Salivary microbiome in patients undergoing hemodialysis and its associations with the duration of the dialysis

Background Chronic kidney disease (CKD) patients, especially those with end stage renal disease (ESRD) undergoing hemodialysis (HD), exhibit high prevalence of periodontitis. This cross-sectional study aimed to investigate the periodontal status of HD patients and its relationship with salivary microbiome. Methods One hundred eight HD patients and one hundred healthy control individuals were recruited. They were subjected to periodontal examination followed by saliva samples collection for 16S rRNA gene sequencing. Results The HD patients were with worse periodontal health status, and exhibited higher salivary microbial diversity and lower richness. The periodontal pathogens were significantly enriched in the HD patients. The inferred functional analyze showed microbes enriched in the HD patients were mainly related to metabolism. Despite the periodontal status and overall structure of the microbiome were not significantly altered as the HD duration prolonged, the abundance of Lachnospiraceae [G-2] sp. |HMT_096| is positively correlated with the duration of HD and the community periodontal index (CPI). Five OTUs (operational taxonomic units) belonging to the phyla Firmicutes were enriched as the duration prolonged, and four OTUs originated from the phyla Proteobacteria were negatively related with the CPI index. ESRD patients undergoing HD exhibited microbiota structural, compositional and functional differences compared with the healthy controls. And the species changed as the duration of hemodialysis prolonged. Conclusions End stage renal disease changes salivary microbiome and is a risk factor for oral dysbiosis.

encompasses 10 times more cells and 100 times more genes than the host [11]. The oral microbiome can damage the dynamic balance between health and disease, locally and systemically [12]. The microbiome in saliva sheds from adhering microbial niches on various intraoral surfaces and appears to be the representative of the oral microbiome integrity. Salivary microbiome has a critical role on the genesis of oral diseases, mainly including dental caries, periodontitis and peri-implantitis [13][14][15]; non-oral diseases such as cardiovascular disease [16], cancer [17,18] and low birth weight neonate [19] are also somewhat related to the dysbiosis of salivary microbiome.
Gut dysbiosis has already been well described in CKD patients [20,21] and ESRD patients undergoing HD [22]. The oral cavity is the primary entry site of the gastrointestinal, additionally, and the oral pathogenic microbes has been described to induce the dysbiosis of gut microbiota in an animal model [23]. Previous studies, using microbiological methods with limited scope, have found that the contents of periodontal pathogens increased in ESRD patients as compared to non-CKD controls [24,25]. However, until now, very few studies analyzed the salivary microbiota compositions in HD patients.
In this study, we aimed to compare the periodontal status and saliva microbiota between HD patients and healthy persons, and further investigated the association between the salivary microbiome and the duration of HD treatment. These results enrich our understanding of the relationship between general disease state, oral microbiome changes and oral health status.

Study population
This study was approved by the Institutional Review Board of the Sichuan University Hospital. The case group comprised 108 patients undergoing HD three times per week at the West China Hospital, Sichuan University, Chengdu, China. They were in stable clinical conditions for at least 3 months. Patients with diabetic nephropathy or kidney transplantation, or patients who had been on absorbable antibiotics or undergoing periodontal disease treatment within the preceding 3 months or have smoking habits were excluded. 100 healthy matched subjects were recruited as controls. All participants gave a written informed consent before participation, and the study was carried out in accordance with those ethical standards in the Declaration of Helsinki.

Oral examination
One trained dentist performed all the clinical measurements. The Community Periodontal Index (CPI) was used to determine the oral periodontal health status, with the score of 6 sites for 10 World Health Organization index teeth [26]. The number of decayed, missing, and filled teeth (DMFT) was assessed according WHO guidelines [27]. To achieve a high quality of the data, an examiner repeatability exercise was undertaken prior to the initiation of the study. The kappa score for intra-examiner reliability was 0.94.

Saliva collection and DNA isolation
After the oral examination, saliva samples were collected by the same dentist. All participates were instructed not to eat or brush their teeth at least 2 h before sampling. A Salivette (Sarstedt, GER) collection swab with hard pack was used for the saliva collection and storage. Whole unstimulated saliva was collected using the swab in floor of the mouth. The swab was placed back into the collection tube, and stored at − 80°C before subsequent analysis. Total DNA was isolated from saliva using the QIAamp DNA micro Kit (QIAGEN Sciences, MD, USA) per the manufacturer's recommendations, with minor modification by adding an extra lysozyme (3 mg ml − 1 , 1.5 h) treatment step to lyse the bacterial cell.

