Field of Science

Do chromosomal proteins on 'donor' DNA affect transformation?

I've started polishing my not-quite-good-enough CIHR proposal for what's called the 'Transitional Open Operating Grant Program' competition.  This is the last-chance-under-the-old-system competition; any future proposals will be evaluated under the new system, which doesn't have much use for pure science or small labs. Proposals are due March 2 2015.

Before then I'd like to do some preliminary experimental work on one of the studies I'm proposing. I expect that the H. influenzae DNA available to H. influenzae cells in the host environment will still have bound to it many of the normal chromosomal proteins (HU, H-NS. Fis), and might even retain aspects of the normal nucleiod structure.  I want to find out how this affects the ability of the DNA to be taken up by competent cells, particularly whether specific sequences or segments are affected more than others.

My poorly thought-out plan is to lyse donor cells carrying one or more antibiotic resistance alleles in a way that doesn't disrupt bindings of proteins to DNA (so not with 1% SDS), and then mix this lysate with competent sensitive cells.

Big problems I forsee:

1.  How to lyse the donor cells without disrupting the nucleoid proteins?

My original plan was to use the H. influenzae phage HP1.

I could probably instead lyse the cells with lysozyme and a small amount of a surfactant that disrupts membranes but not proteins.

2.  How to lyse the donor cells without killing the recipient cells?

If the recipient cells are lysogenic for this phage (I have such a strain in the freezer), then they will be resistant to the free phage in the lysate.  An undiluted lysate has an enormous number of phage (>10^10 per ml), but I'll dilute the lysate to a chromosomal DNA concentration of about 1 µg/ml, which would be saturating for uptake if the DNA had been purified.
(Hmm, what will be the chromosomal DNA concentration of a lysate?  Say 3x10^9 cells/ml, most of them lyse, 1,830 kb of DNA/cell, 10^-12 µg of DNA/kb...  That's only about 4 µg of chromosomal DNA per ml.)
I can dilute the DNA way more than this, because transformation has such a wide sensitivity range.  I could also make it difficult for the phage to infect by eliminating Mg, but that might also hinder DNA uptake (the normal competence medium is 10 mM Mg).  If I had antibodies to the phage I could block infection that way (we used to do this with lambda), but I don't.

If I lysed the cells with surfactant, I could then easily dilute the lysate to reduce the concentration of the surfactant to a concentration that wouldn't harm the recipient cells (10-fold or 100-fold wouldn't be a problem).

3.  How to kill off or remove all the donor cells that aren't lysed, without disrupting the nucleoid proteins?

I don't know if it would be possible to pellet the cells without pelleting the nucleoids, especially since the nucleoids and DNA may be still attached to the cell wall.  Or to filter out the cells without removing or disrupting the nucleoids.

I don't really need intact nucleoids, but I'd like to still have the proteins bound to much of the DNA.

How can I check the state of the DNA?  I hope I wouldn't need to use electron microscopy.  Maybe I could use a simple very-low concentration agarose gel? Like an Eckhart gel, but interested in the DNA+gunk, not the megaplasmid?

4.  How to kill off or remove all the donor cells that aren't lysed, without killing the recipient cells?

If the recipient cells are already resistant to an antibiotic that the donor cells are sensitive to, I can include this antibiotic in my selective plates.

Or maybe I could kill them by adding a bit of chloroform to the lysate, and then diluting or evaporating the chloroform before I add the lysate to the cells.  Chloroform is normally used to sterilize phage lysates, but I don't know if it would affect the nucleoid proteins.

Interaction of a ∆hfq mutation with the rpoD mutation

Here's the last part of the summary of what our senior co-op technician has been doing.  The last set of experiments tested the interaction between a new ∆hfq mutation we've been studying, which reduces competence) and the rpoD mutation (which increases competence).

The Hfq protein binds small regulatory RNAs, helping them to form base-pairs with the mRNAs that they regulate.  In other bacteria we know that this base pairing can either reduce the mRNA expression (usually mediated by RNase E degradation) or increase it (by reducing the effect of otherwise-inhibitory mRNA secondary structure).  Our ∆hfq mutation reduces competence in MIV-induced cells by about 10-fold, suggesting that it increases the translatability of sxy mRNA.

The technician tested whether this effect is still seen in cells with the rpoD mutation.  She first had to construct the double-mutant strain.  This was relatively easy because the ∆hfq mutation is 'marked' with a SpcR cassette, and the honours student who's studying this mutation had already made chromosomal DNA.  So she used his chromosomal DNA to transform strain RR753 (rpoD mutant) to spectinomycin resistance.

She then did a competence time course, following development of competence in rich medium in four strains.  In the graph below we see that the ∆hfq mutant (green line) develops competence later than the wildtype cells (KW20, blue line) and to a lower final level.  This mutation also reduces the competence of the rpoD mutant (compare red and purple lines), although not as severely.

