastant preparation and sensory assessment ...... 229 7.3.3 Study protocol and saliva collection ...... 229 7.3.4 Fungiform papillae density measurement ...... 230 7.3.5 Sample handling and analyses ...... 230 7.3.6 Statistical analyses ...... 231 7.4 Results ...... 233 7.4.1 Participant demographics ...... 233 7.4.2 Taste responses ...... 234 7.4.3 Assessment of heritability of salivary composition, taste perception and fungiform papillae density ...... 234 7.4.4 Assessment of salivary compositional differences in sensitive and insensitive tasters...... 237 12

7.5 Discussion ...... 248 7.5.1 Genetic contributions to taste perception and saliva composition ...... 248 7.5.2 Relationship between salivary composition and taste sensitivity ...... 249 7.5.3 Relationship between intra-oral sucrose and sucrose perception ...... 253 7.5.4 Summary ...... 254 Chapter 8: General discussion ...... 255 8.1 Insight into the salivary metabolome ...... 255 8.2 Limitations of presented work ...... 260 8.3 Future work ...... 263 8.3.1 The role of salivary SCFAs in oral health...... 263 8.3.2 Mechanism of action for salivary metabolite/taste receptor interactions ...... 264 8.3.3 The role of citrate in the physical properties of saliva ...... 265 8.4 Conclusion ...... 266 References ...... 267 Appendix A: Protocol and preliminary findings of the effects of SCFAs on TR146 cells (buccal epithelial like squamous cell carcinoma) ...... 284 Appendix A Protocol ...... 284 Appendix B: Preliminary investigation of causation and mechanism for salivary metabolite alteration of taste ...... 286 Appendix B: In-vivo experiment ...... 286 Appendix B: In-vitro experiment ...... 287

13

Table of Figures

Figure Page Figure 1.1: Illustration of salivary gland anatomy. 19 Figure 1.2: Illustration of salivary acinar unit and intercalated duct. 21 Figure 1.3: A micrograph of a Gram-stained WMS droplet showing the cellular 26 content. Figure 1.4: Growth of literature concerning studies of the human salivary 28 metabolome and proteome. Figure 1.5: Diagram of lingual papillae and taste bud. 38 Figure 1.6: Transduction of basic taste sensations. 40 Figure 1.7: A depiction of different types of suprathreshold intensity scales. 45 Figure 1.8: The bi-directional relationship between saliva and taste. 47 Figure 2.1: A depiction of the scale used to assess GCF volume collected on 59 filter strips. Figure 2.2: Photograph of a lane from a coomassie and PAS stained 61 polyacrylamide gel showing the separated major proteins in a typical saliva sample. Figure 2.3: Illustration of the process of digitising protein samples lanes from 63 polyacrylamide gel scans. Figure 2.4: Illustrates salivary capillary breakup measurement using a CaBER. 67 Figure 2.5: An example of the generalised labelled visual analogue scale 68 (glVAS) used for suprathreshold intensity ratings. Figure 2.6: Illustrates the process of FPD measurement. 69 Figure 3.1: A comparison of the effects of centrifugation and freezing on WMS 102 bacterial viability. Figure 3.2. Inter- and intra-gel stability of digitised salivary protein profiles. 104 Figure 3.3. Saliva samples from four participants run in triplicate on four PAS 105 and coomassie stained polyacrylamide gels. Figure 3.4. Investigation of inter-gel “batch” effects. 106 Figure 3.5: Boxplots of the participant intensity ratings (n = 12) for different 107 concentrations of four tastants. Figure 3.6: comparison of intensity ratings for 0.25 M sucrose and 8 mM 108 caffeine between low and high FPD participants, (n = 4 per group). Figure 3.7: Comparison of flow rate and metabolite output (concentration x 109 flow rate) following oral exposure to 10 ml water control and 0.25 M sucrose solutions. Figure 4.1: An outline of the experimental design with sample and control 116 preparation and the relevant incubation conditions. Figure 4.2: Comparison of taurine in biofluids. Data (n = 11) were analysed by 131 repeated-measures ANOVA with Tukey’s post-hoc test. Figure 4.3: Partial 600 MHz 1D CPMG 1H-NMR spectra (0.7 to 8.5 ppm, 132 excluding 4.5 to 5.5 ppm), comparing parotid saliva and submandibular/sublingual saliva from the same donor. Figure 4.4: Investigating taurine content of buccal epithelial cells. 133 Figure 4.5: Comparison of taurine concentrations in GCF and WMS. 134 Figure 4.6: Ultrasonication effects on salivary protein. 135 Figure 4.7: Example gels showing parotid saliva protein changes following 137 inoculation with oral microbes.

14

Figure 4.8: Comparison of the protein consumption from parotid saliva when 138 inoculated with different sources of oral bacteria/biofilm, as measured by lane densitometry. Figure 4.9: Summary of Log10 CFU/ml measured from the inoculated parotid 139 saliva incubated under anaerobic and aerobic conditions. Figure 4.10: Relationship between final bacterial concentrations and protein 139 consumption from the inoculated parotid saliva under anaerobic and aerobic incubation conditions. Figure 4.11: Partial 1D 600 MHz CPMG 1H-NMR spectra comparing PBS- 141 inoculated parotid saliva (top spectrum) and tongue biofilm-inoculated parotid saliva (bottom spectra). Figure 4.12: PCA plot and k-means cluster analysis of all incubated and non- 146 incubated control sample metabolite profiles. Figure 4.13: PCA plots of sample metabolite profiles separated by inoculum 147 type. Figure 4.14: A summary of the significant (p < 0.05) correlations between 148 protein consumption and change in metabolite concentration under both aerobic and anaerobic incubation conditions. Figure 5.1: Schematic of the order and timing of control/tastant 155 administration and sample collection and analysis. Figure 5.2: Effects of tastants on salivary flow rates. 156 Figure 5.3: Summary of the oral microbial metabolic pathways following exposure to sucrose. 191 Figure 6.1: Illustration of the Euclidean distance measurements made to assess inter- vs. intra-week stability of salivary protein and metabolite profiles. 199 Figure 6.2: Salivary flow rate of all samples collected over the course of this experiment. 201 Figure 6.3: Intra- and inter-individual variation in salivary flow rate. 202 Figure 6.4: PCA plots of saliva major protein profiles of trained panellists (above) and untrained panellists (bottom). 203 Figure 6.5: Stability of major salivary protein profile. 204 Figure 6.6: Comparison of the inter- and intra-individual variation of specific major salivary proteins. 205 Figure 6.7: PCA plots of salivary metabolite profiles of trained panellists (above) and untrained panellists (bottom). 207 Figure 6.8: Variability of salivary metabolite profiles. 208 Figure 6.9: Comparison of the inter- and intra-individual variation of salivary metabolites. 209 Figure 6.10: A comparison of detection and recognition thresholds between two weeks for bitter, sweet and sour tastes. 211 Figure 6.11: Comparison of detection and recognition thresholds for bitter, sweet and sour tastes between trained and untrained panellists. 212 Figure 6.12: Results of salivary metabolites and spectral regions statistically different between sensitive and insensitive tasters on both weeks. 215 Figure 6.13: Comparison of significantly different salivary protein band intensities between trained and untrained panellists. 216 Figure 6.14: Comparison of salivary metabolite concentrations between trained panellists, untrained panellists and age-matched female controls. 217 Figure 7.1: Flowchart summarising the data handling and analyses for assessing salivary composition and intra-oral catabolism in different groups of participants based on suprathreshold taste sensitivity. 233

15

Figure 7.2: Distributions of participant taste ratings for all tastants. The demarcations indicate mean and standard deviation. 234 Figure 7.3: Summary of correlations for salivary flow rate, total protein concentration, and taste ratings between monozygotic twin pairs (n = 26). 235 Figure 7.4: Comparison of the similarity of taste perception (7.4 a.) and salivary composition (7.4 b.) in monozygotic twins, dizygotic twins and unrelated individuals within the dataset. 237 Figure 7.5: Box and whisker plots summarising the distribution of intensity perception ratings for each tastant. 240 Figure 7.6: Summary of spectral regions of unstimulated saliva statistically differing between insensitive and sensitive perceivers for Sucrose, Aspartame, Oleic acid and Black tea. 241 Figure 7.7: Summary of differences in salivary flow rate and composition between insensitive and sensitive sucrose perceivers common to unrelated individuals and twins discordant in their intensity rating for sucrose 243 Figure 7.8: Summary of differences in salivary composition between insensitive and sensitive sucrose perceivers only present when comparing unrelated individuals. 243 Figure 7.9: Summary of flow rate and protein concentration differences observed in sensitive and insensitive aspartame perceiving discordant twins. 244 Figure 7.10: A significant difference in salivary composition was observed for formate between unrelated sensitive and insensitive caffeine perceivers only. 244 Figure 7.11: Summary of differences in salivary metabolite composition between insensitive and sensitive oleic acid perceivers common to unrelated individuals and twins discordant in their intensity rating for oleic acid. 246 Figure 7.12: Additional differences in salivary metabolite composition between insensitive and sensitive oleic acid perceivers only detected in unrelated individuals. 246 Figure 7.13: Summary of different salivary composition between sensitive and insensitive unrelated perceivers of Black tea astringency. 247 Figure 7.14: Summary of citrate:pyruvate and lactate:pyruvate ratios following sucrose rinse in sensitive and insensitive sucrose perceivers. 248 Figure 8.1: Summary of the process involved in shaping the salivary metabolome and how these net events might alter taste perception. 258 Figure 8.2: Theoretical outline of the dynamic events occurring on exposure to tastants such as capsaicin. 262 Appendix A Figure 1: Comparison of the effects of SCFA supplementation on the proliferation of TR146 cells. 285 Appendix B Figure 1: Summary of Sucrose and caffeine taste intensity from ten participants following oral conditioning with water or urea and taurine solutions. 286 Appendix B Figure 2: Summary of Taurine and urea pre-treatment on the intracellular calcium response of TR146 cells to glucose. 288

16

Table of Tables

Table Page Table 1.1: Summary of changes in salivary ion composition following 23 stimulation. Table 1.2: A summary of studies investigating metabolite composition of 29 saliva as a source of disease biomarkers. Table 1.3: A summary of the literature base involving metabolomic study 34 of oral microbial communities in human subjects. Table 3.1: Summary of tastant and solvent concentrations investigated at 73 the initial visit. All concentrations in ppm and percentage are by volume. Table 4.1: Summary of the concentrations of metabolites consumed and 142 generated following 24 h anaerobic incubation of parotid saliva inoculated with oral bacteria relative to inoculation with sterile PBS. Table 4.2: Summary of the concentrations of metabolites consumed and 144 generated following 24 h aerobic incubation of parotid saliva inoculated with oral bacteria relative to inoculation with sterile PBS. Table 5.1: Changes in major salivary proteins and metabolite 159 concentrations and outputs following 0.25 M sucrose. Table 5.2: Changes in major salivary proteins and metabolite 161 concentrations and outputs following 8 mM caffeine. Table 5.3: Changes in major salivary proteins and metabolite 163 concentrations and outputs following 250 ppm menthol. Table 5.4: Changes in salivary rheology following sucrose, caffeine and 165 menthol, p-values are for two-tailed paired t-test, (n=10). Table 6.1 – Summary of the order of taste threshold tests conducted in 197 this study. Participants were blind to the tastants in every session. Table 6.2 Summary of the taste concentrations used in threshold 197 determination for this study, adapted from ISO 3972:2011. Table 6.3: Ranking of the intra and inter-individual stability of salivary 210 proteins and metabolites. Table 6.4: Summary of the salivary analytes that were significantly 213 different between individuals with sensitive or insensitive taste thresholds. Table 7.1: Summary of salivary compositional differences between 242 sensitive and insensitive sucrose tasters. Table 7.2: Summary of salivary compositional differences between 245 sensitive and insensitive oleic acid perceivers. Table 7.3: Summary of metabolite output changes following sucrose 247 exposure in sensitive and insensitive sucrose perceivers, n = 9 per group.

17

18

Chapter 1: Introduction

1.1 Salivary secretion

1.1.1 Macroscopic anatomy of human salivary glands and ducts

Saliva is an exocrine fluid secreted into the oral cavity. In humans, saliva originates from three bilateral pairs of major glands and several hundred minor glands. The major salivary glands are the parotid, submandibular and sublingual glands 1. Parotid gland secretions drain into the mouth adjacent to the maxillary first molars via the parotid (Stensen’s) duct. Submandibular and sublingual gland secretions enter the floor of the mouth below the tongue. Submandibular gland saliva is conducted by the submandibular (Wharton’s) duct and sublingual gland saliva by Bartholin’s duct. Bartholin’s duct converges with the submandibular duct to form a common opening called the salivary caruncle. Sublingual gland secretions also empty into the floor of the mouth via short conduits termed the ducts of Rivinus 2. Minor salivary glands are located directly beneath the buccal, labial and palatal mucosae and secrete directly onto the mucosal surfaces 3. The major salivary glands and other relevant structures are outlined in Figure 1.1.

Figure 1.1: Illustration of salivary gland anatomy. a. shows the major salivary glands and their duct systems. b. shows the submucosal distribution of minor palatal minor salivary glands. Minor glands are also distributed beneath the buccal and labial mucosae (not pictured). Images are adapted from earthslab.com 4. 1.1.2 Histology of salivary glands

19

Salivary glands consist of several cell types organised in a specific architecture. The main saliva-producing cells are called acinar cells and are classified as either mucous or serous depending on the type of saliva they secrete. Acinar cells are organised into small lobular bunches and are surrounded at their basal ends by myoepithelial cells, which are flattened, contractile cells that aid fluid secretion from the acinar cells. The parotid gland is composed of entirely serous acini, minor glands feature entirely mucous acini, and the submandibular and sublingual glands have mixed serous and mucous acini. Each acinar unit merges with ductal cells called intercalated ducts. Intercalated ducts from multiple acinar units converge into striated ducts, which in turn converge into excretory ducts 5 6. The different ductal cells also contribute to the composition of the fluid before entering the oral cavity. Ductal cells contribute about 15 % of the total protein in saliva and modify the ionic composition, converting the isotonic fluid produced by acinar cells into a hypotonic fluid 7.

1.1.3 Salivary secretion

Fluid secretion by acinar cells is dependent on the existence and maintenance of ion concentration gradients across the cell membranes. Acinar cells accumulate sodium, potassium and chloride ions via co-transporters on their basolateral membranes. Acinar cells actively transport excess sodium ions (via sodium/potassium ATPase pumps) at their basal membranes to maintain a constant positive sodium gradient (extracellular to intracellular). Potassium is similarly removed from the cells. This sodium gradient allows ongoing intracellular accumulation of chloride which is subsequently released into the intercalated duct lumen at the apical surface to drive saliva production. Sodium follows chloride and enters the duct lumen by leaking between acinar cells. Water follows the flux of chloride and sodium, by entering between acinar cells or by aquaporins (water channels) on acinar cell membranes. This process results in an isotonic fluid at the point of entry at the intercalated ducts.

At this point the fluid enters striated ducts where the ductal cells actively remove sodium and chloride from the fluid and secrete potassium and bicarbonate. Ductal intercellular junctions do not permit water movement, thus the secreted saliva is hypotonic 8 9. Salivary proteins are mostly packed into membrane bound secretory granules following their synthesis and these granules are transported close to the apical membrane. Upon stimulation, neural signals trigger an intracellular signalling cascade and proteins are released by exocytosis. To balance the negative charges of proteins the secretory granules are rich in calcium. Upon exocytosis the calcium enters the fluid alongside the protein 8 10. A schematic layout of a salivary acinus 20 and duct, along with illustrations of the processes of fluid and protein secretion is illustrated in Figure 1.2.

Figure 1.2: Illustration of salivary acinar unit and intercalated duct. Acinar cells labelled A., B., C. and ductal cell D. illustrate different process in saliva production. A. shows the process of protein secretion. Sympathetic neurotransmitters (noradrenaline or VIP) bind to their respective G-protein coupled receptors (GPCR), activating adenylate cyclase. Cyclic AMP is generated, in turn elevating protein kinase A, triggering exocytosis of stored protein from intracellular vesicles. B. shows the signal transduction for parasympathetic mediated fluid production. Acetylcholine binds a GPCR, activating phospholipase C and producing inositol triphosphate. Increased inositol triphosphate mediates intracellular calcium release from stores in the endoplasmic reticulum, activating Cl- release via Ca2+ gated Cl- channels at the apical membrane. Cl- release is the first step in fluid production. C. illustrates sodium and water following the Cl- flux by permeating between acinar cells, as well as the maintenance of

21 ion balance within acinar cells. Cl-, K+ and Na+ enter the cell via basolateral cotransporters, with excess Na+ and K+ being removed by separate ion pumps. D. shows the conversion of isotonic fluid into hypotonic saliva in the duct system by the removal of Cl- and Na+ and the

+ - secretion of K and HCO3 ions. Water cannot permeate between ductal cells thus the modified fluid remains hypotonic. A contractile myoepithelial cell is shown surrounding the acinar unit.

1.1.4 Neural regulation of salivary secretion

Salivary secretion is an autonomic process and glands continually receive both sympathetic and parasympathetic innervation. These efferent signals can occur simultaneously and are not antagonistic, but together modulate the volume and composition of saliva produced 11. Whilst awake, autonomic inputs regulate a “resting” level of salivary secretion, averaging around 0.35 ml/min. During sleep saliva is still produced, however, flow rate is reduced below 0.1 ml/min 12. Furthermore, salivary secretion displays a circadian rhythm, gradually increasing from waking until evening then declining at the same rate until sleep 13. In the presence of physiological stimuli, autonomic reflexes can stimulate rapid short-lived increases in salivary flow rate and differential activation of major salivary glands. These stimuli can be chemical, such as detection of tastant molecules or transient receptor potential (TRP) agonists by oral tissues or physical, such as thermal changes or mechanoreceptor activation during mastication 8. Resting saliva is of approximately 70% submandibular/sublingual origin and 30% parotid origin, upon stimulation parotid flow rate may increase more than ten-fold, and represent approximately 50% of stimulated saliva 14 15.

1.2 Salivary composition and function

Saliva serves a multitude of physiological functions essential to health. These can be broadly categorised into protection of oral tissues (from potential infection, physical and chemical insults) and facilitation of oral processes. The latter includes lubrication of oral tissues during speech and all steps of eating from solubilising food to allow taste perception, to lubrication of the oral cavity during chewing, bolus formation and swallowing 16. Salivary function is conferred by its biochemical composition. At the point of secretion, saliva is an aqueous solution of ions and proteins. As with flow rate, salivary ion and protein composition is dependent on the presence or absence of stimuli.

1.2.1 Major salivary ions and their function

22

While the major role of ions in saliva could be argued as driving the secretion of fluid from acinar cells as described above, salivary ions do serve some purposes following secretion into the mouth. The major ionic composition of saliva is dependent on the magnitude and duration of stimulation 17. Furthermore, large inter- and intra-individual variation is observed in the absence of stimulation 18. Upon stimulation, the concentrations of sodium, chloride and bicarbonate all increase from unstimulated levels to a peak then slowly reduce in a time dependant manner. Potassium and phosphate ion concertation drops then slowly increases to baseline levels and calcium remains stable. These changes are summarised in Table 1.1 Important ions linked to function include calcium and phosphate in maintaining an equilibrium favouring tooth tissue mineralisation and the large increase in bicarbonate which has a role in buffering dietary acids and maintaining a physiological salivary pH 19.

Table 1.1: Summary of changes in salivary ion composition following stimulation. Adapted from Dawes, 2008. Ion Salivary Salivary concentration Maximum change concentration pre- post-stimulation in concentration stimulation (mM) (mM) (mM) Potassium (K+) 20 14 -6 Sodium (Na+) 5 32 +27 - Bicarbonate (HCO3 ) 5 16 +11 Chloride (Cl-) 16 34 +18 Phosphate (P+) 5 2 -3 + Calcium (Ca2 ) 1 1 0

1.2.2 Major salivary proteins and their function

As with ions, salivary protein composition is variable. Different major glands express different proteins, and thus confer different functionality to saliva. Amylases represent the most abundant proteins in saliva and are produced by all major glands. Salivary amylase is not believed to be crucial in the digestive process of . Rather, amylase aids the breakdown and oral clearance of residual food material, conferring protection of oral tissues by depriving oral microorganisms of substrate 20. Proline-rich proteins (PRPs) are another major component of parotid and submandibular saliva. PRPs are a heterogenous family of proteins containing a large percentage of proline residues which have a well-studied affinity for binding dietary polyphenols 21. Glycated PRPs have also been demonstrated to bind oral bacteria22. Other major salivary proteins which function in the protection of the oral tissues are histatins, cystatins and statherin. Cystatins are a family of cysteine protease inhibitors which are 100- fold more abundant in submandibular saliva compared to parotid saliva. The role of cystatins

23 is in regulation of microbial proteases and binding hydroxyapatite in the process of hard- tissue pellicle formation 23. Statherin was the first salivary protein to be sequenced and displays an affinity to bind mineralised tissue in the mouth. Statherin is thus an important constituent in of hard tissue pellicles and also serves to prevent precipitation of salivary calcium and phosphate from solution 24,25. Histatins are histidine rich proteins which have been demonstrated to have anti-fungal properties, particularly against C. Albicans, a common oral commensal and opportunistic pathogen 26.

The lubricating and surface coating properties of saliva are derived largely from high molecular weight glycoproteins known as mucins. The predominant mucins in saliva are MUC5B and MUC7 (formerly termed MG1 and MG2, respectively). MUC5B is the highest molecular weight salivary protein and has the capacity to form gels whereas MUC7 is smaller and remains monomeric 27. MUC5B has a molecular weight of 20 to 40 MDa, and MUC7 150- 180 kDa 28. The abundant residues in mucins can contribute to over half of the proteins mass and are critical to the mucin’s functionality. Mucins can form complex suprastructures which aid in their rheological properties, helping maintain a thin film of saliva on the surfaces of the mouth reducing friction and preventing desiccation of the soft tissues 29 30. The role of MUC5B and MUC7 in protecting the mouth extends beyond simply the formation and maintenance of a physical barrier. As constituents of both mucosal and hard tissue pellicle, mucins have been demonstrated to bind other anti-microbial salivary proteins and maintain a concentration of these defensive molecules within the mouth 31,32. Furthermore, MUC7 has been shown to bind bacteria in the mouth via sialic acid residues representing a further protective modality of salivary mucin 27.

The complex physical properties of saliva are not explained solely by mucins, however. For example, attempts to replicate the lubricating properties of human saliva using porcine or bovine-derived purified mucin are unsuccessful. It is recognised that the interaction other, lower molecular weight salivary proteins with salivary mucins are critical in conferring lubricating properties to the fluid. Mucin and non-mucin protein interactions are still not fully understood however these represent an important future direction to further understand the link between biochemical composition and physical behaviour of saliva 33. Non-mucin salivary proteins which have been demonstrated to confer lubricity to saliva in the presence of mucins include lactoferrin (which, being positively charged, is believed to bridge negatively charged layers of mucin) 34, and proline-rich proteins, which would be capable of interacting with mucins in a similar manner to lactoferrin 35. Other low molecular-weight salivary proteins such

24 as cystatin-S and statherin have also been implicated in the surface coating properties of saliva 35 36.

1.2.3 Lower abundance salivary components

In terms of abundance, the vast majority of both ions and proteins in saliva are composed of a relatively small number of constituents. However, the diversity of ions and proteins that compose the residual content is large. Elemental analysis has revealed that saliva contains numerous trace elements. Of these low abundance ions, several are important minerals such as zinc, magnesium and copper whereas others likely represent environmental bioaccumulation such as caesium and mercury 37 38.

With respect to proteins, advances in proteomic techniques such as mass spectrometry indicate the number of unique proteins present in saliva is of the order of thousands. A meta- analysis of multiple salivary proteomic studies has revealed 4,833 unique proteins found in saliva 39. There is wide inter-individual variation in the salivary proteome and datamining suggests many salivary proteins have overlapping functions 40 41. It has therefore been suggested that low concentration proteins could be absent from an individual’s saliva without detriment to health, as any functional role would be compensated by other analogous proteins 42.

1.2.4 Non-exocrine components of saliva

The fluid present in the oral cavity at a given time is not simply the net products of the various major and minor salivary glands. The fluid produced on expectoration, termed “whole-mouth saliva” (WMS) inevitably contains tens of thousands of eukaryotic cells either as shed buccal epithelial cells or leucocytes, which have entered the mouth from the circulation, primarily via the junctional epithelium surrounding the teeth as well as tonsillar tissue 43. Furthermore, the number of bacteria present in WMS can be as high as one billion per millilitre. An image of the cellular content of saliva is shown in Figure 1.3. The host and bacterial cells present in WMS contribute their own proteome to that of WMS whilst at the same time providing proteolytic which degrade the proteins produced by salivary glands 44. In terms of number of proteins, proteins derived from host cells have been shown to compose a majority of enamel pellicle proteins (68 %) compared to proteins of glandular origin (14 %). While this finding does not reflect the pellicle composition in terms of protein mass, it does suggest cellular derived proteins have an important role in pellicle formation 45. 25

Figure 1.3: A micrograph of a Gram-stained WMS droplet showing the cellular content. Epithelial cells are stained lightly pink, and the darker purple stains are clusters of oral bacteria.

Other contributions to WMS include gingival crevicular fluid (GCF) and secretions from Von Ebner’s glands (VEG). GCF is a modified blood serum that enters the mouth at the gingival sulcus. Flow rate of GCF is several orders of magnitude below that of saliva and thus GCF represents a small fraction of WMS by volume 46. Despite the low volume produced, GCF is rich in immune cells, cytokines and proteolytic enzymes thus the fluid has long been recognised as a valuable source of biomarkers reflecting oral health status 47. As with GCF, fluid produced by VEG also represent only a small percentage of net WMS. VEG are located on the posterior of the tongue adjacent to taste papillae and produce a serous fluid. As with other salivary glands, VEG flow rate increases upon stimulation 48. In the rat, VEG produce lingual lipase which is demonstrated to have significant lipolytic activity post-swallowing. In humans the role of VEG lipase is less clear as it far less abundant and appears to have negligible role in 49 50.

1.2.5 The oral microbiome

26

The number and diversity of microorganism species in the human mouth is vast. Many oral microbial species cannot be cultured conventionally and techniques such as 16s RNA sequencing place the number of oral bacterial species over 1000. Additionally, such techniques reveal the healthy oral cavity is home to a large number of fungal species, viruses and candidate phyla radiation (“ultrasmall bacteria”) 51.

A considerable majority of oral bacterial species are harmless commensal organisms and many even confer health benefits. Important benefits include preventing opportunistic pathogens from colonising oral surfaces and the conversion of salivary nitrate to nitrite, allowing nitric oxide production which confers cardiovascular benefits to the host. Dysbiosis of oral microbial communities is, however, associated with diseases such as dental caries and periodontal disease. In both cases the majority of the damage is done by a select few species of microorganisms, and restoration of a healthy oral microflora by improved oral hygiene or dietary alteration can cause cessation of disease progress 52 53. It has been proposed that the role of saliva in maintaining and nourishing a healthy bacterial population is just as important to oral health as actively fighting pathogenic bacteria 54.

1.3 Metabolic composition of saliva

1.3.1 Historical context and advent of metabolomics

An important and arguably overlooked group of salivary constituents are metabolites. There is a fairly long history of the study of individual metabolites within saliva including urea, citrate and lactate 55 56,57. Despite this, the biological significance of salivary metabolites has not been scrutinised in the same way as those of salivary proteins. This may be best exemplified by lactate. Historically, blood lactate was considered nothing more than a waste product of anaerobic . Lactate is now increasingly recognised as serving important physiological roles as a metabolic intermediate and cell-signalling molecule 58 59. There is no literature indicating a similar paradigm shift regarding lactate produced orally, and it continues to be widely considered solely as a detrimental waste product.

In the same way that technology-driven proteomic analysis has revealed thousands of unique salivary proteins, continual discovery of metabolites in saliva is uncovering a diverse and complex array of molecules. The adoption of a systematic approach to characterise the metabolic composition of a biological material is termed “metabolomics”. Over recent years there has been rapid growth in the study of salivary metabolomics although studies of the 27 salivary proteome dominate salivary “omics” research. The discrepancy in cumulative publications in each field is illustrated in Figure 1.4.

N u m b e r o f s t u d i e s o f s a l i v a r y

p r o t e o m e v s m e t a b o l o m e

2 5 0

s

n

o i

t 2 0 0 a

c P r o t e o m e

i l M e t a b o l o m e

b 1 5 0

u

p

e

v 1 0 0

i

t

a l

u 5 0

m

u C 0 2 0 0 0 2 0 0 2 2 0 0 4 2 0 0 6 2 0 0 8 2 0 1 0 2 0 1 2 2 0 1 4 2 0 1 6 2 0 1 8 2 0 2 0

Y e a r

Figure 1.4: Growth of literature concerning studies of the human salivary metabolome and proteome. Publications yielded by the search terms “saliva OR salivary AND proteome OR proteomics” or “saliva OR salivary AND metabolome OR metabolomics” were searched in web of science (September 2019). Studies of non-humans were excluded. The imbalance between the study of salivary proteome and metabolome is illustrated, with the former receiving much more attention than the latter.

1.3.2 Current knowledge of salivary metabolomics

Although more is known about the salivary proteome, knowledge of the salivary metabolome is advancing rapidly. The most comprehensive metabolomic study of saliva to date involved multiple platforms and reported 308 metabolites in saliva. The true number is likely to be much higher as the same study combined literature mining to reveal 853 metabolites have been reported in saliva. Like salivary ionic and protein composition it appears the majority of metabolite abundance is composed of a small number of metabolites, with the majority of metabolites being present at very low concentrations 60. Much of the recent literature focuses on the potential for salivary metabolites as biomarkers of various pathologies. A summary of this research is presented in Table 1.2, illustrating the disease studied, the metabolite profiling technology and resulting metabolites proposed as biomarkers.

28

Table 1.2: A summary of studies investigating metabolite composition of saliva as a source of disease biomarkers. Disease Metabolomic Biomarkers identified Reference technique (+ = raised in disease; - = lowered in disease) Oral cancer Capillary Alanine, Taurine, (Sugimoto et al., electrophoresis time- Leucine/isoleucine, 2010) 61 of-flight mass Histidine, Valine, spectrometry (CE-TOF- Tryptophan, Glutamic acid, MS) Threonine, Carnitine, Breast Pipecolic acid (all +) cancer

Alanine, Leucine/isoleucine, Valine, Glutamic acid, (all +); Prostate Taurine, Threonine (-) cancer Alanine, Leucine/isoleucine, Histidine, Valine, Tryptophan, Glutamic acid, Threonine, Carnitine, (all +); Periodontal Taurine (-) disease Alanine, Leucine/isoleucine, Valine, Tryptophan, Threonine (all +); Pipecolic acid, Taurine (-) Oral cancer Ultraperformance , n-Eicosanoic acid (Wei et al., 2010) liquid chromatography (+) 62 coupled with Valine, γ-aminobutyric acid quadrupole/time-of- (GABA), Phenylalanine (all -) Oral flight mass leukoplakia spectrometry (UPLC-QTOFMS) Isoleucine, phenylalanine, threonine, homocysteine, n- tetradecanoic acid, 4- methoxyphenylacetic acid (all +) Oral cancer Reverse phase liquid Propionylcholine, succinic (Wang et al., chromatography and acid, lactic acid (all +) 2014) 63 hydrophilic interaction chromatography with Acetylphenylalanine, time of flight mass carnitine, phytosphingosine spectrometry (all -)

Oral cancer CE-TOF-MS 3-phosphoglycerate, (Ishikawa et al., carnosine, 2016) 64 phosphoenolpyruvate,

29

dihydroxyacetonephosphate (all -)

Extensive list of metabolites (all +) Dental Proton nuclear Propionate, butyrate (+) (Fidalgo et al., caries magnetic resonance 2013) 65 spectroscopy (1H-NMR) Saccharides (-) Dental 1H-NMR Butyrate, acetone (-) (Pereira et al., caries 2019) 66 Extensive list of amino acids and organic acids (all +) Periodontal Ultra-high- Phenylpropionate, (Liebsch et al., disease performance liquid phenylacetate (+) 2019)67 chromatography and tandem mass spectrometry (UHPLC- MS/MS) Periodontal 1H-NMR Acetate, GABA, propionate, (Aimetti et al., disease n-butyrate, succinate, 2012) 68 trimethylamine, valine (all+)

Pyruvate, n-acetyl groups (-) Aggressive 1H-NMR Isoleucine, proline, valine, (Romano et al., periodontitis tyrosine, phenylalanine (all 2018) 69 +) Lactate, pyruvate, n-acetyl Chronic groups (all -) periodontitis Formate, phenylalanine, tyrosine (all +) Sarcosine, Lactate, pyruvate, n-acetyl groups (all -) Primary Gas chromatography‐ Extensive list of metabolites (Kageyama et al., Sjögren’s mass spectrometry (all -) 2015) 70 syndrome (GC-MS) Primary 1H-NMR Alanine, glycine, butyrate, (Mikkonen et al., Sjögren’s taurine, phenylalanine, 2013) 71 syndrome choline, tyrosine (all +) Dementia CE-TOF-MS Tyrosine, arginine (+) (Tsuruoka et al., 2013) 72 Dementia 1H-NMR Acetate, histamine, (Figueira et al., propionate (all +) 2016) 73 Dimethyl sulfone, glycerol, succinate, taurine (all -) Alzheimer’s Faster ultra- Sphinganine-1-phosphate, (Liang et al., disease performance liquid ornithine, phenyllactic acid 2015) 74 chromatography mass (all +) spectrometry (FUPLC- Inosine, 3-dehydrocarnitine, MS) hypoxanthine (all-)

Alzheimer’s 1H-NMR Acetone, propionate (+) (Yilmaz et al., disease 2017) 75 30

While the studies listed in Table 1.2 demonstrate salivary metabolites can be of use in discriminating a wide range of pathologies from healthy controls, application of this knowledge is limited without a thorough understanding of the physiological changes in salivary metabolite profile in health.

Studies have been conducted to investigate the inter- and intra-individual stability of the salivary metabolome, circadian effects on salivary metabolites and a range of exogenous factors that might influence the metabolic composition of saliva. MS metabolic profiling showed 15% of salivary metabolites, the majority of which being amino acids, have been demonstrated to have a circadian rhythm. For tyrosine and arginine this rhythm appears to mirror that of total protein concentration demonstrated by previously by Dawes, indicating these metabolites may reflect overall gland activity 76 13. This circadian nature has also been demonstrated by the fact that intra-day variability of salivary metabolomic profile is greater than inter-day variability. It appears clear that the timing of sample collection is an important factor in the metabolomic study of saliva. Interestingly, the flow rate of saliva was demonstrated to be independent from the metabolic variations 77. Circadian effects have also been demonstrated by NMR metabolic profiling, although metabolite profile was shown to be independent of sex and body mass index (BMI) 78. The intra-individual variation both within- and between-days has been shown to be considerably lower than the inter-individual salivary metabolite profile. The first saliva collected on waking however was found to be significantly richer in many metabolites which may represent changes in saliva production during sleep 79. Saliva has also been found to be independent of diet, as dietary standardisation can reduce inter-individual variation in the metabolic composition of urine, but a similar reduction was not observed with saliva 80.

Other factors influencing the metabolic composition of saliva include the presence of stimulation and smoking 81. Stimulation using citric acid appeared to cause a decrease in the majority of metabolites measured, which although attributable to dilution due to increased fluid content of the saliva, did not dilute metabolites proportionally, indicating metabolite specific effects. Smoking was associated with raised sucrose, lactate, pyruvate and citrate and reduced formate compared to controls. In contrast to Bertram et al., (2009) 78, who considered net metabolite profile Takeda et al. (2009) 81, did report sex based metabolic differences in individual metabolite quantification. Stimulation by chewing also appears to cause contrasting changes in metabolite concentration compared to citric acid stimulation. Whereas citric acid stimulation reduced the concentration of metabolites, chewing has been 31 demonstrated in MS and 1H-NMR studies to cause an increase in the concentration of the majority of metabolites. This was attributed to the mechanical action of chewing liberating metabolites from oral biofilms 82 83.

A large study of physiological and environmental factors demonstrated salivary metabolite profile is not significantly affected by lifestyle factors such as alcohol consumption, fasting, oral hygiene, medication use or supplementation. The largest effects were observed for sex, followed by collection method and then smoking. BMI was shown to have an effect only when comparing underweight individuals to healthy weight, but not when comparing overweight individuals. Menstrual status had no effect when considering female participants 84. Collection method as a source of variation has been corroborated, with the use of collection systems as opposed to passive drooling being identified as a potential source of exogenous analytes in samples 66. Literature is sparse on the effects of stimulation beyond the common citric acid or chewing. One study has shown an oleic acid stimulus causes alterations is metabolic composition. Importantly, the authors found that the changes within saliva were dependant on the participants sensitivity to the stimulus. The fact that the same stimulus can produce different changes in different individuals further underlines the complexity of measuring the salivary metabolome 85.

Although there has been considerable progress in the understanding of the dynamic nature of the healthy oral metabolome research remains lacking with respect to some important questions. These are primarily concerning the source of salivary metabolites and the physiological significance, if any, of the various metabolites. There is a general consensus in literature that many of the metabolites present in saliva are produced by the bacteria in the mouth. This bacterial contribution has been speculated since the earliest metabolomic studies of saliva 86,87. Since then there has been little research directly focused on the host and microbiome contributions to salivary metabolites, although the need for such studies has been recognised 81. Relatively detailed study has been made of the role of oral bacteria in contributing to the salivary proteome, with many proteins being attributed to specific bacteria at the genus level 88. Equivalent study of salivary metabolites has not been conducted. A further source of inconsistency in literature which has not been assessed directly is the substrate for the bacterial generation of metabolites in the mouth. For example, acetate, the most concentrated salivary metabolite, has been speculated as being produced by either carbohydrate , collagen degradation or metabolism 60,68.

32

Furthermore, little is known about the metabolic composition of glandular saliva before it comes into contact with oral bacteria. The few metabolomic studies of parotid saliva have either been performed on samples from a single participant or hindered by technological limitations 89 90. Investigation of the host or microbial origin of salivary metabolites, as well as the substrates for metabolite production is the subject of Chapter 4 of this thesis. Knowledge regarding physiological functions of salivary metabolites is also lacking. There is evidence that the metabolomic profile of salivary metabolites may be related to the oral detection of fat (oleic acid) 91, but beyond this there is minimal research to date. Metabolomic research into the gut has shown bacterially generated short chain fatty acids (SCFAs) such as acetate, propionate and butyrate have important roles including maintaining the epithelial integrity of the colon and immune system function 92. Given these metabolites are all abundant in saliva the potential for an analogous role in supporting mucosal tissues represent an unexplored area of study.

1.3.3 Metabolomics technologies

The growth in salivary metabolomic research is driven by two major analytical platforms, namely mass spectrometry (MS) and nuclear magnetic resonance (NMR) spectroscopy. Both techniques have unique advantages and limitations. The primary advantage offered by mass spectrometry is the detection of metabolites present at extremely low concentrations. MS requires lower volumes of sample than NMR spectroscopy, however NMR spectroscopy is non-destructive and can be performed without physical or chemical alteration of the sample. Additional advantages of NMR spectroscopy over mass spectrometry include the ability to derive information about the chemical structure of the analytes and relatively simple absolute quantification of metabolites 93. A specific advantage when profiling salivary metabolites by NMR spectroscopy is the fact that important metabolites such as acetate and formate are readily detected whereas mass spectrometry can fail to detect these metabolites due to their low mass to charge ratio 94.

In order for the preliminary research of salivary metabolomics to be translated into useful clinical knowledge, particularly in the case of biomarker discovery, there is a need for multi- site studies involving large subject numbers and reproduction of results by different researchers. The need for protocol standardisation has been appreciated for salivary proteomics, and analysis of pre-treatment sample preparation factors for MS proteomics has been analysed 95. Metabolomics of biofluids such as plasma and urine have well established guidelines on sample collection, preparation and analysis by NMR spectroscopy 96. Chapter 3 33 of this thesis describes the investigation of pre-analytical treatments of saliva to facilitate the early development of a standardised protocol for metabolic profiling by NMR spectroscopy.

1.3.4 Metabolomics and the oral microbiome

It has been recognised that sequencing of oral microbial communities does not provide sufficient information about the biological activity of these communities. Metabolomics has been identified and applied as a functional measure of multispecies biofilms 97. While genomics, transcriptomics and proteomics can all be used to provide useful information these only reveal the potential genetic information and proteins within in a system, and only metabolomics demonstrates activity 98,99. There is also a growing awareness that metabolomics can extend beyond biomarker discovery, which much of the literature on saliva is centred around, and move towards revealing biochemical mechanisms 100.

Studies of human models

Beyond the oral cavity, metabolomic study of microbial communities is enhancing understanding of disease and identifying therapeutic pathways 101. In the oral cavity, progress is being made in this area. To date the number of studies applying metabolomics to the oral microbiome in human participants is low. The main studies and their significance are summarised in Table 1.3.

Table 1.3: A summary of the literature base involving metabolomic study of oral microbial communities in human subjects. Study Study design and main findings (Seerangaiyan Comparison of tongue coating of halitosis patients and controls by LC- et al., 2019) MS/MS. Halitosis patients had increased levels of metabolites including 102 branched chain fatty acids, 3-fumaryl pyruvate and acetyl phosphate suggesting altered lingual microbial metabolism producing volatile metabolites. (Prodan et al., Experimentally induced gingivitis led to increased bacterial metabolites 2016) 103 including cadaverine, α-hydroxyisovalerate and GABA as well as host derived metabolites involved in immune response. (Takahashi & CE-MS study of glycolysis and intermediates of plaque Washio, collected from volunteers before and after exposure to glucose and xylitol 2011) 104

34

in the presence and absence of fluoride allowing mechanistic deduction of the inhibitory effects of fluoride and xylitol on oral glucose catabolism. (Feng et al., Compared microbial and metabolite composition of saliva and tongue film. 2018) 105 Tongue film was composed of more Firmicutes, Streptococci and actinobacteria and fewer Bacteroidetes and Prevotella. Tongue film also contained less glucose, butyrate and lactate and more acetate than saliva. This suggests site-specific oral microbial communities and local variation in metabolite concentrations. (Tang et al., Analysis of stool and saliva microbiome and stool and plasma metabolome 2019) 106 and host diet. Dietary fibre intake was linked to the plasma metabolome but no link was found between oral microbiome and diet or plasma metabolome. (DeAngelis et Compared salivary microbiome and metabolome of healthy and treated al., 2016) 107 coeliac disease children. Associations were reported between the oral microbiome and metabolome, including Prevotella species associated with nonanal and 1-chlorodecane (abundant in healthy controls) and Veillonella parvula associated with γ-lactone (abundant in coeliac disease patients). (Ercolini et Analysis of salivary microbiome and metabolome of African coeliac disease al., 2015) 108 children adopting a gluten-free diet. Shifts in microbiome and metabolome were observed including increased Granulicatella, Porphyromonas and Neisseria and decreased Clostridium, Prevotella and Veillonella. Metabolome changes included enhanced amino acid, vitamin and co-factor metabolism following the dietary change. (De Filippis et Compared the salivary microbiome and metabolome of omnivore, al., 2014) 109 vegetarian and vegan participants. No discriminatory microbiome pattern was found however omnivores were discriminated from vegetarian and vegan subjects by increased formate, urea and uridine and decreased proline, propanol and hexanoic acid. (Skarke et al., A study of the temporal stability of the oral microbiome and metabolome 2017) 110 revealed that several microbial species display temporal fluctuations including Streptococci, Veillonella, Rothia and Actinomyces. Several metabolites displayed temporal fluctuations, including cortisol, ornithine, niacin and diphosphoglycerates, although associations between metabolome and microbiome were not investigated.

35

(Zaura et al., A large-scale analysis of the salivary microbiome and metabolome of 2017) 111 healthy adults found that participants could be classified into different microbial and metabolic phenotypes. Primarily these were defined as a proteolytic phenotype characterised by increased Veillonella and Prevotella and metabolites reflective of amino acid degradation and a saccharolytic phenotype classified by streptococci and metabolites related to lipid/carbohydrate catabolism.

Studies of animal and biofilm models A further important approach to enhancing knowledge of complex microbial communities is by the study of model organisms and their interactions either with other species or with different growth conditions. For example, monoculture of several periodontal pathogens Aggregatibacter actinomycetemcomitans, Treponema denticola, Porphyromonas gingivalis and Tannerella forsythia has metabolic biomarkers specific to each species, which were subsequently demonstrated to have diagnostic utility in patients infected with these organisms 112. Such an approach is also exemplified by Takahashi et al. (2010) 94, who studied the metabolism of monospecies cultures to help understand the metabolic differences between in-vivo oral microbial metabolism and oral bacterial species in isolation. Similarly, analysis of the metabolic changes of cariogenic oral bacterial species during carbohydrate fermentation has been utilised to unveil novel mechanisms of microbial adaptation to low pH environments 113. A link between oral health, particularly periodontal disease, and systemic health has been a topic of much debate in recent years. Metabolomic study has recently been applied to mice inoculated orally with the periodontal pathogen Porphyromonas gingivalis, revealing increased serum levels of alanine, histidine, glutamine tyrosine and phenylalanine 114. Such studies enhance understanding of the relationship between dysbiosis, altered metabolism and potential pathology.

Other work with A. Actinomycetemcomitans has suggested that in a cell-microbe model, monolaurin was found to modulate host-pathogen interactions including the metabolome 115. Metabolomic study of oral microorganisms is providing valuable insights into mechanisms of growth inhibition of oral pathogens by probiotic species. For example, production of metabolites including xylitol, L-tyrosine and L-glutamic acid by an oral strain of Bacillus subtilis correlated with inhibitory effects on the growth of oral pathogens Streptococcus mutans, Lactobacillus acidophilus and Actinomyces viscosus 116. Mono- and multi-species biofilms have

36 also been assessed for their metabolic activity and the production of peptidic small molecules. Unique patterns of molecule production were demonstrated at species level, as were dynamic changes with time when exposed to (glucose and sucrose). Although only a small percentage of these molecules were identifiable at present, peptide small molecules present a useful indicator of bacterial activity in future 117.

1.4 Taste and oral perception

1.4.1 Overview

The mouth is capable of detecting a wide range of chemical stimuli. Collectively, these detection mechanisms are important in evaluating and discriminating nutritionally beneficial substances from those that may be toxic or otherwise harmful 118. It has been speculated that unlike other senses such as sight or hearing, the sense of taste is essential to life as those deprived of their sense of taste lose their drive to eat and do not survive long without nutritional support 119. The sense of taste, being one of the five “basic senses”, is probably the most familiar mechanism of oral perception and is attuned to five primary modalities – sweet, bitter, salty, sour and umami. There are also detection systems in the oral cavity beyond those for detecting the primary taste modalities. These include known mechanisms such as the TRP- mediated sensations of capsaicin (burning/tingling) or menthol (cooling) 120, as well as sensations with strong evidence but as yet unproven mechanism such as the detection of fat molecules 121, or the perception of astringency (drying sensation) 122.

1.4.2 Anatomy of taste perception

Basic tastes are all perceived by specialised chemoreceptive cells which are organised into structures called taste buds. Taste receptor cells (TRCs) are classified into four types. Type I cells have a support role, type II cell are receptor cells involved in sweet, bitter and umami perception, type III cells detect salt and sour stimuli. Taste receptor cells undergo turnover approximately every ten days and type IV cells are progenitor cells (basal cells) which can differentiate into the other cell types. Taste buds are collections of several hundred chemosensory cells and have a rounded microscopic appearance. Taste buds are located in specialised papillae on the tongue just below the epithelial surface and are directly connected to the oral cavity by a taste pore 123 124.

37

Taste buds are present within three types of lingual papillae. These are the fungiform papillae, which are located on the anterior two thirds of the tongue, the foliate papillae which are located laterally and posteriorly on the tongue and circumvallate papillae which are located on the posterior dorsal surface of the tongue. The circumvallate papillae house approximately half of all taste buds, and remaining half are divided fairly evenly between fungiform and foliate papillae. A small number of papillae are also found within palatal, laryngeal and oesophageal mucosa. Taste signals from the circumvallate and foliate papillae are transmitted by the glossopharyngeal nerve, signals from fungiform papillae are transmitted by the facial nerve 125. A diagram of the distribution of taste papillae on the human tongue, alongside a depiction of a single taste bud is shown in Figure 1.5.

Figure 1.5: Diagram of lingual papillae and taste bud. a. depicts the different types of taste papillae found in humans. Papillae which house taste buds and are thus capable of chemoreception are the circumvallate, fungiform and foliate papillae. Filliform papillae do not respond to taste but are capable of proprioception. b. depicts a magnified taste bud on the lateral aspect of one of the circumvallate papillae. The different cell types within taste buds is depicted, as is the communication with the oral cavity via a taste pore and the connection to afferent nerve fibres. The image has been adapted from: (Gravina et al., 2013) 124.

1.4.3 Transduction and neural processing of basic tastes

Sweet taste The perception of sweet taste begins with the binding of a sweet tastant molecule to a dimeric transmembrane protein in type II TRCs named the T1R2/T1R3 receptor. A range of molecules activate T1R2/T1R3 notably naturally occurring sugar molecules such as sucrose, 38 glucose and as well a range of non-nutritive molecules (artificial sweeteners) 126. T1R2/T1R3 is a G-protein coupled receptor (GPCR). The main G protein involved in sweet taste transduction is called gustducin, although other G-proteins are involved. It is thought the signalling cascade differs depending on whether a sugar or non-nutritive tastant molecule is bound. Upon binding sugar molecules, G-protein activation of adenylyl cyclase generates cyclic AMP, which phosphorylates membrane bound potassium channels. Blockage of potassium influx causes a depolarisation and subsequent voltage-dependant calcium influx, in turn stimulating neurotransmitter release. Upon binding an artificial sweetener molecule, phospholipase-C is activated generating inositol triphosphate and diacyl glycerol which in turn stimulate release of calcium from intracellular stores, causing depolarisation and neurotransmitter release 127 128 129.

