Understanding the human salivary metabolome - Wiley Online Library

0 downloads 0 Views 235KB Size Report
Published online in Wiley InterScience: 3 March 2009. Understanding the human salivary metabolome. Ienaka Takeda a,ey. , Cynthia Stretch ay. , Pamela ...
Research Article Received: 5 September 2008,

Revised: 7 January 2009,

Accepted: 7 January 2009,

Published online in Wiley InterScience: 3 March 2009

(www.interscience.wiley.com) DOI:10.1002/nbm.1369

Understanding the human salivary metabolome Ienaka Takeda a,ey, Cynthia Stretch ay, Pamela Barnaby a, Kriti Bhatnager a, Kathryn Rankin b, Hao Fu b, Aalim Weljie d, Naresh Jha a and Carolyn Slupsky b,c * Saliva is a readily accessible biofluid that is important for the overall health, aiding in the chewing, swallowing, and tasting of food as well as the regulation mouth flora. As a first step to determining and understanding the human saliva metabolome, we have measured salivary metabolite concentrations under a variety of conditions in a healthy population with reasonably good oral hygiene. Using 1H NMR spectroscopy, metabolite concentrations were measured in resting (basal) and stimulated saliva from the same subject and compared in a cohort of healthy male non-smoking subjects (n ¼ 62). Almost all metabolites were higher in the unstimulated saliva when compared to the stimulated saliva. Comparison of the salivary metabolite profile of male smokers and non-smokers (n ¼ 46) revealed citrate, lactate, pyruvate, and sucrose to be higher and formate to be lower in concentration in smokers compared with non-smokers ( p < 0.05). Gender differences were also investigated (n ¼ 40), and acetate, formate, glycine, lactate, methanol, propionate, propylene glycol, pyruvate, succinate, and taurine were significantly higher in concentration in male saliva compared to female saliva ( p < 0.05). These results show that differences between male and female, stimulated and unstimulated, as well as smoking status may be observed in the salivary metabolome. Copyright ß 2009 John Wiley & Sons, Ltd. Keywords: NMR; metabolomics; metabonomics; saliva; targeted profiling; smoking

INTRODUCTION Saliva is a complex biological fluid that is produced in, and secreted from, the salivary glands. It consists of 99% water with electrolytes, mucus, proteins, and small molecular weight metabolites making up the rest of the components (1,2). It is produced by three pairs of major salivary glands (parotid, submandibular, and sublingual) as well as several minor glands. Saliva is critical for preserving and maintaining the health of oral tissues, as it has a variety of functions including digestion and lubrication. At rest without stimulation, a small, continuous salivary flow, termed basal unstimulated secretion, covers, moisturizes, and lubricates the oral tissues (1). The submandibular gland contributes 65–70% of the total saliva volume, with 20 and 8% due to the parotid and sublingual glands, respectively (1,3). Stimulated saliva is produced primarily by the parotid glands, is released upon smell, taste, mechanical, or pharmacological stimulus (1) and contributes to most of the daily salivary production. Varying physiological, pathological, and environmental factors may cause changes in the amount and composition of both stimulated and unstimulated saliva (1,3). Saliva is an extremely important biofluid, as it is known that loss of salivary gland function, in cases such as radiation-induced xerostomia or Sjo¨gren’s syndrome, has been shown to have a profound effect on quality of life. Xerostomia impairs the ability to taste, chew, and swallow food, and alters oral microbial flora leading to the development of dental caries (4). It also causes problems with speech, and can be greatly demoralizing to patients (4). An understanding of the composition of this fluid may help to further understand the qualities of saliva, which potentially could aid those afflicted with low salivary production.

Advances in global profiling have impacted the number of salivary components that may be studied at one time. Protein profiling in saliva is becoming more advanced and sophisticated (5–7), and has even afforded a potential diagnostic for head and neck cancer (8). However, metabolite profiling of saliva remains largely under-explored despite demonstrated potential for biomarker detection in other biofluids such as urine and blood (9–12).