So we conclude that, whatever Hfq is doing to promote competence, it's still at least partly needed by the rpoD mutant.

The honours student has been analyzing the interactions of the ∆hfq mutant with other factors and other hypercompetence mutations.  I'll do a separate post pulling this together, unless he does it on his blog first.

Effects of cAMP and AMP on competence development by the rpoD mutant

This is a continuation of yesterday's post on the phenotype of our hypercompetent rpoD mutant strain RR753.  Yesterday we wrote about its behaviour under 'normal  growth conditions, and now we're going to consider two new factors, cyclic AMP (cAMP), which induces competence under what are otherwise non-inducing conditions, and AMP, which inhibits competence development under what are normally inducing conditions.

First the effect of adding cAMP: We tested this by adding 1 mM cAMP to cells growing exponentially at an OD600 of 0.1, and measuring transformation 60 min. later.  At this growth stage, normal cells do not transform detectably, but addition of cAMP turns on sxy transcription.  Some of the resulting sxy mRNA is translated and the Sxy protein acts with CRP to stimulate transcription of genes encoding the DNA uptake machinery.  In the tech's experiment, cAMP addition raised the transformation frequency about 500-fold, from 1-3 x 10^-8 (just at the detection limit) to 6.5-8 x 10^-6.  The rpoD mutant is somewhat transformable even with out cAMP (01-3 x 10^-6), and cAMP addition raised this about 100-fold, to ~2 x 10^-4.

So we conclude that the rpoD mutant does not bypass the need for cAMP in competence induction. This rules out the boring hypothesis that changing Sigma 70 activity perturbs cellular metabolism, causing an elevation of baseline cAMP levels that in turn causes the mutant's increased competence. Instead it's consistent with our interesting hypothesis that the rpoD mutation likely changes one or more events after the sxy transcription is stimulated by cAMP and CRP.

Next, the effect of adding AMP:  Maximal competence is normally induced by transferring exponentially growing cells from rich medium to a 'starvation' medium called 'MIV' and incubating them for 100 min.   Previous work has shown that adding purine nucleotides or nucleosides (usually 1 mM AMP) to the MIV prevents normal competence development by reducing the translation of sxy mRNA (Sinha et al. 2013).  The next experiments tested whether AMP has the same effect in the rpoD mutant.

Both wildtype and rpoD mutant cells have high transformation frequencies after incubation in MIV. In these experiments the rpoD cells had slightly higher transformation frequencies (about 4 x 10^-3) than the wildtype cells (about 1 x 10^-3).  Adding AMP to the MIV used to induce competence reduced the transformation of wildtype cells more severely than seen in previous work, about 5000-fold (from 1.7 x 10^-3 to 3.4 x 10^-7) and at least 10,000-fold (from 3 x 10^-4 to 3 x 10^-8).  The AMP also reduced the viability of the cells by several fold. (Both replicates gave no transformants at all with added AMP, so these estimates are upper limits.)

Added AMP also reduced the transformability of the rpoD mutant.  The first replicate gave no transformants (the plated cells were too dilute) indicating that transformability as reduced at least 1000-fold, and the second replicate showed a ~6700-fold reduction.

These numbers are all lower than previous results, but the conclusion is clear that adding AMP to the MIV medium strongly inhibits the development of competence in the rpoD mutant.  We don't know how the added AMP caused the reduction in competence, but, based on other evidence from analysis of purine-biosynthesis mutants, I've hypothesized that the key factor is a decrease in the concentration of another metabolite (PPRPP), which maybe interacts with sxy mRNA.  Production of Sxy by the rpoD mutant is still sensitive to this effect, so... (I don't know what).

Phenotype of the rpoD mutant

This mutation causes H. influenzae cells to become competent prematurely, and to reach levels of competence in rich medium that are about 100 times higher than normal cells.  The mutation causes a single amino acid substitution in domain 3 of the 'housekeeping' transcription factor called 'sigma 70'.  rpoD is an essential genes, needed for transcription of most housekeeping genes.  Since the mutant strain (named RR753) shows only a very slight decrease in exponential growth rate we think the mutation causes only a very minor change in the protein's function

My earlier post this morning said I had never explained my hypothesis about how this mutation causes increased competence, but actually I did, briefly here.  Here's the key sentences from that post:
My hypothesis is that the mutation's effect on transcription of sxy mRNA increases competence by increasing sxy translation.   I've long hypothesized that slowing elongation or increasing pausing in the 100 nt segment of sxy mRNA that forms its regulatory secondary structure will promote sxy translation by increasing the ribosome's access to the sxy ribosome-binding site and start codon. 
Then I listed some low-tech phenotypic analyses that the senior of our two co-op technicians could do:
  • Is RR753 sensitive to the inhibition of competence by added purines?
  • What's the effect of an hfq deletion in this background?
  • How does this strain respond to added cAMP?
  • How does it respond to the standard competence-inducing MIV treatment?
  • Does the mutation increase competence of a sxy mutant (sxy6) that has an extra-stable secondary structure?
  • Does it further increase log-phase competence of the sxy hypercompetence mutants, which have weakened sxy mRNA secondary structures?
She's now done all but the last of these, and we're considering what she should do next.  So here we're going to summarize what she's found and what we think it means.

