Molecular Analysis of Gut Microbiota in Obesity Among Indian Individuals

Molecular analysis of gut microbiota in obesity among Indian individuals

1,† 1 1 1 2 DEEPAK PPATIL , DHIRAJ PDHOTRE , SACHIN GCHAVAN , ARMIYA SULTAN , DHAWAL SJAIN , 3 4 4 4 VIKRAM BLANJEKAR , JAYSHREE GANGAWANI , POONAM SSHAH , JAYSHREE STODKAR , 4 3 1 1, SHASHANK SHAH , DILIP RRANADE , MILIND SPATOLE and YOGESH SSHOUCHE * 1National Centre for Cell Science, Ganeshkhind, Pune 411 007, India 2Institute of Bioinformatics and Biotechnology, University of Pune, Pune 411 007, India 3Microbial Sciences Division, Agharkar Research Institute, Pune 411 004, India 4Laparo Obeso Centre, Ruby Hall Clinic, Sassoon Road, Pune 411 001, India

*Corresponding author (Fax, +91-20-25692259; Email, [email protected]) †Present address: Center for RNA biology, The Ohio State University, Columbus, OH 43202,USA

Obesity is a consequence of a complex interplay between the host genome and the prevalent obesogenic factors among the modern communities. The role of gut microbiota in the pathogenesis of the disorder was recently discovered; however, 16S-rRNA-based surveys revealed compelling but community-specific data. Considering this, despite unique diets, dietary habits and an uprising trend in obesity, the Indian counterparts are poorly studied. Here, we report a comparative analysis and quantification of dominant gut microbiota of lean, normal, obese and surgically treated obese individuals of Indian origin. Representative gut microbial diversity was assessed by sequencing fecal 16S rRNA libraries for each group (n=5) with a total of over 3000 sequences. We detected no evident trend in the distribution of the predominant bacterial phyla, Bacteroidetes and Firmicutes. At the genus level, the bacteria of genus Bacteroides were prominent among the obese individuals, which was further confirmed by qPCR (P<0.05). In addition, a remarkably high archaeal density with elevated fecal SCFA levels was also noted in the obese group. On the contrary, the treated-obese individuals exhibited comparatively reduced Bacteroides and archaeal counts along with reduced fecal SCFAs. In conclusion, the study successfully identified a representative microbial diversity in the Indian subjects and demonstrated the prominence of certain bacterial groups in obese individuals; nevertheless, further studies are essential to understand their role in obesity.

[Patil DP, Dhotre DP, Chavan SG, Sultan A, Jain DS, Lanjekar VB, Gangawani J, Shah PS, Todkar JS, Shah S, Ranade DR, Patole MS and Shouche YS 2012 Molecular analysis of gut microbiota in obesity among Indian individuals. J. Biosci. 37 647–657] DOI 10.1007/s12038-012-9244-0

1. Introduction cannot be ignored (Hill and Peters 1998). With surging numbers, obesity together with the related disorders such Worldwide, approximately 1.6 billion adults are overweight as atherosclerosis, hypertension, type 2 diabetes and and 500 million adults are obese. This epidemic is not just cancer, have become a burden to mankind (Thompson prevalent among the developed nations; developing nations and Wolf 2001; Haslam and James 2005; Shamseddeen such as India are also no exception, where the disease is et al. 2011). more endemic to the metropolitan cities (Deepa et al. 2009). Fundamentally, obesity is a metabolic disorder caused by Cause for such sudden rise in these incidences appears uncontrolled food intake accompanied by a sedentary life- complex and multifactorial (Cummings and Schwartz style. Dietary restriction and exercise are the preliminary 2003); however, a role of the widespread obesogenic factors therapies of choice. Because of their poor performance and

Keywords. Gut microbiota; Indian; microbial diversity; obesity http://www.ias.ac.in/jbiosci J. Biosci. 37(4), September 2012, 647–657, * Indian Academy of Sciences 647

