SEARCH

SEARCH BY CITATION

Keywords:

  • Allograft;
  • intestinal microbiota;
  • monitoring

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

Small bowel transplantation can be a life-preserving procedure for patients with irreversible intestinal failure. Allograft rejection remains a major source of morbidity and mortality and its accurate diagnosis and treatment are critical. In this study, we used pyrosequencing of 16S ribosomal RNA gene tags to compare the composition of the ileal microbiota present during nonrejection, prerejection and active rejection states in small bowel transplant patients. During episodes of rejection, the proportions of phylum Firmicutes (p < 0.001) and the order Lactobacillales (p < 0.01) were significantly decreased, while those of the phylum Proteobacteria, especially the family Enterobacteriaceae, were significantly increased (p < 0.005). Receiver-operating characteristic analysis revealed that relative proportions of several bacterial taxa in ileal effluents and especially Firmicutes, could be used to discriminate between nonrejection and active rejection. In conclusion, the findings obtained during this study suggest that small bowel transplant rejection is associated with changes in the microbial populations in ileal effluents and support microbiota profiling as a potential diagnostic biomarker of rejection. Future studies should investigate if the dysbiosis that we observed is a cause or a consequence of the rejection process.


Abbreviations: 
AR, actively rejecting; NR, nonrejecting; OTUs, operational taxonomic units; PR, prerejecting; qRT-PCR, quantitative real-time PCR; ROC, receiver operating characteristic; SBT, small bowel transplant; TLR, toll-like receptor.

 

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

Small bowel transplantation is a life-preserving treatment for patients with chronic intestinal failure and complications from parenteral nutrition. The small bowel is highly immunogenic and allograft rejection remains a significant factor in determining both graft and patient survival (1). Early detection of rejection is critical to allow rapid treatment and successful reversal of the process (2). Delay in treatment can lead to severe exfoliation and/or patient death (3). The ability to reduce the occurrence of rejection episodes or to diagnose them early would likely improve posttransplant survival, which presently ranges (by age) from 64–89% at 1 year to 33–76% by 5 years (Organ Procurement and Transplantation Network, March 29, 2010). Unfortunately, there is not a reliable noninvasive marker to predict intestinal allograft rejection. Hence, the identification of rapid diagnostic tools is critically needed.

The intestine is unique as a transplanted organ in that it is associated with an extensive microbial population. This microbiota consists of around 108 bacterial cells/mL in the healthy human ileum (4) and recent observations showed that similar bacterial numbers can be found in ileal effluents of transplant patients (5). Although the gut microbiota provides a large range of beneficial functions such as digestion of complex plant polysaccharides and resistance to the establishment of enteric pathogens (6,7), it has also been associated with a spectrum of human diseases, such as inflammatory bowel diseases, autoimmune diseases and allergies (8).

The microbiota in intestinal allograft survival is a logical area of investigation and several observations suggest a relationship between the gut microbiota and graft rejection. First, intestinal allograft rejection bears clinical and pathological overlap with ileal Crohn's disease (9), a chronic inflammation of the gastrointestinal tract that is likely driven by a T-cell response to bacterial antigens (10,11). Second, transplant recipients with mutations in NOD2 have a genetic predisposition to develop rejection and sepsis and the likelihood of allograft rejection in these patients is ≈100 times greater than in patients with wild type allele (9). NOD2 is a pattern recognition receptor found on macrophages, dendritic cells and Paneth cells that sense microbial products (12–14). Defects in this microbial sensor are thought to result in impaired expression of intestinal antimicrobial peptides and other defects in innate immune responses, which could trigger an activation of immune cells through microbes that might contribute to the rejection process (9). Finally, in a murine model of intestinal allograft rejection, graft survival was prolonged in the absence of toll-like receptor 4 (TLR4), suggesting that the gut microbiota associated with a graft may enhance alloimmune responses through this receptor (15).