Her first experiments gave a better characterization of RR753's growth rate.  She's done both Bioscreen growth curves (high-precision analysis of exponential growth) and manual ones (lower precision but better for cultures at low and high densities).

First the Bioscreen results:  Here growth of RR753 (red line) is compared to wildtype cells (KW20) and two other hypercompetence mutants, RR563 (sxy-1) and RR749 (murE).  This clearly shows the slightly slower exponential growth of the rpoD mutant.

The Bioscreen analysis measures OD600, so it doesn't tell us about the actual numbers of viable cells. The tech has also done a number of manual growth curves, mostly as control parts of experiments examining other variables.  These agree with the Bioscreen results in showing usually a slightly slower exponential growth rate, and no obvious differences in later survival.  

Her next time courses replicated my earlier measures of transformation frequency.  It looks like the rpoD mutant differs from the other hypercompetence mutants (in sxy and murE) in having very low competence at very low cell density, perhaps as low as that of wildtype cells.  The other mutants at 100-1000-fold more competent than wildtype cells at very low cell density. The rpoD mutant may also become highly competent at lower cell densities than wildtype cells, but may not be any different than the other hypercompetent mutants - these data are hard to interpret.

What could be the significance of having very low competence at very low cell densities?  I'd been assuming that the moderate competence of the sxy-1 hypercompetent mutants at low cell density reflected a baseline level of sxy transcription and an increased efficiency of translation.  If the rpoD mutation acts as I've hypothesized, it should have the same effect.  Provided the cells are in real exponential growth, the cell density shouldn't matter.  Might the rpoD mutant have two different 'exponential' growth phases, one at very low cell density and another at moderately low cell density?

Is this an important issue?  It would be quite a bit of work to investigate carefully, so let's set it aside for now.

The next experiments analyzed the effects of adding cAMP (known to stimulate transcription of sxy) and AMP (known to reduce translation of sxy mRNA).  I'll leave these for the next post.

What should our 'senior' co-op tech do next?

We have two co-op (undergraduate) technicians at present (paid from the last of our leftover CIHR funds).  Each is with us for 8 months; one started in May and the other in September, so there's a 4 month overlap.

The senior one has done almost all of the work preparing the samples for the RNA-seq project, and lately she's also been doing competence time courses to characterize the phenotype of our hypercompetent rpoD mutant.  She's looked at growth conditions, at the effects of added cAMP (competence up) and added AMP (competence down), and the effects of knocking out the small-RNA regulator Hfq (down).  Writing this post makes me realize that we haven't summarized her results anywhere, so I'll sit down with her and pull it all together this morning (I think that will be another post).

Now we need to decide whether there are still more rpoD phenotype assays she should do, or whether she should move on to another project for her last two months.  Since I hypothesize that the rpoD mutation causes competence by slowing sxy transcription and increasing its mRNA translatability, she could assay the effects of of the rpoD mutation in combination with our various sxy mutations that affect its mRNA translatability.  But this would be very much a fishing expedition, lots of work but probably no new insights (because it's not testing any specific hypotheses).

Hmmm, looking back, I discover that I've never written a blog post clearly explaining the general hypothesis about how the rpoD mutation causes hypercompetence.  I think it's time I did that.

So long since I've posted...

But here's an update on the RNA-seq saga.

We've been struggling to get evidence that our RNA preps are of good enough quality for library preparation and sequencing.  Because the Ribo-zero kits that we use to remove the rRNA from the samples are so expensive, we wanted to be sure the final mRNA concentrations and integrity would be good enough to make the equally expensive sequencing worthwhile.

Unfortunately my attempts at using a Bioanalyzer have been very frustrating (and it's not cheap - several $ per sample), with odd smeary/spiky patterns and poor recognition of the size standards. This is partly because the concentrations are so low and the 'pico-RNA kits are very fussy, and partly because it's just a fussy procedure and easy (for me) to make mistakes.

We sought advice from the person who will be preparing the sequencing libraries for us.  She said that samples that appeared bad on Bioanalyzer results often gave excellent sequencing results, and suggested that we have her prep an initial 8 libraries.  She would then pass these on to the person who does the sequencing, who could just run her initial quality checks on the libraries and tell us whether there were suitable for sequencing.  So that's what we did, and the news was good - these initial libraries look OK.

Now it's time to do the Ribo-zero and final clean-up on the rest of the samples, so they can be handed over for eventual library prep and sequencing.  There's a two-month backlog so we likely will get the results as a Christmas present.