Bitter taste The range of molecules known to stimulate bitter taste is extensive. It is theorised that the unpleasant nature of bitter perception serves to cause an aversion to substances that may be toxic. Bitter molecules are perceived in a similar way to sweet molecules, by the binding of tastant molecules to GPCRs. Bitter receptors are called T2R receptors (or Tas2R). In humans, 25 T2R receptors have been discovered to date. As with the sweet-receptive T1R2/T1R3 dimers, T2R receptors are expressed on type II TRCs, however they tend to be monomeric receptors. The majority of receptors are activated by numerous different molecules, thus the number of bitter substances detectable greatly exceeds the number of receptors. Furthermore, many bitter molecules will activate more than one T2R receptor 130. The signal transduction pathway for bitter tastes is essentially the same as for sweet 131 132.

Umami taste Umami is a specific taste modality typified by monosodium glutamate (MSG) although the amino acid aspartate as well as the ribonucleotides guanosine monophosphate and inositol monophosphate can elicit the sensation. The receptors for umami taste are found in type II taste cells and are called T1R1 and T1R3. The umami transduction pathway is the same as for sweet taste 133 134.

Sour taste Sour (acidic) taste is essentially the detection of protons. Specific receptors for sour taste are as yet unidentified. Unlike the tastes discussed previously, which are mediated by type II cells, it is known that type III cells are responsible for sour perception. The general mechanism involves the entrance of protons into the receptor cells which may be directly by ion channels

39 such as PKD2L1, or in the case of organic acids, transmembrane diffusion of an undissociated proton-anion pair, which dissociates intracellularly. This change in intracellular pH then causes the blockage of potassium channels and cell depolarisation. Due to crossing membranes and dissociating intracellularly, at the same pH, organic acids are perceived as more intensely sour than fully dissociated strong acids 135 136.

Salt taste The taste of salt occurs by the detection of sodium ions. Influx of sodium is mediated by channels such as the epithelial sodium channel (ENaC) and causes cells to depolarise. It is not certain which cell type mediates salt taste 137 138. The transduction pathways of the five basic taste modalities are summarised in Figure 1.6.

Figure 1.6: Transduction of basic taste sensations. Type II TRCs recognise sweet, umami, and bitter stimuli via G-protein coupled receptors. These tastes share a broadly similar transduction pathway. Bitter, umami and artificial sweetener molecules activate phospholipase C, generating inositol triphosphate. This causes release of calcium form intracellular stores, and sodium influx via TRPM5 sodium channels. Natural sugars activate

40 adenylate cyclase, generating cyclic AMP and blocking potassium efflux. These events depolarise the cell and cause ATP release via Panexin 1 channels. ATP may then activate afferent neurones directly or activate GPCRs on adjacent type III TRCs. This intercellular communication via ATP leads to inositol triphosphate mediated intracellular calcium release causing exocytosis of serotonin (5HT) which in turn activates action potentials from afferent neurones. Type III TRCs also respond to sour taste. This can be by protons entering via specific channels or the intracellular dissociation of organic acids into protons and their anion. This leads to further depolarisation by blocking of potassium channels preventing potassium efflux and opening voltage gated calcium channels allowing calcium influx. Salt taste depolarises cells by sodium entry into the cell via channels such as the epithelial sodium channel (ENaC). For purposes of illustration this diagram shows salt taste detection by type III TRCs. This diagram summarises information from (Gravina et al., 2013) 124, (Chaudhari & Roper, 2010) 137, and (Deepankumar et al., 2019) 129.

1.4.4 Central perception of taste

Upon depolarising peripheral TRCs, taste signals must be relayed by afferent neurones to the brain to be perceived as taste. There are two proposed systems for how this process occurs. These are the “labelled line” model and the “across fibre” model. The former suggests that each TRC will recognise a specific taste modality and this information is conveyed via synapsing with a single neurone dedicated to that TRC. The “across fibre” model involves TRCs being activated in specific patterns, the combination of activation being integrated into the correct neural signal 139. There has been debate over which model is correct, and while current evidence strongly supports the labelled-line hypothesis, there are likely to be modifications to this model such as the temporal nature of neurone firing which are important in how taste signals are ultimately perceived 140 141 142. The possibility remains that both models of neural inputs may be used in a complimentary way. The central processing of taste is complex. At the insular cortex (sometimes referred to as the primary taste cortex) there is evidence that taste type, intensity and affective (emotional response) are integrated. When considering the process of eating, a myriad of other sensory information (olfactory, thermal, textural) must be processed in parallel with primary taste signals 143.

1.4.5 Additional oral sensations

41

Trigeminal/TRP mediated sensation The mouth responds to a wide range of chemical stimuli beyond those inducing the primary taste modalities. Some sensations are mediated by a family of ion channels known as transient receptor channels (TRP). These are a diverse group of receptors expressed in tissues throughout the body, involved in many physiological functions notably nociception and thermoreception. In the oral cavity TRP channels relay sensory information via the trigeminal nerve 144. The literature on oral TRP channels is focussed heavily on TRPV1 (vanilloid receptor) which is activated by heat greater than 43° Celsius, and molecules such as capsaicin (chilli) or allicin (garlic). Other oral TRP channels are TRPA1 and TRPM8. TRPA1 is co-expressed with TRPV1 and shares some agonists such as allicin but is also activates by other chemicals such as allyl-isothiocyanate (horseradish, wasabi). TRPM8 responds to cold stimuli and is activated by chemicals such menthol and eucalyptol. There is debate as to the precise localisation of these TRP within the oral cavity, particularly whether they are expressed within taste buds. Much of the underlying evidence looking at TRP localisation is derived on mice and rats and results have been conflicting. The net evidence for both TRPV1 and TRPM8 is that these receptors are not expressed directly within taste buds but are expressed on nerve fibres in close proximity to taste buds, as well as other oral tissues that do not respond to basic tastes such as the lips and cheeks 145 146.

Fat There are several means by which the mouth is involved in the detection of fat. The mouth aids retronasal olfaction of fat and is also involved in the tactile perception of fat. Over recent years there is growing evidence to support the concept of “fat taste” mediated by specific oral receptors. These are speculated to act on free fatty acids released from dietary fat by lipase activity in the oral cavity. Two potential receptors for free fatty acids have been demonstrated in the human lingual papillae – CD36, a transmembrane glycoprotein capable of binding a diverse range of molecules and GPR120, a GPCR. Free fatty acids are also capable of transmembrane diffusion and may act intracellularly within TRCs 147 148. Although a precise mechanism for fat taste has not been elucidated there is additional evidence that oral exposure without ingestive, tactile or olfactory cues initiates physiological responses indicating oral sensation must occur 149. Intra-oral lipolytic activity has been postulated to play a role in fat taste, however this remains controversial in humans. There is some evidence of a site and substrate specific intra-oral in a minority of individuals, however whether this is due to lipase from Von Ebner’s glands or other sources such as oral microbes is not definitive 150. Furthermore, lipolytic activity has been found not to differ significantly between sensitive and insensitive perceivers of oleic acid 151.

42

Astringency Another important oral sensation that can influence the ingestion of foods and drinks is astringency. Astringency is defined as a feeling of drying or loss of lubrication in the oral tissues. Astringency is caused by a range of stimuli, notably plant polyphenols found in tea and wine, although other molecules such as acids and ions such as zinc can elicit the sensation 152. Astringency is recognised as a “mouth feel” sensation and mechanoreceptive processes are thought to be prominent in mediating the sensation. While it is known that many astringent molecules are precipitated by salivary proteins, the role of this process in astringency perception is unknown 153. It is likely astringency perception is multifactorial and may be mechanistically different depending on the astringent substance detected. It has been speculated that astringency involves the disruption of mucosal pellicle proteins exposing underlying mechanoreceptors, alongside the formation of salivary protein-astringent complexes causing a sensation of “grittiness” 122. Another interaction that has been demonstrated to occur between astringent substances and saliva is a disruption of the adsorbed mucin-rich surface film, causing a measurable loss of lubrication 154. Polyphenols are theorised to alter the physical microstructure and water holding capacity by exposing hydrophobic side groups in the macromolecules of the surface film 155.

1.4.6 Sensory assessment of taste

Taste sensation varies between individuals, both in the magnitude of overall intensity as well as the interpretation of the underlying taste modalities and hedonic associations (“liking” or “disliking” the sensation). There are numerous approaches to measure taste perception each with inherent advantages and methodological challenges. There is no overall consensus on what method is superior in the assessment of taste perception, in fact the optimal method probably depends both on the sensation being investigated, the method of delivery of the stimulus and the underlying research question 156. For instance, academic research into taste perception mechanisms would probably be better focused on measuring the magnitude of perception of single tastants in solution whereas industrial product development might study the combined perceptual and hedonic responses to the food or drink in question.

Broadly, there are two methods to measure taste sensitivity – measurement of taste threshold and measurement of suprathreshold taste intensity. Measurement of taste threshold can be further divided into the measurement of detection threshold and recognition threshold. Detection threshold is the concentration of a substance at which a 43 taster can determine the presence of something above the relevant baseline control. Recognition threshold is the concentration at which the primary taste modality of a substance can be identified. Threshold measurement protocols can be varied in literature although International Standards Organisation guidelines do exist, theoretically enabling comparison between data gathered across different studies. Threshold is assessed by delivering sequentially increasing concentrations of the substance tested and recording when the participant reports detection and recognition of the substance 157. Discrimination threshold can also be a useful measurement of taste sensitivity. Tests such as the triangle test or 3- alternative forced-choice test (3-AFC) measure the ability of a taster to correctly identify a substance when presented with the substance and two controls (e.g. the same substance at lower concentrations) 158. This method has some objectivity as participants have to correctly identify the substance rather than simply self-report their identification of the substance. Methodologies for determining taste threshold are not without their limitations. Problems include participant fatigue and adaptation to the stimulus 159. The discriminatory ability is also limited as the threshold can only be defined as between the concentration at which the taste is recognised and the preceding taste concentration and individuals with different thresholds could fall within the same range. Threshold tests are also time consuming to set up and perform compare to supra threshold scaling. Discrimination tests are also subject to a range of biases from participants. These are related to factors such as the magnitude of the difference between the positive sample and controls, and participant tendencies to switch between certain “strategies” in attempting to discriminate the correct sample 160.

Suprathreshold taste assessment has a long history in sensory research. It offers advantages over threshold estimation notably in the efficiency of use and theoretical discriminatory ability if used correctly. Suprathreshold intensity measurements are performed using specifically developed scales. It would be inaccurate to describe the measurement of taste by sensory scales as “simple” as great care needs to be taken in their use and it is not uncommon for invalid scaling techniques to be used. A significant challenge when rating tastes is in knowing whether the same descriptor or intensity rating represents the same sensory experience between individuals. This is sometimes addressed by the use of a stimulus of different sense, typically auditory or visual, to provide a standard between individuals 161. Of course, the assumption is then that these different participants perceive the sound or bright light as equally intense. Given there are many factors that would affect auditory acuity or photosensitivity (e.g. age, occupation, mood) this approach is not a panacea 162.

44

The commonest and probably the best studied sensory scale used in taste assessment is the generalised labelled magnitude scale (gLMS). This is a well validated scale in taste research, is generalised across taste and sensory modalities and is a graduated scale anchored at the low end by “no sensation” and the high end by the descriptor “strongest sensation experienced”. Other scale types are also used, such as the generalised visual analogue scale (gVAS) which differs in the lack of graduations and descriptors between the high and low end 163. A summary of scales is depicted in Figure 1.7. Further modifications to these scales have been made and comparisons suggest that data gathered by gLMS or gVAS are probably not comparable, however there is no indication one scale is superior to the other 164 165. While the gLMS was in part designed to mitigate ceiling effects that were problematic with earlier iterations of psychophysical scaling (hence the “logarithmic” type structure of the gLMS) it is thought that perhaps it leans too heavily towards high intensity thus lower and moderate intensity stimuli may be inappropriately clustered at the lower end of the gLMS. Furthermore, it has also been shown that the use of graduated descriptors leads to a tendency for ratings to cluster at the descriptors rather than be uniformly spaced out 165. The gLMS can also be adapted for use in hedonic assessment. The hedonic gLMS is similar to the design for intensity data but spans from a central zero or neutral point to high and low extremes, reflecting liking or disliking 166.

Figure 1.7: A depiction of different types of suprathreshold intensity scales. gLMS = generalised labelled magnitude scale, gVAS = generalised visual analogue scale, glVAS = generalised labelled visual analogue scale, gwVAS = generalised words-only visual analogue scale. From: (Kershaw & Running, 2019 164).

Studies comparing sensory scales generally return to the concept that the choice of scale should be considered in the context of the specific experiment. What these different scales have in common is the need for careful explanation of their use to participants, familiarisation 45 with the scales, and a form of “screening” to have confidence in the ratings. If, for example, a participant was unable to conceptualise that a whisper would be rated as less intense than a conversation, which in turn would be rated at lower intensity than a shout on the scale then it would be reasonable to assume their taste intensity ratings would be unreliable 167. What is also critical is that the study design in the context of any scale used is standardised. It has been showing that taste assessment by “sip and spit” versus the same stimulus assessed by “sip and swallow” approaches yields different sensory ratings 168. Rather than designing an experiment to fit a specific scale, it may be more logical to design a scale to fit a planned experiment based on some preliminary data. The protocol development and scale design used throughout this thesis, which was essentially a generalised labelled visual analogue scale (glVAS), is described in Chapter 3 of this thesis.

1.4.7 Biological measurement of taste perception

Alongside psychometric measurements, taste perception has also been measured using biological features as a classifier of individuals taste acuity. The main methods are often examined alongside each other. These are measurement of fungiform papillae density (FPD) and genetic sequencing of the TAS2R38 gene. TAS2R38 is a gene that encodes a bitter receptor (T2R28) which has three single nucleotide polymorphisms that in combination result in two primary haplotypes. These are proline-alanine-valine (PAV) and alanine-valine- isoleucine (AVI) which in combination give rise to PAV/PAV homozygotes, AVI/AVI homozygotes and PAV/AVI heterozygotes. The TAS2R38 genotype is responsible for the bitter perception of molecules such as 6-n-propylthiouracil (PROP) and phenylthiocarbamide (PTC). PAV/PAV homozygotes will taste these compounds as strongly bitter whereas AVI/AVI homozygotes do not perceive bitterness on exposure to PROP or PTC. These individuals are often termed “tasters” or “non-tasters” 169. Although the underlying molecular genetics explaining PROP/PTC taste disturbances is a relatively recent finding, the Mendelian nature underlying differences in PTC perception was discovered shortly after the first differences in PTC taste were noted in the early 1930’s 170 171. PTC or PROP taste sensitivity was used as a proxy for assessing TAS2R38 genetics and taster status. Today, PTC/PROP sensitivity is still used in taste studies as a quick and cheap method of taste sensitivity screening although DNA sequencing is also used 172.

The term “supertasters” was adopted for PROP sensitive individuals (PAV/PAV TAS2R38 homozygotes) as it was noticed that sensitivity to PROP was also associated with sensitivity to other taste modalities. It was observed that supertasters had significantly higher density of 46 fungiform papillae, as well as higher density of taste pores (and by proxy taste buds) than non- tasters 173. This biological difference in fungiform papillae was suggested as a means of validating the ability of a suprathreshold taste scales to allow cross-group participant comparisons 174. A number of studies have subsequently found that FPD and PROP status are not linked 175 176 156, although others continue to find differences in FPD between groups of PROP sensitive and PROP insensitive individuals 177. The reasons for this variability may be due to methodological differences in the measurement of FPD 178. Both the use of FPD and PROP status have been questioned in their use as indicators of taste sensitivity. FPD is questioned due to inconsistent results, on balance favouring the view that no such link genuinely exists 179. However, FPD has also been implicated in the tactile sensitivity of the tongue thus fungiform papillae may still have an important role in perceiving texture of foods 180. PROP status has been questioned as ratings of other stimuli including sodium chloride, citric acid, sucrose and quinine hydrochloride all correlated more strongly with each other than ratings of PROP, hence they may represent better markers of taste acuity. Since PROP status is a marker of polymorphisms in a single bitter receptor, a wider range of taste stimuli may give a better representation of an individual’s general level of taste acuity 181.

1.5 The relationship between saliva and taste

Associations between salivary composition and taste perception are typically studied from two different perspectives (Figure 1.8). Taste perception has long been recognised as the most potent modulating stimulus for salivary gland activity 182. Consequently, numerous studies have looked at how different taste stimuli alter salivary flow rate, composition and physical properties. More recently, studies focusing on potential mechanisms whereby variation in baseline salivary composition can lead to different degrees of taste perception are gaining traction.

Figure 1.8: The bi-directional relationship between saliva and taste. 1.5.1 Tastant-stimulated salivary changes

47

Sour taste (typically from citric acid) is one of the most widely studied modulators of reflex salivation. Citric acid stimulation has been demonstrated as a potent stimulus of salivary flow rate as well as causing increases in sodium and chloride concentrations for the initial two minutes post stimulation 183 184. Citric acid has also been shown to stimulate the largest output of protein from the parotid glands when controlling for flow rate relative to other tastants. This approach to matching the salivary flow from different stimuli indicated that to produce a similar flow response to 0.01 M citric acid would require 0.25 M monosodium glutamate (MSG), 0.5 M sucrose and 0.5 M sodium chloride. Sodium chloride was shown to stimulate an increase in salivary protein concertation whereas other taste stimuli (sucrose and MSG) appeared to cause reduction in salivary protein concertation 185. In contrast to this data, Dawes demonstrated that sodium chloride stimulation caused significant increases in salivary protein relative to sugar, citric acid and quinine, although did not report changes in the relative protein abundance of saliva stimulated by any of the stimuli 184. More recently proteomic analysis has suggested taste stimuli alters minor salivary protein composition in the order of sour>bitter>umami>sweet using nitric acid, calcium nitrate, inosine monophosphate, and glucose as the respective stimuli 186. Further proteomic analysis of taste stimulated samples has confirmed citric acid stimulation as modifying the greatest number of salivary proteins followed by sodium chloride then the TRPA1 agonists 6-gingerol and hydroxy-α- sanshool. Importantly, increased lysozyme production following citric acid was demonstrated to have a functional role in limiting growth of certain bacterial species in-vitro 187.

Taste stimulation has also been shown to alter the physical properties of saliva. Citric acid stimulation results in a highly elastic but relatively low viscosity fluid compared to chewing- stimulated saliva, despite matching for total flow rate. These differences were attributed to differential gland activation by citric acid taste perception possibly stimulating increased mucin content from minor glands 188. Stimulation by monosodium glutamate also produces saliva with different rheological properties relative to chewing stimulation. Monosodium glutamate taste produced saliva with greater spinnbarkeit (filament forming ability) and lower contact angle (i.e. enhanced wetting properties) than chewing stimulation 189. Additional study of salivary extensional rheology following stimulation by TRP agonists also reveals an increase in spinnbarkeit following nonivamide stimulation (a capsaicin analogue) relative to control stimuli although the effect was transient limited to the first minute post-stimulation 190.

Compared to flow rate and protein composition, modulation of salivary metabolite composition by taste stimulation is less studied. Citric acid stimulation appears to cause a 48 general dilution of salivary metabolites, although not all metabolites are diluted proportionally 81. Study of the salivary metabolome following oleic acid stimulation has revealed changes in metabolite concertation that seem to be dependent on the participants’ sensitivity to the oleic acid stimulus 85.

1.5.2 Tastant mediated salivary compositional changes independent of gustatory reflexes

Oral exposure to certain substances can alter the composition of saliva independent from taste perception and gustatory reflex saliva secretion. This tends to be by chemical interaction with saliva that has already been secreted into the oral cavity. The best studied substances that interact with saliva in this way are astringent compounds. Kallithraka et al. (1998) 191, compared saliva collected before and after exposure to polyphenol rich wine or wine substitute. Saliva collected after exposure to polyphenols displayed reduced intensity of HPLC peaks corresponding to salivary protein and the appearance of new peaks likely reflecting protein-polyphenol complexes 191. Interactions between salivary proline-rich proteins and plant polyphenols have been studied extensively, often on purified salivary proteins 192. In- vitro analysis of whole mouth saliva interactions with various astringent compounds including tannin, alum and hydrochloric acid suggests the protein interactions between astringent compounds vary depending on the substance in question. While tannins precipitate most proline-rich proteins, hydrochloric acid precipitated mucins and alum precipitated PRPs and mucins 193.

Another process by which taste molecules can alter salivary composition is by catabolic processes in the mouth. A prominent example of this is the intra-oral breakdown of sucrose. Salivary pH reductions following oral sucrose exposure and the role of bacterially produced lactic acid in this process has long been recognised 194. More recent metabolomic approaches to analysis of intra-oral by measuring the metabolite content of dental plaque has allowed a more comprehensive analysis of the metabolic intermediates in this process 94. Given the ubiquity in sensory research of sucrose and its constituent monomers (glucose and fructose) as sweet tastants, intra-oral metabolic processes could represent a source of considerable inter-individual variation in the composition of the net saliva-tastant mixture. The possibility for such processes to account for some of the individual variability in taste response to sweet stimuli is addressed in Chapter 7.

49

1.5.3 The influence of saliva composition on taste perception

Salivary composition has been shown to influence the perception of various tastants. There are several proposed mechanisms by which these processes can occur. Salivary components involved in taste perception also tend to be associated with specific taste modalities. Certain salivary constituents have been implicated with multiple taste modality associations implicating a nonspecific mechanism of action.

Interaction between salivary components on taste receptors Endogenous salivary sodium was one of the first analytes to be recognised as having an association with the taste perception of salt taste (sodium chloride). McBurney & Pfaffmann (1963) 195, demonstrated that NaCl thresholds were lower following adaptation of the oral cavity to distilled water compared to NaCl thresholds following adaptation to saliva. Subsequently, Morino & Langford (1978) 196, demonstrated a correlation between participants’ salivary sodium concentrations and their NaCl detection thresholds. Increasing salivary sodium concentration by chewing has also been demonstrated at an individual level to reduce sensitivity to NaCl 197. It can therefore be concluded that taste receptor cells undergo a degree of adaptation to dissolved sodium in their surrounding fluid environment.

The potential for taste receptor adaptation to salivary glutamate (the primary stimulus of umami taste) has been investigated. It was found that taste thresholds for glutamate as well as suprathreshold intensity rating did not differ between participants with low vs. high salivary glutamate. Hedonic perception was found to differ, however, with participants with low endogenous salivary glutamate rating MSG solutions as less pleasant compared to high salivary glutamate individuals 198. The evidence is therefore less strong for a causative link between salivary glutamate concentrations and umami perception. Furthermore, unlike for sodium where salivary concentrations are typically much closer to taste threshold concentrations salivary glutamate concentration is much lower than the concentration require for detection (0.02 mM vs. 0.6 mM, respectively) 199.

Interaction between salivary components and tastants Salivary ions and proteins can also interact with tastants upon ingestion. This is typified by the buffering capacity of saliva being capable of diminishing the sour taste of acid stimuli 200. While there are several buffering systems in saliva including protein, phosphate and bicarbonate, the latter is the most potent buffering agent. A direct link between individual

50 salivary bicarbonate concentrations and sour taste perception has not been demonstrated. Salivary bicarbonate has been found to be significantly more concentrated in chronic kidney disease patients compared to healthy controls, and suprathreshold intensity rating of citric acid was significantly lower 201.

A further saliva-tastant interaction, already alluded to in the previous section, is that of salivary proteins and astringent substances. The type of salivary protein-astringent interactions described have been demonstrated to reduce in vivo astringency perception in a number of studies. These have shown that participants capacity to replenish salivary proteins following the respective astringent stimuli (both tannic acid and alum) correlates inversely with the perceived intensity of astringency 202 203. Measurement was made only on total salivary protein however, and the inclusion of astringent stimuli in a model food system as opposed to solution did not yield the same finding. Pre-conditioning the mouth by chewing stimulation has also been shown to reduce the intensity of astringency perception by increasing salivary protein relative to a water rinse 204. A subsequent analysis of salivary protein changes following tannic acid stimuli found that several salivary proteins were diminished by tannic acid in certain individuals including PRPs (glycosylated and non- glycosylated), amylase and to a lesser extent cystatins and histatins 205. Salivary protein therefore appears to be capable of mitigating the perceived astringency of certain stimuli by complexing astringent molecules thus reducing subsequent interactions with salivary protein films at oral mucosal surfaces.

A further important interaction between salivary constituents and sensory perception of food emulsions occurs via the interaction between salivary proteins, ions and ingested emulsions. For example, interaction of salivary anions with positively charged emulsifying proteins (lactoferrin) has been shown to disrupt emulsion droplets and cause their aggregation 206. There are several other mechanisms by which salivary constituents can interact with emulsified foods in order to alter their sensory properties, including flocculation either by interaction with negatively charged mucins (bridging flocculation) or via free mucins increasing osmotic pressure around droplets causing depletion flocculation. These flocculation mechanisms are associated with markedly different sensory properties, the former is associated with an astringent like loss of lubrication whereas the latter is associated with greater perception of a creamy/fatty mouthfeel 207.

51

Salivary proteins modulating tastant-receptor interactions There is a small but growing literature base looking at salivary protein composition in association with taste perception. One of the earliest studies on this topic looked at the salivary proteome of individuals sensitive and insensitive to the bitterness of caffeine. It was found that insensitive individuals had higher salivary levels of the protease inhibitor cystatin- SN and reduced levels of protein fragments originating from other salivary proteins. It was proposed that this may reflect reduced protease activity in insensitive caffeine perceivers. More intact proteins would therefore contribute to the mucosal pellicle thus preventing the passage of tastant solution through the pellicle to activate TRCs, diminishing perceived intensity 208. A relationship between salivary cystatins and bitterness acceptance of 6 month old children to urea was subsequently discovered 209. Differences in salivary cystatin have also been found between supertasters and non-tasters of PROP, although the difference was significant only after PROP stimulation. In line with other findings for bitter taste perception, cystatin abundance was higher in non-tasters of PROP 210. Amongst children, obesity has been reported as a cofounder of the relationship between salivary s-type cystatins and bitter sensitivity to caffeine. Obese, caffeine-insensitive children had higher levels of salivary s-type cystatins than obese, caffeine-sensitive children, however amongst normal weight participants’ the association was the inverse211.

Beyond bitter tastants, proteomic study has also revealed salivary cystatins are present in higher abundance in insensitive perceivers of NaCl. This was theorised that the cystatins may inhibit proteases that cleave a subunit of sodium channels in TRCs which reduces the entry of sodium into TRCs and diminishes depolarisation and subsequent sodium perception 212. An association between salivary cystatin and sweet sensitivity has also recently been reported however this relationship was the opposite (sensitive tasters having greater salivary cystatin abundance) of what much of the literature on bitter stimuli reports 213.

These collective findings suggest salivary proteases and protease inhibitors may influence perception of tastants in a taste specific manner. There is also the possibility that multiple interactions occur simultaneously such as receptor alterations combined with pellicle integrity modification.

Salivary composition and intra-oral enzymatic activity Enzymatic activity of salivary proteins is a further mechanism that could underly an individual’s sensitivity to different tastants. Salivary enzymes studied in this regard include amylase and lipase. Salivary amylase has been found to correlate with sucrose sensitivity in

52 children 214, and salivary amylase activity was significantly higher amongst sucrose sensitive adult males, although no difference was found for females 213. Additionally, amongst real foods, salivary amylase activity has been linked with structural breakdown of starch thickened foods and consequent reduced mixing efficiency, sodium release and salt taste perception 215. Salivary lipolytic activity has also been investigated in relation to the basal fatty acid content of saliva, something which has been speculated to have a role in determining individual fat perception sensitivity. A number of studies have suggested a relationship between individual intra-oral lipolytic activity and sensitivity to fat. Fat sensitive subjects have been demonstrated to produce higher concentrations of free oleic acid following an oral triglyceride challenge compared to fat-insensitive individuals. The same study also reported that a number of different lipases were expressed in human lingual tissue 216. Furthermore, lipolytic activity positively correlates with salivary free fatty acid concentration, however lipocalin concentrations did not correlate with total lipolytic activity, suggesting additional lipolytic enzymes are involved in shaping basal salivary fatty acid composition 217.

Trophic effects of salivary constituents on taste and oral tissues A further mechanism by which salivary components may relate to taste sensitivity is by trophic effects on taste receptor cells. An example of this process is the role of epidermal growth factor (EGF), a salivary protein, in maintaining the taste buds of rats 218. The authors reported reduction in taste bud quality and quantity following gland resection in rats, although the effect was reversed upon EGF supplementation. Such an effect has not been investigated in humans, although it is known that salivary EGF does increase following various types of tissue insult and injury 219. A further example of a salivary protein implicated in taste bud health is carbonic anhydrase VI (CAVI, formerly termed “gustin”). CAVI was observed to be reduced in individuals following a viral illness where taste and smell were impaired, and biopsies showed degradation of taste receptors 220. As a cofactor of CAVI, zinc has also received attention in relation to taste perception. Zinc supplementation was found to be effective in correcting taste deficiency only in a subset of CAVI deficient patients 221. There is currently insufficient evidence to support the use of zinc as a general enhancement of taste perception in disease or health 222.

Salivary metabolites and taste and perception There are very few studies that have analysed the salivary metabolome with respect to sensory perception. The first published study of salivary metabolomic composition and sensory perception was in relation to oleic acid. It was found that several metabolites differed between sensitive and insensitive perceivers of oleic acid. These were the short chain fatty

53 acids acetate and butyrate, which were overexpressed in insensitive oleic acid perceivers, and spectral regions corresponding to fatty acids, pyruvate, proline and lysine were overexpressed in sensitive oleic acid perceivers 91. A recent study looked at the dynamic nature of salivary metabolomic changes following citric acid stimulation in relation to sodium perception. Most of the metabolites profiled reduced in concentration post-stimulation (likely due to the increased flow rate) and were not associated with sensory perception. Citric acid stimulation increased endogenous salivary sodium secretion which was associated with reduced salt detection 187.

1.6 Summary

Saliva and taste perception are linked bi-directionally. It is well established that gustatory stimulation alters salivary flow and composition, although the extent of modulation is stimulus-specific. Similarly, it is known that salivary composition can influence taste perception, although this latter association is less understood and less predictable than the former. Individual variation in taste perception is increasingly being studied as a driver of nutrient intake and subsequent diet-related health outcomes. Recent advances in knowledge of the metabolomic composition of saliva offer new avenues of investigation that could link salivary composition to taste and other oral sensations. Investigating the complex interactions involved during taste perception is more important than ever given the current population level trends for excessive consumption of high calorie foods to the detriment of health. Investigation of taste perception with respect to salivary metabolite composition with emphasis on host-microbiome interactions could form the basis for modulating taste perception to improve nutrition.

1.7 Aims and Objectives

Metabonomics can be defined as “the quantitative measurement of the dynamic multiparametric metabolic response of living systems to pathophysiological stimuli…” 223. In the context of the current understanding of the human salivary metabolome, the field is well placed to begin moving from profiling and measuring metabolites in saliva to considering the role(s) of individual metabolites in health and disease. Indeed, there are numerous specific questions about the salivary metabolome that merit answering. As taste perception is one of the major physiological modifiers of salivary production, study of taste stimuli on salivary metabolic profiles represents a key area of research. Taste receptors are continually bathed in saliva and as such are exposed to the metabolic content of saliva. The potential for metabolic differences in saliva to account for inter-individual taste differences is therefore also a critical

54 question in the study of salivary metabonomics. Explaining why taste differences exist between individuals can help develop understanding of why different dietary choices are made, resulting in individual experience of diet-mediated health outcomes. Furthermore, compared to other fluids commonly used in metabolomic profiling such as plasma, serum or urine, saliva has a rich microbial content. Recent study of the interactions between gut microbiomes and their respective hosts has shown that, via metabolite generation, there are multiple links between the health of the host and the microbial communities they harbour. Analogous study of the salivary and oral metabolomes and microbiomes represents a further important area of study.

The primary aim of this thesis is to investigate the relationship between salivary composition and the oral perception of basic tastes (sweet and bitter) and TRP mediated oral sensations (warming and cooling). Specific objectives relating to this aim are as follows:

1. Determine the sources of the predominant metabolites in saliva With emphasis on determining the metabolites produced by oral bacteria and those of host origin, either from salivary glands or other sources. Further objectives were to investigate salivary proteins as a substrates for microbial metabolite production and whether biofilms adjacent to taste papillae might produce different metabolite profiles from salivary proteins. This is explored in chapter four.

2. Investigate taste perception as a physiological modifier of salivary metabolite composition Including the effects of basic tastes (sweet and bitter) and TRP mediated mouthfeel sensations (warming and cooling) on salivary metabolite content, protein content, and extensional rheology. Investigate whether metabolic changes may relate to protein and rheological changes. This is explored in chapter five.

3. Investigate salivary metabolite composition in relation to threshold determination of taste sensitivity Compare differences in salivary metabolite and protein concentrations between sensitive and insensitive tasters measured using taste threshold assessments. Assess inter- and intra- individual stability of salivary metabolite and protein profiles in association with taste perception. Compare the salivary composition of trained and untrained tasters. This is the topic of chapter six.

55

4. Investigate salivary metabolite composition as a modifier of suprathreshold taste perception Controlling for genetic and anatomical factors by studying a twin population, analyse differences in the oral environment (salivary flow rate, protein and metabolite content) of sensitive and insensitive tasters of sucrose, aspartame, caffeine, oleic acid and black tea. Examine genetic contributions to salivary metabolic and protein composition. Investigate the dynamic nature of the intra-oral catabolism of sucrose in relation to sweet sensitivity. This is the topic of chapter seven.

56

Chapter 2: Materials and Methods

This chapter is intended to serve as a concise, yet thorough, description of the experimental protocols referred to throughout this thesis. Validation of the application of these protocols is described in Chapter 3. Any modifications to the protocols detailed here specific to individual experiments are described within the relevant chapter methodology section.

2.1 Ethical approval and recruitment

Ethical approval for conducting experiments on human volunteers was granted by King’s College Research Ethics Committee (HR-15/15–2508). Participants expressing interest were screened for eligibility and provided with written and verbal information about the study. Written consent was obtained from all participants.

The study was open to all individuals over 18 and under 70 years old. Exclusion criteria included disturbances of salivary function or taste perception, active oral disease, antibiotic use within the previous three months and any sensitivity or aversion to the administered tastants.

2.2 Sample collection, processing and storage

In the hour preceding sampling, participants were required to abstain from any activity that could introduce exogenous material into saliva. Activities included eating, drinking (except water), chewing gum, brushing teeth, smoking, and use of lipstick. Samples were immediately placed on ice to minimise degradation after collection, unless a specific experiment required otherwise.

2.2.1 Whole mouth saliva Unstimulated WMS was collected by asking participants to clear their mouths. A timed collection was conducted during which participants expectorated into pre-weighed sterilised universal tubes.

Stimulated WMS was collected as for unstimulated saliva, however a 10 ml tastant solution or control was first passively held in the mouth for 30 seconds. The tastant solution was expectorated prior to saliva collection.

57

WMS samples were then centrifuged (15,000 g for 10 min. at 4 °C) and supernatant was aliquoted and stored at -80 °C. If needed, (e.g. for microbial culture or cell composition analysis) an aliquot of WMS was taken for these purposes prior to centrifugation.

2.2.2 Parotid saliva Parotid saliva was collected using sterilised Lashley cups. The stimulus for parotid salivary flow was 2 % food grade tartaric acid (Sigma, Gillingham, UK), administered to the anterior of the tongue in 0.25 ml increments. Parotid saliva was treated and stored as for WMS. Although sterile at the point of secretion, there is transient contact of parotid saliva with mucosal tissues adjacent to the parotid papilla. Where sterility was imperative, parotid saliva was filtered through a 0.2 µm filter.

2.2.3 Submandibular/sublingual saliva Residual WMS was cleared from the mouth by requesting participants swallow. Sterile cotton wool rolls were placed over the parotid ducts and participants were instructed to hold their tongue in a retracted position, tipping their heads slightly forward. Upon secretion, submandibular and sublingual saliva was collected from the floor of the mouth into sterile, pre-weighed universal tubes using a sterile, single-use pipette.

As for parotid saliva, for purposes of consistency, submandibular and sublingual gland saliva samples were subject to centrifugation prior to storage.

2.2.4 Plasma Blood was collected into heparinised capillary tubes (Sigma) by lancing the finger with a sterile single-use lancet following disinfection with isopropanol. Blood capillaries were sealed with capillary clay (Sigma) and centrifuged (1,600 g for 15 min. at 4 °C). The plasma layer was collected with a pipette and stored at -80 °C. The cellular layer was disposed of appropriately.

2.2.5 Tongue biofilm Residual salivary film was removed from the tongue with sterilised gauze. Tongue biofilm (plaque) was collected by gently scraping the tongue surface using single use plastic scrapers. Biofilm was collected from the anterior of the tongue (fungiform papillae region) and adjacent to circumvallate papillae posteriorly. Biofilm was weighed (10 mg) and utilised immediately.

58

2.2.6 Buccal epithelial cells Buccal surfaces were patted dry with sterilised gauze and a plastic scraper was then used to scrape some buccal epithelial cells from the cheek. Buccal epithelial cells were then resuspended in sterile PBS and adjusted to the desired cell density based on cell density counts using trypan blue.

2.2.7 Gingival-crevicular fluid Sterile cotton wool rolls were used to isolate the anterior maxillary teeth and any residual saliva was removed. Filter paper strips (5mm width) were then placed into the gingival margins and GCF was collected. Using the visual scale below (Figure 2.1), this process was repeated until approximately 2 µl of GCF was obtained.

Figure 2.1: A depiction of the scale used to assess GCF volume collected on filter strips. The strips shown have been exposed to 0.25, 0.5, 1 and 2 µl of blue dye, used as a visual aid. Each strip is 5 mm wide. Collection strips were then placed in 0.5 ml Eppendorfs with 50 µl PBS and a hole was made in the bottom of the tube with a sterile syringe needle. The 0.5 ml Eppendorf was placed in a 1.5 ml Eppendorf and GCF was eluted into the PBS by centrifuging at 15,000 g for ten minutes.

2.3 Salivary analyses

2.3.1 Flow rate Salivary flow rate was calculated in g/min by subtracting the initial mass of the collection tube from the final mass of the tube plus saliva and dividing by the number of minutes over which collection occurred.

2.3.2 Cellular and microbial composition Eukaryotic cells were analysed by mixing one-part saliva and one-part 0.4% trypan blue (Sigma) and viewing the mixture in a modified Fuchs-Rosenthal haemocytometer chamber under a light microscope at 500x magnification. Eukaryotic cells (epithelial or leukocytic) were

59 counted within the four corner haemocytometer chambers (total volume 0.8 µl) then multiplied by 2500 to calculate the cell density per ml.

Microbial composition was assessed by calculating colony forming units (CFU) per ml saliva. Growth media was prepared by mixing 45.6 g/l fastidious anaerobe agar with distilled water (LabM, Lancashire, UK), autoclaving at 121 °C, 2.25 bar pressure for 15 minutes, cooling to 50 °C in a water bath then adding 5% by volume defibrinated horse blood (Fisher Scientific, Loughborough, UK) and pouring into sterile petri dishes in a laminar flow cabinet. Saliva samples were diluted ten-fold from 1:1000 to 1:100,000 and 25 µl was plated onto petri dishes with sterile plastic spreaders. Petri dishes were incubated in an anaerobic cabinet (80%

N2, 10% CO2, 10% H2) for 48 hours then colonies were counted, and CFU/ml was calculated.

2.3.3 Protein composition

Total protein Samples were analysed for total protein using a Pierce bicinchoninic acid (BCA) assay kit (ThermoScientific, Rockford, IL, USA). The manufacturers protocol was followed, diluting samples 1:10 with purified water. The provided bovine serum albumin was used to generate a standard curve, read at 545 nm.

Protein band densitometry Salivary proteins were separated by 1-dimensional sodium dodecyl sulphate polyacrylamide gel electrophoresis (1D-SDS-PAGE). NuPAGE LDS sample buffer, NuPAGE 4–12% Bis-Tris Gel pre-cast polyacrylamide gels and NuPAGE MES SDS running buffer were all purchased, (ThermoFisher Scientific, Carlsbad, CA, USA).

Supernatant (13 µl) was mixed with 5 µl sample buffer and 2 µl 0.5 M dithiothreitol (DTT, Sigma). Samples were vortexed, heated at 100 °C for 3 mins then centrifuged at 4600 g for 3 mins. Buffered sample was loaded 10 µl per lane into precast gels and electrophoresed at 200 V, 250 mA, for 32 mins in a 5% running buffer solution.

To compare samples from different gels, every gel contained one lane with a sample of pre- aliquoted reference saliva prepared in the same way. Following electrophoresis, gels were stained with Coomassie brilliant blue (Sigma) diluted 2:3 with acetic acid for 30 minutes. Gels were destained with 10% acetic acid and imaged using a ChemiDocTM MP imaging system and analysed in ImageLab 6.0.1 (Bio-Rad Laboratories, Hercules, CA, USA). Lanes and bands were

60 automatically detected, corrected manually and band intensity peaks were referenced to the cystatin band on the reference saliva lane. A uniform background filter (equivalent to the signal of an empty lane) was applied to every lane. Gels were then stained for glycoproteins with periodic acid Schiff’s stain (PAS). Gels were stained in 2% periodic acid (Sigma) for 20 minutes, washed three times in purified water for 5 minutes then stained with Schiff’s reagent (VWR, Lutterworth, UK) for thirty minutes in a light-proof tray. Gels were destained in purified water, imaged and analysed as for coomassie stained gels. An example of the major proteins measured by band densitometry in saliva is shown below in Figure 2.2.

Figure 2.2: Photograph of a lane from a coomassie and PAS stained polyacrylamide gel showing the separated major proteins in a typical saliva sample. The major proteins measured by band densitometry are labelled. glPRP = glycated proline-rich protein, PRP1 and PRP2 refer to separate, unnamed proline-rich protein bands of different molecular weights. Band annotations have been adapted from Carpenter, (2013)20.

Digitisation and multivariate analysis of lane profiles from scanned polyacrylamide gels Scanned gels containing a standard lane were viewed in Image Lab 6.0.1. A standard level of background subtraction was applied to all lanes and gels in the dataset. The lanes were automatically detected then manually corrected to ensure that: a) the full length of all lanes is visualised, b) the full width of each lane is visualised without any encroachment from samples in adjacent lanes, c) the maximum height of each lane is visible and d) the cursor is positioned along the relative front at a standard point (typically the maximum cystatin peak of the

61 standard sample was chosen). An image of each lane was captured to export for digitisation. An example lane is illustrated in Figure 2.3 a., below.

The image of each scanned lane was imported into WebPlotDigitiser version 4.1 (https://automeris.io/WebPlotDigitizer/, San Francisco, USA). The lane profile axes were assigned in WebPlotDigitiser using the origin as points X1 and Y1. Point X2 was assigned as the intercept of the pre-positioned cursor with the x axis, and Y2 was the maximum intensity on the image y axis (60,000 units). This process is illustrated in Figure 2.3 b. i.

The lane profile was then manually traced, and the foreground colour selected and filtered, Figure 2.3 b. ii. The “X Step w/ Interpolation” algorithm was assigned to extract the data points. X and Y min were both set as 0, Y max was 1 and X max was set at 1.5. The ΔX Step was set at 0.002, giving around 650 data points per lane. Figure 2.3 d. illustrates this process and a zoomed view of the data points at the amylase peak are shown to demonstrate the digitisation. The data was then exported from WebPlotDigitiser to Excel 2013 and lane profiles were re-created (Figure 2.3 e.).

Digitised lane profiles were then analysed using principal component analysis (PCA) algorithms and k-means cluster analysis in Knime (Konstanz, Germany).

62

Figure 2.3: Illustration of the process of digitising protein samples lanes from polyacrylamide gel scans. a. – careful capturing of an image of the scanned lane from ImageLab. b i. – setting the axes of the lane image in WebPlotDigitiser. X1, X2, Y1 and Y2 axes alignment points are inset. b ii. – manual tracing of the lane profile and background extraction. c. – digitising algorithm settings, showing zoomed in amylase peak illustrating the lane conversion into data points. d. – the exported data re-visualised in excel, where the data can be subject to more in- depth multivariate analysis. 63

2.3.4 Metabolite composition - 1H-NMR spectroscopy

Preparation All reagents and consumables were purchased from Sigma. NMR buffer was prepared with 0.5

2 mM sodium trimethylsilyl-[2,2,3,3- H4]-propionate (TSP), 0.2 M Na2HPO4, 44 mM NaH2PO4 and

50 % by volume deuterium oxide (D2O), pH 7.4. Samples were mixed with buffer at a ratio of 4:1 sample: NMR buffer, giving a final solution concentration of 0.1 mM TSP and 10% by volume D2O.

Final sample volumes were 550 µl (440 µl sample 110 µl buffer) for analysis in 5 mm ED tubes or 300 µl (240 µl sample 60 µl buffer) for 3 mm ED tubes. In cases where these ratios were altered (i.e. low sample volume) protocols are described in the relevant chapter methods section.

Acquisition 1H-NMR spectra were acquired on a Bruker Avance III spectrometer (Bruker Biospin, Karlsruhe, Germany) operating at a proton frequency 600.2 MHz. Samples were maintained at 4 °C prior to analysis. Samples were analysed at 25 °C following a 5-minute period for temperature equilibration. One dimensional 1H-NMR spectra were acquired with a Carr−Purcell−Meiboom−Gill (CPMG) spin−echo pulse sequence with water presaturation to filter out broad macromolecule resonances, a total echo time of 64 ms, relaxation delay of 4 seconds, acquisition time of 2.32 seconds, and 256 transients collected with 64,000 data points following four dummy scans, with a spectral width of 20 ppm (−5 to 15 ppm). In some experiments a representative sample was analysed by two dimensional 1H-1H correlation spectroscopy (COSY) to aid in peak assignment. Spectra were acquired with 4096 data points, 400 increments, with 8 scans per increment, a relaxation delay of 2 seconds and spectral width 11,160 Hz (15.9 ppm).

Spectral Processing and data analyses Spectra were analysed in TopSpin 4.0.6 (Bruker BioSpin). A 0.3 Hz exponential line broadening function was applied before Fourier transformation and automatic phase correction. Baselines were inspected and polynomial baseline correction applied. Where required, manual phase and baseline adjustments were made. Metabolite assignments were made using Chenomx NMR suite 8.3 (Chenomx Inc., Edmonton, CAN), the human metabolite database (http://www.hmdb.ca) and literature values.

64

Metabolite peaks were manually integrated and quantified relative to the TSP peak in each spectrum. For multivariate analyses, spectra were integrated using MestReC V (Mestrelab, Santiago de Compostela, ES). Spectra were integrated in 0.01 ppm buckets between 0.7 and 8.5 ppm, excluding the water peak 4.5 between 5.5 ppm. Deviations from these parameters are noted in the relevant sections. When running samples of the same fluid type, the TSP peak (0.02 to -0.02 ppm) was integrated to scale spectra. When comparing more heterogenous fluids, the spectral buckets were normalised to the total spectral area.

Spectral buckets were analysed in Knime by PCA and k-means cluster analysis. To calculate relative proximity of projected data points (i.e. the relative similarity of two or more spectral profiles) Euclidean distance in three dimensions, weighted for the relative PCA weighting of each dimension, was calculated.

Individual salivary metabolite assays Quantification of several individual salivary metabolites was performed using specific benchtop assays.

Urea A urea assay kit was acquired (ThermoFisher) and used in accordance with the manufacturer’s instructions. The provided urea standard was used to construct a standard curve and samples were prediluted 1:5 for saliva and 1:10 for plasma samples.

Citrate A citrate assay kit was purchased (Sigma). The assay was conducted as per the manufacturer’s instructions. Samples were heated at 110 °C in sealed Eppendorfs for 10 minutes to denature residual enzymes that could cause potential interference and an additional background well was analysed for each sample. The standard curve was prepared with citrate standard included in the kit. The assay was read in fluorescence mode although it can be read as absorbance or fluorescence.

Taurine A taurine assay was obtained (Abcam, Cambridge, UK). Eluted GCF samples were prepared in accordance with the manufacturer’s instructions. Taurine standard included with the kit was used to prepare a standard curve. Due to the dilution effect of elution process, (2 µl GCF in 50 µl purified water) the protein content was sufficiently reduced and the maximum volume of sample (25 µl) was added per well.

65

2.3.5 Extensional Rheology Analysis of salivary shear rheology can be complicated by factors such as required sample volume, experimental time and surface deposition of salivary protein all of which may cause artefacts in the resulting data 224-226. Extensional rheology of saliva, whilst less common in literature, offers several advantages such as low sample volume requirement and rapid data collection 226. A common and simple rheological test is the measurement of the spinnbarkeit (thread forming) property of saliva 190,227,228, however this test yields only a single measurement (maximum thread length) and can be prone to inconsistency. For these reasons salivary extensional rheology measured by capillary breakup was used as the method provides a balance of a reasonable amount of data extrapolated per sample with a rapid experimental time and low sample volume.

Rheological analysis was conducted with a HAAKE CaBER-1 capillary-breakup extensional rheometer (ThermoFisher). Rheological analysis was conducted on unprocessed WMS immediately post-expectoration since treatments such as centrifugation, freezing or even simple delays in analysis can cause rheological properties to naturally dissipate. Samples were loaded between 6 mm diameter plates set at an initial gap of 2 mm. To avoid introducing bubbles when loading, 60 µl of sample was pipetted to ensure a slight excess remained in the pipette tip. Capillary formation was set to a sample end height of 10.8 mm with a stretch time of 50 ms at a constant rate of 0.176 m/s. Three measurements were made per sample and averaged. Total capillary breakup time and extensional viscosity over a range of strains (9, 9.5, 10 and 10.5) were analysed.

As the CaBER requires a fresh sample for each reading, the plates were cleaned between readings with distilled water. Between participants, the rheometer plates were cleaned with followed by distilled water and then air dried. A diagram illustrating the principles of the capillary breakup measurement is shown in Figure 2.4. Initially, a sample loaded between two plates is drawn apart from a starting length, L0, to a final length, Lf, with an exponential stretch profile at a constant extension rate (ε˙0), modelled by the equation: L = L0exp(ε˙0t). Following this, the capillary diameter of the fluid between the two plates is measured at the midpoint and the extensional flow is used to measure apparent extensional viscosity. In depth explanation of capillary breakup extensional rheology is described by Miller et al., (2009) 229, and Bhardwaj et al., (2007) 230.

66

Figure 2.4: Illustrates salivary capillary breakup measurement using a CaBER. a. A sample loaded between two plates (left) is formed into a capillary when the plates are drawn apart (right). b. At the midpoint of the capillary in the horizontal plane, incident light from a laser is used to measure the diameter of the capillary as it breaks up. The image is not to scale. Readings of capillary diameter are made every 10 ms during capillary breakup.

