DOI: https://doi.org/10.21203/rs.3.rs-1826254/v1
Transplant rejection and failure are the primary causes of short life in transplant patients, and the mechanism is yet unknown. A number of recent research findings point to a possible association between the gut microbiome and transplantation failure. However, it is unclear if part of the gut microbiota is the cause of transplantation failure.
A Mendelian randomization study was carried out to test the potential causal relationship between gut microbiota and transplantation failure. Three GWAS results were used, one for the gut microbiome, one for transplantation failure, and one for transplantation status. As instrumental variables, SNPs with a strong correlation to the abundance of gut microbiota were chosen.
The abundance of Bifidobacteriaceae was discovered to be a causal factor for transplantation failure, with a protective effect (IVW p = 0.049, OR = 0.658, 95% CI: 0.433–0.998) after Mendelian randomization analysis. Furthermore, there was no relationship between Bifidobacteriaceae and transplantation status. Gene enrichment analysis revealed that the genes containing the instrumental variables of Bifidobacteriaceae were primarily enriched in synapse and membrane related terms.
These findings suggest that a decrease in the abundance of Bifidobacteriaceae in the gut may increase the risk of transplantation failure. This work contribute novel insights for further explaining the process of transplantation failure, and it is important in intervening and avoiding transplantation failure.
Transplantation of solid organs or tissues is an effective treatment for end-stage organ failure or burns. To avoid immune-mediated transplant rejection, most transplant recipients require lifelong immunosuppressive therapy. Some patients, however, may still experience transplant rejection or even transplant failure. There are numerous reasons for transplantation failure, including genetic factors, infections, complications, and so on (1–3). There is currently no effective way to completely avoid transplant rejection and failure, and the potential mechanisms require further investigation.
The intestine is the largest immune organ in the human body, and the microbiota that live there play critical roles in the regulation of human immune response(4). For example, Bacteroides fragilis promotes the development of regulatory T cells (Tregs) (5), and clusters IV and XIVa of Clostridium contribute to Treg accumulation (6). Faecalibacterium prausnitzii facilitates the augmentation of Treg in peripheral blood cells and splenocytes (7), and Segmented filamentous bacterium induces the appearance of CD4(+) T helper cells in the lamina propria (8). In addition, Alistipes has been shown to improve skin graft survival in mice (9). The relationship between gut microbiota and transplantation failure and rejection is currently unclear.
Causal analysis is an analytical method that examines the causal relationship between variables directly, whereas association analysis examines only the correlation between variables. Understanding the causal relationship between variables is critical for determining the correlation mechanism between them. As a result, causal analysis is increasingly being used in biomedical research presently. Mendelian randomization (MR) is a common approach for causal inference that employs genetic variation to answer questions about how modifiable exposures influence outcomes (10). Before performing MR analysis, genetic variations with a strong correlation to candidate exposures were screened as instrumental variables (IV). The study population of outcomes is then divided into exposed and non-exposed groups based on IV genotypes. In order to establish the causality relationship between the exposures and outcomes, the outcomes for the two groups are finally compared. If the outcome of the two groups differ and the contribution of IV to the exposure is greater than that to the outcome, it is inferred that the exposure is the cause of the outcome.
To better elucidate the mechanism of transplantation failure and explore the potential causal relationship between gut microbiome and transplantation failure, this MR study was carried out. The abundance of gut bacteria such as Bifidobacteriaceae were discovered to be causative variables for transplantation failure, indicating a possible relation between the gut microbiome and transplantation failure and rejection.
Three cohorts were employed in this study. Cohort 1 has 18473 people of European, Hispanic, Middle Eastern, Asian, and African ancestry from 24 sub-cohorts (11). Whole genome genotyping microarray and 16S rDNA sequencing were used to collect the genotypes of all participants' whole genomes and gut microbiome composition. More than 5,720,000 SNPs (after imputation) and 211 taxa were identified. Cohort 2 is a Finnish population cohort with 124 cases of failure and rejection of transplanted organs and tissues (FRTOT) and 199271 healthy controls. Case diagnosis criteria are based on ICD10 T86, which includes individuals with various kinds of FRTOT. Cohort 2 was collected from the FINNGEN biobank (https://www.finngen.fi/en), and a total of 16,380,395 SNPs were genotyped using whole genome genotyping microarrays (after imputation). Cohort 3 was also acquired from the FINNGEN biobank, and it included 1,449 tissue and organ recipients and 195,047 healthy controls. A total of 16,380,406 SNPs (after imputation) were genotyped in cohort 3. The MRCIEU database (https://gwas.mrcieu.ac.uk/) was utilized to query the information of cohort 2 and 3 (12, 13). All of the preceding investigations were carried out in conformity with the ethical standards outlined in the Helsinki Declaration and were approved by the institutional review board.
