Corrigendum: Intestinal microbiome in children with severe and complicated acute viral gastroenteritis

This corrects the article DOI: 10.1038/srep46130.

infection, including 15 with severe AGE (Vesikari score > 10) 15 and 5 with mild-to-moderate AGE (Vesikari score ≦ 10). The 15 complication presentations comprised 10 AGE with convulsion (7 norovirus, 3 rotavirus infections), 2 necrotizing enterocolitis (norovirus infections), 2 gastroenteritis with renal failure or severe electrolyte imbalance (norovirus infections), and 1 gastroenteritis with chronic diarrhea of prolonged duration (> 3 weeks) and malnutrition (norovirus infection). Antibiotics were administered in 2 infants with necrotizing enterocolitis before surgery and a further 1 infant with renal failure. All fecal samples were collected at least 1 week after discontinuation of the antibiotics. Enteric bacteria in the 40 fecal samples were classified into 9 phyla, 62 families, and 159 genera based on 16S rRNA sequences and metagenomic analysis of all selected reads.
Diversity of the intestinal microbiota in children with severe gastroenteritis. The Shannon diversity indices of the fecal samples were calculated and reported as entropy scores 16 . These are indicative of the intestinal microbiota composition and disruption of the microbiota, as scores are lower in the presence of overgrowth of a single bacterial species. The entropy score of the intestinal microbiota of the normal controls, and patients with mild-to-moderate and severe AGE were 0.56 (0.47-0.73), 0.7 (0.56-0.93), and 0.3 (0.05-0.6), respectively (Fig. 1A). The Shannon diversity index (entropy score) of the intestinal microbiota was significantly lower in patients with severe AGE than in normal controls (P = 0.017) and patients with mild-to-moderate AGE (P = 0.011).
The average severity score of the 5 patients with rotavirus infection was 13.8 (range, [11][12][13][14][15], which is significantly higher than that of the 15 patients with norovirus infection (mean, 10.3; range, 8-11) (P = 0.03). The entropy score in patients with norovirus infection was 0.65 (0.2-0.75), and was not significantly different compared with that of the normal controls. The entropy score of patients with rotavirus infection was 0.05 (0.02-0.35), which was significantly lower than that of patients with norovirus infection (P = 0.048) (Fig. 1B).
The intestinal microbiota composition at the phylum level of normal controls, and patients with mild-to-moderate and severe AGE is shown in Fig. 2A. The microbiota composition in patients with rotavirus or norovirus infection and the associated disease severity classification are shown in Fig. 2B. There was no significant difference in microbiota composition between the normal controls and patients with AGE at the phylum level. The feces of normal controls exhibited significant abundance of Rikenellaceae (P = 0.043) and Porphyromonadaceae (P = 0.02) at the family level, and Alistipes (P = 0.038) and Parabacteroides (P = 0.019) at the genus level compared with any severity of AGE patients.
We also analyzed the intestinal microbiota composition of 3 patients with complicated AGE, comprising 1 patient with necrotizing enterocolitis after norovirus infection, 1 with rotavirus infection with convulsions, and 1 with norovirus infection and renal failure ( Supplementary Fig. 1). In the norovirus infection patient with necrotizing enterocolitis, the entropy score of the intestinal microbiota was 0.008833, and Streptococcaceae (63%), Enterobacteriaceae (23%), and Pasteurellaceae (6%) comprised > 90% of the microbiota. The entropy score of patient with rotavirus infection and convulsions was 0.02779, and Bacteroidaceae comprised almost 100% of the intestinal microbiota in this patient. The entropy score of norovirus infection patient with renal failure was 0.005949, and Enterococcaceae, Clostridiaceae, and Streptococcaceae comprised > 95% of the microbiota. Microbiota of viral AGE patients with and without complications. There were 15 children that experienced complications after viral AGE and 5 without complications. The intestinal microbiota of the children with viral AGE differed from that of healthy controls at the family level ( Table 2). The overall differences in healthy children, and those with non-complicated and complicated AGE were in the richness of Porphyromonadaceae in healthy children (P = 0.035) and Streptococcaceae in uncomplicated AGE (P = 0.035), shown by the composition of fecal bacteria. Patients with complicated AGE had greater abundance of Campylobacteraceae (P = 0.0003), Neisseriaceae (P = 0.0115), Methylobacteriaceae (P = 0.0004), Sphingomonadaceae (P = 0.0221), and Enterobacteriaceae (P = 0.0451) than healthy children. Uncomplicated AGE was associated with a decrease in the abundance of Desulfovibrionaceae (P = 0.0198), Ruminococcaceae (P = 0.0125), Veillonellaceae (P = 0.0199) and increase in Carnobacteriaceae (P = 0.0124). Finally, patients with complicated AGE had greater Pasteurellaceae richness (P = 0.0071) than patients with AGE without complications.