2.4 Sensory analyses

2.4.1 Tastant preparation

All tastants and solvents intended for administration to participants were purchased as food grade or pharmaceutical grade. For consistency, the aqueous solvent used throughout was Buxton mineral water (Nestlé, Gatwick, UK). Tastant storage and preparation was conducted in a dedicated laboratory space with single-use equipment to prevent cross-contamination. Relevant details regarding tastants specific to individual experiments are provided. Any 67 tastant concentrations described in percentage or parts per million (ppm) refer to dilution by volume, not mass.

2.4.2 Tastant administration and intensity rating

As for sample collection, tastant ratings were collected from participants at least one hour after oral exposure to exogenous substances. An example of the suprathreshold intensity scale is shown below in Figure 2.5. Participants were blinded to the concentration of tastants although the sensation being assessed was included in the scale. Participants were provided verbal and written instruction on the use of the scale and familiarised with the rating process prior to commencing tasting.

Tastants were administered as 10 ml solutions and participants were asked to hold the liquid passively in the floor of the mouth, then expectorate after 30 seconds. Participants were asked to rate the maximum intensity perceived during this period using a single vertical line across the scale.

Figure 2.5: An example of the generalised labelled visual analogue scale (glVAS) used for suprathreshold intensity ratings. This example describes sweet taste assessment. Other sensory modalities would be modified with appropriate descriptors. Note that the concentration field is left blank until the scale is completed so participants were blinded to concentrations. 2.4.3 Fungiform papillae density measurement

Brilliant Blue FCF food dye (PME, Enfield, UK) was used to stain the dorsal surface of the tip of the tongue. A sterilised 6 mm diameter filter disc was placed on the tongue for calibration purposes and a digital photograph was taken and analysed using ImageJ freeware (NIH, USA). As FPD varies across the tongue surface (reducing when moving posteriorly) a standardised tongue region was measured in all participants. A central point 3 mm lateral to the medial lingual sulcus and 3 mm posterior to the non-keratinised mucosal border of the tongue was

68 identified. A 6 mm diameter circle centred around this point was drawn in ImageJ. The precise diameter of the circular area was measured, calibrated against the 6mm disc included in the image. Papillae within this area were counted and density (papillae/cm2) was calculated. For each image this process was conducted by a single operator and repeated on three non- consecutive days without reference to previous counts to establish a mean FPD. This process is illustrated in Figure 2.6.

Figure 2.6: Illustrates the process of FPD measurement. a. shows the selection of a standardised anatomical reference point on the participants tongue. b. shows the measurement of a circular area of 3mm radius from this point, and the measurement of this area calibrated against the 6mm filter paper disc included in the photograph. In this example the measured area was 27.7 mm2, close to the area of a 6mm diameter circle which is 28.3 mm2. The red quadrant and black dots have been superimposed on the image to aid counting of papillae. Adapted from: (Gardner & Carpenter, 2019 231).

2.5 Statistical analyses

Data collected were subject to a range of different statistical analyses. The specific techniques used, and the comparisons made for each experiment are detailed in the relevant chapters. The following packages were used for statistical analyses: GraphPad Prism 7 (La Jolla, CA, USA), SPSS 24 (IBM, Armonk, NY, USA), Microsoft Excel 2013 (Redmond, WA, USA), RStudio (Boston, MA, USA) and Knime analytical platform (Konstanz, Germany).

69

Chapter 3: Methodological validation

3.1 Introduction

Many of the techniques and their applications used throughout this thesis had not been validated in literature. It was deemed pertinent to conduct experiments ensuring the planned data collection and analytical methodologies were valid and justifiable. This chapter incorporates previously published work, as well as additional unpublished data. The publication “Developing and Standardizing a Protocol for Quantitative Proton Nuclear Magnetic Resonance (1H NMR) Spectroscopy of Saliva”, published in the Journal of Proteome Research, 2018, is incorporated in its unedited form, including the supplementary material 232. This work essentially validates a protocol for collection, preparation and analysis of saliva by 1H-NMR spectroscopy, and the work is contextualised in the manuscript introduction. Author contributions were as follows: Alexander Gardner – study design, sample collection, data analysis and manuscript writing; Harold Parkes – data analysis; Po-Wah So – study design, data analysis; Guy Carpenter – study design, data analysis. All authors edited and approved the final version of the manuscript. Results of a further relevant but unpublished experiment on the centrifugation and freezing effects on salivary bacterial viability are included as additional supporting work for this manuscript.

Additional work that was not previously published on the same pre-treatment conditions (centrifugation, freezing, and storage) prior to SDS-PAGE band densitometry is included. While previous study of pre-treatment conditions on salivary SDS-PAGE analysis has been conducted 224,233, the specific conditions and analytes relevant to this thesis were not featured thus it was necessary to address this. The linearity of SDS-PAGE band densitometry was also assessed to ensure measurements obtained by this technique in future experiments were reliable. Work validating the use of multivariate statistics applied to digitised SDS-PAGE lane profiles, as described in the materials and methods section, in included in this chapter to confirm the utility of this method.

Validation of the sensory techniques used in this thesis, including the suprathreshold intensity scale (glVAS) and tastant preparations is also included in this chapter. Data on these topics from the publication “Anatomical stability of human fungiform papillae and relationship with oral perception measured by salivary response and intensity rating”, published in Scientific Reports, 2019, is incorporated into this chapter 231. The manuscript is not incorporated in full, however relevant data analyses are included which would not have realistically fit into the

70 original publication due to space limitations. Alexander Gardner and Guy Carpenter contributed to study design, data collection and analysis and approved the manuscript. A further experiment relating to the sensory and sample collection protocol was conducted to assess oral clearance and restoration of metabolite concentrations to baseline after perturbations following tastants such as sucrose. Intra-oral metabolism of sucrose raised levels of metabolites such as lactate, pyruvate and succinate. Therefore, for experiments involving multiple tastant stimulated sample collections, particularly where sucrose is administered, it is important to be sure that metabolite levels return to baseline before commencing successive sample collection.

Finally, preliminary data about the expected effects of age and dental status were gathered to help target the recruitment of subsequent studies and identify caveats relating to participant demographics.

3.2 Aims and Objectives

The overall aim of the work conducted in this chapter was to ensure that the sample collection and analyses as well as the sensory data collection would yield valid data and these techniques could be applied in subsequent experiments. Specific objectives were: 1. Investigate sample storage and preparation protocols for saliva samples prior to analysis by 1H-NMR spectroscopy and SDS-PAGE. 2. Determine the validity of using digitised SDS-PAGE lane profiles for multivariate analysis. 3. Determine the optimal specific tastant concentrations and intensity scaling techniques. 4. Determine the optimal approach to resetting the baseline salivary metabolite profile in between different tastant solution/ sample collections. 5. Identify effects of participant age and dental status on salivary metabolite concentrations.

71

3.3 Materials and methods

3.3.1 Assessment of centrifugation and freezing on salivary bacterial viability Unstimulated WMS was collected from ten healthy volunteers. Samples were aliquoted for preparation as follows: 1. no treatment, 2. centrifugation at 15,000 g for ten minutes at 4 °C 3. freezing at -80 °C for one hour, thawing, then centrifuging as for sample 2. 4. centrifuging as for sample 2., freezing at -80 °C for one hour then thawing Samples not undergoing freezing were chilled on ice during the preparation of the other samples. Aliquots were cultured and CFU density assessed as described. CFU densities were compared by repeated measured ANOVA with Bonferroni post-hoc test.

3.3.2 Assessment of multivariate analyses of digitised SDS-PAGE salivary protein profile The following experiments were conducted in order to validate the reproducibility of the lane digitising process.

Initially, data acquired from digitising lanes of the same sample run on the same gel (intra-gel samples) versus different gels (inter-gel samples) were compared. The same sample was run six times on a single gel and then run a further six times each on separate gels. Lanes were scanned and their profiles digitised as described in section 2.3 and the resulting data visualised in excel and as a principal component analysis plot in Knime.

A second experiment was designed to further explore the ability of the lane digitising protocol to identify samples by multivariate analysis as well as to investigate ways of reducing inter-gel differences as a source of variability. Saliva from four participants was collected and three lanes of each sample were run on four different gels. Methods to improve any inter-gel effects were then conducted. Firstly, use of ComBat batch correction software prior to PCA was performed. A second approach was to treat one of the participant lanes on each gel as a standard. These lanes were then aligned between gels and used to normalise the projected data for the other participants inter-gel samples.

3.3.3 Assessment of suprathreshold intensity scale and tastant concentrations Twelve participants were presented with a range of tastant solutions (10 ml) and asked to rate their perceived intensity for each tastant on the glVAS scale. These are summarised in Table 72

3.1. Participants had been given written and verbal instruction on the use of the scale and they were blinded to the concentrations of each tastant. In the first instance, tastant concentrations were selected based on previous literature concentrations for suprathreshold assessment and preliminary assessments. For sucrose and caffeine there are abundant literature reports of different concentrations being administered. Strong concentrations could be considered around 400 mM and 4 mM for sucrose and caffeine respectively 156, although other studies have administered 1.17 M sucrose and 50 mM caffeine 228,234. For capsaicin, concentrations of 1 to 2 ppm are typical and for menthol there are few studies describing an oral administration although 500 ppm has been reported 190,235. Initial tests found 500 ppm menthol was excessively strong. These values were used to decide the range of concentrations in a preliminary study as described in Table 3.1, the results of which were in turn used to help decide definitive tastant concentrations used throughout this thesis.

Table 3.1: Summary of tastant and solvent concentrations investigated at the initial visit. All concentrations in ppm and percentage are by volume. From: (Gardner & Carpenter, 2019 231). Tastant Taste/oral Concentrations Solvent sensation investigated Sucrose Sweet 0.25 M, 0.5 M, 1 M Water Caffeine Bitter 4 mM, 8 mM, 20 mM Water L-Menthol, Cooling 50 ppm, 100 ppm, 250 Water and ethanol (1R,2S,5R sensation ppm (9.5x10-3 %, 0.019 % stereoisomer) and 0.475 % respectively) Capsaicin Warming 0.1 ppm, 1 ppm, 2.5 Water and ethanol sensation ppm (9.5x10-3 %, 0.095 % and 0.24 % respectively)

As a potential biological measurement of taste response, participants fungiform papillae density was recorded as described in section 2.4.

Participant responses were then analysed to confirm that the scale was appropriate for discriminating between the different tastant concentrations, to assess optimal tastant concentration and for association with fungiform papillae density. Tastant concentration responses were assessed based on the following parameters: mean rating centred close to the

73 scale midpoint (i.e. no evidence of ceiling or basement effects), a balanced distribution of responses (i.e. no skew to the data) and ideally similar mean intensity between tastants, suggesting an iso-intense sensation was conferred by the different tastant solutions.

3.3.4 Assessment of protocol to return salivary metabolites to baseline levels Five healthy participants were recruited. Participants were given 10 ml water control to hold in the mouth for 30 seconds, expectorate then saliva was collected into sterilised universal tubes for two minutes. This process was repeated using 10 ml 0.25 M sucrose. Following the post-sucrose saliva collection, three 20 second water rinses were performed then a further saliva sample was collected over two minutes. Finally, the water rinse process and saliva collection were repeated. Salivary flow was measured for each sample. Samples were prepared and analysed by 1H-NMR spectroscopy as described (mixed 440 µl sample and 110 µl NMR buffer in 5 mm ED tubes, analysed by CPMG pulse sequence) and outputs of metabolites that changed post-sucrose were quantified. Data were analysed by Friedman’s test followed by Dunn’s post-hoc test relative to the first post-water control sample.

74

75

3.4 Results

3.4.1 Developing and Standardizing a Protocol for Quantitative Proton Nuclear Magnetic Resonance (1H-NMR) Spectroscopy of Saliva

76

77

78

79

80

81

82

83

84

85

Developing and Standardising a Protocol for Quantitative Proton Nuclear Magnetic Resonance (1H-NMR) Spectroscopy of Saliva

Alexander Gardner†, Harold G. Parkes∥, Guy H Carpenter†, Po-Wah So⊥

† Department of Mucosal and Salivary Biology, Dental Institute, King's College London, London, SE1 9RT, UK

∥ Institute of Cancer Research, London, SW7 3RP, UK

⊥ Department of Neuroimaging, Institute of Psychiatry, Psychology and Neuroscience, King's College London, Maurice Wohl Clinical Neuroscience Institute, 5, Cutcombe Road, London, SE5 9RX, UK Corresponding author: Email: [email protected] Tel: 020 7848 5453 Table of contents Supplementary Figures: Figure S-1: Factors affecting salivary metabolites - Exogenous substances. Figure S-2: Factors affecting salivary metabolites - Intra-oral catabolism of dietary substances. Figure S-3: induced changes. Figure S-4: Comparison of 700 MHz CPMG and NOESY 1D 1H-NMR spectra. Figure S-5: Diagram of a NMR tube with a coaxial insert and the volume ratio calculation. Figure S-6: Confirmation of assignment of acetoin in whole mouth unstimulated saliva. Figure S-7: Investigation of the assignment of propylene glycol in whole mouth unstimulated saliva. Figure S-8: Investigation of whole mouth saliva pre- and post-addition of propylene glycol. Figure S-9: Comparison of different freeze-thaw treatments on the 1H-NMR spectra of saliva. Supplementary Tables: Table S-1: Summary of selected salivary metabolite concentrations collected pre-, during and 2 h post-exercise. Table S-2: Analysis of centrifugation force on metabolite concentrations. Table S-3: Analysis of freeze-thaw considerations on metabolite concentrations. Table S-4: Analysis of quantification method on metabolite concentrations. 86

Supplementary Information Factors affecting salivary metabolites Exogenous substances

Figure S-1: (A) Partial 700 MHz CPMG 1H-NMR spectra (echo time 64 ms, 3.00 – 4.10 ppm) of saliva collected from the same participant (i) before and (ii) twenty minutes after eating. Spectra are of the same vertical scale. Peaks from sucrose, and glucose obscure metabolites such as glycine and taurine and the quartet from lactate (4.12 ppm) in saliva collected twenty minutes after eating. These peaks were not observed in samples collected one hour after eating or drinking. (B) Detection of xylitol from chewing gum in saliva collected one hour after chewing gum. Xylitol peaks did not obscure other assigned metabolites. Samples were centrifuged at 15,000 g prior to freezing. Intra-oral catabolism of dietary substances Certain exogenous substances not only obscure salivary NMR spectra but are readily metabolised in the oral cavity by the complex microbial community, and thus alterations in levels of other salivary metabolites can be observed. This is illustrated in Figure S2, showing the intra-oral catabolism of sucrose. Water (10 ml) was held in the mouth for 30 s before being expectorated. Saliva was then collected over a period of two minutes. This process was repeated after 5 mins with 0.25 M sucrose solution (10 ml). Studies of saliva involving consumption of oral substances (including those administered as sialagogues, e.g. citric acid) therefore need to consider the effects these substances may have on metabolite profile.

87

Figure S-2: Partial 700 MHz CPMG 1H-NMR spectral regions (echo time 64 ms, 1.0 - 2.5 ppm) of saliva, expectorated following (A) a 0.25 M sucrose rinse and (B) a water rinse (B). Elevated acetate, lactate, pyruvate and succinate was observed in the saliva following a sucrose rinse. Samples were centrifuged at 15,000 g prior to freezing. Spectra are of the same vertical scale.

88

Exercise induced changes Changes in salivary metabolite concentration were induced by exercise. Within ten minutes of continuous exercise, expectorated saliva had higher concentrations of all metabolites Levels returned to baseline within two hours post-exercise. The higher levels during exercise may partly be due to dehydration (i.e. less fluid leading to more concentrated metabolites), however metabolite concentrations do not change proportionally, indicating additional factors causing differential generation and consumption of metabolites. Recent exercise therefore presents an additional variable to consider prior to collecting saliva for 1H-NMR spectroscopy.

Figure S-3: Partial 700 MHz CPMG 1H-NMR spectral regions (0.8 - 2.5 ppm) of saliva collected (A) before, (B) during and (C) two hours after exercise. Samples were centrifuged at 15,000 g prior to freezing, with quantification by internal, buffered TSP. Vertical scale is the same for all spectra. The acetate peak has been truncated.

89

Table S-1: Summary of selected salivary metabolite concentrations collected pre-, during and 2 h post-exercise (n=1), illustrating the disproportionate increases in salivary metabolites following exercise. Metabolite Pre-exercise (mM) During exercise (mM) Post-exercise (mM) Acetate 2.08 6.46 2.26 Lactate 0.07 0.37 0.08 Propionate 0.29 0.72 0.29 Succinate 0.13 0.33 0.06 Pyruvate 0.08 0.43 0.09

90

Comparison of saliva with CPMG and NOESY pulse sequences

Figure S-4: Stacked partial spectra of partial 700 MHz 1H-NMR spectral regions (0.1 – 4.1 ppm) of the same saliva sample analysed with a NOESY pulse sequence (top) and a CPMG pulse sequence (echo time 64 ms, bottom). Spectra are of the same vertical scale with the acetate and lactate peaks truncated. Spectra were similar however the CPMG spectra featured a flatter baseline than the NOESY spectrum, without attenuating of the remaining resonances, and so the former was used for quantification. NOESY 1H-NMR spectra were acquired at 700.13 MHz. 32 transients were collected with 64 k data points following four dummy scans, with a spectral width of 20 ppm (-5 to 15 ppm), a relaxation delay of 4 s and a mixing time of 10 ms.

91

Figure S-5: Diagram to show how the volume ratio of a NMR tube and a coaxial insert was calculated. Different standard solutions (A and B) are placed into each tube. Both solutions are of known concentration, chemical shift, and number of protons giving rise to the peak to be integrated. Absolute volumes are not important, provided the volume read by the NMR probehead (rectangular area) is fully covered. Once the spectrum has been acquired and the peak integrals measured the volume ratio can be calculated using the equation: 푉표푙푢푚푒 퐴 퐼푛푡푒푔푟푎푙 퐴 푃푟표푡표푛 푐표푛푐푒푛푡푟푎푡푖표푛 퐵 = ∗ 푉표푙푢푚푒 퐵 푃푟표푡표푛 푐표푛푐푒푛푡푟푎푡푖표푛 퐴 퐼푛푡푒푔푟푎푙 퐵 Where proton concentration is the molar concentration of the solution multiplied by number of protons contributing to the peak that was integrated. Diagram is not to scale.

92

Figure S-6: Confirmed assignment of acetoin in whole mouth unstimulated saliva by 2D NMR. 2D 1H-1H COSY spectra were obtained from saliva samples of two participants, A and B. Spectra were acquired with 4096 data points, 400 increments, with 8 scans per increment, a relaxation delay of 2 s and spectral width 11,160 Hz (15.9 ppm). In both saliva samples, the doublet at 1.37 ppm (arising from the methyl group adjacent to the CH(OH) in acetoin) shows a cross peak at 4.42 ppm, which matches HMDB assignments for acetoin (http://www.hmdb.ca/spectra/nmr_one_d/1939). The quartet at 4.42 ppm from the proton in the CH(OH) group is masked by other resonances in the 1D 1H-NMR spectra of saliva. Samples were centrifuged at 15,000 g prior to freezing, with quantification via external TSP in a coaxial tube.

93

Figure S-7: Investigation of propylene glycol, reported in saliva by Singh et al., 2017. 2D 1H-1H COSY spectra were obtained from the same saliva samples as described for Figure S-3. Cross peaks from the doublet at 1.13 ppm, believed to be propylene glycol, were observed at 3.64 ppm in both participants. This was not in line with the expected literature on propylene glycol (http://www.hmdb.ca/metabolites/HMDB01881) where the methyl group doublet at 1.13 ppm would display a cross peak with the adjacent CH group at 3.87 ppm. Additional investigation by spiking saliva with propylene glycol is described below (Figure S-8), to determine the presence of propylene glycol in saliva by 2D NMR methods.

94

Figure S-8: Partial 700 MHz 1H-NMR CPMG spectra of saliva displaying an unassigned doublet signal at 1.13 ppm, believed to be propylene glycol before (A) and after spiking in 0.1 mM propylene glycol, producing a doublet at the same frequency (B). The corresponding 2D 1H-1H COSY spectra shows a faint cross peak at 1.13 and 3.64 ppm, thought to be propylene glycol (C) but on addition of propylene glycol, another cross peak is evident at 1.13 and 3.87 ppm (D), suggesting the 1.13 ppm resonance in saliva does not arise from propylene glycol as previously assigned from the 1D-1H NMR spectra.. Samples were centrifuged at 15,000 g, and analysed fresh. No standard was used as spectra were calibrated using the acetate peak. Spectra A and B are at the same vertical scale.

95

Comparison of different freeze-thaw treatments on the 1H-NMR spectra of saliva

Figure S-9: Stacked partial spectra of partial 700 MHz CPMG 1H-NMR spectral regions (0.6 - 3.6 ppm) of saliva sample collected at the same time from a representative individual. Spectra are of: (i) – centrifuged at 15,000 g with no freezing; (ii) –frozen and thawed after centrifuging at 15,000 g; (iii) – frozen and thawed before centrifuging at 15,000 g; (iv) – centrifuged at 15,000 g then frozen and thawed four times. Quantification was via external TSP in a coaxial tube for all aliquots. The same degree of similarity was observed in the other participants. Spectra are of same vertical scale. The acetate peak has been truncated.

96

Table S-2: Analysis of centrifugation force on metabolite concentrations.

Metabolite Metabolite concentration (mean ± SEM, µM) of Repeated Bonferroni post-hoc Bonferroni post-hoc test of sample centrifuged at: measures ANOVA test normalised data (p value) (groups compared; (groups compared; *= < 0.05 p value) p value)

0 g 750 g 1500 g 3000 g 15,000 g Acetate 3603.0 3801.0 3708.0 3502.0 3480.0 ± 0.26 N.S. N.S. (0000042) ± 318.8 ± 388.5 ± 348.0 ± 250.0 279.9 Acetoin 58.7 ± 43.9 ± 42.2 ± 39.3 ± 39.2 ± 0.03* N.S. N.S. (0003243) 9.2 4.4 4.0 4.7 4.8 Alanine 90.2 ± 84.3 ± 81.4 ± 79.7 ± 79.1 ± 0.039* N.S. N.S. (0000161) 8.8 8.4 8.1 8.9 9.1 Butyrate 215.2 ± 182.9 ± 169.2 ± 156.3 ± 152.4 ± 0.14 N.S N.S. (0000039) 53.2 34.9 29.8 22.0 20.3 Choline and 11.1 ± 10.2 ± 10.0 ± 9.5 ± 9.8 ± 0.013* N.S. 0g vs. 1500g; 0.019 choline- 2.2 1.9 2.0 2.1 2.0 containing compounds (0000097) Citrate 35.1 ± 35.4 ± 36.5 ± 32.0 ± 35.0 ± 0.34 N.S. N.S. (0000094) 4.9 4.3 4.6 4.4 4.9 Dimethylamine 6.8 ± 7.1 ± 6.8 ± 6.3 ± 6.6 ± 0.16 N.S. N.S. (0000087) 0.8 1.1 0.9 0.9 1.1 Ethanol 74.2 ± 75.6 ± 70.7 ± 69.3 ± 67.5 ± 0.079 N.S. N.S. (0000108) 13.1 13.1 11.7 12.6 11.6 Formate 88.6 ± 111.0 ± 104.4 ± 90.0 ± 91.1 ± 0.25 N.S. N.S. (0000142) 65.0 73.7 70.6 57.9 59.5 Glycine 131.7 ± 132.9 ± 133.2 ± 127.5 ± 130.2 ± 0.26 N.S. N.S. (0000123) 19.8 21.3 20.8 20.6 21.5 Histidine 36.3 ± 36.9 ± 36.9 ± 35.0 ± 37.1 ± 0.32 N.S. N.S. (0000177) 6.1 5.8 5.6 6.4 6.2 Lactate 238.4 ± 171.9 ± 159.1 ± 151.3 ± 144.1 ± 0.015* N.S. 0g vs. 750g; 0.024 (0000190) 57.8 37.3 32.0 37.2 34.5 0g vs. 3000g; 0.012 97

0g vs. 15,000g; 0.008 Methanol 30.6 ± 33.3 ± 32.8 ± 30.6 ± 31.3 ± 0.22 N.S. N.S. (0001875) 4.3 4.3 4.3 3.4 3.9 Methylamine 9.2 ± 9.6 ± 9.4 ± 8.6 ± 8.8 ± 0.20 N.S. N.S. (0000164) 1.0 0.9 1.0 1.3 1.2 Phenylalanine 41.5 ± 41.4 ± 41.9 ± 39.3 ± 34.6 ± 0.29 N.S. N.S. (0000159) 5.0 4.9 5.6 5.1 6.5 Propionate 523.0 ± 564.2 ± 549.6 ± 498.4 ± 492.2 ± 0.25 N.S. N.S. (0000237) 130.9 134.1 127.3 89.4 92.3 Pyruvate 122.8 ± 119.8 ± 119.9 ± 116.9 ± 119.7 ± 0.40 N.S. N.S. (0000243) 19.7 18.6 19.5 19.5 20.9 Succinate 69.7 ± 72.3 ± 71.1 ± 67.5 ± 69.4 ± 0.24 N.S. N.S. (0000254) 13.7 14.8 14.2 13.4 13.8 Taurine 158.5 ± 161.0 ± 156.3 ± 150.6 ± 153.6 ± 0.46 N.S. N.S. (0000251) 28.9 32.7 28.2 31.4 32.5 Tyrosine 35.7 ± 34.6 ± 34.0 ± 32.7 ± 28.6 ± 0.29 N.S. N.S. (0000158) 3.5 3.1 2.9 3.2 4.4 Trimethylamine 2.7 ± 2.8 ± 2.7 ± 2.6 ± 2.6 ± 0.46 N.S. N.S. (0000906) 0.5 0.5 0.6 0.6 0.6

98

Table S-3: Analysis of freeze-thaw considerations on metabolite concentrations.

Metabolite Metabolite concentration (mean ± SEM, µM) of Repeated Bonferroni Bonferroni post-hoc sample subject to freeze-thaw treatment: measures post-hoc test test of normalised ANOVA (p (groups data Centrifuged Centrifuged Frozen, Centrifuged, value) compared; p (groups compared; , not frozen , frozen, thawed, four freeze- *= < 0.05 value) p value) thawed centrifuged thaw cycles Acetate 3366.0 ± 3478.0 ± 3358.0 ± 3560.0 ± 0.14 N.S. N.S. (0000042) 398.1 453.0 406.0 445.4 Acetoin 40.5 ± 40.4 ± 38.6 ± 40.2 ± 0.04* N.S. N.S. (0003243) 5.3 5.2 4.8 5.3 Alanine 107.4 ± 106.0 ± 107.7 ± 106.5 ± 0.58 N.S. N.S. (0000161) 24.9 23.6 23.6 23.1 Butyrate 171.0 ± 169.2 ± 165.1 ± 167.1 ± 0.44 N.S. N.S. (0000039) 23.4 22.3 20.5 20.8 Choline and 14.3 ± 14.1 ± 14.4 ± 14.2 ± 0.66 N.S. N.S. choline- 3.5 3.4 3.5 3.5 containing compounds (0000097) Citrate 57.4 ± 54.4 ± 57.5 ± 55.7 ± 0.44 N.S. N.S. (0000094) 13.1 12.2 13.2 11.4 Dimethylamine 13.2 ± 13.2 ± 12.8 ± 12.9 ± 0.50 N.S. N.S. (0000087) 3.1 3.2 3.0 2.8 Ethanol 79.2 ± 80.2 ± 76.6 ± 80.6 ± 0.23 N.S. N.S. (0000108) 13.3 13.5 11.9 12.9 Formate 80.0 ± 81.3 ± 68.7 ± 84.9 ± 0.14 N.S. N.S. (0000142) 28.1 28.7 25.8 31.9 Glycine 143.2 ± 142.7 ± 142.2 ± 144.4 ± 0.42 N.S. N.S. (0000123) 24.8 24.7 24.1 23.5 Histidine 19.7 ± 18.0 ± 19.6 ± 18.5 ± 0.14 N.S. N.S. (0000177) 5.2 4.6 5.1 5.0 Lactate 333.1 ± 334.8 ± 326.0 ± 344.9 ± 0.17 N.S. N.S. (0000190) 85.2 86.2 84.2 87.6 99

Methanol 20.6 ± 20.2 ± 20.6 ± 21.9 ± 0.18 N.S. N.S. (0001875) 3.3 3.2 3.3 3.8 Methylamine 8.9 ± 9.2 ± 8.7 ± 8.8 ± 0.32 N.S. N.S. (0000164) 1.7 1.6 1.5 1.4 Phenylalanine 53.9 ± 51.5 ± 53.4 ± 52.5 ± 0.36 N.S. N.S. (0000159) 10.5 9.9 9.6 9.6 Propionate 468.9 ± 489.5 ± 468.6 ± 498.9 ± 0.19 N.S. N.S. (0000237) 109.8 114.9 108.5 114.1 Pyruvate 115.0 ± 115.2 ± 116.4 ± 115.6 ± 0.81 N.S. N.S. (0000243) 22.8 23.0 23.5 22.6 Succinate 222.5 ± 217.9 ± 215.3 ± 218.8 ± 0.38 N.S. N.S. (0000254) 48.0 46.4 43.4 46.4 Taurine 209.2 ± 213.4 ± 217.2 ± 220.6 ± 0.25 N.S. N.S. (0000251) 39.7 39.0 39.9 42.5 Tyrosine 40.0 ± 39.9 ± 41.8 ± 41.2 ± 0.07 N.S. N.S. (0000158) 10.4 10.4 11.0 10.6 Trimethylamine 2.7 ± 2.7 ± 2.7 ± 2.6 ± 0.28 N.S. N.S. (0000906) 0.6 0.5 0.6 0.5

100

Table S-4: Analysis of quantification method on metabolite concentrations.

Metabolite Metabolite concentration (mean ± SEM, µM) of Repeated Bonferroni sample quantified by: measures post-hoc ANOVA test Internal, Internal, External TSP (p value) (groups buffered TSP unbuffered TSP *= < 0.05 compared; p-value) Acetate 3336.0 ± 2990.0 ± 3480.0 ± 0.05* External vs. (0000042) 380.8 194.7 279.9 internal, unbuffered; 0.049 Acetoin 38.9 ± 38.6 ± 38.8 ± 0.97 N.S. (0003243) 4.2 4.6 4.5 Alanine 73.8 ± 72.5 ± 77.9 ± 0.31 N.S. (0000161) 9.2 7.3 9.2 Butyrate 154.2 ± 142.2 ± 149.3 ± 0.38 N.S. (0000039) 25.9 18.1 20.9 Choline and 9.2 ± 9.5 ± 9.8 ± 0.61 N.S. choline- 1.8 1.8 2.0 containing compounds (0000097) Citrate 31.9 ± 30.6 ± 33.7 ± 0.34 N.S. (0000094) 5.4 3.9 4.6 Dimethylamine 6.1 ± 5.9 ± 6.5 ± 0.28 N.S. (0000087) 1.0 0.8 1.1 Ethanol 101.1 ± 90.4 ± 99.3 ± 0.35 N.S. (0000108) 31.9 22.5 28.1 Formate 87.6 ± 68.8 ± 88.7 ± 0.26 N.S. (0000142) 61.8 45.4 59.0 Glycine 125.5 ± 122.0 ± 128.9 ± 0.59 N.S. (0000123) 21.3 16.6 21.4 Histidine 35.8 ± 36.0 ± 38.5 ± 0.57 N.S. (0000177) 6.3 5.8 6.6 Lactate 124.7 ± 129.8 ± 144.1 ± 0.37 N.S. (0000190) 24.7 32.8 34.5 Methanol 31.9 ± 27.7 ± 31.0 ± 0.012* External vs. (0001875) 4.5 3.4 4.0 internal, unbuffered; 0.024 Methylamine 9.1 ± 7.5 ± 8.8 ± 0.07 N.S. (0000164) 1.3 0.9 1.2 Phenylalanine 40.7 ± 42.5 ± 39.6 ± 0.32 N.S. (0000159) 5.2 5.0 4.9 Propionate 491.0 ± 456.5 ± 492.2 ± 0.53 N.S. (0000237) 120.7 120.7 92.3 Pyruvate 113.0 ± 112.4 ± 119.3 ± 0.53 N.S. (0000243) 19.4 20.8 21.0 Succinate 146.2 ± 148.4 ± 143.1 ± 0.77 N.S. (0000254) 28.8 20.8 26.9 Taurine 139.4 ± 138.8 ± 151.9 ± 0.50 N.S. (0000251) 25.8 36.8 33.1 Tyrosine 29.3 ± 29.7 ± 30.3 ± 0.73 N.S. (0000158) 3.3 2.3 2.9 Trimethylamine 2.4 ± 2.6 ± 2.4 ± 0.22 N.S. (0000906) 0.6 0.6 0.6

101

3.4.2 Assessment of centrifugation and freezing on salivary bacterial viability CFU densities are presented in Figure 3.1. Centrifugation reduced the CFU density of WMS by approximately two orders of magnitude on average. Freezing had no effect on the viability of salivary bacteria.

Figure 3.1: A comparison of the effects of centrifugation and freezing on WMS bacterial viability. Data are shown as mean ± SEM, n = 10. Analysis was by repeated measures ANOVA with Bonferroni post-hoc test.

3.4.3 Assessment of multivariate analyses of digitised SDS-PAGE salivary protein profile

Gel scans with superimposed lane profiles and a PCA plot of the intra-gel and inter-gel samples are shown in Figure 3.2. Visual inspection of the gel lanes shows the inter-gel samples are less consistent than the intra-gel samples, both in intensity and relative front. This is apparent on the superimposed excel plots of the gel profiles as well as the PCA plot. The differences in the inter-gel profiles include the intensity of the lanes and protein bands and the relative front particularly in the amylase region. These subtle differences could arise from differences in the staining and destaining protocol as well as the imaging process, and differences in electrophoresis units yielding slightly different degrees of separation in the same time period due to voltage/current fluctuations. Furthermore, the samples on the same gel would have been electrophoresed using parameters for a single gel, whereas the other gels were run two at a time which might affect the rate of separation.

The gels from the second experiment are shown in Figure 3.3. Visually, the participants can be grouped both visually and by the PCA analysis. K means cluster analysis showed the algorithm 102 correctly clustered samples within and between gels by participant, however all samples from participants 2 and 4 were clustered together (Figure 3.4 a.). The PCA analysis does reveal an inter-gel effect is being detected and samples from the same individuals run on separate gels are more dispersed than samples from the same individual run on the same gel. This inter-gel differences in the projected data appear to be proportionate for each participant’s samples, indicated by the arrows in Figure 3.4 b. Attempting to correct for between-gel differences by ComBat batch correction did not resolve this issue (Figure 3.4 c.). Treating the first lane of participant three’s samples on each gel as a standard and normalising data within the gels to this was more effective in correcting the inter-gel differences (Figure 3.4 d.).

103

Figure 3.2. Inter- and intra-gel stability of digitised salivary protein profiles. a i. shows a PAS and coomassie stained polyacrylamide gel of the same saliva sample run six times. a ii. shows the same sample run on six separate gels. b i. shows the scanned lanes from the intra-gel samples superimposed and b ii. shows the scanned lanes from the inter-gel samples superimposed. As can be seen on both the gels and the scanned lanes the intra-gel samples are more consistent than the inter-gel samples in both horizontal and vertical axes (i.e. measured intensity and relative front). This difference in consistency is quite apparent when visualised in the grouping on the PCA plot, c.

104

Figure 3.3. Saliva samples from four participants run in triplicate on four PAS and coomassie stained polyacrylamide gels. Visual inspection shows good similarity within an individual both on the same gel and between different gels. Subtle differences can be noticed such as the advancing front of the samples on gel 3 being closer to the bottom of the gel compared to those of gel 2, as evidenced by the lip of the gels which hasn’t been fully cropped. This illustrates how the same electrophoresis parameters (voltage, time, current) can yield different separation between gels.

105

Figure 3.4. Investigation of inter-gel “batch” effects. a. shows PCA and k-means cluster analysis of the digitised lane profiles from the gels shown in Figure 3.3. The clustering algorithm has correctly clustered all lanes from the participants however participants 2 and 4 were clustered together. b. shows the apparent inter-gel “batch effect” and the arrows indicate a proportionate relative spacing of lanes run on different gels. c. shows the projected data following ComBat batch correction process. There is no observable correction of the batch effect. The first lane of participants c’s saliva on each gel has been treated as an inter- gel standard. d. shows the projected data after alignment of the standard lanes. The inter-gel separation of samples has been partially corrected by this method.

It is therefore concluded that an inter-gel “batch effect” is observed when digitising sample lanes on polyacrylamide gels. Therefore, samples to be compared should be run on the same gel. Where it is necessary to compare samples between gels (e.g. to measure the overall 106 variance in profiles from a participant group) differences should be mitigated by normalising samples to the standard lane on each gel.

3.4.4. Assessment of suprathreshold intensity scale and tastant concentrations

Twelve participants providing their blinded intensity ratings for three concentrations of four different tastants (total of 144 ratings). There was only one instance of tastant solutions being incorrectly ranked for taste intensity based on their concentration, giving a correct response rate of 99 percent.

Box plots of the intensity ratings of the different tastant solutions are shown in Figure 3.5. Based on the desired criteria 0.25 M sucrose, 8 mM caffeine, 1 ppm were selected. Mean intensity ratings for these solutions were all approximately at the centre of the scale, with uniform distributions. While ratings for 100 ppm menthol were closer the scale midpoint, the ratings for 250 ppm menthol appeared more uniform and thus 250 ppm was selected as the concentration for menthol.

Figure 3.5: Boxplots of the participant intensity ratings (n = 12) for different concentrations of four tastants.

107

At an individual level, none of the tastant solutions correlated with fungiform papillae density. At group level, however, there was evidence that the scale discriminated between the intensity ratings of the four participants with lowest and highest FPD for the basic tastants (sucrose and caffeine) but not the TRP tastants (menthol and capsaicin). Data are presented in Figure 3.6.

Figure 3.6: comparison of intensity ratings for 0.25 M sucrose and 8 mM caffeine between low and high FPD participants, (n = 4 per group). Error bars are mean ± SEM, data are compared by t-test.

3.4.5 Assessment of protocol to return salivary metabolites to baseline levels

Alongside flow rate data, salivary metabolites significantly differing following oral sucrose exposure are presented in Figure 3.7. Small but consistent increases in flow rate relative to control following sucrose rinse were observed, hence the necessity to adjust metabolite concentrations for flow rate variations. Compared to control, samples collected following 0.25 M sucrose had significantly higher outputs of lactate, succinate, acetoin and residual sucrose. Pyruvate was also raised, and this approached significance (p = 0.06). The first set of 3 x 20 second water rinses returned metabolite outputs to their baseline concentrations. In the case of propionate, a significant reduction in output relative to control was observed following the second rinsing step, indicating a risk of excessive rinsing reducing metabolite concentrations below baseline. It was therefore concluded that a single set of 3 x 20 second water rinses was adequate to return metabolite concentrations to their baseline levels.

108

Figure 3.7: Comparison of flow rate and metabolite output (concentration x flow rate) following oral exposure to 10 ml water control and 0.25 M sucrose solutions. Data are analysed by Friedman’s test following Dunn’s post-hoc to detect significant differences relative to the control samples. N = 5, data are shown as mean ± SEM.

3.5 Discussion

The results presented in this chapter essentially enabled the studies presented hereafter to be conducted with confidence that the subsequent data collection was valid and would allow meaningful interpretation. Regarding the pre-treatment of saliva for 1H-NMR spectroscopy and SDS-PAGE, saliva was found to be a resilient fluid which was not greatly affected by freeze-thaw treatments, including multiple freeze thaw cycles in the case of 1H-NMR spectroscopic analysis. It was found that the timing of centrifugation of saliva in relation to freezing did not affect the analysis by 1H-NMR spectroscopy.

Although SDS-PAGE is one of the most widely used protein separation techniques in bioscience the literature on applying multivariate statistics to sample profiles is sparse. Interestingly, some literature exists on this topic, originating from the food science field. 109

Dewettinck et al. (1997) 236, describe a methodology for applying PCA to the lane profiles of cheese to discriminate different types of cheese. The authors describe a fairly laborious process of converting lane data in volumograms of optical density against migration distance. The resulting data conversion appears like it would lead to a loss of fine detail of the data, however the authors were still able to discriminate cheese types using their technique. More recently Olias et al. (2006) 237, describe a similar SDS-PAGE based approach to studying the protein profiles of ham. The authors convert images of gels into data based on the pixel intensity along the length of the different lanes and apply multivariate analytical techniques such as PCA and PLS (partial least squares) analysis. The main difference between the current technique and these previously described methods is that the data used is generated entirely by the Bio-Rad gel imaging software and simply extracted for further analyses, rather than requiring further manual steps for data manipulation. Our results suggest the technique has reasonable discriminant ability for salivary protein profiles of different donors, however one caveat is the presence of inter-gel “batch” effects, which could be minimised but not eliminated. Further refining of the method in future may help reduce these effects further, but at present it is advised that running samples on different gels should be avoided where feasible.

The sensory data gathered in the blinded validation experiment described suggest the glVAS scale was accurately able to discriminate taste sensitivity at an individual level. The tastants concentrations selected for the majority of future analyses were generally iso-intense (with the exception of 250 ppm menthol, chosen for its more uniform distribution). The lack of correlation at an individual level between taste response and FPD was perhaps unsurprising. FPD continues to be a somewhat controversial measure of taste sensitivity with large studies failing to report correlations between FPD and taste responses 176. Nevertheless, when comparing group means of the low and high FPD participants differences in taste ratings were detected by the scale. Our classification was defined based on the original “supertaster” definition of Bartoshuk et al., (1994) 173, (non-tasters with FPD < 80/cm2 and supertasters FPD > 100/cm2 ). The data suggests the lower FPD individuals had significantly lower taste ratings for 0.25 M sucrose and 8 mM caffeine. Therefore, the scale and sensory methodology appear to show a degree of broad discrimination between biological classifications of tasters. While this approach is advocated by some in designing sensory scales 174, correlative data between taste ratings and other parameters tends to be highly variable and correlations where detected are often weak 238. Comparisons of mean ratings at extreme ends of the scale may therefore aid detection of biological differences between tasters compared to correlation analyses of the full data set. The fact menthol and capsaicin ratings did not differ based on 110

FPD unlike the basic tastes is not surprising, as the former are detected by a wide range of oral tissues whereas the latter act on taste buds within fungiform, foliate and circumvallate papillae 231,239.

The investigation of intra-oral metabolism of sucrose showed rapid and pronounced changes in the salivary concentrations of several metabolites. For metabolomic study of saliva, the primary concern in removing residual tastants from the mouth is to prevent their ongoing metabolism and return metabolite concentrations to baseline levels. Various sensory methodologies, including ISO standards 157, described some form of rinsing step to clear residual taste from the mouth,. Unfortunately, these are not often based on data or strictly defined, and often may be left to the participants discretion. This study showed that a succession of three water rinses each of 20 seconds duration was sufficient to clear salivary metabolites generated by intra-oral sucrose metabolism.

111

Chapter 4: Determining host and microbial contributions to the salivary metabolome

4.1 Introduction

Literature on salivary metabolomic profiling by 1H-NMR spectroscopy tends to focus on detection and quantification of analytes in the saliva of different participant groups with limited investigation as to the biological events that generated the metabolites. Studies generally acknowledge that the oral microbiome must be involved in shaping the net metabolic composition of saliva, although the extent to which bacteria alter saliva upon secretion into the mouth is unclear. Another important detail lacking evidence is the substrate for bacterial metabolism. In one of the earliest metabolomic studies of saliva, Silwood et al. (2002) 240, recognised the role of oral bacteria in contributing to the salivary metabolome however were unable to determine the extent of this contribution. Takeda et al. (2009) 81, suspected the lactate observed in saliva was due to the microbial breakdown of glucose although were conservative in their discussion, calling for additional research to more fully explain the results they observed. The source of acetate in saliva was speculated by Aimetti et al. (2012) 68, to result from an enhanced degradation of collagen and/or amino acids by proteolytic bacteria in the absence of carbohydrates. Contrary to this explanation, Dame et al. (2015) 60, attributed raised acetate in saliva collected in the morning to be a product of increased carbohydrate metabolism overnight.

An apparent necessity was therefore identified for work to determine how salivary metabolites come to be present in the mouth, whether by host or microbial metabolism. Additionally, examination of interactions between microbial and host metabolite production was identified as an important deficit in existing literature. Answering such questions is critical to ultimately reconcile observed results from metabolomic studies of saliva with their biological interpretation and significance.

This chapter incorporates work published previously, in Journal of Oral Microbiology, Volume 11, issue 1, article 1617014 241. The final manuscript has been included in this chapter without modification from the original publication. The published supplemental information is included along with a small amount of additional unpublished data analysis, included as further supplemental results. Author contributions to this work are as follows: Alexander Gardner – study design, sample collection, data analysis and manuscript writing; Harold

112

Parkes – data analysis; Po-Wah So – study design, data analysis; Guy Carpenter – study design, data analysis. All authors edited and approved the final version of the manuscript.

Additional work was conducted building on the findings of the published work. The additional experiment was an in-vitro study of the degradation of salivary proteins by oral bacteria, and the resulting metabolite production and baseline metabolite consumption. This was deemed pertinent in light of the work of Ruhl (2012), who recognised that although oral bacteria produce a range of enzymes that would allow them to catabolise salivary proteins, little is known about this process 54. A further element to this second experiment was the study metabolic activity of biofilm collected from different tongue sites. Biofilm was collected from sites adjacent to fungiform papillae or circumvallate papillae. It has been demonstrated that microbial metabolites can activate taste receptors such as the TAS2R38 bitter receptor 242. Furthermore, the idea that differences in the microbial composition adjacent to taste papillae could be related to differences in taste sensitivity has recently been studied with respect to both fungiform papillae and circumvallate papillae 177 243. It was therefore intended to assess whether bacteria taken from these different oral sites would generate a different pattern of metabolites which may be capable of differential stimulation of local taste receptor cells.

Neither the published study nor additional experimental data satisfactorily resolved the source of taurine in saliva, so further experiments were conducted to provide an answer to this question. As taurine is a known tissue metabolite 244, and has been shown to inversely correlate with salivary flow rate 71, buccal epithelial cells were investigated as a source of salivary taurine. Since there is some evidence of 1H-NMR spectral peaks that could originate from taurine being present in submandibular/sublingual saliva, albeit from a subject with active sialadenitis (glandular inflammation typically due to infection) 90, the metabolite profile of submandibular/sublingual from healthy individuals’ saliva was investigated. Finally, GCF was investigated for the presence of taurine.

4.2 Aims and Objectives

The aims of the work conducted in this chapter were as follows: 1. Examine the compositional differences in metabolite profile between parotid saliva (i.e. saliva at the point of secretion) and whole-mouth saliva, thus determining which metabolites are of host origin. 2. Determine the role of oral bacteria in shaping the metabolite composition of whole mouth saliva. This was achieved by examining the relationship between bacterial load

113

and metabolite concentration in-vivo, as well as looking at the production of metabolites from sterile parotid saliva by oral bacteria in-vitro. 3. Determine the extent to which oral bacteria can utilize salivary proteins as a substrate and examine the consequent metabolic yield from salivary . 4. Determine whether there are any differences between the metabolites generated by bacteria collected from the fungiform papillae region compared to those from the circumvallate papillae region.

4.3. Materials and methods

4.3.1 Additional analyses of published data

Correlations between the concentrations of different metabolites in WMS were analysed in R. The resulting correlation matrix is presented as an additional unpublished supplemental figure, showing only correlations that were significant (p < 0.05).

Concentrations of taurine in WMS, PS and plasma were compared. Due to partial overlap from glucose peaks, plasma taurine was quantified by integrating the triplet peak between 3.24 and 3.28 ppm.

4.3.2 Assessment of potential salivary taurine sources

Submandibular/sublingual saliva Submandibular and sublingual saliva was collected from three healthy volunteers as described in Chapter 2. Samples were stored, prepared and analysed by 1H-NMR spectroscopy, also as described.

Buccal epithelial cells Buccal epithelial cell scrapes and WMS samples were collected from five volunteers. The epithelial cell density of WMS samples was determined as described. Buccal epithelial cells were adjusted to approximately the same or higher density as for WMS using PBS. Epithelial cells were then ultrasonically lysed (ten 2 second pulses, with 5 second intervals to reduce heat). Cell lysate was then centrifuged using 10 kDa molecular weight filters (Sigma) at 14,000 g for ten minutes at 4 °C. WMS samples were treated in the same way for consistency. All samples were stored, prepared and analysed by 1H-NMR spectroscopy as described.

114

Gingival-crevicular fluid GCF was collected from four healthy volunteers. GCF (2 µl) was eluted using 50 µl PBS. Samples were then analysed using a colorimetric taurine assay (Abcam, Cambridge, UK).

4.3.3 In-vitro saliva inoculation study

Sample collection Parotid saliva was collected from a single volunteer to serve as the substrate for the bacterial inoculums in this study. To guarantee sterility, the parotid saliva was filtered through a 0.2 µm filter. The saliva was then aliquoted into sterilised tubes as 500 µl aliquots and stored at -80 °C for one week prior to use.

Bacterial inoculums were sourced from healthy adult volunteers. Antibiotic use in the preceding six months and active oral disease were exclusion criteria. Six volunteers were recruited. Unstimulated WMS was collected from each volunteer. Biofilm from the anterior of the tongue (adjacent to fungiform papillae) and posterior of the tongue (adjacent to circumvallate papillae) was collected using sterilised, pre-weighed plastic scrapers.

Inoculation and incubation conditions Parotid saliva aliquots were thawed on ice. Aliquots were inoculated with 20 µl of WMS or 20 mg of tongue biofilm from either tongue site (i.e. 4% by volume/mass, respectively). Preliminary experiments suggested 4% biofilm inoculum would be adequate to observe differences within a timescale of under 48 hours. It was anticipated that a lower initial inoculum would not produce detectable differences within a reasonable timescale 245. Control samples were prepared with 20 µl of sterile PBS. Samples were duplicated for an aerobic and anaerobic incubation. Eppendorfs lids were pierced with sterilised forceps to ensure gas exchange could occur with the anaerobic or aerobic environment. Eppendorfs were stored inside sterile universal tubes with wet tissue paper in the bottom to minimise fluid loss from dry conditions. Samples were incubated at 37 °C for 24 hours in either aerobic conditions or in an anaerobic cabinet with gas blend 10% H2, 10% CO2 and 80% N2.