The characterization of microbial community shifts during transplant rejection could improve our understanding of disease etiology and the identification of predictive bacterial signatures associated with rejection could yield potential diagnostic markers. The aim of this study was therefore to characterize the ileal microbiota in patients with nonrejecting (NR), prerejecting (PR) and actively rejecting (AR) small bowel transplants (SBTs). We have used deep sequencing for the microbiota analysis, which provided a comprehensive insight into the relationship between the microbiota and rejection.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

SBT patient cohort

Thirty-five ileal effluent samples were obtained from 19 SBT recipients (median age 4.3 years, range 0.7–49.8 years) before surgical closure. The samples were selected from a prospectively collected tissue bank at the University of Nebraska Medical Center. All patients were receiving standard doses of tacrolimus and steroids. The majority of the samples were from pediatric patients. Samples were designated as nonrejection (14 samples from 12 patients, median 17 weeks posttransplant) or rejection (12 samples from 10 patients, median 45 weeks posttransplant) based on clinical course (symptoms, ostomy output), endoscopic examination at the time of collection, histologic diagnosis and absence of confounding factors such as concurrent viral enteritis. For 9/10 rejection patients, we had an ostomy effluent sample from within 2 weeks before the rejection episode (designated prerejection), where patients had normal ostomy outputs and no clinical symptoms suggesting rejection. All samples from rejecting patients were collected at the time of endoscopy and biopsy. This means that samples included corresponded to the exact day rejection was diagnosed and therefore collected before any additional immunosuppressive treatments were initiated. Clinical characteristics of the subjects (e.g. calprotectin concentration, ileostomy output and pathology scores) are summarized in Table S1. This study was approved by the Institutional Review Board at UNMC (#417–02) and informed consent was obtained from all patients.

DNA extraction of ileal effluent

DNA was extracted from ileal effluents according to Martínez et al. with a few modifications (16) and is described in detail in the Supplemental Materials and Methods.

Pyrosequencing of 16S rRNA tags and taxonomic analysis

To investigate the bacterial community composition in the ileostomy samples, massive parallel sequencing was performed on PCR amplicons that covered the V1–V3 regions of the 16S rRNA gene as described by Martínez et al. (17). The bacterial populations represented by the sequences were analyzed with two independent methods (17). The Ribosomal Database Project (RDP) Classifier tool was used for taxonomic assignment of sequences to the phylum, order, family and genus level (18). Sequences were also assigned to operational taxonomic units (OTUs) using the RDP pyrosequencing pipeline (http://pyro.cme.msu.edu/). In depth description of methods used for pyrosequencing and taxonomic analysis can be found in the supplementary text.

Quantitative real-time PCR (qRT-PCR) and Shigella spp./Enteroinvasive Escherichia coli(EIEC)-specific PCR

Quantification of total bacteria, Enterobacteriaceae, lactobacilli, E. coli and Bifidobacterium spp. was performed by qRT-PCR using primers described previously (19–22). Please refer to Supplementary Materials and Methods for complete description of qRT-PCR methods (Table S2). A PCR assay based on the invasion plasmid antigen H (ipaH) gene was performed to test if Shigella spp. and EIEC were present in the ileal effluent samples according to methods described by Ahmed et al. (23).

Statistical analysis

One-way nonparametric ANOVA (Kruskal–Wallis) was performed to identify significant differences in the ileal microbiota composition between patient groups. Dunn's multiple post hoc test was performed for pairwise comparisons between the different groups of samples using GraphPad Prism version 4.00. A p-value of <0.05 was considered to be statistically significant. Data are presented as mean ± SEM unless otherwise stated. Pyrosequencing data was analyzed on the basis of the proportions of reads in each taxon to the total number of sequences in the sample, whereas the average counts obtained through qRT-PCR data were Log10 transformed to conduct the statistical analyses.

