More about the reduced variation at USSs

The analysis I showed in my last post deserves more discussion. First, it could be improved in a couple of ways. Second, is this an expected result for sequences subject to a molecular drive resulting form biased DNA uptake and unbiased homologous recombination?

Improvement one: better controls. As it stands, the comparisons of variation at USS positions is compared to adjacent positions that are not part of the USS motif. This is a bit weak, because some of these positions may contribute to DNA uptake but not show up as 'motif'. A better control comparison would be with random segments of the genome. This is easy to do, as I already have a set of 3500 segments of 39bp (like the USS-centered segments) that I used as a control for the analysis of covariation. Well, 'easy' to see how to do (convert this set to a BLAST database, query it with the three genomes, and extract the alignment positions), but it's still a pain doing the Word-Excel shuffle to extract the position information from the six BLAST searches (each strand of each genomes). But this control would let me draw a band across the graph, probably at about 2% variation, indicating how much variation non-USS sequences have.

Improvement two: more data. Some of the error bars are uncomfortably large, because the datasets don't contain a lot of variation (rarely more than 20 mismatches at any one position even over the ~5000 alignments). Using more datasets would help. Only three complete genomes are available through GenBank, but incomplete versions of 9 more genomes are available. I should concatenate the few largest contigs of at least some of these and use them as queries to the USS database.

Improvement 1.1? Controls with more data? If I do the analysis with more genome sequences, should I also query the control database with them?

What does it all mean? I think we need to start by considering the implications of the molecular drive model for sequence variation. (This is the simplest explanation, invoking forces we already know are acting.) Let's start by imagining that the population of H. influenzae genome sequences are at an evolutionary equilibrium. This means pretending that the environment is unchanging and that all genes have evolved to their optimal sequences at mutation-selection equilibrium (that is, mutations that arise are subject only to purifying selection). Pretend too that the USSs in the genome have also evolved to equilibrium, with the molecular drive balanced by both random mutation and any purifying selection the sequences may be subject to due to their cellular functions.

Consider first some specific location of the USS motif, at a place in the genome that has no cellular function at all. It undergoes random mutation, but its recombination is biased by the specificity of the DNA uptake system. Is this sequence expected to have less 'standing genetic variation' than non-USS positions? Will biased DNA uptake act like stabilizing selection, purging genetic variation that reduces the uptake of the segment?

We need to consider the diversity of this specific sequence in a diverse population of H. influenzae cells. To start with the simplest case, assume that this sequence has the optimal uptake sequence, perfectly matching the bias of the uptake machinery. Also assume that most other cells in the population have the same sequence at this USS site. If it mutates at any USS-motif position, it will be further away from the preferred USS and it will preferentially get converted back. I think this means that biased uptake/recombination acts like stabilizing selection, reducing genetic variation.

What if the sequence isn't a perfect-consensus USS? This requires thinking about a more complex situation. Maybe it isn't 'perfect' because it's already been hit by several mutations but not by replacement. Because both processes are stochastic, some sequences will 'get lucky' and some won't, especially if the molecular drive is only a bit stronger than random mutation (so our Perl model tells us). But assume that this is still the usual version of this site in the population, i.e. the available DNA will usually have this sequence. If this sequence in our cell mutates to a worse matches, it will still be preferentially restored by uptake/recombination, purging the variation. If it mutates to a better match, and then dies, its DNA has a better than average chance of being taken up and recombined into another cell, which I think would increase variation.

I think this means that, if we were to examine many individuals for their sequence at the same particular USS site, we expect to see reduced variation at positions that match the USS consensus, and increased variation at positions that don't match it. But that's not what the analysis I've just done looked at. Instead it summarized the variation at each position across ALL of the USS sites in the Rd genome. Most of these positions did match the consensus, so it's not surprising that, on average, they showed reduced variation.

OK, I seem to have convinced myself that reduced variation is what we should expect at USS sites, if they are being maintained by biased uptake and recombination. But my evolutionarily savvy post-docs are suggesting other causes, so we'll need to put our heads together to see if we can reach agreement.

And the result is....

Between-strain variation in nucleotide sequence is greatly reduced at positions that are part of the USS motif. This is clearly seen in the figure to the left, where the blue bars representing the amount of variation for each position are small at the positions where the motif bases are tall (strong consensus). The error bars are the standard deviations of the six datasets (forward and reverse strands of the three readily available genomes) This was about 5000 alignments.

For the flanking AT-rich segments the magnitude of the reduction is proportional to the strength of the motif consensus, but variation in the USS core is most strongly reduced in the initial As.

This result was predicted last year by one of the post-docs, but at that time I didn't see an easy way to test it. How should we interpret it? Perhaps it means that most of the USSs in the genome have evolved to the optimum compromise between the preference of the uptake machinery and the needs of within-cell functions. Why don't the differences in % variation map precisely onto the logo?

BLAST success, analysis success

Everything is working.

I still can't get BLAST to attend to matches close to the ends of the 39 nt fragments, but I'm treating these as mismatches at the innermost position and 'no information' at positions closer to the end. For example, if a sequence matches at positions 4-39, I assume there's a mismatch at position 3 and that I have no information about positions 1 and 2.

I'm searching for the two USS orientations separately (searching the forward and reverse strands of the query genome separately). I'm analyzing the data separately. So far I've analyzed only the forward searches, but I'll need to flip the results I'll get from analyzing the reverse searches.

I'm collecting the output as pairwise comparisons between query and USSs, because this makes it easiest to pull out the positions of the mismatches without the information about what kind of a mismatch each is.

I'm doing the analysis by bouncing the file back and forth a couple of times between Word and Excel, using Word to first insert tabs between the output lines, and then Excel to delete the columns (formerly lines) I don't want (including the actual sequences) and to sort the results by both the match score and the locations of the ends of the matches. Then I use Word to insert tabs between each position of each alignment, and Excel to count the numbers of mismatches at each position and to graph the results.

Next post I'll describe the results.

I can run BLAST locally!

A few suggestions from the post-doc ("Try keeping all the BLAST files in the same folder.") got my Unix problems solved. Commenters on my last post have provided lots of suggestions, some of which are useful and some of which address problems I've already solved. Neil has even set up a NodalPoint page about this problem, but I haven't yet figured out how to edit in my comments.

I've now used standalone (local) BLAST to search the database of 2136 different 39 nt segments of the H. influenzae Rd genome with the source genome (Rd) sequence. This is the 'positive control' for the searches I want to do with the non-Rd genome sequences. This worked, and let me do some trials to work out which options I should specify. Then I did the same searches with the three other genomes I have. This showed me other problems, some of which I haven't worked out yet.

I have two big remaining problems.

First, I need to understand BLAST searches well enough to optimize the alignments. I understand that mismatches close to the ends of the sequences will be under-represented, because of how BLAST works. I think this appeared in the search results - instead of alignments with single mismatches near the ends I think I got alignments that had been shortened by 4 nt. I may be able to minimize this by setting the options appropriately, but I probably can't eliminate it. Luckily the first and last 5 nt are the least important for this analysis.