Further control samples were prepared immediately prior to analysis. One control was parotid saliva and PBS to control for effects of incubation, the second control was parotid saliva and 20 µl of pooled WMS to control for the metabolite addition present in the WMS inoculum. The experimental design is summarised in Figure 4.1.

115

Figure 4.1: An outline of the experimental design with sample and control preparation and the relevant incubation conditions.

4.3.4 Sample analyses

Bacterial counts Following incubation, samples were vortexed to distribute bacterial cells evenly throughout the solution. An aliquot (20 µl) was then pipetted and logarithmically diluted in sterile PBS down to 1:100,000. Samples (20 µl) of the 1:1000 and 1:100,000 dilutions were plated onto pre-prepared FAA agar with 5% defibrinated horse blood and incubated at 37 °C for 48 hours under anaerobic conditions. The final bacterial load of inoculated parotid saliva was assessed by counting the CFU measured from the incubated plates. Undiluted control samples were plated and incubated as further confirmation of their sterility.

Sample protein and metabolite composition Following vortexing, samples were centrifuged at 15,000 g at 4 °C for 10 minutes to remove residual bacterial debris. Pelleted bacteria were retained. Supernatant was prepared for 1H-NMR spectroscopy by mixing 4:1 with NMR buffer (final concentration of 0.1 mM TSP and 10% D2O) in 5 mm OD NMR tubes. Samples were analysed using a CPMG pulse sequence as described in Chapter 3.

Supernatant was prepared for SDS-PAGE analysis and were electrophoresed and stained with coomassie as described in Chapter 2. On each gel aerobic or anaerobic samples from two participants were run along with a sample of unincubated parotid saliva (with 4% PBS, to account for dilution factors present in test samples) and the relevant (aerobic/anaerobic) incubated PBS inoculated parotid saliva.

116

To ascertain whether any salivary proteins were aggregated by the bacteria, the pellet was resuspended up to the original 500 µl in sterile PBS with 1% general protease inhibitor cocktail set I (Sigma). The pellet was then disrupted by ultrasonication for 20 seconds (10 × 2 second pulses, spaced 5 seconds apart to prevent heating). Lysed pellet was then reconstituted by centrifugation and both the pellet and PBS supernatant were prepared for SDS-PAGE analysis in volumes proportionate to that of the salivary supernatant.

A preliminary analysis of the effects of the ultrasonication process on saliva as well as the effects of protease inhibitor cocktail on SDS-PAGE gel profile was conducted prior to the sample preparation.

Statistical analyses Following processing, 1H-NMR spectra were integrated in 0.01 ppm buckets using MestRec v (MestreLab Research) from 8.5 ppm to 0.7 ppm, excluding the water peak (4.5 – 5.5 ppm) and including the TSP reference peak (-0.02 to 0.02 ppm). Bucket integrals were scaled to the total spectra and spectral profiles were analysed by PCA and K-means cluster analysis in Knime. Differences in metabolite profile resulting from the different inoculums were compared, as were the effects of aerobic and anaerobic incubation.

Integrals of individual metabolites were manually quantified for statistical analysis. Comparisons were made relative to the post-incubation concentrations in PBS-inoculated parotid saliva by one-sample t-test. Comparisons between final metabolite concentrations in bacteria-inoculated samples were compared by Tukey’s multiple comparison post-hoc test following one-way ANOVA.

Protein consumption was measured by densitometric analysis of the total lane profile relative to unincubated control sample lanes on each gel.

Protein consumption by the different inoculums (WMS, fungiform biofilm and circumvallate biofilm) was compared by Tukey’s post-hoc following repeated measures ANOVA. Final CFU of the different inoculums was analysed by Tukey’s post-hoc following repeated measures ANOVA following logarithmic transformation.

The relationship between final CFU and protein consumption of the inoculated samples following anaerobic and aerobic incubation was compared by Pearson’s correlation analysis.

117

The relationship between protein consumption and change in metabolite concentration of the inoculated samples following anaerobic and aerobic incubation was compared by Pearson’s correlation analysis.

118

4.4 Results

4.4.1 Determining bacterial and host contributions to the human saliavary metabolome

119

120

121

122

123

124

125

126

127

Bacterial and host contributions to the human salivary metabolome

Supplemental Material

Supplemental Figure 1: Partial 1D 600 MHz spectrum of tartaric acid, showing a singlet at 4.33 ppm arising from the equivalent protons on carbons 2 and 3. No peak at this region was observed in stimulated parotid saliva samples, indicating no contamination between sample and stimulus occurred.

Supplemental Table 1: Comparison of metabolite and protein concentration in stimulated and unstimulated parotid saliva, (n=8). Protein was measured using a bicinchoninic acid assay (Thermo-Scientific, Rockford, Illinois, USA.) Unassigned metabolite concentration is expressed in relative units as the concentration cannot be determined without knowing the structure. Except citrate, mean concentrations of all analytes decreased upon stimulation, however this was significant (p<0.05) only for urea and protein. Analyte Unit Unstimulated Stimulated P-value (paired mean ± SEM mean ± SEM t-test; * = p < 0.05) Citrate mmol/L 0.03 ± 0.01 0.03 ± 0.01 0.98 Lactate mmol/L 0.10 ± 0.02 0.04 ± 0.01 0.07 Urea mmol/L 2.70 ± 0.22 1.74 ± 0.42 0.02* Unassigned NA 0.24 ± 0.05 0.15 ± 0.04 0.30 Total protein mg/dl 316 ± 49 152 ± 18 0.02*

128

Supplemental Table 2: For metabolites where statistical differences were detected by ANOVA adequate power was confirmed. Power calculations for a repeated measures ANOVA with three multiple comparisons, with an alpha level of 0.05 were conducted. A threshold of power > 0.8 was deemed sufficient to accept the result of the ANOVA, therefore the power to accept the ANOVA result for formate is slightly below threshold. Metabolite Power Lactate 0.97 Propionate 0.92 Formate 0.66 Acetate 0.96 Butyrate 0.92 Urea 0.96

129

Bacterial and host contributions to the human salivary metabolome

Additional Unpublished Supplemental Material

Supplemental Figure 2: A visual summary of the correlations between the concentrations of salivary metabolites. Significance is set at p < 0.05, non-significant results are blank.

130

4.4.2 Assessment of potential salivary taurine sources

Comparison of plasma, PS and WMS taurine concentrations A comparison of taurine concentrations in the various biofluids is shown in Figure 4.2. WMS taurine concentrations were significantly higher than plasma concentrations. Mean plasma concentration was 19 µM, ranging from 11 to 29 µM. These values were slightly below literature values (mean ~ 40 µM). Taurine levels in WMS could not therefore be derived from plasma without a mechanism of actively concentrating taurine into saliva.

Figure 4.2: Comparison of taurine in biofluids. Data (n = 11) were analysed by repeated- measures ANOVA with Tukey’s post-hoc test.

Submandibular/sublingual saliva A comparison of submandibular/sublingual gland saliva and parotid gland saliva is shown in Figure 4.3. Samples are from the same individual, however collected on different days. Spectra have been scaled to the lactate peak in both samples. The spectral profiles are similar for both fluids and the major host derived metabolites are labelled. Although a low level of acetate was present in the submandibular/sublingual gland sample, this likely reflects a degree of inevitable contamination of submandibular/sublingual saliva as it contacts the oral cavity briefly upon collection unlike parotid saliva. No detectable taurine was present in any submandibular/sublingual gland samples, eliminating the possibility of taurine being specific to the submandibular or sublingual glands. Parotid saliva appears to be a good approximation of the other major glands with respect to metabolite composition.

131

Figure 4.3: Partial 600 MHz 1D CPMG 1H-NMR spectra (0.7 to 8.5 ppm, excluding 4.5 to 5.5 ppm), comparing parotid saliva and submandibular/sublingual saliva from the same donor. The aromatic regions have been vertically increased two-fold and horizontally compressed two-fold to fit. Prominent peaks are labelled. Samples were collected and spectra were acquired on separate occasions. Spectra are scaled to the lactate peak to assess relative metabolite concentrations.

Buccal epithelial cells A summary of data for examining the possible epithelial cell derived taurine is presented in Figure 4.4. Buccal epithelial cells collected by a cheek scrape generally showed some degree of membrane damage, although some cells did exclude Trypan Blue. Following ultrasonic lysis, no intact cells remained. The buccal epithelial cell concentrations of cheek scrape samples in PBS were on average higher than those of WMS. Despite this, taurine concentrations were significantly lower, indicating buccal epithelial cannot be concluded as the main source of taurine in WMS. The possibility remains that intracellular contents of superficial, friable epithelial cells may have already leaked into the oral cavity prior to collection.

132

Figure 4.4: Investigating taurine content of buccal epithelial cells. a. and b. show resuspended buccal epithelial cells pre- and post-ultrasonic lysis. Prior to lysis most cells have some degree of trypan blue uptake, indicating membrane damage, although some intact cells (arrows) are present. After lysis only cell fragments are present. c. Buccal cell samples were adjusted to a slightly higher density in PBS than the corresponding participant’s WMS epithelial cell density, this difference was not significant. d. Epithelial cell lysates had minimal taurine, absent in most participants. Concentrations in cell lysate were significantly lower than in WMS despite cell densities being the same. Data analysed by paired t-test, (n = 5), error bars display mean ± SEM.

Gingival crevicular fluid Gingival crevicular fluid was found to have taurine concentrations in excess of two order of magnitude greater than WMS (26 mM in GCF vs. 0.1 mM in WMS, on average). A comparison of WMS and GCF taurine concentrations is depicted in Figure 4.5. Based on the high concentrations of taurine in GCF paired with the relative dilution of WMS by glandular fluids in the mouth, it seems that GCF is the major source of taurine in WMS.

133

Figure 4.5: Comparison of taurine concentrations in GCF and WMS. Taurine concentrations measured in GCF by colorimetric assay were on average 260 times greater than those measured in WMS by 1H-NMR spectroscopy. Sample types were from different individuals, data was analysed by t-test. The y-axis is interrupted to allow visualisation of error bars (mean ± SEM) as WMS taurine concentrations would not be visible on the graph relative to GCF taurine concentrations.

4.4.3 In-vitro saliva inoculation study

Preliminary investigation of ultrasonication of saliva and protease inhibitor cocktail dilutions The gels shown in Figure 4.6 illustrate the effect of ultrasonication protocol on WMS and the presence of any band occurring from the protease cocktail itself at different dilutions. Ultrasonication caused the MUC5B in unstimulated saliva to breakdown and form a diffuse staining pattern rather than a distant band. This effect was not observed for other major protein bands. Therefore, ultrasonication of parotid salivary proteins (lacking mucins) would not cause adverse protein changes. The protease inhibitor cocktail showed a band when analysed neat, and a faint band at a 1:10 dilution (shown in the black box), however no band was visible at the recommended dilution of 1:100.

134

Figure 4.6: Ultrasonication effects on salivary protein. Panel a. shows example comparisons of unstimulated WMS pre-ultrasonication (i.) and post-ultrasonication (ii.), from two participants. The only major protein to be affected by the process was MUC5B which appeared to have been reduced in molecular weight by the ultrasonication process. Ultrasonication therefore would not appear to alter non-mucin proteins found in parotid saliva. Panel b. shows the presence of a band resulting from neat protease inhibitor cocktail. A faint band occurs at a 1:10 dilution however at the recommended concentration (1:100) no visible band is present.

Effects of bacterial inoculation of parotid saliva on protein composition Figure 4.7 shows representative coomassie stained gels of the parotid saliva inoculated with the different sources of oral bacteria from two participants, as well as the bacterial pellets. Incubation of PBS control inoculum samples under both conditions caused some visible change in the parotid saliva statherin band and intensification of the band ~ 3 kDa. Higher molecular weight major proteins were not visibly changed by incubation with sterile PBS inoculum. A general reduction in all protein band intensities compared to controls was observed following incubation of bacterial inoculated samples, with residual amylase often being the only detectable protein after incubation of inoculated samples. No residual proteins were observed in the bacterial pellets or the PBS used to re-suspend the pellets during preparation (Figure 4.7 b.), hence the protein absent from the parotid saliva appeared to be consumed rather than simply aggregated by bacteria.

A significant reduction in total lane intensity was observed following inoculation with all bacterial sources under both incubation conditions. Furthermore, inoculation with tongue biofilm caused a significantly greater consumption of proteins than inoculation with WMS, 135 although no difference was observed between fungiform and circumvallate papillae biofilm, (Figure 4.8).

136

Figure 4.7: Example gels showing parotid saliva protein changes following inoculation with oral microbes. Figure 4.7 a. shows the parotid saliva changes post incubation (top left = aerobic; top right = anaerobic). Lanes are annotated as follows: i. – parotid saliva with 4% PBS, no incubation; ii. – parotid saliva inoculated with 4% PBS; iii. – parotid saliva inoculated with 4% WMS; iv. – parotid saliva inoculated with 4% fungiform papillae biofilm; v. – parotid saliva inoculated with 4% circumvallate papillae biofilm. Figure 4.7 b. confirms that the protein absent relative to controls in Figure 4.7 a. is not aggregated by bacteria by confirming it is absent from both the pellet itself and is also not lost from the pellets during preparation. Lanes are labelled as follows: i. – parotid saliva with 4% PBS, no incubation; ii. – WMS pellet, aerobic incubation; iii. – fungiform papillae biofilm pellet; aerobic incubation; iv. – circumvallate papillae biofilm pellet; aerobic incubation; v. – WMS pellet, anaerobic incubation; vi. – fungiform papillae biofilm pellet; anaerobic incubation; vii. – circumvallate papillae biofilm pellet; anaerobic incubation.

137

Figure 4.8: Comparison of the protein consumption from parotid saliva when inoculated with different sources of oral bacteria/biofilm, as measured by lane densitometry. FFP = fungiform papillae, CVP = circumvallate papillae. P-values are for Tukey’s post-hoc following repeated- measures ANOVA.

Comparison of final CFU resulting from different inoculums and correlation with protein consumption

A comparison of Log10 CFU/ml measured from the inoculated parotid saliva is shown in Figure 4.9. Under anaerobic conditions, bacteria were significantly more concentrated in the solutions inoculated with circumvallate papillae biofilm compared to a WMS inoculum. Under aerobic conditions there were no significant differences in final bacterial concentrations from the different inoculums.

The final bacterial concentrations correlated positively with protein consumption from the parotid saliva. This correlation was true for both aerobically and anaerobically cultivated samples, although the relationship was stronger for anaerobically incubated samples. These relationships are shown in Figure 4.10.

138

Figure 4.9: Summary of Log10 CFU/ml measured from the inoculated parotid saliva incubated under anaerobic and aerobic conditions. FFP = fungiform papillae, CVP = circumvallate papillae. P-values are for Tukey’s post-hoc following repeated-measures ANOVA, n.s. = non- significant.

Figure 4.10: Relationship between final bacterial concentrations and protein consumption from the inoculated parotid saliva under anaerobic and aerobic incubation conditions.

Bacterial metabolism of parotid salivary proteins A comparison of partial 1D 600 MHz CPMG 1H-NMR spectra of anaerobically incubated PBS- inoculated parotid saliva and anaerobically incubated circumvallate papillae biofilm inoculated parotid saliva is shown in Figure 4.11. Samples inoculated with bacteria showed abundant levels of metabolites however consumption of parotid metabolites was also observed. Some of the generated metabolites such as SCFAs, glycine and succinate are typically seen in WMS although many metabolites are not typically observed at high concentrations in healthy WMS. These include phenol, phenylacetate, 3-phenylpropionate, 5-aminopentanoate and putrescine. Higher concentrations of amino acids such as leucine, valine and proline were also present.

139

Changes in individual metabolite concentrations are summarised in Table 4.1 (anaerobic incubation) and Table 4.2 (aerobic incubation). In general, metabolite concentrations tended to be significantly different from PBS-inoculated control samples. Glucose was fully consumed in all samples. Urea was fully consumed in all tongue biofilm-inoculated samples, but not all WMS inoculated samples. Citrate was either fully consumed or significantly consumed in all samples. Lactate was significantly consumed in all samples except anaerobically incubated WMS inoculated samples. Pyruvate was significantly consumed in all tongue biofilm inoculated samples but not WMS inoculated samples.

Under both incubation conditions phenylalanine was not significantly produced by FFP tongue biofilm samples, succinate was not significantly produced by either tongue biofilm inoculated samples, 3-phenylpropinate was not significantly produced by FFP or WMS inoculated samples. Butyrate and propionate were not significantly produced by WMS inoculated samples. Considering differences in metabolite concentrations between different inoculum sources, there were no differences between CVP or FFP inoculums for any metabolite. Significantly higher concentrations of acetate, butyrate, propionate and phenol were generated by CVP and FFP tongue biofilm inoculums compared to WMS inoculums under both incubation conditions. Significantly higher concentrations of phenylacetate were generated by CVP biofilm compared to WMS under both incubation conditions. Significantly higher concentrations of 5-aminopentanoate were generated by CVP and FFP tongue biofilm inoculums compared to WMS inoculums under only aerobic incubation conditions. PCA plots of spectral profile are shown in Figure 4.12 and 4.13. When looking at all spectra (Figure 4.8), a separation between WMS and tongue biofilm inoculums was detected by k- means cluster analysis, with only one participants WMS inoculum profiles being clustered with tongue biofilm inoculum. The spectra of this participants WMS inoculated samples did contain high concentrations of metabolites otherwise associated with the tongue biofilm samples, such as phenol, phenylacetate and putrescine, indicating a different microbial profile of this participants WMS.

Analysis of incubation conditions (Figure 4.13) appeared to show no indication of separation of metabolite profiles by anaerobic or aerobic incubation. When splitting samples by inoculum type, there appeared to be a general separation by participant suggesting the inoculum donor is more important than the subsequent incubation conditions in shaping the consequent metabolite profile.

140

Correlations between protein consumption and changes in metabolite concertation for samples under both incubation conditions are shown in Figure 4.14. Metabolite generation (positive correlations) correlated significantly with protein loss for acetate and putrescine under both incubation conditions. Further correlations with butyrate, propionate, phenol and phenylacetate were observed under anaerobic conditions whereas correlations with formate and 5-aminopentanoate were observed under aerobic incubation conditions. Consumption of metabolites (negative correlations) correlated significantly with protein loss for lactate, citrate and pyruvate under both incubation conditions.

Figure 4.11: Partial 1D 600 MHz CPMG 1H-NMR spectra comparing PBS-inoculated parotid saliva (top spectrum) and tongue biofilm-inoculated parotid saliva (bottom spectra). The water peak region 4.2 – 5.2 ppm has been excluded. Aromatic spectral regions (5.2 – 8.5 ppm) are vertically scaled 16 times greater than aliphatic regions (0.78 – 4.2 ppm). Acetate and propionate peaks have been cropped (black bars).

141

Table 4.1: Summary of the concentrations of metabolites consumed and generated following 24 h anaerobic incubation of parotid saliva inoculated with oral bacteria relative to inoculation with sterile PBS. Significant results (p < 0.05) are presented in bold. NA = statistical test could not be conducted due to absence of metabolite from all samples yielding a S.D. of zero; n.s. = not significant. FFP = fungiform papillae, CVP = circumvallate papillae. Anaerobic incubation Metabolite Baseline Mean (±SEM) inoculated sample conc. (mM), (n P-value between bacterial and P-value between bacteria inoculated PBS- = 6) control inoculum (one-sample t- samples (Tukey’s). inoculum test). control conc. FFP biofilm CVP biofilm WMS CVP FFP WMS CVP CVP FFP (mM), (n = 1) v. v. v. v. v. v. baseline baseline baseline FFP WMS WMS Endogenous parotid saliva metabolites consumed Urea 0.074 0.00 (± 0.0) 0.00 (± 0.0) 0.017 (± 0.01) NA NA 0.012 n.s. n.s. n.s. Lactate 0.18 0.03 0.02 0.15 < 10-5 < 10-5 n.s. n.s. 0.02 0.01 (± 0.008) (± 0.003) (± 0.05) Citrate 0.076 0.006 0.00 0.029 < 10-5 NA 0.004 n.s. n.s. 0.02 (± 0.006) (± 0.0) (± 0.009) Pyruvate 0.05 0.02 0.01 0.03 < 10-5 < 10-5 n.s. n.s. 0.045 0.02 (± 0.003) (± 0.003) (± 0.008) Glucose 0.12 0.00 (± 0.0) 0.00 (± 0.0) 0.00 (± 0.0) NA NA NA n.s. n.s. n.s. Metabolites generated

142

Formate 0.004 0.77 (± 0.07) 0.92 (± 0.07) 0.72 (± 0.11) 0.0001 < 10-5 0.0011 n.s. n.s. n.s. Phenylalanine 0.017 0.08 (± 0.02) 0.11 (± 0.04) 0.05 (± 0.01) 0.015 n.s. 0.02 n.s. n.s. n.s. Phenol 0.00 0.16 (± 0.02) 0.13 (± 0.02) 0.052 (± 0.02) 0.0005 0.0008 0.026 n.s. 0.003 0.02 Proline 0.00 0.37 (± 0.05) 0.45 (± 0.07) 0.32 (± 0.07) 0.0005 0.001 0.006 n.s. n.s. n.s. Valine 0.00 0.19 (± 0.04) 0.20 (± 0.05) 0.10 (± 0.03) 0.005 0.01 0.01 n.s. n.s. n.s. Phenylacetate 0.00 0.09 (± 0.02) 0.05 (± 0.02) 0.01 (± 0.004) 0.01 0.02 0.03 n.s. 0.01 n.s. Glycine 0.02 0.50 (± 0.09) 0.58 (± 0.13) 0.41 (± 0.09) 0.003 0.008 0.008 n.s. n.s. n.s. 5-amino 0.00 1.06 (± 0.17) 1.04 (± 0.18) 0.50 (± 0.12) 0.001 0.002 0.01 n.s. n.s. n.s. pentanoate 3-phenyl 0.00 0.13 (± 0.04) 0.10 (± 0.04) 0.01 (± 0.01) 0.03 n.s. n.s. n.s. n.s. n.s. propionate Succinate 0.03 0.06 (± 0.02) 0.07 (± 0.03) 0.13 (± 0.01) n.s. n.s. 0.0008 n.s. n.s. n.s. Acetate 0.01 4.70 (± 0.42) 4.19 (± 0.42) 1.85 (± 0.34) < 10-5 0.0002 0.003 n.s. 0.0004 0.002 Butyrate 0.00 0.62 (± 0.10) 0.47 (± 0.10) 0.06 (± 0.04) 0.002 0.007 n.s. n.s. 0.001 0.01 Propionate 0.00 2.33 (± 0.10) 1.97 (± 0.10) 0.37 (± 0.10) 0.001 0.002 n.s. n.s. 0.0006 0.003 Leucine 0.02 0.10 (± 0.02) 0.12 (± 0.04) 0.07 (± 0.02) 0.02 0.05 0.02 n.s. n.s. n.s.

143

Table 4.2: Summary of the concentrations of metabolites consumed and generated following 24 h aerobic incubation of parotid saliva inoculated with oral bacteria relative to inoculation with sterile PBS. Significant results (p < 0.05) are presented in bold. NA = statistical test could not be conducted due to absence of metabolite from all samples yielding a S.D. of zero; n.s. = not significant. FFP = fungiform papillae, CVP = circumvallate papillae. Aerobic incubation Metabolite Baseline Mean (±SEM) inoculated sample conc. (mM), (n = 6) P-value between bacterial and P-value between bacteria PBS- control inoculum (one-sample t- inoculated samples inoculum test). (Tukey’s). control conc. FFP biofilm CVP biofilm WMS CVP v. FFP v. WMS v. CVP v. CVP v. FFP v. (mM), baseline baseline baseline FFP WMS WMS (n = 1) Endogenous parotid saliva metabolites consumed Urea 0.072 0.00 (± 0.00) 0.00 (± 0.00) 0.046 (± 0.009) NA NA 0.036 n.s. < 10-4 < 10-4 Lactate 0.18 0.02 (± 0.005) 0.02 (± 0.005) 0.10 (± 0.007) < 10-5 < 10-5 < 10-5 n.s. < 10-5 < 10-5 Citrate 0.073 0.00 (± 0.0) 0.006 (± 0.006) 0.048 (± 0.003) NA < 10-5 0.0005 n.s. < 10-5 < 10-5 Pyruvate 0.06 0.01 (± 0.002) 0.02 (± 0.004) 0.06 (± 0.007) < 10-5 0.0002 n.s. n.s. < 10-5 < 10-5 Glucose 0.12 0.00 (± 0.0) 0.00 (± 0.0) 0.00 (± 0.0) NA NA NA n.s. n.s. n.s. Metabolites generated Formate 0.004 0.65 (± 0.12) 0.60 (± 0.07) 0.36 (± 0.07) 0.003 0.0004 0.005 n.s. n.s. n.s. Phenylalanine 0.007 0.072 (± 0.02) 0.082 (± 0.03) 0.031 (± 0.008) 0.009 n.s. 0.018 n.s. n.s. n.s. Phenol 0.00 0.15 (± 0.02) 0.13 (± 0.01) 0.03 (± 0.02) 0.0005 0.0003 n.s. n.s. 0.001 0.006

144

Proline 0.00 0.36 (± 0.06) 0.45 (± 0.08) 0.30 (± 0.05) 0.001 0.002 0.001 n.s. n.s. n.s. Valine 0.00 0.17 (± 0.03) 0.16 (± 0.04) 0.08 (± 0.03) 0.004 0.01 0.04 n.s. n.s. n.s. Phenylacetate 0.00 0.09 (± 0.03) 0.06 (± 0.04) 0.01 (± 0.03) 0.02 0.04 0.03 n.s. 0.02 n.s. Glycine 0.02 0.43 (± 0.08) 0.47 (± 0.11) 0.28 (± 0.03) 0.003 0.008 0.0005 n.s. n.s. n.s. 5- 0.00 1.00 (± 0.09) 0.88 (± 0.11) 0.42 (± 0.09) 0.0001 0.0006 0.005 n.s. 0.002 0.01 aminopentanoate 3- 0.00 0.12 (± 0.04) 0.10 (± 0.05) 0.01 (± 0.01) 0.02 n.s. n.s. n.s. n.s. n.s. phenylpropionate Succinate 0.04 0.05 (± 0.01) 0.09 (± 0.02) 0.09 (± 0.01) n.s. n.s. 0.004 n.s. n.s. n.s. Acetate 0.01 4.84 (± 0.31) 4.19 (± 0.55) 1.58 (± 0.28) < 10-5 0.0006 0.002 n.s. 0.0001 0.0009 Butyrate 0.00 0.50 (± 0.08) 0.38 (± 0.10) 0.06 (± 0.03) 0.001 0.02 n.s. n.s. 0.003 0.03 Propionate 0.00 2.30 (± 0.29) 1.83 (± 0.43) 0.29 (± 0.14) 0.0005 0.008 n.s. n.s. 0.001 0.008 Leucine 0.02 0.10 (± 0.02) 0.10 (± 0.03) 0.05 (± 0.01) 0.02 n.s. 0.03 n.s. n.s. n.s.

145

Figure 4.12: PCA plot and k-means cluster analysis of all incubated and non-incubated control sample metabolite profiles. Control samples were clustered together. There was reasonable separation between tongue biofilm inoculated samples and WMS inoculated samples, with only WMS inoculated samples from one participant being clustered with tongue biofilm inoculated samples. No distinction was noted between FFP and CVP biofilm inoculums, nor was any separation based on anaerobic or aerobic incubation conditions.

146

Figure 4.13: PCA plots of sample metabolite profiles separated by inoculum type. No evidence of separation between aerobic and anaerobic incubation conditions is observed however there is a tendency for samples from the donor to group together. This suggests that sample donor is a more important factor than subsequent incubation conditions in shaping the final metabolite profile.

147

Figure 4.14: A summary of the significant (p < 0.05) correlations between protein consumption and change in metabolite concentration under both aerobic and anaerobic incubation conditions. Negative correlation indicates consumption of metabolites whereas positive correlation indicates production of metabolites.

148

4.5 Discussion

The work presented in this chapter answers several important questions about the metabolic composition of saliva and the role of oral bacteria in shaping the salivary metabolome. It is apparent that the majority of metabolites detected in WMS by 1H-NMR spectroscopy are absent from parotid saliva and are products of oral microbes. Furthermore, the substrate for metabolite production, at least in part, includes salivary proteins. Importantly, this study looked at the oral microbial metabolism of parotid saliva and found simultaneous protein degradation and metabolite production. In the present study, salivary proteins appeared to be consumed non-discriminately from parotid saliva, with only trace amounts of residual amylase typically remaining after the 24 hour incubation. Furthermore, the host derived metabolites demonstrated in parotid saliva (notably urea, lactate and citrate) were fully consumed by oral bacteria in-vitro. This suggests that host-derived metabolites can also provide a substrate for oral bacteria, alongside proteins.

A recent study of salivary metabolites has yielded some similar findings to the published and unpublished results from this chapter. Liebsch et al. (2019) 67, conducted a mass spectrometry-based analysis of the salivary metabolome in association with analysis of multiple clinical indicators of oral disease. Their findings support the inverse relationship between salivary urea and oral bacteria (as measured by dental plaque indices). Furthermore, they found strong positive associations between metabolites phenylacetate and 3- phenylpriopionate and cumulative periodontal pocket depths > 4mm, which they suggest could be useful biomarkers of periodontal disease. The present in-vitro study found that phenylacetate was produced at significant concentrations following inoculation of parotid saliva with WMS and tongue biofilm samples under aerobic and anaerobic conditions. 3- phenylpropionate was produced at significant concentrations only following inoculation from CVP tongue biofilm, although was observed in samples inoculated with FFP tongue biofilm.

Acetate butyrate and propionate were also produced in high concentrations from tongue- biofilm inoculated parotid saliva, although these SCFAs were not measured by Liebsch et al. (2019) 67.Unlike the SCFAs, phenylacetate and 3-phenylpropionate are not observed in the saliva of healthy individuals. Detecting phenylacetate and 3-phenylpropionate in this experiment suggests that biofilm from healthy individuals is capable of generating these metabolites under the correct ecological conditions. Further work is therefore required to understand the relationship between oral microbial communities, their net metabolism and their local ecology in order to identify new diagnostic and therapeutic avenues. Additional

149 limitations of the in-vitro work presented that merit future study include the time scale and static nature of the experiment. It would be interesting to understand the dynamics of both salivary protein breakdown as well as salivary metabolite generation. Even during sleep, when salivary flow is minimal, a degree of flux will occur in the oral cavity. Therefore, an experimental design with a slow replenishment of sterile saliva balanced with removal of “waste” saliva, measured over multiple timepoints, would more accurately allow mapping of how the salivary metabolome reaches an equilibrium.

Having demonstrated the production of metabolites from sterile parotid saliva via protein catabolism, the secondary aim of this in-vitro experiment was to determine whether oral bacteria from different oral sites produced different metabolite profiles. The metabolic profile of parotid saliva inoculated with fungiform papillae or circumvallate papillae biofilm was not detectably different. No statistical difference in the concentrations of metabolites produced or consumed by either tongue biofilm inoculum was detected for any metabolite measured. Several metabolites were generated at higher concentrations from tongue biofilm compared to WMS inoculums, notably acetate, propionate and butyrate and phenol. Phenylacetate was significantly higher following CVP biofilm inoculation compared to WMS inoculation. 5- aminopentanate was significantly higher following tongue biofilm inoculation compared to WMS inoculation under aerobic incubation only.

These differences in metabolite production may reflect greater bacterial load from tongue biofilm compared to WMS, although final CFU/ml was only significantly different between CVP biofilm and WMS under anaerobic incubation. The biofilm structure of the tongue inoculums compared to the planktonic nature of WMS bacteria may therefore have been an important source of enhanced metabolic activity. It is probable that a study with increased numbers of participants as well as more detailed measurements of the microbial communities within inoculums could reveal differences in the metabolic activity of CVP and FFP biofilms. Both the microbial composition of tongue biofilms as well as salivary concentrations of acetate and butyrate have been implicated in the perception of fat 177 243, and recently Neyraud & Morzel (2019) 246, have described the likely significance of oral biofilms in modulating taste perception. A study from the same research group reported that tongue biofilm yielded significantly higher bacterial CFU density compared to WMS. Metabolic differences were also detected with tongue film having more acetate and propionate and less lactate that WMS 105. Our findings were somewhat similar as we found WMS inoculum did not consume lactate as extensively as tongue film inoculum, and tongue biofilm generated more SCFAs. This reinforces the concept that the oral cavity does not necessarily contain fluid of homogenous 150 composition, but there may be local concentrations of metabolites at different oral sites. Localised concentration of metabolites generated by biofilms adjacent to oral chemoreceptors, such as the SCFAs observed in this in-vitro study, may represent a mechanism for microbial modulation of taste sensitivity.

Neither the biofluid comparison study nor in-vitro experiment found evidence of bacterial production of salivary taurine. Taurine is an important metabolite, involved in a range of physiological functions, and has been identified as a salivary biomarker of various diseases (see Table 1.2, Chapter 1). The experiments described to investigate the source of taurine in WMS collectively showed taurine is not of glandular origin, tissue origin (epithelial cells), nor is it passively derived from circulating plasma. Taurine was highly concentrated in GCF relative to WMS, suggesting GCF is the primary source of taurine in WMS. While GCF is plasma derived, it contains abundant immune cells which are known to be rich in taurine 247, and this cellular exudate likely contributes to taurine concentration in GCF. Proteomic study of GCF has revealed useful information particularly about its role in mediating immune response, and preliminary metabolomic study of GCF has been conducted 248 249. Despite the low sample volumes typically obtained, 1H-NMR has been conducted on diluted GCF. Therefore, the entrance of immune-cell derived taurine and potentially other metabolites into WMS via GCF represents a further mechanism by which oral microbiome and host immune response can shape the WMS metabolome.

151

Chapter 5: Analysis of the effects of taste stimulation on salivary composition and extensional rheology

5.1 Introduction

Taste perception is one of the most potent physiological stimuli for eliciting reflex salivary secretion. Despite response to pathophysiological stimuli being a key element in the metabonomic study of an organism or system 250, taste stimulation has not been studied in great detail by salivary metabolic profiling. The most common salivary stimuli are citric acid and mastication. Both stimuli have been studied with respect to metabolic composition using 1H-NMR and MS based platforms. Takeda et al., (2009) 81, looked at the metabolite content of saliva pre- and post-stimulation with citric acid. It was noted that most metabolites decreased upon stimulation, presumably due to higher fluid output, however there was not an equal degree of dilution for all metabolites. This indicates certain metabolite-specific changes following citric acid stimulation. Interestingly, the metabolites observed to decrease minimally included lactate and pyruvate, and urea was not significantly changed. In light of the work from Chapter 4 this suggests that host-derived metabolites enter saliva at different rates, possibly dependant on mechanism of entry (i.e. passive diffusion or active transport).

Unfortunately, as citric acid was used as a stimulus, no measurements of endogenous citrate could be made. The metabolite peaks that have been assigned to propylene glycol were also found to increase upon stimulation. Spiking experiments have shown these peaks do not correspond to propylene glycol (Chapter 3), although a correct assignment has yet to be made. Increased concentration upon stimulation, however, would suggest a host-derived molecule is a likely candidate.

Bader et al. (2018) 187, studied salivary composition by ion chromatography and mass spectrometry before and after a range of stimuli including citric acid. In accordance with Takeda et al., (2009) 81, a general trend for a dilution of analytes was observed. Exceptions to this were ions such as sodium, calcium and phosphate, which significantly increased. A critical finding of this study was that the dynamic changes following citric acid stimulation, particularly the increases in endogenous sodium concentration, subsequently diminished participant salt taste sensitivity. This serves to underline the complexity of the interaction between salivary and taste physiology and the importance of studying the effects of saliva both at rest and following stimulation.

152

The metabolic content of saliva following chewing stimulation has yielded interesting results when compared to those of citric acid stimulation. Neyraud et al., (2013) 83, found that chewing stimulated saliva generally had higher metabolite concentrations than unstimulated saliva. The exceptions were taurine and propionate. This finding was attributed to the physical action of chewing would mechanically disrupt plaque biofilm and cause greater release of metabolites. This would explain why taurine (shown in Chapter 4 to be GCF derived) would not be subject to this process. It is less clear why propionate would not also increase as it is also microbially derived. It is possible this could reflect site-specific metabolic differences in biofilm metabolism. The same study also found that total fatty acids in saliva also generally increased upon chewing stimulation. The authors proposed two explanations for this, either differences in glandular contributions to stimulated saliva or alterations of epithelial cell- derived . Okuma et al., (2017) 82, found a similar pattern of metabolite concentration increases following chewing stimulation, although they used mass spectrometry rather than NMR.

Beyond citric acid and chewing stimulation there are few metabolomic studies of other salivary stimuli. Bader et al., (2018) 187, looked at a range of other tastants including sodium chloride, monosodium glutamate, aspartame, iso-α-acids, hydroxy-α-sanshool, hydroxy-β- sanshool and 6-gingerol. Generally, there was a decrease in all salivary metabolites following stimulation with the exception of salivary ions. Mounayar et al., (2014) 85, compared saliva stimulated by oleic acid in two participant groups, those who were sensitive and insensitive to the stimulus. Importantly, significant differences in the salivary response to the stimulus were detected. Sensitive perceivers had increased levels of acetate, propionate, formate, GABA and lysine following stimulation, whereas insensitive perceivers had increased levels of glucose, , lactate and threonine. This study unveiled further complexity in taste-saliva interactions, by proving that salivary metabolite modulation can be dependent on participant sensitivity to the stimulus. The authors could not prove causality, although the role of microbial communities on oral tissues was postulated.

A further area in which there are few pre-existing studies is the effects of stimulation on salivary rheology. Additionally, how salivary physical changes post-stimulation might relate to other compositional changes is also poorly understood, especially regarding metabolites. Work of Vijay et al., (2015) 189, has shown that ions such as calcium and bicarbonate are associated with salivary extensional rheology. The authors also found differences between chewing and taste stimulation (0.1 M MSG), with the latter being a more potent stimulus for

153 extensional rheology. Recently, there has been emerging evidence that TRPV1 agonists such as capsaicin and nonivamide are potent stimuli for salivary extensional rheology 190,228. The aims of this chapter were to characterise changes in salivary metabolites, major proteins and extensional rheology following the basic tastes sucrose and caffeine (sweet and bitter) and TRP agonists menthol and capsaicin (TRPM8 and TRPV1, respectively). Relationships between the metabolite composition, protein composition and extensional rheology pre- and post-stimulation were assessed.

This chapter incorporates the publication “Endogenous salivary citrate is associated with enhanced rheological properties following oral capsaicin-stimulation” by Alexander Gardner, Po-Wah So and Guy Carpenter, as published in Journal of Experimental Physiology, 2019. All authors contributed to the study design, data analysis and interpretation. AG conducted the experiments. All authors contributed to and approved the manuscript. This manuscript contains the results of capsaicin-stimulation of saliva. Results for the other stimuli are presented separately.

5.2 Aim

The aim of this chapter was to characterise the changes in salivary flow rate, protein composition, metabolite composition and extensional rheology where stimulated by basic tastes (sucrose and caffeine) and TRP agonists (menthol and capsaicin). A key objective as part of this investigation was to look for relationships between metabolite and protein changes and changes in extensional rheology. Previous research has implicated salivary ions such as bicarbonate and calcium in salivary extensional rheology 189, although metabolomic study in this respect has not been conducted. Given many salivary metabolites are present as dissociated anions, these may be capable of interacting with proteins or ions directly or via corresponding cations (typically protons, leading to pH change), contributing to protein aggregation and altered extensional rheology.

5.3 Materials and methods

Experiments were conducted on ten healthy volunteers. The following food-grade tastant solutions were prepared as described; 0.25 M sucrose, 8 mM caffeine, 250 ppm menthol and 1 ppm capsaicin. Control solutions were water for the basic tastes, 0.475% ethanol for the menthol and 0.095% ethanol for capsaicin. The tastants sucrose and caffeine were selected as they are widely used in sensory science to represent sweet and bitter taste, respectively, and are used in International Standards Organisation protocols 157. Furthermore, they are widely

154 consumed flavourings and food additives and are therefore of significant interest to industry 251,252. Capsaicin is less researched as a tastant, however there is promising work that it can alter salivary flow rate and extensional rheology (as measured by spinnbarkeit) 190,235, therefore study of its effects on salivary protein and metabolite composition is merited. Menthol is the least studied of the four tastants investigated in this chapter and it was selected as it is a TRPM8 agonist producing a sensation of cooling which would complement the TRPV1 mediated warming sensation produced by capsaicin.

The first water control was administered (10 ml held passively in the mouth for 30 seconds and expectorated) then saliva was collected for two minutes. Extensional rheological measurements were conducted with the CaBER immediately following collection, and the participants mouth was given time to return to baseline. This process was repeated with the first tastant solution (0.25 M sucrose). The order of tastants were sucrose, caffeine, menthol and capsaicin, each preceded by their relevant control solution. Participants were blinded to the order.

Intra-oral salivary flow during tasting was measured by weighing the control/tastant solution before and after expectoration.

All samples were centrifuged at 15,000 g for ten minutes at 4 °C and, before storing at -80 °C, divided into two aliquots for SDS-PAGE and 1H-NMR spectroscopy as described in Chapter 2. The experimental design is summarised in Figure 5.1.

Figure 5.1: Schematic of the order and timing of control/tastant administration and sample collection and analysis. This process was repeated for all four tastant samples and their respective controls. The recovery time was typically fifteen minutes, as reported by participants. A minimum of ten minutes was required regardless of residual taste to allow for rheological analysis.

155

Targeted quantitative analysis of metabolites was conducted. Data were inspected for normality and paired comparisons of flow rate (intra-oral and post-expectoration), capillary breakup, extensional viscosity, metabolite concentrations, and major protein band abundance were undertaken. Metabolite and protein values were then adjusted for flow rate to assess output, using the following equation:

퐵푖표푓푖푙푚 표푢푡푝푢푡 (µ푚표푙/푚푖푛) = 푀푒푡푎푏표푙푖푡푒 푐표푛푐. (푚푚표푙/퐿) ∗ 푆푎푙푖푣푎푟푦 푓푙표푤 푟푎푡푒 (푚푙/푚푖푛)

5.4 Results

5.4.1 Intra-oral and post expectoration flow rates

Flow rate data following sucrose, caffeine and menthol are summarised in Figure 5.2. Sucrose caused significant increases in both intra-oral and post expectoration flow rate, caffeine significantly increased intra-oral flow rate but not during the two minute post-expectoration collection and menthol did not significantly affect flow rate.

Figure 5.2: Effects of tastants on salivary flow rates. a. summarises the intra-oral flow rate changes following sucrose, caffeine and menthol stimuli. The “negative” intra-oral flow rate

156 for the sucrose control is due to coating/absorption of fluid onto oral tissues causing retention in the mouth. b. illustrates data for flow rates two minutes post-expectoration. p-values are for two-tailed paired t-test (n=10), N.S. = not significant.

5.4.2 Salivary metabolite concentrations and protein abundance

Data showing the changes in protein abundance and metabolite concentrations relative to control are shown for sucrose, caffeine and menthol in Tables 5.1, 5.2 and 5.3, respectively.

Sucrose stimulation Total protein concentration and abundance of all major salivary proteins did not change following sucrose stimulation. Concentration of several metabolites were significantly increased, particularly those pertaining to sucrose catabolism and glycolysis, such as glucose, lactate, pyruvate and succinate. Other metabolite that can be synthesised from pyruvate including acetoin and alanine also significantly increased. Interestingly, citrate concentrations decreased as did methylamine. When considering outputs, a number of protein outputs were significantly increased. These were amylase, cystatin and MUC5B. Metabolite outputs were almost universally increased, the exceptions being the amines, urea, formate and citrate.

Caffeine stimulation Statherin was the only salivary protein which significantly changed, increasing upon caffeine stimulation. Protein outputs did not significantly change. A number of metabolite concentrations significantly reduced when comparing caffeine stimulation to control. These were all metabolites that significantly increased following sucrose stimulation, however, and these observations may simply represent a delayed return to baseline concentrations as opposed to a genuine consequence of caffeine stimulation. Salivary metabolite outputs displayed a similar pattern of results as for concentrations, which is unsurprising given the similarities in flow rates pre- and post-caffeine.

Menthol stimulation There were no significant changes in protein abundance or output following menthol stimulation. The majority of metabolites did not change in concentration, except lactate and glycine, which were reduced. There were no significant changes in metabolite output.

157

5.4.3 Extensional rheology A summary of the changes in extensional rheology following sucrose, caffeine and menthol is presented in Table 5.4. There were no significant rheological changes following sucrose, caffeine, or menthol. Caffeine was unique in that there appeared to be a trend for reduction in rheology following the stimulus, this approached significance for total capillary breakup time (p = 0.078).

158

Table 5.1: Changes in major salivary proteins and metabolite concentrations and outputs following 0.25 M sucrose. Proteins and metabolites are ranked highest to lowest by fold change in concentration relative to control stimulated levels. p-values are for two-tailed paired t-test (n=10), values in bold indicate significance (p < 0.05). *Total protein concentration is measured in mg/ml and protein output in mg/min. Taurine and glycine were not quantified due to residual sucrose resonances obscuring these peaks. Salivary Salivary concentrations Salivary outputs proteins Control (H2O) Sucrose 0.25 M Mean p-value Control (H2O) Sucrose 0.25 M Mean fold p-value (relative units/ml) (relative units/ml) fold (relative units/min) (relative units/min) change Mean S.D. Mean S.D. change Mean S.D. Mean S.D. Statherin 0.12 0.14 0.15 0.16 1.28 0.57 0.08 0.10 0.15 0.15 1.81 0.24 Cystatin 0.63 0.38 0.71 0.30 1.12 0.17 0.52 0.32 0.83 0.45 1.61 0.001 Amylase 1.74 0.63 1.76 0.48 1.01 0.9 1.47 0.75 1.99 0.65 1.35 0.004 glPRP 0.25 0.19 0.24 0.16 0.96 0.69 0.22 0.19 0.27 0.18 1.23 0.15 Total protein* 1.21 0.35 1.14 0.36 0.94 0.49 1.05 0.48 1.29 0.51 1.23 0.02 MUC5B 0.53 0.28 0.48 0.23 0.9 0.25 0.44 0.27 0.56 0.29 1.26 0.04 PRP 0.44 0.24 0.32 0.16 0.71 0.09 0.37 0.22 0.35 0.16 0.94 0.72 MUC7 0.12 0.12 0.08 0.06 0.65 0.07 0.08 0.06 0.08 0.05 1 0.98

Salivary Control (H2O) Sucrose 0.25 M Mean fold p-value Control (H2O) Sucrose 0.25 M Mean fold p-value metabolites (mmol) (mmol) change (µmol/min) (µmol/min) change Mean S.D. Mean S.D. Mean S.D. Mean S.D.