Nonparametric Spearman's rank correlation test and receiver operating characteristic (ROC) analysis were performed in GraphPad Prism v4.00, as explained in Supplementary Materials and Methods.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

The ileal microbiota of 19 SBT recipients with different clinical histories was studied. A total of 35 ileal effluent samples (1–3 samples per subject) were collected for allograft monitoring purposes. Massively parallel sequencing of rRNA amplicons (V1–V3 region of the 16S rRNA gene) was used to analyze the bacterial communities present. Sequencing resulted in a total of 188 453 sequences after preprocessing and quality control steps. To standardize the number of sequences between samples, we included 1173–4800 sequences per sample in the analysis, which resulted in a total of 149 144 sequences (and an average of 4261 sequences per sample). The mean sequence length was 514 base pairs. Sequences were clustered with a 97% similarity cut-off, yielding 1217 OTUs for the entire sequence set, with an average of 64 OTUs per subject.

Microbial populations of ileal effluents

The ileal microbiota of SBT recipients was dominated by two bacterial phyla, Firmicutes (56.1%) and Proteobacteria (38.1%). Other minor phyla detected were Bacteroidetes (2.9%) and Actinobacteria (2.6%). We found the following bacterial orders in all transplant samples: Lactobacillales (48.3%), Enterobacteriales (36.8%), Clostridiales (7.2%) and Bacteroidales (2.9%). At the family level, the dominant groups were Enterobacteriaceae (36.8%), Streptococcaceae (20.0%), Enterococcaceae (13.9%), Lactobacillaceae (9.5%), Veillonellaceae (3.1%) and Peptostreptococcaceae (2.5%). Genera that prevailed throughout the entire collection of samples were Escherichia/Shigella (21.1%), Streptococcus (19.9%), Enterococcus (13.5%), Lactobacillus (9.4%), Klebsiella (7.6%), unclassified Enterobactericeae (5.5%), Veillonella (2.9%) and Enterobacter (2.3%).

Community analysis revealed interphenotype differences in bacterial populations

Comparisons between samples from NR, PR and AR SBT samples revealed major alterations in the proportions of multiple bacterial taxa associated with active rejection as compared to healthy transplants (Figure 1 and Table 1). Rejection was associated with phylum-level changes in the microbiota. A reduction in Firmicutes from 81 to 29% (p < 0.001) and an expansion of Proteobacteria from 16 to 61% (p < 0.01) were detected during rejection (Figures 2A and B). The decrease in Firmicutes was to a large extent because of a reduction in the order Lactobacillales (Figure 2C). At the family level, we observed a decrease of Streptococcaceae, Enterococcaceae and Lactobacillaceae during rejection (Figure 1C). The increase in Proteobacteria corresponded mainly with an enrichment of Enterobacteriaceae which expanded more than threefold in some rejection samples (p < 0.01; Figure 2D). Although an increase in the abundance of the genera Escherichia/Shigella was clearly detectable in multiple subjects, it did not achieve statistical significance in the entire data set. Interestingly, the three patients with acute rejection, who nonetheless showed no expansion of Escherichia/Shigella, still harbored a high proportion of the family Enterobacteriaceae (Figure S1A). In these patients, proportions of Klebsiella, Enterobacter, Citrobacter and unclassified members of this family were enriched (Figure S1B). Accordingly, the sum of the proportions of the genera Escherichia and Klebsiella increased significantly during active rejection (Figure 2E).

image

Figure 1. Collective ileal microbial composition in patients with nonrejecting (NR), prerejecting (PR) and actively rejecting (AR) intestinal allografts at the (A) phylum, (B) order, (C) family and (D) genus level. Only taxonomic groups ≥2% are shown.

Download figure to PowerPoint

Table 1.  Relative abundance of bacterial groups in the ileal microbiota that were affected by graft rejection status as determined by pyrosequencing of the 16S rRNA tags
 Bacterial taxaPercent abundance of bacterial taxa [Mean ± SEM]
NRPRARp Value
  1. Bacterial populations during different allograft status were compared to each other using Kruskal–Wallis.

  2. Numbers in bold represent proportions that were significantly higher than numbers shown in italics (p < 0.05), according to Dunn's Multiple Comparison Test.