Second, the information I need to get from the non-Rd searches is the locations of mismatches within the 39 nt segments. For this I found that representing the output as pairwise alignments made easiest to extract the information that specifies the positions of the mismatches within the alignments. This is relatively straightforward (yes Neil, with Word and Excel) provided the mismatches are internal to 39 nt alignments, but will need some more sophisticated tricks for alignments that are truncated at one or the other end. Another problem for extracting the position info is that half of the 39 nt sequences are in the opposite orientation to the others, and so they are reversed in the output.

The guy in the next office strongly recommended BBEdit, so I bought that. He said it does a lot of the editing chores that I'd otherwise need to write Perl scripts to do. Sounded great. But BBEdit has wandered far from its "Bare Bones Software" (="BB") roots, and learning how to use it will take some time...

Preparing to run BLAST locally

I asked the guy in the next office for advice about using BLAST to characterize the pattern of variation in and around USS sites in the sequenced H. influenzae strain genomes. His advice was to set up BLAST on one of our own computers (not a big deal, he said), to create a BLAST database containing all of the ~2000 USS sequences (39 nt each) from the H. influenzae Rd genome, and then to 'query' this database with the genome sequences of each of the other strain genomes in turn. This has the big benefit of creating only one output file for each genome, rather than the one file for each SS site I would have if I used the 39 nt USS sequences as queries.

Finding the downloadable files at NCBI was quite easy, and I now have a BLAST folder in the applications folder of our fast Mac (Stingray). I also had no trouble converting my file of ~2000 USS sequences into FASTA format, with each entry given an identifying number (I don't need that but BLAST does). And I downloaded the genome sequence files - only 3 other H. influenzae genome sequences turned out to be directly available from GenBank; the others apparently still must be obtained from the various sequencing centers. One of the post-docs may already have these, but if not I think I don't really need data from more than three anyway.

I also found the instructions for formatting commands for BLAST searches, and (must do first) how to use the 'formatdb' program to convert my FASTA file into a recognized BLAST database file. BUT, as usual, I'm stuck at the very basics of using Unix commands in the Mac Terminal interface. I'm hoping one or other of the post-docs will be able to set me straight today.

Molecular drive on USSs drives tripeptide frequencies!

Here's a nifty new result. I was working on the Results section of the USS manuscript revisions, to include a result I posted about last month.

I had found that the distribution of coding-region USSs between the six possible reading frames paralleled the abundance of the tripeptides these USSs would encode in the H. influenzae proteome. This suggested that the reason some frames had more USSs than others was because USSs in some frames encoded more versatile tripeptides than others.

I now decided that this analysis needed some controls, so I scored the frequencies of the same USS-encodable tripeptides in the proteomes of several other bacteria: E. coli, B. subtilis, and (because it has the same base composition as H. influenzae) Listeria monocytogenes. The relative proportions of these six tripeptides seemed comparable to those of H. influenzae, but I noticed that the raw numbers were smaller than I would expect given the sizes of the respective proteomes (all the control bacteria have larger genomes than H. influenzae). So I normalized the tripeptide counts by dividing by the total number of amino acids in each proteome, to get what I could call tripeptide density.

The results are in the top graph. The blue bars are the H. influenzae tripeptide densities, the other colours are the tripeptide densities of the controls. Note that the H. influenzae proteome has disproportionately high densities of the USS-specifiable tripeptides.

So the 'control' needs another control - the densities of tripeptides that aren't encodable by USSs. I chose these by taking the same amino acids in backwards order (e.g. VAS instead of SAV). This is good because it doesn't change the abundances of the single amino acids making up the tripeptide. Here's that graph. Again the blue bars are H. influenzae and the other colours are the control proteomes. And note that now the blue bars are nothing special - H. influenzae has the same densities of these tripeptides as do the control proteomes.

So what does this mean? It means that the biased uptake and recombination of USS-containing DNA fragments has not been neutral for the proteome. USS not only accumulate at protein-coding positions where they don't change the amino acids, they accumulate at positions that wouldn't otherwise have used these amino acids.

My collaborator in Ottawa has already shown that the corresponding 'preferred ' USS-encodable tripeptides are overrepresented in proteomes containing USS. What I've done here is extended that result to show that even the tripeptides we thought were least versatile have become overrepresented in the H. influenzae genome.

Clarification of the analysis I want to do

Commenters are offering help and advice! Damn, I love open science! But I need to clarify what I'm trying to do.

I have a FASTA file with about 2000 segments of the H. influenzae strain Rd genome, each exactly 39bp (a few are shown on the left (yes, I know this is not FASTA format)).





Genbank has genome sequences of 12 other H. influenzae strains (shown on the left). I want to take each of the 2000 Rd sequences and find the sequences of its homologs in the other strains' genomes. If all the genomes have homologs of all the sequences, that will give me, in principle, 24,000 39bp sequences.






I've done this manually for a few of the 39bp sequences. Often BLAST says it found no matches - I don't really understand why.

When BLAST agrees to do what I want, it produces alignments like those on the left. The upper one shows an Rd sequence to which all the genomes have perfect matches; the lower one shows an Rd sequence several of whose homologs have a single mismatch to (yikes, is that syntax correct?).

What I need to do is automate this process, so BLAST searches the 12 genome sequences with each of the 2000 queries. Ideally the output will be in a form that I can easily use for the next step.

Because each of my ~2000 query sequences is centered on one of the ~2000 USS motifs in the Rd genome, they can all be aligned with each other (look back at the top figure to see this). What I want to do is create a meta-alignment of all the 24,000 homologs to these 2000 sequences. Then I will use this meta-alignment to score, for each of the 39 aligned positions how often one of the homologous sequences has a mismatch to its query sequence. That is, I want to align all 24,000 sequences (or all 24,000 patterns of dots), and create a histogram showing how often each position differs from its Rd homolog.

Only some of the 39 positions make significant contributions to the USS motif; seven internal positions and the five at each end are not part of the motif. I hope this analysis will show me whether positions that contribute to the USS motif are more or less variable than nearby USS-independent positions.

Maybe we CAN do this analysis...

One of the reviewers of our USS manuscript asked that we compare the sequences of the USSs we'd found in the Haemophilus influenzae Rd genome with the homologous sequences in the genomes of other sequenced H. influenzae isolates. We've been planning to dismiss this as both too difficult ("beyond the scope of the present work") and unlikely to yield any relevant information. I still expect that such a comparison will probably show that the USS sequences are just like the rest of the genome, with a few differences here and there. But I've just realized that the analysis might be much easier than I had thought.

The Gibbs motif search provided us with a file containing about 2000 sequences, each about 40bp centered on a USS. In principle it should be straightforward to use BLAST to search the other available H. influenzae genome sequences for matches to each of these USS sequences. The USS sequences are long enough that truly homologous sequences will be found; I expect most of these to be perfectly matched, with the occasional single-base mismatch.

I think I can easily use my highly developed 'find and replace' MS Word skills to convert the list of USS sequences to the FASTA format needed by BLAST. I don't know how to set up a search that uses multiple input sequences to search a few genomes, but I'm pretty sure that the guy in the next office will be able to help me. I'm not sure what form the results will take, and this may be the complicating factor. The standard BLAST display isn't designed for this kind of analysis. But output will probably be provided as a text file..