Published online: 14 August 2012 648 Deepak P Patil et al. delayed results, invasive bariatric surgery is the highly surgically treated group includes individuals who underwent sought treatment for morbid or severe obesity (Choban et sleeve gastrectomy (SG) and adjustable gastric banding (AGB) al. 2002). As the human gut hormonally controls food intake surgeries as a treatment for obesity. These two forms of re- (Murphy and Bloom 2006), these surgeries intend to alter the strictive bariatric surgeries effectively reduce the stomach size, gut structure, thereby limiting the caloric intake and, in turn, leading to an early satiety and reduced caloric intake without weight gain. Nonetheless, despite high success rates, bariat- affecting absorption of the digested food (Schneider and Mun ric surgeries are expensive (Terranova et al. 2012) and most 2005). After identification of dominant bacterial groups of them permanently alter the gut anatomy. This necessitates using 16S rRNA gene libraries, we confirmed their promi- development of effective and noninvasive therapeutic nence by real-time PCR across the 20 subjects. Estimation of approaches, primarily considering the other vital functions fecal short chain fatty acids (SCFAs) was also carried out in of the gut, such as, regulation of immunity, stimulation of parallel to demonstrate the differences in the fermentation angiogenesis, intestinal cell proliferation and differentiation, efficiencies of the obese and treated-obese microbiome. inflammatory immune response, and vitamin supplementa- tion (Stevens and Hume 1998; Rautava and Isolauri 2002; 2. Materials and methods Stappenbeck et al. 2002; Noverr and Huffnagle 2004; Rakoff-Nahoum et al. 2004). 2.1 Collection of samples Apart from the above-mentioned functions, the human gut also harbours trillions of bacteria, which affect our health and physiology to an inconceivable extent. Discovery of Early morning fecal samples were collected from unrelated healthy individuals (21–62 years old) irrespective of gender. association of gut bacteria with fat storage and obesity These were classified based on their body-mass-index (BMI) through sequencing of 16S rRNA libraries is one such ex- into lean (<19 kg/m2, n=5, median age=23 years), normal ample (Backhed et al. 2004; Ley et al. 2005). Obese gut (18–24 kg/m2,n=5,medianage=44years),obese(25– microbiome exhibits an increased transmissible efficiency in 53 kg/m2, n=5, median age=45 years), and treated obese harvesting energy from the diet leading to an obese pheno- (25–36 kg/m2, n=5, median age=50 years, these are obese type; however, no single species of bacteria is identified to individuals regressing to normal BMI after SG and AGB vary with weight gain (Turnbaugh et al. 2006). In a diet- surgeries). For details refer table 1. Apart from BMI to based weight-loss model, analysis of gut microbiota over a confirm leanness among lean subjects, whole body fat per- period of time revealed an increase in the bacteria of phylum centage was also determined using OMRON HBF-306C Bacteroidetes with a significant reduction in Firmicutes (Ley hand-held body fat analyser (Omron healthcare inc., USA) et al. 2006), although researchers have failed to detect such as in reference (Lintsi et al. 2004). No participant used any change in another community (Duncan et al. 2008). A sim- antibiotic, pre- or probiotics within the preceding 3 months ilar study conducted elsewhere (Nadal et al. 2009) reported before sampling. All the fecal samples were stored at 4°C and no change in the Bacteroidetes-to-Firmicutes ratio but transported to laboratory on ice. The samples were either reported an increase in the Bacteroides-Prevotella group. processed immediately or stored at −80°C until processed. In a pregnancy model of obesity, weight gain during preg- The study was approved by the Ruby Hall Clinic Ethical nancy correlates with an increase in bacteria of genus Bac- Committee, Pune, India. Accordingly, the participating lab teroides in women (Collado et al. 2008). A recent study (Laparo Obeso Centre) at Ruby Hall Clinic collected the fecal (Zhang et al. 2009) described the fecal microbiota of patients samples from the volunteers after a verbal consent. with post-gastric bypass anti-obesity surgery. These individuals exhibited a high Firmicutes count along with reduced 2.2 DNA extraction archaeal densities in comparison with the untreated obese. With diverse models and community-specific reports, explo- Total DNA was extracted from each stool sample using ration of the bacterial pathogenesis of obesity remains a QIAamp DNA Stool Mini Kit (Qiagen) with an additional step daunting task; nevertheless, preliminary identification of of bead beating using a mix of silica beads (diameter 0.1, 0.2 gut bacteria remains a cardinal step. and 0.5 mm). Purified DNA samples were qualitatively veri- Obesity-associated changes in bacteria are largely reported fied by agarose gel electrophoresis for integrity and concen- from the Western population (Ley et al. 2006; Duncan et al. trations were estimated by Nanodrop (Thermo Scientific). 2008; Nadal et al. 2009); however, despite unique dietary habits (Sinha et al. 2003), the Indian counterparts are significantly underexplored. Here, we report for the first time, an 2.3 Generation of 16S rRNA gene library and sequencing analysis and comparison of gut microbiota by sequencing 16S rRNA gene libraries from lean, normal, obese and Fecal 16S rRNA libraries were generated for each individual surgically treated obese individuals of Indian origin. The separately. Five replicates of 25 μl PCRs, each containing

J. Biosci. 37(4), September 2012 Gut microbiota in obesity in Indian individuals 649

Table 1. Details of participating individuals understudy

Sample Gender and age BMI before BMI at the time of sample Body fat Surgical Time required for weight Serial No. IDa (in years)b surgery (kg/m2)c collection (kg/m2) percentaged treatmente loss (in days)f

1 L1 M27 - 17.54 17.80 - - 2 L2 M22 - 18.09 18.70 - - 3 L3 F21 - 16.03 21.66 - - 4 L4 M23 - 15.82 12.04 - - 5 L5 F24 - 14.81 19.06 - - 6 N1 M49 - 24.00 - - - 7 N2 M45 - 23.00 - - - 8 N3 M44 - 23.00 - - - 9 N4 M28 - 24.80 - - - 10 N5 F25 - 23.00 - - - 11 O1 F42 - 44.10 - - - 12 O2 F45 - 52.80 - - - 13 O3 M41 - 51.20 - - - 14 O4 M50 - 35.00 - - - 15 O5 M49 - 40.00 - - - 16 T1 F30 37.65 32.05 - SG 93 17 T2 M42 40.68 28.09 - SG 196 18 T3 F50 39.35 34.48 - SG 104 19 T4 M62 45.18 35.86 - AGB 371 20 T5 F52 36.12 27.68 - AGB 278 a Sample ID, where L = Lean, N = Normal, O = Obese, T = Surgically treated-Obese. b M = Male, F = Female; the following two-digit number is age of the participant in years. c BMI before surgery, applicable to obese individuals, who underwent anti-obesity surgeries. d Body fat percentage was calculated using Omron body fat analyser. e,f Only applicable to surgically treated obese; AGB = adjustable gastric banding, SG = sleeve gastrectomy.

25 ng DNA, 1X Thermopol Buffer (NEB), 1 unit of Taq Sequences were converted to sense orientation using Orien- Polymerase (NEB), 200 μM dNTPs, 200 μM concentration tationChecker (Ashelford et al. 2006), aligned in NAST of 8 F-I (5′-GGA TCC AGA CTT TGA TYM TGG CTC aligner (DeSantis Jr. et al. 2006) and screened for chimeras AI-3′) and 907R-I (5′-CCG TCA ATT CMT TTG AGT TI- with Mallard (Ashelford et al. 2006). Non-chimeric sequen- 3′) (Ben-Dov et al. 2006). Cycling conditions were 95°C for ces above 300 bases were analysed for nearest neighbours 3 min, followed by 20 cycles of 95°C for 10 s, 53°C for 45 s, using SEQMATCH tool at RDP-II (version 10.9) (Cole et al. and 72°C for 1 min, with a final extension of 20 min at 72°C. 2005) database. For estimation of phylotype or operational Replicate PCRs were pooled together. The resulting 900 bp taxonomic unit (OTU) richness and microbial diversity, dis- amplicon was gel-purified using the Qiagen extraction kit and tance matrix was generated using DNADIST from Phylip cloned into pGEM-T vector, followed by transformation into E. 3.68 package (Felsenstein J (1989) PHYLIP - Phylogeny coli DH5α. Extraction controls (no fecal material added) did Inference Package (Version 3.2). Cladistics 5: 164-166), not produce detectable PCR products or white/positive and OTUs were determined at 97% sequence similarity by colonies. For each sample, 200–400 representative colonies the furthest neighbour method in DOTUR 1.53 (Schloss and with cloned amplicons were bidirectionally sequenced using Handelsman 2005). Non-parametric species richness esti- BigDye™ Terminator Cycle Sequencing Ready Reaction mates were also determined using the abundance-based cov- Kit v3.1 in an automated 3730 DNA analyser (ABI). erage estimator (ACE) and Chao I estimator at a distance of 0.03. Library comparisons were performed using Uni- frac analysis (Lozupone and Knight 2005). A Genus-based 2.4 Sequence analysis distribution heatmap was constructed using R (R Foundation for Statistical Computing, Vienna, Austria, 2011.). All the Base-calling for each sequence read was performed using non-chimeric sequences were deposited to NCBI’s GenBank PHRED (Ewing et al. 1998). Each clonal sequence was database (Accession No. GU954553-GU957822). Phyloge- assembled using CAP3 from forward and reverse sequences. netic relationship between sequences was determined by the

