Table 1 details the total number of cultured isolates obtained over the course of the study based on the16S rRNA gene sequence analysis and visible colony morphology. The lysozyme extraction protocol effectively isolated bacterial DNA from all the colonies picked from plates. Typical DNA yields for bacterial colonies were in the range of 0.6 to 82.4 ng μl-1. Table 2 shows the results of culturing on the environmental swab samples taken between September 2005 – April 2006. For all days sampled except October 6, 2005, colonies grew on either blood or nutrient agar plates. This means that there were large numbers of viable bacteria consistently present on the surfaces and toys sampled at the daycare center. We isolated as many as 29 putative bacterial species. We considered it a potentially different species of bacteria if it had a unique 16S rRNA gene sequence or a clearly distinct morphology or both (Table 2). Bacillus species were the most commonly culture-isolated bacteria, followed by Staphylococcus spp. Culture methods identified at least 29 viable bacterial species on toy and furniture surfaces over the 6 months of the study (Table 1). Species of Bacillus were isolated every day of sampling (Table 2), and we identified as many as 15 different distinct morphologies over the course of the study (Table 1).
The lysozyme DNA extraction protocol proved effective for direct swab extractions and yielded DNA in the range of 5.1 to 14.7 ng μl-1. Bead-beating methods are typically preferred for isolating environmental DNA because the mechanical shearing of cells by the beads helps extract DNA from particularly "tough" bacterial cells such as bacterial vegetative cells (e.g., Pseudomonas putida), bacterial endospores (e.g., Bacillus spp.), and fungal conidia (e.g., Fusarium moniliforme) . However, our attempts with bead-beating methods failed to isolate sufficient DNA from the swabs (data not shown), whereas the lysozyme method yielded sufficient amounts of DNA for PCR.
Figure 1 and 2 show the results of phylogenetic analysis of selected sequences from the 453 clones obtained from the nine libraries relative to sequences from cultured and uncultured bacteria from other studies. Based on our survey of nine 16S rRNA gene PCR-clone libraries we identified as many as 190 bacterial species (1% divergence; 141 at the more conservative 3% divergence level), most of them with no cultured representatives. The clone library sequence coverage ranged from 48% to 65% (average 54%) for the nine clone libraries. Most sequences found in both groups appeared to be uncultured bacterial species. Members of the Pseudomonadaceae and the Oxalobacteraceae predominated in the clone libraries (Fig. 3). Pseudomonads were particularly abundant and were on every surface sampled on every sampling occasion (Fig. 3).
Since the main purpose of this study was to identify the types of organisms in the daycare center, one-directional sequencing of the first part of the small-subunit rRNA gene, which includes the most variable part of the gene, was enough for our purposes. Sequencing bi-directionally would have been ideal in terms of reducing error, but would have doubled the sequencing costs and added little to our understanding of the diversity. Also, we trimmed the sequences to around ~500 bp in length, and edited out the most problematic part of the reads. Any errors that remained would have had little impact on the phylogenetic analysis. However, we have made glycerol stocks of all the clone libraries we created, which are available from the authors upon request.
The alignment of the edited and trimmed sequences proved straightforward, and the alignment was checked by confirming complementary base-pairing in known stem regions of the alignment. A total of 78 sequences were deposited in GenBank, including sequences from both culture isolates (Table 1) and representative clone library sequences used in the phylogenetic analysis (Fig. 1, 2). Approximately 500 nucleotide positions, corresponding to E. coli positions 25 to 534, were used for all the phylogenetic analyses. There was strong support for the majority of the phylogenetic relationships as judged by both Bayesian posterior probabilities (Fig. 1, 2) and MP bootstrap support (not shown). Bayesian, MP and ML methods produced highly similar tree topologies. The differences in tree topologies produced by the various methods were in regions of the trees not supported by either posterior probabilities (1, 2). Using this information, and the Fastgroup II dereplication information, we were able to assess the relative abundance of various sequences in clone libraries and these are presented in Figure 3.