* Correspondence to: C. Slupsky, Department of Medicine, MRDC, 550A Heritage Medical Research Building, University of Alberta, Edmonton, Alberta, Canada. E-mail: [email protected] a I. Takeda, C. Stretch, P. Barnaby, K. Bhatnager, N. Jha Department of Oncology, University of Alberta, Edmonton, Alberta, Canada b K. Rankin, H. Fu, C. Slupsky Magnetic Resonance Diagnostic Centre, University of Alberta, Edmonton, Alberta, Canada c C. Slupsky Department of Medicine, University of Alberta, Edmonton, Alberta, Canada d A. Weljie Chenomx, Inc., Edmonton, Alberta, Canada e I. Takeda Department of Pathology, Tsurumi University, Yokohama, Japan y

Contributed equally to this work. Contract/grant sponsor: Alberta Cancer Board. Abbreviations used: DSS, sodium 2,2-dimethyl-2-silapentane-5-sulfonate; LDH, lactate dehydrogenase; PCA, principal components analysis; PLSDA, partial least squares discriminant analysis.

577

NMR Biomed. 2009; 22: 577–584

Copyright ß 2009 John Wiley & Sons, Ltd.

I. TAKEDA ET AL. Saliva has advantages over other biofluids, such as blood and urine, since its collection is non-invasive and relatively fast. Some salivary metabolites have been successfully identified using 1 H NMR (13–18), and inter-subject variability has been studied (16,17), but to date chemometrics has not been applied to saliva metabolite datasets to find potential markers for disease. As a first step in the understanding the human saliva metabolome, we employed high-resolution 1H NMR spectroscopy to determine if salivary metabolite composition differs due to gender, stimulation, or smoking status. This study will be the first to establish the composition and concentration of salivary metabolites in a normal human population, and how health choices, such as smoking, may affect the metabolic profile.

METHODS Participants Three studies were undertaken to establish differences between stimulated and unstimulated saliva, and to examine the effects of smoking and gender on unstimulated saliva. None of the participants consumed alcohol on the day of sample collection, and samples were collected at least 1 h after the last meal, initiated at approximately the same time for each participant (11:00 hours). Saliva was collected using aspiration from the floor of the mouth and labial vestibule. Samples were frozen immediately at 808C until ready for acquisition. Written, informed consent was obtained from each participant. The study was reviewed and approved by the Alberta Cancer Board Health Research Ethics Board. For comparison of stimulated versus unstimulated saliva, 62 non-smoker male subjects with no active periodontal disease, dental carries, or serious medical conditions, ranging in age from 40 to 76 years were recruited. Each participant in this part of the study provided two saliva samples: The first was collected under resting conditions (unstimulated) and the second directly after salivary induction with a citric acid solution applied directly to the tongue (stimulated). A subset of these male non-smokers (n ¼ 23) was chosen (based on age) for comparison of unstimulated saliva to a cohort of male smokers (n ¼ 23) who were chosen based on the fact that they smoked at least one cigarette per day. To examine the effect of gender on unstimulated saliva, 20 male and 20 female subjects with no active periodontal disease, dental carries, pregnancy (in the case of female subjects), or serious medical conditions, ranging in age from 21 to 63 years were recruited. For female subjects, menstrual status was not recorded. Of this group, five males and eight females were smokers (at least one cigarette per day). Sample collection and preparation For acquisition of salivary NMR data, the frozen harvested saliva was thawed, and spun at 3000 rpm to remove particulate matter. The salivary supernatant was removed and diluted 1:10 with Chenomx Internal Standard solution (Chenomx Inc., Edmonton, AB, Canada) (consisting of either 5 mM sodium 2,2-dimethyl-2silapentane-5-sulfonate (DSS) and 0.2% sodium azide in 99% D2O (for male stimulated and unstimulated saliva, and smoker saliva), or 5 mM DSS, 100 mM imidazole, 100 mM creatinine, and 0.2% sodium azide in 99% D2O (for male vs. female study)). NMR data acquisition and processing Data were acquired on either a 500 MHz Bruker Avance NMR spectrometer (male vs. female data), or a 600 MHz Varian INOVA