16S rRNA gene amplification and sequencing
Total DNA was extracted from saliva using the QIAamp DNA micro Kit (QIAGEN Sciences, MD, USA) per the manufacturer's recommendations, with minor modification by adding an extra lysozyme (3 mg ml − 1 , 1.5 h) treatment step for lysing the bacterial cell. The V3-V4 regions of the bacterial 16S rRNA gene were PCR amplified using barcoded 338F and 806R. Amplicons were extracted from 2% agarose gels and purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) and quantified using Quanti-Fluor™ -ST (Promega, USA). Purified amplicons were pooled in equimolar and paired-end sequenced (2 × 250) on an Illumina MiSeq platform (Illumina, San Diego, CA) according to the standard protocols.

Sequence information and data analysis
The raw fastq files were first quality-filtered by Trimmomatic and then merged by FLASH. OTUs were clustered by 97% cutoff using UPARSE (version 7.1). To determine the species name and HOT (Oral Taxon ID), sequences were characterized based on the HOMD database. The relative abundance of each bacterial taxon was calculated and typically presented as "Mean ± SD". The core microbiome of salivary community was analyzed based on previous study [28]. Mann-Whitney U-test was used to compare the differences between the two groups (P < 0.05), and Kruskal-Wallis test was used for comparation among three groups (P < 0.05). X 2 test or Fisher's exact test was used to compare the categorical variables in the contingency tables (P < 0.05). Principal co-ordinates analysis (PCoA) and supervised Partial least squaresdiscriminant analysis (PLSDA) analysis were assessed using the SimcaP software (version 13.0, Umetrics, Umea, Sweden). Linear discriminant analysis (LDA) effect size (LEfSe) method was implemented with LEfSe (version 1.0). The functionality of the microbiota was performed using PiCRUST and compared between the two groups. Cytoscape (version 3.6.0) was used to visualize the correlations between the microbiota and duration of HD and other clinical index, with a Spearman's correlation coefficient (|correlation ρ| > 0.3 or 0.4, and P < 0.05).