PhylumFirmicutes81.22 ± 6.1555.68 ± 10.5328.77 ± 6.780.0009
 Proteobacteria16.00 ± 6.2839.97 ± 11.1460.82 ± 8.510.0052
OrderLactobacillales68.27 ± 8.8549.16 ± 11.3025.56 ± 5.620.0123
 Enterobacteriales12.59 ± 5.6739.86 ± 11.1660.54 ± 8.530.0017
FamilyEnterobacteriaceae12.59 ± 5.6739.86 ± 11.1660.54 ± 8.530.0017
 Bifidobacteriaceae0.07 ± 0.052.50 ± 1.630.06 ± 0.050.0159
 Unclass. “Lactobacillales”2.82 ± 1.06 1.55 ± 0.44 0.42 ± 0.11 0.0466
 Peptostreptococcaceae6.07 ± 6.02 0.91 ± 0.88 0.01 ± 0.01 0.0478
 Corynebacteriaceae0.02 ± 0.01 0.00 ± 0.00 0.00 ± 0.00 0.0755
GenusBifidobacterium0.05 ± 0.052.47 ± 1.610.06 ± 0.050.0040
 Unclass. “Peptostreptococcaceae”5.79 ± 5.77 0.02 ± 0.01 0.00 ± 0.00 0.0629
 Klebsiella0.47 ± 0.32 13.24 ± 9.43 10.46 ± 6.08 0.0758
 Unclass. “Streptococcaceae”0.07 ± 0.02 0.02 ± 0.01 0.01 ± 0.01 0.0783
 Corynebacterium0.02 ± 0.01 0.00 ± 0.00 0.00 ± 0.00 0.0802
 Escherichia & Klebsiella6.34 ± 3.0837.80 ± 11.5645.10 ± 9.860.0095
OTUEscherichia/Shigella6.13 ± 2.91 23.77 ± 10.7434.58 ± 10.410.0754
 Clostridium irregulare6.11 ± 6.09 0.00 ± 0.00 0.00 ± 0.00 0.0414
 Bacteroides fragilis0.01 ± 0.01 0.00 ± 0.00 6.99 ± 6.94 0.0650
 Corynebacterium durum0.02 ± 0.010.00 ± 0.00 0.00 ± 0.000.0217
 Actinomyces naeslundii0.03 ± 0.02 0.00 ± 0.00 0.00 ± 0.00 0.0624
RatioLactobacillales:Enterobacteriales511.90 ± 444.2023.42 ± 20.940.59 ± 0.190.0010
image

Figure 2. Relative proportions of representative bacterial groups within the ileal microbiota affected by graft status. (A) Phylum Firmicutes, (B) phylum Proteobacteria, (C) order Lactobacillales, (D) family Enterobacteriaceae, (E) the genera Escherichia and Klebsiella and (F) OTU related to Escherichia coli (similarity of ∼98% by NCBI BLASTn). Groups are determined to be statistically significant different if p-value is (*<0.05, **<0.01, ***<0.001).

Download figure to PowerPoint

To gain insight into the differences of ileal microbial communities between NR, PR and AR at the species level, we analyzed the sequence data on the basis of OTUs. Despite the alterations observed at higher taxonomic levels we found only two OTUs (related to Corynebacterium durum and Clostridium irregulare) that were significantly different between samples of distinct transplant status; specifically, both more prevalent during nonrejection (Table 1). In addition, an OTU related to E. coli/Shigella was more abundant during active rejection, approaching significance (p < 0.1; Figure 2F). As E. coli and Shigella spp. cannot be differentiated using 16S rRNA genes sequences, we tested for the presence of Shigella species using a PCR system that targets the ipaH gene, which is present in all four Shigella species as well as in EIEC (24). This PCR was negative for all samples included in this study (data not shown), indicating that the expansion of enterobacteria associated with rejection can largely be ascribed to commensal E. coli and not Shigella or EIEC.

