Sunday, August 26, 2007

Happy to have been wrong about HiCRP

I conducted some very satisfying bandshifts this week, although I haven't found time to crunch and analyze the data. Thus, I will only be able to briefly describe my first impressions of the results. The first satisfaction came from confirming that my protein preps were very active in DNA binding; because preparing the proteins takes several days, I always live in fear through much of the preparation process that the final protein preps will be bunk.
The second satisfaction came from discovering that, contrary to earlier conclusions, H. influenzae CRP (HiCRP) appears to have a very high affinity for DNA, comperable to that of E. coli CRP (EcCRP). This revelation comes from testing both proteins' affinities for the theoretically optimal CRP-binding site, called "ICAP". In earlier experiments conducted in 2002 and 2005, I tested CRP affinity for a natural CRP site from the Hflu mglBAC promoter; this site was my positive control for DNA binding because it is very similar to the optimal site (ICAP). EcCRP bound the mglB site with very high affinity, but HiCRP did not. From this, I concluded that HiCRP just was not as strong a DNA binding protein. I hypothesized that HiCRP may be a weak DNA binding protein because it does not form very stable dimers (only homodimeric CRP molecules bind DNA). I think that my ICAP results may offer some support for this theory. Although HiCRP and EcCRP demonstrated a similar affinity for ICAP when they were in the 10-500 nM concentration range, my impression was that HiCRP binding was not detectable below 10 nM, whereas EcCRP binding could be detected down to 0.5 nM. In my thesis I lamented that I did not know of a good assay to detect dimerization of CRP molecules at the low concentrations where dimerization would be an important contributer to CRP's ability to bind DNA. However, bandshift assays are very sensitive to inter-molecular interactions at very low protein and DNA concentrations, so perhaps I will be able to glean some more informative data from my bandshifts than I had initially intended. This will require that I get a better handle on the calculations used and underlying assumptions made when measuring protein-DNA binding kinetics.

Monday, August 20, 2007

More bandshifts....

The protein is pure and concentrated, the DNA is hot, and the scientific questions always abound, so tomorrow I start running bandshift gels (many, and perhaps more than humanly possible at one time). Fingers crossed for beautiful bands. Tomorrow, a blog about results (finally)!

Sunday, August 12, 2007

Using bandshifts to study CRP-DNA interactions

Along with cloning new pilABCD promoter constructs for testing the role of putative UP elements in CRP-S promoters (see July 30 post), I am purifying E. coli and H. influenzae CRP proteins to test their in vitro affinities for various gene promoters. I have been conducting similar bandshift experiments off and on throughout my graduate studies and have found some very curious and unexpected results (these unexpected results are undoubtedly part of the reason this work has been sporadic; it's sometimes very hard to know what to conclude and what to do next). In short, I found that E. coli CRP behaves, well, like E. coli CRP, which is all well and good. However, H. influenzae CRP (a highly similar ortholog with the same cellular role) exhibits dramatically different binding behaviour. For example, E. coli CRP has very high affinity and specificity for the lacZYA CRP-binding site, whereas H. influenzae CRP does not treat this site any differently from random DNA sequence. As a first step to figure out why, I have mutated the lacZYA CRP site to better match the CRP-binding site consensus sequence (only a single base pair change makes the lacZYA site nearly "perfect"). If this mutation is sufficient to change this site to a specific H. influenzae CRP-binding site, then we will know that the Hflu protein is highly selective for a specific base pair at position 19 in the 22bp site.

Tuesday, August 07, 2007

Peter, Peter, DNA eater

During her research into the distribution of the competence phenotype in H. influenzae strains, Heather has discovered a strain, PittAA (which I call Peter), that demonstrates transformation frequencies 1000 fold greater than the reference strain (KW20) when cultured in rich medium. Specific mutations in the sxy and murE genes (and possibly also in crp) can cause similarly elevated transformation frequency in KW20, a phenotype we call hypercompetence. If we can track down the genetic cause(s) of PittAA’s hypercompetence, it may help us better understand competence gene regulation in our preferred organism, KW20. Can we track down a region of the PittAA genome that confers hypercompetence when transformed into KW20?

1. Is PittAA sxy responsible for hypercompetence?
Heather has described the amount of difference between the PittAA and KW20 sxy genes: Unfortunately, we don’t know anything yet about how the Sxy protein works, so we don’t know if certain amino acid differences between PittAA and KW20 Sxy are functionally significant.

In KW20, mutations that destabilize the secondary structure of sxy mRNA cause hypercompetence. Examination of the homologous sequence in PittAA revealed three base differences compared to KW20: one in Loop C, one in Stem 3, and one in Stem 1A (see figure of KW20 structure). A base difference in Loop C is not expected to have an effect on 2º structure, and the differences in the stems conserve the base pairing potential seen in KW20 (C:G→U:G and A:U→G:U; differences between KW20 and PittAA are in bold). Therefore, non of these 3 sequence differences are expected to destabilize the mRNA 2º structure and thus do not readily explain the hypercompetence phenotype. Though unlikely, these base differences in PittAA may favour the formation of a different 2º structure than what we have characterized in KW20. I will see if the RNA-folding program Mfold predicts an alternate, but thermodynamically favourable, folding structure for PittAA sxy mRNA. One caveat is that we don’t know for sure whether transcription of the PittAA sxy gene initiates at the same base as in KW20; I will check for the position of the CRP site and RNAP binding sites in the PittAA sxy gene promoter and this should help us predict where transcription initiates in PittAA.

2. Is PittAA crp responsible for hypercompetence?
One possibility is that the PittAA CRP protein is constitutively active (ie. cAMP independent); in E. coli, multiple mutations have been identified in crp that make CRP constitutively active. To test whether PittAA CRP responds to cAMP, cAMP can be added to exponentially growing PittAA cells. Addition of cAMP is expected to increase transformation frequency, and if it does, we know that CRP activity is still limiting to competence gene induction during exponential growth.

The most exciting possibility is that PittAA has gain-of-function mutations in Sxy or CRP that allow one protein to act at CRP-S promoters in the absence of the other. If we could identify such a mutation(s), it would tell us a lot about how CRP and/or Sxy activate transcription at CRP-S promoters. To test whether PittAA has GOF mutations in CRP or Sxy, we can transform PittAA with KW20 crp- and sxy- constructs, but would have to use PCR amplified knockout genes in order to avoid transforming other parts of the PittAA genome. We predict that clean knockout mutations in PittAA crp or sxy will abolish transformation, as they do in KW20. If not, than a GOF mutation is the best explanation.

It is tempting to hope that only one genetic difference can explain the “non-KW20” phenotype in each H. influenzae strain that is non-competent, non-transformable, or hypercompetent. However, it seems likely that non-competent strains will have accumulated mutations in unused competence genes. On the other hand, a hypercompetence phenotype and a non-transformable phenotype can each be generated in KW20 with single mutations. Thus, it should be tested whether KW20 genes can "restore" the KW20 phenotype in the non-transformable strains.