159

Glucose 0.10 0.086 2.05 1.80 21.54 0.007 0.07 0.07 2.58 2.44 36.72 0.009 Lactate 0.18 0.25 1.65 1.00 9.27 0.0003 0.17 0.31 2.16 1.84 12.78 0.003 Succinate 0.064 0.054 0.24 0.11 3.76 0.0002 0.06 0.06 0.3 0.19 5.25 0.001 Pyruvate 0.07 0.05 0.18 0.12 2.69 0.004 0.06 0.05 0.22 0.15 3.75 0.002 Acetoin 0.038 0.026 0.079 0.068 2.05 0.02 0.034 0.032 0.10 0.11 2.94 0.03 Alanine 0.07 0.05 0.094 0.073 1.37 0.04 0.06 0.05 0.11 0.09 1.86 0.005 Propionate 0.46 0.34 0.60 0.47 1.32 0.09 0.42 0.37 0.75 0.60 1.77 0.006 Choline 0.024 0.016 0.027 0.023 1.12 0.26 0.019 0.014 0.028 0.022 1.48 0.02 Urea 0.14 0.10 0.16 0.17 1.11 0.58 0.11 0.07 0.17 0.17 1.58 0.13 Histidine 0.01 0.006 0.01 0.007 1.04 0.65 0.008 0.005 0.012 0.008 1.44 0.002 Ethanol 0.077 0.049 0.078 0.048 1.01 0.97 0.07 0.05 0.09 0.07 1.41 0.03 Acetate 1.91 1.17 1.91 1.26 1 0.96 1.71 1.32 2.31 1.67 1.35 0.003 Tyrosine 0.021 0.012 0.02 0.01 0.99 0.89 0.018 0.013 0.025 0.018 1.36 0.005 Butyrate 0.16 0.08 0.15 0.08 0.94 0.25 0.14 0.08 0.17 0.10 1.28 0.01 Phenylalanine 0.024 0.012 0.022 0.012 0.93 0.35 0.02 0.01 0.026 0.016 1.35 0.009 Formate 0.076 0.096 0.062 0.053 0.83 0.43 0.072 0.112 0.078 0.091 1.09 0.53 Dimethylamine 0.018 0.016 0.015 0.016 0.82 0.1 0.015 0.013 0.016 0.015 1.07 0.53 Trimethylamine 0.006 0.004 0.004 0.006 0.75 0.27 0.005 0.004 0.005 0.006 0.9 0.68 Citrate 0.027 0.019 0.017 0.016 0.63 0.023 0.02 0.015 0.017 0.014 0.81 0.14 Methylamine 0.006 0.004 0.003 0.003 0.52 0.027 0.005 0.003 0.003 0.003 0.67 0.08

160

Table 5.2: Changes in major salivary proteins and metabolite concentrations and outputs following 8 mM caffeine. Proteins and metabolites are ranked highest to lowest by fold change relative to control stimulated levels. *Total protein concentration is measured in mg/ml and protein output in mg/min. p-values are for two-tailed paired t-test (n=10), values in bold indicate significance (p < 0.05). Salivary Salivary concentrations Salivary outputs proteins Control (H2O) Caffeine 8 mM Mean fold p-value Control (H2O) Caffeine 8 mM Mean fold p-value (relative units/ml) (relative units/ml) change (relative units/min) (relative units/min) change Mean S.D. Mean S.D. Mean S.D. Mean S.D. Statherin 0.08 0.07 0.11 0.09 1.38 0.025 0.08 0.04 0.11 0.08 1.44 0.07 Amylase 1.41 0.41 1.54 0.47 1.09 0.23 1.51 0.58 1.57 0.38 1.04 0.62 Total protein* 1.06 0.38 1.12 0.28 1.06 0.5 1.11 0.42 1.16 0.29 1.04 0.48 MUC5B 0.42 0.21 0.42 0.21 1.0 1.0 0.45 0.21 0.44 0.23 0.98 0.71 Cystatin 0.60 0.38 0.58 0.32 0.97 0.8 0.62 0.38 0.61 0.33 0.98 0.83 glPRP 0.21 0.17 0.2 0.15 0.96 0.73 0.23 0.18 0.22 0.16 0.96 0.69 PRP 0.26 0.18 0.24 0.12 0.95 0.71 0.27 0.18 0.26 0.13 0.97 0.79 MUC7 0.09 0.10 0.06 0.05 0.7 0.28 0.09 0.11 0.07 0.05 0.75 0.36

Salivary Control (H2O) Caffeine 8 mM Mean fold p-value Control (H2O) Caffeine 8 mM Mean fold p-value metabolites (mmol) (mmol) change (µmol/min) (µmol/min) change Mean S.D. Mean S.D. Mean S.D. Mean S.D. Formate 0.09 0.067 0.10 0.12 1.17 0.50 0.09 0.07 0.11 0.16 1.22 0.54 Choline 0.018 0.012 0.019 0.017 1.04 0.71 0.018 0.011 0.019 0.017 1.03 0.83 161

Dimethylamine 0.0129 0.0094 0.013 0.0108 1.01 0.9 0.0132 0.0092 0.0135 0.011 1.03 0.79 Histidine 0.01 0.004 0.01 0.004 1.00 0.95 0.01 0.004 0.01 0.004 1.01 0.94 Tyrosine 0.02 0.009 0.02 0.012 0.99 0.81 0.022 0.010 0.021 0.014 0.97 0.74 Urea 0.182 0.141 0.18 0.14 0.99 0.93 0.19 0.13 0.18 0.11 0.94 0.67 Ethanol 0.09 0.03 0.09 0.05 0.98 0.82 0.10 0.043 0.09 0.06 0.97 0.76 Glycine 0.068 0.059 0.066 0.065 0.97 0.69 0.07 0.064 0.07 0.08 1.02 0.84 Acetate 1.94 1.01 1.86 1.28 0.96 0.50 2.13 1.27 2.01 1.61 0.94 0.64 Phenylalanine 0.024 0.006 0.022 0.008 0.93 0.18 0.026 0.01 0.024 0.010 0.92 0.4 Butyrate 0.14 0.053 0.13 0.063 0.93 0.25 0.15 0.061 0.13 0.07 0.92 0.34 Trimethylamine 0.0035 0.0044 0.0031 0.0035 0.89 0.33 0.0037 0.0045 0.0031 0.0035 0.83 0.17 Citrate 0.029 0.015 0.025 0.017 0.88 0.12 0.029 0.013 0.026 0.015 0.88 0.27 Alanine 0.068 0.045 0.06 0.047 0.88 0.008 0.072 0.049 0.064 0.053 0.89 0.12 Methylamine 0.0029 0.0037 0.0026 0.0035 0.87 0.27 0.0033 0.0041 0.0026 0.0035 0.79 0.23 Taurine 0.08 0.05 0.06 0.04 0.81 0.16 0.08 0.043 0.06 0.043 0.74 0.12 Propionate 0.67 0.48 0.49 0.38 0.73 0.007 0.74 0.50 0.51 0.42 0.7 0.022 Succinate 0.09 0.07 0.06 0.049 0.67 0.07 0.09 0.08 0.06 0.06 0.67 0.07 Pyruvate 0.09 0.06 0.06 0.046 0.65 0.003 0.10 0.07 0.06 0.053 0.64 0.002 Acetoin 0.061 0.052 0.032 0.025 0.54 0.013 0.064 0.053 0.035 0.03 0.54 0.004 Glucose 0.14 0.14 0.04 0.07 0.30 0.014 0.13 0.125 0.04 0.07 0.31 0.016 Lactate 0.72 0.731 0.16 0.17 0.23 0.013 0.75 0.74 0.17 0.20 0.23 0.01

162

Table 5.3: Changes in major salivary proteins and metabolite concentrations and outputs following 250 ppm menthol. Proteins and metabolites are ranked highest to lowest by fold change relative to control stimulated levels. p-values are for two-tailed paired t-test (n=10), values in bold indicate significance (p < 0.05). *Total protein concentration is measured in mg/ml and protein output in mg/min. Ethanol was not quantified as the exogenous ethanol in the control and tastant solutions would prevent accurate measurements. Butyrate was not quantified as residual menthol obscured butyrate resonances. Salivary Salivary concentrations Salivary outputs proteins Control (Ethanol Menthol 250 ppm Mean p-value Control (Ethanol Menthol Mean p-value 0.0475%) (relative units/ml) fold 0.0475%) 250 ppm fold (relative units/ml) change (relative units/min) (relative units/min) change Mean S.D. Mean S.D. Mean S.D. Mean S.D. MUC7 0.07 0.14 1.97 0.27 0.08 0.18 2.29 0.30 Statherin 0.11 0.17 1.55 0.18 0.11 0.19 1.70 0.13 Cystatin 0.60 0.65 1.09 0.38 0.64 0.75 1.16 0.14 MUC5B 0.41 0.44 1.08 0.44 0.43 0.51 1.16 0.13 Amylase 1.53 1.55 1.01 0.84 1.55 1.71 1.10 0.24 Total protein* 1.21 1.11 0.92 0.40 1.29 1.27 0.99 0.92 PRP 0.29 0.26 0.89 0.49 0.31 0.29 0.92 0.48 glPRP 0.23 0.20 0.89 0.56 0.25 0.23 0.94 0.64

Salivary Control (Ethanol Menthol 250 ppm Mean p-value Control (Ethanol Menthol 250 ppm Mean p-value metabolites 0.0475%) (mmol) fold 0.0475%) (µmol/min) fold (mmol) change (µmol/min) change

163

Mean S.D. Mean S.D. Mean S.D. Mean S.D. Citrate 0.023 0.015 0.025 0.017 1.10 0.29 0.023 0.013 0.028 0.018 1.25 0.13 Phenylalanine 0.027 0.016 0.027 0.015 0.99 0.76 0.029 0.018 0.032 0.020 1.13 0.07 Histidine 0.008 0.005 0.008 0.004 0.96 0.55 0.009 0.005 0.009 0.005 1.09 0.17 Tyrosine 0.021 0.013 0.019 0.012 0.93 0.09 0.022 0.017 0.024 0.017 1.05 0.25 Acetoin 0.037 0.023 0.034 0.020 0.93 0.47 0.039 0.026 0.041 0.028 1.07 0.44 Urea 0.17 0.11 0.16 0.07 0.93 0.59 0.149 0.103 0.158 0.092 1.06 0.66 Choline 0.019 0.016 0.017 0.014 0.92 0.16 0.018 0.015 0.020 0.017 1.07 0.24 Alanine 0.062 0.052 0.057 0.038 0.91 0.42 0.066 0.056 0.068 0.052 1.04 0.53 Succinate 0.10 0.09 0.09 0.08 0.91 0.23 0.11 0.13 0.11 0.12 0.99 0.86 Glycine 0.08 0.08 0.07 0.08 0.87 0.046 0.09 0.12 0.09 0.12 0.98 0.61 Formate 0.05 0.06 0.04 0.055 0.86 0.07 0.06 0.10 0.05 0.09 0.92 0.38 Methylamine 0.003 0.004 0.0026 0.0036 0.85 0.25 0.003 0.004 0.003 0.004 1.01 0.92 Acetate 2.58 1.95 2.16 1.50 0.84 0.18 2.83 2.53 2.71 2.35 0.96 0.52 Pyruvate 0.07 0.05 0.06 0.036 0.82 0.11 0.08 0.055 0.07 0.05 0.94 0.51 Lactate 0.15 0.09 0.12 0.06 0.79 0.041 0.15 0.10 0.14 0.08 0.89 0.21 Dimethylamine 0.015 0.012 0.012 0.007 0.78 0.11 0.015 0.012 0.014 0.009 0.89 0.25 Taurine 0.06 0.046 0.04 0.03 0.75 0.07 0.06 0.05 0.05 0.04 0.88 0.14 Trimethylamine 0.004 0.005 0.003 0.004 0.75 0.09 0.005 0.005 0.004 0.005 0.88 0.15 Propionate 0.44 0.35 0.31 0.24 0.72 0.053 0.47 0.41 0.40 0.35 0.85 0.06

164

Table 5.4: Changes in salivary rheology following sucrose, caffeine and menthol, p-values are for two-tailed paired t-test, (n=10). Tastant Rheological parameter Control Stimulus Mean fold p-value Mean S.D. Mean S.D. change Sucrose Capillary breakup time (s) 1.48 1.14 1.71 1.42 1.16 0.69 0.25 M Apparent extensional viscosity at strain 9 (mPa.s) 17.50 18.96 30.25 46.46 1.73 0.40 Apparent extensional viscosity at strain 9.5 (mPa.s) 52.55 59.98 92.76 143.56 1.77 0.44 Apparent extensional viscosity at strain 10 (mPa.s) 103.82 116.90 175.21 297.91 1.69 0.51 Apparent extensional viscosity at strain 10.5 (mPa.s) 172.90 194.21 289.80 437.05 1.68 0.46 Caffeine Capillary breakup time (s) 1.56 1.27 0.92 1.11 0.59 0.08 8 mM Apparent extensional viscosity at strain 9 (mPa.s) 18.89 25.18 15.21 25.80 0.81 0.56 Apparent extensional viscosity at strain 9.5 (mPa.s) 90.84 116.34 35.56 50.47 0.39 0.15 Apparent extensional viscosity at strain 10 (mPa.s) 209.81 326.41 79.65 117.45 0.38 0.26 Apparent extensional viscosity at strain 10.5 (mPa.s) 429.15 777.57 140.77 249.16 0.33 0.29 Menthol Capillary breakup time (s) 1.35 1.34 1.91 2.10 1.41 0.34 250 ppm Apparent extensional viscosity at strain 9 (mPa.s) 10.00 10.11 26.92 41.89 2.69 0.21 Apparent extensional viscosity at strain 9.5 (mPa.s) 33.67 48.39 74.22 111.99 2.20 0.16 Apparent extensional viscosity at strain 10 (mPa.s) 61.71 91.28 151.29 248.78 2.45 0.14 Apparent extensional viscosity at strain 10.5 (mPa.s) 96.50 136.52 259.71 424.44 2.69 0.14

165

5.4.4 Endogenous salivary citrate is associated with enhanced rheological proerties following capsaicin stimulation

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

5.5 Discussion

The findings of this study indicate that there at least two mechanisms by which salivary metabolite composition can be altered by tastants. In the case of sucrose, metabolite changes are via intra-oral catabolism of the tastant itself. In the case of capsaicin, changes are via increase in endogenous metabolite (citrate) production from glands. Although the breakdown of sucrose by oral bacteria has been known for decades 253, literature studying this process by an in-vivo metabolomics approach remains sparse. Takahashi & Washio (2011) 104, profiled glycolysis pathway and citric acid cycle intermediates in dental plaque samples following glucose rinses using MS. While this approach was useful in detecting many intermediates that would not be observed by NMR due to their low concentration, drawbacks such as the inability to measure formate and acetate due to low mass-to-charge ratio occurred 94. Our data found that sucrose exposure increased the concentrations of several metabolites including glucose, pyruvate and succinate. Additionally, salivary metabolites such as alanine and acetoin were found to increase in concentration. Alanine is synthesised from pyruvate via the action of alanine aminotransferase (alanine transaminase) 254. It is unclear which oral species would be predominantly involved in this process. For example, in E. coli models, multiple alanine transaminases have been detected 255,256. Additionally, host-derived alanine transaminase is present in saliva 257. Similarly, there are few definitive studies of oral bacteria that produce acetoin. S. mutans strains have been shown to produce acetoin from pyruvate under aerobic but not anaerobic conditions, although lactate dehydrogenase deficient strains persisted in anaerobic acetoin production 258.

Inter-individual differences in the intra-oral catabolism of sucrose may have significant bearing on taste perception. Sucrose is overconsumed in the UK, with added sugar contributing to 13% of average daily calorie intake 259. As such, sucrose is widely used as a sweet tastant when studying taste perception. Sucrose and its constituent , glucose and fructose, have different sweet intensities at the same molarity 260. The same sucrose solution may rapidly become a different mixture of sucrose, glucose and fructose in the mouths of different participants, based on variations in the generation and consumption of glucose and fructose. Measurement of salivary fructose following sucrose by 1H-NMR is difficult as there is considerable spectral overlap with the residual sucrose peaks. Fructose is metabolised differently to glucose in humans as fructose is mostly metabolised in the whereas glucose is widely used by different tissues 261. Whether oral bacteria prefer glucose or fructose is unclear as intra-oral microbial glycolysis and has not yet been studied by metabolomics. There is in-vitro evidence from a S. mitis model that fructose is less

189 efficiently metabolised that glucose, although these findings cannot be extrapolated to an in- vivo system 262. Other work with S. mutans has suggested that glucose and fructose are consumed concurrently, although there is a preference for glucose over other monosaccharides 263. Sucrose also appears not to be the only tastant that is catabolised by oral bacteria. The work of Bader et al., (2018) 187, suggests a similar effect following oral exposure to the non-caloric sweetener aspartame. Salivary concentrations of phenylalanine and aspartic acid (the constituent amino acids of aspartame) increase following exposure, whereas concentrations of all other amino acids decrease 187.

The results of caffeine and menthol stimulation were minimal in this study. Following caffeine, a number of metabolites were found to significantly decrease in concentration. It should be noted however, that with the exception of propionate, these metabolites were all significantly raised following the sucrose solution. It therefore appears more likely that this observation simply represents a delayed return to baseline for these metabolites. While time was given in between tastes for participants mouths to return to normal, there may have been delayed metabolic events that extended beyond this period. The rinsing protocol described in Chapter 3 was not employed, meaning the only source of oral clearance would be saliva replenishment. The decrease in lactate following menthol stimulation may similarly reflect a delayed return of lactate to baseline. The fall in glycine should also be interpreted with caution as this was a relatively small fold change close to the limit of significance and could represent a false positive.

The salivary changes following capsaicin stimulation were amongst the most striking. The changes in protein have previously been observed 231, as have changes in extensional rheology (spinnbarkeit) following nonivamide 190. With respect to metabolite content, this has not previously been assessed. This chapter showed that citrate was significantly increased following capsaicin stimulation. Citrate appears to be an interesting metabolite with regards to salivary stimulation. Neyraud et al., (2013) 83, showed that chewing stimulation increased salivary citrate concentrations. While the use of citric acid as a stimulus would preclude the assessment of change in endogenous citrate, our supplementary results from Chapter 4 also indicate that tartaric acid stimulation causes parotid citrate output to increase proportionately to fluid secretion. In contrast, the results from this chapter regarding sucrose stimulation found citrate was one of a few metabolites to significantly decrease in concentration. This is despite the fact that the salivary response as measured by flow rate was greater for sucrose than capsaicin, suggesting endogenous salivary citrate output is stimulus specific. This net decrease in citrate following sucrose exposure may represent a rapid shift in 190 the net metabolic pathways of the oral microbiome, with the majority of the pyruvate being produced undergoing conversion to lactate. Efficient microbial lactate producers such as S. mutans possess reduced citric acid cycles, although are capable of citrate uptake and utilisation by other metabolic pathways 264,265. The observed increase in succinate following sucrose may also reflect such a metabolic shift. With reduced citrate availability for oral microbes with complete citric acid cycles, build-up of oxaloacetate would cause a shift towards succinate output from the cycle 94. Alternatively, succinate is known to be synthesised by a variety of pathways by gut bacteria and is an intermediate in propionate production 97,266. These processes are summarised in Figure 5.3.

Figure 5.3: Summary of the oral microbial metabolic pathways following exposure to sucrose. The central role of pyruvate, which is common to multiple pathways is illustrated. PEP = phosphoenolpyruvate, OAA = oxaloacetate, MAL = malate, FUM = fumarate. Information is summarised from Takahashi, (2015) 97; Takahashi & Washio, (2010) 94; Ajdic et al., (2002) 264; Korithoski et al., (2005) 265 and Fernández-Valero & Vendrell, (2019) 266. 191

In summary, this chapter demonstrates that salivary changes following basic tastes and TRP agonists are stimulus specific. Caffeine and menthol did not greatly alter the metabolic composition of saliva, however sucrose and capsaicin did. Capsaicin caused significant increases in major salivary proteins, extensional rheology and citrate. Changes in citrate were also proportional to the magnitude of the rheological parameters. Endogenous citrate may, therefore, play a role in mediating the complex biochemical events surrounding mucin unfolding and assembly. The changes following sucrose were due to microbial catabolism of the tastant. Sucrose catabolism appeared to have a degree of interindividual variation, which might account for some of the interindividual differences in taste perception for the same sucrose stimulus.

192

Chapter 6: Determining the intra-individual variability of salivary metabolite and major protein profiles and the relationship with detection and recognition thresholds of basic tastes.

6.1 Introduction:

A large degree of day-to-day variation is reported when assessing individual suprathreshold response to taste 174. The intra-individual variations reported by Prutkin et al. (2000) 174, to PROP bitterness show ratings from the same participant fluctuating from “moderate” to “very strong” when rated on consecutive days. This degree of variability clearly cannot be explained by the same biological factors that are attributed to the explaining the large inter-group differences in PROP bitterness perception that are sometimes observed. For example, TAS2R38 haplotype is genetically determined and unchanging and fungiform papillae density, although declining slowly with age at a rate of 1-2 papillae per cm2 per year 176, does not change on a daily timescale 231. Therefore, it is a reasonable hypothesis that changes in the oral environment are a critical factor in the perception of tastes, even more so than genetic and anatomical factors such as papillae density and taste receptor genetics.

It is possible that the changes seen in an individual’s salivary composition and fluctuations in their taste perception may be related to each other, given that individual salivary composition is known to change between days and within the same day. This is true for ions such as phosphate and calcium and major proteins including amylase and histatins 18 267 268. While compositional changes of salivary content may in part be due to the circadian nature of salivary secretion, circadian effects are not the sole factor contributing to salivary compositional variation. Unstimulated saliva is known to show circadian rhythms for flow rate, protein composition and ions such as sodium and chloride 13. A recent metabolomic studies found that only around fifteen percent of salivary metabolites showed a circadian effect. Of these, the amino acids tyrosine and arginine appear to closely mirror the circadian cycle of total salivary protein demonstrated by Dawes in 1972, however the overall finding was that most salivary metabolites are free of circadian control 76.

Despite being a relatively new field, recent studies have contributed important data in the area of salivary metabolomic stability. Following the preliminary work of Dallman et al. (2012) 76, in investigating the circadian control of salivary metabolites the temporal stability of the salivary metabolome has been studied in reasonable detail by both 1H-NMR spectroscopy and

193

Mass Spectrometry metabolomic studies. Wallner-Liebmann et al. (2016) 79, found that the salivary metabolome as studied by 1H-NMR showed good intra- and inter-day stability across participants with intra-individual variation being less than inter-individua variation. The main difference in metabolite composition found was invariably between the first saliva collection immediately upon waking and those collected later throughout the day 79. Metabolites including alanine, methylamine, trimethylamine, valine and choline were found to be significantly more concentrated in the first saliva sample produced on waking. The authors did not speculate heavily on why these differences were present. Given our findings from Chapter 4 where methylamine, trimethylamine and choline were correlated with bacterial load and valine was produced by bacterial degradation of salivary proteins it would seem that reduced salivary flow and increased bacterial metabolism of salivary proteins would be a likely explanation.

Kawanishi et al. (2019) 77, conducted a non-targeted mass-spectrometry based study of the salivary metabolome. Their findings led to the conclusion that intra-day variation in salivary metabolome was not significant whereas inter-day variation was. Furthermore, they investigated the effects of stimulation and found that stimulated saliva showed reduced metabolic variability. An important limitation of the current studies of salivary metabolomic stability is in the analytical techniques employed. These studies have focused on the net metabolite profile of the samples using multivariate techniques such as PCA and measurement of Euclidean distance between projected data points. An important factor to address in this chapter, therefore, was measurement of the variability of individual salivary metabolites by targeted quantification. Such knowledge is particularly important, for example, in determining the natural variability of potential metabolite biomarkers to define the expected normal range in health (Table 1.2).

The protein composition of saliva is another variable feature that may contribute to taste perception. Studies focused on proteomic analysis of saliva and the stability of specific proteins have suggested that while intra-individual variation in protein profile is lower than inter-individual variation a large proportion of salivary proteins will typically display fluctuations in abundance of approximately two fold 269 40. We found that certain salivary proteins (PRP) are associated with the oral perception of tea astringency (see chapter 7) and there is literature evidence that salivary cystatin may be associated with the perception of caffeine bitterness 208. A further potential mechanism by which salivary proteins could be involved in taste perception, particularly in light of our data from Chapter 4, would be in their subsequent conversion into metabolites by oral bacteria. For these reasons, the stability of 194 the net salivary protein profile as well as individual abundance of major salivary proteins may be important in relation to taste perception.

Unlike other chapters, the sensory analysis for this chapter was conducted by threshold assessments. There were several justifications to explore this approach. Firstly, threshold assessment provides a somewhat more objective assessment of taste perception as participants have to correctly identify the taste sensation. Secondly, there may be differences in taste-saliva interactions when exposed to different concentrations of tastant. Additionally, as the maximum participant number was known before commencing the study, it was estimated that there may insufficient numbers to yield an appropriate separation of low and high tasters using suprathreshold measurements. Threshold measurements may therefore yield more definitive separation of sensitive and incentive tasters. Finally, as there is ongoing debate as to what sensory methodology is most appropriate, investigating different sensory methodologies in the novel study of salivary metabolomics and taste perception was deemed important 156. The salivary metabolites are invariably present at low millimolar or sub- millimolar concentrations. Metabolite effects on taste may be more noticeable during threshold determination as salivary metabolite concentrations are closer to tastant concentrations used in threshold measurement compared to concentrations used in suprathreshold assessment.

6.2 Aims and Objectives

The work presented in this chapter was designed to answer three primary questions: 1. Do the salivary analytes relate to tastant recognition and/or detection thresholds? 2. How variable from day-to-day and week-to-week are the net profiles and individual analyte concentrations of salivary metabolites and major proteins? 3. Is there any difference in salivary composition or stability between a trained taste panel and untrained participants that could serve as a biological marker for sensitive and consistent taste ability?

6.3 Methods

6.3.1 Participants

Participants in this study were 23 adults aged 30 to 70 years. Participants reported good general and oral health. The participant group was composed of 13 all-female trained panellists, who had previously been identified as consistent tasters and trained in the

195 description of tastes and oral sensations. The 13 trained taste panellists had all been working in a commercial sensory panel for at least two years and had previously undertaken the ISO method of threshold assessment at least twenty-five times. The remaining ten participants were 4 males and 6 females who were untrained panellists. The total participant number was known prior to commencing the study hence all analyses utilised the maximum number of participants.

6.3.2 Study design

This study was conducted over multiple days. Participants attended on three consecutive days, then two weeks later they attended for an additional three conductive days. All experiments were conducted at 10:00 a.m. on each day, and participants abstained from oral activity for at least one hour before commencing.

At each visit an unstimulated saliva sample was collected before any tasting. Recognition and detection thresholds were then obtained for either sweet, bitter or sour. A single taste modality was measured on each of the three days. The same tastes were measured over the second three-day session however the order of the three modalities was different.

6.3.3 Sensory methodology

Recognition and detection thresholds were measured following ISO standard 3972:2011 “Method of investigating sensitivity of taste”. Fresh solutions for each primary tastant were prepared on the morning of each experiment at a range of concentrations. Participants were presented with each solution in ascending order of concentration, rinsing in between with water. Participants indicated the first concentration at which the solution could be discriminated from water (recognition threshold). Participants then continued tasting the increasingly concentrated solutions and indicated the first concentration at which the solution could be identified. Participants were blind to the taste investigated on each day. The order of taste thresholds investigated is summarised in Table 6.1.

196

Table 6.1 – Summary of the order of taste threshold tests conducted in this study. Participants were blind to the tastants in every session. Week Day Taste threshold investigated One One Bitter (Caffeine) One Two Sour (Citric acid) One Three Sweet (Sucrose) Two One Sweet (Sucrose) Two Two Bitter (Caffeine) Two Three Sour (Citric acid)

ISO standards use citric acid as a sour tastant however the hydration is not specified in the ISO protocol, therefore this study used citric acid monohydrate. The initial ISO recommended concentrations for sour were found to be excessively high on the first week, therefore on the second week these were reduced four-fold. Concentrations of each tastant are summarised in Table 6.2, below.

Table 6.2 Summary of the taste concentrations used in threshold determination for this study, adapted from ISO 3972:2011. Dilution Week one and Week one and two Week one sour Week two sour two sweet bitter threshold threshold threshold threshold (caffeine conc.) (citric acid conc.) (citric acid (sucrose conc.) conc.) g/L mmol/L g/L mmol/L g/L mmol/L g/L mmol/L D1 12 35.06 0.27 1.39 0.6 2.86 0.15 0.71 D2 7.2 21.03 0.22 1.13 0.48 2.28 0.12 0.57 D3 4.32 12.62 0.17 0.88 0.38 1.81 0.10 0.45 D4 2.59 7.57 0.14 0.72 0.31 1.48 0.08 0.37 D5 1.56 4.56 0.11 0.57 0.25 1.19 0.06 0.30 D6 0.94 2.75 0.09 0.46 0.2 0.95 0.05 0.24 D7 0.55 1.61 0.07 0.36 0.16 0.76 0.04 0.19 D8 0.34 0.99 0.06 0.31 0.13 0.62 0.03 0.15

6.3.4 Saliva collection, storage and analysis

197

Prior to tasting, unstimulated saliva was collected into pre-weighed, sterile universal tubes. Tubes were re-weighed and stored at -20 °C. Frozen samples were transported on ice to -80 °C storage within 16 days of collection of the first batch of samples. Samples were then thawed on ice and prepared for SDS-PAGE and 1H-NMR analysis.

Samples were electrophoresed, stained, destained and imaged as described in Chapter 2. The NMR analysis was conducted in 5mm external diameter NMR tubes with internal NMR buffer prepared as outlined in Chapter 3. Samples were analysed in a single batch with intra-batch control samples composed of pooled 20 µl from all samples dispersed evenly throughout the run. CPMG spectra were acquired. Spectra were corrected by automatic phase and baseline corrections with manual adjustments. Imaged gels were digitised, and overall protein profile was analysed by PCA in Knime. Average Euclidean distance of relevant data points was conducted. Individual band intensity was also measured for MUC5B, MUC7, glPRP, amylase, PRP bands 1 and 2, cystatin and statherin. NMR spectra were bucketed into 0.01 ppm buckets from 0.7 to 8.5 ppm (excluding the water peak from 4.5 to 5.5 ppm). Bucketed spectra were subject to PCA and Euclidean distance measurements. Individual metabolites were also quantified.

6.3.5 Statistical analyses

Recognition and detection thresholds of trained panellists compared were compared to untrained pannelists by Mann-Whitney U test.

The intra-individual variability of salivary flow rate was assessed by comparing the mean relative standard deviation of all participant’s flow rates to the relative standard deviation of the entire data set by one-sample t-test.

The inter- and intra-individual stability of salivary protein profiles were compared. Intra- individual variability was calculated by the mean Euclidean distance between PCA points of all samples from the same individual. An estimate of total mean Euclidean distance of all data points in the sample determined inter-individual variability. As there would theoretically be over 6500 measurements between all data points, mean inter-individual variability was estimated based on 696 measurements. An assessment of the convergence on the true mean was conducted to verify an adequate sample of data points. Inter-individual variability was measured with and without inter-gel standard corrections. Comparisons were made by ANOVA. 198

Intra-week versus inter-week salivary protein stability was performed by comparison of mean Euclidean distance of individual samples collected within the same week versus samples collected on different week (illustrated in Figure 6.1). Means were compared by two-tailed t- test.

Figure 6.1: Illustration of the Euclidean distance measurements made to assess inter- vs. intra- week stability of salivary protein and metabolite profiles. The blue circles represent projected PCA points of samples collected from one participant over six days – three in one week (numbered 1-3) and three two weeks later (numbered 4-6). Intra-week measurements are shown by the red lines (n = 6) and inter-week measurements are shown by the green lines (n = 9).

The individual stability of quantified salivary protein band was compared by calculating the average relative standard error for each band within each individual.

Comparisons of protein band abundance between trained panellists and untrained panellists was performed by two-tailed t-test of the measured band intensities. Participants were ranked by taste thresholds on each day and comparisons of protein abundance between sensitive (low threshold) and insensitive (high threshold) tasters was performed by two-tailed t-test.

Assessment of inter vs. intra-individual salivary metabolite profile stability and assessment of intra-week vs. inter-week stability was conducted as for protein profiles. Since metabolomic data was gathered in a single batch no correction was necessary, thus a two-tailed t-test was used to compare inter and intra-individual variability.

199

Individual metabolite peaks were quantified and compared between sensitive and insensitive tasters by paired t-test using the same rankings as for the individual protein bands. The same comparison was also made for spectral buckets corresponding to unassigned peaks.

Comparison of metabolite concentrations between trained panellists and untrained panellists was initially performed by two-tailed t-test of the measured band intensities. To account for biases in the sex demographic of the all-female trained panellists and mixed sex untrained panellists, age-matched female controls were also sought from the participants of Chapter 7. The analysis of salivary metabolite levels between the three groups was repeated by ANOVA with Tukey’s post-hoc test.

All PCA analysis and Euclidean distance measurements were performed in Knime. ANOVA, t- tests and Wilcoxon signed-rank assessments were performed in GraphPad Prism.

6.4 Results

6.4.1 Intra-individual salivary flow rate is consistent between days

Individual flow rates across all dates are presented in Figure 6.2. Visual inspection shows that intra-individual flow rate tends to be consistent. The largest range in flow rate within a participant is from 0.54 to 1.01 ml/min, an increase of 88%, whereas across all participants the range is from 0.24 to 1.85 ml/min, an increase of 660%. The mean relative standard deviation of flow rate within individual participants is compared to the mean relative standard deviation of all flow rates is shown in Figure 6.3. Intra-individual flow rate was significantly less variable than inter-individual salivary flow rate, as shown by one-sample t-test.

200

S u m m a r y o f a l l f l o w r a t e m e a s u r m e n t s 2 . 0

W e e k 1 d a y 1

W e e k 1 d a y 2 )

n W e e k 1 d a y 3 i

1 . 5 m

/ W e e k 2 d a y 3

g (

W e e k 2 d a y 2

e t

a W e e k 2 d a y 3

r

1 . 0

w

o

l

f

y

r

a

v i

l 0 . 5

a S

0 . 0

1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 1 1 1 1 1 1 1 1 1 1 2 2 2 2

P a r t i c i p a n t n u m b e r

Figure 6.2: Salivary flow rate of all samples collected over the course of this experiment. Missing bars indicate a session where a particular participant did not attend. Flow rate appears to be more consistent within the same individual than between individuals.

201

Figure 6.3: Intra- and inter-individual variation in salivary flow rate. 6.3 a. shows the relative standard deviation in salivary flow rate for all participants. The dotted line indicates the relative standard deviation of all samples. 6.3 b. shows a one-sample t-test of the average relative standard deviation of all participants against the relative standard deviation of all samples (dotted line).

6.4.2 Salivary protein composition is significantly more stable within than between individuals

PCA plots of major salivary protein profiles are shown in Figure 6.4. To aid visualisation the points for trained and untrained panellists are separated. The points from different participants are shown in different colours and the samples collected on different days are shown by a different point size. As can be seen, samples from the same individual tend to be well grouped. Intra-individual salivary protein profile was significantly lower than inter- individual variation (61 percent lower). Alignment of standard lanes in the measurement of total inter-individual variation across different gels made a small but significant reduction of 9 % in total variability, Figure 6.5 a. No statistical difference in variability between trained and untrained panellist profiles was detected, nor was any difference in stability of salivary major proteins within week compared to between weeks, Figure 6.5 b and c. Total inter-individual variability was based on 696 measurements between samples of unrelated individuals and this was more than sufficient for the net variability to converge on the true value, Figure 6.5 d.

Inter-individual stability of the major protein bands is presented in Figure 6.6. The intra individual variation of each individual major salivary protein was significantly lower than the inter-individual variation across all participants. Cystatin had the lowest intra and inter- individual variation whereas statherin had the highest intra- and inter-individual variation.

202

Figure 6.4: PCA plots of saliva major protein profiles of trained panellists (above) and untrained panellists (bottom). Samples have been split onto two backgrounds purely as a visual aid. Different sizes of points represent the day of sample collection.

203

Figure 6.5: Stability of major salivary protein profile. a – shows the significantly reduced intra- individual stability of salivary protein profiles compared to inter-individual stability. Columns B and C illustrate a small but significant effect of batch correction in reducing inter-gel sources of non-biological variability. 6.5 b. shows no statistical difference in stability of salivary protein profiles from samples collected within compared to between weeks. 6.5 c. shows no significant difference in stability of salivary protein profiles between trained and untrained panellists (n.s. = not significant). 6.5 d. illustrates the convergence of the sample estimation of inter-individual mean Euclidean distance based on number of inter-individual measurements taken. Good stability is present after around 200 measurements.

204

I n t e r a n d i n t r a - i n d i v i d u a l v a r i a t i o n i n

i n d i v i d u a l s a l i v a r y p r o t e i n a b u n d a n c e

2 0 0

y

t

i

d s

r 1 8 0 n

a - 4 e

d * * = p < 1 0

t 1 6 0

n

n

i

a

t d

s 1 4 0

n

e

a

v i

b 1 2 0

t

a

n

l

i e

e 1 0 0

t

r

o r

M 8 0

p * *

E

n

S

i

6 0 * * * * 

n * *

* *

o n i 4 0

t * * * * a

a * *

i e

v 2 0

M

e d 0

M U C 5 B M U C 7 g l P R P A m y l a s e P R P 1 P R P 2 C y s t a t i n S t a t h e r i n

P r o t e i n b a n d

Figure 6.6: Comparison of the inter- and intra-individual variation of specific major salivary proteins. Intra-individual variation is the mean ± SEM of the relative standard deviation of each participant’s protein band intensity (n=23). Inter-individual variation is indicated by the dotted lines above each column, representing the relative standard deviation of the full data set for that particular protein band. p-values are for one-sample t-test between mean relative standard deviation of each participant against total relative standard deviation of the entire sample.

6.4.3 Salivary metabolite composition shows comparable or greater relative stability than salivary major protein composition

PCA plots of salivary metabolite profiles are shown in Figure 6.7. To aid visualisation trained and untrained panellists have been displayed on separate backgrounds. The data from two outliers (one from each participant group) is not included to allow a closer view of the remaining participants which would otherwise appear too closely packed to discriminate properly. Data from the intra-batch controls is also included.

The relative intra- and inter-individual stability is shown in Figure 6.8 a. This is also presented alongside the batch corrected data for protein profiles in Figure 6.8 b., showing a comparable ratio of intra- to inter-individual variability for both salivary metabolite and protein profiles. No significant difference was noticed for the metabolite profile stability within and between weeks, (Figure 6.8c.). Figure 6.8 d. shows the comparison between metabolite profile stability

205 within trained panellists and untrained panellists. Although the Euclidean distance was lower for trained panellists (i.e. greater stability) this was not significant (p=0.1).

Data for the inter- and intra-individual stability of individual salivary metabolites is presented in Figure 6.9. As with individual protein bands, intra-individual variation is consistently significantly below inter-individual variation. By comparison to Figure 6.6, it can be seen that most individual metabolites have less intra-individual variation than most major proteins, however the inter-individual variation (relative standard deviation of the sample) is higher for most metabolite than for most proteins. These trends were not statistically significant, however. This comparison is summarised in Table 6.3. Half of all proteins had an intra- individual variation greater than 35 %, whereas all but two metabolites showed intra- individual variations below 35 %. Five of eight protein band displayed inter-individual variation below 70 %, whereas nine of thirteen metabolites had inter-individual variation greater than 70 %. The mean largest fold change in salivary analyte concentration in also included. As can be seen the majority of salivary analytes show a maximum fluctuation of around two-fold although this is higher for metabolites such as propionate, lactate and particularly formate. The proteins glPRP, MUC5B, MUC7 and statherin all show relatively high fold changes.

206

Figure 6.7: PCA plots of salivary metabolite profiles of trained panellists (above) and untrained panellists (bottom). Samples have been split onto two backgrounds purely as a visual aid. Different sizes of points represent the day of sample collection. Participants 1 and 17 are not shown on the plots as if their data was fit to scale the other participants data would be too compressed to discriminate easily. The red samples are from five intra-batch controls, prepared from equal volumes of all other samples.

207

Figure 6.8: Variability of salivary metabolite profiles. a. shows the intra-individual stability of salivary metabolite profiles is significantly lower than inter-individual variability (t-test). The variability of intra-batch controls (0.68 % of total variability) is included to show the excellent within-batch reproducibility of 1H-NMR spectroscopy. 6.8 b. shows the close similarity in relative intra-individual stability compared to inter-individual stability of protein and metabolite profiles. For both measurements intra-individual stability was 57% lower than inter-individual stability. 6.8 c. shows no statistical difference in stability of salivary metabolite profiles from samples collected within compared to between weeks. 6.8 d. shows no significant difference in stability of salivary metabolite profiles between trained and untrained panellists (n.s. = not significant).

208

I n t e r a n d i n t r a - i n d i v i d u a l v a r i a t i o n i n

i n d i v i d u a l s a l i v a r y m e t a b o l i t e c o n c e n t r a t i o n

2 0 0

n

o i

t 1 8 0

d

a

r

r

t

a n

d 1 6 0

e n

c - 4

a t

n * * = p < 1 0

1 4 0

s

o

c

e

v e

i 1 2 0

t

t

i l

a * *

l o

e 1 0 0

b

r

a

t M

e 8 0

E

m

S * *

n 6 0

i 

* * n

n * * * * o

a 4 0 * * i * * * * * * * * * *

t * *

e * *

a

i M

v 2 0

e d 0

A c e t o i n A l a n i n e A c e t a t e B u t y r a t e M e t h a n o l P y r u v a t e S u c c i n a t e T a u r i n e U r e a P r o p i o n a t e G l y c i n e L a c t a t e F o r m a t e

M e t a b o l i t e

Figure 6.9: Comparison of the inter- and intra-individual variation of salivary metabolites. Intra-individual variation is the mean ± SEM of the relative standard deviation of each participant’s metabolite concentration (n=23). Inter-individual variation is indicated by the dotted lines above each column, representing the relative standard deviation of the full data set for that particular metabolite. P-values are for one-sample t-test between mean relative standard deviation of each participant against total relative standard deviation of the entire sample.

209

Table 6.3: Ranking of the intra and inter-individual stability of salivary proteins and metabolites. Proteins are in bold to aid discrimination Salivary analytes ranked Salivary analytes ranked Salivary analytes ranked by by intra-individual by inter-individual mean maximum intra-individual variation (mean relative variation (relative fold change standard deviation across standard deviation of all all participants) samples) Cystatin 20.5 Cystatin 41.86 Cystatin 1.86 Acetoin 23.81 PRP2 50.03 Amylase 1.87 Alanine 24.12 Amylase 56.09 Acetoin 1.91 Amylase 24.77 Urea 56.41 Alanine 1.93 Acetate 24.94 Pyruvate 59.62 Methanol 2.02 Butyrate 26.22 Acetoin 64.39 Butyrate 2.05 Methanol 26.6 Succinate 64.57 PRP3 2.08 PRP2 26.83 PRP1 66.54 Pyruvate 2.16 Pyruvate 27.34 MUC5B 68.23 Acetate 2.19 Succinate 31.18 Alanine 73.99 Succinate 2.38 Glycine 31.66 Methanol 78.22 Glycine 2.42 Taurine 32.66 Butyrate 80.22 Urea 2.44 MUC5B 32.93 MUC7 85.75 Taurine 2.58 Propionate 33.87 Acetate 87.01 PRP2 2.62 Urea 34.23 Taurine 88.06 Propionate 2.79 PRP1 36.23 glPRP 90.27 MUC5B 2.87 MUC7 41.53 Lactate 92.88 glPRP 3.93 glPRP 41.7 Glycine 94.52 Lactate 4.43 Lactate 53.93 Propionate 115.95 MUC7 5.23 Statherin 62.83 Formate 141.67 Statherin 7.61 Formate 91.65 Statherin 196.94 Formate 42.44

6.4.4 Taste thresholds of trained and untrained panellists

Detection and recognition thresholds for all participants on both weeks are presented in Figure 6.10. Participants who did not complete both sessions have been left out. The majority of participants showed fluctuations in detection and recognition thresholds for sweet and bitter tastes between weeks. These fluctuations were generally small in magnitude, and

210 thresholds for some participants were stable between weeks. The sour taste shows on week one the majority of participant detected the taste at the lowest concentration, hence the data for the second week, using a more dilute citric acid solution shows better discrimination. Comparisons between taste thresholds for trained and untrained panellists are shown in Figure 6.11. These are the data from both weeks for sweet and bitter, but for sour only the second week data is used. Mann-Whitney U test shows that trained panellists had statistically lower detection and recognition thresholds for bitter, and recognition thresholds for sweet approached significance (p = 0.06).

Figure 6.10: A comparison of detection and recognition thresholds between two weeks for bitter, sweet and sour tastes. The concentration range used on the second sour week was one third that of the first week as the original concentration was excessively strong with all but one participants detecting the taste at the lowest concentration. Concentration ranges for sweet and bitter were the same on both weeks.

211

Figure 6.11: Comparison of detection and recognition thresholds for bitter, sweet and sour tastes between trained and untrained panellists. Data from both sessions is used for bitter and sweet taste, for sour only the second week data is used as sour taste solutions in the first week were excessively concentrated. Data are compared by Mann-Whitney U test.

212

6.4.5 Relationship between taste thresholds and salivary composition

For each taste session, participants were ranked by detection and recognition thresholds. Concentrations of salivary proteins and metabolites were compared between high threshold (insensitive) and low threshold (sensitive) tasters. Results that were observed on both weeks are presented graphically in Figure 6.12. These included significantly higher urea in the saliva of lower detection threshold tasters for sweet and an unassigned spectral region (5.64 – 5.68 ppm) which was more concentrated in participants with a lower recognition threshold for sour.

Salivary analytes that differed statistically between sensitive and insensitive tasters only on one week are presented in Table 6.4. Unassigned spectral peaks are shown in Figure 6.12 c.

Table 6.4: Summary of the salivary analytes that were significantly different between individuals with sensitive or insensitive taste thresholds. As the number of protons contributing to the unassigned peaks is unknown, they have not been converted into mmol/l concentrations. N.B. high and low cut off points were based on the ranked sensory threshold of all participants and therefore are composed of both trained and untrained participants.

Taste Low cut off; High cut off; Salivary Mean and Mean and p- threshold no. of no. of analyte SEM for SEM for high value participants participants (units) low threshold (two- threshold (insensitive) tailed (sensitive) t-test) Bitter < 0.11 g/l > 0.14 g/l MUC7 0.06 0.13 (0.027) 0.02 recognition caffeine; caffeine; (no units) (0.015) n = 10 n = 9 Bitter < 0.11 g/l > 0.14 g/l Glycine 0.37 (0.04) 0.61 (0.11) 0.02 recognition caffeine; caffeine; (mmol/l) n = 10 n = 9 Bitter < 0.11 g/l > 0.14 g/l Taurine 0.28 (0.03) 0.57 (0.13) 0.03 recognition caffeine; caffeine; (mmol/l) n = 10 n = 9 Bitter < 0.06 g/l > 0.07 g/l 3.16 ppm 0.08 0.13 (0.017) 0.01 detection caffeine; caffeine; (unknown) (0.004)

213

n = 9 n = 10 Sweet < 0.34 g/l > 0.95 g/l Flow rate 0.8 (0.09) 0.55 (0.038) 0.04 detection sucrose; sucrose; (g/min) n = 9 n = 10 Sweet < 0.34 g/l > 0.95 g/l 3.16 ppm 0.07 0.13 (0.019) 0.015 detection sucrose; sucrose; (unknown) (0.007) n = 8 n = 8 Sweet < 0.34 g/l > 0.95 g/l Lactate 1.35 (0.28) 0.57 (0.23) 0.03 detection sucrose; sucrose; (mmol/l) n = 8 n = 8 Sour < 0.04 g/l > 0.05 g/l 1.16 – 1.13 0.07 (0.01) 0.13 (0.027) 0.02 detection citric acid; citric acid; ppm n = 12 n = 7 (unknown) Sour < 0.13 g/l > 0.16 g/l Flow rate 0.6 (0.061) 0.81 (0.055) 0.02 recognition citric acid; citric acid; (g/min) n = 10 n = 8

214

Figure 6.12: Results of salivary metabolites and spectral regions statistically different between sensitive and insensitive tasters on both weeks. 6.12 a. shows the results for sweet detection thresholds where insensitive tasters had significantly lower concentrations of urea on both sessions. 6.12 b. shows the results for sour recognition threshold where a spectral region 5.64 – 5.68 was consistently less concentrated in insensitive tasters. This region is identified on an example partial 1H-NMR spectrum in 6.12 c. For reference, other unassigned peaks from Table 6.4 that appeared to associate with taste threshold only on one session are also illustrated in 6.12. c. 215

6.4.6 Comparison of salivary composition between trained panellists, untrained panellists and age and gender matched controls

No differences in the relative stability of any of the measured major salivary proteins or metabolites was detected between the trained and untrained panellists. A number of major proteins and salivary metabolites differed significantly between trained and untrained panellists.

These were the proteins glPRP, MUC5B and PRP2 and the metabolites urea, taurine, methanol, glycine, acetate, propionate and butyrate. To control for factors such as sex given the all-female composition of the trained panellists, appropriate age and sex matched controls were sought from the twin participants of the subsequent chapter.

Comparisons could not be made between the protein band intensity between studies due to differences in gel loading and background filter when imaging, however as the 1H-NMR spectroscopy was conducted in the same manner, relative to the same standard, metabolite comparisons were performed. Comparisons between protein composition are shown in Figure 6.13 and comparisons in metabolite composition are shown in Figure 6.14.

Figure 6.13: Comparison of significantly different salivary protein band intensities between trained and untrained panellists. P-values are for two-tailed t-test.

216

Figure 6.14: Comparison of salivary metabolite concentrations between trained panellists, untrained panellists and age-matched female controls to address the sex imbalance between the trained and untrained panellists. p-values are for Tukey’s post-hoc following one-way ANOVA. 217

6.5 Discussion

6.5.1 Stability of salivary protein composition, salivary metabolite composition and taste thresholds

The results of repeated taste threshold assessment showed that the majority of individuals did display a degree of fluctuation in their detection and recognition thresholds for sweet and bitter taste. This was true for both trained and untrained panellists. This study showed that both the major protein and metabolite profiles of saliva are significantly more variable between individuals than on separate days within individuals. Both measures showed a surprisingly similar level of intra-individual stability relative to the inter-individual variation of (0.39:1).

With respect to individual major protein bands these findings are in line with those of Hsiao et al. (2019) 269. They reported relative standard deviations (using the terminology “coefficient of variability”) for 90 lower abundance salivary proteins, analysed by mass-spectroscopy. The reported inter-day variation in net protein profile was 43% and inter-individual variation was 69%. Our results for the eight major salivary proteins measured were 36% and 82% respectively. The latter value may be slightly skewed by the highly variable protein statherin, the exclusion of which changed mean inter-individual variation to 66%.

Studies of the stability of salivary metabolite profiles are sparse in literature however our findings were similar to those of Wallner Liebmann et al., (2016) 79, who found that PCA of salivary 1H-NMR spectra from the same individuals would allow fairly robust discrimination of the sample donor. As only one sample was collected per day in the present study, analysis of intra-day changes in saliva could not be performed. Kawanishi et al. (2019) 77, reported that intra-day salivary metabolite profiles did not change whereas inter-day samples did. Furthermore, the stability of salivary metabolite profile appears to be stable over at least sixteen days as there was no significant difference between the stability of intra-week and inter-week saliva samples.

While the global profiles of salivary proteins and metabolites are relatively stable between days, there is scope for individual analytes to fluctuate considerably day-to-day, independent of the net metabolic or protein profile. Hsiao et al. (2018) 269, reported that minor salivary proteins typically displayed intra-individual fold changes of around two-fold. Our results find that the majority of salivary analytes show comparable or greater average maximum fold

218 change. This is an important finding concerning the potential effect of salivary composition on taste perception. Such intra-individual variation in fold change of specific metabolites is typically slightly greater than the differences in mean metabolite concentrations found between sensitive and insensitive tasters. Therefore, if salivary metabolites were to directly or indirectly alter taste sensitivity, individuals experiencing large day-to-day fluctuations in salivary metabolite concentrations would experience resulting changes in taste sensitivity. Between-day fluctuation in the concentrations of salivary metabolites could be a contributing factor to the large inter-day variability in suprathreshold taste perception reported by Prutkin et al. (2000) 174.

As well as showing salivary proteins and metabolites display comparable levels of stability, this study provided further insight into the relationship between salivary protein and metabolite composition. We found that a number of metabolites (glycine, taurine, acetate, butyrate, succinate, propionate) correlated positively with salivary proteins MUC5B, amylase and statherin. Urea was also the only metabolite to correlate inversely with these salivary proteins. Alongside the results from Chapter 4, where urea correlated inversely with salivary bacterial load but the majority of other metabolites showed a positive correlation, this finding supports the role for salivary proteins as a major source of nutrients for oral bacteria. As the most abundant salivary proteins, MUC5B and amylase would present a particularly rich food source that would be continuously regenerated in the absence of residual exogenous nutrients. Additionally, amylase may be liberating nutrients from residual in the mouth. The concept of salivary proteins serving as a substrate for oral bacteria has been recognised, although there is a lack of in-vivo data of the nature of which proteins are consumed by which bacteria (Scannapieco, 1994; Rudney, 2000; Ruhl, 2012) 54,270,271. This is particularly true for the role of salivary protein-bacteria interactions in contributing to the salivary metabolome.

