The three-way crosses were designed to examine the possibility that multiple parents could be involved in generation of a single recombinant progeny. We saw no evidence of a three-way cross in any of our selection experiments or in any genome sequence analysis, even though multiple independent two-way crosses

were recovered from those experiments. If the probability of a three-way event is a function of the probability of two independent recombination events, it is likely that not enough individual recombinants were screened to identify an extremely rare progeny clone. There is, however, one issue that is addressed by the absence of any evidence for contribution of three parents in a selleck chemical cross. In many of the recombinants

identified by our group and in studies by Demars and colleagues [4, 38], multiple fragments from each parental genome are found in a recombinant progeny, often in regions of RXDX-101 purchase the chromosome that were not selected for with the tested antibiotics (Figures 3 and 5). It is possible that these differently recombined fragments involve sequential and independent recombination events occurring during the mixed infections used in this procedure. If involvement of multiple chlamydiae was a common occurrence in the generation of a cross, we hypothesized that some progeny from the three-way crosses should carry fragments of each parent. As no single progeny strain was identified with fragments of each of the parents in the three-way cross, our results do not support this hypothesis. Therefore our current model is that the generation of recombinant progeny is the result of a single exchange event between two parents, and that these exchanges can involve very large fragments of the chromosomal DNA. This latter result is consistent with analyses DNA ligase by other laboratories [4, 9, 33, 35, 38]. Subsequent recombination events will then lead to differential integration of fragments of the exchanged DNA, leading to the mosaicism

seen in many of the recombinants. The attachment efficiency in the absence of centrifugation measured for the different recombinants revealed groups having either a high attachment efficiency, as exhibited by LGV strains, or a low attachment efficiency, as exhibited by non-LGV urogenital strains (Figure 6). Genome wide association analysis of this phenotype revealed a number of loci that were quantitatively associated with the attachment efficiency phenotype seen in cell culture. While the list of candidate alleles that might be associated with this phenotype includes a wide variety of genes (i.e. type III secretion –associated ORFs [28, 29]), we focus this discussion on proteins known or hypothesized to be on the surface of the chlamydial elementary body.