Background
Diabetes mellitus (DM) is one of the most common and fastest growing chronic diseases worldwide and it is estimated that there will be 642 million people with DM in 2040 [
1]. Diabetic nephropathy (DN), one of the microvascular complications of DM, occurs in almost 20–40% of these patients [
2]. DN is characterized clinically with persistent high urinary albumin-to-creatinine ratio over 30 mg/g and/or a continuous decline in estimated glomerular filtration rate to less than 60 ml/min/1.73 m^2 [
3]. Patients with DN generally accompany with increased risk of cardiovascular events and progress to end-stage renal disease, which brings a heavy burden on social and high morbidity [
4‐
6]. Despite the management and therapeutic strategies of DN have been established for decades, the quest for truly effective measures continues.
Gut microbiota (GM) is considered as a pivotal “organ” and participates in health maintenance in our whole life [
7]. In recent years, intervening the gut-kidney axis for renal diseases treatment tends to be a new research spotlight [
8]. Numerous evidences suggested the intestinal flora disorder existed in patients with DN and contributed to DN progression, while probiotics improved DN [
9‐
13]. Imbalance of GM involves in DN progression via GM-derived metabolites, which mainly consist of short-chain fatty acids (SCFAs), bile acids, tryptophan and uremic toxins [
14]. Li et al. found the anti-inflammation and anti-fibrosis capability of SCFAs binding with G protein-coupled receptors (GPR)43 or GPR109A in DN mice [
15]. Enrichment of SCFAs-producing bacteria may protect against DN. However, another study showed dysbiosis of GM-regulated GPR43 activation aggravated albuminuria in DN through podocyte insulin resistance [
16]. These contradictory evidences make confusion and the quality of the evidence of traditional epidemiological studies is concerned due to the limitation of confounding factors or reverse causality. Whether there is a causal relationship between GM and DN is still unclear. Accordingly, it is necessary to figure out the causal connection between GM and DN at the genetic level.
Mendelian randomization (MR) analysis is a study method to explore the causal effect of exposures and outcomes by using genetic variations as instrumental variables (IVs) [
17]. Naturally, genetic variation is inherited randomly and the DNA phenotype is dependent on parent, thus the causality between exposures and outcomes cannot be influenced by multifarious confounding factors [
18]. Similarly, the outcome does not change the intrinsic genetic variations and thus avoids reverse causation. Compared to the traditional observational studies, MR analysis provides a possibility to study the causal effect between exposures and outcomes with mitigating the bias from confounding factors and reverse causation [
19].
To identify the link between GM and DN risk, we conducted a two-sample MR method using genome-wide association study (GWAS) summary data. The findings of the study might provide new insights into the mechanism of GM on DN, detection and the potential therapeutic target.
Discussion
High prevalence and morbidity of DN drives us to pursuit the pathogenesis and new therapeutic targets for patients with DN. This study marked the inaugural effort to elucidate the causal effect of GM on DN through a series of MR analysis. The current investigation not only suggested that genus Ruminococcus gauvreauii was predominantly and causally related with a reduced risk of DN via gene FCRG2B, but also revealed another 7 bacterial taxa had causal effects on DN. The findings signified the essential contribution of GM to the progression of DN and provided a genetic-level reference for further research.
Previously, a great deal of studies investigated the interaction between gut microbiota and several diseases, like diabetes mellitus, autoimmune disease, chronic kidney disease, etc [
43‐
45]. Alteration of gut microbiome attributed to curative effect and clinical outcomes [
45,
46]. Similarly, numerous studies explored the adverse effect of GM dysbiosis on DN and the beta diversity of GM in DN differed from healthy control [
11,
47,
48]. In our study, order Bacteroidales, genus Akkermansia, Coprococcus 1, Marvinbryantia and Parasutterella were found to be risk factors for DN while the others were the protective factors. A recent meta-analysis summarized the changes of abundance of gut microbiota based on current reported studies and found that the abundance of genus Akkermansia increased in patients with DN while genus Coprococcus varied from distinct studies [
11]. Augmented abundance of genus Akkermansia might be a risk factor of DN which was consistent with our study. However, the actual mechanism of genus Akkermansia affecting DN was still uncovered. Genus Akkermansia was reported to negatively modulate glucose metabolism via interferon-γ and improved insulin sensitivity [
49,
50]. Restoral of abundance of genus Akkermansia might prevent DN development via SCFAs producing [
48]. Conversely, genus Akkmeransia played a crucial role in gut-immune axis and promoted M1 macrophages polarization which secreted excessive inflammatory factors and aggravated kidney injury [
51]. Besides, the abundance of genus Akkmeransia was positively correlated with renal failure biomarkers and increased along with CKD progression, which indicated that genus Akkmeranisa participated in uremic toxin production and led to kidney dysfunction [
52‐
54]. More extensive investigations were warranted to characterize and offer a comprehensive understanding of the contentious effect of genus Akkmeransia on DN. Besides, genus Parasutterella was positively related with inflammatory cytokines like lipopolysaccharide and interleukin (IL)-8 but negatively with IL-10 in type 2 diabetes mellitus (T2DM) mice, while probiotic treatment maintained barrier integrity and ameliorated inflammation through decreasing enrichment of genus Parasutterella [
55].
