Salivary protein glycopatterns for natural regulation of oral microbiota

The oral microbiota is the direct precursor of dental caries and periodontitis, which are the most common microbial-induced diseases worldwide. The distinct microenvironment at the oral barrier breeds unique microbial communities, which are regulated by host factors (inammation or dietary sugars)1. Increasing evidence indicates that dysbiosis of oral microbial communities is associated with many human diseases2-5. Our studies demonstrated that human disease could induce different alterations in salivary protein glycopatterns6. However, the relationship between salivary protein glycopatterns and oral microbial communities is unknown. Here, we report that altered salivary protein glycopatterns, namely, fucosylated or sialylated structures induced by gastric cancer (GC) or type 2 diabetes mellitus (T2DM), respectively, are also drivers of dysbiosis of oral microbial communities and ultimately dental caries and periodontitis. The fucosylated neoglycoproteins and sialic acid (SA) α2-3 galactose (Gal) structure can inhibit the growth or/and adhesion of Aggregatibacter segnis and Candida albicans from the oral cavity of patients with GC and T2DM, respectively. These ndings provide a novel theory that dynamic communities of oral microbiota are regulated naturally by host salivary protein glycopatterns, having important implications for developing new carbohydrate drugs for oral and body health.


Main Text
The microbiota in the oral cavity plays an important role in maintaining oral health and even body health.
The nososymbiocity of the polymicrobial communities involved in both dental caries and periodontitis is largely regulated by host factors, predominantly dietary sugars and in ammation. Recently, increasing evidence has shown that the oral microbiota is associated with human diseases, including diabetes, obesity, and cancer 7,8 . Our previous studies demonstrated that different types of diseases (e.g., gastric cancer (GC), type 2 diabetes mellitus (T2DM), hepatopathy, and breast disease) could induce various alterations in salivary protein glycopatterns (e.g., fucosylation in GC and sialylation in T2DM) 9-12 . Furthermore, can altered salivary protein glycopatterns also impact the dynamic and polymicrobial oral microbiome? We hypothesize that altered salivary protein glycopatterns are possible drivers leading to dysbiosis of oral microbial communities and ultimately dental caries and periodontitis.
To address this hypothesis, we rst studied the changes in the oral microbiota associated with GC.
Among the 24 samples, an estimated 79 thousand raw sequence reads per sample with a median length of 423 base pairs is obtained. After quality trimming and chimera checking, 75 thousand clean sequence reads per sample remained, which are recovered for downstream analysis. The detailed sequencing data of each sample are shown in Extended Data Table S1. In total, 789 unique species are identi ed in the oral cavity of healthy volunteers (HVs) and patients with GC; 525 species are common in the HVs and GC patients, while 128 and 136 species are observed in only the HVs and GC patients, respectively (Fig. 1A).
The alterations in the microbial communities are evaluated between the HV and GC groups. The alphadiversity analysis shows that there are no signi cant differences between the groups (Fig. 1B-E), while the beta-diversity analysis shows that the oral microbiota pro les of the two groups are different, and can be distinguished using PCA, PCoA and NMDS plots (Fig. 1F-H).

