Salivary microbiome in chronic kidney disease: what is its connection to diabetes, hypertension, and immunity?

This study was approved by the Institutional Review Board of Wuxi Second Hospital of Nanjing Medical University (Ref. 2018051). All participants gave written informed consent before participation, and the study was carried out in accordance with the ethical standards in the Declaration of Helsinki. The inclusion criteria for the groups of CKD and HC were as follows: To be a participant,  ≥ 18 years, no antibiotic use for  ≥ 1 month before sampling, without acute intercurrent disease and infections/diarrhea/kidney transplantation/pregnancy/breastfeeding, have not been administered medications such as antibiotics/probiotic/immunosuppression within 1 month before enrollment. The diagnosis of CKD was based on the criteria: either kidney damage or Egfr  < 60 ml/min/1.73 m² present for  ≥ 3 months. The markers of kidney damage include albuminuria [albumin excretion rate  ≥ 30 mg/24 h; urinary albumin creatinine ratio (UACR)  ≥ 30 mg/g], urine sediment abnormalities, electrolyte and other abnormalities due to tubular disorders, abnormalities detected by histology, and structural abnormalities detected by imaging [19]. Kidney damage refers to pathologic abnormalities or markers of damage, including abnormalities in blood or urine tests or imaging studies [20]. CKD patients in our present study had never undergone hemodialysis. Subjects with kidney damage (the markers of kidney damage as described above) or eGFR above 90 ml/min/1.73 m², current illness including diabetes and hypertension, and acute or chronic infections, were excluded from the HC group.

We collected demographic and health status information through face-to-face interviews and medical review. Fasting blood glucose, HbA1c, and blood pressure were assessed on the day of recruitment. An immunoturbidimetric test was used to assess serum immunological parameters on the day of sample collection by Freelite test (AU5421; Beckman Coulter, USA).

To investigate the influence of co-occurrence of diabetes and hypertension in CKD on the salivary microbiome, we divided the CKD patients into subgroups of diabetic CKD (DM-CKD) and non-diabetic CKD (nonDM-CKD), and subgroups of hypertensive CKD (HTN-CKD) and non-hypertensive CKD (nonHTN-CKD).

1 mL saliva was collected into an RNA-free sterile container after the participants refrained from eating, drinking, smoking, and brushing at least 1 h before saliva collection. The samples were added to 500 μL lysis buffer and stored at −80 ℃ until DNA isolation. The procedures of DNA extraction, 16S rRNA quantitative PCR, qualitative 16S targeted metagenomic sequencing, and bioinformatic analysis are described in our previous study [21]. In addition, a receiver operating characteristic (ROC) curve, was used to illustrate the diagnostic ability of bacterial genus. Linear discriminant analysis effect size (LefSe) was used to find biomarkers between groups. R studio (version 3.6.3) was used to perform the bioinformatics analysis. Functional metagenomes were predicted based on the 16S rRNA sequencing data of the salivary microbiome using PICRUSt2 (phylogenetic investigation of communities by reconstruction of unobserved states) v2.0.3 (https://​github.​com/​picrust/​picrust2) to predict functional gene abundance with Kyoto Encyclopedia of Genes and Genomes (KEGG) orthologs.

Descriptive statistics for demographics and clinical and immunological parameters of groups are presented. To compare demographics and clinical characteristics between the groups, continuous variables were assessed using independent t-tests and categorical variables using Chi-square or Fisher’s exact tests, when appropriate. One-way analysis of variance was employed to compare quantitative variables among the number of groups  ≥ 3. The Wilcoxon rank-sum test/Kruskal–Wallis one-way test was applied to compare bacterial diversity and the relative abundance of bacterial taxa (≥ 1% of total abundance) between/among groups, and a Benjamini Hochberg false discovery rate (FDR)-corrected P-value was calculated for comparative tests. A P < 0.05 was used as a cut-off for comparative statistical tests. SPSS (v. 24.0) was used to compare the demographics between/among groups.

A total of 200 salivary samples were analyzed, including 100 from CKD patients and 100 from HC (Table 1). The participants in the groups of CKD and HC were age-gender-BMI matched. In the CKD group, 29, 64 and 58 patients were with co-occurring type 2 diabetes mellitus (T2DM), hypertension, and detectable ASO respectively (Additional file 1: Table S1, Additional file 2: Table S2 and Additional file 3: Table S3). Among the 29 T2DM patients, 25 of them had hypertension, and 18 had detectable ASO; among the 64 hypertension patients, 25 of them had T2DM, and 35 of them had detectable ASO; among the 58 patients with detectable ASO, 18 of them had T2DM, and 35 had hypertension. As expected, the CKD group had a declined eGFR, and increased levels of serum urea, serum creatinine, serum uric acid, urine protein, urine creatinine, and blood pressure (P < 0.05). Also, the patient group had increased levels of serum C-reactive protein, κ FLC, and λ, FLC (P < 0.05), and decreased levels of C3 and C4. Among the CKD patients, ASOPOS subjects accounted for 58% (Additional file 3: Table S3).

After excluding samples with invisible bands in PCR, we characterized the bacterial composition of salivary samples from 198 subjects, including 98 diagnosed with CKD and 100 healthy subjects. A total of 16,504,359 raw reads were identified, after removing low-quality or ambiguous reads 14,869,458 valid reads have remained. After rarifying to 15,193 reads, we obtained 10,388 ASVs. Good’s coverage index in each sample had a value of 100% measured at the amplicon sequence variant (ASV) level.

