A limitation of our study is the lack of comparison of our sequences with that of the upper respiratory flora. This could possibly be obtained by performing 16S rDNA sequencing on a matching nasal lavage sample for each mouse. This should be done in the future. Our lung tissue samples showed some clustering that could indicate a sampling problem. In our study we sampled the distal tip of the left lung lobe after the BAL procedure was performed. The clustering could be a result of this BAL procedure not being equally effective between samples in the very low airways, sometimes leaving the distal

tip un-flushed. This would predict a clustering showing one community equal to the one found in the BAL and one more rich and diverse representing the less rinsed tissue. If we were especially interested in the tissue associated E7080 microbiota, BAL should not be performed before sampling and mouse cells should not be removed from the BAL fluid before extraction. CP673451 concentration Our results show

that there are fewer OTUs in the BAL-plus samples with mouse cells and that the lung this website tissues samples have a large variation. This suggests that the removal of tissues and host cells is a viable approach, when extracting DNA for the examination of the lung microbiome. Another challenge when working with low bacterial loads is the risk of contamination from the environment or sampling procedures. Some contamination must be expected and taken into account when interpreting data. We believe that we have taken large precautions to insure sterility during procedures and we have used excess controls to check

that our sampling procedure or experimental chemicals did not produce any sequences on their own LY294002 in the PCRs. Culturing of the BAL used for DNA extraction did not yield many bacteria either. Furthermore, our sequences were very consistent between mice. This would suggest that any contamination was either negligible or at least distributed evenly between mice. We did find large variation within the vaginal samples resulting in subclustering into groups we designated S1 and S2 (Figure 1C and D). S1 (vaginal samples 2, 5 and 8) was found to be much more distantly related to caecum and lung communities than the S2 group, which more closely resembled the lung microbiota. We believe this could be the result of a possible infection in the S1 vaginas, as these 3 samples contained 56-97% Streptococcus. In the present study, we did not monitor the stage of the estrous cycle at the time of sampling, which has been shown to change the bacterial profile of the vagina in animals and humans [28, 29]. Mice have a daily fluctuation in estrous cycle, which in part could explain the subclustering of the vaginal microbiota. This should be taken into account in futures studies.