6.5.2 Associations between taste thresholds and salivary composition

Urea was observed to be more concentrated in the saliva of individuals with a lower sucrose detection threshold on both sweet tasting sessions. There are a number of potential mechanisms for the role of salivary urea in taste perception. While urea is a taste active compound (bitter in sufficient concentrations), salivary urea concentrations are reported to be well below taste threshold 272. Endogenous salivary urea has, however, been shown to affect the suprathreshold perception of taste intensity for sour, umami and bitter tastes in chronic kidney disease patients 201. This patient demographic experience raised circulating 219 urea due to renal impairment which consequently causes raised salivary urea. Manley et al. (2012) 201, found that chronic kidney disease patients rated bitter, sour and umami stimuli as significantly less intense than a small number of controls. The salivary urea in chronic kidney disease patients was around four-fold that of healthy individuals, so still well below the threshold required to cause a bitter taste. Furthermore, potassium and bicarbonate were also found to be raised in chronic kidney disease patients and likely many other oral and systemic factors affecting taste would be different in a kidney disease patient group.

Nevertheless, there is a possibility urea could be active on taste receptors without necessarily being detectable itself. This may be via sub-threshold taste active properties or alternative mechanisms independent of any taste sensation of urea. Considering the interactions between bitter urea and sweet sucrose, the literature is somewhat lacking. Keast & Breslin (2003) 273, summarised binary taste interactions. Sweet and bitter are generally inhibitory at moderate and high concentrations, however at low concentrations, which salivary urea and the detection threshold of sucrose certainly fall under, the interaction can be variable or even enhancing. Constant low levels of urea bathing taste receptors via saliva may theoretically enhance the detection of sucrose at low concentrations. McBurney and Bartoshuk 274 studied the effects of urea adaptation to various tastes including sucrose, albeit at relatively high suprathreshold concentrations of urea and sucrose. It was found that at lower sucrose concentrations of 18 mM, urea conditioning (0.82 M for thirty seconds) caused a more intense net sensation compared to water, possibly in part due to the development of some salty and sour qualities to the overall taste.

Urea has been demonstrated to have a functional role in supporting keratinocytes including the upregulation of production of proteins involved in barrier function and microbial defence 275. Whether urea has a similar effect on taste receptor cells or oral epithelial cells has not been investigated. An analogous role for urea in the maintenance of taste receptor cell function would present a possible mechanism for urea enhancing taste function independent from its own taste properties at higher concentrations.

Higher urea levels in the saliva of individuals with more sensitive sucrose detection could, of course, be independent of causation. As shown in Chapter 4, urea inversely reflects oral bacterial load. Therefore, high salivary urea could be a marker of fewer oral bacteria that may produce a myriad of biologically active substances that alter taste perception but are not detected directly by the analyses used in this chapter. The spectral region from 5.64 to 5.68 ppm that was associated with sour recognition threshold on both visits might exemplify this. 220

Although in proximity to the broad urea peak the region does not appear to be a distinct peak but rather a small baseline disruption. This may reflect a macromolecule not fully supressed by the CPMG pulse sequence.

The salivary components found to associate with taste thresholds on only one occasion may still be of significance. Taurine was found to inversely associate with bitter sensitivity. Taurine is known to have important bioactive properties in relation to neural processes, as described in the previous chapter. The inverse association of taurine with sweet suprathreshold sensitivity (see Chapter 7) and the inverse association with reduced bitter recognition noted in this chapter adds weight to the inhibitory nature of salivary taurine in taste processes. The peak at 3.16 ppm was found to inversely associate with bitter and sweet thresholds, indicating a molecule that might display a general suppression of taste processes. The peak between 1.13 and 1.16 ppm associated with sour detection is widely assigned in literature as propylene glycol or 1,2-propanediol. The results from Chapter 3 disprove this assignment although the actual molecule is still unknown. The molecule has recently been identified (still referred to as propylene glycol) as being raised in oral squamous cell carcinoma and thus could be of biological significance 276.

6.5.3 Comparisons of taste thresholds and saliva composition of trained and untrained panellists

Differences in the salivary composition between trained and untrained panellists, as well as age matched female controls were marked. Salivary urea was significantly higher and salivary SCFAs (butyrate, propionate and acetate) were significantly lower in trained panellists than the other groups. Salivary taurine was also significantly lower in trained panellists than untrained panellists and approached significance (p = 0.06) for the age matched female controls. Together with the results from Chapter 4, these indicate a lower bacterial load and potential differences in the immune cell infiltrate of GCF in trained panellists. Lower levels of molecules such as taurine and SCFAs were associated with greater suprathreshold sensitivity to sucrose and oleic acid in the subsequent chapter.

These metabolomic differences in saliva most likely reflect a cleaner mouth in trained panellists. This might in part be due to better oral hygiene practices, as the trained panellists would have a high incentive to care for their mouths which is effectively their livelihood. Improved oral hygiene has been shown to confer better salt and sweet detection in elderly

221 participants and tongue brushing has been shown to improve detection thresholds of sucrose, sodium chloride, caffeine and citric acid in both young and old groups 277 278.

A further tongue brushing study showed improve suprathreshold sensitivity to bitter and salty taste however a negligible reduction of bacterial load was reported, although tongue coverage did reduce 279. As advocated by Takahashi (2015), the net metabolic activity of the oral microbiome may be a more meaningful measure than simply quantifying the total bacteria or the species diversity of oral bacteria 94. For example, an abundant relatively metabolically inactive microbiome may be preferable to a numerically small but metabolically active oral microflora.

Beyond oral hygiene, there may be additional biological factors explaining these metabolic differences between trained and untrained panellists such as differences in salivary protein abundance. MUC5B was significantly lower in trained panellists. Ruhl suggests that glycoproteins are a preferred food source for oral bacteria over proteins lacking abundant carbohydrate side groups 54. As the most abundant salivary glycoprotein, higher MUC5B levels in untrained panellists could sustain a larger population of oral bacteria.

Trained panellists also had significantly higher levels of glPRP and PRP1 bands. Proline-rich proteins are known to have a role in binding a wide range of oral bacteria 280. Higher levels of PRP in saliva has been associated with reduced dental caries in children 281 282. Other bacterial species aggregated by salivary PRP include fusobacterium nucleatum, a species involved in periodontal disease 283.

6.5.6 Summary

This chapter addressed several questions in relation to salivary composition and taste perception. While it was found that salivary protein and metabolite composition is generally stable over a time span of at least two weeks, individual proteins and metabolites can fluctuate several-fold between days. Therefore, changes in certain metabolites and proteins between days might explain the variable nature of taste perception better than non-changing biological factors such as taste receptor genetics or fungiform papillae density. An association between endogenous salivary metabolite concentrations and taste thresholds was found with urea being consistently associated with improved sucrose detection. Urea was also found to be significantly more concentrated in the saliva of trained panellists compared to untrained panellists, whereas SCFAs were less concentrated in the saliva of trained panellists. These 222 findings implied reduced bacterial load in trained panellists’ mouths, something associated in literature with more sensitive tasting. Microbial metabolism may constitute a mechanistic explanation for these previous studies.

223

Chapter 7: Investigation of salivary composition as an oral environmental factor in relation to suprathreshold oral perception

7.1 Introduction

The concept of the oral environment, including salivary composition, influencing taste perception is not new. In the early 1960’s, McBurney & Pfaffmann (1963) 195,explored the hypothesis that salivary sodium concentrations could modulate the thresholds for sodium chloride perception 195. Despite this history there remain relatively few studies on the topic. Most studies have focused on the interactions between a single salivary component, identified in advance, in association with a specific taste. Delwiche & O’Mahony (1996) 197, built on preceding work relating salivary sodium levels to salt perception, Scinska-Bienkowska et al., (2006) 198, found endogenous salivary glutamate levels improved hedonic response to monosodium glutamate (umami taste) and Christensen et al., (1987) 200, found that increased salivary flow rate and pH mitigated the sour taste of acids.

Studies looking more broadly at salivary composition and whether it has a bearing on taste perception are fewer still. In recent years, however, several studies have looked at proteomic composition and taste sensitivity. Many of the results emerging implicate salivary protease inhibitors in taste sensitivity. Dsamou et al. (2011) 208, studied the salivary protein abundance in subjects sensitive and insensitive to the bitter taste of caffeine and found significant differences. The authors found that insensitive subjects had lower levels of amylase and albumin fragments and higher levels of the protease inhibitor cystatin-SN. These findings implied reduced proteolysis in the saliva of insensitive subjects and the authors suggested insensitive subjects had a more intact protein film protecting TRCs from exogenous caffeine molecules. Rodrigues et al. (2017) 213, studied the proteome of young adults with respect to sucrose thresholds. They reported somewhat opposing results to those of caffeine bitterness reported by Dsamou et al., (2011) 208, with insensitive subjects having higher levels of amylase and carbon anhydrase VI and lower levels of cystatins 213. The study also analysed salivary glucose and reported higher levels of glucose in insensitive females. This is suggestive of a sweet conditioning effect of salivary glucose on taste receptors. Rodrigues et al. (2019) 211, subsequently studied the saliva of obese and non-obese children with respect to caffeine sensitivity. The authors observed that in overweight children higher levels of cystatin were associated with bitter sensitivity whereas the opposite was true in normal weight participants. Stolle et al. (2017) 212, studied the salivary proteome in relation to sodium chloride perception.

224

Insensitive individuals were found to have greater abundance of cystatin-D, cystatin-S and cystatin-SN.

Recently, a growing recognition of the potential for the salivary metabolome to interact with taste perception processes is emerging. Feron (2019) 284, recognised that many salivary metabolites are themselves taste-active molecules and therefore their presence in saliva, which may approach threshold levels could explain variations in taste sensitivity. Furthermore, the interaction between host and microbiome, particularly on surface films of the tongue and oral soft tissues has been described by Neyraud & Morzel (2019) 246, is an important phenomenon that is recently emerging in current research. To date, there is arguably only one study applying metabolomics to saliva in association with oral perception, that of Mounayar et al. (2014) 91. This study found that the short chain fatty acids acetate (which is the most concentrated salivary metabolite) and butyrate were overexpressed in the saliva of insensitive perceivers of oleic acid. The work of Mounayar et al. (2014) not only adds to a growing body of evidence that gustation or similar oral chemoreceptive processes are involved in the recognition of fats, but also demonstrates 1H-NMR driven metabolomics of saliva is a useful technique in the study of oral perception. Bader et al. (2018) 187, studied metabolomic changes in saliva following citric acid stimulation in relation to salt sensitivity. The compositional changes reported generally related only to salivary ionic composition with sodium itself influencing salt taste sensitivity.

In light of these recent studies and the growing recognition of salivary metabolites as putative modifiers of taste sensitivity, further exploration of this relationship is merited. A challenge in studying taste perception, particularly in association to saliva, stems from the complexity of the genetic and environmental factors that are involved. The genetics of human taste perception are known to be relatively complex. Genetic screening in relation to taste phenotypes is a relatively new field, only emerging over the two decades, however knowledge is expanding continually. A significant majority of literature on human taste genetics focuses on the bitter receptors TAS2R38. Variants of this gene featuring three common SNPs are known to confer the sensitivity to the bitterness of PROP and PTC. TAS2R38 dependant sensitivity to PROP/PTC in turn has been linked to many other taste phenotypes largely under the umbrella of the “supertaster” concept 285. Beyond TAS2R38, SNPs in the genes encoding several other bitter receptors as well as those for umami and sweet taste (TAS1R2 and TAS1R3) have been linked to taste perception 286.

225

A useful approach to studying human genetics without specifically screening individual DNA is twin studies 287. Twin and family studies predate the age of molecular genetics and, with respect to taste perception, were critical in the early identification of the Mendellian nature of PROP/PTC sensitivity 288. Despite having a long tradition in biomedical science, twin studies of taste perception remain relatively limited 289. Concerning bitter perception, a number of studies have examined various bitter stimuli in twin pairs and have reported varying results. Threshold perception of quinine has been reported by Kaplan et al. (1967) 290, as largely not heritable and by Smith & Davies (1973) 291, as relatively strongly heritable (~0.55). Hansen et al. (2006) 292, found the heritability of suprathreshold intensity of quinine and caffeine to be much lower than that of PROP (~ 0.3 compared to ~ 0.7) and did not find common genetic factors explaining heritability of the different substances.

Twin studies of sweet taste have yielded similarly variable results, although the general consensus is that aspects of sweet taste perception are only weakly heritable. Knapilla et al. (2012) 293, found the heritability of sucrose intensity to be around 0.18, compared to a heritability estimate of 0.52 for PTC intensity (applying Falconer’s heritability formula to their reported data). Sweetness liking and preference for sucrose was measured by Keskitalo et al. (2007) 294, and Bretz et al. (2006) 295, respectively, with similar degrees of heritability reported for the different measures (0.49 – 0.55). The genetics governing hedonic response to sucrose therefore appear distinct from those governing intensity perception. Hwang et al. (2015) 296, reported relatively low but comparable heritability for sweet intensity of glucose, fructose and aspartame of around 0.3 – 0.35. Unlike for bitter stimuli as reported by Hansen et al. (2006) 292, where a common genetic factor was not detected between the different stimuli, Hwang et al. (2015) 296, reported evidence that the perception of sugars and artificial stimuli might be governed by a common set of genes.

The evidence for heritability of other basic tastes is limited. Beauchamp et al. (1987) 297, reported a negligible genetic effect on salt taste perception, a finding subsequently corroborated by Wise et al., (2007) 298. Wise et al. (2007) 298, also reported a relatively strong genetic component to sour thresholds for citric acid (accounting for 53 % of variance). This finding was somewhat supported by Törnwall et al. (2012) 299, who reported genetic factors explained around 31 % of the variation in suprathreshold citric acid intensity rating. Recently, Costanzo et al. (2018) 300, studied the genetics of fat taste sensitivity has also been investigated and no genetic element to fat detection was found with dietary fat intake being cited as the main environmental modifier.

226

A limited number of twin studies focusing on salivary factors such as composition and flow rate have been conducted, offering some insight into the genetic contributions to the oral environment. Smith et al. (1976) 301, looked at the heritability of parotid saliva protein composition by SDS-PAGE and developed a system of measuring band intensity to estimate similarity between monozygotic twins, dizygotic twins, siblings and unrelated individuals. While they found the highest degree of concordance in monozygotic twins, a large degree of concordance was also seen even between unrelated individuals. Rudney et al. (1994) 302, looked at salivary flow rate, total protein concentration and the concentrations of selected individual salivary proteins. A degree of heritability was found for total protein, lactoferrin and total peroxidase concentrations however flow rate, secretory IgA, lysozyme and myeloperoxidase were not found to be heritable. While a twin study of the salivary metabolome has not yet entered publication, a longitudinal study of the oral microbiome has been conducted. Stahringer et al. (2012) 303, found that twins had more similar microbial communities than unrelated individuals, this effect dissipated from age 12 to age 22. The authors concluded shared environment during childhood is a greater factor than genetics in shaping the oral microbiome.

A study encompassing both basic taste perception and assessment of saliva in twins has not been performed. Considering the general evidence that salivary composition can be a modifier of taste response and that taste perception is generally only weakly heritable, a twin study encompassing salivary analysis would clarify the effect of oral environment on taste perception. Interestingly, however, saliva and astringency perception has been studied in a twin cohort. Tornwall et al., (2011) 304, found that whilst astringency perception of tannic acid containing apple juice was not related in monozygotic twins, hedonic response to some astringent stimuli (drinks and fruits) were. Importantly, the most similar salivary proteins amongst monozygotic twins were mucins and PRPs. Given the implicated role of these proteins in interacting with astringent molecules, leading to polyphenol-mucin interactions and consequent lubrication failure at mucosal surfaces 154,204,205, the findings might imply a genetic salivary protection from astringency. The work of Tornwall et al., (2011) 304, on the heritability of astringency was not conclusive, however, and further study into this area is merited.

A further avenue not yet explored in existing literature is whether the intra-oral microbial catabolism of sucrose, observed in the previous chapter, has any influence on taste perception. Studies linking perception of natural sugars to food consumption often report mixed results, often differing based on the tastant used, whether sucrose, glucose or fructose 227

252. Given the rapid nature of sucrose breakdown in the mouth, shown in Chapter 5, a sucrose solution will quickly become a mixture of sucrose, glucose, fructose and metabolic products such as lactate and pyruvate. Different sugars have different intensities and sweet qualities 305. An identical sucrose solution stimulus could ultimately become a different composition of sugars and metabolites based on-inter individual differences in oral microbial metabolic activity. Such intra-oral events could potentially explain some of the inconsistencies in the results of the studies reviewed by Tan & Tucker (2019) 252.

7.2 Aims and Objectives

The work presented in this chapter was conducted to evaluate the genetic and oral environmental factors contributing to taste perception. Participants were same-sex twin pairs recruited from the TwinsUK registry at St. Thomas’s hospital, London. The following questions were addressed: 1. To what extent is salivary flow rate, major protein composition and metabolome shaped by genetics? 2. To what extent do genetics influence fungiform papillae density and suprathreshold intensity perception of sucrose, aspartame, caffeine, oleic acid and tea astringency? 3. Does salivary flow rate, major protein composition or metabolite composition relate to oral perception in unrelated participants, twins with discordant taste perception or twins with concordant taste perception? 4. Do differences in intra-oral sucrose metabolism have a bearing on taste perception?

7.3 Methodology

7.3.1 Participant recruitment

Participants in this study were recruited from the TwinsUK registry based at St Thomas’s Hospital, London. A mixture of monozygotic and dizygotic twin pairs was recruited although all twin pairs were of the same sex. Any willing adult participants were eligible and an upper age limit of 70 years was imposed. Exclusion criteria included antibiotic use within the previous six months, acute oral or systemic illness e.g. mouth ulcers or sinus blockage. Medication use did not exclude participants provided salivary function and taste perception was not reported to have been affected. Individuals who commenced the study but were subsequently found to

228 have an unstimulated salivary flow below 0.3 g/min were excluded from completing the study.

7.3.2 Tastant preparation and sensory assessment

Five tastants were evaluated by participants. The tastants were 0.25 M sucrose, 3 mM aspartame, 8 mM caffeine, 16 mM oleic acid and 28 g/l black tea. All tastants were purchased as food grade from Sigma-Aldrich, except the tea which was Yorkshire brand (Bettys and Taylors, Harrogate, UK). Sucrose, caffeine and aspartame were prepared by dissolution at room temperature in Buxton mineral water on the morning of the study. Tea was prepared by dissolving 7 g of tea in 250 ml of freshly boiled tap water, gently agitating for ten minutes then filtering and cooling to room temperature. Oleic acid was prepared by dispersing an aliquot of oleic acid (stored under N2) in Buxton water immediately prior to tasting, making an emulsion.

Suprathreshold intensity assessment of the five tastants was conducted using the scale described in Chapter 2. Participants were initially familiarised with the scale and the rating process and were provided both written and verbal instruction on the correct use of the scale. The descriptors used on the scale to describe the primary sensation assessed were sweetness for sucrose and aspartame, bitterness for caffeine and astringency for tea (described as dryness and loss of lubrication of lips and cheeks). For oleic acid, since any taste quality is less definitively agreed, it was described simply as total oral sensation and the scale was left without descriptors of a specific sensation.

7.3.3 Study protocol and saliva collection

Taste assessment and sample collection were conducted in the afternoon between 2:00 to 4:00 p.m., at least one hour since the participant’s most recent food or drink consumption. Both twins completed the test simultaneously. Twins were seated apart for the taste assessments and told not to discuss their thoughts about any of the tastes until the end. An unstimulated saliva sample was collected by passive spitting into a sterile pre-weighed universal tube. A 10 ml control solution (Buxton water) was then given, passively held in the mouth for thirty seconds, expectorated, and then a two-minute saliva sample collection period was completed. The thirty second tasting followed by expectorating and two-minutes of saliva collection was completed for each of the tastants in the order sucrose, aspartame, caffeine, tea, and oleic acid. Twins were instructed to rinse with Buxton water between tastes

229 and take as long as necessary for the mouth to return to a resting state before progressing to the next solution.

After collection, samples were weighed to determine flow rate and immediately stored on ice prior to preparation.

7.3.4 Fungiform papillae density measurement

Fungiform papillae density was measured in twin pairs using a simplified method to that described in Chapter 2. The tongue was stained with brilliant blue FCF, then the papillae area was determined using paper discs with 6mm diameter central holes and 3mm wide borders. The discs were positioned on the tongue with the outer border touching the midline laterally and the dorsal-ventral junction anteriorly. The tongue was then photographed and the area to count papillae density (28.3 mm2) was immediately visible on the final image without the need for additional measurements and calibrations. Papillae in this area were identified and counted as described in Chapter 2.

7.3.5 Sample handling and analyses

Samples were centrifuged at 15,000 g at 4 °C for ten minutes then supernatant was divided into aliquots for future analysis and stored at -80 °C.

1H-NMR spectroscopy All unstimulated saliva samples were analysed by 1H-NMR spectroscopy, as were post-water control and post-sucrose samples from a subset of sensitive and insensitive sucrose perceivers.

Sample preparation, spectral acquisition and processing was as described in Chapter 2. Supernatant thawed at 4 °C was mixed 4:1 with NMR buffer (yielding a final TSP concentration of 0.1 mM, 10 % D2O), transferred to 3mm ED NMR tubes and analysed using the CPMG pulse sequence. Control samples made from pooling 20 µl of saliva from all participants were placed at the start, end and evenly dispersed throughout the sample queue (every 13 samples) to confirm analytical reproducibility.

Following automated phase and baseline correction with manual correction where necessary, processed spectra were bucketed into 0.01 ppm buckets in (MestRec v). Regions integrated 230 were from 0.7 to 8.5 ppm, excluding the regions close to the water peak between 4.5 and 5.5 ppm. The standard peak (- 0.2 to 0.2 ppm) was included to standardise bucket volumes in each sample. Metabolite peaks of interest were subsequently integrated manually. For the water control/sucrose stimulated samples, metabolites relating to sucrose metabolism including α and β glucose were manually integrated.

Total protein assay Total protein was quantified in unstimulated saliva samples by BCA assay as described in Chapter 2.

1D SDS-PAGE Unstimulated saliva samples were prepared for SDS-PAGE, electrophoresed, stained by Coomassie and PAS staining, imaged and digitised as described in Chapter 2. Band intensity was also measured for MUC5B, MUC7, glPRP, amylase, PRP1, PRP2, cystatin and statherin bands. Total PRP region intensity (between amylase and cystatin) was also measured. Lanes were loaded with 15 µl of buffered sample. Samples from the same twin pair were run on the same gel and a standard reference sample was included on each gel. Thus, any inter-gel effects on assessment of inter-twin salivary protein profile similarity were negated.

7.3.6 Statistical analyses

Assessment of genetic contributions to salivary composition and taste perception The final dataset was unbalanced with respect to MZ and DZ twin pairs (20 MZ and 6 DZ), precluding common assessments of heritability such as Falconer’s formula. Two methods of estimating the genetic contributions to salivary composition and taste perception were therefore employed.

Firstly, Pearson’s correlation amongst monozygotic twins for taste responses and unstimulated salivary flow rate and total protein concentration were conducted. The heights and weights, both biometric measures for which heritability is well characterised, for the twin pairs were correlated between the twin pairs to provide a relative measure of heritability. Intraclass correlation coefficients were also assessed alongside Pearson’s, however resulting data was comparable for both measures.

Secondly, the Euclidean distance between the biometric data for monozygotic twins, dizygotic twins and unrelated individuals in the dataset was calculated and presented for comparison.

231

The bucketed NMR spectral data and digitised protein lane profiles for unstimulated saliva were analysed by PCA in Knime. For the protein data, inter-gel corrections were made by aligning the projected data points relative to the standard lanes run on each gel. Mean Euclidean distance was calculated in three dimensions for the projected data of monozygotic twins, dizygotic twins and unrelated individuals in order to assess the relative genetic contributions to salivary metabolite and major protein composition. For purpose of comparison, mean Euclidean distance was normalised to that of unrelated individuals (i.e. the total variability in the dataset).

Fungiform papillae density measurements were ultimately collected in a wider twin cohort, allowing estimation by Falconer’s formula (Heritability = 2*(rMZ - rDZ)), where rMZ = Pearson’s correlation coefficient of monozygotic twins and rDZ = Pearson’s correlation coefficient of dizygotic twins.

Assessment of salivary composition in relation to taste perception Participants were ranked in order of taste response for each tastant. Initially, comparison was made between unrelated individuals categorised into sensitive (high intensity rating) and insensitive (low intensity rating) perceivers. If twin pairs were grouped together based on similar taste responses the lower perceiving twin was included if both twins were in the insensitive category and the higher perceiving twin was included if both twins were in the sensitive category. On rare occasions where taste responses were identical, neither twin was included.

Differences in flow rate, protein band intensity and 1H-NMR spectral bucket volume were compared between low and high perceivers by two-tailed t-test. Twin pairs were then ranked based on the magnitude of the difference between their response to each tastant. Discordant twins for each taste were defined as having a difference in perceived taste intensity greater than 1.5 units. Salivary data from higher perceiving twins were compared to that of their lower perceiving counterpart by paired t-test. Data from concordant (similar rating) twins was analysed in the same way.

This analysis was conducted both on the concentration of salivary analytes in unstimulated saliva as well as the output (i.e. concentration multiplied by flow rate). Significant differences in isolated spectral buckets from spectral regions with no detectable metabolite/macromolecule peak were assumed to be false positives, reflecting the random nature of baseline fluctuations. 232

Assessment of intra-oral sucrose catabolism in relation to taste perception Metabolite outputs from biofilms were measured in saliva following water control stimulus and the 0.25 M sucrose stimulus. Differences in metabolite output (sucrose solution output - control solution output) were compared between unrelated sensitive and insensitive sweet perceivers by two-tailed t-test. A flow chart of the data analyses with respect to the sensory responses is summarised in Figure 7.1.

Figure 7.1: Flowchart summarising the data handling and analyses for assessing salivary composition and intra-oral catabolism in different groups of participants based on suprathreshold taste sensitivity.

7.4 Results

7.4.1 Participant demographics Participants completing the study included 52 individuals (26 twin pairs) composed of 20 monozygotic twin pairs and 6 dizygotic twin pairs. Female twin pairs were approximately two thirds of the dataset (18 female pairs: 8 male pairs), in keeping with the general demographics of the TwinsUK cohort. Participant ages ranged from 21 to 67 years, with a mean of 44 and standard deviation of 13 years.

233

7.4.2 Taste responses Taste responses for each tastant are depicted in Figure 7.2. Sucrose, aspartame and caffeine were well matched for mean intensity, and were fairly evenly distributed around the scale midpoint. There appeared to be a minimum sucrose threshold of around 2, with no participant rating sucrose below this intensity. Black tea intensity were slightly positively skewed and oleic acid ratings were slightly negatively skewed. A clear separation between sensitive and insensitive tasters could be drawn for all tastes.

Figure 7.2: Distributions of participant taste ratings for all tastants. The demarcations indicate mean and standard deviation.

7.4.3 Assessment of heritability of salivary composition, taste perception and fungiform papillae density

Correlation in monozygotic twins Pearson’s correlation coefficients of unstimulated salivary flow rate, total salivary protein concentration and taste ratings between monozygotic twins are presented in Figure 7.3. Correlations for the established biometric measures of height and weight are included for reference.

234

Figure 7.3: Summary of correlations for salivary flow rate, total protein concentration, and taste ratings between monozygotic twin pairs (n = 26). The correlations for height and weight in these twins are included (horizontal bars). Significant correlations (p < 0.05) are included in bold.

Salivary flow rate and total protein both showed comparable weak positive correlations, considerably below that of body weight. Intensity rating of oleic acid was the strongest correlation, greater than that of weight. Correlations for aspartame approached that of body weight. Correlations for sucrose rating, black tea and caffeine were weaker and non- significant.

Assessment of similarity of salivary composition and taste by comparison of mean Euclidean distance. A comparison of mean Euclidean distance between monozygotic twins, dizygotic twins and unrelated individuals is shown in Figure 7.4. Data for taste perception is shown in Figure 7.4 a. and data for salivary parameters including metabolite composition and major protein composition is shown in Figure 7.4 b. Height showed a relative mean Euclidean distance of 0.15 and weight 0.48. The relative mean Euclidean distance for perception of all tastes between monozygotic twins was greater than for body weight in all cases. Euclidean distance between perception of oleic acid and aspartame were significantly lower in monozygotic twins than in unrelated individuals (0.53 and 0.60, respectively). No significant differences in Euclidean distances between monozygotic twin pairs and unrelated individuals were observed for sucrose, black tea or caffeine perception (0.62, 0.70, 0.88, respectively). Dizygotic twins did not differ between unrelated individuals, although this could reflect the low number of twin pairs (six).

235

For the salivary parameters measured, mean Euclidean distances between monozygotic twins for metabolomic profile and major protein profiles were significantly lower (0.57 and 0.62, respectively) than in unrelated individuals. Significant differences were not detected for flow rate and total protein concentration (0.66 and 0.65, respectively). In all cases the mean relative Euclidean distance was greater than that for body weight.

These results taken together suggest that taste perception of oleic acid and aspartame, as well as salivary metabolite and protein composition have statistically significant genetic elements, however the degree of heritability is relatively low, less than that for body weight.

Fungiform papillae density Estimation of fungiform papillae density heritability was derived from a larger but minimally overlapping cohort of twins. Data from 39 pairs of monozygotic and 39 pairs of dizygotic twins were acquired and analysed by Falconer’s formula. Pearson’s correlation coefficients were 0.902 and 0.418 for MZ and DZ pairs respectively, yielding a heritability estimate of 0.97 (2*(0.902 – 0.418). This indicates fungiform papillae density is a strongly heritable trait, comparable to, or even exceeding, height.

236

Figure 7.4: Comparison of the similarity of taste perception (7.4 a.) and salivary composition (7.4 b.) in monozygotic twins, dizygotic twins and unrelated individuals within the dataset. Height and weight have been included to provide a reference to estimate heritability. The closer the relative Euclidean distance is to zero the more similar the twin pairs are relative to the general inter-individual variability. Data were compared by ANOVA with Tukey’s post-hoc between the three groups for each parameter measured. Data are displayed on the same axes to aid visualisation.

7.4.4 Assessment of salivary compositional differences in sensitive and insensitive tasters.

An overview of the taste perception ranges for the different tastants and different perceiver groups is presented in Figure 7.5. A summary of statistically differing 1H-NMR spectral regions

237 between unrelated individuals, discordant twins and concordant twins is presented in Figure 7.6.

Sucrose perception A number of spectral regions were significantly different between sensitive and insensitive sucrose perceivers for unrelated individuals as well as discordant twins, with a high degree of overlap between the different regions within both groups (Figure 7.6). These differences, including differences in protein band intensity, are summarised in Table 7.1. Without exception, protein and metabolite concentrations were higher in the insensitive tasters. No significantly different spectral regions or protein band intensities were detected between sensitive and insensitive concordant twins.

Flow rate was significantly higher for sensitive tasters in both unrelated and discordant twin groups. Differences in flow rates, protein band intensity and known metabolites are summarised in Figure 7.7, showing differences common to unrelated individuals and discordant twins. Further significant differences only found when comparing sensitive and insensitive unrelated individuals are summarised in Figure 7.8. Analysis of protein and metabolite outputs removed the differences detected when measuring concentration, (Figure 7.6).

Aspartame perception Significant differences in salivary composition between insensitive and sensitive aspartame perceivers were detected only discordant twin pairs. These were for flow rate, MUC7 and a considerable majority of spectral regions. Flow rate and MUC7 differences are depicted in Figure 7.9, where flow rate is significantly higher in sensitive perceivers. MUC7 was significantly higher in the saliva of insensitive perceivers. The spectral regions significantly differing appeared to correspond to macromolecule regions but not metabolite peaks. For example, between the regions 1.8 ppm and 3 ppm the only spectral regions not significantly different correspond to major metabolite peaks acetate, 5-aminopentanoate, propionate and succinate, or are relatively flat baseline regions (~ 2.9 ppm), (Figure 7.6). For all spectral regions, insensitive perceivers had a higher mean concentration than sensitive perceivers. When examining output, these spectral regions differences were no longer present.

238

Caffeine perception Results for caffeine are not depicted in Figure 7.6 because only a single metabolite difference was detected. This was for the salivary concentration of formate between sensitive and insensitive individuals, shown in Figure 7.10.

Oleic acid perception As with sucrose, a number of differences in salivary composition were detected between sensitive and insensitive oleic acid perceivers in unrelated individuals and discordant twins. There was a reasonable degree of overlap in these differences in both groups, summarised in Table 7.2 and Figure 7.11. Additional significant differences in salivary acetate, propionate and butyrate and taurine were found between sensitive and insensitive unrelated individuals (Figure 7.12). A similar pattern of higher salivary SCFA concentration in insensitive discordant twin perceivers was observed which approached significance for acetate (p = 0.06). Corresponding p-values for butyrate and propionate were 0.10 and 0.14, respectively.

Black Tea astringency perception Differences in salivary composition between sensitive and insensitive perceivers of astringency were mostly found in unrelated individuals. These were for PRP1, PRP2 and total PRP band intensity (Figure 7.13), and a considerable proportion of spectral regions corresponding to macromolecules. In all cases the relationship was inverse i.e. higher abundance in the saliva of insensitive perceivers. Total PRP output was also significantly higher in insensitive astringency perceivers (Figure 7.13), as were the outputs for a larger proportion of spectral regions (Figure 7.6). A relatively lower number of spectral regions differed significantly between sensitive and insensitive discordant twins although these regions were limited to macromolecule regions as opposed to metabolite peaks.

239

Figure 7.5: Box and whisker plots summarising the distribution of intensity perception ratings for each tastant. These are divided into six groups – sensitive and insensitive perceivers for unrelated individuals, discordant twin pairs and concordant twin pairs.

240

Figure 7.6: Summary of spectral regions of unstimulated saliva statistically differing between insensitive and sensitive perceivers for Sucrose, Aspartame, Oleic acid and Black tea. The numbered lines for each taste correspond to the following comparisons: 1. – Unrelated sensitive v. insensitive individuals, 2. – Sensitive v. insensitive discordant twins, 3. – Sensitive v. insensitive concordant twins. Lines 4., 5., and 6. are as for 1., 2., and 3., respectively, but for output rather than concentration (i.e. adjusting for salivary flow rate). Green indicates no significant difference; red indicates a significant difference (p < 0.05). Unrelated individuals were compared by two-tailed t-test, twin pairs by two-tailed paired t-test. The corresponding partial 1D 1H-NMR 600 MHz CPMG spectra shown above is from the pooled control sample, showing regions 0.8 to 8.49 ppm, excluding regions 4.5 to 6.88 ppm. Although the region 5.5 – 6.88 ppm was included in the analyses the only major metabolite peak, urea, did not show significant differences for any comparisons.

241

Table 7.1: Summary of salivary compositional differences between sensitive and insensitive sucrose tasters. Group(s) compared Differing spectral Magnitude of differences (fold change) regions/salivary expressed as insensitive/sensitive analytes Common differences Macromolecule peaks: when comparing: 6.92 – 6.97 ppm; 1.69-fold unrelated, 1.39-fold disparate twins 4.30 – 4.45 ppm; 1.48-fold unrelated, 1.26-fold disparate twins Sensitive with 3.19 – 3.24 ppm; 1.68-fold unrelated, 1.37-fold disparate twins insensitive unrelated 2.37 – 2.47 ppm 1.56-fold unrelated, 1.36-fold disparate twins individuals (excluding pyruvate and succinate peaks); and 2.27 – 2.33 ppm; 1.52-fold unrelated, 1.31-fold disparate twins 1.99 – 2.05 ppm; 1.54-fold unrelated, 1.31-fold disparate twins Sensitive with 1.83 – 1.91 ppm; 1.47-fold unrelated, 1.30-fold disparate twins insensitive discordant 1.69 – 1.74 ppm; 1.57-fold unrelated, 1.34-fold disparate twins twins 1.39 – 1.53 ppm; 1.61-fold unrelated, 1.35-fold disparate twins 0.79 – 0.88 ppm; 1.47-fold unrelated, 1.33-fold disparate twins

Metabolites: Citrate 1.51-fold unrelated, 1.35-fold disparate twins Dimethylamine 1.45-fold unrelated, 1.30-fold disparate twins

Protein bands: PRP2 1.65-fold unrelated, 1.34-fold disparate twins Macromolecule peaks: Additional differences 3.88 – 4.11 ppm; 1.73-fold greater when comparing 2.08 – 2.16 ppm; 1.48-fold greater sensitive with insensitive unrelated Metabolites: individuals Taurine 2.24-fold greater Pyruvate 1.60-fold greater

Protein bands: PRP1 1.75-fold greater Total PRP 1.76-fold greater

242

Figure 7.7: Summary of differences in salivary flow rate and composition between insensitive and sensitive sucrose perceivers common to unrelated individuals and twins discordant in their intensity rating for sucrose. Data shown are mean ± SEM compared by two-tailed t-test (paired for twin comparisons, otherwise unpaired).

Figure 7.8: Summary of differences in salivary composition between insensitive and sensitive sucrose perceivers only present when comparing unrelated individuals. Data shown are mean ± SEM compared by two-tailed t-test.

243

Figure 7.9: Summary of flow rate and protein concentration differences observed in sensitive and insensitive aspartame perceiving discordant twins. Data presented are mean ± SEM analysed by paired two-tailed t-test.

Figure 7.10: A significant difference in salivary composition was observed for formate between unrelated sensitive and insensitive caffeine perceivers only. Data presented is mean ± SEM analysed by two-tailed t-test.

244

Table 7.2: Summary of salivary compositional differences between sensitive and insensitive oleic acid perceivers. Taste (stimulus); Differing spectral Magnitude of differences (fold change) groups compared regions/salivary expressed as insensitive/sensitive analytes Common differences Macromolecule peaks: when comparing: 7.49 – 7.53 ppm; 2.10-fold unrelated, 1.48-fold disparate twins 7.35 – 7.41 ppm; 1.68-fold unrelated, 1.34-fold disparate twins Sensitive with 7.26 – 7.33 ppm; 1.55-fold unrelated, 1.27-fold disparate twins insensitive unrelated 7.15 – 7.21 ppm; 1.81-fold unrelated, 1.31-fold disparate twins individuals 6.84 – 6.87 ppm; 1.75-fold unrelated, 1.26-fold disparate twins

and Metabolites: Sensitive with Trimethylamine 1.89-fold unrelated, 1.39-fold disparate twins insensitive disparate Methylamine 1.55-fold unrelated, 1.37-fold disparate twins twins 5-aminopentanoate 1.89-fold unrelated, 1.39-fold disparate twins Additional differences Metabolites: when comparing Acetate 1.71-fold greater sensitive with Propionate 1.96-fold greater insensitive unrelated Butyrate 1.76-fold greater individuals Taurine 2.03-fold greater

Protein bands: Statherin 3.32-fold greater

245

Figure 7.11: Summary of differences in salivary metabolite composition between insensitive and sensitive oleic acid perceivers common to unrelated individuals and twins discordant in their intensity rating for oleic acid. Data shown are mean ± SEM compared by two-tailed t-test (paired for twin comparisons, otherwise unpaired).

Figure 7.12: Additional differences in salivary metabolite composition between insensitive and sensitive oleic acid perceivers only detected in unrelated individuals. Data shown are mean ± SEM compared by two-tailed t-test.

246

Figure 7.13: Summary of different salivary composition between sensitive and insensitive unrelated perceivers of Black tea astringency. Data shown are mean ± SEM compared by two- tailed t-test.

7.4.5 Assessment of intra-oral sucrose catabolism in relation to taste perception A summary of metabolite outputs measured is presented in Table 7.3. The only metabolite to significantly differ between insensitive and sensitive sucrose perceivers was citrate. When analysing metabolite ratios, the ratios of lactate to pyruvate and citrate to pyruvate differed between insensitive and sensitive perceivers. In sensitive perceivers, the lactate to pyruvate ratio was lower and the citrate to pyruvate ratio as higher compared to insensitive perceivers. No difference in flow rate change post-sucrose was detected, Figure 7.14.

Table 7.3: Summary of metabolite output changes following sucrose exposure in sensitive and insensitive sucrose perceivers, n = 9 per group. Glucose = sum of α and β glucose peaks. Data were compared by two-tailed t-test and ranked by p-value, p-values in bold reflect significance (p < 0.05). Metabolite Output post-sucrose Output post-sucrose Two tailed t-test for insensitive for sensitive p-value perceivers perceivers mean (µmol/min) (µmol/min) Mean S.D. Mean S.D. Citrate -0.02 0.03 0.04 0.05 0.02 Pyruvate 0.41 0.40 0.61 0.25 0.24 Alanine 0.06 0.12 0.11 0.09 0.42 Butyrate -0.05 0.12 0.01 0.21 0.44 Succinate 1.04 0.99 1.28 0.54 0.54 Glucose 8.23 13.09 6.22 6.34 0.69

247

Acetoin 0.13 0.17 0.11 0.06 0.75 Lactate 7.95 7.47 8.45 3.59 0.86 Formate -0.012 0.074 0.001 0.238 0.88 Acetate -0.61 2.24 -0.72 3.97 0.94 Propionate 0.48 0.98 0.49 0.99 0.98

Figure 7.14: Summary of citrate:pyruvate and lactate:pyruvate ratios following sucrose rinse in sensitive and insensitive sucrose perceivers. Relative flow rate differences post-sucrose between the two groups are also compared. Data are mean ± SEM analysed by two-tailed t- test.

7.5 Discussion

7.5.1 Genetic contributions to taste perception and saliva composition

Heritability of biological traits can be difficult to estimate, and data will often differ between different study populations of comparable size. Silventoinen et al. (2003) 306, reported that height, possibly the simplest biometric measurement, correlates strongly between monozygotic twins (correlation coefficient range 0.84 to 0.94). Carmichael & McGue (1995) 307 , reported a similar range of correlation coefficients between the height of monozygotic twins and found that body weight correlated between 0.66 and 0.85. This study used height and weight, biometric measurements for which heritability is well studied, as benchmarks for the analysis of taste perception and saliva flow rate and composition, for which there is minimal published data.

Of the five tastes and oral sensations studied, only oleic acid and aspartame intensity ratings showed significant correlations between monozygotic twins. These appeared to correlate to a similar strength to body weight measurements within our sample. The strength of correlation 248 found between monozygotic twin perceptions of oleic acid were considerably higher than those reported by Costanzo et al. (2018) 300. The strength of correlation for aspartame perception was similarly higher than that reported by Hwang et al. (2015) 296. An explanation for this may lie in differences between the taste assessment methodologies. Costanzo et al. (2018) 300, analysed a comparable number of monozygotic twins and reported similar correlation coefficients for height to the present study. Oleic acid sensitivity was determined by threshold assessment, with textural and olfactory cues being masked 300. Therefore, the participants in the present study may have received additional sensory inputs to help them rate the overall perception of oleic acid. Regarding aspartame, Hwang et al. (2015) 296, conducted a large number of sensory assessments and a degree of participant fatigue may have occurred.

Analysis of salivary flow rate, total protein, metabolite profile and protein profile generally showed a low level of genetic influence. When comparing monozygotic twin correlations, our flow rate result was similar to that of Rudney et al. (1994) 302, however we found salivary total protein to be less strongly correlated. When comparing mean Euclidean distance measurements between monozygotic twins and unrelated individuals, no significant differences were detected for flow rate or total protein concentrations. Metabolite and protein profiles were significantly more similar in monozygotic twins than unrelated individuals although the degree of similarity was less than that for body weight. The present study suggests a stronger genetic contribution to salivary metabolite and protein composition than to the oral microbiome, reported by Stahringer et al. (2012) 303. Similarly, our findings suggest a greater genetic contribution to the salivary metabolome than that reported by Zierer et al. (2018) 308, who studied the gut metabolome. Nevertheless, the genetic contribution to the salivary components measured appeared to be relatively small. One exception to this was FPD, which was found to be a strongly heritable trait, comparable to height. Alongside the findings that sensitivity to the majority of tastes was not strongly correlated in monozygotic twins this finding provides further evidence for the limited role of fungiform papillae in mediating taste intensity.

7.5.2 Relationship between salivary composition and taste sensitivity

This study found salivary compositional differences between sensitive and insensitive perceivers of sucrose and oleic acid. Salivary differences showed a fairly high degree of overlap between unrelated individuals and discordant twin perceivers. The same was true, but to a lesser extent for tea astringency perception. This substantiates the theory that the oral 249 environment is of greater importance than underlying genetics in mediating taste perception, as even in twin pairs salivary differences were associated with discordant taste sensitivity. Several differences in the saliva of insensitive and sensitive perceivers for the various tastants were detected. Considering the results for all tastes collectively it can be seen that, without exception, salivary analytes were more concentrated in the saliva of insensitive taste perceivers. This would suggest that salivary components tend to provide an inhibitory effect on taste perception. A second observation is that generally there was no single salivary component common to different tastes suggesting that any relationship between saliva and taste is likely multifactorial and generally tastant-specific.

The only exception to the latter observation was for taurine, detected at higher concentrations in the saliva of insensitive perceivers of both sucrose and oleic acid compared to unrelated sensitive perceivers. Taurine is an important molecule to almost all animals and has a wide range on physiological functions. Importantly, taurine is known to have neuroinhibitory properties. Taurine hyperpolarises excitable cells and antagonises stimulatory neurotransmitters in the central nervous system 309 310. A direct action of taurine on peripheral excitatory cells within taste receptors may be a potential mechanism for taste inhibition. Amine molecules were common to insensitive perceivers of sucrose and oleic acid in both unrelated and discordant twin pairs. This was true for dimethylamine and sucrose perception and methylamine and trimethylamine for oleic acid perception. While these amines are volatile flavour active compounds with characteristic “fishy” qualities. With the exception of trimethylamine, which has been proven to significantly taint the flavour of milk 311, it is unlikely the concentrations observed in saliva would exceed taste thresholds 312. As volatile molecules it would be possible that salivary amines could contribute to retronasal olfactory cues as the odour thresholds are lower than salivary levels 313. Alternatively, as amines were found to correlate with oral bacterial load it is possible that they may serve as markers for bacteria that may in turn generate substances that modulate taste sensitivity but are not directly detectable.

When looking at salivary differences between unrelated sensitive and insensitive oleic acid perceivers, the results for salivary SCFA concentration corroborated those of Mounayar et al. (2014) 91, who were the first to demonstrate metabolic differences in the saliva of sensitive and insensitive oleic acid detectors. They reported acetate was the most discriminatory metabolite in their dataset, and also found butyrate was overexpressed in the saliva of insensitive perceivers. This study found the same differences regarding acetate and butyrate, and also propionate and 5-aminopentanoate. This result was despite some methodological 250 differences in the oleic acid preparation and sensitivity assessment (threshold vs. suprathreshold rating). Our finding of propionate alongside acid and butyrate is not surprising as salivary SCFAs appear to be strongly correlated (Chapter 4). 5-aminopenatnoate (δ- aminovaleric acid) was not observed at significant levels in the saliva of participants from Chapter 4 although it is a microbial product of lysine metabolism and is raised in the saliva of individuals with periodontal disease 314. This study corroborates the findings of Mounayar et al. (2014) 91, with respect to salivary SCFA and oleic acid sensitivity. Alongside data from Chapter 4, these findings suggest that host-microbial interactions in the mouth play a role in sensing nutrients.

Considering the results for aspartame and sucrose together provides additional insights into the genetic and environmental influence of taste perception. Salivary compositional differences were detected between unrelated individuals and discordant twins for sucrose, but only between sensitive and insensitive discordant twins for aspartame. This finding could bely an underlying role of genetics in mediating aspartame perception that is further modulated by salivary factors. Sucrose is perceived by the molecules binding both TAS1R2 and TAS1R3 receptors whereas aspartame binds only the TAS1R2 receptor 315. Genes encoding the TAS1R3 receptor have been shown to be conserved 316, whereas those encoding the TAS1R2 receptor are known to be particularly prone to polymorphisms 317. TAS1R2 genotype has been shown not to influence sucrose perception although less frequently occurring SNPs in TAS1R3 do contribute to a small (~ 15 percent) percentage of variability in sucrose perception 318. The relative contributions of TAS1R2 and TAS1R3 to aspartame perception has not been explicitly studied. Therefore, aspartame perception may be under greater genetic influence from variation in TAS1R2 genes, and in the absence of the same underlying receptor genetics to mediate the sensation salivary composition is less critical. Our data suggest that only when the same underlying receptor-aspartame interactions occur does salivary composition further modulate response.

An additional important finding with respect to the sweet tastes was that in all groups where salivary compositional differences were detected, flow rate differences were also present. The nature of the flow rate differences was the opposite of that of salivary analyte concentrations. Insensitive perception was associated with reduced flow rate and increased analyte concentration and sensitive perception was associated with high flow rate and reduced analyte concertation. When adjusting for flow rate, the salivary outputs of the various analytes were not different. This therefore suggest salivary flow rate could be critical to influencing taste response in the presence of the same glandular protein output or oral 251 biofilm metabolite output. By serving as the solvent for salivary proteins and metabolites, salivary flow rate ultimately determines the final concentration of analytes that continually bathe the mouth and taste receptor environment. This further emphasises the multifactorial nature of taste perception with receptor genetics, oral microbial metabolic activity and endogenous salivary protein, metabolite and fluid production simultaneously contributing to tastant perception.

This study found caffeine perception has the lowest genetic influence of all the sensation tested. There also appeared to be the lowest number of salivary compositional differences between sensitive and insensitive caffeine perceivers. The only salivary analyte differing between the saliva of insensitive and sensitive caffeine perceivers was formate. In a mouse model, Hughes et al. (2017) 319, showed formate is a biomarker of inflammation and dysbiosis in the gut. Formate functions as an electron donor facilitating the development of a microbiome shift towards facultative aerobes rather than the typical obligate anaerobe population 319. Formate is also produced by and modulates growth of oral pathogens in vitro 320 321. It is unclear how exactly how salivary formate would impair the taste perception of caffeine. Bitter perception of caffeine has been shown to relate to regular dietary exposure 322. It seems likely a complex multifactorial relationship between dietary intake, taste sensitivity and the oral microbiome and metabolome exists.