Our sample processing had been a bit scattershot, so now I needed to go through the sample information and make sure we have the right samples, and that the numbers add up.  They do, so I think we're all set to quickly complete the remaining Ribo-zero treatments and use RNeasy-Min-elute columns to clean up each RNA sample after Ribo-zero.  Then the remaining 64 samples can be handed over.  (The 'we' here is our excellent UBC-Co-op program technician.)

We have accidentally wasted two of our Ribo-zero treatments on duplicate samples, and I was thinking that this meant that we'd have to buy another 6-sample kit ($650).  But now I think we can get by without this, either by not treating two samples (the G6 samples in the table above) or by using slightly less of the reagent for a subset of samples.

We had been using only 1 µg of total RNA for Ribo-zero treatment, but for these samples we'll sue all the DNase-treated RNA we have (typically several µg).  In some cases the volume of sample will be too high (max 28 µl for Ribo-zero), but singe the samples are in H2O we can just evaporate them a bit to decrease the volume.

Possible work with the new rpoD mutation

I'm up to my ears preparing materials for the new version of Useful Genetics, but we have a great new technician through UBC's Science Co-op program.  She's nearly finished the work preparing the RNA samples for our big RNA-seq project, so it's time to consider what she should work on next.

One possibility is characterizing the newly identified rpoD mutant strain.  This strain (RR753) has a point mutation in HI0533, which encodes the sigma factor that regulates initiation and early elongation of transcription of 'housekeeping' genes, especially during exponential growth.

One analysis we need is BioScreen growth curves of the mutant and a wildtype control, to confirm the preliminary growth curve data suggesting that this mutation causes slightly slower cell growth (lower graph).  I think this slower growth results from a general slowing or minor disruption to normal transcription of many genes, and is not specific to its effects on competence. 

My hypothesis is that the mutation's effect on transcription of sxy mRNA increases competence by increasing sxy translation.   I've long hypothesized that slowing elongation or increasing pausing in the 100 nt segment of sxy mRNA that forms its regulatory secondary structure will promote sxy translation by increasing the ribosome's access to the sxy ribosome-binding site and start codon.  We're not in a position to dive into the molecular analyses of RNA and protein that will probably be needed, but I wonder if there are some genetic or culture-conditions approaches that will shed light on the situation.
  • Is RR753 sensitive to the inhibition of competence by added purines?
  • What's the effect of an hfq deletion in this background?
  • How does this strain respond to added cAMP?
  • How does it respond to the standard competence-inducing MIV treatment?
  • Does the mutation increase competence of a sxy mutant (sxy6) that has an extra-stable secondary structure?
  • Does it further increase log-phase competence of the sxy hypercompetence mutants, which have weakened sxy mRNA secondary structures?
I think the technician is going to be doing a lot of competence time-courses...

CIHR depressing

The results have been released for the latest competition for operating grants from the Canadian Institutes for Health Research (Canada's equivalent of NIH).  Our proposal was ranked 13/72 by its assessment committee (Microbiology and Infectious Disease), but they only funded 11.

This is our 7th failed CIHR proposal in a row.  They all had good scores, and several, like this one, were very close to being funded.  The scores show that our proposals do keep getting better (this time 4.5/5), but the funding cutoffs also keep rising and we're never quite good enough.

Will I try again?  Perhaps not.  This was the last round under the old funding system, and the new system is even less favourable to the small-lab fundamental research that we do. 

I'm not quitting research.  For this year we have funds left from a previous CIHR grant (must be all spent by the end of March), and after that we'll potter along on our very small NSERC grant.  Luckily most of what we do isn't very expensive, but I can now only support one grad student, supplemented with several excellent undergrads who'll work mostly for free.

More cell preps for RNA-seq analyses

Unfortunately we have to repeat most of the cell preps I made for the big planned RNA-seq analysis , because our RNA preps used up all the cells without producing any RNA.

The main problem was that we were using a Qiagen RNeasy Plus kit (designed to remove contaminating DNA) instead of the usual plain RNeasy kit.  I had bought the Plus kit because of a cheap introductory price, planning to just leave out the final 'on-column DNase digestion step, since this hadn't been very effective in the past.  However Qiagen had changed the kit without changing its name, replacing the final 'on-column' digestion with an initial pass through a 'G-DNA eliminator column' (before the usual RNeasy column steps).

The kit instructions didn't say anything about its suitability for bacterial cells, so I contacted Qiagen technical support.  They assured me that, although the kit would not remove DNA from bacterial cells, it would give a normal recovery of RNA.  What they didn't tell me was that the initial detergent treatment would not lyse bacterial cells - at least I think this must be the reason we didn't find any RNA when we ran all our RNA samples in a gel and a Nanodrop spec.  We had also processed some samples with the normal RNeasy kit, but most of these can't be used because we need complete sets of samples from replicate cultures.