The abundance of gut bacteria was classified as an exposure in this study, while FRTOT was designated as an outcome. Before performing MR analysis, the IV in cohort 1 were screened. IV screening criteria include: a strong association with exposure; independence from confounding factors that impact exposure and outcome; and the absence of pleiotropy effects. To achieve the best results, we selected the IV based on five p value levels (P < 10− 4, P < 10− 5, P < 10− 6, P < 10− 7 and P < 10− 8). The MR analysis was carried out on Cohorts 2 and 3. If the number of IV was greater than one for each taxon, the MR-Egger, Weighted Median, Inverse Variance Weighted (IVW), Weighted Mode, and Simple Mode methods were used. Otherwise, the Wald ratio model was used. Sensitivity analysis was also conducted to assess the heterogeneity and pleiotropy effects of the selected IV. In addition, for each taxon, we determined the F value of the selected IV. The IV had a strong association with taxa was characterized as F > 10. The following formula is used to calculate F: \(F=\frac{(\text{n}-\text{k}-1)/\text{k}}{{r}^{2}/(1-{r}^{2})}\). Where n represents the sample size, k represents the number of IV, and r2 represents the variance of exposure explained by IV. The MR analysis was carried out using the R package TwoSampleMR (version 0.5.6). Gene enrichment analysis was performed using David (https://david.ncifcrf.gov/home.jsp).
At P < 10− 4 and P < 10− 5 levels, all 211 taxa successfully screened out the IV, however at P < 10− 6, P < 10− 7 and P < 10− 8 levels, only 157, 30 and 9 acquired IV respectively. Most of taxa only screened out one IV, at P < 10− 7 and P < 10− 8 levels.
At P < 10− 4 level, 17 significantly causative associations between taxa and FRTOT were identified, while at P < 10− 5, P < 10− 6, P < 10− 7 and P < 10− 8 levels, the numbers were 16, 4, 1 and 0 respectively.
At P < 10− 4, P < 10− 6 and P < 10− 7 levels, the abundance of the genus Oxalobacter appeared to have a positive causal effect on FRTOT. Unknowngenus (id.1000001215), unknownfamily (id.1000001214), order Gastranaerophilales and class Melainabacteria were discovered to be the risk factor for FRTOT, whereas family Bifidobacteriaceae and order Bifidobacteriales were revealed to be protective factors for FRTOT at both the P < 10− 4 and P < 10− 5 levels. In addition, another 25 taxa including Anaerofilum, Bifidobacterium and Butyricicoccus were shown to be causally associated to FRTOT at least at one P value level. Table 1 displays all of the MR analysis results.
Biological classification | Name | Nsnps | Level | Model | Β | SE | P val |
---|---|---|---|---|---|---|---|
Phylum | Euryarchaeota | 89 | P < 10− 4 | MR Egger | -1.274 | 0.588 | 0.033 |
Class | Clostridia | 108 | P < 10− 4 | MR Egger | 1.748 | 0.851 | 0.042 |
Melainabacteria | 78 | P < 10− 4 | MR Egger | 1.681 | 0.646 | 0.011 | |
10 | P < 10− 5 | MR Egger | 3.739 | 1.420 | 0.030 | ||
Order | Bifidobacteriales | 124 | P < 10− 4 | IVW | -0.419 | 0.213 | 0.049 |
15 | P < 10− 5 | IVW | -1.436 | 0.612 | 0.019 | ||
Gastranaerophilales | 77 | P < 10− 4 | MR Egger | 1.470 | 0.678 | 0.033 | |
9 | P < 10− 5 | MR Egger | 4.672 | 1.467 | 0.015 | ||
Family | Bacteroidales | 8 | P < 10− 5 | IVW | 1.289 | 0.616 | 0.036 |
Bifidobacteriaceae | 124 | P < 10− 4 | IVW | -0.419 | 0.213 | 0.049 | |
15 | P < 10− 5 | IVW | -1.436 | 0.612 | 0.019 | ||
Oxalobacteraceae | 4 | P < 10− 6 | Weighted median | 1.842 | 0.789 | 0.020 | |
Peptococcaceae | 92 | P < 10− 4 | IVW | 0.457 | 0.216 | 0.035 | |
Unknownfamily (ID:1000001214) | 77 | P < 10− 4 | MR Egger | 1.470 | 0.678 | 0.033 | |
9 | P < 10− 5 | MR Egger | 4.672 | 1.467 | 0.015 |
Analysis of heterogeneity revealed that all of Gastranaerophilales, unknownfamily (ID:1000001214) and unknowngenus (ID:1000001215) have heterogeneity (Table 2). Clostridia, Melainabacteria, unknownfamily (id.1000001214), Oxalobacter, Dialister, LachnospiraceaeNC2004group, unknowngenus (id.1000001215), unknowngenus (id.1868), Gastranaerophilales and Euryarchaeota displayed pleiotropic effects, according to pleiotropy analyses. After excluding taxa that did not match the MR analysis requirements, only Bifidobacteriaceae and Bifidobacteriales shown to be significantly causative associations with FRTOT at more than one P value levels. We further compared the IV of these two taxa, and found that they were identical. As a result, we concentrated on Bifidobacteriaceae and its subordinate genus in the subsequent analysis.
