The results showed that MW qualitatively contained almost the same metabolome information as that of US and SS. Quantitatively, a limited number of metabolites were characteristic of MW and SS, but individual differences were reflected in all three sampling methods (Table 3). These findings suggest that MW is a promising alternative to saliva samples for analyzing oral metabolome profiles.
With respect to qualitative analysis, the composition of metabolites in the MW samples was more than 80% identical to that of the US and SS samples, which have been previously used for oral metabolome analysis. Furthermore, the samples collected using the different methods contained almost the same composition (Fig. 1). Since the diversity of the oral microbiota in US, SS, and MW is similar according to the oral microbiome studies conducted by Jo et al.13,14, it can be expected that MW would reflect US and SS metabolome profiles as well. Figueira et al.19 reported findings from a comparative analysis of unstimulated, stimulated, and parotid saliva using nuclear magnetic resonance spectroscopy-based metabolome analysis, and detected 45, 44, and 44 metabolites, respectively. Although there are differences in the composition ratios, the metabolites detected are almost identical. This is consistent with our current results and validates the study because the types of metabolites detected in US and SS signified almost the same metabolome information.
The relative area of the metabolites not detected in MW samples was small (Supplementary Fig. S1), suggesting that the MW sampling method failed to detect few metabolites, perhaps due to the dilution of metabolites present at low-concentrations in the oral cavity, resulting in these metabolites not being detected in MW. Metabolites not detected in MW but detected in more than half of the US and SS samples included thymine, octopine, and O-phosphoserine, which are pyrimidine metabolites, a derivative of arginine and alanine, and a phosphate ester of serine, respectively. These metabolites have either not been quantified or detected in saliva or their quantitative values have been reported to not be large20. Overall, these results indicate that MW contains almost the same metabolome information as US and SS, but there are some components that are characteristic of only US and SS. These points should be considered when conducting research using MW sampling for oral metabolome analysis.
With respect to quantitative analysis, we observed that the oral metabolites obtained using the three collection methods were divided into three major clusters by hierarchical clustering analysis: clusters characteristic of MW, of SS, and of all three sampling methods with similar metabolome profiles (Fig. 2A, B and Table 3). Cluster 1, a cluster characteristic of SS, contained most of the amino acids (Supplementary Table S2). Figueira et al.20 reported that the metabolite compositions of US and SS differ and speculated the reason for the difference to be the increase in the proportion of parotid saliva with masticatory stimulation, which was reflected in the SS composition. In addition, as reported by Neyraud et al.21, amino acids are found to be more abundant in SS than in US. In MW, 15 metabolites were characteristic and most of them were considered to have oral origin, i.e., saliva, tartar, GCF, or the tongue coating (Table 4). The reason for the greater abundance of these metabolites in MW compared to that in US or SS is unclear. However, it is possible that these metabolites were more easily obtained by rinsing the mouth with water than by salivation, with or without stimulation. Accordingly, care should be taken when performing comparative analysis of these metabolites, as they may behave differently when compared to those in US or SS.
Hierarchical clustering and comparative analysis of the 108 oral metabolites commonly detected in the three sampling methods with no significant differences (Figs. 3 and 4) revealed that the differences among the subjects were larger than the differences among the three sampling methods. In other words, the profiles of these metabolites reflected individual differences rather than differences depending on the sampling method, suggesting that MW, like US and SS, may be an oral specimen that reflects individual differences. We have reported that the differences in the microbiome reflect individual differences rather than sampling method differences (MW, SS, US, and tongue) in our previous comparative study13. We, therefore, believe that similar results were obtained in the current study on metabolites.
The findings of this study must be seen in the light of some limitations. Salivary composition is well known to exhibit circadian rhythms. Kawanishi et al. reported significant differences in metabolite concentrations in unstimulated saliva between morning (8:00–9:00) and evening (17:00–18:00), and in stimulated saliva between morning and daytime (12:00–13:00) and between daytime and evening22. In the current study, as the samples were collected between 14:30 and 17:20, it is presumed that the observed differences between the subjects are unlikely to be greatly affected by the diurnal variation. However, its effects cannot be completely ruled out and need to be addressed in future studies. In addition, we profiled hydrophilic metabolites using CE-TOFMS. Since 99% of saliva is water, CE-TOFMS is suitable for profiling a wide range of metabolites. However, poorly water-soluble chemical compounds, such as lipids, have been found to be present in saliva23,24. Therefore, further analysis is needed to perform a fully comprehensive oral metabolome profiling. Moreover, although MW appears to be a useful method to collect oral samples from subjects who have difficulty producing saliva, such as patients with dry mouth and the elderly, to verify the usefulness of MW sampling for these individuals, it will be necessary to include subjects with low saliva volume, such as those with xerostomia.