Qiagen responded bvery well to my complaint about our results, providing us with two new RNeasy kits and 500 ml of the RNA Protect solution.  But we still have to do the work of regenerating the samples.  Luckily we have a very competent new technician, hired through UBC's Biology CO-op program, and she's doing most of the work.  In particular she's regenerating the samples of cells during induction of competence by our MIV starvation medium.  But tomorrow I'm going to (finally) spend a day in the lab generating all the samples from cells growing in rich medium.  This is 9 cultures, with 3 samples from each.

I think I can do them all in one day, if I plan carefully.  So here's some planning:

I'm collecting cells at three different densities:  OD600 = 0.02, 0.6 and 1.0.  I freeze pellets from 2 ml of cells for RNA prep (3 tubes with 0.67 ml cells and 1.3 ml RNA-Protect) and 1 ml of cells with 0.25 ml 80% glycerol, for later testing of transformation frequency if needed.  The OD 0.02 cells need to be concentrated 10-fold before freezing so we'll get enough RNA from them, so I need to start with a larger-than-usual volume of cells - last time I used 75 ml.

I'll be starting with the OD=0.02 cells frozen in glycerol in the last preps, rather than from fresh cultures.  These cells are already in log phase; I'll just pellet them to remove the glycerol and resuspend them in the 75 ml sBHI.  The initial density will be about OD=0.003, but previous experiments suggest that not all these cells will be viable, so the cultures may take several hr to grow back to OD=0.02. Then I'll concentrate 40 ml by filtration, resuspend the cells in 4 ml, and freeze 2 ml for RNA and 1 ml for competence assay.  The remaining culture (~ 20 ml after OD sampling) will continue shaking to OD=0.6 and OD=1.

I'll need to have the filters ready, and all the tubes to put the samples into, each labelled and preloaded with RNA-Protect or glycerol.

What are the cultures?

  • 3 replicates of KW20 (names on tubes: FK, GK, HK)
  • 2 replicates of murE749 (names: G7, H7)
  • 2 replicates of sxy1 (RR563; names: G5, H5)
  • 1 replicate of ∆crp (RR668; name: FC)
  • 1 replicate of ∆sxy (RR648; name: G6)

Why isn't competence regulated by the availability of DNA?

Most bacteria tightly regulate the genes that enable them to take up DNA from their surroundings.  This makes sense, since the uptake machinery is complicated, probably expensive to produce, and may interfere with other membrane functions, and since the benefits of DNA uptake may arise only under particular circumstances.

The regulatory signals include diverse physiological and environmental cues. In other posts I've discussed the signals that regulate H. influenzae competence, and here's a couple of recent reviews for Gram-negative and Gram-positive bacteria (;  (Unfortunately neither is open access, sorry.)

But none of these bacteria are known to regulate competence by what should be the most important information - whether or not there's any DNA in their environment to take up.  I say 'known to' because in fact there is almost no published information addressing this point, and the researchers I've contacted don't know of any relevant data. 

It's possible that bacteria don't need this information because DNA is always within reach in their environments, but we don't really have data addressing this point either.

My lab has a new undergrad who doesn't yet have a project to work on.  I wonder if he'd like to test this point, first in H. influenzae and then maybe in other competent species?  Given the lack of positive evidence I'd expect negative results (external DNA doesn't affect competence), but I think this might be a case where negative results would be publishable.

The tests are tricky because addition of external DNA is also how we sensitively measure competence, so we might need a way to get rid of the 'inducing' DNA before measuring competence with the 'assaying' DNA. And good controls...

UBC's new website

Here's the email UBC sent everyone the day before they launched their new web presence:
"To the UBC Community,

We are pleased to announce that a significantly redesigned website will launch on April 25th, 2014.

Consultation and research showed that the current site could do a better job reflecting the true nature and scope of our university. The navigation, content
and functionality of the current site makes it difficult for our visitors to find the information they are looking for and the overall look is considered conservative and dated. The redesign addresses these challenges; it is bold and experiential, offering improved design, navigation and content. A few highlights of these changes are:

1. Moving from a solely internally-focused navigation structure to one that is also audience-based

2. Expanding the opportunity for faculties and units to share their content through a new homepage section called “UBC Now”

3. Creating innovative and in-depth stories on the homepage that illustrate the impact UBC is making in the world

4. Implementing secondary pages that provide a stronger introduction to internal UBC partner sites to improve site navigation and hand-off

We are encouraged by the positive feedback we have received from those who have seen the site while in development and will continue to monitor its performance and fine-tune as required following launch.

We wish to thank the large number of contributors from across our campuses for bringing their energy and ideas to the project.

We invite you to explore the site in the days ahead:

Kari Grist
Managing Director
Communications & Marketing

UBC a place of mind"

 And here's the home page:

Actually there are several other versions with different photos, all of people evidently thrilled by what they're discovering at UBC.  The little dots in the ring around his head are links to pages about specific aspects of UBC's wonderfulness.