Sociodemographic and Oral health characteristics of the study population
One hundred eight HD patients and one hundred healthy subjects were recruited. The two groups matched in age, gender and body mass indices (BMIs) ( Table 1). For the HD group, individuals with CPI 1 (23.0% v. 0.9%) and CPI 2 (39.0% v. 10.2%) were less common, while the percentages of CPI 3 (27.0% v. 41.7%) and CPI 4 (3.0% v. 47.2%) individuals were all significantly higher than those in the healthy controls. Both groups have no individuals with CPI X. There were 45 (41.7%) individuals with moderate periodontitis and 51 (47.2%) individuals with severe periodontitis in the HD group, compared with 27 (27.0%) and 3 (3.0%) in the control group. Namely, the percentage of individuals with CPI 3 and CPI 4 in the HD group was nearly 3 times that in the control group. Meanwhile, the HD group demonstrated greater number of missing teeth compared with healthy group (P = 0.013).
Overview of salivary microbial differences among the patients In total, 29 known phyla and 631 genera were identified, and 1430 OTUs were detected at 3% dissimilarity. To study the richness and diversity of the microbiome from the samples, we compared a series of alpha diversity metrics between the HD patients and the healthy controls, including the Chao index, Shannon index and observed OTUs (Fig. 1a). The HD patients had a lower richness estimator (Chao, P = 0.012), and a higher diversity index (Shannon, P = 0.016). Uncultivated phylotypes did not significantly differ between the two groups (P = 0.610). We generated PCoA plots to investigate the relationship of the composition. The unweighted UniFrac plot showed a distinct separation of the HD samples and the healthy controls based on PC1 (13.28%) and PC2 (8.91%) (Fig. 1b). Consistently, the dissimilarity tests, Adonis and ANOSIM, also revealed significant difference between the two microbial communities (supplementary Table S1). We constructed a Venn diagram and found that 696 OTUs were common in the two communities, while 298 and 436 OTUs were unique to the salivary microbiome of healthy controls and that of HD patients, respectively (Fig. 1c). "-" = vacancy CPI: The data are presented as n (%).The categories of CPI were defined as: 0 = healthy gingiva; 1 = bleeding observed directly or by using a mouth mirror after probing; 2 = calculus detected during probing but with all of the black bands on the probe visible; 3 = pocket of 4 to 5 mm (gingival margin within the black band on the probe); 4 = pocket of 6 mm (black band on the probe was not visible); X = excluded sextant (< 2 teeth present) Next, we evaluated the oral microbial community characteristics of the two niches. The taxonomic compositions at the phylum level varied widely between groups (Fig. 2a). Firmicutes, Proteobacteria, Bacteroidetes, Actinobacteria and Fusobacteria phyla were the dominant phylum (97.66% v. 95.91%) in both groups. The relative abundances of Firmicutes (P = 0.008), Bacteroidetes (P = 0.002), Spirochaetae (P = 0.001), Synergistetes (P < 0.001), Tenericutes (P < 0.001), Gracilibacteria (P < 0.001), and the total relative abundance of the other 21 phyla (P < 0.001) were significantly higher in the HD patients, whereas the abundances of Proteobacteria (P = 0.074) and Actinobacteria (P < 0.001) decreased (Fig. 2b). The microbial shift was illustrated in more detail at the genus level. We analyzed the top 30 genera (supplementary Table S2), and the taxa with significant difference were shown in supplementary Table S3, where more than half of these 30 genera were listed. Compared with the healthy cohort, the relative abundances of Ruminococcaceae (P < 0.001), Capnocytophaga (P < 0.001), Porphyromonas (P = 0.001), Veillonellaceae (P < 0.001), Granulicatella (P < 0.001) and other 9 genera were significantly enriched in the HD

Species associated with periodontal diseases
Considering CKD as a risk factor for periodontal health, we evaluated the microbes that are the important periodontal pathogens belonging to the "red complex" and "orange complex" [29]. HD patients appeared to have Fig. 3 Core salivary microbiome with relative abundance > 1% or relative abundance of < 1% while presenting statistically difference in the HD patients or the healthy controls. The core microbiome was defined as species present in 75% or greater of samples. The inner box labeled with 1 indicates the species is more abundant in the healthy controls than in the HD patients. The inner box labeled with 2 indicates the species is more abundant in the HD patients than in the controls. The significance of differences between groups was assessed using Mann-Whitney U-test (*P < 0.05, **P < 0.01, ***P < 0.001) higher levels of periodontal pathogens except for Prevotella nigrescens (P = 0.147) (supplementary Table S4).

Predicted microbial function
We generated PICRUSt to predict the gene functions in the salivary microbiome (Fig. 4). Microbes with greater relative abundances in the healthy community had functions that were most related to cellular processes and signaling (ABC transporters, secretion system and pores ion channels) and energy metabolism (oxidative phosphorylation). In contrast, the microbes in the samples of the HD patients were more likely to have functionality related to metabolism including carbohydrate metabolism (fructose and mannose metabolism), amino acid metabolism (valine leucine and isoleucine degradation, beta alanine metabolism), xenobiotics biodegradation and metabolism, as well as lipid metabolism. Moreover, microbes with the function of sporulation also increased in the HD patients.
Associations of the salivary microbiome with duration of HD therapy and clinical indices Next, we performed a correlation analysis of the differential microbiota at the OTU level with the duration of HD therapy and oral clinical indices. Figure 5 showed the constructed related network (|correlation ρ| > 0.3 or 0.4, and P < 0.05), and the supplementary  To determine whether the duration of HD therapy could be a potential effect on the ESRD patients' salivary microbiome, we divided all the patients into three groups: (i) HD1 group (n = 23): 3 m < duration of HD ≤1y; (ii) HD2 group (n = 61): 1y < duration of HD ≤5y and (iii) HD3 group (n = 24): duration of HD >5y. All the patients in the HD3 group were with periodontitis and the percentage of individuals with CPI 3 and CPI 4 was slightly higher than the other two groups (supplementary Table  S6). The overall structure of the microbial community (supplementary Figure S1, supplementary Table S7) were not altered significantly. No significant difference was observed at the phylum level (supplementary Table S8) and among periodontal pathogens (supplementary Table S9).

