ID EMI-2017-0646 Re-submission Title: Bangladeshi children with acute diarrhea show fecal microbiomes with increased Streptococcus abundance, irrespective of diarrhea etiology Short title: Fecal Streptococcus increase in acute diarrhea Authors: Silas Kieser1*, Shafiqul A. Sarker2*, Olga Sakwinska1*, Francis Foata1*, Shamima Sultana2, Zeenat Khan2, Shoheb Islam2, Nadine Porta1, Séverine Combremont1, Bertrand Betrisey3, Coralie Fournier3, Aline Charpagne3, Patrick Descombes3, Annick Mercenier1, Bernard Berger1° , Harald Brüssow1+ * these authors contributed equally to this work ° Corresponding author Bernard Berger P.O. Box 44 CH-1000 Lausanne 26 Switzerland e-mail: [email protected]
Phone: +41 21 785 8955 + Current address: KU Leuven, Division of Animal and Human Health Engineering, Kasteelpark, Arenberg 21, Box 2462, 3001 Leuven, Belgium, e-mail:
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an ‘Accepted Article’, doi: 10.1111/1462-2920.14274 This article is protected by copyright. All rights reserved.
Page 2 of 47
1 Gut Ecosystem Department, Institute of Nutritional Science, Nestlé Research Centre, Verschez-les-Blanc, CH-1000 Lausanne 26, Switzerland 2 International Centre for Diarrheal Diseases Research, Bangladesh (icddr,b), 68 Shaheed Tajuddin Ahmed Sharani, Mohakhali, Dhaka 1212, Bangladesh, 3 Nestlé Institute of Health Sciences, EPFL Innovation Park, CH-1015 Lausanne, Switzerland
2 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 3 of 47
Summary We report streptococcal dysbiosis in acute diarrhea irrespective of etiology. Compared to 20 healthy local controls, 71 Bangladeshi children hospitalized with acute diarrhea (AD) of viral, mixed viral /bacterial, bacterial and unknown etiology showed a significantly decreased bacterial diversity with loss of pathways characteristic for the healthy distal colon microbiome (mannan degradation, methylerythritol phosphate and thiamin biosynthesis), an increased proportion of fecal streptococci belonging to the Streptococcus bovis and Streptococcus salivarius species complexes, and an increased level of E. coli-associated virulence genes. No enteropathogens could be attributed to a subgroup of patients. Elevated lytic coliphage DNA was detected in 2 out of 5 investigated enteroaggregative E. coli (EAEC)-infected patients. Streptococcal outgrowth in AD is discussed as a potential nutrient-driven consequence of glucose provided with oral rehydration solution.
Keywords: acute diarrhea; gut microbiota; streptococcal dysbiosis; oral glucose; bacteriophages; Bangladesh.
Originality-Significance Statement Our observations are challenging widely shared concepts in acute diarrhea pathogenesis, by raising questions on its etiology and the potential impact of the carbohydrates supplied in the oral rehydration solution used in the WHO recommended treatment.
3 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 4 of 47
Introduction Worldwide, diarrhea is the second most important cause of childhood morbidity and mortality (Black et al., 2008). The introduction of oral rehydration solution (ORS) (Munos et al., 2010) and zinc supplementation (Walker and Black, 2010) have decreased diarrheal mortality. Efficient vaccines have been introduced for rotavirus (RV) (Breiman et al., 2012), but not yet for Escherichia coli (Ahmed et al. 2013), which was described as an important cause of childhood diarrhea in developing countries (Qadri et al., 2005). Since E. coli is resistant to many antibiotics (Jiang et al., 2002), we previously explored bacteriophages for the treatment of diarrhea in Bangladeshi children hospitalized mostly with enterotoxigenic E. coli (ETEC) infection (Sarker and Brüssow, 2016). No clinical amelioration was achieved with two oral bacteriophage cocktails over standard of care consisting of reduced glucose ORS-zinc therapy in a placebo-controlled, randomized clinical trial (Sarker et al., 2016). Strikingly, stool microbiota analysis in these children showed only low median ETEC titers of 105 cfu / g stool (Sarker et al., 2016). ETEC titers were probably lower than needed for in vivo replication of T4 phage (Wiggins and Alexander, 1985). Low pathogen titers of 3x105 cfu/ g stool, just 10fold higher than in controls, were also reported for enteropathogenic E. coli (EPEC) infections in Peruvian children (Barletta et al., 2011). Furthermore, large epidemiological studies reported a low pathogenicity index for several pathotypes of diarrhea-associated E. coli in children from low-income countries (Kotloff et al., 2013; Liu et al., 2015). In Bangladeshi diarrhea patients, pathogenic E. coli was frequently detected in association with other pathogens (Taniuchi et al. 2013), raising doubts about their pathogenic role. In the phage trial, fecal Streptococcus but not Escherichia abundance correlated with diarrhea symptoms both quantitatively and temporally (Sarker et al., 2016). Fecal streptococci are considered as commensals of the small intestine (Booijink et al., 2010), but were also suggested as a candidate diarrhea pathogen in children lacking enteropathogens in the stool (Jin et al., 2013). 4 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 5 of 47
In the present report, we specifically asked whether elevated abundance of streptococci in stools as described in E. coli diarrhea (Sarker et al., 2016) occurs in pediatric diarrhea patients independent of etiology. Since the pathogenesis of diarrhea in the absence of known pathogen detection remains poorly understood, we also investigated with molecular pathogen detection tests and metagenome analysis a subgroup of patients without pathogen diagnosis and compared them to health controls and to patients with diagnosed enteroaggregative E. coli (EAEC) diarrhea.
