4/4 is even better

Today I received the sequencing results from my second putative clone of the ICAP-9G,14C variant and it was good! Thus, I now have all of my ICAP variants and so can proceed with more bandshift analysis, while at the same time conducting CRP-Sxy bandshifts along with Sunita.

Further analysis of the first ICAP-9G,14C clone revealed that it lacked a restriction site where the sequencing primer was targeted to bind, consistent with the failed sequencing reaction that I described in my last post. That plasmid is now in the trash and I can file the experience away as another one of those funny little cloning complications.

3 out of 4 ain’t bad

(It’s better than Meat Loaf’s 2 out of 3.) This week I attempted to clone 4 modified versions of the ICAP CRP-binding site. In a previous post I described my attempt to use site-directed mutagenesis to change ICAP. That approach failed, so I returned to the method used to originally clone ICAP: complementary synthetic oligonucleotides containing the ICAP sequence (or variant sequences) were allowed to anneal and the resulting dsDNA 28 bp molecules were cloned.

As described in the earlier blog post, the ICAP palindrome poses a serious problem because of the formation of very strong hairpins, which directly competes with inter-molecular annealing. Consequently, cloning synthetic oligos has a very low efficiency (I was lucky with ICAP; the single positive clone had the ICAP sequence!), and this is why I had opted to try site-directed mutagenesis. This time around, oligos were melted at 95º and then given 10 hrs to cool to room temperature, an approach that should hold oligos at high temperatures where inter-molecular bonds are stable and hairpins are not.

This round of cloning was slightly more successful. Each ICAP variant had at least 5 positive clones (ie. white colonies in blue/white screening on X-gal) except for ICAP-9G,14C (2 white colonies), which has the strongest hairpin structure of all ICAP variants.

My sequencing results came in on Friday and all but ICAP-9G,20C came back positive as having the desired sequences. Unfortunately, there was no sequencing read for ICAP-9G,20C, thus I may in fact have the desired clone, I just don’t know it. I will talk to NAPS on Monday about what may have gone wrong. The strong ICAP secondary structure had been a problem in earlier sequencing reactions, but this appears to have been circumvented by addition of betaine to sequencing reactions. Even if the ICAP-9G,20C secondary structure remains problematic during sequencing, the reaction should have at least yielded sequence up to the hairpin; the “no sequence read” is a mystery, so I really hope the folks at NAPS have a good idea of what to try next. This weekend I extracted plasmid from the other ICAP-9G,20C positive clone, so I’ll take that in for sequencing on Monday as well. Given the success of the other three clones, I remain confident that ICAP-9G,20C will also turn out well.

Covariance in CRP sites

Yesterday in lab meeting I described how I will introduce base changes at various positions in ICAP to test which positions are most important for HiCRP binding (see previous blog). One important feature of CRP sites, both in E. coli and H. influenzae, is that they have A/T rich runs at either end (positions 1-3 and 20-22; see figure showing a sequence logo from alignment of 45 H. influenzae CRP sites). These A/T rich regions allow a small degree of bending in the DNA, which strengthens protein-DNA interactions. EcCRP does not directly “read” the base sequence at positions 1-3 and 20-22, instead, contacts between the protein and the phosphate backbone of the DNA are favoured when these positions are A/T rich because A/T runs are more flexible. Thus, there has been no reason to think that it matters whether there is a T or an A at position #1 (for example); either base will do. However, Heather made the great point in lab meeting that covariance between adjacent bases at positions 1-3 and 20-22 may be important and would not be apparent in a sequence logo. In other words, if position #1 is a T, it may be much better for CRP binding if position #2 is also a T; conversely, if #1 is an A, position #2 is more likely to be A.

I used MatrixPlot (which has been previously used in our lab for USS analysis, see here) to detect covariance in the 45 experimentally determined H. influenzae CRP sites. Absolutely no covariance was detected between any positions over the 22bp length. I think 45 sites should be adequate for detecting significant covariance. This result is good news because it means that when I make a point mutation in ICAP, I probably don’t have to worry that this same mutation would have a dramatically different effect on CRP affinity if it were introduced into a different site (as I had speculated before). In other words, base contributions to CRP binding may be additive and not cumulative, which makes experiments more straightforward and easier to interpret.

Mutating the ICAP binding site

In my last post I reasoned that I should introduce mutations into the perfect CRP binding site, called ICAP, to measure HiCRP’s preference for certain bases at certain positions within the site. This past week I did just that (I hope). However, the mutagenesis procedure presented some technical difficulties, prompting me to experiment with a new approach.

The big problem is that the ICAP site is a perfect palindrome (see the DNA sequence in the figure; the vertical dashed line shows the axis of symmetry) and the vector sequence on either side of the ICAP clone is also palindromic, resulting in a palindrome >40bp long. The problem arises because mutagenic primers are normally designed to be perfectly complementary and extend by >20 bases on either side of the mutation being introduced. I want mutations near the axis of symmetry, resulting in primers that in most cases are themselves perfect palindromes (purple primers in figure, labeled as “old method”; not to scale). Thus, it is highly favourable for primers to form hairpins and become double stranded along their entire length; these intra-molecular bonds form more readily than do inter-molecular bonds with template DNA.