Our analysis found only two taxa to show significant differences (p < 0.05) differences between NR and PR: the family Bifidobacteriaceae and the genus Bifidobacterium (Figure S2). However, this increase was mostly because of higher levels of bifidobacteria in only two individuals, hence it does not seem to be a general characteristic of prerejection. Although we did not detect significant differences between NR and PR for most taxa, the bacterial taxa that showed significant changes during active rejection showed the same trend during prerejection (Table 1, Figures 1 and 2). For example, the numbers of Escherichia and Klebsiella were very similar in PR and AR samples and clearly differed from individuals that showed no signs of rejection (Figure 2E). As the ROC analysis below demonstrates, a number of PR and AR samples had Escherichia or Klebsiella levels that were higher than any observed in the NR samples.

We were prompted to test if the differences in ileal microbiome composition observed between the NR versus AR group were a consequence of a differential use of antibiotics in these patients. Most patients were treated with antibiotics at the time of sampling as shown in Table S1. Because treatment with Bactrim was used consistently across the different patient groups, we compared the proportion of bacterial taxa that were affected by rejection status (Table 1) in patients that had and had not received antibiotics and in patients that had and had not received treatment with Vancomycin and Zosyn. This analysis revealed no significant differences (data not shown), suggesting that alterations in gut microbiome composition observed in NR and AR samples cannot be attributed to the use of antibiotics. We did not observe any differences between groups in the hospitalization rates around the time of sampling.

Group- and species-specific qRT-PCR

To validate the findings obtained with pyrosequencing and to determine absolute cell numbers of the bacterial taxa that showed links with graft rejection, we performed qRT-PCR using universal primers (to assess total bacterial counts) and primers specific for Enterobacteriaceae, lactobacilli (and related genera Pediococcus, Leuconostoc and Weissella), E. coli and the genus Bifidobacterium. Cell counts obtained by qRT-PCR in ileal samples showed strong positive correlations when compared with the relative proportions determined by pyrosequencing. Correlations were very high for E. coli (r = 0.903, p < 0.0001) and Bifidobacterium spp. (r = 0.898, p = 0.0002; Figure S3). There was also good agreement in both methods for Enterobacteriaceae (r = 0.548, p = 0.0017) and the Lactobacillus group (r = 0.792, p < 0.0001; Figure S3). This analysis demonstrated that the relative proportions of bacterial taxa obtained with pyrosequencing correlate well with absolute densities of bacteria and showed an enrichment of enterobacteria during rejection (Figure S4A). However, despite the good correlations between the two methods, qRT-PCR analysis failed to detect statistically significant differences between NR and AR (Figure S4A). This indicates that absolute quantification of bacterial taxa by qRT-PCR is not as discriminative as the relative proportions obtained by pyrosequencing. Relative proportions can also be studied by qRT-PCR. We did this by determining the enterobacteria/total bacteria ratio and as shown in Figure S4B, this analysis showed an improvement in detecting differences between healthy transplants and rejection when compared to absolute cell numbers (Figure S4A).

Bacterial density and diversity in transplant specimens

To determine if rejection of SBT was associated with alterations in bacterial numbers in the ileum, we quantified the total number of bacteria present in all ileal effluent samples by qRT-PCR using universal bacterial primers. Cell densities in our ileal effluents ranged from 2.4 × 105 to 5.91 × 109 cells/mL; these numbers were in general agreement with a previous study (5). Total bacteria counts were not significantly different between transplant phenotypes (p = 0.9186; Figure S5A). Diversity measurements via Shannon and Simpson's index revealed no significant alterations in microbial diversity in the different groups (Figures S5B–S5D). Although no significant differences in bacterial diversity were associated with rejection, we did observe variability between patients and temporal variability within individual patients.