NMR spectrometer (stimulated vs. unstimulated and smoking vs. non-smoking) using standard Chenomx parameters which utilize a 1D NOESY pulse sequence with an acquisition time of 4 s, a sweep width of 12 ppm, and a pre-acquisition delay of 1 s as described previously (12,19). All spectra were zero-filled to 128k points and a weighted Fourier transform was applied to the time domain data with a 0.5 Hz line-broadening followed by manual phase and baseline correction in preparation for targeted profiling analysis. Concentration determination Quantification of salivary metabolites was achieved using Chenomx NMRSuite v4.6 (Chenomx Inc.). Briefly, the Chenomx profiler is linked to a database containing more than 250 metabolite NMR spectral signatures encoded at different spectrometer 1 H frequencies, including 500 and 600 MHz. Comparison of the spectral data obtained for each saliva sample with the Chenomx metabolite library results in a list of compounds together with their respective concentrations based on the known concentration of the added internal reference compound, DSS. The database compounds identified in this work have been verified for accuracy and precision, and this has been previously published (12). For each sample, up to 25 metabolites were assigned and quantified and included: acetate, acetone, alanine, arginine, citrate, dimethylamine, ethanol, fructose, formate, glucose, glycine, histidine, isopropanol, lactate, leucine, methanol, phenylalanine, propionate, propylene glycol, pyruvate, succinate, sucrose, taurine, tyrosine, and urea. Some metabolites (e.g. arginine, ethanol, fructose, histidine, and leucine) were observed in only a few samples, and were thus excluded from multivariate and univariate statistical analysis. When comparing stimulated and unstimulated saliva, citrate was excluded from the analysis since citrate was used for stimulation. Statistical analysis Multivariate data analysis (principal components analysis (PCA) and partial least squares discriminant analysis (PLSDA)) was conducted using SIMCA-P 11.0 (Umetrics, Umea˚, Sweden) with the application of mean centering and unit variance scaling. PLSDA was used because of its ability to deal with multivariate data and account for known class membership (12). Metabolite concentrations were log10-transformed prior to analysis. Metabolite concentrations were also subjected to Wilcoxon-rank sum tests using GraphPad Prism 4.0c software for MacIntosh (GraphPad Software Inc., San Diego, CA) to determine statistical significance. Significance was set at a ¼ 0.05 for all tests.

RESULTS A sample 1H NMR spectrum of unstimulated saliva from a non-smoking male participant with several resonances identified is shown in Fig. 1. Resonance assignments were made using Chenomx NMRSuite software. Spectra of both unstimulated and stimulated (not shown) saliva were dominated by lactate, acetate, and isopropanol. An investigation into the source of isopropanol revealed that it was used to sterilize the surfaces of the equipment between subject collections, and thus its presence in saliva was deemed to be a contaminant of the collection process. Therefore, isopropanol was removed from further analysis. Most other metabolites (except for arginine, ethanol, fructose,

578 www.interscience.wiley.com/journal/nbm

Copyright ß 2009 John Wiley & Sons, Ltd.

NMR Biomed. 2009; 22: 577–584

HUMAN SALIVARY METABOLOME

Figure 1. A typical 600 MHz 1H NMR spectrum of unstimulated saliva from a healthy non-smoking male: (a) aliphatic region and (b) aromatic region with vertical scale increased by a factor of 10. Various metabolites are indicated.

histidine, and leucine) were present in concentrations that were quantifiable in at least 13 out of 62 spectra. In cases where quantitation was not reliable for a particular compound (related to a particular spectrum), concentrations were either not used (for generation of a model using multivariate statistics), or reported as being less than 1 mM. Citrate was excluded in the comparison of stimulated and unstimulated saliva, as this compound was used as a stimulant. Analyses of the same spectrum by three separate analysts revealed quantitative agreement to within 10% for high-concentration metabolites and 50% for low-concentration metabolites. To determine differences between unstimulated and stimulated saliva, 62 healthy male non-smoking subjects were recruited and both unstimulated and stimulated saliva were collected from each. PCA of targeted profiling data for 18 metabolites obtained from analysis of NMR spectra of stimulated versus unstimulated saliva revealed class distinction (Fig. 2a). For better visualization, and to optimize class separation, PLSDA was performed using stimulation status as a classifier (Fig. 2b). Based on variable importance, among the most important metabolites