Discussion
The intestinal microbiota diversity was less in patients with severe viral AGE compared with healthy controls and those with mild-to-moderate AGE, suggesting the counterpoise of microbiota being changed or disrupted into some overgrowing bacterial species. Recent studies suggested that disruption of the gut microbiota is associated with antibiotic use following viral infections 13,14,17 . Rotavirus infection resulted in decreased microbiota diversity compared with norovirus infection, which has not to our knowledge been reported previously although in our study, the mean severity of rotavirus AGE is higher. Recent studies showed that norovirus infection of B cells in vivo was facilitated by bacteria expressing an appropriate histo-blood group antigen 18,19 . This could explain the difference in microbiota composition between patients with rotavirus and norovirus infections.
Richness or reduction in specific families or genera was related to the presence of abdominal pain, longer hospitalization, multi-system involvement, and convulsion, but not diarrhea, vomiting, or fever also gender and age. An association between intestinal symptoms and the intestinal microbiota, as well as a significant inverse correlation between Bifidobacteria and abdominal pain, has been reported in adults 20 ; no such association was found in children in this study. Nevertheless, our results showed that several bacterial taxa-such as Prevotellaceae, Staphylococcaceae, and Coriobacteriaceae-were more abundant in children with abdominal pain. The relative instability of the intestinal microbial community in children might explain this discrepancy.
Another study reported that patients with irritable bowel syndrome had significantly higher counts of Veillonella than controls, which appeared to be associated with high acetic acid or propionic acid levels in the bowel 21 . Probiotic supplementation could stabilize the microbiota and alleviate symptoms of irritable bowel syndrome 22,23 . However, this was not the case in our study: the major features of patients with abdominal pain were overgrowth of particular bacterial species, and no deficiency in specific microbial taxa was evident. Therefore, supplementation should not be recommended for AGE patients with significant abdominal pain.
The microbiota of stool specimens from patients with complicated AGE exhibited greater richness either at the phylum, family, or species level, suggesting that several important gut species remain to be characterized. The characterization offers a new approach to identifying individuals at high risk of developing specific complications following viral AGE. During early life, the intestinal microbiota changes with age 24  and norovirus infection complicated by renal failure, extremely lower microbiota diversity was found. This suggests that these complications could be associated with disruption of the microbiota. When compared with the intestinal microbiota in healthy early life 8,24 , the dominant taxa in complicated AGE were Streptococcaceae, Enterobacteriaceae, and Pasteurellaceae rather than Bifidobacteriaceae, Clostridiaceae, and Lactobacillaceae in a representative necrotizing enterocolitis patient; Bacteroidaceae, Lachnospiraceae, and Porphyromonadaceae rather than Clostridiaceae, Lachnospiraceae, and Bifidobacteriaceae in a representative patient with convulsions; and Enterococcaceae, Clostridiaceae, and Streptococcaceae rather than Clostridiaceae, Lachnospiraceae, and Ruminococcaceae in a representative acute renal failure patient. Our data showed that, in addition to antibiotic use (which may influence the microbiota composition), viral infection might also affect the diversity and predominant taxa of the intestinal microbiota, particularly in patients with severe disease or complications.
The use of probiotics such as Lactobacillus rhamnosus GG and Saccharomyces boulardii, or the combination of Streptococcaceae faecalis, Clostridium butyricum, and Bacillus mesentericus to treat or prevent AGE has been evaluated 25,26 . In the present study, Streptococcaceae richness was significantly greater in patients with uncomplicated compared with complicated AGE, and Parabacteroides and Porphyromonadaceae abundance was greater in healthy controls than in patients with severe AGE and in the complicated ones individually and this would trigger more follow-up investigations for future clinical application.
This study had several limitations. First, a limited number of cases were evaluated. Second, use of antibiotics by some of the patients prior to confirmation of a viral etiology could have affected the intestinal microbiota. Third, acute clinical symptoms and intestinal microbiota composition were investigated at a single time point; no longitudinal follow-up was performed.
In conclusion, we evaluated the intestinal microbiota composition of children with viral AGE and its correlations with various severities also with or without complications. The results will facilitate further studies of the interaction between the intestinal microbiota and viral pathogens, and use of probiotics to treat AGE caused by rotavirus or norovirus.