Results Patient characteristics Between November 2014 and March 2015, 71 children, hospitalized with acute watery diarrhea (AD), were enrolled into an observational study. AD cases differed from 20 local healthy control (HC) children for increased stool frequency, vomiting, elevated body temperature, increased blood pressure, heart and respiration rates (Table 1). Anthropometric parameters defining underweight (weight for age Z-score < -2 standard deviation of WHO reference values), stunting (height for age Z score < -2) and wasting (weight for height Zscore< -2) did not differ significantly between cases and controls (Table 1). Controls were 3 months older, had less illiterate mothers and a higher rate of completed immunizations than cases (Table 2). Cases and controls did not differ for age of mother, number of siblings, prevalence of low family income, length of exclusive breastfeeding or home medication (one case received antibiotics at home, antibiotics were not used in the hospital) (Table 2). Pathogen analysis Etiology was established with standard methods at icddr,b (Harris et al., 2008). Twenty-six patients revealed a RV infection without detection of a bacterial co-pathogen. Sixteen patients 5 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 6 of 47
showed RV associated with enteroaggregative E. coli (EAEC), while 7 patients displayed only E. coli infections (6 EAEC, 1 ETEC). Seven patients harbored other bacterial pathogens (Aeromonas; Aeromonas + RV; Aeromonas + Vibrio ; Vibrio; Vibrio + ETEC; Campylobacter + RV; Shigella + ETEC). Fifteen patients did not yield a pathogen (AD-p) with standard procedures. TaqMan Array Card analysis tested for the presence of 19 pathogens in 11 randomly selected healthy control children and these 15 AD-p patients. Only two enteropathogens, namely rotavirus, and Cryptosporidium, returned significantly higher detection in AD-p (Fig. 1A and suppl. Table 4). Surprisingly, half of the controls showed higher PCR signals than AD-p for Campylobacter jejuni. Only the higher prevalence of C. jejuni in HC was confirmed by metagenome sequencing (Fig 1C, see below and suppl. Table 4 for the metagenomics results), possibly because the other identified pathogens detected in the diagnostic tests did not represent a sufficient part of the fecal microbiota to be measured by standard metagenomics sequencing depth. In all samples but one, a positive signal was observed for more than one enteropathogen, with no difference in the distribution of infection multiplicity between HC and AD-p (Fig 1 B). Stool microbiota composition by 16S rRNA gene sequencing Next we investigated the stool microbiota composition by 16S rRNA gene sequencing in 20 controls and 71 AD patients, including 26 RV, 16 RV + EAEC, 7 ETEC + EAEC, 15 AD-p and 7 patients with a diversity of other bacterial pathogens. There was no consistent correlation between iccdr,b diagnosis and 16S rRNA gene sequencing (data not shown), partly explained by the low resolution of the latter particularly for Enterobacteriaceae. Compared to controls, a higher abundance of Streptococcus was observed in all AD categories (Fig. 2A) and in most individual AD patients (Fig. 3A). The increase was statistically significant (Fig. 2D). On the species level, most streptococci were represented by two species complexes (Gao et al., 2014; Pontigo et al., 2015): S. bovis (including S. equinus 6 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 7 of 47
and S. gallolyticus) and S. salivarius (Fig. 3B). Other typical small intestine bacterial groups either were not detected (Bilophila) or were decreased in cases over controls (Veillonella) (P= 0.0008) (Fig. 3A). In contrast to cases, controls showed a variety of commensal bacteria typical for a healthy colon: Bifidobacterium, Prevotella, Streptococcus, Escherichia, Megamonas, Blautia, Bacteroides, Ruminococcus, Faecalibacterium and Megasphaera (Fig. 2A, 3A). Cases showed lower α (Fig. 2B) and within group β diversity than controls (Fig. 2C) suggesting a loss of diversity and convergence to a similar state. Cases also showed a decreased abundance (P< 10-5) of Prevotella, Bifidobacterium, Bacteroides, and Megamonas compared to controls (Fig. 2A, 3A). Escherichia abundance was increased in all diarrhea groups except RV patients (Fig. 2F, 3A), but overall did not reach statistical significance in cases over controls (Fig. 2F). No significant difference was seen for the stool microbiota composition with respect to Streptococcus and Escherichia genus between day 1 and day 2 of the hospitalized diarrhea patients (Fig. 2D, F). Abundance analysis based on 16S rRNA sequencing only indicates relative proportions of stool bacteria. qPCR determination of 16S rRNA genome equivalents per gram stool showed for controls a significant 8-fold higher titer than cases (Fig. 2G). As intestinal Streptococci (enriched in cases) have 25% more rRNA gene copies than Bifidobacterium or Prevotella (both enriched in controls) [5 copies versus 4 and 4 copies (Stoddard et al, 2015)], the difference in bacterial counts is approximatively 10 fold. Although cases produce an estimated 2-3 fold higher daily stool output (S. Sarker, unpublished observations), an about four-fold higher bacterial output per day in controls than in AD cases remains, representing a potentially very significant biological difference. Stool metagenome sequencing
7 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 8 of 47
Metagenome sequencing was conducted with 9 controls, 5 patients with EAEC as only pathogen, and 11 AD-p cases. Stool samples collected at the day of hospitalization were analyzed. The classification of the metagenome sequences showed a marked decrease (P= 0.0025) for Prevotella copri and Bifidobacterium (which affected more B. bifidum than B. breve and B. longum), for Megamonas, Megasphaera, and Lactobacillus ruminis in cases compared to controls (Fig. 4A), confirming and extending the 16S rRNA sequencing analysis. E. coli and Streptococcus belonging to the S. bovis complex (S. infantarius, S. lutetiensis, S. pasteurianus) and the S. salivarius complex (S. salivarius, S. vestibularis) were higher in cases than controls. Pathway analysis identified a higher mannan degradation capacity in controls attributed to Bacteroides (Table 3), which concurs with the observed decrease of Bacteroides in cases. Bacteroides is known to digest yeast mannan in the human gut (Cuskin et al., 2015). Controls are also enriched for methylerythritol phosphate pathway (contributed by B. longum in controls) and the biosynthesis of thiamin (contributed by Bacteroides in controls), two pathways which are prominent in the distal colon microbiome of healthy humans (Gill et al., 2006). Conversely, stool metagenomes from cases were enriched for a number of biosynthetic and degradation pathways without a common denominator (data not shown) including lactose and galactose degrading genes contributed by streptococci (Fig. 5A) and anaerobic and to a lesser extent aerobic respiration genes contributed by E. coli (Fig. 5B). Antibiotic resistance genes contributed by E. coli were increased in EAEC cases over controls, but not in AD-p (Fig. 5C). We also investigated bacterial genera for a higher ratio of origin-over-terminus DNA (OT ratio) suggesting actively replicating bacteria (Korem et al., 2015). Cases showed a significant higher OT ratio for Bifidobacterium over controls, but not for Streptococcus or Escherichia (Fig. 5D). 8 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 9 of 47
Virulence genes For the 11 AD-p and 5 EAEC cases and 9 controls, we screened the metagenome sequences for bacterial virulence genes with annotations of toxin, adherence or invasion characteristics (Chen et al., 2016) (Fig. 4C). Overall, cases showed a higher number of bacterial virulence genes than controls (P= 0.05) (except for Campylobacter jejuni). Virulence genes in cases were mainly associated with E. coli pathogens: fimbriae of uropathogenic CFT073, adhesins and toxins from enterohemorrhagic O157:H7 and EAEC O44:H18, pili from EPEC, followed by adherence and flagella genes from Salmonella, and Yersinia enterocolitica, respectively.