J. Biosci. 37(4), September 2012 650 Deepak P Patil et al. neighbour-joining method. Neighbour-joining tree was 3. Results and discussion constructed with 1000 bootstrap and Kimura-2 parameter as a model of nucleotide substitution in MEGA4 3.1 Study design and the groups under study (Tamura et al. 2007). In this study, we investigated the representative gut microbiota of lean, normal, obese and surgically treated 2.5 Quantification of genus Bacteroides and archaea obese individuals of Indian origin irrespective of any age and diet restriction (details provided in table 1). All the All the real-time PCR assays were performed on 7300 real- subjects were classified based on their BMI. Additionally, to time PCR system (Applied Biosystems) using 2X Power confirm leanness of the lean individuals, whole body fat SYBR green master mix (ABI) or 2X TaqMan® Universal was also assessed using a body fat analyser (OMRON), PCR Master Mix (ABI) for Taqman probe–based assays. which works on the principle of bioelectric impedance. Primer and probe details are given in table 2. A tenfold The groups had a close age match, as reflected by their dilution series of pCR4-TOPO vector with respective 16S median age; however, rarity of older lean individuals rRNA fragment was used to generate the standard curve. All constrained us to include the readily available younger the DNA samples were assayed in triplicates at appropriate subjects (median age=23 years). Earlier studies employed dilutions, which gave least PCR inhibition. pregnancy and diet-based models (Ley et al. 2006; Duncan et al. 2008) to achieve this aim, but no study performed a cross-sectional comparison except one (Zhang et 2.6 Estimation of fecal SCFAs al. 2009). Further, our study differs from earlier studies with respect to the approach, adopted primer set and participating Estimation of fecal SCFAs was done in triplicates on individuals, who were under no dietary restrictions. Diet- Chemito 8510 (Chemito Technologies Pvt. Ltd., Mum- control-based models considerably involve prolonged active participation of the subjects over time. Moreover, the pregnan- bai, India) using 10% FFAP and 2% H3PO4 packed column fitted with flame ionization detector following stan- cy model involves only female participants and diet control dard procedure. makes this model too clumsy. In the midst of these compli- cations, we sought to compare the randomly available individuals. We hypothesized that considering variations in genetic, 2.7 Statistical analysis dietary, and age factors, such comparison may be inconclusive; however, a robust association may be interpreted. To determine the correlation between BMI and Bacter- oides distribution, Spearman Rank Correlation Coefficient 3.2 Clone libraries and classification of clones was determined using two-tailed test of significance using OriginPro SR4 software. For other comparisons, we used To assess the dominant microbiota, roughly 100–400 good- unpaired t-test at 95% confidence interval (Student’s t-test). quality sequences were obtained per library. All the libraries Genus-level statistical comparison of the libraries was carried were constructed with the recently reported primers, 8FI and out based on the concept adopted from elsewhere (Romualdi 907RI, which are reported to have more coverage across the et al. 2003). Fisher’s exact test was used with Bonferroni bacterial kingdom for 16S rRNA gene amplification (Ben-Dov correction (significance threshold=5.482456e−06). et al. 2006). Chimeric and short sequences (<300 bases)

Table 2. Details of the primers and probes used in this study

Domain/Phylum/ Genus Primers Reference

Eubacteriaa Eub338 (5′-ACT CCT ACG GGA GGC AGC AG-3′) Fierer et al. 2005 Eub518 (5′-ATT ACC GCG GCT GCT GG-3′) Archaeaa 931f (5′-AGG AAT TGG CGG GGG AGC A-3′) Einen et al. 2008 m1100r (5′-BGG GTC TCG CTC GTT RCC-3′) Bacteroidesb HuBac566f (5′-GGG TTT AAA GGG AGC GTA GG-3′) Layton et al. 2006 HuBac692r (5′-CTA CAC CAC GAA TTC CGC CT-3′) HuBac594Bhqf(5′-(FAM) TAA GTC AGT TGT GAA AGT TTG CGG CTC (TAMRA)-3′) a SYBR green–based assay; b Taqman probe–based assay.

J. Biosci. 37(4), September 2012 Gut microbiota in obesity in Indian individuals 651 were exempted from the analysis. All the non-chimeric sequen- treated individual T4 demonstrated a remarkable composition ces were deposited to NCBI (accession number GU954553- with 19% (31 out of 166) of novel uncultured bacterial species. GU957822). Identification and classification of these sequen- Moreover, T4 along with N5 had an unusually low sequence ces were performed using SEQMATCH tool at the RDP-II identity (48.19% and 25.9% respectively) with the cultured database (Cole et al. 2005) against both total (uncultured) bacterial species. and cultured bacterial isolate databases (for details, refer table 3). The taxonomic assignments were confirmed by 3.3 Estimation of OTU richness phylogenetic analysis. All the individuals showed 80–99% of bacteria with a significant identity (>97%, species level) Total number of OTUs in each library was estimated using with the uncultured bacterial database sequences. Surgically ACE and Chao1 estimators (refer table 4) employing DOTUR program. Diversity indices such as Shannon index and Simp- son’s diversity index were also determined. Microbial diver- Table 3. BLAST hit summary sity indices reflected no correlation to BMI or group, which Percent identity Percent identity is contradictory to an earlier report (Turnbaugh et al. 2009). with culturable with culturable Possibly this could be due to the necessity of additional and uncultured bacterial isolate sampling or sequencing as no library exhibited a saturated Total non- bacterial reference reference chimeric sequences (at sequences (at rarefaction curve (figure 1). Library/Total sequences RDP-II) RDP-II)

