Molecular studies neglect apparently gram-negative populations in the human gut microbiota.

Studying the relationships between gut microbiota, human health, and diseases is a major challenge that generates contradictory results. Most studies draw conclusions about the gut repertoire using a single biased metagenomics approach. We analyzed 16 different stool samples collected from healthy subjects who were from different areas, had metabolic disorders, were immunocompromised, or were treated with antibiotics at the time of the stool collection. The analyses performed included Gram staining, flow cytometry, transmission electron microscopy (TEM), quantitative real-time PCR (qPCR) of the Bacteroidetes and Firmicutes phyla, and pyrosequencing of the 16S rRNA gene amplicons targeting the V6 region. We quantified 10(10) prokaryotes per gram of feces, which is less than was previously described. The Mann-Whitney test revealed that Gram-negative proportions of the prokaryotes obtained by Gram staining, TEM, and pyrosequencing differed according to the analysis used, with Gram-negative prokaryotes yielding median percentages of 70.6%, 31.0%, and 16.4%, respectively. A comparison of TEM and pyrosequencing analyses highlighted a difference of 14.6% in the identification of Gram-negative prokaryotes, and a Spearman test showed a tendency toward correlation, albeit not significant, in the Gram-negative/Gram-positive prokaryote ratio (ρ = 0.3282, P = 0.2146). In contrast, when comparing the qPCR and pyrosequencing results, a significant correlation was found for the Bacteroidetes/Firmicutes ratio (ρ = 0.6057, P = 0.0130). Our study showed that the entire diversity of the human gut microbiota remains unknown because different techniques generate extremely different results. We found that to assess the overall composition of bacterial communities, multiple techniques must be combined. The biases that exist for each technique may be useful in exploring the major discrepancies in molecular studies.