Virome analysis Elevated amounts of phage DNA sequences were detected in cases over controls (Fig. 4B, 6A). While significant, it is important to point out that there are very clearly only few patients that are skewing the data (Fig. 6A). Within the 5 EAEC patients, 2 showed high abundance of coliphage DNA, suggesting phage replication on the infecting E. coli strains (Fig. 4B, 6B). One EAEC patient contained large amounts of the rV5-like E. coli Myophage phAPEC8 (Tsonos et al., 2012), another EAEC patient showed large amounts of phiEco32-like E. coli Podophage (Savalia et al., 2008). Both phages are lytic and were found in patients with high fecal E. coli abundance (Fig. 6B) suggesting ongoing coliphage replication in two out of five EAEC patients. Within the 11 AD-p patients, four displayed phage sequences (Fig. 4B, 6B). One patient showed high amounts of a Sfi19-like Streptococcus Siphophage, another a moderately elevated amount of an Lj965-like Lactobacillus Siphophage. Both are temperate phages observed in presence of their host bacterium, suggesting the detection of prophages (no purification of virus-like particles was done). A third AD-p patient showed an pYD38-like vibriophage lacking genes for prophage integration, suggesting a lytic phage. A fourth AD-p patient displayed lytic T5-like phages. Within the 9 controls, 1 showed a lytic Bacillus and a 9 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 10 of 47
lytic Klebsiella phage. Small amounts of human viruses were found in 3 subjects (polyoma-, adeno- and bocavirus, the latter in a healthy control).
Discussion A survey of 71 Bangladeshi children hospitalized with acute watery diarrhea revealed a stool microbiota characterized by elevated levels of streptococci irrespective of etiology. Sequence analysis attributed the fecal streptococci to the S. bovis and S. salivarius species complexes. Similar observations were previously reported for adult and pediatric diarrhea patients from Bangladesh hospitalized with cholera (Hsiao et al., 2014) and E. coli infections (David et al., 2015; Sarker et al., 2016), Chinese children hospitalized with AD-p (Jin et al., 2013), as well as African children from Kenya, Mali and The Gambia hospitalized with AD and dysentery (Pop et al., 2015). The studies differed slightly in methods and resulting species annotation, Hsiao et al. (2014) annotated S. bovis and S. thermophilus, a subspecies of S. salivarius, and Pop et al. (2015) added S. mitis to these streptococci. Apparently, fecal abundance increase in streptococci represents a biomarker for AD independent of etiology and geography. Most groups observing fecal streptococci in diarrhea patients have not attributed a pathogenic role to these bacteria, except Jin et al. (2013), who suggested Streptococcus lutetiensis, a member of the S. bovis group, as potential new diarrhea pathogen based on the detection of purported virulence factors observed in genome sequencing of fecal isolates. This observation would fit with our observation that stool output in children with mostly ETEC diarrhea correlated with fecal Streptococcus, but not Escherichia abundance (Sarker et al., 2016). However, sequenced streptococcal isolates from these children did not show known bacterial virulence genes. While this does not exclude unknown virulence genes, one might search for alternative interpretations for the high streptococcal percentage in the stool of diarrhea patients. 10 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 11 of 47
Streptococci belonging to the S. bovis and S. salivarius species complexes were also found in control children albeit with lower proportion. Likewise, members of the S. bovis species complex were also described in healthy European adults as normal commensals of the small intestine (Booijink et al., 2010). S. bovis is also known as food microbe (Jans et al., 2016), while pathogenic isolates are rare (Danne et al., 2011). S. salivarius is a major oral commensal with low pathogenic potential, only associated with opportunistic infection (Chaffanel et al., 2015), but has also been reported as commensal of the small intestine (Mignolet et al., 2016). Recent studies have stressed the importance of getting from relative to absolute abundance estimates for the interpretation of the fecal microbiome (Stämmler et al., 2016; Vandeputte et al., 2017). In addition, intestinal transit time should also be taken into account. In diarrhea patients, transit times (as demonstrated by the increased stool frequency in the patients) are dramatically reduced and this should affect particularly the colon microbiota characterized by a slow transit. Diarrhea would thus not allow the outgrowth of colon bacteria to their normal population sizes (1011-1012 bacteria per gram of content) simply due to a lack of time and therefore skew the fecal microbiota to bacterial groups that are typically abundant in the small intestine. Some observations from this study speak for this interpretation: the stool of diarrhea patients showed a lower bacterial diversity typical for the small intestinal microbiota and a lower bacterial density than controls (although the difference was estimated to be about fourfold, while the ileal bacterial density is with 108 bacteria per gram of content much smaller than this 4-fold reduction). Therefore intestinal streptococci must have experienced a specific growth stimulation since other typical small intestinal bacteria like Veillonella and Bilophila did not show such an outgrowth. Glucose in ORS can stimulate rapid growth particularly of intestinal streptococci (van den Bogert et al., 2013) and might explain the high frequency of streptococci in diarrhea patients independent of etiology in this and the above-mentioned 11 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 12 of 47