>97%a <97%b >97%a <97%b 3.4 Comparative microbial composition at phylum, genus, L1 282 279 3 220 62 and species levels L2 218 211 7 160 58 L3 291 291 0 192 100 Microbial composition was determined using SEQMATCH L4 218 217 1 159 59 tool at the RDP-II database. SEQMATCH-based hit summary L5 236 227 9 154 82 is presented in table 3. Phylum-based comparison is shown in figure 2. At the phylum level, all the individuals exhibited Total (L) 1245 20 361 dominance of bacteria of phyla, Bacteroidetes and Firmi- N1 95 94 1 91 4 cutes, which corroborates with earlier reports (Ley et al. N2 98 96 2 88 10 2005;Ley et al. 2006). Additionally, a small number of phyla N3 140 136 4 130 10 N4 162 160 2 146 16 100 N5 166 154 12 80 86 L1 Total (N) 661 21 126 L2 L3 O1 100 98 2 88 12 L4 80 O2 153 147 6 126 27 L5 O3 102 100 2 98 4 N1 N2 O4 91 88 3 70 21 N3 60 O5 71 63 8 40 31 N4 Total (O) 517 21 95 N5 O1 T1 178 176 2 161 17 O2 T2 175 161 14 139 36 40 O3 No. of Phylotypes T3 164 157 7 139 25 O4 T4 166 135 31 43 123 O5 T1 T5 164 150 14 140 24 20 T2 Total (T) 847 68 225 T3 Total sequences 3270 T4 (L+N+O+T) T5 0 0 100 200 300 The table represents summary of SEQMATCH tool based sequence No. of sequenced clones identity percentage to the deposited uncultured and culturable bacterial isolate sequences at RDP II database. a Sequence identity above 97% denotes similarity at species level. b Sequence identity Figure 1. Rarefaction curves. The curves represent the captured bac- below 97% denotes novel species. L, N, O & T are described in the terial diversity depicted as number of phylotypes in each library (at 97% table 1 legend. cut off). (L = lean, N = Normal, O = Obese and T = Treated-obese).

J. Biosci. 37(4), September 2012 652 Deepak P Patil et al. such as Proteobacteria and Actinobacteria were also and Prevotella population accounts between 12% and 62% recorded. Some of the sequences referred to as ‘unclassified of the total eubacterial population in rumen contents of sheep bacteria’ remained unidentified, which could be putative and cow (Wood et al. 1998). Their dominance among these novel bacteria. Based on the distribution of sequences at urban individuals unders tudy is intriguing, as their associa- the phylum level, no specific pattern specific to any group tion with cattle is minimal; however, this association cannot was apparent (figure 2). be neglected among their ancestors. A huge population in Differences in the bacterial distribution were clearly illus- villages still uses dried cow dung as fuel and resides in trated by the comparison of 16S rRNA gene libraries at the houses coated with cow feces (Raghavachari 1961). Recent- genus level (figure 3). Unifrac analysis and neighbour- ly, Butyrivibrio fibrisolvens, a cattle rumen bacterium (Fir- joining-based clustering of the representative OTUs is rep- micute), was reported in the feces of 52% adults and 60% resented in the heatmap in figure 3. Library cluster analysis children from an Indian village, but not in urban children clearly demonstrates clustering of lean and normal libraries (Balamurugan et al. 2009). Additionally, the role of diet in except L3 (which has an unusually high Bacteroides genus determining herbivore-like gut microbiota cannot be counts). Interestingly, libraries O1 and O2 cluster in the neglected as Indians eat a lot more plant-based food in normal/lean clade. comparison to their Western counterparts (Sinha et al. All the individuals exhibited a variety of genera. Although, 2003). A comparative analysis of gut microbiota from vari- Prevotella genus dominated the libraries, bacteria of genus ous mammals showing different dietary habits, ranging from Bacteroides were seen to be dominant in obese and surgi- carnivore to herbivore, demonstrated variable microbial cally treated obese individuals in comparison with normal community characterized by their diet and phylogeny and lean except L3 (P<0.05). Interestingly, the Bacteroides (Ley et al. 2008). Dominance of Bacteroides among the obese individuals was further confirmed by means of real-time PCR (figure 4), Table 4. Distance-based OTUs, species richness estimates, and which demonstrated a positive correlation between Bacter- diversity indices oides distribution and BMI in our subjects. These well- Observed Richness known bacteria have numerous beneficial and adverse Libraries OTUs estimators Diversity indices effects on human health (Wexler 2007). Their large genomes encompass an enriched set of enzymatic machinery for di- ACE Chao I Shannon Simpson gestion of indigestible carbohydrates making them efficient (H) (1-D) polysaccharide (cellulose, hemicellulose, starch and pectin) Lean L1 30 60.48 45.60 2.5020 0.1212 foragers (Flint et al. 2008). Interestingly, Bacteroides were L2 53 76.82 64.88 3.3013 0.0636 earlier correlated to weight gain and diabetes in pregnant L3 63 120.10 103.10 3.0537 0.0937 women (Collado et al. 2008) and also diabetes-prone rats L4 26 57.74 42.50 2.4989 0.1088 (Roesch et al. 2009). These bacteria are also shown to be L5 66 118.10 122.10 3.5338 0.0463 associated with ulcerative colitis (Lucke et al. 2006). More- Normal N1 37 64.31 54.00 3.2777 0.0403 over, some of the Bacteroides species produce faecapen- N2 25 35.54 29.00 2.6200 0.1153 taenes (Huycke and Gaskins 2004), which is a potent N3 22 78.49 67.33 1.0658 0.6467 in vitro mutagen. Some Firmicutes also show an enriched N4 36 60.87 58.67 2.8924 0.0919 set of glycoside hydrolases, e.g. Eubacterium rectal (Turnbaugh N5 63 111.00 107.00 3.6737 0.0369 100 Obese O1 29 45.43 38.75 2.8442 0.0775 Bacteroidetes Firmicutes O2 44 63.50 63.00 3.0723 0.0918 80 Proteobacteria unclassified Bacteria O3 25 58.63 46.00 2.1740 0.2353 Actinobacteria O4 22 22.36 22.00 2.6563 0.1047 60 O5 50 270.80 186.70 3.6865 0.0217 40 Treated- T1 33 59.96 63.00 2.6815 0.1145 Obese 20 T2 55 250.80 211.00 3.1267 0.0809 Percent distribution of bacteria of distribution Percent T3 46 74.08 69.00 2.8881 0.1189 0 3 1 2 1 2 3 T4 78 172.10 137.40 3.8024 0.0475 L1 L2 L3 L4 L5 N1 N2 N N4 N5 O O O3 O4 O5 T T T T4 T5 Individuals T5 43 91.97 80.50 2.8849 0.1043

