Gut microbial dysbiosis is associated with development and progression of radiation enteritis during pelvic radiotherapy

Abstract Radiation enteritis (RE) is the most common complication of radiotherapy for pelvic irradiation receivers. Herein we investigated the alterations in gut microbial profiles and their association with enteritis in patients undergoing pelvic radiotherapy. Faecal samples were collected from 18 cervical cancer patients during radiotherapy. Microbiota profiles were characterized based on 16S rRNA sequencing using the Illumina HiSeq platform. Epithelial inflammatory response was evaluated using bacterial‐epithelial co‐cultures. Dysbiosis was observed among patients with RE, which was characterized by significantly reduced α‐diversity but increased β‐diversity, relative higher abundance of Proteobacteria and Gammaproteobacteria and lower abundance of Bacteroides . Coprococcus was clearly enriched prior to radiotherapy in patients who later developed RE. Metastat analysis further revealed unique grade‐related microbial features, such as more abundant Virgibacillus and Alcanivorax in patients with mild enteritis. Additionally, using bacterial‐epithelial co‐cultures, RE patient‐derived microbiota induced epithelial inflammation and barrier dysfunction, enhanced TNF‐α and IL‐1β expression compared with control microbiota. Taken together, we define the overall picture of gut microbiota in patients with RE. Our results suggest that dysbiosis of gut microbiota may contribute to development and progression of RE. Gut microbiota can offer a set of biomarkers for prediction, disease activity evaluation and treatment selection in RE.

treatment efficacy. 3,4 There is neither standardized prophylactic nor therapeutic strategies available proven to mitigate the radiation enteritis (RE) symptoms or allow safe radiation dose escalation for better cancer control. In 2002, Pietro Delia 5 reported that use of Lactobacillus rhamnosus could prevent the occurrence of diarrhoea in patients receiving radiotherapy. Later, Crawford and Gordon 6 revealed that germ-free mice were markedly resistant to lethal RE. And after faecal microbiota transplantation, the intestinal function and survival rate were significantly improved in irradiated mice. 7 These reports suggest a human-microbiome link upon irradiation and provided an insight into potential radio-protective therapeutics.
The advancement of recent next-generation sequencing technologies and bioinformatics has changed the way research is done in microbial ecology. It analyses the complete bacterial genome sequences and provides enormous amounts of information. 8  In this study, using the high-throughput 16S rRNA gene sequencing, we identified specific faecal microbial signatures in patients with RE and sought to elucidate potential biomarkers or mechanistic principles how the gut microbiota dysbiosis may impact the pathogenesis of RE. The results will also provide useful information about the therapeutic value of microecological preparation for RE.

| Sample collection
Eighteen patients with stage II-IV cervical cancer (CCa) who had not received any treatments for those conditions and were undergoing pelvic radiotherapy were recruited in our department from June 2015 to January 2016. The detailed clinical parameters are shown in Table 1.
The exclusion criteria were as follows: recent (<2 months prior) use of any antibiotic or probiotic therapy, recent (<2 weeks prior) use of any proton pump inhibitors, known any other enteritis, 9,10 known autoimmune condition, significant gastrointestinal disorder, age <18 years, vegetarians, abnormal BMI value (<18.5 or >24), known history of any other cancer, 12 and significant liver, renal, or peptic ulcer disease.
Baselines of bowel habit and symptom were recorded, and patients with prior higher bowel symptoms were excluded as well. Pelvic radiotherapy was delivered at total doses of 50. 4

| In vitro studies
The human normal colonic epithelial cell line consisting of FHC (foetal colon) was provided by Dr Liang Peng (The First Affiliated Hospital of Guangzhou Medical University), and cultured routinely in RPMI1640 medium. Faecal bacterial suspension was obtained as previously described. 13

| Bioinformatics and statistical analysis
Paired-end reads were assigned to samples based on their unique barcode and bioinformatic processing was done as previously described. 15,16 Briefly, quality control was performed using QIIME V1.7.0 software package to obtain high-quality clean tags. For in vitro studies, data from three or more independent experiments was expressed as mean ± SE and processed using SPSS18.0 statistical software. Statistical analysis was performed with t test, non-parametric Mann-Whitney and factorial analysis. P < 0.05 from two-sided tests was taken as statistical significance.

| RE SULTS
In total, we generated 1 617 140 paired-end reads of high-quality sequences (average 57 755 per sample). The total number of OTUs  Patient samples were further classified into three groups according to the grade of RE using the Radiation Therapy Oncology Group (RTOG) grading system. 19 Three patients in the RE group developed grade 1 radiation toxicity (RE1), three developed grade 2 toxicity (RE2) and another four developed grade 3 toxicity (RE3). No grade 4 or 5 was recorded. Relative to those with mild toxicity, patients with severe enteritis had a significantly reduced α-diversity (ρ = 0.034, Figure 5A) but a non-significantly increased β-diversity ( Figure 5B).
Patients at grade 3 RE had lowest α-diversity and highest β-diversity among all the RE patients, indicating a potential trend of gradual microbial response to radiation inflammation. Analysis of microbial composition revealed a grade-related microbial feature ( Figure 5C).
An epithelial monolayer cell co-culture model was used to explore the effects of radiation-induced microbial dysbiosis on epithelial inflammatory response. Following co-culture with bacterial suspensions from RE3 patients, compared with suspension from non-RE patients, inflammatory and barrier markers were significantly downregulated. Membrane Syndecan-1 was released from the cell surface.