studies since glucose-based ORS represents the WHO recommended treatment for acute diarrhea. One should, however, not rapidly reject the association of streptococci with diarrhea as reverse causation. Several classical diarrhea pathogens particularly some E. coli pathotypes have in epidemiological surveys low pathogenicity indices (Kotloff et al., 2013; Liu et al., 2014). Many diarrhea patients harbor several enteric pathogens and are thus candidates for polymicrobial infections. In addition, the pathobiont and the dysbiosis concept invokes that gut commensals can develop a pathogenic potential when increased in number (CerfBensussan et al., 2010; Kamada et al., 2013). Therefore there is the possibility that a marked frequency increase of gut commensals, being it streptococci or E. coli, could cause or contribute to diarrhea as seen in cases of antibiotic-associated diarrhea without Clostridium difficile diagnosis (Larcombe et al., 2016). There is another, clinical reason to ask about a potential role of intestinal streptococci increases in diarrhea. A recent Cochrane review concluded that glucose polymer (i.e. starch)based ORS resulted in a reduced first day diarrhea output and a reduced diarrhea duration compared to glucose-based ORS when diarrhea etiology was not considered (Gregorio et al., 2016). Furthermore, several (Alam et al., 1992; Molla et al., 1982, 1985; 1989), but not all (Hossain et al., 2003) clinical trials from Bangladesh showed a reduced stool output in patients with cholera treated with a starch- compared to a glucose-based ORS. Since intestinal streptococci cannot digest starch, polymer ORS might have its beneficial effect by suppressing the intestinal streptococcal outgrowth. Other effects might also explain the superior efficacy of polymer over glucose ORS (Kühn et al., 2014). To settle these points, microbiota studies in polymer ORS treated diarrhea patients are clearly warranted, as well as studies in patients treated with intravenous, but not oral rehydration.
12 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 13 of 47
Virome analysis in the stool is increasingly complementing gut microbiota analyses (Lim et al., 2015; Manrique et al., 2016) and starts to be applied to pathological gut conditions (Norman et al., 2015). The metagenome analysis revealed that phage-related sequences are enriched in the AD case samples compared to controls. This is quite an intriguing finding since two of the 5 EAEC-infected AD patients showed high abundance of virulent (lytic) coliphages concomitantly with an elevated frequency of their potential bacterial hosts. The biological relevance of these findings remains unclear because of the small sample numbers analyzed by metagenome sequencing in this study. In a previous study with AD patients showing -despite a diagnosed ETEC infection- only low fecal E. coli titers, no substantial coliphage titers were seen (Sarker et al., 2016). However, in malnourished AD patients who showed high fecal E. coli titers, 7 of 19 patients showed elevated titers of fecal coliphages (Kieser et al, 2018) suggesting in vivo coliphage replication when the host cell crosses a threshold titer.
We should mention limitations of the study. The group sizes were uneven with 20 healthy controls versus 71 AD individuals. This unevenness may have skewed the conclusions. In a large diarrhea hospital like icddr,b, no healthy controls present to the clinicians, obtaining time- and place-matched controls represent therefore logistic challenges, which explain also that cases and controls were not matched for all epidemiological parameters. However, these differences are relatively small, and the consistency of the observation of fecal streptococcal increases in diarrhea patients over many published studies (David et al., 2015; Hsiao et al., 2014; Jin et al., 2013; Pop et al., 2015; Sarker et al., 2016) makes it unlikely that we report a sampling bias. Although the methodologies based on sequencing provide relative quantification only, we made an attempt to evaluate the difference of absolute output of fecal streptococci per day between cases and controls. Transforming relative frequencies of 13 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 14 of 47
bacterial groups into absolute titers is however difficult by the used method: qPCR estimates with universal primers represent numerous difficulties (distinct efficiency of DNA extraction, primer fit, different rRNA gene copy numbers for different bacterial groups) and more reliable methods like flow cytometry (Vandeputte et al., 2017) and stool quantification are needed to obtain reliable absolute titers.
Diarrhea patients without an identified pathogen (AD-P) by standard clinical microbiology remain a challenge for the understanding of diarrhea pathogenesis. Using molecular tests we could detect pathogens in a number of AD-P patients. However, signals were low, sometimes not much higher than the highest titers for these pathogens found in control patients. In the case of C. jejuni even higher titers were found in controls than in diarrhea cases. This is a repeated observation at icddr,b (S. Sarker, unpublished observations) and one might question the role of C. jejuni as an enteropathogen in pediatric diarrhea patients from Bangladesh.
Overall, one gets the impression that widely shared concepts in diarrhea pathogenesis are challenged by a wealth of recent observations from children living in developing countries. Here, we raised attention on the difficulty to assign etiology despite the progress in diagnostic tools, on the potential impact of the carbohydrates supplied in the ORS, and on the commensal microbiota and their hypothetical conversion into pathobiont (streptococci dysbiosis) that may contribute to the diarrhea. Further detailed microbiome including virome analyses are clearly needed to clarify the situation.