Using pcr, the mutagenic primers generate full-length linear amplicons from plasmid DNA. Complementarity between the primers allows the newly synthesized sequences to anneal as double stranded circular molecules with staggered nicks, which are repaired by the host bacterium after transformation. I reasoned that the complementarity is key, but the mutagenic primers do not have to overlap for their full length. Thus I designed primers (green in the figure; not to scale) that have a region of complementarity with one another, but only one primer has the mutation and the other does not overlap this region. This allows me to design primers that contain short palindromes, thus greatly increasing their preference for annealing to template DNA and reducing the strength of intra-molecular pairing.

I was pleased to later discover that a biotech company now markets a mutagenesis system that uses the same approach of staggering primers, giving me confidence that this approach would work. I got the satisfying result today: my positive control of introducing a stop codon in pSU20’s lacZalpha gene had a 75% success rate (ie. 75% of colonies were white when plated on Xgal). This is lower than the >95% success rate I enjoyed with the old method (which used mutations in pUC18’s lacZalpha as a positive control), but all I care is that I circumvented the primer hairpin problem and got mutants. Unfortunately, my positive control mutagenesis is not exactly like my ICAP mutagenesis, so I will need to wait for sequencing results later this week to know if the experiment really worked.

Should I make a bad site better or a wonderful site worse?

In order to understand competence gene regulation in H. influenzae, we have been using the huge body of knowledge surrounding E. coli CRP to inform us about how the ortholog in H. influenzae functions as a transcription factor. Recently, bandshift analysis of H. influenzea CRP (HiCRP) binding to DNA has confirmed that both HiCRP and E. coli CRP (EcCRP) have the same perfect binding site (called ICAP, see an earlier blog about ICAP). However, I also discovered that HiCRP is much more selective about which sites it will bind; for example, HiCRP will not bind to E. coli’s lacZYA promoter, even though EcCRP has very high affinity for this site.

Eleven amino acids in EcCRP’s helix-turn-helix DNA-binding domain make direct contact with bases and/or phosphates in the ICAP DNA site; EcCRP makes fewer direct contacts with the lacZYA CRP site because some of the base pairs in this site are not optimal. HiCRP’s highly selective binding behaviour suggests that it needs to make a greater number of direct contacts with the DNA to manifest a binding strength that EcCRP can accomplish with fewer interactions. In other words, more amino acids in HiCRP have to make specific base interactions and/or non-specific phosphate interactions with DNA for the protein to be able to hold on.

I have already identified a single mutation (lacZ-19A) that converts the lacZYA site into something that HiCRP can bind. However, HiCRP still binds to ICAP with >100-fold greater affinity than binding to lacZ-19A. Now it becomes a simple question of which bases in lacZ-19A prevent perfect binding by HiCRP? Six out of twenty-two bases differ between ICAP and lacZ-19A. Three of these differences are likely irrelevant or make minor contributions to binding, so I will ignore them at first. Of the remaining three, it is possible that no change alone will have a dramatic difference but that a particular combination of two key bases will have a dramatic effect. Thus, multiple possible permutations of mutated lacZ-19A sites arise, and it will be a lot of work to test all of them (so I won’t).

So, the simplest experiment is to make 3 single point mutations, but do I change ICAP to resemble lacZ-19A and measure the deleterious effects on binding, or do I change lacZ-19A to resemble ICAP and measure the positive effects? Because there is such a dramatic difference in HiCRP affinity for these two sites, I have good power to detect beneficial or deleterious changes. I am inclined to work with ICAP because I can measure the effect of each mutation while all other bases remain “perfect”. With lacZ-19A, I don’t know if the mutations I introduce would have greater or lesser effects on affinity if studied in the context of other naturally occurring CRP sites. The key question remains “which bases in a CRP site are most critical for HiCRP binding?” and not “why is the lacZYA site such a crap site for HiCRP”. Therefore, I will work with ICAP, and hopefully this will provide insight into the lacZYA issue.

CRP and Sxy plans

Having completed all but a few final touches on the revised Sxy manuscript, I can now return my focus to the next most pressing piece of science/manuscript: an investigation into the molecular mechanisms by which CRP binds and activates CRP-S promoters. The initial (already submitted to and rejected by a journal) version describes experiments that compare how E. coli and H. influenzae CRP bind to DNA, followed by experiments examining how converting a CRP-S site to a CRP-N-like site influences CRP binding and transcription initiation, then some experiments testing Sxy dependence at CRP-S promoters, and finally experiments designed to test the influence of putative UP elements in CRP-S promoters.

This manuscript began life as a chapter in my thesis designed to pull together several good, but not very cohesive, experiments. New experiments conducted over the past few months have allowed this manuscript to evolve into a more thorough analysis of how CRP binds and activates CRP-S promoters. The constantly evolving nature has paved the way for some very neat experiments, however this means that the manuscript won’t be done for another couple of months yet. On the plus side, it will be a much more cohesive manuscript and the science will be more complete and exact (no more hodgepodge of results).

I now feel that the manuscript should be only about CRP and that experiments testing Sxy’s role in DNA binding and transcription activation should be saved for a Sxy-specific manuscript that combines my work with Sunita’s. I think Sunita’s excellent molecular biology skills are bringing us rapidly towards the ability to directly test how Sxy functions on a molecular level. Furthermore, I have recently acquired some V. cholerae DNA which I hope Sunita can use in her Sxy and CRP cross-species complementation experiments. Because very few regions of the Sxy protein are conserved between H. influenzae, E. coli and V. cholerae, if all three proteins work in all three species, this result will very quickly narrow down which regions of the protein are required for whatever function we discover Sxy to have. She and I will meet tomorrow to discuss our experimental and manuscript plans.