Lactobacillus elicits a 'Marmite effect' on the chicken cecal microbiome

www.nature.com/npjbiofilms BRIEF COMMUNICATION OPEN Lactobacillus elicits a 'Marmite effect' on the chicken cecal microbiome Angela Zou1,2, Shayan Sharif3 and John Parkinson1,2,4 The poultry industry has traditionally relied on the use of antibiotic growth promoters (AGPs) to improve production efficiency and minimize infection. With the recent drive to eliminate the use of AGPs, novel alternatives are urgently required. Recently attention has turned to the use of synthetic communities that may be used to ‘seed’ the developing microbiome. The current challenge is identifying keystone taxa whose influences in the gut can be leveraged for probiotic development. To help define such taxa we present a meta-analysis of 16S rRNA surveys of 1572 cecal microbiomes generated from 19 studies. Accounting for experimental biases, consistent with previous studies, we find that AGP exposure can result in reduced microbiome diversity. Network community analysis defines groups of taxa that form stable clusters and further reveals Lactobacillus to elicit a polarizing effect on the cecal microbiome, exhibiting relatively equal numbers of positive and negative interactions with other taxa. Our identification of stable taxonomic associations provides a valuable framework for developing effective microbial consortia as alternatives to AGPs. npj Biofilms and Microbiomes (2018)4:27 ; doi:10.1038/s41522-018-0070-5 INTRODUCTION The association of antibiotic growth promoter (AGPs) usage with antimicrobial resistance is prompting the poultry industry to seek alternative feed supplements.1 AGPs are used to increase production efficiency and reduce flock infections.1 While their precise mode of action is not known, AGPs are thought to work through altering the microbial community (microbiome) in the livestock gastrointestinal tract.2 Currently, interest lies in finding combinations of previously identified probiotics that can be used to promote the development of a healthy microbiome. To better understand stably associating taxa, we present a meta-analysis of published 16S rRNA surveys of the chicken ceca to identify key interactions/influencers in the chicken cecal microbiome. Previous publications have reported microbiome responses under a variety of conditions; including the effects of feed additives, Eimeria challenge, and breeding conditions. However, experimental biases of individual studies have led to conflicting results, especially when investigating the effects of AGPs.3 By combining datasets, it may be possible to discern general patterns of microbiome behaviour that are consistently found across all studies. RESULTS AND DISCUSSION Limitations of technical biases on microbiome meta-analyses 16S rRNA gene sequences from 1572 chicken cecal samples were collated from 19 studies (Supplementary Table 1). We assigned ~22 million 16S rRNA gene sequences to 3300 OTUs (See Supplemental Information). Consistent with previous studies,4 Bacteroidetes, Firmicutes, and Proteobacteria were the dominant phyla, with relative proportions varying by breed (Fig. 1a and Supplementary Fig. 1). Relative to other breeds, broilers from commercial primary breeders, Cobb and Ross, exhibited similar profiles albeit Cobb exhibited a higher proportion of Christensenellaceae and Lactobacillus. Of the two layers included in this study (White leghorn and Lohmann), the microbiome profile of commercial Lohmann layers closely resembled the profiles of Chinese Tibetan chicken samples, which were sequenced and extracted by the same study, potentially reflecting study bias. Indeed, PCoA revealed that microbiome structure segregated by individual studies (Fig. 1b, Supplementary Fig. 2), suggesting they may be influenced by technical biases present, similar to the results of other microbiome meta-analyses.5,6 Moreover, sequencing region strongly influenced alpha diversity comparisons; we observed that AGP-treated samples sequenced using the V4, V3, and V6-V8 hypervariable regions exhibited significantly higher diversity (t-test; p-value < 0.05) than non-AGP-treated samples, most of which were sequenced by V1V3 and 454 Roche (Supplementary Fig. 3). However, after partitioning data based on the region of the 16S rRNA gene targeted for sequencing, AGP-treated samples consistently display equal or lower diversity compared to control groups regardless of hypervariable region used (Supplementary Fig. 4, Supplementary Table 2), consistent with previous studies. Given that different regions of the 16S rRNA gene vary in length and sequence diversity,7 it is not unexpected that phylogenetic resolutions and subsequent within-diversity analysis were also found to differ for each region (Supplementary Fig. 3). Furthermore, sequencing platforms differ in error rates and sequencing depth, both of which were found to impact the number of OTUs detected within a sample (Supplementary Fig. 3). This is consistent with findings from other meta-analyses5,8 and highlights the need to be cautious when interpreting results from 16S rRNA-based metaanalyses, particularly when datasets may be generated using different methodologies. 1 Program in Molecular Medicine, Hospital for Sick Children, Peter Gilgan Center for Research and Learning, 686 Bay Street, Toronto, ON M5G 0A4, Canada; 2Department of Biochemistry, University of Toronto, Toronto, ON, Canada; 3Department of Pathobiology, Ontario Veterinary College, University of Guelph, Guelph, ON N1G 2W1, Canada and 4 Department of Molecular Genetics, University of Toronto, Toronto, ON, Canada Correspondence: John Parkinson () Received: 8 February 2018 Accepted: 18 October 2018 Published in partnership with Nanyang Technological University Lactobacillus elicits a 'Marmite effect' on the chicken cecaly A Zou et al. 2 A Relative abundance 100 1 1 1 1 1 1 1 130 1122 163 56 59 35 Bacteroides Parabacteroides Alistipes Rikenellaceae RC9 gut group Lactobacillus Christensenellaceae R-7 group Clostridiales (vadin BB60 group) Lachnoclostridium Lachnospiraceae (NK4A136 group) [Ruminococcus] torques group Lachnospiraceae Anaerotruncus Butyricicoccus Faecalibacterium Ruminoclostridium 5 Ruminoclostridium 9 Ruminococcaceae UCG-005 Ruminococcaceae UCG-014 Ruminococcaceae Helicobacter Escherichia-Shigella 80 60 40 20 1234567890():,; 0 Breeds B Hypervariable region PC2 (8.57 %) (48) V1-V2 (955) V1-V3 (30) V2-V3 (104) V3 (7) V3-V4 (30) V3-V5 (278) V4 (120) V6-V8 PC1 (15.38 %) PC3 (4.17 %) Fig. 1 Microbial diversity of 1572 cecal samples from chicken. a Relative abundance of the most abundant genera by chicken breeds. Number on top of bars represent the number of sequencing samples for each breed, note that certain samples are pooled from multiple chicken cecal samples (see supplementary table 1). Only taxa present at greater than 1% were included. b Principal-coordinate analysis plot of unweighted UniFrac distances coloured according to hypervariable region. Numbers in brackets are the number of samples sequenced (...truncated)