Page 3/12
To further investigate the composition of oral microbial communities, the Wilcoxon test is used to compare the variance at each taxonomic level between the HV and GC groups ( Table 1). The results show that there are no signi cant differences at the phylum, class and order levels; however, the members of Carnobacteriaceae are increased in the GC patients compared with the HVs at the family level ( Fig. 2A).
The proportions of the genera Alloprevotella and Megasphaera are signi cantly increased whereas those of Granulicatella, Bregeyella and TM7 [G-6] are decreased in the GC group (Fig. 2B). Five species (Oral taxon 392, Oral taxon 308, Aggregatibacter segnis (A. segnis), Megasphaera micronuciformis (M. micronuciformis), and Oral taxon 396) are signi cantly increased (p<0.05) and 2 (Streptococcus salivarius (S. salivarius) and Oral taxon 870) are signi cantly decreased (p<0.01) in the GC patients compared with the HVs (Fig. 2C). Furthermore, the relative abundances of A. segnis, M. micronuciformis and S. salivarius are veri ed by PCR and qPCR, and their results are consistent with the results obtained by 16S rDNA sequencing (Extended Data Fig. S1). Our previous studies demonstrated that GC could induce alterations in salivary protein glycopatterns, especially fucosylated glycoproteins, in the saliva of GC patients 13 15 ; it is a normal commensal in oral humans but is rarely reported as a pathogen. Infective endocarditis is the most well-known infection but is still rarely reported to be associated with A. segnis 16 . Periodontal disease, acute appendicitis, pancreatic abscess, and cholecystitis have also been associated with A. segnis on occasion 17,18 . Previous studies have focused mainly on case reports, and the true prevalence of this bacterium in human infections is probably underreported; the underlying mechanism associated with diseases is not yet explicitly known. To understand the related mechanisms, the glycopatterns of the outer membrane and total protein of A. segnis and A. segnis treated with 30 and 100 μg/mL Fuc-BSA are analyzed by lectin microarrays. There are 9 lectins (e.g., Jacalin, GSL-II, and SNA) that show decreased expression levels of the glycopatterns (e.g., Galβ1-3GalNAcα-Ser/Thr, GalNAcα-Ser/Thr and Sia2-6Gal/GalNAc), while there are 3 lectins (PHA-E, LEL, and MAL-I) that show extremely increased expression levels of the glycopatterns (Bisecting GlcNAc, high mannose-type N-glycans, and Galβ-1,4GlcNAc) on the outer membrane of A. segnis treated by Fuc-BSA (Extended Data Fig. S5A). However, 4 lectins (PHA-E, PNA, LEL and GSL-I) show increased expression levels of glycopatterns (bisecting GlcNAc, Galβ1-3GalNAcα-Ser/Thr, high mannose-type Nglycans, and αGalNAc/αGal) in the total protein of A. segnis treated with Fuc-BSA (Extended Data Fig.  S5B). These results implied that Fuc-BSA mainly affects glycopatterns on the outer membrane of A. segnis. Annotation of A. segnis in the carbohydrate-active enzymes database (CAZy) revealed that 24 putative glycosyltransferases (GTs) are distributed among 11 glycosyltransferase families. Of these, 8 GTs are associated with lipopolysaccharide (LPS) biosynthesis, 7 GTs are associated with cell wall formation, and others are associated with glycosylation, glycogen synthase, and carbohydrate metabolism (Extended Data Table S2) 19 . The qPCR results show that 3 mRNAs of LPS-associated GTs (NCTC10977_00605 (Rfaq_1), NCTC10977_00729, and NCTC10977_00649 (WaaA)) are signi cantly decreased and that the mRNA of mannosyltransferase OCH1 is signi cantly increased in A. segnis treated with Fuc-BSA ( Fig. S7A and S7B), which indicate that Fuc-BSA interfered with the biosynthesis of core oligosaccharides and O-antigenic polysaccharides of LPS. These results suggest that Fuc-BSA mainly altered the glycopatterns of the outer membrane of A. segnis.
Second, Candida albicans (C. albicans) is the most prevalent human fungal pathogen causing infections and oral diseases in T2DM patients 20,21 . Our previous work showed that the expression level of the SAα2-3 structure in the saliva of T2DM patients is signi cantly lower than that of healthy people, which is related to T2DM susceptibility to avian in uenza virus 10 . Here, sialoglycoproteins from bovine milk are used to inhibit the growth and adhesion of C. albicans 22 . The results show that their growth can be inhibited gradually by different concentrations (50, 100, 200, and 400 μg/mL) of sialoglycoproteins (Fig.   4A) and be inhibited entirely by 400 μg/mL sialoglycoproteins with an IC50 of 142.8 μg/mL (Fig. 4B). To determine the roles of Siaα2-3/6Gal glycans, 400 μg/mL desialoglycoproteins (sialoglycoproteins treated with sodium periodate), sialoglycoproteins without Siaα2-3Gal (sialoglycoproteins treated with α2-3 neuraminidase S), sialic acid, and BSA are also used to inhibit the growth of C. albicans. The results show that they cannot inhibit the growth of C. albicans, which implies that the Siaα2-3 moieties of sialoglycoproteins play a major role against C. albicans. Furthermore, 200 μg/mL sialoglycoproteins and desialoglycoproteins are used to blockade C. albicans to adhere to CAL-27 cells. Notably, sialoglycoproteins signi cantly inhibit the adhesion of C. albicans to CAL-27 cells; however, desialoglycoproteins cannot inhibit this process. These results also demonstrate that the Siaα2-3Gal moieties of sialoglycoproteins play a major role in inhibiting the adhesion of C. albicans to CAL-27 cells (Fig. 4D, E).
Currently, there are several processes that underlie the transition of a microbial community to a state of dysbiosis. In particular, diet can affect the community composition, and overexposure to dietary carbohydrates and host factors promotes the production of extracellular polymeric substances (EPS) and acidic metabolites, which cause the accumulation of acidogenic and aciduric microorganisms, thus driving the transition to a pathogenic bio lm community [23][24][25] . Bacteria can obtain nutrients from glycoproteins in saliva and gingival crevicular uid during fasting 26,27 . Glycoproteins can be broken down into sugars and proteins and metabolized by bacteria that can gain energy to survive. During metabolism, sugars and proteins are further converted into acidic and basic small molecules, which can neutralize each other to maintain a neutral state in the mouth. If sugar consumption is low and infrequent, the microbial communities on teeth remain stable. Despite being able to produce acids that demineralize enamel, the episodic pH decrease can be readily neutralized by saliva, which restores and maintains the mineralization of enamel 28 . However, once a community has transitioned to a dysbiotic state, the structural stability of functionally specialized components will allow the condition to persist for an extended period of time, promoting the development of oral diseases such as periodontitis and dental caries [29][30][31] .
However, our studies demonstrated that human disease could induce different alterations in salivary protein glycopatterns 6 . The altered salivary protein glycopatterns (e.g., fucosylated or sialylated structures) induced by GC or T2DM are also drivers of dysbiosis of oral microbial communities and ultimately dental caries and periodontitis. The fucosylated neoglycoproteins and SAα2-3Gal structure can inhibit the growth and adhesion of A. segnis and C. albicans from the oral cavity of GC and T2DM, respectively. These ndings provide a novel theory that dynamic communities of oral microbiota are regulated naturally by host salivary protein glycopatterns, having important implications for developing new carbohydrate drugs for oral and body health.