14 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 15 of 47
Experimental Procedures Cohort recruitment Seventy-one boys and girls between 6 months and 2 years of age hospitalized at the Dhaka Hospital of the International Center for Diarrheal Diseases Research in Bangladesh (icddr,b) with a diagnosis of acute diarrhea were enrolled into an observational study during the winter season 2014/2015. Twenty healthy children were in the same time period recruited as controls from Nandipara, a peri-urban community of Dhaka (12 km from icddr,b hospital). The study was approved by the Ethical Review Committee of icddr,b as protocol #PR-14081 and a written informed consent was obtained from the parents or caregiver of the children. Assessments in the hospital All subjects underwent a physical examination and nutritional assessment (including past and present breastfeeding and supplementary food). Socioeconomic and medical data were collected at anamnesis as part of the standard process. Low income was defined as less than 8000 Takas per month. Home medication including antibiotics was documented for the month preceding hospitalization. AD patients received standard care consisting of reduced glucose ORS supplemented with zinc, but no antibiotics. Antibiotics are at icddr,b only used in dysentery and cholera patients or patients with other co morbidities e.g., acute respiratory tract infection, severe acute malnutrition (< -3WHZ) or sepsis. Two stool samples were obtained for each AD patient at day 1 and day 2 of hospitalization. Pathogen detection Fresh stool samples obtained on day 1 of hospitalization were used for pathogen detection. Rotavirus (RV) was detected by ELISA, Vibrio, Salmonella, Shigella, Aeromonas, and Campylobacter by culture and Vibrio cholerae was confirmed by dark field microscopy;
15 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 16 of 47
ETEC, EPEC, EAEC, enterohemorrhagic (EHEC) and Shiga-toxin producing E. coli (STEC) were diagnosed by multiplex PCR after culture (Harris et al., 2008). Total nucleic acids from stool samples of AD patients without identified pathogen (QIAamp MinElute Virus Spin Kit, Qiagen) was investigated by TaqMan Array Card for 19 different enteropathogens (Liu et al., 2016) : Rota-, Adeno-, Astro-, Noro- and Sapovirus; Vibrio, Salmonella, Shigella, Aeromonas, Campylobacter, Clostridium difficile, ETEC, EPEC, EAEC; Cryptosporidium, Giardia, Entamoeba histolytica, Ascaris, Trichuris. The results are expressed as semi-quantitative cycle threshold (CT) values for each pathogen. We normalized the pathogen PCR signal to the signal for total bacterial count based on 16S rRNA gene CT.
Fecal DNA extraction, sequencing of 16S rRNA genes, and 16S rRNA quantification by qPCR Total DNA was extracted using the QIAamp DNA Stool Mini Kit (QIAGEN), following the manufacturer's instructions, except for the addition of a series of mechanical disruption steps (4 × 60 s) using a FastPrep apparatus and Lysing Matrix B tubes (MP Biochemicals). Then, the 16S rRNA gene variable regions V3 to V4 were PCR amplified using universal primers (Klindworth et al., 2013) with dual indexing (Kozich et al., 2013) and sequenced with Miseq reagent kit V3 (Illumina Inc., San Diego, California) as previously described (Caporaso et al., 2012). Total bacterial load (expressed as average genome equivalent) in the stool samples was determined by real-time PCR as described by Nadkarni et al. (2002) using E.coli K12 strain for the standard curve and an average bacterial genome weight of 3.34 x 10-15 g to calculate the average genome equivalent. 16S rRNA gene data analysis
16 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 17 of 47
Raw sequence data were analyzed using Mothur V.1.33.0 21 (Schloss et al., 2009) and QIIME V.1.8 22 (Caporaso et al., 2010) software packages. Paired-end sequences were demultiplexed and joined as described (Kozich et al., 2013). Then, sequences were split in separated fasta files for each sample using Mothur commands [deunique.seqs(), degap.seqs(), and split.groups()]. Conversion to QIIME format using add_qiime_labels.py and subsequent analytical steps were performed in QIIME. Chimera were identified de novo and against Greengenes V13.8 with usearch61. Using a new approach (see below), sequences of the genera Bifidobacterium, Lactobacillus and Streptococcus were annotated to the species level and enumerated. Standard OTUs picking at 97% identity was performed on the remaining sequences using pick_otus.py, with options usearch_ref (Edgar et al., 2011), optimal uclust, and without additional chimera detection. Taxonomy assignment was performed on representative sequences using RDP Classifier (Wang et al., 2007). After merging the results of both picking approaches, OTU and species representative sequences were aligned using PyNast method (Caporaso et al. 2010) and Uclust as pairwise alignment method. The resulting multiple alignments was then filtered and used to build a phylogenetic tree with the FastTree method (Price et al., 2009). After quality filtering of the OTUs (Bokulich et al., 2013) diversity analyses were performed in QIIME. Species level analysis of the 16S rRNA gene belonging to the genera Bifidobacterium, Lactobacillus, and Streptococcus We first defined species and subspecies-level reference signatures on the 16S rRNA gene sequences. Sequences from isolates assigned to species from Bifidobacterium, Lactobacillus, and Streptococcus genera were retrieved from the RDP v10 and Silva v119 databases. Subsequent steps were performed separately for each genus. Using Mothur, all good quality sequences of >1200 bp corresponding to isolates of the species of interest and to all other type strains of the genus were aligned on the Silva reference v119. To increase the quality of our 17 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 18 of 47
reference dataset, clear outlier sequences within each species were removed based on phylogenic trees analyses. The signatures were determined as minimal combinations of nucleotides at defined positions on the alignment which were conserved within species, but variable among species. In some instances, closely related species could not be distinguished based on the V3 to V4 16S rRNA gene region targeted by primers for Illumina sequencing. In these cases, they were pooled as groups of species identified by common signatures. Signatures are reported in Suppl. Table 1 for Bifidobacterium species, Suppl. Table 2 for Lactobacillus species, and Suppl. Table 3 for Streptococcus species. The accuracy of the approach was assessed with annotated sequences from RDP and Silva databases, and from well-characterized sequences from isolates of the Nestlé Culture Collection. To analyze the Illumina sequencing datasets using our species-level reference signatures, chimera checked 16S rRNA sequences were classified at genus level with the RDP Classifier v10 (Wang et al., 2007) implementation in Mothur. Sequences annotated as Bifidobacterium, Lactobacillus, or Streptococcus, were then aligned on the Silva reference v119. Based on the combinations of nucleotides searched in our species-level reference signatures, each assembled Illumina sequence was annotated. Sequences unassigned at species level kept their genus-level annotation (with sp. suffix). Metagenomics Twenty five stool samples were investigated with shotgun metagenomics. We randomly selected 11 of the 15 AD-p patients for metagenome sequencing because we were particularly interested to identify potential pathogens in AD patients without enteropathogens diagnosed by standard procedures of the clinical microbiology laboratory (AD-p). Nine samples were randomly selected among the 20 healthy controls. As a comparator, we used five samples from AD patients with EAEC as the only detected pathogen.