Biological classification | Name | Level | Q pval* | Egger intercept | SE | P val# |
---|---|---|---|---|---|---|
Phylum | Euryarchaeota | P < 10− 4 | 0.888 | 0.157 | 0.071 | 0.029 |
Class | Clostridia | P < 10− 4 | 0.770 | -0.118 | 0.051 | 0.023 |
Melainabacteria | P < 10− 4 | 0.322 | -0.176 | 0.065 | 0.008 | |
P < 10− 5 | 0.111 | -0.479 | 0.156 | 0.015 | ||
Order | Bifidobacteriales | P < 10− 4 | 0.851 | 0.023 | 0.037 | 0.542 |
P < 10− 5 | 0.213 | -0.108 | 0.158 | 0.505 | ||
Gastranaerophilales | P < 10− 4 | 0.245 | -0.148 | 0.068 | 0.033 | |
P < 10− 5 | 0.021 | -0.639 | 0.167 | 0.007 | ||
Family | Bacteroidales | P < 10− 5 | 0.310 | -0.228 | 0.242 | 0.382 |
Bifidobacteriaceae | P < 10− 4 | 0.851 | 0.023 | 0.037 | 0.542 | |
P < 10− 5 | 0.213 | -0.108 | 0.158 | 0.505 | ||
Oxalobacteraceae | P < 10− 6 | 0.227 | 0.382 | 1.094 | 0.760 | |
Peptococcaceae | P < 10− 4 | 0.939 | -0.037 | 0.046 | 0.424 | |
Unknownfamily (ID:1000001214) | P < 10− 4 | 0.245 | -0.148 | 0.068 | 0.033 | |
P < 10− 5 | 0.021 | -0.639 | 0.167 | 0.007 | ||
Genus | Anaerofilum | P < 10− 4 | 0.798 | 0.039 | 0.073 | 0.596 |
Bifidobacterium | P < 10− 4 | 0.297 | 0.068 | 0.043 | 0.112 | |
Bilophila | P < 10− 5 | 0.534 | 0.266 | 0.225 | 0.262 | |
Butyricicoccus | P < 10− 4 | 0.966 | 0.023 | 0.046 | 0.615 | |
DefluviitaleaceaeUCG011 | P < 10− 5 | 0.817 | -0.026 | 0.218 | 0.909 | |
Dialister | P < 10− 5 | 0.341 | -0.524 | 0.192 | 0.023 | |
ErysipelotrichaceaeUCG003 | P < 10− 4 | 0.473 | 0.073 | 0.052 | 0.167 | |
Eubacteriumfissicatena | P < 10− 5 | 0.847 | 0.378 | 0.280 | 0.219 | |
Eubacteriumnodatum | P < 10− 6 | 0.903 | NA | NA | NA | |
Intestinibacter | P < 10− 5 | 0.947 | -0.102 | 0.139 | 0.474 | |
LachnospiraceaeNC2004 | P < 10− 5 | 0.172 | -0.648 | 0.237 | 0.029 | |
LachnospiraceaeUCG008 | P < 10− 4 | 0.635 | 0.035 | 0.066 | 0.596 | |
Olsenella | P < 10− 4 | 0.685 | -0.105 | 0.071 | 0.145 | |
Oscillibacter | P < 10− 5 | 0.782 | 0.010 | 0.174 | 0.953 | |
Oxalobacter | P < 10− 4 | 0.510 | -0.176 | 0.066 | 0.009 | |
P < 10− 6 | 0.220 | 0.161 | 0.801 | 0.860 | ||
P < 10− 7 | 0.829 | NA | NA | NA | ||
Ruminococcus2 | P < 10− 5 | 0.486 | 0.000 | 0.105 | 0.997 | |
RuminococcaceaeNK4A214 | P < 10− 6 | 0.349 | -0.188 | 0.743 | 0.843 | |
Unknowngenus (ID:1000001215) | P < 10− 4 | 0.245 | -0.148 | 0.068 | 0.033 | |
P < 10− 5 | 0.021 | -0.639 | 0.167 | 0.007 | ||
Unknowngenus (ID:1000005479) | P < 10− 5 | 0.310 | -0.228 | 0.242 | 0.382 | |
Unknowngenus (ID:1868) | P < 10− 4 | 0.880 | 0.148 | 0.063 | 0.021 | |
Note: *Calculate by IVW method; # Calculate by MR egger method; |
Based on IVW method, the abundance of Bifidobacteriaceae was identified to be a protective factor for FRTOT (p = 0.049, OR = 0.658, 95%CI: 0.433–0.998) at P < 10− 4 level. The results of Single SNP analysis revealed that the MR effect size and 95% confidence interval of FRTOT of rs7174549, rs13020688 and rs79593173 was less than 0 (Fig. 1A). Leave one out analysis found that none of IV had a great impact on the MR model, while the majority of IV were negatively related with FRTOT (Fig. 1B). According to the scatter diagram, the