Research Article: Microbiota data from low biomass milk samples is markedly affected by laboratory and reagent contamination

Date Published: June 13, 2019

Publisher: Public Library of Science

Author(s): Josef Dahlberg, Li Sun, Karin Persson Waller, Karin Östensson, Mark McGuire, Sigrid Agenäs, Johan Dicksved, Juan J. Loor.


Discoveries of bacterial communities in environments that previously have been described as sterile have in recent years radically challenged the view of these environments. In this study we aimed to use 16S rRNA sequencing to describe the composition and temporal stability of the bacterial microbiota in bovine milk from healthy udder quarters, an environment previously believed to be sterile. Sequencing of the 16S rRNA gene is a technique commonly used to describe bacterial composition and diversity in various environments. With the increased use of 16S rRNA gene sequencing, awareness of methodological pitfalls such as biases and contamination has increased although not in equal amount. Evaluation of the composition and temporal stability of the microbiota in 288 milk samples was largely hampered by background contamination, despite careful and aseptic sample processing. Sequencing of no template control samples, positive control samples, with defined levels of bacteria, and 288 milk samples with various levels of bacterial growth, revealed that the data was influenced by contaminating taxa, primarily Methylobacterium. We observed an increasing impact of contamination with decreasing microbial biomass where the contaminating taxa became dominant in samples with less than 104 bacterial cells per mL. By applying a contamination filtration on the sequence data, the amount of sequences was substantially reduced but only a minor impact on number of identified taxa and by culture known endogenous taxa was observed. This suggests that data filtration can be useful for identifying biologically relevant associations in milk microbiota data.

Partial Text

The introduction of DNA based methods to study bacterial communities has in recent years stimulated interest and substantially challenged previous knowledge about environments thought to be sterile. Milk, placenta and airways are examples of environments that previously were considered sterile in healthy individuals, but when studied with DNA based methods revealed to harbor their own microbiome [1–5]. Simultaneously publications on problems with laboratory and reagent contamination in microbiota studies have become increasingly common and a list of commonly occurring contaminating genera has been created [6, 7]. Occasionally discoveries of a microbiome in a previously believed sterile environments have been questioned and attributed to methodological artefacts [8].

The animal experiment for this study was designed to compare the milk microbiota within quarter over time, as well as between quarters and between animals over time. A large proportion of quarters had a similar bacteriological finding by culture throughout the study period indicating that bacteria findings by culture dependent methods were stable over time. Corynebacterium spp. was the most commonly isolated bacteria by culture and was repeatedly detected in milk from the same quarters adding further support that bacteriological response was stable over time. We aimed to study the composition and temporal stability of the milk microbiota using 16S amplicon sequencing. Despite a very careful treatment of the samples, with all DNA isolations carried out in a laminar air flow hood pretreated with UV light and cleaned with both 10% bleach and 70% ethanol, contamination from reagents and the laboratory environment had a pronounced effect on the results. Due to this contamination, characterization and assessment of temporal stability of the bovine milk microbiota via 16S rRNA sequencing proved difficult and is further discussed below.