OTU grouping is reported at 0.03 level, which refers to 3% distance Figure 2. Phylum-based comparison. The stacked bar graph and corresponds to the bacterial species level. L, N, O & T are describes the percent distribution of fecal microbiota across the subjects described in the table 1 legend. understudy (L = lean, N = Normal, O = Obese and T = Treated-obese).

J. Biosci. 37(4), September 2012 Gut microbiota in obesity in Indian individuals 653

Bacteroides

Parabacteroides Bacteroidetes Paludibacter unclassified Prevotellaceae unclassified Porphyromonadaceae Tannerella Hallella Prevotella unclassified Rikenellaceae Alistipes-O5-87 374 bases Alistipes Sutterella-LI2-44 374 bases Sutterella Comamonas-T3-125 374 bases Comamonas unclassified Enterobacteriacea 374 ... unclassified Enterobacteriaceae Paracoccus-LI1-102 374 bases Paracoccus Agrobacterium-T2-48 374 bases Agrobacterium Streptococcus-O2-134 374 bases Streptococcus Lactobacillus-O5-5 374 bases Lactobacillus Erysipelotrichaceae 374 bases Erysipelotrichaceae Incertae Sedis Allobaculum-N5-126 374 bases Allobaculum unclassified Erysipelotrichace 374 ... unclassified Erysipelotrichaceae Catenibacterium-O1-104 374 bases Catenibacterium Coriobacterineae-N5-140 374 bases Coriobacterineae Veillonella-T3-31 374 bases Veillonella Dialister-O5-188 374 bases Dialister

Megasphaera-O5-158 374 bases Megasphaera Firmicutes Acidaminococcus-O2-181 374 bases Acidaminococcus Mitsuokella-O2-43 374 bases Mitsuokella Megamonas-T1-52 374 bases Megamonas unclassified Veillonellaceae-T 374 ... unclassified Veillonellaceae unclassified Ruminococcaceae 374 bases unclassified "Ruminococcaceae" Color key (%) Papillibacter-O5-174 374 bases Papillibacter 0 Sporobacter-O2-90 374 bases Sporobacter <=1 Ruminococcaceae 374 bases Ruminococcaceae Incertae Sedis 1-2 Ruminococcus-O5-184 374 bases Ruminococcus 2-3 Anaerotruncus-LI5-6 374 bases Anaerotruncus 3-4 Subdoligranulum-O5-71 374 bases Subdoligranulum 4-5 Faecalibacterium-O5-157 374 bases Faecalibacterium Peptostreptococcaceae 374 bases Peptostreptococcaceae Incertae Sedis 5-10 Clostridiaceae 374 bases Clostridiaceae 10-20 unclassified Incertae 374 bases unclassified Incertae Sedis XI 20-30 Anaerovorax-O5-92 374 bases Anaerovorax 30-40 Butyrivibrio-LI5-307 374 bases Butyrivibrio 40-50 Dorea-O5-72 374 bases Dorea 50-60 Lachnospira-O2-183 374 bases Lachnospira 60-70 unclassified Lachnospiraceae 374 bases unclassified "Lachnospiraceae" 70-80 Roseburia-O2-186 374 bases Roseburia 80-90 Lachnospiraceae 374 bases Lachnospiraceae Incertae Sedis 90-100 Coprococcus-N2-169 374 bases Coprococcus

0.15 0.10 0.05 0.00

Figure 3. Genus-based comparison and microbial similarities. The heatmap represents clustering of bacterial communities with their percent distribution across the subjects at the genus level. The bacterial communities or OTUs were clustered phylogenetically by neighbour-joining method and the libraries were clustered by profile pattern by Unifrac analysis. Color key for the distribution is presented in the right corner (L = lean, N = Normal, O = Obese and T = Treated-obese). et al. 2006). However, we did not find dominance of this bacteria actively participate in butyrate synthesis, which acts bacterium in any of the obese individuals under study. as an energy source for colonocytes and is known to exhibit Butyrate producing bacteria such as Faecalibacterium were beneficial effects on the gut health (Flint et al. 2007). Faeca- seen in smaller numbers but comparatively higher percen- libacterium prausnitzii is shown to have anti-inflammatory tages in the normal gut in comparison with the others. These properties and epidemiology of Crohn’s disease (Sokol

J. Biosci. 37(4), September 2012 654 Deepak P Patil et al.

Bacteroides distribution with BMI 50 L (Yost et al. 1977). Such a condition can also be speculated in N the human subjects under study. In obese individuals under O T study, the dominance of H2-producing and polysaccharide- 40 foraging bacterial species such as Prevotella (Avgustin et al. 1997; Wood et al. 1998) and Bacteroides (Samuel and Gordon 2006), along with methanogenic archaea, may result 30 in an increased glucose and SCFA supply, thereby leading to

16S rRNA 16S copies perturbations in the host metabolism. Elevated glucose levels with increased adiposity are renowned hallmarks of obesity 20 (Haslam and James 2005). Comparatively lower occurrences Bacteroides of bacteria of genus ‘Bacteroides’ and domain ‘Archaea’ in lean, normal and treated-obese individuals represent a rever-

(percent per total eubacterial 16S rRNA sequences) 10 sal of the previously discussed change in the metabolism, thereby leading to a normal BMI.