There was no traditional epidemiological study on the relationship of other significant bacterial taxa found in our study and DN. Gut metabolites, insulin resistance, local RAS activation, inflammation and mucosal immunity disorder accounting for GM dysbiosis were the primary mechanisms to DN [
14]. Acetate, butyrate and propionate, which were generated by anaerobic fermentation of dietary fibers, were main components of SCFAs. It was known that Bacteroidetes and Firmicutes stood as the predominant phyla in human intestinal microbiota, with Bacteroidetes members were primary source of acetate and propionate, whereas Firmicutes were mainly engaged in butyrate synthesis [
56,
57]. Acetate played a pivotal role in dysregulation of cholesterol homeostasis through the activation of GPR43, and consequently contributed to tubulointerstitial injury in DN [
58]. This might be the potential mechanism of order Bacteroidales affecting DN. Regarding to butyrate, several molecular pathways took part in DN protection. (i) Butyrate was speculated to improve insulin resistance via GPR43. SCFAs could bind to GPCRs like GPR43 and involved in GPCRs mediated signaling pathways [
59]. A previous study demonstrated dysregulation of GPR43 modulated by dysbiosis of GM resulted in podocyte insulin resistance and glomerular injury in DN while another research discovered butyrate reversed insulin resistance through GPR43 mediated suppression of oxidative stress and NF-κB signaling in mice model, which prevented mesangial matrix deposition and renal fibrosis [
16,
60]. (ii) Dysbiosis-induced acetate overproduction was implicated in the renal damage observed in early DN through the activation of intrarenal RAS [
10]. Lei W and colleagues demonstrated that sodium butyrate ameliorated angiotensin II-induced kidney injury via inhibition of renal (pro)renin receptor and intrarenal RAS [
61].
Apart from the previous elaboration of protective effects of butyrate on DN, a few works explored the role of butyrate in immunity regulation [
62‐
64]. Man Y, et al. reported that sodium butyrate was capable to alleviate vacuolar degeneration of renal tubules and tubular epithelial cells exfoliation, via attenuating inflammation activation mediated by PI3K/Akt/NF-κB pathway in high glucose induced human monocyte-macrophages [
65,
66]. Butyrate could regulate Treg/Th17 equilibrium by promoting regulatory T cell differentiation while inhibiting Th17 helper T cell [
63]. Kathrin Eller, et al. claimed the role of CD4(+)Foxp3(+) Tregs in improving insulin sensitivity and diabetic nephropathy [
67]. Oppositely, Th17 cell mediated inflammation and produced IL-17, a proinflammatory cytokine, and hence increased the risk of T1DM due to islet inflammation and β-cells destruction [
68,
69]. Thereby, butyrate might engage in prevention from DN by adjusting Treg/Th17 ratio. Both genus Eubacterium ventriosum, Ruminococcus gauvreauii and Erysipelotrichaceae (UCG003) belongs to Firmicutes phylum and all these bacterial taxa mainly produce butyrate. Based on the aforementioned protective effects of butyrate on DN and the results in MR analysis, we hypothesized these taxa might played an essential role in DN improvement mediated by butyrate.
Although genus Coprococcus 1 and Marvinbryantia also produced butyrate, the current studies on interaction among genus Coprococcus 1, diabetes and kidney disease were discrepant and whether genus Coprococcus 1 took part in DN development was still unknown. Further study should be done to figure out the potential mechanism. Additionally, some researchers showed that genus Marvinbryantia was positively correlated with inflammatory cytokines and might induce inflammatory response in high glucose environment [
70,
71]. According to our study, we did transcriptome MR analysis and found deleterious effect of VNN2 on DN. VNN 2 was detected high expression in kidney tissue [
72]. The protein Gpi80, encoded by the VNN2, exhibited surface aggregation on activated migrating neutrophils and potentially modulated neutrophil adhesion and migration [
73]. Thus, we implied that genus Marvinbryantia involved in pathological exacerbation of DN by inflammation activation.