PCoA for the significantly differential ASVs indicated that the bacterial community in CKD patients’ saliva was different from that in the HC group (P < 0.05; Fig. 1A). Of 10,388 observed ASVs, 28.26% of them was shared by the two groups (Fig. 1B). However, when the bacterial richness and diversity were compared, the index of Chao 1, Shannon, and Simpson did not show a difference between the groups of CKD and HC (P > 0.05; Fig. 1C).

As Fig. 1D shown, the most abundant bacterial phylum in both CKD and HC groups were Firmicutes (44.25% in CKD and 43.38% in HC) and its genus Streptococcus (30.50% in CKD and 28.45% in HC). Firmicutes was followed by Proteobacteria (26.61% in CKD and 22.40% in HC), and its bacterial genus such as Neisseria was accounted as the second most abundant bacteria in both CKD and HC (14.00% and 12.94%, respectively). Actinobacteria exhibited a decline in the CKD group (10.77%) compared to HC (13.77%). Also, Rothia, a bacterial genus belonging to Actinobacteria, showed a decreasing trend in the CKD (6.99%) relative to HC (8.37%).

When the bacterial phylum was compared between the groups of CKD and HC. Significantly decreased abundances of Actinobacteria and Bacteroidetes were observed in CKD group, whereas Proteobacteria increased in CKD (P < 0.05; Fig. 2A).

When the Wilcoxon rank-sum test was applied, the bacterial genus of Actinomyces, Prevotella, Provotella 7, and Trichococcus significantly decreased in CKD than those in HC, while Lautropia and Pseudomonas increased in CKD (P < 0.05; Fig. 2B). ROC measured by AUC showed that Pseudomonas had an AUC value of 0.804 (Fig. 2C). LefSe showed that Proteobacteria and its genus Pseudomonas can be classified as CKD biomarkers, while Prevotella, Prevotella 7, and Trichococcus can be considered healthy biomarkers in saliva (Fig. 2D).

In order to determine the functional implication of microbial composition in CKD, we used PCIRUSt2 to infer microbial gene content from 16S rRNA gene data and aggregated relative abundance of functional genes into metabolic pathways. CKD salivary samples exhibited significantly higher levels of amino acid metabolism, lipid metabolism, xenobiotics biodegradation and metabolism, etc., whereas had lower levels of metabolism of cofactors and vitamins, glycan biosynthesis and metabolism, cardiovascular diseases, etc.(P < 0.05; Fig. 3).

When we separated the CKD patients into subgroups of DM-CKD and nonDM-CKD, both of them displayed significantly different bacterial communities from that in HC (P < 0.05; Fig. 4A), whereas no difference between DM-CKD and nonDM-CKD (P > 0.05; Fig. 4A). The bacterial richness and diversity, including indices of Chao 1, Shannon, and Simpson did not show significant differences among the groups of DM-CKD, nonDM-CKD, and HC (P > 0.05; Fig. 4B). The top 5 abundant bacterial genera were Streptococcus, Neisseria, Rothia, Gemella, and Haemophilus in the DM-CKD group; and the most abundant bacterial genera in the nonDM-CKD group were the same as the DM-CKD group; Streptococcus, Neisseria, Rothia and Haemophilus were also accounted for the most abundant bacteria in the HC group, and Prevotella 7 was stand out in the HC (Fig. 4D). Surprisingly, Lautropia and Pseudomonas only showed a decline in group HC compared to the nonDM-CKD group instead of the DM-CKD group (P < 0.05; Fig. 4D). Similarly, Prevotella 7 only increased in the HC group with respect to the nonDM-CKD instead of the DM-CKD group (P < 0.05; Fig. 4D).

Similar to the alteration trend in DM-CKD, both the groups of HTN-CKD and none-CKD displayed a significant difference in the bacterial community from the group of HC (P < 0.05; Fig. 5A), while the group of HNT-CKD did not differ from nonHTN-CKD (P > 0.05; Fig. 5A). Also, the bacterial richness and diversity were not different among the groups of HTN-CKD, nonHTN-CKD, and HC (P > 0.05; Fig. 5B). Figure 5C displayed that the sequence of the most five abundant bacterial genera in the group of HTN-CKD was Streptococcus, Neisseria, Rothia, Gemella, and Haemophilus, and the most abundant bacterial genus in the nonHTN-CKD group was ranked as Streptococcus, Neisseria, Rothia, Haemophilus and Gemella. When the bacterial genus was compared, both HTN-CKD and the nonHTN-CKD group had a significantly lower level of Prevotella 7 than group HC. In contrast, both of them had a higher level of Pseudomonas than group HC (P < 0.05; Fig. 5D).

To seek the association between bacterial microbiome and immunity, we performed Pearson correlation analysis between the bacterial genus showing significantly increased or decreased in CKD and the parameters of immunological indicators in CKD and noticed that Pseudomonas was significantly correlated to the level of serum IgG (P < 0.05; Fig. 6A). When the CKD patients were separated into subgroups of ASOPOS-CKD and ASONEG-CKD, both of them showed a different bacterial community from HC, whereas no difference was observed between the subgroups of ASOPOS-CKD and ASONEG-CKD (P < 0.05; Fig. 6B). ASONEG-CKD patients had a significantly higher level of bacterial richness of Chao 1 compared to ASOPOS-CKD patients and HC (P < 0.05; Fig. 6C); both Prevotella 7 and Trichococcus showed significantly lower levels in groups of ASOPOS-CKD and ASONEG-CKD comparing to HC, while Pseudomonas enriched in ASOPOS-CKD and ASONEG-CKD comparing to HC (P < 0.05; Fig. 6D).