Discussion
Our previous study have found that the ESRD patients undergoing HD suffered a high prevalence and severity of periodontitis [8], further related with the survival of this population [30,31]. Periodontitis has been identified as a novel risk factor for morbidity and mortality in patients undergoing long-term hemodialysis treatment [32][33][34][35]. Moreover, nonsurgical periodontal treatment, as a relatively simple intervention, has beneficial impact on periodontal status, circulating inflammatory markers and nutritional status in ESRD patients [36][37][38] that underlines the importance of oral health and efficient periodontal treatment to maintain prolong survival and quality of life in these patients. However, very few patients have regular dental care [39,40] suggesting that more effort should be paid to provide reasonable clinical guideline for dental treatment [41] and to improve the oral health status of this population.
ESRD with HD alters the gut microbiome and may have an impact on the human oral microbiome. Alterations in microbial ecology are related to both oral and systemic diseases, including periodontitis. A recent study described the profile of the subgingival microbiome of peritoneal dialysis patients with periodontitis [42]. In this study, we aimed to investigate the microbial salivary profiles of HD patients and further explore the association between salivary microbiome, periodontal health status and duration of HD. Our finding of variations in oral health status and in salivary microbiome between the HD patients and the healthy individuals suggests that ESRD with HD obviously disturbed the salivary microbiome and periodontal health; duration of HD may further alter the abundance of some specific species, meanwhile, alterations of species may also influence the periodontal health states of the HD patients. Saliva is a microbial reservoir that contains microbes shedding from both supra and niches. Hence, intraoral transmissions of pathogenic bacteria are likely to be mediated through saliva [43]. Salivary microbiome can enter the tract by swallowing, which may cause imbalance of the intestinal flora [23]. To elucidate the changes of salivary microbiome in ESRD patients can expand the understanding of the interactions between oral microbiome and systemic health. A number of previous studies have used saliva sample as an easy, noninvasive and inexpensive diagnostic tool to assess healthy or disease conditions over the past decades [44][45][46], including periodontitis [47]. Salivary changes have already been reported by previous studies [48], and understanding of the changes in salivary microbiome may present a new point to explore the relationship between ESRD undergoing HD and the other oral manifestations caused by this status, such as taste change, calculus formation and fungal infection. Moreover, previous study reported the salivary microbiome of immunoglobulin A nephropathy (IgAN) differed from that of healthy controls [49]. Inspired by these findings, in this study, we collected saliva samples from the 208 individuals.
Taxonomic differences were detected between the HD patients and the healthy controls at phylum, genus and OTU level. Besides "the others" sort, 7 of the top 10 phyla show significant difference compare with the healthy controls, indicating obvious overall changes in the compositions of the two microbial communities. Changes at phylum level in this study are consistent with previous finding, suggesting that the development of periodontal diseases is intimately linked to shift from a symbiotic dental biofilm, composed mostly of facultative anaerobic bacteria (e.g. Streptococci and Actinomyces), to a dysbiotic microbial community of anaerobic microbes belonging to the phyla essentially comprised of Firmicutes, Spirochaetes, Synergistetes, and Bacteroidete [50]. An increase in Firmicutes has also been reported in the gut microbiome of the ESRD patients during HD [22].
The core salivary microbiome of both HD patients and healthy controls comprised of 104 species (58 species shared by the two groups, while 46 differed). The data indicate that there exists a core salivary microbiome composed of microbes that are most suited to healthy individuals, whereas ESRD with HD therapy might modify this environment. Actinomyces odontolyticus, core microbiome of the healthy community with relative abundance > 1%, is a gram-positive facultative aerobic bacterium, which decreased significantly in the HD community. Previous studies have reported that A. odontolyticus was the host of the candidate phylum Saccharibacteria, being associated with human inflammatory mucosal diseases, thereby killed by a parasitic phase [51,52]. By contrast, Porphyromonas gingivalis, core microbiome of the HD community with relative abundance > 1%, is a convincedly proposed keystone bacterium in the genesis and development of periodontitis, which was at significantly higher level in the HD group. It is worth mentioning that all the three important periodontal pathogens belonging to the "red complex" were significantly elevated in HD patients, suggesting that ESRD with HD creates a pathogen-rich oral community that could further inducing a high prevalence and a serious degree of periodontitis in this cohort. On the contrary, early study has failed to find the periodontitisassociated taxa differed between CKD and control group [53]. The reasons might include that participants were in different stages of CKD and different primers and sequencing platform were used. Notably, there is a bidirectional relationship between oral microbiome and kidney disease. Evidence suggests that periodontal pathogens and those metabolic products have adverse effects on the survival of ESKD patients who are undergoing dialysis [30]. Moreover, moderate-to-severe periodontal disease also may increase the risk of cardiovascular mortality in the dialysis patients [54]. Though nephropathies also favor the appearance of oral diseases, periodontal examination is still not part of their daily standard medical assessments. Hence, periodontitis in this population must be carefully evaluated by dental professionals, and clinical dental treatment should be taken to reverse this amendable inflammation status.
According to the LDA score, it inferred that the metagenomic pathways that were enriched in the patients undergoing HD were mainly those involved in metabolism, ranging widely from carbohydrate, amino acid and lipid. Porphyromonas, showing significantly higher abundance in the HD group, were amino acid degrading species, which can break down peptides and proteins into amino acids and then degrade them to produce short-chain fatty acids, sulfur compounds, ammonia, and indole/skatole, which all act as virulent influences in periodontitis [55]. The relationship between cellular processes and signaling and the healthy population is unclear.
The severity of periodontal disease slightly worsened as the HD duration prolonged. Further, we found that OTUs, positively related with the duration of HD, were mainly originated from the phyla Firmicutes, while OTUs being negatively related with CPI index of the HD patients were principally from the phyla Proteobacteria, which is similar with the higher ratio of Firmicutes and Proteobacteria in the HD community. Interestingly, Lachnospiraceae [G-2] sp. | HMT_096 |, from the genera Firmicutes, was enriched as the extension of the duration of HD and was more abundant in the patients with severe periodontitis. The duration may not change the overall composition and structure of the salivary microbial community significantly, but contribute to changes of some specific species that may further stimulate a bad effect on the patients' periodontal status. Considering that the available data about oral microbiota of HD patients remain limited, all these correlated nods may be helpful in finding potential periodontal pathogens or probiotic bacteria and new reasonable periodontal treatment with the HD population in the future.