| D ISCUSS I ON
Since its first report in 1897, 20   be related to differences in the severity of local mucosal inflammation, or changes in epithelial permeability or barrier. Factors which were not investigated here might also play a part. However, these results have strongly suggested the importance of gut microbiota and the need for further and more detailed investigation.
We did some preliminary in vitro experiments to explore whether and how the microbiota affected the radiation-associated tissue damage. Incubating colonic epithelial cells with faecal bacteria from patients with severe RE impaired cell layer integrity, increased cell layer permeability and stimulated cytokine secretion and NF-κB activation. Thus the dysbiotic microbiota might in part directly induce barrier dysfunction and inflammatory response on the epithelial cells. Otherwise, microbiota significantly differed following local radiation treatment. And these changes paralleled with the cytokine profile in patients with or without RE. It was suggested the host immune response upon irradiation shapes the microbial community structure. When cells are exposed to radiation, leucocytes infiltrate into the irradiated normal cells. Various signalling pathways are activated, accompanied with secretion of pro-inflammatory cytokines, shedding of mucosa, disruption of barrier and initiation of coagulation cascade. 23 For example, pathway analysis on the gene expression profiles has identified radiation-induced time-, dose-and even segment-dependent up-regulation of TNF-α, claudin-2, MMP7 and EDA2R. 24 Radiation also provokes increase in MPO activity and CXC chemokine levels. 25 Activation of these pathways suggest that colon sustains severe mucosal inflammation and barrier disruption, and might influence and disturb the balance of microecology. We previously demonstrated that loss of Syndecan-1 in the inflamed intestine impaired normal intestinal barrier and led to bacterial translocation through mucosa. 14 Winter et al 26,27 found that host-derived nitrate F I G U R E 6 Faecal microbiota is associated with intestinal inflammation and barrier function. Foetal colon cells were co-cultured with irradiated microbiota from patients with grade 3 RE. (A) Barrier-associated proteins and NF-κB activity were determined using Western blot. (B) Cytokine secretion was determined using quantitative PCR. (C) Epithelial integrity was determined with TEER. (D) Epithelial permeability was determined with FITC-dextran flux. Patients' serum were collected. Levels of shed Syndecan-1 (E) and cytokines TNF-α and IL-1β (F) were detected by ELISA in response to mucosal inflammation conferred a growth advantage to commensal Escherichia coli or pathogenic Salmonella enterica in the mice intestine. Taken together, although it is tricky to decipher the question of cause and effect, these data are still sufficient to confirm that the unique radiation-induced dysbiosis is closely associated with inflammatory response.
Our results suggested that the pre-existing changes in gut microbial ecology may serve as a predictive marker to identify patients who are more likely to progress to RE during pelvic irradiation, in agreement with Wang et al's proposal. 21 Moreover, our data suggested possibility to prevent or treat RE by targeting the gut microbiota. In mouse model, gavage of Lactobacillus rhamnosus GG before radiation repositioned COX-2 expression through TLR-2/MyD88 signalling and reduced epithelial apoptosis and crypt loss from radiation injury. 28 In patients, prevention of radiotherapy-induced mucositis by probiotics has been investigated in several clinical trials. 29,30 Although the results were inconsistent, and strong evidence is lacking, there was still some promising data. Chitapanarux's randomized study included 63 patients treated with pelvic radiotherapy concurrent with weekly cisplatin chemotherapy. As compared with placebo, treatment with live Lactobacillus acidophilus plus Bifidobacterium bifidum resulted in improved stool consistency and less usage of anti-diarrhoeal medication. 31 In Urbancsek's larger randomized trial with 206 irradiated patients, supplementation with Lactobacillus rhamnosus led to less frequently needed anti-diarrhoeal drugs. 32 Furthermore, L. Fuccio 33 systemically reviewed clinical trials including the above two. However, no significant differences were confirmed between probiotic supplementation and placebos.
Despite the few available trials and the presence of significant clinical and statistical heterogeneity might limit the analysis, encouraging results have been indeed observed in some patients. Because not all probiotics exert favourable effects, possibly owing to variability of probiotics and patient characteristics, the importance of identifying the classification of patients and the ideal type and dose of bacterial strains need to be addressed in further high-quality clinical trials. Moreover, cancer patients are generally at risk of disease-or treatment-related immunosuppression. Microbial preparation may induce detrimental effects in these individuals; note some published reports of septic complications because of probiotics. 34 Therefore, safety concerns about the use of probiotics should also be carefully investigated.
In conclusion, we reported the comprehensive analysis of gut microbiota in patients with RE using faecal samples by high-throughput 16S rRNA sequencing. We identified the radiation-induced impaired gut microbiota and its relationship with RE. Our results will be helpful for the prediction and treatment of cancer patients receiving pelvic irradiation and suffering from RE. Furthermore, multicenter, randomized and placebo-controlled trials are needed to confirm this.

ACK N OWLED G EM ENT
This work was supported by the National Natural Science Foundation of China (81602677 and 81872472).

CO N FLI C T O F I NTE R E S T
The authors declare that there are no conflicts of interest.

AUTH O R S' CO NTR I B UTI O N S
ZQW and ZYY designed the study. ZQW, QXW, XW performed the actual laboratory analyses. QXW, LZ, JC and BLZ obtained the samples and analysed the data. ZQW, WW, YC and ZYY wrote and revised the manuscript.