And here's the obligatory xkcd cartoon.  Rich FitzJohn found this and points out how obedient UC remains to its precepts.

A new mutation causing hypercompetence

I described last month how we were revisiting on old hypercompetent mutant whose causative gene was unknown.  I rechecked its phenotype and prepped DNA to sequence, from both the original EMS-induced mutant strain (strain RR735) and from a 'backcross' strain where the unknown mutation had been transferred to an unmutagenized genetic background by transformation (strain R753). 

Here's the phenotype again.  The lower graph shows that it grows slightly slower than wildtype, and the upper graph that it has a 10-200 times higher transformation frequency in the rich medium sBHI.

The postdoc just emailed me the sequencing results.  Both the original and backcross strains have the same single mutation, an amino acid substitution in the rpoD gene, which encodes the sigma-70 transcription factor.

This is a surprisingly clear result.  We had expected to find many EMS-induced mutations in the original strain, and probably several mutations in the segments of DNA transferred in the backcross transformation, and were planning another series of analyses to sort out which mutation causes the phenotype.  But both strains have only the one rpoD mutation, suggesting that our EMS mutagenesis wasn't nearly as heavy as we had thought.  As controls we had sequenced the original and backcross strains of another hypercompetence mutant (RR749, known to have a mutation in I, and both these strains also had only the single known mutation.

A hypercompetence mutation in rpoD fits very nicely into my thinking about how competence is regulated.  I'll write a separate post about this.

Cell preps for RNAseq are all done

I'm pretty sure that I've now done all of the cell preps for our big planned RNAseq analysis, more or less as diagrammed in the previous post. 

Instead of a cya knockout mutant as a negative control (pink in the diagram I used a crp knockout.  cya encodes the enzyme that synthesizes cyclic AMP (cAMP), and crp encodes the transcriptional activator CRP, whose ability to induce transcription is entirely dependent on cAMP, so the two mutants have the same phenotype - inability to induce both the competence genes and the energy-balance genes in the CRP regulon.  I decided to use the crp mutant partly because that strain grew up first and partly just in case there are traces of cAMP in our sBHI medium.  I did one MIV-competence time course with this mutant (4 samples) and one sBHI time course (3 samples rather than the 2 in the diagram).

I also did 3 sBHI-timecourse samples of the sxy knockout as another negative control (also pink in the diagram).  I think I now have two more samples than will fit in 3 lanes of sequencing (24 samples multiplexed per lane), so I'll probably omit the OD=0.6 samples of the negative control sBHI timecourses.  But I'll process the RNA from them just in case something goes wrong with another sample.

I found my missing DNase.  I hadn't lost it, just ordered the wrong kind.  So now I have $250 worth of a high-quality DNase I don't really need, and will need to order the right kind.

I'll get to the RNA preps once I get some teaching responsibilities under control - partly the grading for my face-to-face Human Ecology course but mainly the need to rerecord ~150 lecture videos for Useful Genetics/Genetics for Life.

RNA-seq progress, problems and plans

I've been growing the cell preps for the RNA-seq analysis, as shown in the planning figure below.

I did Day C's cultures, freezing 1 ml of cells for later transformation testing if needed, and 2 ml of cells for RNA purification.  The first snag was the invisibility of the cells!  Following instructions from the former RA, I mixed 2 ml cells from each of the first samples with 4 ml of the magic 'RNAprotect' reagent, let the mixture sit for 5 min at room temperature, and spun down the cells (2 ml of mixture in each of three 2 ml tubes) in our mini-spin microcentrifuge.  The plan was to discard the liquid and freeze the cell pellets at -80 °C, for later RNA preps.  But there was no visible cell pellet!  This amount of cells typically forms a small but easily visible pellet when centrifuged, but the bottoms of the plastic tubes looked perfectly clean.

I spun the tubes again - still no pellet.  So I pretended a pellet was present, discarded the liquid, and froze the apparently empty tubes anyway.  I checked the RNAprotect instruction booklet which reassuringly said that sometimes the pellet might be invisible, but I also contacted the RA, who said that her preps had given clearly visible pellets.  Later I thawed out one tube and did a wilful-suspension-of-disbelief RNA prep using the RNeasy kit, which produced the same concentration of (high-quality) RNA as the RA's original preps.  So I'm now assuming that there are invisible cell pellets in all the tubes, and I've done Day D's preps.

The RNAeasy kit also had some surprises, a solution that was supposed to be clear (or with a bit of particulates that could be removed by centrifugation), instead was very cloudy, and centrifuging it raised the cloudy material to the top (as a diffuse scum) rather than pelletting it.  I couldn't get rid of the scum (it just redispersed when I tried to pipette it), so I went on to the next step (adding 100% ethanol), which eliminated the cloudiness completely.  (Maybe this cloudiness came from the presence of too much residual RNAprotect in my tubes - because I couldn't see any pellet I didn't thoroughly drain the tubes before freezing them.)