OK, I'm doing a test run, searching for one fake USS sequence against the 13 (wow!) H. influenzae genome sequences in the BLAST microbial genomes database. Result: "No significant similarity found. For reasons why click here." OK, I should be using the 'Search for short nearly exact matches' rather than regular BLAST. Hmmm..., various complications. I think I need to try this using a real USS sequence rather than a made up one. I'll play around more later.

USS manuscript angst

I'm not making much progress on revising the USS manuscript, I think partly because, deep down, I'm worried that our logic is a bit weaker and more circular than I'm happy with. So I'll try to lay it out below.

Our goal for our USS research is to find out whether these sequence motifs are abundant simply because of the molecular drive caused by the biased DNA uptake machinery, or because natural selection favours cells that take up DNAs from closely related cells/species. This latter explanation arises from the common assumption that cells take up DNA because recombination generates benefits, and is inconsistent with our proposal that cells take up DNA simply as food.

We seem to be trying to discriminate between these explanations by finding out whether the bias of the uptake machinery matches the motif accumulated in the genome. But I think this is not a good test, because such a correspondence is consistent with both explanations. Furthermore, the DNA uptake data in the manuscript aren't compelling enough to push this very weak argument.

But why am I thinking of doing a big rewrite at this stage? Why don't we just make the minimum improvements necessary to satisfy the reviewers and send the damn thing back in? Now that I've reread the whole manuscript and reminded myself of what we said, I guess I'd better reread the reviews, and the responses suggested by the post-doc author, and see if I think we can do this.

Revising the USS manuscript

Our manuscript about the within-genome diversity of uptake signal sequences (USSs) and their effect on DNA uptake was provisionally accepted, but we need to do substantial revisions to satisfy the reviewers, especially as we really don't want to have to do more experiments for this manuscript. On rereading the Introduction, I realized that changing the perspective might be good.

In the present version of the Introduction we do the following (each point is a paragraph):

1. Introduce natural competence.
2. Describe the general process of DNA uptake.
3. Explain that H. influenzae and Neisseria preferentially take up DNA fragments with specific sequences.
   
4. Describe how the H. influenzae uptake specificity was determined.
5. Describe how the USSs in the H. influenzae genome were characterized once the genome sequence became available in 1995.
6. Describe the USSs in the genome and possible non-uptake functions for these USSs.
7. Describe the distribution of USSs in the genomes of related bacteria.
8. Say that USSs arise by point mutation (not insertion) and that molecular drive can explain their accumulation in the genome.
9. Provide a mechanistic explanation of why the DNA uptake machinery might be biased towards a specific sequence.
   
10. Explain why we need a better understanding of the USS motif and the specificity of the uptake machinery.
11. Summarize the results that the manuscript presents.

My alternative perspective would frame the results in the context of repeated sequence evolution rather than of competence. It would go as follows:

1. All genomes contain repeats, which usually fall into one of three kinds: dispersed multi-copy elements that spread by copying and insertion (selfish DNA); motifs that arise by point mutation and interact with specific cellular DNA-binding proteins (e.g. binding sites for transcription factors); and simple-sequence repeats that in bacteria arise by polymerase slipping and function to promote gene activation/inactivation by frameshifting. This paper is about an abundant repeat that doesn't fit any of these.
   
2. These repeats are called uptake signal sequences. Describe what's known about length, copy number, consensus, distribution in genome and between species. How we know they are motifs and not insertions.
   
3. Possible within-cell functions and relevant information from bioinformatics.
4. Occur only in a few naturally competent bacteria. Discovered because are preferred by uptake machinery. Similarity between preferred sequence and genome abundance can't be coincidental.
   
5. Describe competence. Usual explanation for USS is selection as mate-choice marker. Why this is problematic.
6. Describe alternative explanation = molecular drive: USSs accumulate as a side effect of uptake bias and recombination.
7. Provide a mechanistic explanation of why the DNA uptake machinery might be biased towards a specific sequence.
   
8. Explain why we need a better understanding of the USS motif and the specificity of the uptake machinery. Focus on molecular drive as an important phenomenon in its own right, as well as on why cells take up DNA.
   
9. Summarize the results that the manuscript presents.

I think this might be a better outline - appealing more broadly to people interested in genome evolution as well as to those interested in competence. But I wonder if it's worth making this big a change to the manuscript at this stage.

Should we write a proposal to NIH?

Yesterday one of the post-docs and I discussed whether we should submit a proposal to NIH. Yes, we did just get one proposal funded, to work on the regulation of competence genes in H. influenzae and E. coli. But her project is completely different, and some parts of it are going to be expensive.

She's studying when, how and why different lineages of H. influenzae lose (or maybe gain) the ability to take up DNA and recombine it into the chromosome. We already knew that this occurs in various bacteria, and she's now completing a thorough analysis of the variation in DNA uptake and recombination ability in a broad selection of H. influenzae strains.

Some of these strains were chosen because their genomes have been or are in the process of being completely sequenced (one of the benefits of working on a sometime pathogen), and her next goal is to analyze these sequences for differences that could explain their different phenotypes. This bioinformatics work won't cost much except her time; we've already bought a nice fast computer for it. And her time doesn't cost the lab anything, because she's supported by a lovely post-doctoral fellowship from NIH.

But the next steps will be expensive. She wants to use the bioinformatics information to design investigations into the genetic differences of strains that haven't been sequenced. Her original plan was to develop a microarray chip containing all the genes and alleles that the bioinformatics and other work suggested might be involved. This still seems like a good approach, but the field is changing so fast that better ways to survey genomes are becoming available faster than we can keep up. One thing they have in common is that they'll all cost a bundle.

Subsequent work will also be pricey. We'll probably want to follow up the H. influenzae findings with investigation into related bacteria. This will be beyond the scope of the present post-doc, so we'll need new post-docs or grad students or technicians, as well as money for the tools and techniques.

The other reason to write a proposal to NIH is that proposal-writing is the best framework I know of for clear thinking about research plans.

Why I don't work on sex in eukaryotes

I just got back from the SMBE meeting and the final meeting of the CIfAR (formerly CIAR) Evolutionary Biology program. I made a point of going to most of the sessions on deep phylogeny, and they reminded me of why I decided not to work on the evolutionary origins of sexual reproduction in Eukaryotes.

John Logsdon's talk reminded me of why and how I originally was planning to work on this. "The Evolution of Sex" is still one of the biggest unsolved problems in evolutionary biology; we don't know how or why eukaryotic sexual reproduction (by meiosis and gamete fusion) originated and we don't know why it's been so evolutionarily successful. Most research into this is theoretical, or involves comparing populations that reproduce with and without sex.

My plan (10 or 15 years ago) was to instead investigate the origin of sexual reproduction by first identifying the deepest-branching evolutionary lineage that has sexual reproduction or homologs of meiosis-specific genes, and then determining the functions of these homologs and/or of sexual reproduction in the present-day members of this lineage.