20 30 40 50 Fecal SCFA profile was studied through gas chromato- BMI (Body mass index, in kg/m2) graphic analysis of the fecal samples. It revealed a significant overproduction of SCFAs (figure 6A) in the obese gut, indi- Figure 4. Distribution of genus Bacteroides with BMI. The graph cating increased saccharolytic fermentation as reported ear- presents real-time PCR based quantification of Bacteroides in percent lier (Turnbaugh et al. 2006). Obese and treated-obese copies detected per total 16S rRNAs in each sample per gram of individuals exhibited approximately twofold higher produc- feces vs BMI of the individuals understudy. Linear regression (R2= tion of total SCFAs in comparison with lean and normal 0.4378) is also shown (Spearman’s correlation coefficient r=0.6491, P= individuals. The treated-obese individuals show a signifi- 0.0020). (L = lean, N = Normal, O = Obese and T = Treated-obese). cantly reduced synthesis of SCFA in comparison with obese individuals, correlating to reduced archaeal populations in et al. 2008). Reduced levels of this bacterium and lower the distal gut (figure 6A). A mutualistic association of H2- levels of butyrate in the obese may put them at risk of inflam- producing bacteria, Bacteroides and Prevotella with archaea matory disorders of gastrointestinal tract. Some lean and almost all the obese individuals exhibited dominant presence of the 8.0×106 bacteria of family Ruminococcaceae, which remain unclassi- * fied and unidentified. Many other species belonging to genera, ‘ ‘ 6.0×106 Lachnospiraceae Incertae Sedis , Veillonella, Streptococcus, * Ruminococcus, Porphyromonas, Parabacteroides, Mitsuo- kella, Megamonas, Eubacterium and Coprococcus were also × 6 found in the individuals understudy but in low abundance. 4.0 10

2.0×106 3.5 Distribution of archaea and fecal SCFAs 1.5×104

To assess the distribution of archaea in the individuals un- × 4 ** derstudy, real-time PCR-based quantification was carried 1.0 10

out. As previously reported, we found a significantly high feces) g of per copies (gene Average archaeal 16S rRNA ** archaeal density in the obese subjects (Samuel and Gordon 5.0×103 2006; Zhang et al. 2009) and comparatively lesser numbers were detected in the treated obese (figure 5). Normal and lean individuals exhibited a remarkably low archaeal DNA 0 amplification, with some individuals showing no amplification L N O T at all. Methanobrevibacter smithii, a dominant methanogenic Group archaeon of human gut (Eckburg et al. 2005), in consortium with Bacteroides thetaiotaomicron produces a high amount Figure 5. Distribution of archaeal 16S rRNA gene copies among of SCFAs (especially acetate and propionate). The overpro- the groups under study. The graph presents real-time PCR based duced acetate stimulates hepatic de novo lipogenesis in mice, distribution of archaeal 16S rRNA gene copies in the groups un- causing an increased adiposity (Samuel and Gordon 2006). derstudy (Student’s t-test, *P=0.0341, **P=0.0596) (L = lean, N = Another SCFA, propionate, is also shown to be gluconeogenic Normal, O = Obese and T = Treated-obese).

J. Biosci. 37(4), September 2012 Gut microbiota in obesity in Indian individuals 655 ab 15000 10000 * * ** L N O T * 8000

10000 6000 mol/g of feces) ** mol/g of feces) µ µ

** 4000 5000 ***

2000 Concentration ( in Concentration Concentration ( in Concentration

0 0

L N O T te ra Group etate ionate y p t Ac Bu Pro Group

Figure 6. Distribution of fecal SCFAs. (A) Comparative representation of total fecal SCFAs (*, **P<0.05) and (B) Individual fecal SCFAs (*, **, ***P<0.05) among various groups understudy. Student’s t-test was used for the statistical analysis. (L = lean, N =Normal, O = Obese and T = Treated-obese). in obese individuals, leads to overproduction of SCFAs archaea and lowered fecal SCFAs are some compelling (Samuel and Gordon 2006). Humans are reported to have observations. However, a change in gut microbiota due to hydrogen-consuming methanogenic archaeons predominant- gastrectomy-induced early satiety requires further study and ly, M. smithii (Eckburg et al. 2005). This association leads to better understanding of human physiology. Role of gut hor- an efficient extraction of energy from relatively indigestible mones especially ghrelin cannot be ignored, which is re- polysaccharides such as fructans. Bacteria of genus Bacter- duced significantly after gastrectomy (Langer et al. 2005). oides are capable of digesting various polysaccharides (Flint In addition, dietary differences, age, ethnicity and the num- et al. 2008), leading to the release of several folds of free ber of participating individuals could be other limitations of sugars such as fructose from fructans. These free sugars are this study. To study a direct role of any bacteria in the also converted to acetate through saccharolytic fermentation, pathogenesis or treatment of obesity, a much more in-depth thereby releasing formate, which is used by M. smithii for study is required; however, these data are a stepping stone in methanogenesis. A high prevalence of archaea with bacteria appreciation of the associated gut microbiota in obesity of genus Bacteroides directs Bacteroides to focus on fermen- among the Indian subjects. In light of availability of next- tation of dietary polysaccharides to acetate. An increased generation sequencing technologies such as 454, Illumina amount of acetate enters the host energy metabolism and and Iontorrent sequencing, a study designed to assess the leads to an increased adiposity in the obese individuals. total diversity and metagenomic composition of gut micro- Majorly, three types of SCFAs, vis-à-vis, acetate, propionate biota can be highly informative but analytically challenging. and butyrate, were seen in all the individuals along with smaller proportions of other SCFAs such as valerate and iso- Acknowledgements valerate. As stated earlier, acetate is produced in significantly high amounts in the obese group (figure 6B). Interestingly, a We would like to thank the volunteers for their participation remarkably high butyrate production is seen in normal indi- in the study. We also thank Ms Tulika M Jaokar for proof- viduals in comparison with any other groups under study. reading the manuscript. We also gratefully acknowledge the The ability of the gut microbiota to produce butyrate can active support and encouragement provided by the former vary considerably in response to environmental factors, as Director of National Centre for Cell Science (NCCS) has been shown in vitro in response to pH (Walker et al. Padmashree Dr GC Mishra. 2005) and in vivo in response to diet (Duncan et al. 2008). References