ROC curve analysis

To assess the value of microbiota profiling as a diagnostic tool to discriminate and predict rejection, we performed a ROC analysis. This analysis revealed that population levels of Firmicutes could be used for the discrimination of NR and AR samples with excellent accuracy, resulting in an area under the curve (AUC) of 0.96 ± 0.04 (Figure 3A). A cut-off point of <49.7% of Firmicutes would hereby discern active rejection with 90.0% sensitivity and 90.9% specificity (Table S3). Other taxa or ratios between taxa were also highly accurate in predicting rejection (Table S3). As shown in Figures 3B–D, the phylum Proteobacteria, family Enterobacteriaceae and ratio between the orders Lactobacillales and Enterobacteriales can also be used for excellent discrimination between NR and AR (AUC of >0.9; 25). Importantly, population levels of the order and family Enterobacteriales and Enterobacteriaceae can be used to distinguish between NR and PR with moderate accuracy (AUC of 0.8), but with lower sensitivity and specificity compared to NR versus AR. All thresholds and discriminatory bacterial taxa for NR versus AR, and NR versus PR are shown in Table S3.

image

Figure 3. Receiver operating curves (ROC) showing the optimal cut-off values for the phyla. (A) Firmicutes, (B) Proteobacteria, (C) the family Enterobacteriaceae and (D) the genera Escherichia and Klebsiella to discriminate between presence or absence of acute rejection. p-Values and area under the curve (AUC) as determined by ROC analysis are also shown. Optimal threshold values for each are reported in Table S3.

Download figure to PowerPoint

Correlations between bacterial populations and clinical markers

We tested for associations between bacterial populations and the clinical markers calprotectin and ostomy output. Overall, the bacterial taxa that performed well in ROC analyses showed weak associations with these markers. Here we show the correlations for the phylum Firmicutes and the genera Escherichia/Klebsiella with calprotectin levels and ostomy output (Figure S6). Other bacterial taxa also showed weak associations with clinical markers used in this study (Table S4).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

In this study, pyrosequencing of 16S rRNA tags was used to characterize the bacterial populations present in the small bowel lumen (ileostomy output) of SBT recipients and to detect associations between bacterial taxa and rejection status. Overall, our analysis revealed that the microbial populations present in posttransplant samples resembled those found in the ileum of nontransplant subjects, with dominant bacterial groups belonging to the genera Streptococcus, Lactobacillus, Veillonella, Enterococcus, Escherichia and Klebsiella (26,27). In accordance with previous findings, ileal effluents contain low levels of strict anaerobes, most likely because of the introduction of oxygen into the normally anaerobic ileum (5). However, in-depth, global community analysis by pyrosequencing performed here indicated that the bacterial communities in ileal effluents are more complex than previously suggested by Hartman et al., who showed by qRT-PCR that the posttransplant microbial community was mainly dominated by lactobacilli and enterobacteria. In contrast, our study revealed that streptococci and enterococci were also dominant in ileostomy effluents.

The comparison between the ileal microbiota in NR, PR and AR samples revealed that acute graft rejection was associated with significant alterations in the microbial population structure. The most striking finding was the increase in the Proteobacteria/Firmicutes ratio during active rejection. This change was to a large degree because of an expansion in the family Enterobacteriaceae (and especially the species E. coli and Klebsiella pneumoniae; Figure 1). This family comprised, on average, 61% of the total microbial population in SBT patients during occurrence of active rejection (Figure S1A). The reduction in Firmicutes was largely because of a decrease in the order Lactobacillales. Significant differences in bacterial populations were not observed at lower taxonomic levels, indicating that multiple bacterial groups were negatively affected by rejection. Accordingly, the reduction in the Lactobacillales during rejection was caused through a decrease in the genera Streptococcus, Enterococcus and Lactobacillus, with high interindividual variation among these taxa.

The increase of enterobacteria in AR samples could also be confirmed with qRT-PCR (Figure S4A). However, differences in absolute cell numbers of enterobacteria did not reach statistical significance when AR samples were compared to NR samples. This was consistent with findings by Hartman et al., who did not detect an expansion of absolute numbers of enterobacteria by qRT-PCR during rejection episodes (5). Therefore, an important implication of our study was that relative shifts in the proportions of taxa, as obtained by deep sequencing, were more discriminative of rejection status than absolute cell numbers.