It is easier to theorise about the mechanisms relating to the salivary differences seen between sensitive and insensitive astringency perceivers in the context of pre-existing literature. The precipitation of salivary proteins, particularly proline-rich proteins, by polyphenols is widely acknowledged as a critical step in conferring the sensation of astringency. Unfortunately, much of the existing literature is conducted in vitro and few studies combine sensory data on astringency with salivary analysis of the participants. Nayak & Carpenter (2008) 204, found that reducing total salivary protein by rinsing with water generally increased astringency perception whereas increasing total protein by chewing stimulation reduced astringency perception of tea. Dinnella et al. (2009) 203, found that individuals who restored their total salivary protein levels rapidly after tannic acid rinses perceived less astringency than individuals who had experienced a prolonged reduction in salivary protein. Fleming et al. (2016) 202, demonstrated an inverse correlation between unstimulated total salivary protein concentration and astringency perception of alum and tannic acid in solution. The present study found higher levels of salivary proline-rich proteins in unstimulated saliva were associated with reduced astringency perception. A number of spectral macromolecule regions were also observed to differ significantly, being more concentrated in insensitive astringency 252 perceivers. These may reflect resonances of salivary proteins, although further work would be required to understand the nature of these relationships.

7.5.3 Relationship between intra-oral sucrose catabolism and sucrose perception

Insensitive sucrose perceivers showed a net reduction in citrate output following oral exposure to sucrose solutions, whereas sensitive perceivers showed a net increased citrate output. The results from Chapters 4 and 5 show citrate enters saliva from the host and the concentration increases upon stimulation by substances such as capsaicin. Consumption of parotid salivary citrate was also observed in vitro. In Chapter 5, while capsaicin stimulation increased citrate concentration, in the same participants sucrose stimulation caused a decrease in citrate concentration. Therefore, it seems more likely that the citrate changes observed in the present experiment are a result of changes in bacterial citrate metabolism rather than changes in host endogenous citrate output. This is supported by the lack of any difference in relative flow rate change between the insensitive and sensitive sucrose perceivers, indicating no difference in the degree of salivary gland activity. Takahashi & Washio (2011) 104, analysed metabolic intermediates relating to glycolysis, pentose phosphate and citric acid cycle pathways in dental plaque following a glucose rinse. They reported a significant rise in lactate and pyruvate output from plaque, although citrate did not change. The previous work of Takahashi & Washio (2010) 94, found changes in citrate following glucose differed between oral bacterial species in in-vitro monoculture models.

Pyruvate represents an important intermediate of the glycolysis pathway, prior to conversion into lactate or entrance into the citric acid cycle. In plasma, an elevated lactate:pyruvate ratio can be used as a clinical indicator of various disease states 323-325. Lactate:pyruvate ratios have also been measured in parotid saliva as a measure of host metabolism following sucrose ingestion, as salivary lactate and pyruvate concentrations correlate with circulating plasma concentrations 326. The present study found that for insensitive sucrose perceivers a significantly higher amount of lactate per molecule of pyruvate was produced, whereas for sensitive perceivers a significantly higher amount of citrate was generated per molecule of pyruvate. These metabolic differences may be indicative of differences in the oral microflora of sensitive and insensitive sucrose perceivers. Increased citric acid cycle activity following sucrose exposure in sensitive perceivers may reflect a more aerobic or facultative oral microflora whereas increased lactate conversion, detected in insensitive sucrose perceivers, may indicate a more anaerobic oral microflora. Of particular relevance to this observation may be the Streptococci composition of these participants’ respective oral microbiomes. 253

Streptococci, including S. mutans, one of the major oral lactate producers, feature absent or diminished citric acid cycles which are not primarily involved in energy production 264,327.

A particular limitation of the current study is in the fact that the metabolomic analyses are limited to static observations post control and post sucrose and in reality, the metabolic events are highly dynamic. It is quite possible that important information may be not be captured in this experimental design. Despite this, the study demonstrates potential for metabolomic profiling of saliva to reveal new insights into oral host-microbiome interactions, and how these might relate more widely to health. Studying net metabolic activity of host- microbiome environments has recently been identified as arguably more important than more established analysis of microbial communities by sequencing technologies 97.

7.5.4 Summary

This chapter found that salivary metabolite composition is largely dictated by environmental, rather than genetic, influence. Taste perception was generally also only weakly heritable although heritability was tastant specific. Conversely, FPD was found to be a highly heritable trait. This chapter also identifies several key findings regarding the association between salivary composition and taste sensitivity as measured by suprathreshold intensity assessment. Firstly, similarities in salivary composition were associated with sensitivity to sucrose and oleic acid in both unrelated individuals and twin pairs, indicating saliva composition is an important oral environmental factor in the perception of these substances. Salivary composition and taste perception appear to be associated in a tastant-specific fashion as the relationships observed with sucrose and oleic acid were not found for the perception of aspartame and caffeine. This suggests factors beyond genetics and oral environment, e.g. habitual exposure, may be more important in shaping perceived intensity of these taste modalities. Salivary flow rate was also found to be an important consideration in shaping sensory response. A proposed explanation for this observation is that saliva serves as the carrier for biofilm-generated metabolites, and hence dictates the concentration of metabolites bathing and interacting with taste receptors. Finally, there was evidence suggesting that the individual response to sucrose intensity is related to the intra-oral catabolism of the sucrose. Collectively, these findings indicate that saliva-taste interactions are multifactorial and tastant specific.

254

Chapter 8: General discussion

8.1 Insight into the salivary metabolome

The overarching aim of this thesis was to investigate the relationship between salivary composition, particularly metabolic composition using 1H-NMR spectroscopy, and taste perception. In doing so, several important aspects of the salivary metabolome were investigated. The terms “metabolomics” and “metabonomics” can vary in their intended definition and are sometimes used interchangeably. Using the definitions of Nicholson (2006), the former term refers to the “quantitative analysis of all the metabolites of a specified biological sample”. Metabonomics is defined as “the quantitative measurement of the multiparametric time-related metabolic responses of a complex (multicellular) system to a pathophysiological intervention…” 250. Based on this distinction, much of the literature looking at salivary metabolic composition could be considered as metabolomics studies, given the focus on identifying and measuring the metabolites in a sample 60,78. Some literature looks as pathological states 71, and some looks at physiological stimuli 81, hence could be defined as metabonomic studies. Nevertheless, there was considerable scope for further metabonomic study of the dynamic nature of the salivary metabolic content as well changes brought about by physiological stimulation.

The “complex (multicellular) system”, as defined by Nicholson (2006) 250, must be extrapolated to all metabolically active cells contributing to the sample in question. This work indicated that, at least by metabolite abundance, the majority of the salivary metabolome is microbially derived. Not only are oral bacteria a primary source of metabolites in the mouth but they exist in a complex metabolic balance with their host. This work found that the net metabolism of oral microbes differs based on nutrient availability. In the absence of exogenous nutrients, oral bacteria readily catabolise salivary proteins, shaping the salivary metabolome with SCFAs, amino acids and their respective degradation products. Such a finding likely explains why many metabolite concentrations appear to increase following sleep, as reported by Wallner-Liebmann et al. (2016) 79. In-vitro study also found that, as for proteins, many endogenous metabolites (e.g. citrate, urea and lactate) in saliva will also be consumed by bacteria when no alternative source is available. Importantly, the fluid secreted by salivary glands in not the only source of detectable metabolites in saliva. GCF, although representing a small proportion of oral fluid by volume, appears to concentrate taurine in saliva in concentrations far in excess of those that would be predicted based on simple exudation from serum. The proposed mechanism of taurine entry into WMS is via 255 concentration in immune cells. Choline also appears to be derived from buccal epithelial cells shed into saliva. These findings support those of Mikkonen et al., (2013) 71,who found inverse associations between taurine and choline concertation and salivary flow rates.

Host-microbiome interactions were also found to shape the salivary metabolome following exposure to exogenous sucrose. While oral microbial catabolism of sucrose has been recognised for decades, particularly with emphasis on lactate production 253, using a metabolomics approach is somewhat novel. While not the first to use such as approach 94, our study found that oral sucrose exposure raised not only lactate and pyruvate concentrations, but also those of alanine and acetoin. Alanine and acetoin can both be generated from microbial conversion of pyruvate 254,328. Thus, pyruvate is a key metabolic intermediate in the oral cavity. Furthermore, differences in the fate of pyruvate produced by oral catabolism of sucrose were provisionally linked to host taste sensitivity. Insensitive perceivers were found to have more efficient lactate production, and reduced pyruvate conversion to citrate than sensitive perceivers. At present it cannot be claimed that the association observed was causative, although the metabolic differences presumably reflect differences in the net microbial composition which may be responsible for altering taste perception by other means such as production of uncharacterised peptidic molecules.

Aside from the microbial catabolism of sucrose, the other tastants analysed generally had a fairly minimal effect on salivary metabolite content. This was true for caffeine and menthol. Capsaicin had a pronounced effect on salivary protein composition and extensional rheology and also caused increased citrate secretion. The extent of citrate secretion also correlated with the capillary breakup time of the secreted saliva. Citrate was one of the more interesting metabolites detected throughout this work. While it is present in glandular saliva, citrate secretion appears to be stimulus specific, with capsaicin increasing secretion whereas sucrose does not despite both stimuli causing approximately equal flow rate increases. Potential for future work concerning salivary citrate is discussed below.

Assessing the relationship between taste sensitivity and salivary metabolite composition was approached from two different angles. Firstly, a twin study model was used to study the oral environment as a confounder of taste sensitivity. The data set was divided to look at differences in salivary metabolite and protein composition between unrelated individuals sensitive and insensitive to the taste stimuli (sucrose, aspartame, caffeine, black tea and oleic acid). Then the dataset was used to compare twin-pairs discordant for taste sensitivity and concordant for sensitivity. For a number of stimuli, associations were found the salivary 256 metabolite content of insensitive and sensitive perceivers from the unrelated individuals and discordant twin groups. No such differences were found between concordant tasting twins. These differences were notable for sucrose and oleic acid, in particular. In the former, citrate and dimethylamine were detected as discriminatory metabolites, in the latter SCFAs were discriminatory (although among discordant twins the p-values only approached 0.05). Taurine was a discriminatory metabolite common to both stimuli. Importantly, this work also found that FPD (which is controversially associated with taste sensitivity) was very highly heritable. FPD therefore did not seem to have much bearing on explaining the taste differences observed, given discordant twins would have essentially the same FPD. A further key finding was that adjusting metabolite concentration for salivary flow rate tended to negate the compositional differences seen between sensitive and insensitive perceivers. Therefore, salivary flow rate may be a key variable in determining the salivary concentration and thus putative biological actions of bacterially generated metabolites or other products. These processes are summarised graphically in Figure 8.1.

257

Figure 8.1: Summary of the process involved in shaping the salivary metabolome and proposed interactions for how these net events might alter taste perception. Saliva entering the oral cavity is rich in proteins and endogenous metabolites. Metabolites such as taurine can enter the mouth via GCF. Microbial breakdown of these proteins leads to metabolite production. Similar actions occur with ingested food, where sugars are catabolised by oral bacteria. The overall salivary flow rate will then dictate the concentrations of metabolites produced by oral bacteria or entering via GCF. By continual bathing of taste receptors on the tongue, metabolites may interact with taste receptors by an as yet undetermined mechanism. Alternatively, biofilm may also produce other products that may affect taste receptors, with their metabolite products being indicators of this type of microbial activity.

258

Given the novelty of metabolic profiling of saliva with respect to taste perception, there are few publications to compare our findings against. Nevertheless, our findings agree with one of the only preceding reports, that of Mounayar et al., (2014) 91, focusing on oleic acid perception. Both works found that insensitive perceivers of oleic acid have higher salivary acetate and butyrate. Our results also found propionate existed in the same relationship, which is unsurprising given how closely correlated salivary SCFA concentrations are. Mounayar et al. (2014) 91, also reported differences in other spectral regions possibly corresponding to basal fatty acids in saliva although we did not corroborate this finding. This may be due to technical differences in the spectral acquisition. The fact that the same relationship between salivary SCFA and oleic acid sensitivity was reported despite certain protocol differences aids speculation of an explanation for the findings. As SCFAs are volatile substances, they may contribute to baseline olfaction via the retronasal route. Butyrate producing nasal bacteria have recently been implicated with impaired olfaction 329. Hence high salivary SCFA concentrations could mask olfactory cues associated with the net sensory response to oleic acid. The use of nose-clips by Mounayar et al. (2014) 91, however, negates this explanation and suggests a purely oral mechanism.

Our finding linking astringency perception of black tea with salivary proline-rich protein abundance also adds to the existing literature base on this topic. While various studies have implicated salivary protein in association with astringency perception 202,204,205, between-group differences in baseline salivary PRP abundance in-vivo has not been directly demonstrated. Reconciling the protein and NMR data in relation to black tea astringency perception also raises an interesting question about the utility of salivary macromolecule profile measured by NMR. Salivary macromolecules are typically removed from NMR signals either by physical or electronic means, although inevitably some residual signal remains. Macromolecule profiling by NMR spectroscopy of fluids such as plasma has yielded useful data, and perhaps with further study such an approach could be applied to salivary macromolecules.

Associations between salivary metabolite composition and taste perception were also found when measuring recognition/detection thresholds. In particular, urea was found to positively associate with the detection threshold of sucrose. This was somewhat unique as associations between salivary metabolites and taste were otherwise inverse i.e. higher metabolite concentration was associated with reduced taste sensitivity. The relative concentrations of mean salivary urea (0.45 mM) were somewhat approximate to those of the lower concentrations of sucrose in the experiment (1 mM). Salivary urea has previously been inversely associated with reduced intensity perception in renal disease patients for sour and 259 bitter suprathreshold taste intensity 201, although study with sucrose detection thresholds has not previously been reported. As urea is a taste active (bitter) molecule there could be interactions at low concentrations between urea and sucrose, however the salivary urea concentrations are considerably below threshold 330. Urea was also a unique metabolite with respect to the fact that salivary concentrations inversely correlated with bacterial load. Therefore, salivary urea may simply be a biomarker of reduced bacterial activity, and consequently reduced microbial products (e.g. the small proteins, < 50 amino acid, described by Sberro et al. 2019 331) that diminish sucrose sensitivity.

Other data relevant to salivary metabonomics yielded by this work includes the analysis of genetic composition to the salivary metabolic profile as well as the inter- and intra-individual stability of salivary metabolite profile. Salivary metabolome did not appear to have a particularly strong genetic influence. This might be expected given the significant contribution of the oral microbiome in shaping the salivary metabolome. The oral microbiome has previously been shown to differ between twins, with differences increasing with time and separate environment in adulthood. Our twin participants were all adults no longer sharing a living environment. An interesting comparison of salivary metabolome might be made amongst a younger twin cohort, still sharing the same environment. With respect to the inter- and intra- individual stability of salivary metabolic profile, intra-individual profiles were less variable than inter-individual profile. This was in keeping with the results of Wallner-Liebmann et al., (2016) 79. Importantly, however, concentrations of individual metabolites were found to exhibit varying degrees of stability. Certain metabolites such as lactate, propionate and formate were found to vary considerably within the same participant. Knowledge of the natural variability of salivary metabolite concentration in health is particularly important given the trend for salivary metabolomic profiling in biomarker discovery, as highly variable metabolites might be of limited use as biomarkers, or at least merit repeated sampling to improve diagnostic accuracy.

8.2 Limitations of presented work

The experimental work presented in this thesis has several limitations. One of the major caveats regarding claims that can be made on the basis of the current results is that the work presented is largely based on associations between salivary composition and oral perception. While this is not an uncommon approach in studies relating taste perception to wider biological parameters, it must be borne in mind that the relationships reported cannot be

260 used to state a direct causality. Possible approaches to build on the current work to address this issue in future have been detailed in the future work section.

Additional limitations concern the specific analytical techniques used. The work presented indicates that there is no single salivary analyte (of those measured) that has a universal association with sensory perception. Rather, the relationships appear to multifactorial and tastant-specific. There is undoubtedly a high degree of complexity involved in the oral events before, during and after tastant-receptor interactions that explain inter-individual differences in perception. Due to the complex biochemical nature of saliva, the amount of information that could theoretically be obtained from a single sample is vast. The metabolic activity of saliva and oral microflora has been identified as an important area of research 97. This work predominantly used 1H-NMR spectroscopy to profile salivary metabolites with a combination of targeted and untargeted analyses. Metabolomic data was complimented by analysis of host salivary proteins, extensional rheology and conventional microbiological culture. Sensory data gathered was primarily suprathreshold intensity measurements, and some detection/recognition threshold data.

It is likely that there may exist relationships between saliva and taste perception that would go undetected using the analytical techniques described. For example, there are a large number of small molecules and peptides that may influence taste perception that would not be measurable by 1H-NMR spectroscopy but would be readily detected by MS platforms. Indeed, only recently it has been reported that the gut microbiome produces hundreds of previously uncharacterised small proteins whose functions remain to be determined 331. The possibility remains that the metabolites associated with taste in this study are simply metabolic markers of microbial communities which are simultaneously producing bioactive small proteins. While approaches to salivary analytics using combined analytical platforms do exist 91,332, these remain the exception. As “omics” technologies continually become more widespread and accessible it would seem logical that combined approaches will maximise the data gathered per sample.

Technical limitations due to analytical methodology also extend to the physical and sensory measurements conducted. While data was gathered from separate experiments using suprathreshold intensity and detection/recognition threshold assessment, several other techniques may have been applicable. Firstly, rather than measure intensity as the primary perception, hedonic response (liking/disliking) may have been a useful outcome to record. Hedonic data may also be more applicable than intensity data when making inferences about 261 dietary consumption of sugar 252. A further consideration relating to sensory data is the notion of temporal change when assessing taste. In some ways this was addressed with respect to intra-individual changes in saliva and taste perception in Chapter 6, however temporal change in sensory perception in the mouth is also dynamic. Time-intensity data may therefore have been a useful information source, or techniques such as temporal dominance of sensation (TDS) would even allow measurement of multiple sensations simultaneously 333 334. This may have been useful to gather data on mouthfeel alongside taste perception for the various tastants.

Such sensory approaches could have complemented additional measurement of saliva’s physical properties. While extensional rheology is often used to make inferences about the function and surface coating ability of saliva 189,190,335, tribology may be more directly relevant as the surface film in the oral cavity is known to be important in many oral sensory processes 336 246. Systems have been constructed for modelling oral tribology 337,338, however such analyses were beyond the scope of this work, particularly considering tastants were administered as liquid solutions. The findings from Chapter 5, illustrating the rapid capsaicin- stimulated changes in salivary rheology and composition underline the dynamic nature of taste-saliva relationships. These are summarised in Figure 8.2. Extrapolating these finding to a real food, with repeated exposure to the tastant would require investigation of both dynamic sensory perception and analyses of salivary surface properties including tribology, film thickness, protein composition.

Figure 8.2: Theoretical outline of the dynamic events occurring on exposure to tastants such as capsaicin. Following initial tastant-receptor interaction, tastant-stimulated salivary changes may subsequently interfere with the temporal perception of the stimulus.

Given the emerging interest in the salivary film which coats oral mucosal surfaces 246, more in depth study of this film coating is likely to yield important information in future. Quantification of biological films at the nanoscale level can be achieved using quartz crystal 262 microbalance with dissipation (QCM-D). This technique allows measurement of the mass of a biological film during formation (in vitro) as well as characterisation of the viscoelastic properties of the film 339-341 . The technique has been applied to measuring film formation on dental materials as well as looking at interactions between saliva and polyphenols 340,342, aiding in the understanding of the process of astringency. Further work looking at interactions with other tastant molecules would be highly informative in understanding tastant behaviour at the surface salivary film, particularly if differences could be found in the behaviour of saliva collected from different individuals.

8.3 Future work

The work presented in this thesis has identified several avenues that would merit further exploration regarding the roles of salivary metabolites. Potential hypotheses generated by this thesis as well as suggestions for experimental protocols to begin addressing these hypotheses are discussed in this section.

8.3.1 The role of salivary SCFAs in oral health.

It is well recognised that the bacterial generation of SCFAs in the gut is implicated in a range of physiological functions. These functions include modulating host metabolism, immune signalling and in the case of butyrate, anti-tumorigenic properties 343 344 345. Whether an analogous role is exerted on epithelial cells of the oral cavity by orally generated SCFAs is a question for which little published work exists. One study of TR146 cells, a common buccal epithelial cell line derived from an oral squamous cell carcinoma, found that acetate concentrations of 20 mM or lower did not impair cell growth 346. There is currently wide scope for investigation into potential oral health functions of SCFAs. A further role of butyrate in gastrointestinal health is in the maintenance of colonic barrier function via upregulation of tight junction expression 347 348. As evidenced by the findings of Mounayar et al., (2014) 91, and the results presented in Chapter 7, there appears to be an association between salivary concentrations of SCFAs and oral sensitivity to oleic acid. Interactions between oral mucosal cells, salivary SCFAs and oral expression of candidate fat-detecting receptors, such as GPR120 349, may reveal useful insights into the role of host-microbiome interactions in fat detection.

An experiment is described in appendix A showing proof of concept and preliminary data for the interactions of acetate, butyrate and propionate on the viability of TR146 cells. As shown in appendix A Figure 1, 5 mM butyrate, 5 mM propionate and to a lesser extent 1 mM

263 butyrate all inhibited the proliferation of TR146 cells. Acetate had no bearing on the proliferation of these cells, in line with the work of De Ryck et al. (2014) 346. Typical mean salivary concentrations of propionate and butyrate are around 0.5 mM and 0.1 mM, respectively, as reported in Chapters 6 and 7. The concentrations of SCFAs that inhibited TR146 cells in this experiment are therefore likely in excess of those that would be found in the saliva on average by as much as ten-fold. This preliminary finding does, however, suggest that the antitumorigenic properties of butyrate reported in the gut could also apply to oral squamous cell carcinomas. There is scope for refining of the concentrations and incubation times of SCFAs delivered and developing more accurate cell models of the events in the oral cavity. Interestingly the lowest concentration (0.2 mM) of butyrate appeared to increase TR146 proliferation, particularly at the earlier timepoints, however this finding was not significant.

An attempt was made to assess whether TR146 cells would respond to oleic acid and whether the SCFAs would interact with this process in line with the observed in-vivo associations. Oleic acid caused artefacts in the intracellular calcium measurements thus data could not be acquired. Oleic acid has been used to elicit intracellular calcium responses in cell lines, measured using a monochromator and CCD to capture fluorescence, so this issue would not likely be insurmountable 350.

8.3.2 Mechanism of action for salivary metabolite/taste receptor interactions

The results from Chapters 6 and 7 suggested some relationships between salivary metabolite concentrations and taste perception, primarily for the sweet taste of sucrose. Chapter 7 found taurine was one of several metabolites that was inversely associated with sucrose sensitivity. Taurine was the metabolite with the greatest salivary concentration difference between sensitive and insensitive perceivers, on average 2.24-fold more concentrated in the saliva of insensitive perceivers. Chapter 6 found that salivary urea was positively associated with the detection threshold of sucrose, which was a finding consistent on repeating the same trial with the same participants on separate weeks.

As detailed in the previous section, the data presented in this thesis generally describes associations and these alone cannot be used to infer causality. The investigation of a causative relationship between salivary metabolites such as urea and taurine, in addition to potential mechanisms of actions of any such causation is a logical next step stemming from the current work. Appendix B describes some preliminary experiments on this topic. An in-vitro 264 intracellular calcium signalling experiment assessing the effects of taurine and urea on the response of TR146 to glucose was conducted. This was followed by an in-vivo experiment in which participants rated the intensity of sucrose and caffeine before and after conditioning the mouths with solutions of taurine and urea.

As shown in Appendix B Figure 1, taurine and urea conditioning of the mouths did not significantly alter intensity perception of sucrose or caffeine relative to water. An increasing trend in mean intensity was observed, however this could simply reflect an intensifying perception of repeated tastant exposure, and further controls would be necessary to adjust for this. The preliminary data suggests that should metabolites such as urea and taurine have direct action on sensory perception, any such effects presumably require prolonged exposure of metabolites to taste receptors. As such, the in-vitro experiment exposed TR146 to taurine and urea for 24 hours. The results shown in Appendix B Figure 2 indicate that 5 mM taurine and urea had no effect on TR146 response to glucose, however 1 mM taurine and urea caused a diminished response approaching significance. At the lowest concentration investigated (0.2 mM) taurine had no effect however 0.2 mM urea had a highly significant inhibitory effect. While this finding conflicts with the in-vivo association from Chapter 6, suggesting urea was associated with enhanced sucrose detection, salivary urea has been found to inhibit taste in other literature 201. The cell culture model used in these experiments was imperfect. TR146 cells did not respond to sucrose, thus the observed glucose response cannot be mediated by TAS1R2/3. A more appropriate in-vitro model could be created for further testing of metabolite-tastant interactions, for example using cultured human taste receptor cells351.

8.3.3 The role of citrate in the physical properties of saliva

The results of Chapters 4 and 5 found citrate to be an interesting salivary metabolite. As shown in Chapter 4, citrate enters saliva from the host but is also consumed by oral bacteria in the absence of dietary substrate. Chapter 5 found salivary citrate concentration increased upon capsaicin stimulation, but not sucrose stimulation despite both stimuli increasing flow rate. Citrate was also the only salivary metabolite that correlated with the physical properties of saliva in capsaicin-stimulated samples. The potential for TRP agonists, in particular capsaicin, as therapeutic agents for the management of dry mouth has been raised previously 190. The results of Chapter 5 suggest citrate could have a role in conferring the beneficial physical properties of saliva following TRP stimulation.

265

Humans express different types of citrate transporters, those of the SLC13 gene family (plasma membrane type) and those of the SLC25 family, associated with mitochondrial transport. The SLC25 family encompasses the primary citrate transporters in the prostate 352, where defective citrate transport is associated with malignancy 353. The main citrate transporter described in human salivary glands is the sodium-dependant NaCT transporter, encoded by SLC13A5 354,355. Unlike other SLC (solute carrier) proteins which can transport a range of molecules, the NaCT is specialised for citrate transport and to a lesser degree succinate 356. The NaCT receptor does not appear to have been directly studied with respect to any salivary gland pathologies.

A first step in further elucidating the role of citrate in salivary function could be demonstrating differences in the expression of these transporters between the salivary glands of individuals with salivary gland pathology and controls. Obtaining salivary gland tissue from individuals with suspected pathology such as Sjögren’s syndrome could be possible as minor glands are biopsied in the diagnosis of the condition 357. However, there is the possibility that any citrate transporter defects of the major glands would not be reflected in minor glands. Furthermore, it would be difficult to obtain control glands from healthy volunteers due to the invasion nature of collection. It may be possible to use an animal model, however it may difficult to be sure potential underlying pathology would reflect that experienced by humans. Nevertheless, a knockout model for NaCT in mice has been studied with respect to certain aspects of physiology, although data for salivary function is lacking 356. Should citrate transport be identified as causative of adverse salivary rheology the respective SLC protein could be targeted pharmacologically 358.

8.4 Conclusion

This thesis aimed to determine whether salivary components, particularly metabolites, were associated with taste perception and/or other oral sensations. This aim was achieved, in that the findings of this thesis collectively indicate that saliva has a multifactorial, stimulus-specific influence on taste and oral perception. Certain salivary metabolites were associated with taste sensitivity, as measured by suprathreshold intensity perception and recognition/detection thresholds. Furthermore, there is strong evidence for the role of host- microbiome interactions in shaping individual salivary metabolite profiles. The main processes by which intra-oral host-microbiome interactions appear to associate with taste perception are summarised diagrammatically in Figure 8.1.

266

References

1 Holmberg, K. V. & Hoffman, M. P. Anatomy, biogenesis and regeneration of salivary glands. Monogr. Oral Sci. 24, 1-13, doi:10.1159/000358776 (2014). 2 Holsinger, F. C. & Bui, D. T. in Salivary gland disorders 1-16 (Springer, 2007). 3 Aps, J. K. & Martens, L. C. The physiology of saliva and transfer of drugs into saliva. Forensic Sci. Int. 150, 119-131 (2005). 4 Earthslab. https://www.earthslab.com/anatomy/salivary-glands/, 2019). 5 Amano, O., Mizobe, K., Bando, Y. & Sakiyama, K. Anatomy and histology of rodent and human major salivary glands—Overview of the japan salivary gland society-sponsored workshop—. Acta histochemica et cytochemica 45, 241-250 (2012). 6 Humphrey, S. P. & Williamson, R. T. A review of saliva: Normal composition, flow, and function. The Journal of Prosthetic Dentistry 85, 162-169, doi:https://doi.org/10.1067/mpr.2001.113778 (2001). 7 Baum, B. J. Principles of saliva secretion. Ann. N. Y. Acad. Sci. 694, 17-23 (1993). 8 Proctor, G. B. The physiology of salivary secretion. Periodontol. 2000 70, 11-25 (2016). 9 Catalán, M. A., Nakamoto, T. & Melvin, J. E. The salivary gland fluid secretion mechanism. The Journal of Medical Investigation 56, 192-196 (2009). 10 Turner, R. J. & Sugiya, H. Understanding salivary fluid and protein secretion. Oral Dis. 8, 3-11 (2002). 11 Garrett, J. The proper role of nerves in salivary secretion: a review. J. Dent. Res. 66, 387-397 (1987). 12 Iorgulescu, G. Saliva between normal and pathological. Important factors in determining systemic and oral health. J. Med. Life 2, 303-307 (2009). 13 Dawes, C. Circadian rhythms in human salivary flow rate and composition. The Journal of physiology 220, 529-545 (1972). 14 de Almeida, P. D. V., Gregio, A., Machado, M., De Lima, A. & Azevedo, L. R. Saliva composition and functions: a comprehensive review. J. Contemp. Dent. Pract. 9, 72-80 (2008). 15 Pijpe, J. et al. Progression of salivary gland dysfunction in patients with Sjögren’s syndrome. Ann. Rheum. Dis. 66, 107-112 (2007). 16 Dodds, M., Roland, S., Edgar, M. & Thornhill, M. Saliva A review of its role in maintaining oral health and preventing dental disease. Bdj Team 2, 15123, doi:10.1038/bdjteam.2015.123 (2015). 17 Dawes, C. The effects of flow rate and duration of stimulation on the concentrations of protein and the main electrolytes in human parotid saliva. Arch. Oral Biol. 14, 277- 294 (1969). 18 Larsen, M., Jensen, A., Madsen, D. & Pearce, E. Individual variations of pH, buffer capacity, and concentrations of calcium and phosphate in unstimulated whole saliva. Arch. Oral Biol. 44, 111-117 (1999). 19 Dawes, C. Salivary flow patterns and the health of hard and soft oral tissues. The Journal of the American Dental Association 139, 18S-24S (2008). 20 Carpenter, G. H. The secretion, components, and properties of saliva. Annu. Rev. Food Sci. Technol. 4, 267-276 (2013). 21 Williamson, M. P. The structure and function of proline-rich regions in proteins. Biochem. J. 297, 249 (1994). 22 Manconi, B. et al. The intriguing heterogeneity of human salivary proline-rich proteins: Short title: Salivary proline-rich protein species. J. Proteomics 134, 47-56 (2016). 23 Bobek, L. A. & Levine, M. J. Cystatins—inhibitors of cysteine proteinases. Crit. Rev. Oral Biol. Med. 3, 307-332 (1992).

267

24 Schlesinger, D. H. & Hay, D. Complete covalent structure of statherin, a tyrosine-rich acidic peptide which inhibits calcium phosphate precipitation from human parotid saliva. J. Biol. Chem. 252, 1689-1695 (1977). 25 Raj, P. A., Johnsson, M., Levine, M. J. & Nancollas, G. H. Salivary statherin. Dependence on sequence, charge, hydrogen bonding potency, and helical conformation for adsorption to hydroxyapatite and inhibition of mineralization. J. Biol. Chem. 267, 5968-5976 (1992). 26 Xu, T., Levitz, S., Diamond, R. & Oppenheim, F. Anticandidal activity of major human salivary histatins. Infect. Immun. 59, 2549-2554 (1991). 27 Frenkel, E. S. & Ribbeck, K. Salivary mucins in host defense and disease prevention. J. Oral Microbiol. 7, 29759 (2015). 28 Oppenheim, F. G., Salih, E., Siqueira, W. L., Zhang, W. & Helmerhorst, E. J. Salivary proteome and its genetic polymorphisms. Ann. N. Y. Acad. Sci. 1098, 22-50 (2007). 29 Tabak, L. A., Levine, M. J., Mandel, I. D. & Ellison, S. A. Role of salivary mucins in the protection of the oral cavity. J. Oral Pathol. Med. 11, 1-17 (1982). 30 Tabak, L. A. In defense of the oral cavity: structure, biosynthesis, and function of salivary mucins. Annu. Rev. Physiol. 57, 547-564 (1995). 31 Gibbins, H., Proctor, G., Yakubov, G., Wilson, S. & Carpenter, G. Concentration of salivary protective proteins within the bound oral mucosal pellicle. Oral Dis. 20, 707- 713 (2014). 32 Helmerhorst, E. J. & Oppenheim, F. G. Saliva: a Dynamic Proteome. J. Dent. Res. 86, 680-693, doi:10.1177/154405910708600802 (2007). 33 Sarkar, A., Xu, F. & Lee, S. Human saliva and model saliva at bulk to adsorbed phases– similarities and differences. Adv. Colloid Interface Sci., 102034 (2019). 34 Xu, F. et al. A Self‐Assembled Binary Protein Model Explains High‐Performance Salivary Lubrication from Macro to Nanoscale. Advanced Materials Interfaces 7, 1901549 (2020). 35 Yakubov, G. E., Macakova, L., Wilson, S., Windust, J. H. C. & Stokes, J. R. Aqueous lubrication by fractionated salivary proteins: Synergistic interaction of mucin polymer brush with low molecular weight macromolecules. Tribology International 89, 34-45, doi:https://doi.org/10.1016/j.triboint.2014.12.025 (2015). 36 Proctor, G. B., Hamdan, S., Carpenter, G. H. & Wilde, P. A statherin and calcium enriched layer at the air interface of human parotid saliva. Biochem. J. 389, 111-116 (2005). 37 Chicharro, J. L. et al. Trace elements and electrolytes in human resting mixed saliva after exercise. Br. J. Sports Med. 33, 204-207 (1999). 38 Cleto, D. A. M., Andrello, A. C., Netto, Í. J. V. & Appoloni, C. R. Analysis of saliva and gingival crevice fluid by total reflection X‐ray fluorescence (TXRF). X‐Ray Spectrometry 45, 220-224 (2016). 39 Schweigel, H., Wicht, M. & Schwendicke, F. Salivary and pellicle proteome: a datamining analysis. Sci. Rep. 6, 38882 (2016). 40 Jehmlich, N. et al. Quantitative analysis of the intra‐and inter‐subject variability of the whole salivary proteome. J. Periodontal Res. 48, 392-403 (2013). 41 Prodan, A. et al. A Study of the Variation in the Salivary Peptide Profiles of Young Healthy Adults Acquired Using MALDI-TOF MS. PLoS One 11, e0156707, doi:10.1371/journal.pone.0156707 (2016). 42 Yan, W. et al. Systematic comparison of the human saliva and plasma proteomes. Proteomics–clinical applications 3, 116-134 (2009). 43 McManus, L., Marze, B. & Schiess, A. Deficiency of salivary PAF in edentulous individuals. J. Periodontal Res. 25, 347-351 (1990). 44 Siqueira, W. L. & Dawes, C. The salivary proteome: challenges and perspectives. PROTEOMICS–Clinical Applications 5, 575-579 (2011).

268

45 Siqueira, W. L., Zhang, W., Helmerhorst, E. J., Gygi, S. P. & Oppenheim, F. G. Identification of Protein Components in in vivo Human Acquired Enamel Pellicle Using LC−ESI−MS/MS. J. Proteome Res. 6, 2152-2160, doi:10.1021/pr060580k (2007). 46 Goodson, J. M. Gingival crevice fluid flow. Periodontol. 2000 31, 43-54 (2003). 47 Uitto, V. J. Gingival crevice fluid–an introduction. Periodontol. 2000 31, 9-11 (2003). 48 Spielman, A. I., D'Abundo, S., Field, R. B. & Schmale, H. Protein Analysis of Human von Ebner Saliva and a Method for its Collection from the Foliate Papillae. J. Dent. Res. 72, 1331-1335, doi:10.1177/00220345930720091301 (1993). 49 Mukherjee, M. Human digestive and metabolic lipases—a brief review. Journal of Molecular B: Enzymatic 22, 369-376 (2003). 50 Neyraud, E., Palicki, O., Schwartz, C., Nicklaus, S. & Feron, G. Variability of human saliva composition: possible relationships with fat perception and liking. Arch. Oral Biol. 57, 556-566 (2012). 51 Baker, J. L., Bor, B., Agnello, M., Shi, W. & He, X. Ecology of the oral microbiome: beyond bacteria. Trends Microbiol. 25, 362-374 (2017). 52 Wade, W. G. The oral microbiome in health and disease. Pharmacol. Res. 69, 137-143 (2013). 53 Dewhirst, F. E. et al. The human oral microbiome. J. Bacteriol. 192, 5002-5017 (2010). 54 Ruhl, S. The scientific exploration of saliva in the post-proteomic era: from database back to basic function. Expert review of proteomics 9, 85-96, doi:10.1586/epr.11.80 (2012). 55 Barker, S. & Summerson, W. H. The colorimetric determination of lactic acid in biological material. J. Biol. Chem. 138, 535-554 (1941). 56 HENCH, P. S. & ALDRICH, M. THE CONCENTRATION OF UREA IN SALIVA. JAMA 79, 1409-1412, doi:10.1001/jama.1922.02640170035011 (1922). 57 Shulman, E. H. & Robinson, H. B. Salivary citrate content and erosion of the teeth. J. Dent. Res. 27, 541-544 (1948). 58 Garrote, G. L., Abraham, A. G. & Rumbo, M. Is lactate an undervalued functional component of fermented food products? Front. Microbiol. 6, doi:10.3389/fmicb.2015.00629 (2015). 59 Brooks, G. A. The science and translation of lactate shuttle theory. Cell Metab. 27, 757-785 (2018). 60 Dame, Z. T. et al. The human saliva metabolome. Metabolomics 11, 1864-1883 (2015). 61 Sugimoto, M., Wong, D. T., Hirayama, A., Soga, T. & Tomita, M. Capillary electrophoresis mass spectrometry-based saliva metabolomics identified oral, breast and pancreatic cancer-specific profiles. Metabolomics 6, 78-95 (2010). 62 Wei, J. et al. Salivary metabolite signatures of oral cancer and leukoplakia. Int. J. Cancer 129, 2207-2217 (2011). 63 Wang, Q., Gao, P., Wang, X. & Duan, Y. The early diagnosis and monitoring of squamous cell carcinoma via saliva metabolomics. Sci. Rep. 4, 6802 (2014). 64 Ishikawa, S. et al. Identification of salivary metabolomic biomarkers for oral cancer screening. Sci. Rep. 6, 31520 (2016). 65 Fidalgo, T. K. et al. Salivary metabolite signatures of children with and without dental caries lesions. Metabolomics 9, 657-666 (2013). 66 Pereira, J. et al. Saliva NMR metabolomics: analytical issues in paediatric oral health research. Oral Dis. (2019). 67 Liebsch, C. et al. The Saliva Metabolome in Association to Oral Health Status. J. Dent. Res., 0022034519842853 (2019). 68 Aimetti, M., Cacciatore, S., Graziano, A. & Tenori, L. Metabonomic analysis of saliva reveals generalized chronic periodontitis signature. Metabolomics 8, 465-474 (2012). 69 Romano, F. et al. Analysis of salivary phenotypes of generalized aggressive and chronic periodontitis through nuclear magnetic resonance‐based metabolomics. J. Periodontol. 89, 1452-1460 (2018). 269

70 Kageyama, G. et al. Metabolomics analysis of saliva from patients with primary Sjögren's syndrome. Clin. Exp. Immunol. 182, 149-153 (2015). 71 Mikkonen, J. J. et al. Metabolic profiling of saliva in patients with primary Sjögren's syndrome. Metabolomics 3, 1 (2013). 72 Tsuruoka, M. et al. Capillary electrophoresis‐mass spectrometry‐based metabolome analysis of serum and saliva from neurodegenerative dementia patients. Electrophoresis 34, 2865-2872 (2013). 73 Figueira, J. et al. NMR analysis of the human saliva metabolome distinguishes dementia patients from matched controls. Mol. Biosyst. 12, 2562-2571 (2016). 74 Liang, Q. et al. Metabolomics-based screening of salivary biomarkers for early diagnosis of Alzheimer's disease. Rsc Advances 5, 96074-96079 (2015). 75 Yilmaz, A. et al. Diagnostic biomarkers of Alzheimer’s disease as identified in saliva using 1H NMR-based metabolomics. J. Alzheimers Dis. 58, 355-359 (2017). 76 Dallmann, R., Viola, A. U., Tarokh, L., Cajochen, C. & Brown, S. A. The human circadian metabolome. Proceedings of the National Academy of Sciences 109, 2625-2629 (2012). 77 Kawanishi, N. et al. Effects of inter-day and intra-day variation on salivary metabolomic profiles. Clin. Chim. Acta 489, 41-48 (2019). 78 Bertram, H. C., Eggers, N. & Eller, N. Potential of human saliva for nuclear magnetic resonance-based metabolomics and for health-related biomarker identification. Anal. Chem. 81, 9188-9193 (2009). 79 Wallner-Liebmann, S. et al. Individual Human Metabolic Phenotype Analyzed by 1H NMR of Saliva Samples. J. Proteome Res. 15, 1787-1793, doi:10.1021/acs.jproteome.5b01060 (2016). 80 Walsh, M. C., Brennan, L., Malthouse, J. P. G., Roche, H. M. & Gibney, M. J. Effect of acute dietary standardization on the urinary, plasma, and salivary metabolomic profiles of healthy humans–. The American journal of clinical nutrition 84, 531-539 (2006). 81 Takeda, I. et al. Understanding the human salivary metabolome. NMR Biomed. 22, 577-584 (2009). 82 Okuma, N. et al. Effect of masticatory stimulation on the quantity and quality of saliva and the salivary metabolomic profile. PLoS One 12, e0183109 (2017). 83 Neyraud, E., Tremblay-Franco, M., Gregoire, S., Berdeaux, O. & Canlet, C. Relationships between the metabolome and the fatty acid composition of human saliva; effects of stimulation. Metabolomics 9, 213-222 (2013). 84 Sugimoto, M. et al. Physiological and environmental parameters associated with mass spectrometry-based salivary metabolomic profiles. Metabolomics 9, 454-463 (2013). 85 Mounayar, R. et al. Nutri-metabolomics applied to taste perception phenotype: human subjects with high and low sensitivity to taste of fat differ in salivary response to oleic acid. OMICS 18, 666-672 (2014). 86 Silwood, C., Lynch, E., Claxson, A. & Grootveld, M. 1H and 13C NMR spectroscopic analysis of human saliva. J. Dent. Res. 81, 422-427 (2002). 87 Silwood, C. J. et al. 1H‐NMR analysis of microbial‐derived organic acids in primary root carious lesions and saliva. NMR Biomed. 12, 345-356 (1999). 88 Jagtap, P. et al. Deep metaproteomic analysis of human salivary supernatant. Proteomics 12, 992-1001 (2012). 89 Figueira, J. et al. Metabolite quantification by NMR and LC-MS/MS reveals differences between unstimulated, stimulated, and pure parotid saliva. J. Pharm. Biomed. Anal. 140, 295-300, doi:https://doi.org/10.1016/j.jpba.2017.03.037 (2017). 90 Yamada-Nosaka, A. et al. Preliminary nuclear magnetic resonance studies on human saliva. Arch. Oral Biol. 36, 697-701, doi:https://doi.org/10.1016/0003-9969(91)90025- P (1991).

270

91 Mounayar, R. et al. Salivary markers of taste sensitivity to oleic acid: a combined proteomics and metabolomics approach. Metabolomics 10, 688-696 (2014). 92 Rooks, M. G. & Garrett, W. S. Gut microbiota, metabolites and host immunity. Nature Reviews Immunology 16, 341 (2016). 93 Emwas, A.-H. M. in Metabonomics 161-193 (Springer, 2015). 94 Takahashi, N., Washio, J. & Mayanagi, G. Metabolomics of supragingival plaque and oral bacteria. J. Dent. Res. 89, 1383-1388 (2010). 95 Vitorino, R., Guedes, S., Manadas, B., Ferreira, R. & Amado, F. Toward a standardized saliva proteome analysis methodology. J. Proteomics 75, 5140-5165 (2012). 96 Beckonert, O. et al. Metabolic profiling, metabolomic and metabonomic procedures for NMR spectroscopy of urine, plasma, serum and tissue extracts. Nat. Protoc. 2, 2692, doi:10.1038/nprot.2007.376 https://www.nature.com/articles/nprot.2007.376#supplementary-information (2007). 97 Takahashi, N. Oral microbiome metabolism: from “who are they?” to “what are they doing?”. J. Dent. Res. 94, 1628-1637 (2015). 98 Nyvad, B., Crielaard, W., Mira, A., Takahashi, N. & Beighton, D. Dental Caries from a Molecular Microbiological Perspective. Caries Res. 47, 89-102, doi:10.1159/000345367 (2013). 99 Duran-Pinedo, A. E. & Frias-Lopez, J. Beyond microbial community composition: functional activities of the oral microbiome in health and disease. Microbes and Infection 17, 505-516, doi:https://doi.org/10.1016/j.micinf.2015.03.014 (2015). 100 Johnson, C. H., Ivanisevic, J. & Siuzdak, G. Metabolomics: beyond biomarkers and towards mechanisms. Nature reviews Molecular cell biology 17, 451-459 (2016). 101 Shaffer, M., Armstrong, A. J. S., Phelan, V. V., Reisdorph, N. & Lozupone, C. A. Microbiome and metabolome data integration provides insight into health and disease. Translational Research 189, 51-64, doi:https://doi.org/10.1016/j.trsl.2017.07.001 (2017). 102 Seerangaiyan, K., Maruthamuthu, M., van Winkelhoff, A. J. & Winkel, E. G. Untargeted metabolomics of the bacterial tongue coating of intra-oral halitosis patients. Journal of breath research 13, 046010 (2019). 103 Prodan, A. et al. Effect of experimental gingivitis induction and erythritol on the salivary metabolome and functional biochemistry of systemically healthy young adults. Metabolomics 12, 147, doi:10.1007/s11306-016-1096-4 (2016). 104 Takahashi, N. & Washio, J. Metabolomic Effects of Xylitol and Fluoride on Plaque Biofilm in Vivo. J. Dent. Res. 90, 1463-1468, doi:10.1177/0022034511423395 (2011). 105 Feng, Y. et al. The Associations between Biochemical and Microbiological Variables and Taste Differ in Whole Saliva and in the Film Lining the Tongue. Biomed Res Int 2018, 2838052-2838052, doi:10.1155/2018/2838052 (2018). 106 Tang, Z.-Z. et al. Multi-omic analysis of the microbiome and metabolome in healthy subjects reveals microbiome-dependent relationships between diet and metabolites. Frontiers in genetics 10, 454 (2019). 107 De Angelis, M. et al. Salivary and fecal microbiota and metabolome of celiac children under gluten-free diet. Int. J. Food Microbiol. 239, 125-132, doi:https://doi.org/10.1016/j.ijfoodmicro.2016.07.025 (2016). 108 Ercolini, D. et al. From an imbalance to a new imbalance: Italian-style gluten-free diet alters the salivary microbiota and metabolome of African celiac children. Sci. Rep. 5, 1- 11 (2015). 109 De Filippis, F. et al. The same microbiota and a potentially discriminant metabolome in the saliva of omnivore, ovo-lacto-vegetarian and vegan individuals. PLoS One 9 (2014). 110 Skarke, C. et al. A pilot characterization of the human chronobiome. Sci. Rep. 7, 1-12 (2017).

271

111 Zaura, E. et al. On the ecosystemic network of saliva in healthy young adults. ISME J 11, 1218-1231, doi:10.1038/ismej.2016.199 (2017). 112 Bregy, L. et al. Differentiation of oral bacteria in in vitro cultures and human saliva by secondary electrospray ionization – mass spectrometry. Sci. Rep. 5, 15163, doi:10.1038/srep15163 https://www.nature.com/articles/srep15163#supplementary-information (2015). 113 Edlund, A. et al. Meta-omics uncover temporal regulation of pathways across oral microbiome genera during in vitro sugar metabolism. ISME J 9, 2605-2619 (2015). 114 Kato, T. et al. Oral Administration of <span class="named-content genus- species" id="named-content-1">Porphyromonas gingivalis</span> Alters the Gut Microbiome and Serum Metabolome. mSphere 3, e00460-00418, doi:10.1128/mSphere.00460-18 (2018). 115 Silva, V. d. O. et al. Effects of Monolaurin on Oral Microbe–Host Transcriptome and Metabolome. Front. Microbiol. 9, doi:10.3389/fmicb.2018.02638 (2018). 116 Huang, H. et al. Untargeted metabolomics analysis for a strain of Bacillus subtilis isolated from the oral cavity. Int. J. Clin. Exp. Med. 11, 3729-+ (2018). 117 Edlund, A. et al. Metabolic fingerprints from the human oral microbiome reveal a vast knowledge gap of secreted small peptidic molecules. MSystems 2, e00058-00017 (2017). 118 Lindemann, B. Chemoreception: tasting the sweet and the bitter. Curr. Biol. 6, 1234- 1237 (1996). 119 Breslin, P. A. & Huang, L. in Taste and smell Vol. 63 152-190 (Karger Publishers, 2006). 120 Gerhold, K. A. & Bautista, D. M. Molecular and cellular mechanisms of trigeminal chemosensation. Ann. N. Y. Acad. Sci. 1170, 184-189 (2009). 121 Keast, R. S. & Costanzo, A. Is fat the sixth taste primary? Evidence and implications. Flavour 4, 5 (2015). 122 Gibbins, H. & Carpenter, G. Alternative mechanisms of astringency–what is the role of saliva? Journal of texture studies 44, 364-375 (2013). 123 Feng, P., Huang, L. & Wang, H. Taste Bud Homeostasis in Health, Disease, and Aging. Chem. Senses 39, 3-16, doi:10.1093/chemse/bjt059 (2013). 124 Gravina, S. A., Yep, G. L. & Khan, M. Human biology of taste. Ann. Saudi Med. 33, 217- 222 (2013). 125 Witt, M. & Reutter, K. Anatomy of the tongue and taste buds. Handbook of olfaction and gustation, 3rd edn. John Wiley & Sons, Hoboken, NJ, 637-663 (2015). 126 DuBois, G. E. Molecular mechanism of sweetness sensation. Physiol. Behav. 164, 453- 463 (2016). 127 Margolskee, R. F. Molecular mechanisms of bitter and sweet taste transduction. J. Biol. Chem. 277, 1-4 (2002). 128 Meyers, B. & Brewer, M. Sweet taste in man: a review. J. Food Sci. 73, R81-R90 (2008). 129 Deepankumar, S., Karthi, M., Vasanth, K. & Selvakumar, S. Insights on modulators in perception of taste modalities: a review. Nutr. Res. Rev., 1-16 (2019). 130 Meyerhof, W. et al. The Molecular Receptive Ranges of Human TAS2R Bitter Taste Receptors. Chem. Senses 35, 157-170, doi:10.1093/chemse/bjp092 (2009). 131 Shaik, F. A. et al. Bitter taste receptors: extraoral roles in pathophysiology. The international journal of biochemistry & cell biology 77, 197-204 (2016). 132 Roper, S. D. & Chaudhari, N. Taste buds: cells, signals and synapses. Nature Reviews Neuroscience 18, 485 (2017). 133 Kinnamon, S. C. Umami taste transduction mechanisms. The American journal of clinical nutrition 90, 753S-755S (2009). 134 Zhao, G. Q. et al. The receptors for mammalian sweet and umami taste. Cell 115, 255- 266 (2003).