I would have also tested the DNA-elimination step, which uses Turbo Dnase and a DNase-inactivator chemical.  But my brand new box of TurboDNase ($250) has gone missing.  I've searched the freezer a couple of times, and racked my brain in case I put it somewhere special for 'safekeeping', with no success. My next step is to go down to Stores and have them show me exactly what the TurboDNase box looks like, so I know I have the right search image.

I've also revised my plans for the cells I'll test.  As I wrote earlier, I'm only going to do one replicate of the ∆hfq mutant in MIV, since we really should use special precautions to avoid losing small RNAs from the prep (and maybe to do strand-specific sequencing).  Since we planned on three full lanes of Illumina sequencing, each 24-fold multiplexed, this change opens up space for 8 additional samples. Four of these will come from a MIV time-course using a crp or cya knockout strain (in Day E, which will be tomorrow).  This is an excellent control since it lets us identify all of transcripts dependent on the transcription factor CRP.  I'll also include both crp (or cya) and sxy knockout strains in the rich medium cultures on Day H; taking two time points for each will give the four additional cultures to complete the first two lanes of sequencing. 

Big prep of MAP7 DNA

We're almost out of the standard DNA that we use in our transformation assays.  It's chromosomal DNA of a strain called MAP7 (because it contains point mutations conferring resistance to seven different antibiotics).

So I grew up a liter of cells and prepped DNS from them.  Now I have 25 ml of nicely viscous DNA solution.  It's transparent and colourless but I suspect it's not really pure yet, so I'm doing a second purification on 0.5 ml just to check if the apparent concentration or purity changes when examined with the Nanodrop spectrophotometer.  I'll also do test transformations, with the last of the old DNA preps as a control.

Later:  The Nanodrop spec found that the repurified DNA had half the concentration of the big stock. So I ran a gel and found that the big stock still contains a lot of RNA.  My original RNase step must not have worked very well, probably because of the high concentration of SDS and of proteinase K.  So I';ve now incubated the prep with more RNase overnight.

Getting ready for RNA-seq cell/RNA preps

The RA's missing notebook hasn't turned up, so I don't have her notes of how she prepared the samples for the RNA-seq analysis. Luckily the main procedures are ones she used in many experiments and are described in her earlier notebooks and in an email she sent me.

The basic procedures:

Collecting samples:
  1. Grow cells to desired state in rich medium (sBHI) or competence medium (MIV).
  2. Mix 2 ml with 4 ml RNAprotect reagent (Qiagen); leave 5 min at RT.  We have 100 ml and can get more quickly through LSC Stores.
  3. Mix 2 x 1 ml with 0.25 ml 80% glycerol and freeze at -80°C. (for later competence assays).
  4. Pellet RNAprotect cells and freeze at -80°C.
Preparing RNAs:
  1. Thaw cell pellets
  2. Use Qiagen RNA prep kit.  We have lots and can get more quickly through LSC Stores.
  3. Don't use the DNase step.
  4. Elute in 40 µl H20.
  5. Measure concentration of 1 µl with Nanodrop.
  6. Run 4 µl in a 1% agarose TAE gel at 60V.
Treat to remove DNA:
  1. Use volume containing 1 µg RNA (using RNA concentration from Nanodrop)
  2. Use Ambion Turbo DNase 
  3. Use protocol in RA's notebook #1 (not missing); 2 x 30 min incubations
Check RNA quality (and concentration?):
  1. Use the 'Bioanalyzer' (high-tech equivalent of gel electrophoresis) to check the size distribution of the RNAs in each prep.  Expect to see 2 strong rRNA peaks.
Treat to remove rRNA:
  1. Use Ribo-Zero kit to remove the rRNA from each sample.
Check RNA quality (and concentration?) again:
  1. Use the 'Bioanalyzer' (high-tech equivalent of gel electrophoresis) to check the size distribution of the RNAs in each prep.  Expect to see no rRNA peaks.

Here's the chart showing the MIV-competence samples I had planned.  I'm only going to do one set of the ∆hfq strain now, because our procedures aren't optimized for small RNAs (poor recovery and no strand information).  One of our summer-Honours students will be working on this mutant, and he can take my preliminary data and use it to help design an optimized RNAseq procedure.  So I think on Day C I'll only do the three strains (sxy-, ∆659 and ∆6759/660), and if this goes smoothly scale up to four strains on Day D.  Maybe on Day E I'll replace the ∆hfq strain with something else I want preliminary data about.

She lost me at the Central Limit Theorem

I've been saying for ages that I need to learn the statistical programing language R, so that I can work with all the bioinformatic data we're generating.  So yesterday I looked through the Coursera offerings and found an introductory statistics course that taught R (Data Analysis and Statistical Inference, taught by Mine Cetinkaya-Rundel of Duke University.  It started a few weeks ago, and I've spent yesterday watching the Week 1 videos and doing the Week 1 R lab. 