4. Summary and future directions Ashelford KE, Chuzhanova NA, Fry JC, Jones AJ and Weightman AJ 2006 New screening software shows that most recent large In summary, although the gut microbiota differed from indi- 16S rRNA gene clone libraries contain chimeras. Appl. Environ. vidual to individual, surgically treated individuals displayed Microbiol. 72 5734–5741 few notable and relative trends as in normal individuals. Avgustin G, Wallace RJ and Flint HJ 1997 Phenotypic diversity Comparative decrease in Bacteroides population, decreased among ruminal isolates of Prevotella ruminicola: proposal of

J. Biosci. 37(4), September 2012 656 Deepak P Patil et al.

Prevotella brevis sp. nov., Prevotella bryantii sp. nov. and Pre- Haslam DW and James WP 2005 Obesity. Lancet 366 1197–1209 votella albensis sp. nov. and redefinition of Prevotella rumini- Hill JO and Peters JC 1998 Environmental contributions to the cola. Int. J. Syst. Bacteriol. 47 284–288 obesity epidemic. Science 280 1371–1374 Backhed F, Ding H, Wang T, Hooper LV, Koh GY, Nagy A, Huycke MM and Gaskins HR 2004 Commensal bacteria, redox Semenkovich CF and Gordon JI 2004 The gut microbiota as stress, and colorectal cancer: mechanisms and models. Exp. Biol. an environmental factor that regulates fat storage. Proc. Natl. Med. (Maywood. ) 229 586–597 Acad. Sci. USA 101 15718–15723 Langer FB, Reza Hoda MA, Bohdjalian A, Felberbauer FX, Balamurugan R, Chittaranjan SP, Chandragunasekaran AM and Zacherl J, Wenzl E, Schindler K, Luger A, Ludvik B and Prager Ramakrishna BS 2009 Molecular detection of the ruminal bac- G 2005 Sleeve gastrectomy and gastric banding: effects on terium, Butyrivibrio fibrisolvens, in feces from rural residences plasma ghrelin levels. Obes. Surg. 15 1024–1029 of Southern India.. Microb. Ecolo. Health Dis. 21 38–43 Layton A, McKay L, Williams D, Garrett V, Gentry R and Sayler G Ben-Dov E, Shapiro OH, Siboni N and Kushmaro A 2006 Advan- 2006 Development of Bacteroides 16S rRNA gene TaqMan- tage of using inosine at the 3' termini of 16S rRNA gene based real-time PCR assays for estimation of total, human, and universal primers for the study of microbial diversity. Appl. bovine fecal pollution in water. Appl. Environ. Microbiol. 72 Environ. Microbiol. 72 6902–6906 4214–4224 Choban PS, Jackson B, Poplawski S and Bistolarides P 2002 Ley RE, Backhed F, Turnbaugh P, Lozupone CA, Knight RD and Bariatric surgery for morbid obesity: why, who, when, how, Gordon JI 2005 Obesity alters gut microbial ecology. Proc. Natl. where, and then what?. Cleve. Clin. J. Med. 69 897–903 Acad. Sci. USA 102 11070–11075 Cole JR, Chai B, Farris RJ, Wang Q, Kulam SA, McGarrell DM, Ley RE, Hamady M, Lozupone C, Turnbaugh PJ, Ramey RR, Garrity GM and Tiedje JM 2005 The Ribosomal Database Bircher JS, Schlegel ML, Tucker TA, Schrenzel MD, Knight Project (RDP-II): sequences and tools for high-throughput R and Gordon JI 2008 Evolution of mammals and their gut rRNA analysis. Nucleic Acids Res. 33 D294–D296 microbes. Science 320 1647–1651 Collado MC, Isolauri E, Laitinen K and Salminen S 2008 Distinct Ley RE, Turnbaugh PJ, Klein S and Gordon JI 2006 Microbial composition of gut microbiota during pregnancy in overweight ecology: human gut microbes associated with obesity. Nature and normal-weight women. Am. J. Clin. Nutr. 88 894–899 444 1022–1023 Cummings DE and Schwartz MW 2003 Genetics and pathophysi- Lintsi M, Kaarma H and Kull I 2004 Comparison of hand-to- ology of human obesity. Annu. Rev. Med. 54 453–471 hand bioimpedance and anthropometry equations versus dual- Deepa M, Farooq S, Deepa R, Manjula D and Mohan V 2009 energy X-ray absorptiometry for the assessment of body fat Prevalence and significance of generalized and central body percentage in 17–18-year-old conscripts. Clin. Physiol Funct. obesity in an urban Asian Indian population in Chennai, India Imaging 24 85–90 (CURES: 47). Eur. J. Clin. Nutr. 63 259–267 Lozupone C and Knight R 2005 UniFrac: a new phylogenetic DeSantis TZ, Jr., Hugenholtz P, Keller K, Brodie EL, Larsen N, method for comparing microbial communities. Appl. Environ. Piceno YM, Phan R and Andersen GL 2006 NAST: a multiple Microbiol. 71 8228–8235 sequence alignment server for comparative analysis of 16S Lucke K, Miehlke S, Jacobs E and Schuppler M 2006 Prevalence of rRNA genes. Nucleic Acids Res. 34 W394-W399 Bacteroides and Prevotella spp. in ulcerative colitis. J. Med. Duncan SH, Lobley GE, Holtrop G, Ince J, Johnstone AM, Louis P Microbiol. 55 617–624 and Flint HJ 2008 Human colonic microbiota associated with diet, Murphy KG and Bloom SR 2006 Gut hormones and the regulation obesity and weight loss. Int. J. Obes. (London) 32 1720–1724 of energy homeostasis. Nature 444 854–859 Eckburg PB, Bik EM, Bernstein CN, Purdom E, Dethlefsen L, Nadal I, Santacruz A, Marcos A, Warnberg J, Garagorri M, Moreno Sargent M, Gill SR, Nelson KE and Relman DA 2005 Diversity LA, Martin-Matillas M, Campoy C, et al. 2009 Shifts in clos- of the human intestinal microbial flora. Science 308 1635–1638 tridia, bacteroides and immunoglobulin-coating fecal bacteria Einen J, Thorseth IH and Ovreas L 2008 Enumeration of Archaea associated with weight loss in obese adolescents. Int. J. Obes. and Bacteria in seafloor basalt using real-time quantitative PCR (Lond) 33 758–767 and fluorescence microscopy. FEMS Microbiol. Lett. 282 182– Noverr MC and Huffnagle GB 2004 Does the microbiota 187 regulate immune responses outside the gut? Trends Micro- Ewing B, Hillier L, Wendl MC and Green P 1998 Base-calling of biol. 12 562–568 automated sequencer traces using phred. I. Accuracy assess- Raghavachari TN S 1961 Cow dung and Insanitary Flooring. JAMA ment. Genome Res. 8 175–185 178 680 Fierer N, Jackson JA, Vilgalys R and Jackson RB 2005 Assessment of Rakoff-Nahoum S, Paglino J, Eslami-Varzaneh F, Edberg S and soil microbial community structure by use of taxon-specific quan- Medzhitov R 2004 Recognition of commensal microflora by titative PCR assays. Appl. Environ. Microbiol. 71 4117–4120 toll-like receptors is required for intestinal homeostasis. Cell Flint HJ, Bayer EA, Rincon MT, Lamed R and White BA 2008 118 229–241 Polysaccharide utilization by gut bacteria: potential for new Rautava S and Isolauri E 2002 The development of gut immune insights from genomic analysis. Nat. Rev. Microbiol. 6 121–131 responses and gut microbiota: effects of probiotics in prevention Flint HJ, Duncan SH, Scott KP and Louis P 2007 Interactions and treatment of allergic disease. Curr. Iss. Intest. Microbiol. 3 and competition within the microbial community of the 15–22 human colon: links between diet and health. Environ. Microbiol. Roesch LF, Lorca GL, Casella G, Giongo A, Naranjo A, Pionzio 9 1101–1111 AM, Li N, Mai V, et al. 2009 Culture-independent identification