Figure 2. Comparison of unstimulated and stimulated male saliva. (a) PCA model based on targeted profiling of 1H NMR spectra of human saliva using 18 measured metabolites (excluding citrate). Stimulated saliva (&, n ¼ 62) and unstimulated saliva ( , n ¼ 62) from healthy adult males. (b) PLSDA model of stimulated saliva (&, n ¼ 62) versus unstimulated saliva ( , n ¼ 62) using stimulation status as a classifer. (c) Box and whisker plots comparing some of the measured salivary metabolite concentrations. p < 0.01 and p < 0.001.

579

NMR Biomed. 2009; 22: 577–584

Copyright ß 2009 John Wiley & Sons, Ltd.

www.interscience.wiley.com/journal/nbm

I. TAKEDA ET AL. for distinguishing stimulated from unstimulated saliva were glucose, succinate, propionate, acetate, taurine, glycine, alanine, sucrose, and dimethylamine. Non-parametric statistical analysis of individual metabolites demonstrated nearly all of them to have decreased significantly in concentration upon stimulation ( p < 0.05) (Table 1, Fig. 2c). However, although stimulated saliva is less concentrated than unstimulated saliva, the extent of dilution was not the same for each metabolite. The largest decrease in concentration was seen for propionate, acetate, glycine, succinate, formate, acetone, and taurine. Several metabolites (glucose, sucrose, phenylalanine, and tyrosine) that were not observable in all samples also decreased for those subjects where these metabolites were measurable. These results are reflected in the differences observed in the metabolite concentration ranges (Table 1). Metabolites that decreased the least in concentration upon stimulation included alanine, dimethylamine, lactate, and pyruvate. Methanol and urea did not appear to be significantly different between stimulated and unstimulated saliva, although the concentration of urea may be unreliable due to slight differences in pH between samples. Propylene glycol decreased in concentration in some subjects and increased in others (data not shown), but overall there were higher concentrations of propylene glycol in stimulated versus unstimulated saliva (Table 1). Since it is known that smoking cigarettes has an impact on the protein composition of saliva (20,21), we sought to determine whether smoking has an impact on the metabolite composition of saliva. To determine differences in salivary composition due to smoking status, unstimulated saliva from 23 of the male volunteers employed in the above study (chosen to have approximately the same age as the smoking volunteers, 50  8 years) was compared with unstimulated saliva from a cohort of 23 smoking males. PCA of targeted profiling data obtained from analysis of NMR spectra revealed class distinction with some overlap between the two groups (Fig. 3a). PLSDA was performed using smoking status as a classifier (Fig. 3b), and analysis revealed citrate, sucrose, tyrosine, pyruvate, formate, methanol, glucose, and alanine to be the most important variables separating the cohorts. A summary of metabolite concentrations for the unstimulated smoker saliva and significance using nonparametric analysis (compared with the cohort of 23 nonsmoking males) is shown in Table 1. Citrate, formate, lactate, pyruvate, and sucrose are all significantly different between the two groups with p < 0.05 (Table 1, Fig. 3c). Methanol and propylene glycol also appeared to be slightly higher in smoker saliva compared to non-smoker saliva, although not significant (Fig. 3c). Of particular interest, sucrose was in very high concentration in smoker saliva as compared to non-smoker saliva (Table 1, Fig. 3c). It is also known that gender has an influence over urinary metabolite concentrations (10–12). To determine differences in salivary composition due to gender, unstimulated saliva obtained from an additional 40 volunteers (20 males and 20 females) was obtained and analyzed using 1H NMR spectroscopy coupled with targeted profiling. The females (43  10) were slightly older than the males (35  12 years). Approximately one-third of the individuals in each group were smokers (smoked at least one cigarette per day). Except for glucose, which was higher in this group, and phenylalanine, which could be measured in more males in this group than the group of 62 non-smokers or 23 smokers, comparison of male median metabolite concentrations between this group of males and the combination of smokers

and non-smokers analyzed above revealed no significant differences (Table 1). PCA and PLSDA of targeted profiling data obtained from analysis of 1H NMR spectra of the cohort of male and female subjects revealed some class separation (Fig. 4a,b). Table 1 provides the median concentrations for the unstimulated female and male saliva, with an indication of the significantly different metabolites. Of interest, nearly all metabolites that were significantly different between male and female saliva were generally higher in concentration in the male subjects, and included acetate, formate, glycine, lactate, methanol, propionate, propylene glycol, pyruvate, succinate, and taurine (Table 1). Interestingly, while not significant, citrate appeared to be slightly higher in concentration in female saliva, while acetone appeared to be slightly higher in male saliva (Table 1, Fig. 4c).