When I first made this plan, the field of deep phylogeny was full of optimism that we would soon have a reliable phylogenetic tree showing the true relationships of all eukaryotes. The figure to the left shows the state of the tree then. The groups called Diplomonads, Trichomonads and Microsporidia were thought to be the first to have branched from the line that eventually led to plants and animals. Microsporidians were known to have (various weird forms of) sexual reproduction, while sex was unknown in Trichomonads and Diplomonads, consistent with an origin between the times of these divergences. The precise branching order wasn't yet firm, but most researchers felt that all we needed was a few more sequences of a few key taxa, and maybe a few refinements to the analytical methods used to infer relationships from sequences.

But the adjacent figure shows the present state of this phylogeny. It's from a recent review by a collaboration of CIfAR researchers, all but one of whom were at last week's meetings (Keeling et al. Trends in Ecology & Evolution 20:670-676). There's a pdf here; I think it's open access.

The new tree organizes the eukaryotes into five main lineages, with no real information about the order in which these groups originated (the 'root' (the center of the figure) is shown as what's called a 'star' phylogeny). The division into five groups is itself very speculative, as there are very deep divergences within each group. Bottom line? we don't know which eukaryote lineages arose first, and few researchers are now confident that we will ever know.

And where is sexual reproduction in this tree? Just about everywhere; the apparently asexual groups are dispersed with the sexual groups. So I don't regret my decision 7 years ago to abandon my goal of placing the origin of sex on the eukaryote tree.

And what about the goal of identifying the genes specific to meiosis, and characterizing their distribution in eukaryotes? A number of genes have been described as 'meiosis-specific', and in his SMBE talk John reported on their distribution and relationships to other genes, especially in those unicellular eukaryotes that might be representatives of early branches. Most of these genes are present in most of the species he's looked at, and in most cases they appear to have arisen by gene duplication.

One of the biggest challenges with this work is being sure that a gene is indeed meiosis-specific in all of the species it's found in - a gene that is meiosis-specific in yeast could have a related but not meiosis-specific role in a distant relative. This is very difficult to determine, especially because sex has never been observed in some of these species (if they are sexual, they're also very secretive). I'm glad that John Logsdon has been working on this, rather than me.

We're funded!

Our proposal to study the regulation of CRP-S promoters in H. influenzae and E. coli was successful; we have $127,567 per year for 5 years.

SMBE slides

I've just uploaded the slides from my SMBE talk to SlideShow, as suggested by a commenter. The slide show's name there is "R. Redfield's SMBE talk slides". I gave it tags for 'evolution', 'bioinformatics' and 'science'.

I think the talk went well. It was reasonably well attended, given that it was the last long talk of the meeting. I had streamlined it, and given it a very simple straightforward focus on how the issue of uptake sequences relates to the big question of whether bacteria have any processes that evolved by natural selection for producing recombinational variation.

I talk about sex!

Mingle² - Online Dating

Why our work is important

On Monday I ran my ideas for my SMBE talk past the postdocs, who politely trashed them. I was making all the errors I know not to make. The worst of these was I wasn't telling the audience why they should care about what I was telling them. One of the reasons I don't worry much about competition is that nobody else is working from the perspective I am, but this means I have to spell out the issues at the start of every talk, starting from the very basics.

The big issue is the evolution of 'sex'. The word sex has lots of meanings; here I mean any biological process that evolved because of the benefits of creating new combinations of genes (new genes or new alleles of genes). (If you've stumbled onto this blog by searching for 'sex' with another meaning in mind, you might want to cut your losses now.) I use this definition because it captures the big unsolved question of why so many eukaryotes engage in 'meiotic sex', that is, they produce a diploid genome by merging two haploid genomes and later produce from this four new haploid genomes with new combinations of genes.

Evolution of sex in eukaryotes is a big issue because we biologists don't know why it's worth the trouble. That sounds feeble, but generations of the best minds have rigorously analyzed the genetic consequences without producing any compelling explanation of why the recombined genomes would be sufficiently better than the original ones to compensate for all the biological costs of sex. The costs aren't such a big deal for facultatively sexual organisms like yeast or paramecium (who can reproduce just fine without meiotic sex), but they're enormous for obligately sexual organisms like ourselves and most other plants and animals, which must use meiotic sex to reproduce.

My approach to this problem is to ask whether bacteria have sex; that is, whether they have any processes that evolved because of benefits of creating new combinations of genes. The key word here is 'because', by which I of course mean 'by natural selection for'. We know that bacteria and archaea have processes that cause recombination, and that these processes have been important in the long-term evolution of their genetic capabilities. But I want to find out whether this happens by accident (as side effects of processes evolved by natural selection for other effects), or by natural selection for the new combinations. If the answer is yes (bacteria do have sex), we'll have shown that this selection is ubiquitous, and we'll have an independent (non-meiotic) system in which to investigate it. If the answer is no, we'll have shown that the reasons for meiotic sex are specific to eukaryotes, and that bacteria get all the genetic recombination they need by accidental effects of other processes.

The reason I have almost no competitors is that researchers have traditionally assumed that the processes that cause genetic recombination in bacteria exist because of selection for such recombination, and very few are willing to seriously consider that this assumption should be rigorously tested. This is a good place for one of my favourite quotations:

"I know that most men, including those at ease with problems of the greatest complexity, can seldom accept even the simplest and most obvious truth if it be such as would oblige them to admit the falsity of conclusions which they have delighted in explaining to colleagues, which they have proudly taught to others, and which they have woven, thread by thread, into the fabric of their lives."

Leo Tolstoy

How are we testing this assumption? By looking for evidence (at the molecular level) of how selection has shaped the processes that cause recombination. There are three such processes, but two of them, conjugation and transduction, can be easily shown to cause recombination only as side effects of genetic parasitism by plasmids and phages respectively. That leaves natural competence (DNA uptake) and its genetic consequence (transformation), which is what we work on. Transformation itself arises when the so-called recombination machinery in the cell acts on DNA the cell has taken up. But this machinery exists not because of selection for making new combinations of genes using DNA brought in from outside, but because of selection for the ability to repair and replicate the cell's own DNA. Because transformation itself is an unselected side effect of the replication and repair machinery, we concentrate on understanding how natural selection has acted on natural competence (the DNA uptake process).

I'll explain how we do this in a later post.

The proposal reviews are in...

Yesterday I got the emails telling me that the scientific reviews of our grant proposals were available. The actual funding decisions won't be made for a few weeks.

Regulation of Sxy and CRP-S genes: Score 4.25/5, ranking 9/37. So it might be funded.

Mechanism of DNA uptake: Score 3.8/5, ranking 21/40. This one won't be funded.

Pushing for more open science

I'm going to take advantage of my SMBE talk to proselytize for open science and research blogs.

The idea of open-access scientific publication is spreading. It's relatively easy for scientists to see the benefits of making the results of their research accessible to more people. But hardly anybody seems to consider the potential benefits of opening up the process of doing the science (as I'm trying to do with this blog). The reason is competition; most researchers are afraid that they'll be scooped if their competitors find out what they're doing.