J. Biosci. 37(4), September 2012 Gut microbiota in obesity in Indian individuals 657

of gut bacteria correlated with the onset of diabetes in a rat Tamura K, Dudley J, Nei M and Kumar S 2007 MEGA4: Molec- model. ISME J. 3 536–548 ular Evolutionary Genetics Analysis (MEGA) software version Romualdi C, Bortoluzzi S, D'Alessi F and Danieli GA 2003 4.0. Mol. Biol. Evol. 24 1596–1599 IDEG6: a web tool for detection of differentially expressed Terranova L, Busetto L, Vestri A and Zappa MA 2012 Bariatric genes in multiple tag sampling experiments. Physiol Genomics surgery: cost-effectiveness and budget impact. Obes. Surg. 22 12 159–162 646–653 Samuel BS and Gordon JI 2006 A humanized gnotobiotic mouse Thompson D and Wolf AM 2001 The medical-care cost burden of model of host-archaeal-bacterial mutualism. Proc. Natl. Acad. obesity. Obes. Rev. 2 189–197 Sci. U S A 103 10011–10016 Turnbaugh PJ, Hamady M, Yatsunenko T, Cantarel BL, Duncan A, Schloss PD and Handelsman J 2005 Introducing DOTUR, a com- Ley RE, Sogin ML, Jones WJ, et al. 2009 A core gut micro- puter program for defining operational taxonomic units and biome in obese and lean twins. Nature 457 480–484 estimating species richness. Appl. Environ. Microbiol. 71 Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER and 1501–1506 Gordon JI 2006 An obesity-associated gut microbiome with Schneider BE and Mun EC 2005 Surgical management of morbid increased capacity for energy harvest. Nature 444 1027–1031 obesity. Diabetes Care 28 475–480 Walker AW, Duncan SH, Leitch EC M, Child MW and Flint HJ 2005 Shamseddeen H, Getty JZ, Hamdallah IN and Ali MR 2011 Epi- pH and peptide supply can radically alter bacterial populations and demiology and economic impact of obesity and type 2 diabetes. short-chain fatty acid ratios within microbial communities from the Surg. Clin. North Am. 91 1163–72, vii human colon. Appl. Environ. Microbiol. 71 3692–3700 Sinha R, Anderson DE, McDonald SS and Greenwald P 2003 Wexler HM 2007 Bacteroides: the good, the bad and the nitty- Cancer risk and diet in India. J. Postgrad. Med. 49 222– gritty. Clin. Microbiol. Rev. 20 593–621 228 Wood J, Scott KP, Avgustin G, Newbold CJ and Flint HJ 1998 Sokol H, Pigneur B, Watterlot L, Lakhdari O, Bermudez-Humaran Estimation of the relative abundance of different Bacteroides LG, Gratadoux JJ, Blugeon S, Bridonneau C, et al. 2008 Fae- and Prevotella ribotypes in gut samples by restriction enzyme calibacterium prausnitzii is an anti-inflammatory commensal profiling of PCR-amplified 16S rRNA gene sequences. Appl. bacterium identified by gut microbiota analysis of Crohn disease Environ. Microbiol. 64 3683–3689 patients. Proc. Natl. Acad. Sci. US A 105 16731–16736 Yost WM, Young JW, Schmidt SP and McGilliard AD 1977 Stappenbeck TS, Hooper LV and Gordon JI 2002 Developmental Gluconeogenesis in ruminants: propionic acid production from regulation of intestinal angiogenesis by indigenous microbes via a high-grain diet fed to cattle. J. Nutr. 107 2036–2043 Paneth cells. Proc. Natl. Acad. Sci. USA 99 15451–15455 Zhang H, DiBaise JK, Zuccolo A, Kudrna D, Braidotti M, Yu Y, Stevens CE and Hume ID 1998 Contributions of microbes in Parameswaran P, Crowell MD, Wing R, Rittmann BE and vertebrate gastrointestinal tract to production and conservation Krajmalnik-Brown R 2009 Human gut microbiota in obesity and of nutrients. Physiol. Rev. 78 393–427 after gastric bypass. Proc. Natl. Acad. Sci. USA 106 2365–2370

MS received 12 March 2012; accepted 29 June 2012

Corresponding editor: SUDHA BHATTACHARYA

J. Biosci. 37(4), September 2012