abundance of Bifidobacteriaceae had a negative causal effect on FRTOT (Fig. 1C). The funnel plot revealed that the IV of Bifidobacteriaceae exhibited no obvious asymmetry, indicating that no heterogeneity existed (Fig. 1D). The F value of the 124 IV of Bifidobacteriaceae was 23.787, indicating that the selected IV had a strong association with the abundance of Bifidobacteriaceae. Similar findings were obtained at the P < 10− 5 level. The abundance of Bifidobacteriaceae had a negative causal effect on FRTOT (IVW p = 0.019, OR = 0.238, 95%CI: 0.072–0.790), and the F value for 15 selected IV was 26.841. At P < 10− 4 level, Bifidobacterium, a main genus of Bifidobacteriaceae was also identified as a protective factor for FRTOT (MR Egger p = 0.033, OR = 0.252, 95%CI: 0.072–0.881, Figure S2). The F value of the116 IV of Bifidobacterium was 23.022.
The causal relationship between Bifidobacteria and the risk of organ or tissue transplantation was also analyzed, and no causal effect of Bifidobacteriaceae and Bifidobacterium was discovered (Table S1), implying that the abundance of gut Bifidobacteria was unrelated to the risk of organ or tissue transplantation.
The IV of Bifidobacteriaceae are located in 91 genes, which are enriched in synapse and membrane related terms, according to gene set enrichment analysis (Table 3). The top 10 enriched terms include three terms related to synapses, such as glutamatergic synapse, modulation of synaptic transmission, and syntaxin binding, and five terms related to membrane, such as endoplasmic reticulum membrane, integral component of membrane, transmembrane transport, sodium ion transmembrane transport, and calcium ion transmembrane transport.
Category | Term | Count | Percentage | P value |
---|---|---|---|---|
GOTERM_CC_DIRECT | GO:0098978 ~ glutamatergic synapse | 6 | 6.593 | 0.008 |
GOTERM_BP_DIRECT | GO:0050804 ~ modulation of synaptic transmission | 3 | 3.297 | 0.017 |
GOTERM_MF_DIRECT | GO:0019905 ~ syntaxin binding | 3 | 3.297 | 0.018 |
GOTERM_CC_DIRECT | GO:0005789 ~ endoplasmic reticulum membrane | 9 | 9.890 | 0.021 |
GOTERM_CC_DIRECT | GO:0016021 ~ integral component of membrane | 27 | 29.670 | 0.025 |
GOTERM_MF_DIRECT | GO:0008013 ~ beta-catenin binding | 3 | 3.297 | 0.033 |
KEGG_PATHWAY | hsa01100:Metabolic pathways | 11 | 12.088 | 0.035 |
GOTERM_BP_DIRECT | GO:0055085 ~ transmembrane transport | 4 | 4.396 | 0.036 |
GOTERM_BP_DIRECT | GO:0035725 ~ sodium ion transmembrane transport | 3 | 3.297 | 0.039 |
GOTERM_BP_DIRECT | GO:0070588 ~ calcium ion transmembrane transport | 3 | 3.297 | 0.052 |
Note: Count represents genes involved in the term; Percentage represents the ratio of involved genes to total genes of the term. |
To the best of our knowledge, this is the first study to investigate the causal relationship between gut microbiome and FRTOT. The abundance of Bifidobacteriaceae was discovered to be a protective factor for FRTOT. Gene enrichment analysis revealed that genes harboring SNPs that associated with Bifidobacteriaceae abundance were primarily enriched in synapse and membrane related terms.