Although competition can be a positive force in science, stimulating us to try to do better work than our competitors, I also think that the secrecy that arises from fear of competitors is also a very negative force. Competitors are all potential collaborators, and focus on competition shuts out the potential for collaboration. I was impressed by a NIH grant-writing workshop that told me that the first think NIH does when people call to ask about funding opportunities for a new line of investigation is to connect them with people doing related work, encouraging them to contact these as possible collaborators.

Of course it's easy for us, because we're working on a problem whose importance isn't widely recognized....

Stringency of Gibbs analysis

In revising our USS manuscript I've been thinking again about the stringency of the Gibbs motif searches. The stringency is set by telling the program how many sites it should expect to find. For most analyses we set this at 2000 (and it would find about 1350) but for the analysis of covariation between positions we set it at 3000 so we would have many poorer matches (it would find about 1650) and thus have more variation to analyze.

The Gibbs analysis assigns a score to each site it finds. With 'expect 2000' most sites have scores close to 1.0, but there are always some (~40 in the set I counted) with scores of zero and a few with scores between 0.5 and 0.9.

I only yesterday discovered how to get Excel to make a histogram; here's the histogram showing the distribution of scores for one of the 'expect 3000' searches. As with the 'expect 2000' analyses, most of the USS sites have scores close to 1.0, and few have scores int he middle range. But there are lots more sites with scores=0, and today I've been checking out how bad these sites are.

I had feared that they were garbage included by mistake, just increasing the background noise but not really resembling the USS consensus. So I was pleasantly surprised to see that these sites have a strong USS signal when displayed as a sequence logo (the top logo in the figure). For comparison I made another logo using only the sites that scored 1.0; that's the bottom logo in the figure.

Of course this analysis doesn't tell us whether the low-scoring sites found by the Gibbs analysis actually function as USS in DNA uptake. Most of them miss the consensus at at least one core position, but the ~25% with perfect cores also have stronger flanking consensus (I'm not sure how to interpret this).

Not that kind of framing

I'm working on the revisions for the USS paper, writing up the new analysis of reading frames and constraints. I put the data in the table on the left; the "Relative to best codons" values estimate how easily the USS-encoded tripeptide would be translated. I'm puzzling over why only 49 USSs are in reading frame A.

The graph below shows USS number plotted as a function of the postulated cause, proteome number (this is plotted the opposite way to that in my previous post on this analysis). This plot more clearly shows that one point (frame A) is an outlier; it has a lot fewer USSs than we

would expect given how often the tripeptide it specifies appears in the proteome.

Could this be because of codon constraints, i.e. because USSs in frame A require the tripeptide KVR to be encoded by inconvenient codons? No, the codon score is quite high (0.77) meaning that this frame's USS-specified codons are commonly used in the proteome.

If anything there should be more USSs in this frame than predicted by the proteome abundance of its tripeptide, because the consensus is weak for the first and second positions of the first codon and the second position of the second codon.

Motifs and elements

The reviewers of our USS manuscript didn't feel that our Gibbs motif analysis of USSs was much of an advance on the previous analyses. It's true that the motif identified by the Gibbs analysis is very similar to that found by searching for perfect USS cores. But the results could have been otherwise, and it's important to have found this out. So in our revisions we need to do a better job of explaining why the motif analysis was needed.

First I should clarify that the USS should be viewed not as a replicative 'element' but as a 'motif'. Both terms can refer to sequences or sequence patterns that are present at multiple sites in the genome but, at least for the purposes of this blog, a replicative element is a DNA sequence whose repeats have arisen by copying and insertion. Transposons and insertion sequences are examples of replicative elements that code for their own replication; Alu sequences are elements that are passively replicated. USSs could have also been elements produced by some sort of copying and insertion process, but we now know they are not.

The term 'sequence motif ' can be used for any detectable sequence pattern that occurs in multiple locations or genomes. It is commonly applied to DNA sequence patterns that have been selected for binding by specific DNA-binding proteins such as polymerases, transcription factors, and repressors, and to amino acid sequence patterns that perform specific functions in proteins. These motifs arise by point mutations in preexisting sequences, not by copying and insertion. Typically they are short (5-25bp), and have much weaker consensuses than do replicative elements, with most or all instances differing at one or more positions from the consensus. (Different copies of replicative elements are often identical over hundreds of bp.)

I won't go into the compelling evidence here, but we now know that individual USSs clearly arise by normal point mutations, not by copying. Although previous analyses of genomic USSs did not explicitly consider the distinction between replicative elements and motifs, they were limited by the need to search for specific sequences. The results therefore reflected only the properties of those specific sequences, with no allowance for the true diversity of functional USSs.

Given our new knowledge that USSs are motifs, any analysis of USS evolution had to be built on a solid understanding of their true diversity. The availability of the Gibbs motif sampler program let us search the whole H. influenzae genome for any patterns, without having to specify any sequence. Once the program found a pattern, it created a list of all the sites in the genome fitting the pattern, with a measure of the strength of each match. Thus it provided an unbiased analysis of the full diversity of USS-related sequences in the genome.

Introducing my USS talk

I'll be giving a talk at the upcoming meeting of the Society for Molecular Biology ad Evolution (SMBE) in Halifax. The focus of the session is on the evolutionary consequences of the mechanisms that cause recombination, and I'll be talking about how the sequence specificity of the H. influenzae DNA uptake machinery has affected the evolution of its genome and proteome.

I'm going to start the talk by describing what usually happened at the end of the talks I used to give on the regulation and evolutionary function of competence. I'd conclude that the regulatory evidence supported the hypothesis that bacteria take up DNA for food, not for its genetic information. Most people find this a very unwelcome idea, and one of the first questions would always be why I thought the uptake specificity wasn't compelling evidence that bacteria take up DNA to get new genetic information. I'd answer by saying that uptake specificity needn't have evolved to promote recombination, and that we were beginning to investigate alternative explanations.

I'm hoping that this introduction will capture people's attention - "There's a controversy! Everyone thinks she's wrong!"

We now have tons of analyses to report: the Perl model of USS evolution, the uptake assays with mutated USSs, the Gibbs motif analyses, the reading frame analysis, the evidence that USSs are not insertion elements but motifs that arise by mutation, the correlation of uptake specificity with Pasteurellacean phylogeny, the lack of coding constraints due to the USS int he H. influenzae genome, the presence of competence genes and genomic USSs across the Pasteurellaceae (including species that can't be transformed in the lab)... Unfortunately for me (fortunately for my audience?) I only have 30 minutes including the question period. I'm going to try to pull a draft talk into shape for Monday, when it will be my turn to do lab meeting.

Reading frames vs USSs

Sorry for the dead air; I was away unexpectedly with no internet access.

We got the reviews back for our USS manuscript. Not too bad. Both reviewers asked for more analysis of the role of reading frames in USS locations. (See previous posts here and here.) This turned out to be both easy and fun to do.

Many USSs are in the protein-coding parts of the genome, and these can be sorted by which of the 6 possible reading frames the respective proteins are encoded. The first two figures show the relationships of the USSs to the reading frames.

The frames aren't equally used. Frames A, B and C have 49, 125 and 425 USSs respectively, and frames D, E and F have 474, 125 and 157. The differences are too large to be explained by chance alone, and we think they arise because USS in the less-used reading frames impose more severe constraints on protein function, so that many USSs arising in these frames are eliminated by natural selection.