272

135 Ye, W. et al. The K+ channel KIR2. 1 functions in tandem with proton influx to mediate sour taste transduction. Proceedings of the National Academy of Sciences 113, E229- E238 (2016). 136 Huang, A. L. et al. The cells and logic for mammalian sour taste detection. Nature 442, 934 (2006). 137 Chaudhari, N. & Roper, S. D. The cell biology of taste. The Journal of cell biology 190, 285-296 (2010). 138 Chandrashekar, J. et al. The cells and peripheral representation of sodium taste in mice. Nature 464, 297 (2010). 139 Chandrashekar, J., Hoon, M. A., Ryba, N. J. & Zuker, C. S. The receptors and cells for mammalian taste. Nature 444, 288 (2006). 140 Lee, H., Macpherson, L. J., Parada, C. A., Zuker, C. S. & Ryba, N. J. Rewiring the taste system. Nature 548, 330 (2017). 141 Barretto, R. P. et al. The neural representation of taste quality at the periphery. Nature 517, 373 (2015). 142 Ohla, K. et al. Recognizing taste: coding patterns along the neural axis in mammals. Chem. Senses 44, 237-247 (2019). 143 Stevenson, R. J., Francis, H. M. & Ragg, C. J. in Island of Reil (Insula) in the Human Brain 133-145 (Springer, 2018). 144 González-Ramírez, R., Chen, Y., Liedtke, W. B. & Morales-Lázaro, S. L. in Neurobiology of TRP Channels. 2nd edition (CRC Press/Taylor & Francis, 2017). 145 Running, C. A. Oral sensations and secretions. Physiol. Behav. 193, 234-237 (2018). 146 Roper, S. D. in Mammalian Transient Receptor Potential (TRP) Cation Channels 827- 871 (Springer, 2014). 147 Running, C. A. & Mattes, R. D. A Review of the Evidence Supporting the Taste of Non‐ esterified Fatty Acids in Humans. Journal of the American Oil Chemists' Society 93, 1325-1336 (2016). 148 Tucker, R. M., Mattes, R. D. & Running, C. A. Mechanisms and effects of “fat taste” in humans. Biofactors 40, 313-326 (2014). 149 Guichard, E., Galindo-Cuspinera, V. & Feron, G. Physiological mechanisms explaining human differences in fat perception and liking in food spreads-a review. Trends in Food Science & Technology 74, 46-55 (2018). 150 Lai, W. Y. W., Chua, J. W. M., Gill, S. & Brownlee, I. A. Analysis of the Lipolytic Activity of Whole-Saliva and Site-Specific Secretions from the Oral Cavity of Healthy Adults. Nutrients 11, 191 (2019). 151 Mounayar, R., Septier, C., Chabanet, C., Feron, G. & Neyraud, E. Oral Fat Sensitivity in Humans: Links to Saliva Composition Before and After Stimulation by Oleic Acid. Chemosensory Perception 6, 118-126, doi:10.1007/s12078-013-9152-1 (2013). 152 Bajec, M. R. & Pickering, G. J. Astringency: mechanisms and perception. Crit. Rev. Food Sci. Nutr. 48, 858-875 (2008). 153 Green, B. G. Oral astringency: a tactile component of flavor. Acta Psychol. (Amst.) 84, 119-125 (1993). 154 Laguna, L. & Sarkar, A. Oral tribology: update on the relevance to study astringency in wines. Tribology - Materials, Surfaces & Interfaces 11, 116-123, doi:10.1080/17515831.2017.1347736 (2017). 155 Ma, S., Lee, H., Liang, Y. & Zhou, F. Astringent mouthfeel as a consequence of lubrication failure. Angewandte Chemie International Edition 55, 5793-5797 (2016). 156 Webb, J., Bolhuis, D. P., Cicerale, S., Hayes, J. E. & Keast, R. The relationships between common measurements of taste function. Chemosensory perception 8, 11-18 (2015). 157 ISO. International Standards Organisation 3972:2011 Sensory analysis -- Methodology -- Method of investigating sensitivity of taste. (2011).

273

158 ISO. ISO 13301:2018 Sensory analysis -- Methodology -- General guidance for measuring odour, flavour and taste detection thresholds by a three-alternative forced-choice (3-AFC) procedure. (2018). 159 Snyder, D. J., Prescott, J. & Bartoshuk, L. M. in Taste and smell Vol. 63 221-241 (Karger Publishers, 2006). 160 O'Mahony, M. Who told you the triangle test was simple? Food Qual. Prefer. 6, 227- 238 (1995). 161 Marks, L. E. et al. Magnitude-matching: the measurement of taste and smell. Chem. Senses 13, 63-87, doi:10.1093/chemse/13.1.63 (1988). 162 Fast, K., Duffy, V. B. & Bartoshuk, L. M. New psychophysical insights in evaluating genetic variation in taste. Olfaction, taste and cognition, 391-407 (2002). 163 Hayes, J. in Rapid Sensory Profiling Techniques 53-69 (Elsevier, 2015). 164 Kershaw, J. C. & Running, C. A. Data approximation strategies between generalized line scales and the influence of labels and spacing. Journal of Sensory Studies 34, e12507 (2019). 165 Hayes, J. E., Allen, A. L. & Bennett, S. M. Direct comparison of the generalized visual analog scale (gVAS) and general labeled magnitude scale (gLMS). Food Qual. Prefer. 28, 36-44 (2013). 166 Kalva, J. J., Sims, C. A., Puentes, L. A., Snyder, D. J. & Bartoshuk, L. M. Comparison of the hedonic general labeled magnitude scale with the hedonic 9‐point scale. J. Food Sci. 79, S238-S245 (2014). 167 Kershaw, J. C. & Running, C. A. Data approximation strategies between generalized line scales and the influence of labels and spacing. Journal of Sensory Studies, e12507. 168 Running, C. A. & Hayes, J. E. Sip and spit or sip and swallow: Choice of method differentially alters taste intensity estimates across stimuli. Physiol. Behav. 181, 95-99 (2017). 169 Hayes, J. E., Bartoshuk, L. M., Kidd, J. R. & Duffy, V. B. Supertasting and PROP Bitterness Depends on More Than the TAS2R38 Gene. Chem. Senses 33, 255-265, doi:10.1093/chemse/bjm084 (2008). 170 Blakeslee, A. F. Genetics of sensory thresholds: taste for phenyl thio carbamide. Proc. Natl. Acad. Sci. U. S. A. 18, 120 (1932). 171 Kim, U.-k. et al. Positional Cloning of the Human Quantitative Trait Locus Underlying Taste Sensitivity to Phenylthiocarbamide. Science 299, 1221-1225, doi:10.1126/science.1080190 (2003). 172 Hayes, J. E. & Keast, R. S. Two decades of supertasting: where do we stand? Physiol. Behav. 104, 1072-1074 (2011). 173 Bartoshuk, L. M., Duffy, V. B. & Miller, I. J. PTC/PROP tasting: anatomy, psychophysics, and sex effects. Physiol. Behav. 56, 1165-1171 (1994). 174 Prutkin, J. et al. Genetic variation and inferences about perceived taste intensity in mice and men. Physiol. Behav. 69, 161-173 (2000). 175 Galindo-Cuspinera, V. et al. Reliability of threshold and suprathreshold methods for taste phenotyping: characterization with PROP and sodium chloride. Chemosensory perception 2, 214 (2009). 176 Fischer, M. E. et al. Factors related to fungiform papillae density: the beaver dam offspring study. Chem. Senses 38, 669-677 (2013). 177 Cattaneo, C. et al. New insights into the relationship between taste perception and oral microbiota composition. Sci. Rep. 9, 3549 (2019). 178 Piochi, M., Dinnella, C., Prescott, J. & Monteleone, E. Associations between human fungiform papillae and responsiveness to oral stimuli: effects of individual variability, population characteristics, and methods for papillae quantification. Chem. Senses 43, 313-327, doi:10.1093/chemse/bjy015 (2018).

274

179 Dinnella, C. et al. Individual Variation in PROP Status, Fungiform Papillae Density, and Responsiveness to Taste Stimuli in a Large Population Sample. Chem. Senses 43, 697- 710, doi:10.1093/chemse/bjy058 (2018). 180 Essick, G. K., Chopra, A., Guest, S. & McGlone, F. Lingual tactile acuity, taste perception, and the density and diameter of fungiform papillae in female subjects. Physiol. Behav. 80, 289-302 (2003). 181 Lim, J., Urban, L. & Green, B. G. Measures of individual differences in taste and creaminess perception. Chem. Senses 33, 493-501 (2008). 182 Hector, M. & Linden, R. Reflexes of Salivary Secretion. Neural mechanisms of salivary glands. Frontiers of oral biology 11, 196-217 (1999). 183 Watanabe, S. & Dawes, C. The effects of different foods and concentrations of citric acid on the flow rate of whole saliva in man. Arch. Oral Biol. 33, 1-5 (1988). 184 Dawes, C. Stimulus effects on protein and electrolyte concentrations in parotid saliva. The Journal of physiology 346, 579-588 (1984). 185 Neyraud, E., Heinzerling, C., Bult, J., Mesmin, C. & Dransfield, E. Effects of different tastants on parotid saliva flow and composition. Chemosensory Perception 2, 108-116 (2009). 186 Neyraud, E., Sayd, T., Morzel, M. & Dransfield, E. Proteomic analysis of human whole and parotid salivas following stimulation by different tastes. J. Proteome Res. 5, 2474- 2480 (2006). 187 Bader, M. et al. Chemosensate-Induced Modulation of the Salivary Proteome and Metabolome Alters the Sensory Perception of Salt Taste and Odor-Active Thiols. J. Agric. Food Chem. 66, 7740-7749 (2018). 188 Stokes, J. R. & Davies, G. A. Viscoelasticity of human whole saliva collected after acid and mechanical stimulation. Biorheology 44, 141-160 (2007). 189 Vijay, A., Inui, T., Dodds, M., Proctor, G. & Carpenter, G. Factors that influence the extensional rheological property of saliva. PLoS One 10, e0135792 (2015). 190 Houghton, J. W. et al. Sensory effects of transient receptor potential channel agonists on whole mouth saliva extensional rheology. Journal of Texture Studies 48, 313-317, doi:doi:10.1111/jtxs.12260 (2017). 191 Kallithraka, S., Bakker, J. & Clifford, M. Evidence that salivary proteins are involved in astringency. Journal of sensory studies 13, 29-43 (1998). 192 Bennick, A. Interaction of plant polyphenols with salivary proteins. Crit. Rev. Oral Biol. Med. 13, 184-196 (2002). 193 Lee, C. A., Ismail, B. & Vickers, Z. M. The role of salivary proteins in the mechanism of astringency. J. Food Sci. 77, C381-C387 (2012). 194 Neuwirth, I. & Klosterman, J. A. Demonstration of Rapid Production of Lactic Acid in Oral Cavity. Proc. Soc. Exp. Biol. Med. 45, 464-467 (1940). 195 McBurney, D. H. & Pfaffmann, C. Gustatory adaptation to saliva and sodium chloride. J. Exp. Psychol. 65, 523 (1963). 196 Morino, T. & Langford, H. G. Salivary sodium correlates with salt recognition threshold. Physiol. Behav. 21, 45-48 (1978). 197 Delwiche, J. & O'Mahony, M. Changes in secreted salivary sodium are sufficient to alter salt taste sensitivity: Use of signal detection measures with continuous monitoring of the oral environment. Physiol. Behav. 59, 605-611, doi:https://doi.org/10.1016/0031-9384(95)02120-5 (1996). 198 Scinska-Bienkowska, A. et al. Glutamate concentration in whole saliva and taste responses to monosodium glutamate in humans. Nutr. Neurosci. 9, 25-31, doi:10.1080/10284150600621964 (2006). 199 Dawes, C. et al. The functions of human saliva: A review sponsored by the World Workshop on Oral Medicine VI. Arch. Oral Biol. 60, 863-874 (2015).

275

200 Christensen, C. M., Brand, J. G. & Malamud, D. Salivary changes in solution pH: A source of individual differences in sour taste perception. Physiol. Behav. 40, 221-227, doi:https://doi.org/10.1016/0031-9384(87)90211-3 (1987). 201 Manley, K. J., Haryono, R. Y. & Keast, R. S. Taste changes and saliva composition in chronic kidney disease. Renal Society of Australasia Journal 8 (2012). 202 Fleming, E. E., Ziegler, G. R. & Hayes, J. E. Salivary protein levels as a predictor of perceived astringency in model systems and solid foods. Physiol. Behav. 163, 56-63, doi:https://doi.org/10.1016/j.physbeh.2016.04.043 (2016). 203 Dinnella, C., Recchia, A., Fia, G., Bertuccioli, M. & Monteleone, E. Saliva Characteristics and Individual Sensitivity to Phenolic Astringent Stimuli. Chem. Senses 34, 295-304, doi:10.1093/chemse/bjp003 (2009). 204 Nayak, A. & Carpenter, G. H. A physiological model of tea-induced astringency. Physiol. Behav. 95, 290-294, doi:https://doi.org/10.1016/j.physbeh.2008.05.023 (2008). 205 Dinnella, C., Recchia, A., Vincenzi, S., Tuorila, H. & Monteleone, E. Temporary modification of salivary protein profile and individual responses to repeated phenolic astringent stimuli. Chem. Senses 35, 75-85 (2009). 206 Sarkar, A., Goh, K. K. & Singh, H. Colloidal stability and interactions of milk-protein- stabilized emulsions in an artificial saliva. Food Hydrocolloids 23, 1270-1278 (2009). 207 Sarkar, A., Ye, A. & Singh, H. Oral processing of emulsion systems from a colloidal perspective. Food Funct. 8, 511-521 (2017). 208 Dsamou, M. et al. Salivary protein profiles and sensitivity to the bitter taste of caffeine. Chem. Senses 37, 87-95 (2011). 209 Morzel, M. et al. Salivary protein profiles are linked to bitter taste acceptance in infants. Eur. J. Pediatr. 173, 575-582 (2014). 210 Rodrigues, L. et al. Relationship between saliva protein composition and 6‐n‐ Propylthiouracil bitter taste responsiveness in young adults. Journal of Sensory Studies 32, e12275 (2017). 211 Rodrigues, L. et al. Comparison of salivary proteome of children with different sensitivities for bitter and sweet tastes: association with body mass index. Int. J. Obes. 43, 701-712, doi:10.1038/s41366-018-0289-5 (2019). 212 Stolle, T. et al. Salivary proteome patterns affecting human salt taste sensitivity. J. Agric. Food Chem. 65, 9275-9286 (2017). 213 Rodrigues, L. et al. Salivary proteome and glucose levels are related with sweet taste sensitivity in young adults. Food Nutr. Res. 61, 1389208, doi:10.1080/16546628.2017.1389208 (2017). 214 Marquezin, M. C. S. et al. Descriptive analysis of the masticatory and salivary functions and gustatory sensitivity in healthy children. Acta Odontol. Scand. 74, 443-448, doi:10.1080/00016357.2016.1191085 (2016). 215 Ferry, A.-L. S. et al. In-mouth amylase activity can reduce perception of saltiness in starch-thickened foods. J. Agric. Food Chem. 54, 8869-8873 (2006). 216 Voigt, N. et al. The role of lipolysis in human orosensory fat perception. J. Lipid Res. 55, 870-882 (2014). 217 Neyraud, E. et al. The basal free fatty acid concentration in human saliva is related to salivary lipolytic activity. Sci. Rep. 7, 5969 (2017). 218 Morris-Wiman, J., Sego, R., Brinkley, L. & Dolce, C. The Effects of Sialoadenectomy and Exogenous EGF on Taste Bud Morphology and Maintenance. Chem. Senses 25, 9-19, doi:10.1093/chemse/25.1.9 (2000). 219 Redman, R. S. in Salivary Gland Development and Regeneration 75-91 (Springer, 2017). 220 Henkin, R. I., Brian M, M. & Raghunath P, A. Decreased Parotid Saliva Gustin/Carbonic Anhydrase VI Secretion: An Disorder Manifested by Gustatory and Olfactory

276

Dysfunction. The American Journal of the Medical Sciences 318, 380-391, doi:https://doi.org/10.1016/S0002-9629(15)40663-9 (1999). 221 Henkin, R. I., Martin, B. M. & Agarwal, R. P. Efficacy of exogenous oral zinc in treatment of patients with carbonic anhydrase VI deficiency. The American journal of the medical sciences 318, 392-405 (1999). 222 Kumbargere Nagraj, S. et al. Interventions for managing taste disturbances. Cochrane Database of Systematic Reviews, doi:10.1002/14651858.CD010470.pub3 (2017). 223 Nicholson, J. K., Lindon, J. C. & Holmes, E. 'Metabonomics': understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica 29, 1181-1189 (1999). 224 Schipper, R. G., Silletti, E. & Vingerhoeds, M. H. Saliva as research material: biochemical, physicochemical and practical aspects. Arch. Oral Biol. 52, 1114-1135 (2007). 225 Stokes, J. R., Boehm, M. W. & Baier, S. K. Oral processing, texture and mouthfeel: From rheology to tribology and beyond. Current Opinion in Colloid & Interface Science 18, 349-359, doi:https://doi.org/10.1016/j.cocis.2013.04.010 (2013). 226 Haward, S. J., Odell, J. A., Berry, M. & Hall, T. Extensional rheology of human saliva. Rheologica Acta 50, 869-879, doi:10.1007/s00397-010-0494-1 (2011). 227 Chaudhury, N., Shirlaw, P., Pramanik, R., Carpenter, G. & Proctor, G. Changes in saliva rheological properties and mucin in dry mouth. J. Dent. Res. 94, 1660- 1667 (2015). 228 Pushpass, R.-A. G., Daly, B., Kelly, C., Proctor, G. & Carpenter, G. H. Altered salivary flow, protein composition and rheology following taste and TRP stimulation in older adults. Front. Physiol. 10, 652 (2019). 229 Miller, E., Clasen, C. & Rothstein, J. P. The effect of step-stretch parameters on capillary breakup extensional rheology (CaBER) measurements. Rheologica Acta 48, 625-639, doi:10.1007/s00397-009-0357-9 (2009). 230 Bhardwaj, A., Miller, E. & Rothstein, J. P. Filament stretching and capillary breakup extensional rheometry measurements of viscoelastic wormlike micelle solutions. Journal of Rheology 51, 693-719 (2007). 231 Gardner, A. & Carpenter, G. H. Anatomical stability of human fungiform papillae and relationship with oral perception measured by salivary response and intensity rating. Sci. Rep. 9, 9759, doi:10.1038/s41598-019-46093-z (2019). 232 Gardner, A., Parkes, H. G., Carpenter, G. H. & So, P.-W. Developing and Standardizing a Protocol for Quantitative Proton Nuclear Magnetic Resonance (1H NMR) Spectroscopy of Saliva. J. Proteome Res. 17, 1521-1531, doi:10.1021/acs.jproteome.7b00847 (2018). 233 Pramanik, R. et al. Effects of the UK Biobank collection protocol on potential biomarkers in saliva. Int. J. Epidemiol. 41, 1786-1797, doi:10.1093/ije/dys166 (2012). 234 Calviño, A. M., García-Medina, M. R., Cometto-Muniz, J. E. & Rodríguez, M. B. Perception of sweetness and bitterness in different vehicles. Percept. Psychophys. 54, 751-758 (1993). 235 Nasrawi, C. W. & Pangborn, R. M. Temporal gustatory and salivary responses to capsaicin upon repeated stimulation. Physiol. Behav. 47, 611-615, doi:https://doi.org/10.1016/0031-9384(90)90066-D (1990). 236 Dewettinck, K., Dierckx, S., Eichwalder, P. & Huyghebaert, A. Comparison of SDS-PAGE profiles of four Belgian cheeses by multivariate statistics. Le Lait 77, 77-89 (1997). 237 Olias, R. et al. Sodium dodecyl sulphate‐polyacrylamide gel electrophoresis of proteins in dry‐cured hams: Data registration and multivariate analysis across multiple gels. Electrophoresis 27, 1288-1299 (2006). 238 Bartoshuk, L. M., Duffy, V. B., Hayes, J. E., Moskowitz, H. R. & Snyder, D. J. Psychophysics of sweet and fat perception in obesity: problems, solutions and new 277

perspectives. Philosophical Transactions of the Royal Society B: Biological Sciences 361, 1137-1148 (2006). 239 Carpenter, G. H. Do transient receptor protein (TRP) channels play a role in oral astringency? Journal of texture studies 44, 334-337 (2013). 240 Silwood, C. L., Grootveld, M. & Lynch, E. 1H NMR investigations of the molecular nature of low-molecular-mass calcium ions in biofluids. J. Biol. Inorg. Chem. 7, 46-57, doi:10.1007/s007750100264 (2002). 241 Gardner, A., Parkes, H. G., So, P.-W. & Carpenter, G. H. Determining bacterial and host contributions to the human salivary metabolome. J. Oral Microbiol. 11, 1617014, doi:10.1080/20002297.2019.1617014 (2019). 242 Verbeurgt, C. et al. The human bitter taste receptor T2R38 is broadly tuned for bacterial compounds. PLoS One 12, e0181302 (2017). 243 Besnard, P. et al. Obese subjects with specific gustatory papillae microbiota and salivary cues display an impairment to sense lipids. Sci. Rep. 8, 6742 (2018). 244 Ito, T. et al. Tissue depletion of taurine accelerates skeletal muscle senescence and leads to early death in mice. PLoS One 9, e107409-e107409, doi:10.1371/journal.pone.0107409 (2014). 245 Cleaver, L. M., Moazzez, R. & Carpenter, G. H. Mixed aerobic-anaerobic incubation conditions induce proteolytic activity from in vitro salivary biofilms. J. Oral Microbiol. 11, 1643206 (2019). 246 Neyraud, E. & Morzel, M. Biological films adhering to the oral soft tissues: Structure, composition, and potential impact on taste perception. Journal of texture studies 50, 19-26 (2019). 247 Schuller-Levis, G. B. & Park, E. Taurine: new implications for an old amino acid. FEMS Microbiol. Lett. 226, 195-202 (2003). 248 Khurshid, Z., Mali, M., Naseem, M., Najeeb, S. & Zafar, M. S. Human Gingival Crevicular Fluids (GCF) Proteomics: An Overview. Dent J (Basel) 5, 12, doi:10.3390/dj5010012 (2017). 249 Schirra, H. J. & Ford, P. J. in Oral Biology 79-105 (Springer, 2017). 250 Nicholson, J. K. Global systems biology, personalized medicine and molecular epidemiology. Mol. Syst. Biol. 2 (2006). 251 Heckman, M. A., Weil, J. & De Mejia, E. G. Caffeine (1, 3, 7‐trimethylxanthine) in foods: a comprehensive review on consumption, functionality, safety, and regulatory matters. J. Food Sci. 75, R77-R87 (2010). 252 Tan, S.-Y. & Tucker, R. Sweet Taste as a Predictor of Dietary Intake: A Systematic Review. Nutrients 11, 94 (2019). 253 Miller, B. F., Muntz, J. A. & Bradel, S. Decomposition of carbohydrate substrates by dental plaque material. J. Dent. Res. 19, 473-478 (1940). 254 Owen, O. E., Kalhan, S. C. & Hanson, R. W. The key role of anaplerosis and cataplerosis for citric acid cycle function. J. Biol. Chem. 277, 30409-30412 (2002). 255 Kim, S. H., Schneider, B. L. & Reitzer, L. Genetics and regulation of the major enzymes of alanine synthesis in Escherichia coli. J. Bacteriol. 192, 5304-5311 (2010). 256 Price, M. N. et al. Filling gaps in bacterial amino acid biosynthesis pathways with high- throughput genetics. PLOS Genetics 14, e1007147, doi:10.1371/journal.pgen.1007147 (2018). 257 Vinod, K. S. et al. Salivary and Serum Aspartate Aminotransferases and Alanine Aminotransferases in Insulin-Dependent Diabetes Mellitus and Normal Children: A Comparative Study. J Int Soc Prev Community Dent 8, 229-234, doi:10.4103/jispcd.JISPCD_60_18 (2018). 258 Hillman, J. D., Andrews, S. W. & Dzuback, A. L. Acetoin production by wild-type strains and a lactate dehydrogenase-deficient mutant of Streptococcus mutans. Infect. Immun. 55, 1399-1402 (1987).

278

259 MacGregor, G. A. & Hashem, K. M. Action on sugar—lessons from UK salt reduction programme. The Lancet 383, 929-931 (2014). 260 McBride, R. L. Sweetness of binary mixtures of sucrose, fructose, and glucose. J. Exp. Psychol. Hum. Percept. Perform. 12, 584 (1986). 261 Sun, S. Z. & Empie, M. W. Fructose metabolism in humans - what isotopic tracer studies tell us. Nutr. Metab. (Lond.) 9, 89-89, doi:10.1186/1743-7075-9-89 (2012). 262 Lagerlof, F., Dawes, R. & Dawes, C. The effects of different concentrations of sucrose, fructose, and glucose on pH changes by Streptococcus mitior in an artificial mouth. J. Dent. Res. 64, 405-410 (1985). 263 Colby, S. & Russell, R. Sugar metabolism by mutans streptococci. J. Appl. Microbiol. 83, 80S-88S (1997). 264 Ajdić, D. et al. Genome sequence of <em>Streptococcus mutans</em> UA159, a cariogenic dental pathogen. Proceedings of the National Academy of Sciences 99, 14434, doi:10.1073/pnas.172501299 (2002). 265 Korithoski, B., Krastel, K. & Cvitkovitch, D. G. Transport and metabolism of citrate by Streptococcus mutans. J. Bacteriol. 187, 4451-4456 (2005). 266 Fernández-Veledo, S. & Vendrell, J. Gut microbiota-derived succinate: Friend or foe in human metabolic diseases? Reviews in Endocrine and Metabolic Disorders, doi:10.1007/s11154-019-09513-z (2019). 267 Harmon, A. G., Towe-Goodman, N. R., Fortunato, C. K. & Granger, D. A. Differences in saliva collection location and disparities in baseline and diurnal rhythms of alpha- amylase: A preliminary note of caution. Horm. Behav. 54, 592-596, doi:https://doi.org/10.1016/j.yhbeh.2008.05.019 (2008). 268 Gusman, H. et al. Human salivary gland-specific daily variations in histatin concentrations determined by a novel quantitation technique. Arch. Oral Biol. 49, 11- 22, doi:https://doi.org/10.1016/S0003-9969(03)00182-1 (2004). 269 Hsiao, Y. C. et al. Variability Assessment of 90 Salivary Proteins in Intraday and Interday Samples from Healthy Donors by Multiple Reaction Monitoring‐Mass Spectrometry. PROTEOMICS–Clinical Applications 12, 1700039 (2018). 270 Rudney, J. D. Saliva and dental plaque. Adv. Dent. Res. 14, 29-39 (2000). 271 Scannapieco, F. A. Saliva-bacterium interactions in oral microbial ecology. Crit. Rev. Oral Biol. Med. 5, 203-248 (1994). 272 Spielman, A. Interaction of saliva and taste. J. Dent. Res. 69, 838-843 (1990). 273 Keast, R. S. J. & Breslin, P. A. S. An overview of binary taste–taste interactions. Food Qual. Prefer. 14, 111-124, doi:https://doi.org/10.1016/S0950-3293(02)00110-6 (2003). 274 McBurney, D. H. & Bartoshuk, L. M. Interactions between stimuli with different taste qualities. Physiol. Behav. 10, 1101-1106, doi:https://doi.org/10.1016/0031- 9384(73)90194-7 (1973). 275 Grether-Beck, S. et al. Urea uptake enhances barrier function and antimicrobial defense in humans by regulating epidermal gene expression. The Journal of investigative dermatology 132, 1561-1572, doi:10.1038/jid.2012.42 (2012). 276 Mikkonen, J. J. W. et al. Potential role of nuclear magnetic resonance spectroscopy to identify salivary metabolite alterations in patients with head and neck cancer. Oncol. Lett. 16, 6795-6800, doi:10.3892/ol.2018.9419 (2018). 277 Langan, M. J. & Yearick, E. S. The effects of improved oral hygiene on taste perception and nutrition of the elderly. J. Gerontol. 31, 413-418 (1976). 278 Hyde, R., Feller, R. & Sharon, I. Tongue brushing, dentifrice, and age effects on taste and smell. J. Dent. Res. 60, 1730-1734 (1981). 279 Quirynen, M. et al. Impact of tongue cleansers on microbial load and taste. J. Clin. Periodontol. 31, 506-510 (2004). 280 Gibbons, R. J. Bacterial adhesion to oral tissues: a model for infectious diseases. J. Dent. Res. 68, 750-760 (1989). 279

281 Bhalla, S., Tandon, S. & Satyamoorthy, K. Salivary proteins and early childhood caries: A gel electrophoretic analysis. Contemp. Clin. Dent. 1, 17 (2010). 282 Levine, M. Susceptibility to dental caries and the salivary proline-rich proteins. International journal of dentistry 2011 (2011). 283 Lynge Pedersen, A. M. & Belstrøm, D. The role of natural salivary defences in maintaining a healthy oral microbiota. J. Dent. 80, S3-S12, doi:https://doi.org/10.1016/j.jdent.2018.08.010 (2019). 284 Feron, G. Unstimulated saliva: Background noise in taste molecules. Journal of texture studies 50, 6-18 (2019). 285 Garcia-Bailo, B., Toguri, C., Eny, K. M. & El-Sohemy, A. Genetic variation in taste and its influence on food selection. OMICS A Journal of Integrative Biology 13, 69-80 (2009). 286 Feeney, E., O'brien, S., Scannell, A., Markey, A. & Gibney, E. Genetic variation in taste perception: does it have a role in healthy eating? Proc. Nutr. Soc. 70, 135-143 (2011). 287 Reed, D. R., Tanaka, T. & McDaniel, A. H. Diverse tastes: Genetics of sweet and bitter perception. Physiol. Behav. 88, 215-226, doi:https://doi.org/10.1016/j.physbeh.2006.05.033 (2006). 288 Snyder, L. H. Inherited taste deficiency. Science (1931). 289 Kim, U. K., Breslin, P. A. S., Reed, D. & Drayna, D. Genetics of Human Taste Perception. J. Dent. Res. 83, 448-453, doi:10.1177/154405910408300603 (2004). 290 Kaplan, A. et al. Taste Thresholds in Twins and Siblings 1 2. Acta geneticae medicae et gemellologiae: twin research 16, 229-243 (1967). 291 Smith, S. & Davies, P. Quinine taste thresholds: a family study and a twin study. Ann. Hum. Genet. 37, 227-232 (1973). 292 Hansen, J. L., Reed, D. R., Wright, M. J., Martin, N. G. & Breslin, P. A. S. Heritability and Genetic Covariation of Sensitivity to PROP, SOA, Quinine HCl, and Caffeine. Chem. Senses 31, 403-413, doi:10.1093/chemse/bjj044 (2006). 293 Knaapila, A. et al. Genetic Analysis of Chemosensory Traits in Human Twins. Chem. Senses 37, 869-881, doi:10.1093/chemse/bjs070 (2012). 294 Keskitalo, K. et al. Same genetic components underlie different measures of sweet taste preference. The American journal of clinical nutrition 86, 1663-1669 (2007). 295 Bretz, W. A. et al. Heritability estimates for dental caries and sucrose sweetness preference. Arch. Oral Biol. 51, 1156-1160 (2006). 296 Hwang, L.-D. et al. A common genetic influence on human intensity ratings of sugars and high-potency sweeteners. Twin Research and Human Genetics 18, 361-367 (2015). 297 Beauchamp, G. K. The human preference for excess salt. Am. Sci. 75, 27-33 (1987). 298 Wise, P. M., Hansen, J. L., Reed, D. R. & Breslin, P. A. S. Twin Study of the Heritability of Recognition Thresholds for Sour and Salty Taste. Chem. Senses 32, 749-754, doi:10.1093/chemse/bjm042 (2007). 299 Törnwall, O. et al. Genetic contribution to sour taste preference. Appetite 58, 687- 694, doi:https://doi.org/10.1016/j.appet.2011.12.020 (2012). 300 Costanzo, A. et al. Effect of dietary fat intake and genetics on fat taste sensitivity: a co- twin randomized controlled trial. The American Journal of Clinical Nutrition 107, 683- 694, doi:10.1093/ajcn/nqy022 (2018). 301 Smith, Q., Shapiro, B. & Hamilton, M. Polyacrylamide gel electrophoresis of twin and nontwin parotid salivary proteins. Proc. Soc. Exp. Biol. Med. 151, 535-538 (1976). 302 Rudney, J. D., Michalowicz, B. S., Krig, M. A., Kane, P. K. & Pihlstrom, B. L. Genetic contributions to saliva protein concentrations in adult human twins. Arch. Oral Biol. 39, 513-517, doi:https://doi.org/10.1016/0003-9969(94)90148-1 (1994). 303 Stahringer, S. S. et al. Nurture trumps nature in a longitudinal survey of salivary bacterial communities in twins from early adolescence to early adulthood. Genome Res. 22, 2146-2152 (2012). 280

304 Törnwall, O. et al. Astringency Perception and Heritability Among Young Finnish Twins. Chemosensory Perception 4, 134-144, doi:10.1007/s12078-011-9098-0 (2011). 305 Moskowitz, H. R. Ratio scales of sugar sweetness. Percept. Psychophys. 7, 315-320 (1970). 306 Silventoinen, K. et al. Heritability of adult body height: a comparative study of twin cohorts in eight countries. Twin Research and Human Genetics 6, 399-408 (2003). 307 Carmichael, C. M. & McGue, M. A cross-sectional examination of height, weight, and body mass index in adult twins. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences 50, B237-B244 (1995). 308 Zierer, J. et al. The fecal metabolome as a functional readout of the gut microbiome. Nat. Genet. 50, 790-795, doi:10.1038/s41588-018-0135-7 (2018). 309 Huxtable, R. J. Taurine in the central nervous system and the mammalian actions of taurine. Prog. Neurobiol. 32, 471-533 (1989). 310 Huxtable, R. Physiological actions of taurine. Physiol. Rev. 72, 101-163 (1992). 311 Mehta, R., Bassette, R. & Ward, G. Trimethylamine responsible for fishy flavor in milk from cows on wheat pasture. J. Dairy Sci. 57, 285-289 (1974). 312 Fuquay, J. W., McSweeney, P. L. & Fox, P. F. Encyclopedia of dairy sciences. (Academic Press, 2011). 313 Ruth, J. H. Odor thresholds and irritation levels of several chemical substances: a review. Am. Ind. Hyg. Assoc. J. 47, A-142-A-151 (1986). 314 Syrjänen, S., Piironen, P. & Markkanen, H. Free amino-acid content of wax-stimulated human whole saliva as related to periodontal disease. Arch. Oral Biol. 32, 607-610, doi:https://doi.org/10.1016/0003-9969(87)90032-X (1987). 315 Fernstrom, J. D. et al. Mechanisms for Sweetness. The Journal of Nutrition 142, 1134S- 1141S, doi:10.3945/jn.111.149567 (2012). 316 Kim, U.-k., Wooding, S., Riaz, N., Jorde, L. B. & Drayna, D. Variation in the Human TAS1R Taste Receptor Genes. Chem. Senses 31, 599-611, doi:10.1093/chemse/bjj065 (2006). 317 Bachmanov, A. A. et al. Genetics of sweet taste preferences. Flavour and fragrance journal 26, 286-294 (2011). 318 Fushan, A. A., Simons, C. T., Slack, J. P., Manichaikul, A. & Drayna, D. Allelic Polymorphism within the TAS1R3 Promoter Is Associated with Human Taste Sensitivity to Sucrose. Curr. Biol. 19, 1288-1293, doi:https://doi.org/10.1016/j.cub.2009.06.015 (2009). 319 Hughes, E. R. et al. Microbial respiration and formate oxidation as metabolic signatures of inflammation-associated dysbiosis. Cell host & microbe 21, 208-219 (2017). 320 Grenier, D. & Mayrand, D. Nutritional relationships between oral bacteria. Infect. Immun. 53, 616-620 (1986). 321 Abbe, K., Takahashi, S. & Yamada, T. Involvement of -sensitive pyruvate formate-lyase in mixed-acid fermentation by Streptococcus mutans under strictly anaerobic conditions. J. Bacteriol. 152, 175-182 (1982). 322 Lipchock, S. V. et al. Caffeine Bitterness is Related to Daily Caffeine Intake and Bitter Receptor mRNA Abundance in Human Taste Tissue. Perception 46, 245-256, doi:10.1177/0301006616686098 (2017). 323 Pedersen, M. E. et al. Peritoneal Microdialysis. Earlydiagnosis of Anastomotic Leakage after Low Anterior Resection for Rectosigmoid Cancer. Scand. J. Surg. 98, 148-154 (2009). 324 Feldman, A. G. et al. Lactate and Lactate: Pyruvate Ratio in the Diagnosis and Outcomes of Pediatric Acute Liver Failure. The Journal of pediatrics 182, 217- 222.e213, doi:10.1016/j.jpeds.2016.12.031 (2017).

281

325 Debray, F.-G. et al. Diagnostic Accuracy of Blood Lactate-to-Pyruvate Molar Ratio in the Differential Diagnosis of Congenital Lactic Acidosis. Clin. Chem. 53, 916, doi:10.1373/clinchem.2006.081166 (2007). 326 Kelsay, J., Behall, K., Holden, J. & Crutchfield, H. Pyruvate and lactate in human blood and saliva in response to different carbohydrates. The Journal of nutrition 102, 661- 666 (1972). 327 Cvitkovitch, D. G., Gutierrez, J. A. & Bleiweis, A. S. Role of the citrate pathway in glutamate biosynthesis by Streptococcus mutans. J. Bacteriol. 179, 650-655 (1997). 328 March, J. C., Eiteman, M. A. & Altman, E. Expression of an anaplerotic enzyme, , improves recombinant protein production in Escherichia coli. Appl. Environ. Microbiol. 68, 5620-5624, doi:10.1128/aem.68.11.5620-5624.2002 (2002). 329 Koskinen, K. et al. The nasal microbiome mirrors and potentially shapes olfactory function. Sci. Rep. 8, 1296, doi:10.1038/s41598-018-19438-3 (2018). 330 Visser, J., Kroeze, J., Kamps, W. & Bijleveld, C. Testing taste sensitivity and aversion in very young children: development of a procedure. Appetite 34, 169-176 (2000). 331 Sberro, H. et al. Large-scale analyses of human microbiomes reveal thousands of small, novel genes. Cell 178, 1245-1259. e1214 (2019). 332 Francavilla, R. et al. Salivary microbiota and metabolome associated with celiac disease. Appl. Environ. Microbiol. 80, 3416-3425 (2014). 333 Pineau, N. et al. Temporal Dominance of Sensations: Construction of the TDS curves and comparison with time–intensity. Food Qual. Prefer. 20, 450-455, doi:https://doi.org/10.1016/j.foodqual.2009.04.005 (2009). 334 Noble, A. C. Application of time-intensity procedures for the evaluation of taste and mouthfeel. American Journal of Enology and Viticulture 46, 128-133 (1995). 335 Inoue, H. et al. Rheological properties of human saliva and salivary mucins. J. Oral Biosci. 50, 134-141 (2008). 336 Pradal, C. & Stokes, J. R. Oral tribology: bridging the gap between physical measurements and sensory experience. Current Opinion in Food Science 9, 34-41, doi:https://doi.org/10.1016/j.cofs.2016.04.008 (2016). 337 Sarkar, A., Andablo-Reyes, E., Bryant, M., Dowson, D. & Neville, A. Lubrication of soft oral surfaces. Current Opinion in Colloid & Interface Science 39, 61-75, doi:https://doi.org/10.1016/j.cocis.2019.01.008 (2019). 338 Carpenter, G. et al. A study of saliva lubrication using a compliant oral mimic. Food Hydrocolloids 92, 10-18, doi:https://doi.org/10.1016/j.foodhyd.2019.01.049 (2019). 339 Barrantes, A., Arnebrant, T. & Lindh, L. Characteristics of saliva films adsorbed onto different dental materials studied by QCM-D. Colloids and Surfaces A: Physicochemical and Engineering Aspects 442, 56-62, doi:https://doi.org/10.1016/j.colsurfa.2013.05.054 (2014). 340 Dixon, M. C. Quartz crystal microbalance with dissipation monitoring: enabling real- time characterization of biological materials and their interactions. Journal of biomolecular techniques: JBT 19, 151 (2008). 341 Marx, K. A. Quartz crystal microbalance: a useful tool for studying thin polymer films and complex biomolecular systems at the solution− surface interface. Biomacromolecules 4, 1099-1120 (2003). 342 Yao, J.-W., Lin, C.-J., Chen, G.-Y., Lin, F. & Tao, T. The interactions of epigallocatechin- 3-gallate with human whole saliva and parotid saliva. Arch. Oral Biol. 55, 470-478, doi:https://doi.org/10.1016/j.archoralbio.2010.04.005 (2010). 343 Morrison, D. J. & Preston, T. Formation of short chain fatty acids by the gut microbiota and their impact on human metabolism. Gut microbes 7, 189-200 (2016). 344 Flint, H. J., Scott, K. P., Louis, P. & Duncan, S. H. The role of the gut microbiota in nutrition and health. Nature reviews Gastroenterology & hepatology 9, 577 (2012).

282

345 Koh, A., De Vadder, F., Kovatcheva-Datchary, P. & Bäckhed, F. From dietary fiber to host physiology: short-chain fatty acids as key bacterial metabolites. Cell 165, 1332- 1345 (2016). 346 De Ryck, T. et al. Development of an oral mucosa model to study host-microbiome interactions during wound healing. Appl. Microbiol. Biotechnol. 98, 6831-6846, doi:10.1007/s00253-014-5841-1 (2014). 347 Wang, H.-B., Wang, P.-Y., Wang, X., Wan, Y.-L. & Liu, Y.-C. Butyrate enhances intestinal epithelial barrier function via up-regulation of tight junction protein Claudin-1 transcription. Dig. Dis. Sci. 57, 3126-3135 (2012). 348 Peng, L., Li, Z.-R., Green, R. S., Holzman, I. R. & Lin, J. Butyrate enhances the intestinal barrier by facilitating tight junction assembly via activation of AMP-activated protein kinase in Caco-2 cell monolayers. The Journal of nutrition 139, 1619-1625 (2009). 349 Galindo, M. M. et al. G Protein–Coupled Receptors in Human Fat Taste Perception. Chem. Senses 37, 123-139, doi:10.1093/chemse/bjr069 (2011). 350 Yu, T., Shah, B. P., Hansen, D. R., Park-York, M. & Gilbertson, T. A. Activation of oral trigeminal neurons by fatty acids is dependent upon intracellular calcium. Pflugers Archiv : European journal of physiology 464, 227-237, doi:10.1007/s00424-012-1116-9 (2012). 351 Ozdener, H., Spielman, A. I. & Rawson, N. E. Isolation and culture of human fungiform taste papillae cells. JoVE (Journal of Visualized Experiments), e3730 (2012). 352 Mazurek, M. P. et al. Molecular origin of plasma membrane citrate transporter in human prostate epithelial cells. EMBO reports 11, 431-437 (2010). 353 Mycielska, M. E. et al. Citrate transport and metabolism in mammalian cells: prostate epithelial cells and prostate cancer. Bioessays 31, 10-20 (2009). 354 Sun, Q. F., Sun, Q. H., Du, J. & Wang, S. Differential gene expression profiles of normal human parotid and submandibular glands. Oral Dis. 14, 500-509 (2008). 355 Huard, K. et al. Optimization of a dicarboxylic series for in vivo inhibition of citrate transport by the solute carrier 13 (SLC13) family. J. Med. Chem. 59, 1165-1175 (2016). 356 Pajor, A. M. Sodium-coupled dicarboxylate and citrate transporters from the SLC13 family. Pflügers Archiv - European Journal of Physiology 466, 119-130, doi:10.1007/s00424-013-1369-y (2014). 357 Varela-Centelles, P. et al. Minor salivary gland biopsy in Sjögren's syndrome: a review and introduction of a new tool to ease the procedure. Med. Oral Patol. Oral Cir. Bucal. 19, e20-e23, doi:10.4317/medoral.19131 (2014). 358 Colas, C., Ung, P. M.-U. & Schlessinger, A. SLC transporters: structure, function, and drug discovery. Medchemcomm 7, 1069-1081 (2016).

283

Appendix A: Protocol and preliminary findings of the effects of SCFAs on TR146 cells (buccal epithelial like squamous cell carcinoma)

Appendix A Protocol

TR146 (ECACC 10032305) cells were grown in DMEM/F-12 cell culture media, supplemented with 10% FBS and 1% Penicillin-Streptomycin (10,000 U/ml). Reagents were obtained from Sigma.

Cells were trypsinised with 0.25% trypsin with EDTA, an aliquot of cells was measured for cell density using Trypan blue and the remaining cells were adjusted to a density of 5x105 cells/ml. Cell culture plates (96 wells) were pre-prepared with 50 µl per well of either media (positive control), media and 10% Virkon (negative control) and, into separate wells, sodium salts of acetate, butyrate and propionate, each at concentrations of 10 mM, 2 mM and 0.4 mM. Cell suspension (50 µl) was added to the wells to yield a final cell density of 2.5x105 cells/ml and final SCFA concentrations of 5, 1 and 0.2 mM.

Cells were incubated at 37 °C with 5% CO2 for 24 hours in order to reach confluency. Cell proliferation was assayed using an alamarBlue assay (Thermo Fisher Scientific). The alamarBlue solution was filter sterilised (0.22 µm) and 11 µl was added per well. Cells were returned to the incubator, removing for measurement of fluorescence readings (560/590 nm excitation/emission) at 1, 2.5 and 4 hours.

For each treatment condition data from twelve wells, measured from three biological repeat experiments was analysed by Dunnett’s multiple comparison test relative to the positive control wells following ANOVA. Results are presented in Appendix A Figure 1.

284

Appendix A Figure 1: Comparison of the effects of SCFA supplementation on the proliferation of TR146 cells. n = 12 samples averaged over three experiments. Data are mean ± SEM, p- values denote results of Dunnett’s post-hoc test relative to the positive control.

285

Appendix B: Preliminary investigation of causation and mechanism for salivary metabolite alteration of taste

Appendix B: In-vivo experiment

Solutions of food grade urea (1 mM) and taurine (0.5 mM) were prepared in Buxton mineral water. Concentrations were determined based on the mean salivary concentrations previously observed and deliberately selected to be slightly higher than concentrations typically seen in saliva.

Taste intensity was assessed after oral rinsing and conditioning to water (control), urea and taurine solutions. Tastants assessed were sucrose (0.25 M) and caffeine (8 mM). Ten participants performed the rinse protocol described in Chapter 3 (rinse and expectorate every 20 seconds, for two minutes). A further 3-minute period of passive conditioning to the solution in question was conducted (holding in mouth without rinsing). The sequence of rinsing/conditioning was water, urea then taurine with each solution followed by assessment of sucrose intensity. This protocol was then repeated for caffeine. Intensity data was analysed by ANOVA with Dunnett’s post-hoc test relative to intensity following water control rinses. Results are presented in Appendix B Figure 1.

Appendix B Figure 1: Summary of Sucrose and caffeine taste intensity from ten participants following oral conditioning with water or urea and taurine solutions. Data are mean ± SEM.

286

Appendix B: In-vitro experiment

TR146 cells were cultured as described in Appendix A in preparation for intracellular calcium signalling assay. Cells were incubated for 24 hours in 96 well plates (100 µl/well at densities of 2.5x105/ml) in either plain media (control), or media supplemented with taurine or urea (5 mM, 1 mM, 0.2 mM).

Media was then removed, and the cells were incubated for one hour with Fura-2 AM solution composed of Fura-2 AM 2.5 μM (Abcam), probenecid 1 mM, NaCl 140 mM, KCl 5 mM, MgCl2 1 mM, CaCl2, 10 mM HEPES and the corresponding concentrations of urea/taurine for the relevant wells, (all from Sigma).

Taking care to minimise light exposure, cells were transferred to a FLEXstation 3 (Molecular Devices, San Jose, CA, USA). Fluorescence readings (dual excitation 340nm/380nm for Ca2+ bound and Ca2+ free Fura-2 Am, respectively, and 510 nm emission) were taken every five seconds for two minutes. At 30 seconds, a compound delivery of either Ca2+ free PBS (negative control), glucose 0.5 M in Ca2+ free PBS, or ionomycin 10 µM in Ca2+ free PBS (positive control) was transferred to the cell culture plate.

Maximum change in the 340/380nm excitation ratio post compound delivery relative to baseline levels was calculated for each well. A total of three separate repeat experiments was conducted with three wells per treatment type per plate. Data were analysed by ANOVA with Dunnett’s post-hoc test relative to control wells incubated in plain media. Results are presented in Appendix B Figure 2.

287

Appendix B Figure 2: Summary of Taurine and urea pre-treatment on the intracellular calcium response of TR146 cells to glucose. Data are mean ± SEM, n = 9 (three wells per plate for three repeat experiments).

288