Several GM related genes had been found to have causal relationship with DN in this work. Among these potential genes, FCGR2B relevant to ‘true factor’ genus Ruminococcus gauvreauii might reduce risk of DN. FcγRIIB, a unique inhibitory Fcγ receptor known to be expressed on various immune cells like B cells, macrophages and granulocytes, is the product of gene FCGR2B [
74]. FcγRIIB has been widely studied in autoimmune-mediated kidney diseases [
75‐
77]. Acute kidney injury aggravated lupus activity through spleen tyrosine kinase (Syk)/neutrophil extracellular traps pathways in FcγRIIB deficient mice [
75]. A recent study on lupus nephritis revealed that FcγRIIB conducted inhibitory effect on IL-1β production, which was elevated in several nephritis, in kidney macrophages through Syk signaling pathways [
77]. Furthermore, FcγRIIB limited adaptive immunity by inducing CD8 + T cell apoptosis while suppressing CD8 + T cell response mitigated renal injury and fibrosis [
78,
79]. Therefore, we inferred the influence of genus Ruminococcus gauvreauii on DN through FCGR2B. Regardless of other genes found causation with DN in MR analysis had not been reported to be related to DN, it might provide new insights for future investigations into the mechanism underlying the interaction between GM and DN.
Although mature management and treatment on DN have been established in these years like blood glucose monitoring and application of RAS inhibitors or sodium-glucose cotransporter 2 inhibitors, a more effective strategy should be explored to stop DN progression [
80]. Based on the results of our study, maintaining intestinal microenvironment and promoting the dominance of butyrate-producing bacteria in gut may be a potential approach for DN amelioration. A few clinical trials have shown oral intake of probiotic supplementation conduces to renal function in patients with diabetes [
81,
82], while fecal microbiota transplant is also an effective way [
83]. To our knowledge, dietary education is another efficient method for patients with DN. Supplementation with SCFAs or adoption of a high-fiber diet or could mitigate renal inflammation [
15,
60]. Currently, a few novel drugs targeting modulation of intestinal mucosal immunity showed a distinguished effect on glomerular disease treatment, which gave us prospects in the development of drugs targeting the gut-kidney axis for kidney disease treatment [
84]. Qi et al. reported that microRNA-16 had the capacity to inhibit mesangial cells proliferation via toll-like receptor 4 signaling pathway in FCGR2B deficient mice [
85]. Therefore, we conceive targeting the FCGR2B related signaling pathway, such as microRNA-16, might be a prospective therapy for DN and the future study is worthwhile to explore.
Our study owns a plenty of advantages. While a significant number of investigations have reported correlations and demonstrated variations in bacterial abundances in different cohorts, however, the specific reasons for these alterations and their causal relevance often remain uncertain. By using genetic variants, MR-Egger intercept and MR-PRESSO method, we prevented confounding factors and reverse causal effect to provide a robust causation between GM and DN. Besides, instrumental variations of GM were obtained from the latest comprehensive GWAS summary data which ensured the convincingness of MR results. We also performed a brand new-proposed multivariable MR methods (MR-BMA) to identify the most influential bacterial taxa on DN. Eventually, a PPI network was applied to understand the connection among GM related protein, while transcriptome MR analysis based on mapped genes supplied several potential biomarkers and therapeutic targets on DN.
Notably, this study also has some limitations: (i) The GWAS summary data from MibioGen only classified from phylum to genus and the causal effect of specific species belonged to each genus on DN could not be analyzed. (ii) The involved SNPs from MiBioGen and FinnGen dataset were derived from various cohort studies involved European. Thus, generalizing the MR results of this research to other racial populations might be not viable. (iii) The confounders recognized by Phenoscanner were removed according to the current studies and our clinical experience which indicated the potential bias. (iv) There were only 3283 DN cases in FinnGen cohort and such small sample size lacked great beta effect contributing to less statistical power. Nevertheless, this work was still worthy and provided an initial study using large scale genetic data to explore the correlation of GM and DN. Future study to include larger sample size from different database was expected. Eventually, we carried out a series of methods to validate the robustness of MR results and thus we thought our work was very valuable.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.