18 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 19 of 47
Stool DNA was extracted using MoBio PowerMag Microbiome DNA Isolation Kit on an epMotion M5073 (Vaudaux-Eppendorf AG, Basel) followed by Zymo DNA Clean & Concentrator Kit. Library preparation was done according to the Nextera XT protocol from Illumina. The quality and quantity check was based on LabChip GX Touch HT (Perkin Elmer) results. Sequencing was performed on HiSeq 2500 using chemistry HighOutput v4 PE125. The paired-end reads were filtered using KneadData v0.5.1 (https://bitbucket.org/biobakery/kneaddata), which included quality filtering based on Trimmomatic and excluded reads mapping to the human genome. Taxonomic profiles were generated with MetaPhlAn2 2.5.0 (Truong et al., 2015). Functional annotation was performed with HUMANn2 v0.7.1 (Abubucker et al., 2012) and integrated into pathways from the MetaCyc database (Caspi et al., 2014). The mapped reads are normalized first by gene length, then by million reads (counts per million). Antibiotic resistance genes were retrieved from CARD database (McArthur et al., 2013). Virulence factors were retrieved from VFDB database (Chen et al., 2016) using ShortBRED (Kaminski et al., 2015). The mapped reads were normalized (first by million mapped reads and then by kilobase of reference database, RPKM). Growth rate estimates of origin-toterminus-ratio were produced according to the pipeline described by Korem et al. (2015). Data availability. 16S rDNA and metagenome reads are available under the Bio Project accessions SRP100410, SRP100895, SRP119744, and SRP120020. Statistics As microbiota abundance have a non-normal distribution, non-parametric test were used. If not mentioned otherwise, the two-sided Mann-Whitney-test was used for continuous variables and a χ2- test for categorical variables. 19 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 20 of 47
References Abubucker S, Segata N, Goll J, et al. Metabolic Reconstruction for Metagenomic Data and Its Application to the Human Microbiome Eisen JA, ed. PLoS Comput Biol 2012;8:e1002358. Ahmed S, Byrd W, Kumar S, et al. A directed intimin insertion mutant of a rabbit enteropathogenic Escherichia coli (REPEC) is attenuated, immunogenic and elicits serogroup specific protection. Vet Immunol Immunopathol 2013;152:146–155. Alam NH, Ahmed T, Khatun M, et al. Effects of food with two oral rehydration therapies: a randomised controlled clinical trial. Gut 1992;33:560–562. Barletta F, Ochoa TJ, Mercado E, et al. Quantitative Real-time Polymerase Chain Reaction for Enteropathogenic Escherichia coli: A Tool for Investigation of Asymptomatic Versus Symptomatic Infections. Clin Infect Dis 2011;53:1223–1229. Black RE, Cousens S, Johnson HL, et al. Global, regional, and national causes of child mortality in 2008: a systematic analysis. Lancet Lond Engl 2010;375:1969–1987. Bokulich NA, Subramanian S, Faith JJ, et al. Quality-filtering vastly improves diversity estimates from Illumina amplicon sequencing. Nat Methods 2013;10:57–59. Booijink CCGM, El-Aidy S, Rajilić-Stojanović M, et al. High temporal and inter-individual variation detected in the human ileal microbiota. Environ Microbiol 2010;12:3213– 3227. Breiman RF, Zaman K, Armah G, et al. Analyses of health outcomes from the 5 sites participating in the Africa and Asia clinical efficacy trials of the oral pentavalent rotavirus vaccine. Vaccine 2012;30 Suppl 1:A24-29. 20 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 21 of 47
Caporaso JG, Bittinger K, Bushman FD, et al. PyNAST: a flexible tool for aligning sequences to a template alignment. Bioinformatics 2010;26:266–267. Caporaso JG, Lauber CL, Walters WA, et al. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J 2012;6:1621–1624. Caporaso JG, Kuczynski J, Stombaugh J, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods 2010;7:335–336. Caspi R, Altman T, Billington R, et al. The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of Pathway/Genome Databases. Nucleic Acids Res 2014;42:D459–D471. Chaffanel F, Charron-Bourgoin F, Libante V, et al. Resistance Genes and Genetic Elements Associated with Antibiotic Resistance in Clinical and Commensal Isolates of Streptococcus salivarius Schaffner DW, ed. Appl Environ Microbiol 2015;81:4155– 4163. Chen L, Zheng D, Liu B, et al. VFDB 2016: hierarchical and refined dataset for big data analysis—10 years on. Nucleic Acids Res 2016;44:D694–D697. Cuskin F, Lowe EC, Temple MJ, et al. Human gut Bacteroidetes can utilize yeast mannan through a selfish mechanism. Nature 2015;517:165–169. Danne C, Entenza JM, Mallet A, et al. Molecular Characterization of a Streptococcus gallolyticus Genomic Island Encoding a Pilus Involved in Endocarditis. J Infect Dis 2011;204:1960–1970. David LA, Weil A, Ryan ET, et al. Gut Microbial Succession Follows Acute Secretory Diarrhea in Humans. mBio 2015;6:e00381-15. 21 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 22 of 47
Edgar RC, Haas BJ, Clemente JC, et al. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 2011;27:2194–2200. Gao XY, Zhi XY, Li HW, Klenk HP, Li WJ. Comparative genomics of the bacterial genus Streptococcus illuminates evolutionary implications of species groups. PLoS One. 2014 9:e101229. Gill SR, Pop M, DeBoy RT, et al. Metagenomic Analysis of the Human Distal Gut Microbiome. Science 2006;312:1355–1359. Gregorio GV, Gonzales MLM, Dans LF, et al. Polymer-based oral rehydration solution for treating acute watery diarrhoea. In: The Cochrane Collaboration, ed. Cochrane Database of Systematic Reviews. Chichester, UK: John Wiley & Sons, Ltd; 2016. Harris AM, Chowdhury F, Begum YA, et al. Shifting prevalence of major diarrheal pathogens in patients seeking hospital care during floods in 1998, 2004, and 2007 in Dhaka, Bangladesh. Am J Trop Med Hyg 2008;79:708–714. Hossain MS, Salam MA, Rabbani GH, et al. Rice-ORS versus glucose-ORS in management of severe cholera due to Vibrio cholerae O139 Bengal: a randomized, controlled clinical trial. J Health Popul Nutr 2003;21:325–331. Hsiao A, Ahmed AMS, Subramanian S, et al. Members of the human gut microbiota involved in recovery from Vibrio cholerae infection. Nature 2014;515:423–426. Jans C, Wouters T de, Bonfoh B, et al. Phylogenetic, epidemiological and functional analyses of the Streptococcus bovis/Streptococcus equinus complex through an overarching MLST scheme. BMC Microbiol 2016;16.