Study Population
The collection and use of human saliva samples for the research presented here were approved by the Ethical Committee of Northwest University (Xi'an, China), First A liated Hospital of Xi'an Jiaotong University (Xi'an, China). Written informed consent was received from participants for the collection of their saliva samples. This study was conducted in accordance with the ethical guidelines of the Declaration of Helsinki. After a standardized endoscopic procedure and histopathological evaluation, individuals who were diagnosed with GC (N=12) were enrolled in this study. The age-and sex-matched HVs (N=12) were free of any precancerous lesions, as con rmed by gastroscopy. All the participants were ≥ 40 years old, and no signi cant differences were observed between the two groups regarding demographic, socioeconomic, and lifestyle characteristics. Patients who received preoperative radiotherapy, chemotherapy, chemoradiotherapy or antibiotic therapy were excluded from the study. The clinical characteristics of the HVs and GC patients are summarized in Extended Data Table S3.

Sequencing and Data Analysis
Bacterial genomic DNA was extracted by a Bacterial DNA Isolation Kit D3350-01 (OMEGA, USA) following the manufacturer's instructions. The 16S rDNA V3-V4 regions were ampli ed by speci c bacterial primers 341F-806R with barcodes 341F: CCTAYGGGRBGCASCAG, and 806R: GGACTACNNGGGTATCTAAT. All PCRs were carried out with Phusion High-Fidelity PCR Master Mix (New England Biolabs, USA). The PCR products were extracted and quanti ed, pooled in equimolar concentrations and sequenced using the Ion S5 TM XL platform according to the manufacturer's recommendations. The methods for sequence screening, diversity analysis and taxonomy-based analysis were performed as described in previous reports 32 . Differences between two arbitrary data sets were tested by the Wilcoxon test for each species using SPSS statistics 20.