DISCUSSION This study demonstrates that analysis of salivary metabolites in healthy subjects, using NMR-based metabolomics, reveals differences due to stimulation status, smoking status, and gender. Multivariate and univariate statistical analysis of metabolite concentrations obtained from stimulated and unstimulated saliva showed significant differences in metabolite concentration. Most of the metabolites decreased in concentration using citrate as a stimulant of salivary production. Our results show that some metabolites diluted to a lesser extent than others with stimulation. In particular, alanine, dimethylamine, lactate, and pyruvate were diluted less than acetate, acetone, formate, glucose, glycine, succinate, sucrose, phenylalanine, propionate, taurine, and tyrosine (citrate was not used in the analysis since it was used as a stimulant), whereas methanol and urea were not significantly different, and propylene glycol was higher in many subjects after stimulation. It is known that saliva secreted from the parotid and submandibular glands differ in composition, as saliva from the submandibular gland is more viscous than serous saliva secreted by the parotid glands (22). Moreover, stimulated and unstimulated saliva have been shown to have different protein compositions (23). During unstimulated or resting conditions, the submandibular salivary glands are the predominant source of saliva (24), while after stimulation with citric acid, the contribution to whole saliva from the parotid salivary gland doubles from 25 to 50% (25). In addition, previous studies have also shown that a direct relationship exists between parotid saliva composition and blood (26–30), and that parotid gland saliva has been shown to contain amino acids (31), pyruvate, and lactate (28). Our results are in agreement with the published data (26–32). All of the metabolites in the ‘less diluted’ group upon salivary stimulation are known constituents of blood or secreted by the parotid salivary gland (28,29,32). One interesting metabolite was methanol. While methanol has been shown to be a normal constituent of saliva (33,34), its source has not been identified. The minimal change in concentration between stimulated and unstimulated saliva would suggest that this metabolite is not simply from mouth flora, but is somehow secreted by the parotid glands, as we would expect the concentration to decrease if it simply came from mouth flora. The increase of propylene glycol with salivary stimulation in some cases was also unexpected. This chemical is a common ingredient in oral hygiene products such as mouthwash and toothpaste, and is commonly found as a solvent for intravenous, oral, and topical

580 www.interscience.wiley.com/journal/nbm

Copyright ß 2009 John Wiley & Sons, Ltd.

NMR Biomed. 2009; 22: 577–584

581

NMR Biomed. 2009; 22: 577–584

Copyright ß 2009 John Wiley & Sons, Ltd.

a

1.92 2.22 1.47, 2.55, 2.72 8.45 3.24, 3.49, 3.74, 4.64, 3.54 1.32, 3.35 3.10, 7.32, 1.04, 1.13, 2.36 2.40 3.47, 3.76, 3.82, 4.20, 3.25, 3.04, 7.18 5.76

H chemical shift (ppm)

3.55, 3.80, 3.85, 5.40 3.42 3.20,

3.94, 6.89,

3.65, 3.67, 3.81, 3.82, 3.88, 4.04,

3.27, 3.99, 7.37, 7.42 2.17 3.44, 3.54, 3.88

4.10

3.40, 3.41, 3.47, 3.53, 3.71, 3.72, 3.82, 3.84, 3.90, 5.23

3.78 2.71

1

Median concentrations.

Urea

Taurine Tyrosine

Propionate Propylene Glycol Pyruvate Succinate Sucrose

Glycine Lactate Methanol Phenylalanine

Acetate Acetone Alanine Citrate Dimethylamine Formate Glucose

Metabolite

Assignment

(1–586) (