The new analysis considers the two factors likely to contribute to this disparity. (Because the flanking segments exert only modest constraints on amino acid sequence, I've limited the new analysis to only the most frequent tripeptides encoded by the USS core in each frame.)

The first factor likely to affect USS reading frame usage is the differing biochemical properties of the tripeptides that USS cores in these reading frames will encode. Some combinations are intrinsically more versatile than others, useful at many different locations in a wide variety of proteins, whereas other tripeptide combinations will only be useful in particular contexts.

The second factor is 'codon bias'. Most amino acids are specifiable by at least 2 (often 4 and sometimes 6) different codons, some of which are more efficiently translated than others, and cells preferentially use the easiest-to-translate codons for proteins they need to make a lot of.

The new analysis evaluates the first factor by comparing the number of USSs in each frame to the total usage of the six tripeptides in all the proteins of the genome (the 'proteome'). If the differing versatilities of the tripeptides is responsible for their differing use in USSs, we should see a correlation between total number and USS-encoded number. The results are shown by the red symbols and line in the graph below. The predicted correlation exists. (The line fitted to the points has a confidence score of 0.86. I know enough statistics to know that this is a reasonably strong correlation.)

The blue symbols in the graph show the results of the other analysis. To look for a correlation between codon bias and USS reading frame usage, I needed a crude score indicating how easily each USS-encoded tripeptide would be translated. I was able to get a table of codon usage for the H. influenzae proteome from the TIGR website; this gave the percent usage of each codon. So for each tripeptide I calculated a 'USS-codons' score as the sum of the percentages of the codons specified by USSs in that reading frame, and a 'best-codons' score as the sum of the percentages of the most commonly used codons for its three amino acids. Then I calculated a 'codon cost' as the ratios of these scores, and multiplied it by 100 so it would fit neatly on the graph.

If codon cost contributes to the disparity in reading frame usage by USS, we expect the blue points to show an inverse relationship; the highest codon costs should be for the least-used reading frames, so the blue line should slope down to the right. Instead we see no correlation at all. This tells us that codon bias makes little or no contribution to persistence of USSs in different reading frames.

We're not surprised by this second result. (In fact one of the postdocs was so sure of it that she didn't think the analysis was worth doing.) Other analysis we've done has indicated that USS are rarely found in categories of genes known to be subject to strong codon bias. But that's for a different paper, and this new analysis will nicely address concerns raised by both reviewers of our present paper.

Next?

OK, so the induction of CRP-S genes experiments established clearly that the ppdA gene's CRP-S promoter isn't being induced by the growth conditions I tried. And that changes in beta-gal activity don't necessarily result from changes in promoter activity. What to do now?

To make progress in these experiments, I need better tools for manipulating genes. I want to put the ppdD::lacZ fusion into the chromosome (it's on a plasmid now) and I want to test the effect of knocking out various genes, and I want to try a 'wild' E. coli strain. All of these require recombining desired genes into the chromosome, so I need to get the recombineering technique working for me. One of the post-docs was working on this, so I just need to take up where she left off.

I also need to get back to the tests I was setting up of components of the laser-tweezers project. I was just about ready to test attaching DNA to beads (have the beads with the bound streptavidin, have the biotinylated nucleotides to put on the ends of the DNA, have the DNA prep). Separately, I need to work with the new post-doc on an antibody-based method to attach H. influenzae cells to other beads, so they can easily be pushed around under the microscope.

Not the result I was hoping for....

This is beta-galactosidase activity in cultures that have been transferred from LB (= rich medium) to M9 salts plus a tiny bit of casamino acids (starvation medium). Transfer was at time=0.

All cells carried the ppdA::lacZ fusion on the same plasmid. The black line is cells that are sxy+ and crp+. The blue line cells are sxy-, and the red line cells are crp-.

It's clear that beta-gal activity increased similarly in all three cultures. As seen for the cultures in rich medium (previous post), the crp- cells had slightly lower activity, while the sxy+ and sxy- cells had almost identical activity.

So I conclude that the treatments I've tried so far have not changed the Sxy-dependent activity of the ppdA promoter.

Disappointing result

The detailed time course of beta-galactosidase activity in the E. coli strain with the ppdA::lacZ fusion confirmed the results I posted a few days ago. But I'm not going to post these results, because they are superseded (made uninteresting?) by the results of the next experiment.

Having shown that beta-gal activity decreases as cells enter exponential growth, and increases once growth stops, I needed to find out whether these changes depended on the presence of the transcription activators Sxy and CRP. If the answer is yes, then they are due to changes in the activity of the ppdA gene's CRP-S promoter. If the answer is no, then the beta-gal changes are due to something much less interesting (from my present perspective of wanting to find out how E. coli CRP-S promoters are activated). Unfortunately the answer seems to be NO.

Here's the data. The top graph shows culture density as a function of time for the three strains I tested. The first points (t=0) are before the cells were diluted 300-fold. You can see the 'lag' for the first ~30 minutes, then the cells begin growing exponentially. (You can tell that they're doubling at a constant rate because the points fall on a straight line on this log scale.) After about 200 minutes growth slows. The lines aren't joined to the last points because there's a 700 minute gap separating them (this part of the graph isn't to scale). You can also see that one strain grows slower than the others (red line and points); this is the strain whose crp gene is knocked out.

The second graph shows the amount of beta-gal activity in the cultures. Some of the values for the crp- strain (again the red line and points) are a bit low, perhaps because of its slow growth. However the sxy+ and sxy- strains show almost identical patterns (black and blue lines respectively). This means that the changes in beta-gal activity do not depend on Sxy, and thus almost certainly do not reflect changing activity of the ppdA gene's CRP-S promoter. Again the last points are for samples taken after 1200 minutes, when the cultures had been in stationary phase for quite a while.

What could cause the changes in beta-gal activity? One possibility is that the number of copies of the ppdA::lacZ plasmid per cell might continue to increase after cell growth slows. Another is that the lacZ mRNA might be more stable in stationary phase than other mRNAs. There are probably other possibilities too. The only important possibility is that we're wrong about Sxy and CRP being transcriptional activators of this promoter.

What next? I'm right now testing whether the stronger induction seen when the crp+ sxy+ cells were transferred to starvation conditions depends on CRP and Sxy. I hope it does.

Why are we still using an archaic procedure to make H. influenzae competent?

Last week I did an experiment in parallel with one of the post-docs. Her preparations of competent H. influenzae had not been very competent at all, and we wanted to check if there was a problem with her procedure. We both produced cells that were reasonably (not excellently) competent, so we concluded that her past problems were not due to faults in her procedure.

In recent posts I've been describing my preliminary steps to finding a way to induce E. coli's 'cryptic' competence genes, and this and the above have started me thinking about the procedure we use for H. influenzae. It's a simple sprocedure: cells growing in a rich medium called sBHI are abruptly transferred to a starvation medium called M-IV. After about 90-100 minutes in this medium the cells are as competent as they are going to get. I deally this means that all ofthe cells are competent, each ready to take up a few hundred kb of chromosomal DNA.

