How to Count Our Microbes? The Effect of Different Quantitative Microbiome Profiling Approaches.

Gianluca Galazzo,Christel Driessen,Liene Bervoets,Pascalle H Van Eijck,John Penders,Melissa Oomen,Mathias W Hornef,Kevin Janssen,Paul H Savelkoul,Heike E F Becker,Frank R M Stassen,Niels Van Best,Birke J Benedikter,Petra F Wolffs,Mayk Lucchesi

doi:10.3389/fcimb.2020.00403

Abstract

Next-generation sequencing (NGS) has instigated the research on the role of the microbiome in health and disease. The compositional nature of such microbiome datasets makes it however challenging to identify those microbial taxa that are truly associated with an intervention or health outcome. Quantitative microbiome profiling overcomes the compositional structure of microbiome sequencing data by integrating absolute quantification of microbial abundances into the NGS data. Both cell-based methods (e.g., flow cytometry) and molecular methods (qPCR) have been used to determine the absolute microbial abundances, but to what extent different quantification methods generate similar quantitative microbiome profiles has so far not been explored. Here we compared relative microbiome profiling (without incorporation of microbial quantification) to three variations of quantitative microbiome profiling: (1) microbial cell counting using flow cytometry (QMP), (2) counting of microbial cells using flow cytometry combined with Propidium Monoazide pre-treatment of fecal samples before metagenomics DNA isolation in order to only profile the microbial composition of intact cells (QMP-PMA), and (3) molecular based quantification of the microbial load using qPCR targeting the 16S rRNA gene. Although qPCR and flow cytometry both resulted in accurate and strongly correlated results when quantifying the bacterial abundance of a mock community of bacterial cells, the two methods resulted in highly divergent quantitative microbial profiles when analyzing the microbial composition of fecal samples from 16 healthy volunteers. These differences could not be attributed to the presence of free extracellular prokaryotic DNA in the fecal samples as sample pre-treatment with Propidium Monoazide did not improve the concordance between qPCR-based and flow cytometry-based QMP. Also lack of precision of qPCR was ruled out as a major cause of the disconcordant findings, since quantification of the fecal microbial load by the highly sensitive digital droplet PCR correlated strongly with qPCR. In conclusion, quantitative microbiome profiling is an elegant approach to bypass the compositional nature of microbiome NGS data, however it is important to realize that technical sources of variability may introduce substantial additional bias depending on the quantification method being used.

Highlights

Next-generation sequencing (NGS) has instigated microbiome research and resulted in many novel insights on the role of the microbiome in health and disease
As the DMM-clustering was based on a limited number of samples, having potentially a limited accuracy, we examined whether the clustering was in concordance with the classification according to the reference-based enterotype classification model fitted on MetaHIT samples (Costea et al, 2018) DMM-based clustering and referencebased classification was in accordance for all samples, with the exception of three samples that were classified into the Firmicutes enterotype according to the classification model
Cell counting resulted in a concentration very similar to the expected concentration as provided by the manufacturer as quantified by a cellometer (Table S1)

Summary

INTRODUCTION

Next-generation sequencing (NGS) has instigated microbiome research and resulted in many novel insights on the role of the microbiome in health and disease. Jian et al (2020) used quantitative PCR (qPCR) as a simple and cost-effective alternative to determine the bacterial load and estimate the absolute taxon abundance from NGS data. Vandeputte et al (2017) showed only a moderate correlation between quantification of microbial load by flowcytometry and qPCR, a direct comparison between cellbased and molecular-based methods to estimate absolute taxon abundances from NGS data has not yet been conducted. As such the level of potential bias that could be introduced when applying quantitative microbial profiling remains unknown. We compared the (lack of) sensitivity of qPCRbased methods for microbial quantification to digital droplet PCR (Hindson et al, 2013) as a more precise, discriminating, and reproducible molecular quantification method (Kim et al, 2014; Gobert et al, 2018)

MATERIALS AND METHODS

RESULTS AND DISCUSSION

ETHICS STATEMENT