Methods
Ethics. This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (CGMH100-4283A3), and the participants or their guardians provided informed consent for specimen and clinical data collection. All applied methods were performed in accordance with the approved guidelines.

Patient enrollment and sample collection. Patients less than 18 years old hospitalized at Chang Gung
Children's Hospital from August 2012 to July 2013 whose major clinical manifestation was AGE and who fulfilled the inclusion criteria were offered enrollment. All medical records of enrolled patients were reviewed for the clinical and demographic data. We evaluated the severity of gastroenteritis using the Vesikari score system 15 . A total score of > 10 indicates severe gastroenteritis (maximum total score 20, total score < 7, mild; 7 ≤ 10, moderate; and > 10, severe). Children with gastroenteritis symptoms and any one of the following clinical complications or conditions were included in the complicated gastroenteritis group: extraintestinal or unusual presentations of viral gastroenteritis such as electrolyte imbalance, renal failure, shock, convulsive disorder, prominent hyperthermia (> 39 °C), hypotension or hypovolemic shock (systolic blood pressure < 70 mmHg), hypoglycemia (serum sugar < 70 mg/dL), chronic diarrhea (> 3 weeks) resulting in malnutrition, and necrotizing enterocolitis. Involvement of multiple systems indicated involvement of systems other than the gastrointestinal tract. Twenty otherwise healthy children were enrolled as negative controls. Their fecal specimens were collected within 3 d of hospitalization or complication events and were stored at − 70 °C before extraction.
Extraction of stool DNA. Stool total DNA was extracted using a QIAamp ® DNA Stool Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's recommendations including the procedure of heating suspension for 5 min at high temperature of 95 °C and this can help to lyse Gram positive bacteria and was stored at − 80 °C 27 . Universal primers for the 16S variable regions V1-3 and V3-5 were used for polymerase chain reaction (PCR) amplification. The V3-534 and V5-926 reverse primers included a unique sequence tag to barcode the samples (http://www.hmpdacc.org/doc/16S_Sequencing_SOP_4.2.2.pdf).

NGS.
The purified amplicon mixtures were sequenced using the 454 GS Junior System according to the protocols recommended by the manufacturer (454 Life Sciences, Branford, CT, USA). The regions of 16S rRNA genes were amplified using FastStart ™ HiFi Polymerase (Roche, Manheim, Germany). The PCR amplicons were purified using Agencourt ® AMPure ® XP Reagent (Beckman Coulter, Brea, CA, USA) and quantified using an Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).

Statistical and comparative genomics analyses.
To analyze the data, ~1 × 10 5 reads not aligned to the human genome were randomly selected from all reads with an average length of ~100 base pairs, and similar sequences were searched for in the NCBI nt (National Center for Biotechnology Information nucleotide) database using BLASTn (Basic Local Alignment Search Tool nucleotide database) with a threshold e-value of 5 × 10 −5 . The full-length reference genome was used for complete genome analysis. Sequences were aligned using the CLUSTAL W software included in MEGA v. 5.0 (http://mega.software.informer.com/5.0/) and the bootstrap was plotted for a window of 300 bp, moving in increments of 10 bp along the alignment, using Simplot software (http://sray.med.som. jhmi.edu/SCRoftware/simplot/). All reference sequences used in this study were obtained from GenBank (http:// www.ncbi.nlm.nih.gov/genbank/). The fecal microbiota composition was characterized compared to microbiota across body sites in the Human Microbiome Project (HMP) database (https://commonfund.nih.gov/hmp/index).
Continuous clinical data were analyzed using Student's t-tests and expressed as means ± standard deviation (SD). Binary data were analyzed using the χ 2 test. A value of P < 0.05 was considered to indicate statistical significance. All tests were performed using SAS software v. 8 for Windows (SAS Institute Inc., Cary, NC, USA).