Synthesis of neoglycoproteins
The synthesis of neoglycoprotein is divided into three steps (Extended Data Fig. S6A): the monosaccharides coupled with the linker (5-(4-hydroxyphenyl)pentanoic acid) to form glycan conjugates, and the carboxylates (-COOH) from glycan conjugates reacted to NHS in the presence of EDC, resulting in a semistable NHS, followed by reaction with amines (-NH2) of BSA to form amide crosslinks. Compared with the band of identically processed BSA, the upward migration bands of neoglycoproteins are obviously observed in the SDS-PAGE results (Extended Data Fig. S6B).

Lectin Microarrays and Data Analysis
A lectin microarray was produced using 37 lectins with different binding preferences covering N-and Olinked glycans 9 . The Cy3-labeled glycoproteins or bacteria were incubated on a lectin microarray at 37 °C for 3 h with gentle rotation in the dark. The slides were washed three times with PBST and PBS, centrifuged dry and scanned immediately with 70% photomultiplier tubes and 100% laser power settings using a Genepix 4000B confocal scanner (Axon Instruments, USA). The acquired images were analyzed at 532 nm for Cy3 detection by Genepix 3.0 software. Differences between two arbitrary data sets or multiple data sets were tested by Wilcoxon test or one-way ANOVA for each lectin signal using SPSS statistics 20.

Real-Time PCR (qRT-PCR)
Total RNA was extracted using the Bacteria Total RNA Isolation Kit (Sangon Biotech, China) and converted into cDNA using the PrimeScript RT reagent Kit (TaKaRa, Japan). The cDNAs were then subjected to real-time quantitative polymerase chain reaction (qRT-PCR) analysis (ABI ViiA TM 7, Bio-Rad, USA) using gene-speci c primers (Extended Data Table S4) and TB Green Fast qPCR Mix (TaKaRa, Japan). The 16S rRNA gene was used as an endogenous control for normalizing the relative expression of target genes.

Isolation of Sialoglycoproteins from Bovine Milk
According to the protocol in the literature 22 . In brief, serotonin-functionalized magnetic particle conjugates were thoroughly washed with binding buffer (20 mM Na 3 PO 4 , pH 6.0) and incubated sequentially with 1 mL of binding buffer containing 0.2 mL bovine milk with gentle shaking at room temperature for 3 h without light. The unbound proteins were removed from the conjugates with binding buffer until there were no detectable proteins. Finally, the sialoglycoproteins were eluted from the conjugates with eluting buffer (0.

Adhesion Assays
A. segnis or C. albicans were washed with fresh medium three times, centrifuged and diluted with fresh medium containing 100 μM FITC-d-Lys (Xiamen Bioluminor Bio-Technology, Xiamen, China). After 30 min of incubation at 37 °C, the cells were centrifuged, washed with fresh medium three times, and then resuspended in PBS. CAL-27 or HOEC cells (1×10 5 cells) were inoculated into confocal culture dishes (JingAn Biotechnology, Shanghai, China) and cultured in complete medium. Cells were grown to 60-70% con uence and then synchronized by serum starvation overnight. One hundred microliters of serum-free DMEM with 5 μM DiD (AAT Bioquest, USA) was pipetted onto the dishes and gently agitated until all the cells were covered. After 30 min of incubation at 37 °C, the cells were washed three times with DMEM. C.
albicans or FITC-labeled A. segnis (1×10 6 cells/mL) was added to the dishes in 500 μL serum-free DMEM, which was allowed to infect HOEC or CAL-27 cells for 45 min for adhesion. The infected cells were washed three times with 1× PBS and immobilized by incubating with 0.2% Triton X-100 in 4% paraformaldehyde for 30 min at RT. After DAPI staining (Thermo Fisher Scienti c, Waltham, USA), a laser scanning confocal microscope FV 1000 (Olympus, Tokyo, JPN) was used to acquire the images with the merged channels of FITC (Ex/Em(nm)=488/520), DiD (Ex/Em(nm)=650/670) and DAPI (Ex/Em (nm)=358/461). A uorescence microscope (Olympus, Tokyo, JPN) was used to observe the adhesion of C. albicans to cells.