Bifidobacteriaceae is a common probiotic in the human gut that plays important roles in intestinal homeostasis and inflammation (14, 15). Previous research found that Bifidobacteriaceae improved Treg suppressive activity by activating the IL-10/IL10Ra signaling loop (16). Tregs are a subgroup of T cells that secrete inhibitory cytokines to reduce immunological response. Treg activation improves T cell suppression and lowers the likelihood of transplant rejection (17). Furthermore, CXCL2 activation has been identified as one of the primary causes of lung transplantation failure (18). Short-chain fatty acids (SCFAs) can influence CXCL2 expression, and SCFA treatment inhibits TNF-induced CXCL2 production in BALB/c mice (19). Bifidobacterium is a major genus of Bifidobacteriaceae that generates SCFAs (mainly acetate and lactic acid) during carbohydrate fermentation (14, 20, 21). As a result, by suppressing CXCL2 activation, Bifidobacterium may decrease the likelihood of transplantation failure. In general, a reduction in the abundance of gut Bifidobacterium may raise the likelihood of transplantation failure. The abundance of Bifidobacteriaceae and Bifidobacterium exhibited a negative causal effect on FRTOT in our investigation, which was consistent with this hypothesis. Because Bifidobacterium is one of the most important probiotic bacteria and has been developed into a therapeutic product, supplementing Bifidobacterium to lower the risk of transplantation failure will be possible in the near future.
The microbiome-gut-brain axis is a bidirectional connection between intestinal bacteria and the brain, and synaptic plasticity is important in this axis(22). Interestingly, we discovered that genes harboring Bifidobacteriaceae abundance associated SNPs were primarily enriched in synapse related terms. This finding suggested that polymorphisms in synapse-related genes might affect brain-to-intestine signaling, and then altering the abundance of some intestinal bacteria. Moreover, aberrant synaptic plasticity has been related to a number of brain diseases(23). As a result, some brain diseases may alter the composition of the gut microbiome via affecting synaptic signaling pathways, therefore influencing the prognosis of organ or tissue transplantation. According to recent studies, patients with Alzheimer's disease have dysbiosis of the gut microbiome, and the abundance of Bifidobacterium is down regulated when compared to healthy controls (24, 25). In the future, studies on how brain changes impact the prognosis of organ or tissue transplantation should also be concerned.
This study still has several shortcomings. First, in the MR analysis, different categories of transplantation failure were combined into one group due to data and sample size limitations. The effect of confounders would increase as different forms of organ or tissue transplant failure may be attributed to various causes. Furthermore, we were unable to use patients without failure or rejection after transplantation as controls in the MR study of FRTOT. In order to exclude the influence of Bifidobacteriaceae on the risk of transplantation, we performed a MR analysis on transplant status, and the findings revealed that Bifidobacteriaceae was not a causal factor for the occurrence of tissue and organ transplantation. This indicated that the abundance of Bifidobacteriaceae and FRTOT had a causal relationship. Second, few research have suggested an association between intestinal bacteria and transplantation failure, and our findings need to be validated by more clinical and mechanism investigations. Nonetheless, our findings add to the body of knowledge on transplantation failure.
In conclusion, using MR approach, the current study investigated the causal relationship between gut microbiota and transplantation failure and discovered that lower abundance of gut Bifidobacteriaceae may be one of the contributions of transplantation failure. Our findings contributed new ideas for further explaining the mechanism of transplantation failure, which was important for intervening and avoiding transplantation failure.
Data availability
The datasets used and/or analyzed during the current investigation are accessible upon reasonable request from the corresponding author.
Acknowledgments
For supplying the data for analysis, we are grateful to the MRCIEU database (https://gwas.mrcieu.ac.uk/), MiBioGen (https://mibiogen.gcc.rug.nl/), and FINNGEN biobank (https://www.finngen.fi/en).
Author Contribution Statement
Xi Li, Ying Wang and Han Yan participated in research design. Han Yan, Gongbin Lan and Wei Zhang participated in the writing of the paper. Han Yan and Ying Wang participated in the performance of the research. Xi Li and Han Yan participated in data analysis.
Funding
This research was supported by Chinese National Science Foundation (No.81803583), Hunan Provincial Natural Science Foundation of China (2021JJ31074, 2019JJ50854), Open Fund Project of Hunan Universities Innovation Platform (18K006), the Project Program of National Clinical Research Center for Geriatric Disorders (Xiangya Hospital, Grant No. 2020LNJJ06), and the National Key Research and Development Program of China (No. 2021YFA1301203).
Ethical Approval
The study was approved by the ethics committee of Institute of Clinical Pharmacology, Central South University.
Competing Interests
The authors declare no conflict of interest.