This procedure was developed about 45 years ago, by more-or-less systematic fiddling with growth conditions. It hasn't been altered, simply because nobody has had a strong reason to bother trying. But now we know much more about what is going on when competence genes are induced, and we should be able to improve on this procedure. Improvements would certainly be useful, especially for the post-doc's ongoing survey of competence in 'wild' strains of H. influenzae. Being able to improve the induction procedure is also a good test of whether we really do understand how the genes are regulated.

What should we try? Adding cAMP should take the place of carbohydrate starvation, activating CRP to make its contribution to expression of CRP-S promoters. If we're right that nucleotide starvation is needed to induce sxy translation, we would like to try simpler ways to mimic this; I wonder if there's a specific inhibitor pf purine (or pyrimidine) synthesis? But this won't be much use if the cells are still in rich medium, as they are getting their nucleotide precursors from the medium not from de novo synthesis. Maybe we could simulate nucleotide starvation another way (slowing polymerase with an antibiotic resistance mutation? as we suggested in one of our grant proposals???).

Done and to do

Yesterday I did a big time-course experiment with the ppdA::lacZ fusion strain, to get a more detailed picture of the results in my last post. The information is now in my notebook: 50+ samples, each with sampling time and OD600 (cell density at that time) and volume assayed and assay time and OD420 (ONPG hydrolyzed by beta-galactosidase), but I haven't yet entered it into Excel and analyzed it.

The next important experiment is to demonstrate that the changes in expression are dependent on both CRP and Sxy. I will do this by doing simplified time-course analyses using cells carrying the ppdA fusion plasmid and a knockout of either crp or sxy. I grew up both these strains yesterday and hope to have time to test them this afternoon. If the changes are indeed due to increased activity of the ppdA CRP-S promoter, they shouldn't happen in the knockout backgrounds. I won't do this experiment until after I have analyzed yesterday's data, so I'll know which parts of the time course are most informative.

I didn't have time yesterday to repeat the transfer-to-starvation-medium analysis; I'll wait till I have the knockout results before doing this. That way I'll be sure that the expression changesI'm seeing are indeed due to the cause I want to investagate (changes in activity of CRP-S promoters).

Biocurious about tweezers

In between time courses I've been reading a M.Sc. thesis by the student who was developing the laser-tweezers analysis of DNA uptake. (You may know him as PhilipJ of Biocurious.)

I'm not an examiner of this thesis; I'm reading it to learn more about how laser-tweezers work. It's very well written, and pitched at exactly the right level for a neophyte like me. Any day now I should decide I've taken the time courses far enough for now, and get back to my attempt to follow in Philip's footsteps.

Time course of ppdA expression

On Saturday I did a time course examining whether the expression of the ppdA::lacZ fusion (and I hope thus of the sxy gene) changed as growth conditions changed while the cells were growing in rich medium (LB) and after transfer to a starvation medium.

The green line shows how the cells grew. First the density fell sharply because I diluted the cells 1/250 in fresh LB (+Amp to maintain the plasmid carrying the fusion). After a brief period of slow growth while the cells adjusted their metabolism to the improved conditions ("lag" phase), the cells grew exponentially, doubling about every 30 minutes ("log" phase). After about 220 minutes growth slowed down as the medium became depleted of nutrients (still growing so not yet in "stationary" phase).

The blue line in the second graph shows the expression of the lacZ fusion in LB. The first sample was from the dense culture (it had been left on my bench overnight). Once those cells were diluted into fresh LB the amount of beta-gal activity slowly fell, presumably because expression of the lacZ fusion decreased. Part of the reason the decrease is slow is that the cells still contain beta-galactosidase that will either be degraded or diluted out by cell growth. The amount of beta-gal activity remained low while the cells grew, until the cell density got quite high (OD600 about 1.0), when the activity began to increase. I know that the low activity reflects lacZ expression by cells in exponential growth, and not residual enzyme from earlier induction, because I diluted part of the log-phase culture 100-fold into fresh LB+Amp, and found the same activity after these cells had spent an additional 2 and 3 hours in exponential growth (blue circles).

But once the cells got so dense that growth slowed, the beta-gal activity increased, and by 300 minutes had become at least as high as that of the overnight cells from my bench. This is a weaker version of the gene induction we see as Haemophilus influenzae approaches stationary phase.

In H. influenzae, rapid transfer to a medium lacking most nutrients causes strong induction of CRP-S genes, including the homolog of ppdA. To mimic this I transfered the log-phase cells to a medium consisting of minimal salts (M9) plus the same small amount of amino acids we use for H. influenzae. The two red bars show that this transfer caused a more dramatic increase in beta-gal activity.

So, these are weak but nice results. They suggest that CRP-S genes in E. coli are regulated by similar factors to those regulating CRP-S genes in H. influenzae.

[Just to check, I also induced lacZ expression in wildtype (lac+) cells by adding IPTG. Even though these cells have only one copy of the lacZ gene (in the chromosome, rather than one on each of many copies of a plasmid), after 60 minutes they expressed 7 times as much beta-gal activity as the starved ppdA cells did after 120 minutes. This is slightly less than the ppdA cells expressed after Sxy was overexpressed from the pASKA plasmid, and suggests that the CRP-S promoter is being only partially induced by my starvation conditions.]

What next? I need to repeat this time course, starting with a proper overnight culture and taking more time points, especially after log phase. I need to get a longer log-phase series too. And a time course in the starvation medium.

Which E. coli is best?

I haven't yet tested whether different growth/non-growth conditions alter the expression of the ppdA fusion (though I did get the sxy manuscript resubmitted today), but a paper I came across reminded me of another issue I need to consider.

The paper is an opinion piece titled "Laboratory strains of Escherichia coli: model citizens or deceitful delinquents growing old disgracefully?"; it just came out in the journal Molecular Microbiology (Mol. Micb. 64:881-885). The authors argue that the standard K-12-derived strains of E. coli that microbiologists and molecular biologists typically use are not at all representative of the strains out in the natural environment. For one thing, lab strains have lost about 20% of their genomes. These are (by definition) non-essential genes, but their loss no doubt affects cellular metabolism, so that the metabolic interactions we see in K-12 strains may be quite different than those in natural strains. Furthermore, the culture conditions we use (rich broth, lots of oxygen, no competitors) are unlikely to ever occur in nature. Their long maintenance under these and other unnatural conditions means that the lab strains will have evolved by accumulating cryptic mutations that are beneficial under lab culture conditions but that may have very different and perhaps harmful effects in the natural environment.

I already knew this. It has important implications for my search for conditions that induce expression of competence genes in E. coli. Right now I'm working with standard lab strains, but it's all too possible that one of their ancestors lost the ability to express the genes encoding homologs of H. influenzae's competence genes, or to assemble the proteins into functional DNA uptake machinery.