22 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 23 of 47
Jiang Z, Lowe B, Verenkar MP, et al. Prevalence of Enteric Pathogens among International Travelers with Diarrhea Acquired in Kenya (Mombasa), India (Goa), or Jamaica (Montego Bay). J Infect Dis 2002;185:497–502. Jin D, Chen C, Li L, et al. Dynamics of fecal microbial communities in children with diarrhea of unknown etiology and genomic analysis of associated Streptococcus lutetiensis. BMC Microbiol 2013;13:141. Kaminski J, Gibson MK, Franzosa EA, et al. High-Specificity Targeted Functional Profiling in Microbial Communities with ShortBRED Noble WS, ed. PLOS Comput Biol 2015;11:e1004557. Kieser S, Sarker SA, Berger B, Sultana S, Khan Z, Islam SB, et al. Antibiotic treatment leads to fecal Escherichia coli and coliphage expansion in severely malnourished diarrhea patients. Cell Mol Gastroenterol Hepatol 2018, 5: 458-460.e6. Klindworth A, Pruesse E, Schweer T, et al. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res 2013;41:e1. Korem T, Zeevi D, Suez J, et al. Growth dynamics of gut microbiota in health and disease inferred from single metagenomic samples. Science 2015;349:1101–1106. Kotloff KL, Nataro JP, Blackwelder WC, et al. Burden and aetiology of diarrhoeal disease in infants and young children in developing countries (the Global Enteric Multicenter Study, GEMS): a prospective, case-control study. Lancet Lond Engl 2013;382:209–222.
23 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 24 of 47
Kozich JJ, Westcott SL, Baxter NT, et al. Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Appl Environ Microbiol 2013;79:5112–5120. Kühn J, Finger F, Bertuzzo E, et al. Glucose- but Not Rice-Based Oral Rehydration Therapy Enhances the Production of Virulence Determinants in the Human Pathogen Vibrio cholerae Small PLC, ed. PLoS Negl Trop Dis 2014;8:e3347. Larcombe S, Hutton ML, Lyras D. Involvement of Bacteria Other Than Clostridium difficile in Antibiotic-Associated Diarrhoea. Trends Microbiol 2016;24:463–476. Lim ES, Zhou Y, Zhao G, et al. Early life dynamics of the human gut virome and bacterial microbiome in infants. Nat Med 2015;21:1228–1234. Liu J, Kabir F, Manneh J, et al. Development and assessment of molecular diagnostic tests for 15 enteropathogens causing childhood diarrhoea: a multicentre study. Lancet Infect Dis 2014;14:716–724. Liu J, Platts-Mills JA, Juma J, et al. Use of quantitative molecular diagnostic methods to identify causes of diarrhoea in children: a reanalysis of the GEMS case-control study. The Lancet 2016;388:1291–1301. Manrique P, Bolduc B, Walk ST, et al. Healthy human gut phageome. Proc Natl Acad Sci 2016;113:10400–10405. McArthur AG, Waglechner N, Nizam F, et al. The Comprehensive Antibiotic Resistance Database. Antimicrob Agents Chemother 2013;57:3348–3357.
