The patterns of inhibition seen with the class II and class III Trx- LTTRs are inconsistent with a single nonspecific mechanism of this kind. The class III Trx-LTTR fusions collectively interact with all eight of the cI fusions, but there seems to be no obvious pattern for which cI fusions are sensitive to the overexpression of each class III Trx fusion. Within the subset of possible interactions we could test, the pattern of cross-interaction also does not seem to fall into disjoint clusters, as we might expect if specific interactions behaved as stable traits over the evolution of the LTTR family. This lack of pattern suggests that the cross-interaction among LTTRs may reflect independent evolution of interactions, which need not even involve the same interface residues. Consistent with this view, phylogenetic analysis of interacting and noninteracting pairs showed no evidence that the interactions we observed are generally correlated with evolutionary distance. Although half of the eight cI fusions did interact with their closest relative, all of them interacted with LTTRs that are more distantly related than proteins with which they failed to interact in our assay. This is not surprising for the highly divergent paralogs we tested with pairwise sequence identies ranging from 5.45% to 33.1%. Previous systematic studies of MG132 oligomerization in OxyR and CynR, showed that although the surfaces involved in homodimer- ization of their regulatory domains are superficially similar, distinct residues and residue interactions are important for oligomerization of these LTTRs. Cross-interactions could thus reflect the plasticity of subunit interfaces, allowing evolution to find different combinations of interactions to build similar quaternary structures. The physiological significance of the cross-interactions is unclear. It is formally possible that heteromultimeric LTTRs form functional transcription factors with different physiological roles from the homomultimers. However, we do not think this is likely in most cases of cross-interaction. Formation of heteromultimers could interfere with the normal function of LTTRs, just as it interferes with the l repressor fusions. However, it is likely that some heteromultimerization can be tolerated. Our assay cannot determine the relative affinities of LTTRs homomultimers vs. heteromultimers, so it is possible that homomultimers are favored and heteromultimers are only observable under our artificially high overexpression of the Trx fusions. But even if homomultimers and heteromultimers form with equal affinity, note the concentration of inhibitors in vivo may be inadequate to have a significant inhibitory effect. Transcription factors are often expressed at low steady state levels, of the thirty-eight LTTRs we examined, only two had detectable peptides in a mass-spectrometry based catalog of protein abundance in E. coli. As with all fusion-based systems, ours has limitations and is likely to have false positives and false negatives. Negative dominance occurs only if the inhibitory heteromultimerization can drive the concentration of MK-1775 an active homomultimeric cI fusion below the threshold needed to to block phage infection. Empirically, we find that this requires a large excess of the inhibitory Trx fusion to drive the equilibrium depicted in Figure 1 far enough to the right. However, the expression system used for the Trx-LTTR fusions could be so far above physiological concentrations of the LTTR that interactions that are not biologically relevant might be detected; the physiological levels of each native LTTR has not been determined. However, differences in the expression of the Trx-LTTRs is unlikely to account for the differences in promiscuity of the observed interactions.