Unlike other types of transplantation (e.g. liver, renal), the intestine lacks a reliable and minimally invasive marker to predict rejection. Protocol biopsies and histological analysis remain the gold standard for allograft monitoring, but neither is free of complications, especially in smaller grafts (28,29). In addition, up to 30% of biopsies are nondiagnostic and multiple biopsies may be required to exclude rejection (30,31). Although several molecules have been suggested as biomarkers of rejection (e.g. calprotectin, citrulline, granzymeB and perforin), they generally lack sufficient sensitivity and specificity to be widely applicable (3,32–37). The results obtained during this study suggest that microbial profiling may be a potential diagnostic marker for graft rejection that could be used in conjunction with existing diagnostic tools to monitor SBT.

The ROC analysis suggests that microbial population shifts, and especially the proportion of the phylum Firmicutes and the family Enterobacteriaceae, are sensitive and specific indicators of rejection (Table S3). From our data, microbiome profiling has the potential to be used as an independent marker for allograft rejection, as the bacterial shifts detected showed weak correlations with ileostomy output and calprotectin levels (Figure S6, Table S4). It will require additional studies to validate microbiome profiling as a diagnostic and monitoring tool. If population shifts within the ileal microbiota are indeed predictive of transplant rejection, these tests could be valuable as a means of identifying patients “at risk” of rejection for more intense monitoring or altered immunosuppression. In future work, it will be especially important to determine if microbiome profiling can play a role in distinguishing between allograft rejection and infectious enteritis, as the clinical management of these conditions can be very different (3,38).

An important question raised by our findings is what causes the “dysbiosis” that occurs during rejection and whether it has a causative role in the rejection process. High levels of Proteobacteria and, more specifically, Enterobacteriaceae have been described in cases of diverse enteric inflammatory conditions (39–41). Our finding of increased of Proteobacteria and Enterobacteriaceae in intestinal inflammation and during active rejection suggests nonspecific alterations as a consequence of the inflammatory milieu. Intestinal inflammation could hereby disrupt the balance of the intestinal microbial community and promote the overgrowth of resident or introduced facultative anaerobes such as Enterobacteriaceae (42,43). In addition, the rejection process might result in secondary population shifts because of a reduced production of Paneth cells and enterocyte antimicrobial peptides (AMPs; Ref. 9). It is therefore possible that the structural shifts observed during rejection are a result rather than a cause of an exacerbated immune response.

However, it is also possible that the increased levels of Enterobacteriaceae found in PR and AR samples contribute to the aggravation of the rejection process. Enterobacteria encompass many bacterial strains that possess pathogenic traits such as adherence to epithelia and motility and enteroaggregative E. coli subtypes have been shown to be associated with Crohn's disease and to cause intestinal inflammation in several animal models (44–46). Likewise, Enterobacteriaceae (specifically Klebsiella pneumoniae and Proteus mirabilis) were described as provocative in a new murine model of ulcerative colitis (47). Prolonged graft survival observed in TLR4 knockout mice suggests that bacterial lipopolysaccharides (LPS) may play a role in the rejection process (15) and expression of TLR4 was increased 200-fold during allograft rejection in mice (48). LPS from E. coli and Klebsiella, when they are enriched during acute rejection, might contribute to immune reactions through an activation of the TLR4 pathway. Likewise, the 100-fold higher allograft failure in patents with NOD2 polymorphisms (9) suggests that the microbial populations shifts (potentially caused by a lack of AMPs) do contribute to the rejection process. Strategies that suppress the levels of enterobacteria might therefore constitute a viable therapeutic alternative to improve small intestinal allograft survival. But future studies are needed to separate the impact of the host genotype and the microbiota as dependent or independent risk factors for rejection.