24 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 25 of 47
McCallin S, Alam Sarker S, Barretto C, et al. Safety analysis of a Russian phage cocktail: from metagenomic analysis to oral application in healthy human subjects. Virology 2013;443:187–196. Mignolet J, Fontaine L, Kleerebezem M, et al. Complete Genome Sequence of Streptococcus salivarius HSISS4, a Human Commensal Bacterium Highly Prevalent in the Digestive Tract. Genome Announc 2016;4:e01637-15. Molla AM, Sarker SA, Hossain M, Molla A, Greenough WB 3rd. Rice-powder electrolyte solution as oral-therapy in diarrhoea due to Vibrio cholerae and Escherichia coli. Lancet. 1982 Jun 12;1(8285):1317-9 Molla AM, Ahmed SM, Greenough WB. Rice-based oral rehydration solution decreases the stool volume in acute diarrhoea. Bull World Health Organ 1985;63:751–756. Molla AM, Molla A, Rohde J, et al. Turning off the diarrhea: the role of food and ORS. J Pediatr Gastroenterol Nutr 1989;8:81–84. Munos MK, Walker CLF, Black RE. The effect of oral rehydration solution and recommended home fluids on diarrhoea mortality. Int J Epidemiol 2010;39:i75–i87. Nadkarni MA, Hunter N, Jacques NA, et al. Determination of bacterial load by real-time PCR using a broad-range (universal) probe and primers set. Microbiology 2002;148:257– 266. Norman JM, Handley SA, Baldridge MT, et al. Disease-Specific Alterations in the Enteric Virome in Inflammatory Bowel Disease. Cell 2015;160:447–460. Póntigo F, Moraga M, Flores SV. Molecular phylogeny and a taxonomic proposal for the genus Streptococcus. Genet Mol Res. 2015 14:10905-18. 25 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 26 of 47
Pop M, Walker AW, Paulson J, et al. Diarrhea in young children from low-income countries leads to large-scale alterations in intestinal microbiota composition. Genome Biol 2014;15:R76. Price MN, Dehal PS, Arkin AP. FastTree: computing large minimum evolution trees with profiles instead of a distance matrix. Mol Biol Evol 2009;26:1641–1650. Qadri F, Svennerholm A-M, Faruque ASG, et al. Enterotoxigenic Escherichia coli in Developing Countries: Epidemiology, Microbiology, Clinical Features, Treatment, and Prevention. Clin Microbiol Rev 2005;18:465–483. Sarker SA, Brüssow H. From bench to bed and back again: phage therapy of childhood Escherichia coli diarrhea. Ann N Y Acad Sci 2016;1372:42–52. Sarker SA, Berger B, Deng Y, et al. Oral application of E scherichia coli bacteriophage: safety tests in healthy and diarrheal children from Bangladesh: Oral application of E. coli bacteriophage. Environ Microbiol 2017;19:237–250. Sarker SA, Sultana S, Reuteler G, et al. Oral Phage Therapy of Acute Bacterial Diarrhea With Two Coliphage Preparations: A Randomized Trial in Children From Bangladesh. EBioMedicine 2016;4:124–137. Savalia D, Westblade LF, Goel M, et al. Genomic and Proteomic Analysis of phiEco32, a Novel Escherichia coli Bacteriophage. J Mol Biol 2008;377:774–789. Schloss PD, Westcott SL, Ryabin T, et al. Introducing mothur: open-source, platformindependent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol 2009;75:7537–7541.
26 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 27 of 47
Stämmler F, Gläsner J, Hiergeist A, Holler E, Weber D, Oefner PJ, Gessner A, Spang R. Adjusting microbiome profiles for differences in microbial load by spike-in bacteria. Microbiome. 2016 Jun 21;4(1):28. Stoddard S.F, Smith B.J., Hein R., Roller B.R.K. and Schmidt T.M. (2015) rrnDB: improved tools for interpreting rRNA gene abundance in bacteria and archaea and a new foundation for future development. Nucleic Acids Research 2014; doi: 10.1093/nar/gku1201 Taniuchi M, Sobuz SU, Begum S, et al. Etiology of Diarrhea in Bangladeshi Infants in the First Year of Life Analyzed Using Molecular Methods. J Infect Dis 2013;208:1794– 1802. Truong DT, Franzosa EA, Tickle TL, et al. MetaPhlAn2 for enhanced metagenomic taxonomic profiling. Nat Methods 2015;12:902–903. Tsonos J, Adriaenssens EM, Klumpp J, et al. Complete Genome Sequence of the Novel Escherichia coli Phage phAPEC8. J Virol 2012;86:13117–13118. Van den Bogert B, Boekhorst J, Herrmann R, et al. Comparative Genomics Analysis of Streptococcus Isolates from the Human Small Intestine Reveals their Adaptation to a Highly Dynamic Ecosystem Schaik W van, ed. PLoS ONE 2013;8:e83418. Vandeputte D, Kathagen G, D'hoe K, Vieira-Silva S, Valles-Colomer M, Sabino J, et al. Quantitative microbiome profiling links gut community variation to microbial load. Nature 2017 551:507-511. Walker CLF, Black RE. Zinc for the treatment of diarrhoea: effect on diarrhoea morbidity, mortality and incidence of future episodes. Int J Epidemiol 2010;39:i63–i69.
27 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 28 of 47
Wang Q, Garrity GM, Tiedje JM, et al. Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol 2007;73:5261–5267. Wiggins BA, Alexander M. Minimum bacterial density for bacteriophage replication: implications for significance of bacteriophages in natural ecosystems. Appl Environ Microbiol 1985;49:19–23.
28 Wiley-Blackwell and Society for Applied Microbiology This article is protected by copyright. All rights reserved.
Page 29 of 47
Table 1 Clinical comparison of acute diarrhea patients (AD) with healthy controls (HC)
Weight(kg) Height(cm) Mid upper arm circumference(cm) Weight for age Z score Height for age Z score Body mass index Weight for height Z score BMI Z score MUAC_Z Temperature/°C Pulse/min Respiration Rate /min Vomiting/ day Diarrhea at home(days) Stool frequency/ day Systolic Blood Pressure(mmHg) Diastolic Blood Pressure(mmHg) Exclusive breastfeeding(months) Diarrhea in hospital( days)
AD (N=71) Mean std median
HC (N=20) mean std median
8.2 71.3 14.1
1.4 5.7 1.2
7.8 70.2 14.0
8.7 73.5 13.6
1.3 5.1 0.7
8.4 73.5 13.3
NS 0.09 0.02
-1.0 -0.9 16.0 -0.6
1.0 1.3 1.7 1.2
-1.1 -0.9 15.7 -0.9
-0.9 -1.1 16.1 -0.4
1.1 1.3 1.5 1.0
-1.2 -1.4 16.0 -0.5
NS NS NS NS
-0.6 -0.3 37.0 136.5 35.6 1.9 2.2
1.2 1.1 0.7 4.9 2.0 2.8 1.3
-1.0 -0.4 37.0 138.0 36.0 1.0 2.0
-0.3 -0.8 36.3 109.0 30.4
1.0 0.6 0.5 9.1 1.5
-0.3 -0.9 36.5 110.0 30.0