So I should perform my tests on a less lab-adapted, more 'natural' strain as well as on the K-12 strain the ppdA::lacZ fusion is in. But this raises two problems. First, which strain should I use? I do have a fairly-ancestral K-12 strain, but really I should use a 'wild' strain recently isolated from the environment. The NCBI Microbial Genomes page lists 8 completely sequenced E. coli genomes; their sizes range from 4.6 million bp (K-12) to 5.6 million bp (O157:H7). And the various wild strains have genomes that are quite different from each other - does this mean I would need to test many strains before giving up? I'm also not meticulous enough to be trusted with a strain that's seriously pathogenic to humans, so the two O157:H7 strains are out, as is the uropathogenic strain*.

Second, I'll need to transfer the necessary genes into this strain; first a lacZ mutation making it Lac-, then the ppdA::lacZ fusion so I can assay induction by Sxy. This might mean that I need to get my P1 transductions working after all (I had been thinking I could let them slide now that I've found that the ppdA::lacZ fusion can be used as an indicator of CRP-S induction by Sxy). I don't need P1 to move in the ppdA fusion because it's on a plasmid, but transduction would certainly be the easiest way to move a lac- mutation in. If I'm lucky, maybe some of the wild strains are naturally Lac-. (Probably not; most screens for wild E. coli start by treating anything that's Lac- as not E. coli.)

* NCBI also lists 6 sequenced 'Shigella' genomes - we now know that the bacterial strains assigned to the genus Shigella are really variants of E. coli. But I have absolutely no intention of working with these very pathogenic bacteria.

Manuscripts progress

Last week I finally submitted our manuscript about how CRP acts at CRP-S sites in H. influenzae and E. coli. The most interesting result in it was obtained by the gsnpiw just before he left for Belize. He showed that CRP-S sites contain regulatory sequences just upstream of the FCRP-S site that are needed for transcriptional activation; these sequences aren't typically present in CRP-N promoters.
Still to resubmit is our manuscript about Sxy. It too has been enhanced by a new result from the gsnpiw. The experiments are described here). He only had time to do it once before he left, so we're not treating them as part of the paper's Results section, but the results are sufficiently good to be briefly described in the Discussion section. They show directly that mutations in sxy do alter the ability of the sxy mRNA to be translated. Two of the reviewers had suggested we do a different experiment ("toeprinting") which would have tested whether the mutant mRNAs differ in their ability to serve as templates for a polymerase. In my cover letter to the editor of our manuscript I'll explain why we think our new experiments are more informative than toeprinting would have been. I'll also explain that they're not ready to be part of the Results, and that, if the editor thinks that the experiments need to be completed and included in the Results, we'd like a two-month extension of our revisions deadline so they can be completed after the gsnpiw returns from Belize.

I spent much of yesterday making changes to the manuscript and figures and writing responses to the many points raised by the reviewers. A downside of getting four thorough reviews of a manuscript is the very large number of issues they raised. I'm not complaining, as almost all of of these issues lead to improvements; either the reviewer is right, and we make the suggested change, or the reviewer misunderstood what we meant, and we clarify our writing to prevent the misunderstanding.

Only a few points remain to be dealt with. We used lacZ fusions to examine the effect of sxy secondary structure, and one reviewer wants more background information about the behaviour of the reference fusion. This data is in the PhD thesis of a former grad student (an author on the paper), so we may be able to simply refer to it there rather than adding the data to the manuscript's Results.

Time course and IPTG dependence

I've finished the two experiments I described in the last post: (1) a time course to find out how long it takes for induction of sxy by IPTG to give production of beta-galactosidase by the ppdA::lacZ gene fusion, and (2) a 'dose-response curve' to see how much IPTG is needed to induce the fusion. And I'm posting the resulting graphs, along with a sketch of the E. coli cells I'm using for this experiment.

The sketch shows that the cells contain two plasmids. The one on the left is pASKAsxy; it carries the E. coli sxy gene (red) under the control of the Plac promoter (blue arrow). This promoter is normally OFF in these cells so the sxy gene is not expressed, but the promoter can be activated by adding a lactose-analog called IPTG. The plasmid on the right carries the lacZ gene, which codes for the enzyme beta-galactosidase. Here the lacZ gene has been placed under the control of the ppdA gene's promoter, so when ppdA would be expressed, the cells make beta-galactosidase. Beta-galactosidase normally digests lactose, but I can easily detect it because it also digests another lactose analog called ONPG, producing a bright yellow chemical (ONP).

So I give these cells IPTG, wait a while, give them ONPG, and measure how much yellow colour they have made. That's what I did yesterday. (Measuring the yellow colour was complicated by a problem with our spectrophotometer - it kept forgetting what the standard brightness was - but I managed to keep it working well enough to get my data.)

Here's the time course. I set up three cultures. One got no IPTG, one got the standard amount of IPTG (1.0 mM) and one got tenfold less (0.1 mM). It's easy to see that tenfold less IPTG gave just as strong a response as the standard amount, that neither dose produced a significant increase in beta-gal until at least 40 minutes had elapsed, and that there was a bit more beta-gal at 90 minutes than at 120 minutes.

So I have the most basic information I need. When I go on to test induction of the ppdA promoter by treating cells in ways that I think might induce expression of the normal chromosomal sxy gene, I now know that I should allow at least an hour to see the effect of my treatment.

There are no error bars because I only tested each combination of dose and time once. I'll need to repeat the experiment with more replicates to get publishable results. I'll probably also want to use more time points in the interval between 40 minutes and 90 minutes, and to try a denser culture as well as the quite dilute one I used this time.

It is worth repeating this experiment and getting some more details, because I find this long delay quite surprising. One control I didn't do is to examine IPTG induction of the normal lac operon (the lac promoter driving expression of the lacZ, lacY and lacA genes), but this is a very standard experiment done in most introductory biochemistry lab courses, and as I recall the ONPG starts to accumulate within a few minutes. A Google search just turned up a paper about mutations that affect transcription, which used IPTG induction of lacZ as an assay; their control showed that beta-gal began to accumulate after 2.3 minutes. If we assume that induction of the pASKAsxy promoter by IPTG starts producing Sxy by 2.3 minutes, why does it take so long for Sxy to induce the ppdA promoter and cause beta-gal to start accumulating? Is the sxy mRNA translated very inefficiently, so accumulation of enough Sxy protein takes a long time? Is the ppdA promoter activated only once a lot of Sxy has accumulated? The results may have implications for competence induction in H. influenzae too. We've found that cells need about 45 minutes to become competent, but I have been assuming (perhaps wrongly) that this time is mainly needed for assembly of the uptake machinery. We have some fusions of lacZ to H. influenzae competence gene promoters; I (or someone) should test the kinetics of induction of these.

Here are the results from the dose-response analysis. In this graph the X-axis is IPTG concentration, not time (and note the log scale on both axes). IPTG was added at time 0. We see that any IPTG concentration above 10 micromolar gives full induction (about 40-fold) after 120 minutes, and that even the highest concentration doesn't give any induction at 25 minutes. This may not be very useful information; because I don't have an easy to measure Sxy production directly, it may just tell me about the sensitivity of the lac repressor to IPTG.

So what next? I'm going to grow the cells containing the ppdA::lacZ plasmid (no pASKAsxy) and try the same changes to culture conditions I previously tried with the ppdD::lacZ fusion (which I now know couldn't have worked because it isn't inducible by Sxy).