The shifts in microbiome composition associated with rejection might also contribute to pathology because of a loss of protective bacterial species. Protective bacteria could be those microbes that by direct secretion of antimicrobial compounds, lowering the pH of the environment or by competitive exclusion, limit the growth and density of proinflammatory organisms found in the Proteobacteria phylum. The ileal microbiota of patients with healthy transplants was dominated by the phylum Firmicutes and the order Lactobacillales, particularly members of the families Streptococcaceae, Enterococcaceae and Lactobacillaceae (Figure 1C) and these groups diminish during rejection. Therapy by replenishing bacterial groups associated with healthy transplants, such as streptococci, enterococci and lactobacilli, could potentially be used to correct specific deficiencies in protective bacteria. Lactobacilli, for example, have been shown to both prevent colonization pathogenic species by competitive exclusion and to exert immunoregulatory activities (49,50).

Concluding Remarks

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

The cost and feasibility of DNA sequencing has opened new opportunities for its use in human medicine, placing the analysis of the gut microbiota at our fingertips. The use of universal primers and pyrosequencing provides the most unbiased approach to measure the composition of the gut microbiota and in our case provided better discriminatory power than using a PCR-based approach with targeted primers. The concept of abnormal or detrimental microbial communities has been previously discussed (6) and our findings during this study indicate that gut transplants represent a scenario where a normal microbiota is associated with healthy transplants, although a deviation from “normal” (dysbiosis) accompanies rejection. Dysbiosis is associated with an increase in Enterobacteriaceae and a reduction in several members of the phylum Firmicutes and order Lactobacillales. These shifts associated with acute rejection could constitute a factor that predisposes transplant recipients to developing graft rejection or simply a consequence of rejection induced changes in the gut habitat where this microbial community resides. Future studies with a larger number of patients should be aimed at the validation of microbial profiling as a diagnostic marker for graft rejection and also toward the therapeutic approaches aimed at the microbiota for preventing or treating rejection episodes.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

We would like to thank Jaehyoung Kim, Min Zhang and the Core for Applied Genomics and Ecology (CAGE, University of Nebraska-Lincoln) for conducting the pyrosequencing. This work was supported by Grant NIH-K08AI076609 (DAP) and Funds from the Nebraska Tobacco Settlement Biomedical Research Enhancement Funds (DAP).

Disclosure

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Concluding Remarks
  8. Acknowledgments
  9. Disclosure
  10. References
  11. Supporting Information

Supplemental Materials and Methods

Table S1: Clinical characteristics of 19 SBT patients

Table S2: List of PCR primers and annealing temperatures (Tm) used in this study

Table S3: ROC analyses to determine the potential use of bacterial populations in ileal effluents to predict SBT rejection

Table S4: Correlations between the relative abundance of bacterial populations (as determined by pyrosequencing) and calprotectin concentration, total calprotectin and ostomy output

Figure S1: Microbiota composition of individual ileal samples (n &equals; 35) at the (A) family and (B) genus level. Samples were grouped according to their allograft status: nonrejection (NR), prerejection (PR) and active rejection (AR). Samples marked with &ast; eventually developed active rejection.

Figure S2: Changes observed in relative proportions of the (A) family Bifidobacteriaceae and (B) genus Bifidobacterium as determined by pyrosequencing.

Figure S3: Spearman correlations between relative abundance of specific bacterial taxa as determined by pyrosequencing and absolute cell numbers measured by qRT-PCR.

Figure S4: Absolute cell numbers (A) and relative proportions (B) of Enterobacteria in ileal effluents determined by qRT-PCR.

Figure S5: Total bacterial density and microbial diversity in ileal effluents of patients with NR, PR and AR transplants. (A) Total bacterial cell numbers as determined by qRT-PCR using universal bacterial primers. Microbial diversity as determined by (B) Shannon diversity index, (C) Simpson diversity index and (D) total number of species in samples.

Figure S6: Bacterial populations in the ileal effluent samples such as (A) phylum Firmicutes and (B) genera Escherichia and Klebsiella that showed low associations with calprotectin and ileal fluid output.

FilenameFormatSizeDescription
AJT_3860_sm_SuppMat.pdf1198KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.