Conclusion
ESRD patients undergoing HD exhibited more severe periodontal health status, and had microbiota structural, compositional and functional differences compared with the healthy controls. Furthermore, we identified the species changed as the duration of hemodialysis prolonged. End stage renal disease changes salivary microbiome and is a risk factor for oral dysbiosis.
Additional file 1: Table S1. Dissimilarity Tests between the Healthy Controls and the HD Patients.

Acknowledgments
We wish to thank all participants in this study.
Authors' contributions XBD contributed to data acquisition, analysis and interpretation, drafted and critically revised the manuscript; QY and YQL contributed to design and critically revised the manuscript; XLC, MG and DS contributed to conception and critically revised the manuscript; HXX, QCX, TX and DL contributed to data acquisition; ACM, XDZ and XT contributed to data analysis and interpretation. All authors gave final approval and agree to be accountable for all aspects of the work.

Funding
The corresponding author QY is grateful to the National Natural Science Foundation of China and Sichuan Province Science and Technology Innovation Team Program for financially supporting this research (NSFC81722014 and 2017TD0016), respectively. The co-corresponding author YQL is grateful to the National Natural Science Foundation of China for financially supporting this research (NSFC31870065). Author XDZ thanks the National Key R & D Program of China during the 13th Five-Year Plan for providing financial support for this study (2016YFC1102700). The funding body had no role in study design, data collection and analyses, decision to publish or preparation of the manuscript.

Availability of data and materials
The raw sequence data and the patient data analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The design and procedures of this study were approved by the Institutional the West China Hospital of Stomatology, Sichuan University. Participants provided informed consent via a signed statement before participation.

Consent for publication
Not applicable.