The labs are excellent.  They use a web-based R learning platform called DataCamp - each lab is a long series of guided exercises: with each exercise you're given a bit of instruction and asked to use it to display something or graph something or calculate something.  Integrated multiple-choice questions check your understanding - DataCamp automatically sends your results back to Coursera.

It's also very good that they're part of a basic statistics course, since I've always been disgracefully ignorant of this.  The video lectures are good -short but many, and aimed at the complete beginner.  I was initially quite pleased to be learning what 'marginal' means, and the differences between variance and standard deviation and standard error.  The course materials are very well designed.

But I started getting frustrated when I tried to think more deeply about quantifying variation.  I can sort-of understand why we want to know how variable the members of the population are, but this was never really explained, and I have no idea why we measure it the way we do.  To me it seems simplest to measure variation by calculating how different each individual is from the mean (the deviations), summing the absolute values of these and dividing by the number of individuals.  But that's not what's done.  Instead we square each of the deviations and sum that, to get the 'variance'.  But we don't use the variance (at least not yet), instead we take its square root, which we call the 'standard deviation' and use as our measure of variation.  Why is this cumbersome measurement better that just taking the mean of the deviations?  The instructor doesn't explain; the textbook doesn't explain.

In the first lecture of Week 3 we come to something called The Central Limit Theorem'.  This is apparently a big deal, but I don't know what the point is.  We've moved from considering the properties of a single sample from a population to considering the properties of many independent samples (size n) from the same population - I have no idea why.  The Central Limit Theorem tells us that, if we take many samples and calculate the mean of each one, the mean of these means will be the same as the population mean (is this not expected?), and that the shape of the distribution of means will be 'nearly normal', and that the standard deviation of the means will be smaller than that of the population, by a factor of 1/√n.  So what?  What do we gain by repeating our sampling many times.  Seems like a lot of work, to what end?

Then we're supposed to learn a list of conditions under which the Central Limit Theorem applies.  But without understanding the point, this was too much like rote memorization to me, and why should I bother?

Does fructose inhibit development of competence?

I gave a seminar at Michigan State yesterday, invited by the Microbiology graduate students (Thanks guys!).  While I was there I met with a research group that works on Actinobacillus succinogenes, a relative of H. influenze.  They were interested in improving the competence levels of this species, and I explained that induction of the H. influenzae competence regulon was controlled by CRP and cAMP, with cAMP levels determined by the availability of external fructose to the phosphotransferase system’ (PTS) sugar-uptake system.

The PI then asked me whether we’d tested the obvious prediction that adding frucose to the culture medium should inhibited the development of competence, and I was shocked to realize that we hadn’t.  At least I don’t think we have.  The PTS and fructose studies were done by a very competent PhD student about 15 years ago, and I’m pretty sure I would have remembered the result of this experiment, since it would either have nicely confirmed our hypothesis or disproved it.  I’ll check her thesis when I get back (I ‘m writing this in the East Lansing airport).

Assuming she didn’t do this, I’ll do it right away.  Two experiments really, a time course of competence development in rich medium and induction by transfer to MIV starvation medium.  I’ll grow wildtype cells in regular sBHI, and at OD =0 0.2 (log phase) I’ll collect them and resuspend them in either MIV or fresh sBHI containing either fructose or glucose (as control).  What sugar concentration should I use - is 0.5% standard?

Hmm, I just remembered some old experiments I did (20 years ago?) testing the effect of adding glucose to MIV  I vaguely remember that the cells became very unhappy  - did they die from unbalanced nutrients???  I know that glycerol added to MIV doesn’t have any effect on competence (very old publication confirmed by me).  I could use a different sugar (ribose? xylose?  one that we know H. influenzae can use).

Woohoo! RRResearch made a top-ten list!

If you've come from the list of The 10 Must Read #womeninscience Blogs, you might be a bit disappointed to see that the most recent posts are descriptions of day-to day work in my lab.  That's typical for RRResearch, but here are a few posts you might find more interesting:

Checking the hypercompetence mutants

Here's the results of the transformation time courses of the control strain 'K' (= KW20) and the hypercompetence mutants (RR735 and its backcross descendant RR753).

The lower graph is just culture growth (CFU/ml).  You can see that the two mutants (red and green lines) grew slightly slower than the control (blue line).  The upper graph is the transformation frequencies.  You can see that the two mutants behave identically, and that their transformation frequencies are 100-fold higher than wildtype in log-phase growth and 10-fold higher as growth slows. The dotted red circle indicates two samples that gave no transformants, so the values plotted are upper-limit estimates of the true transformation frequencies.

I also made the chromosomal DNA preps from these two mutants and from a parallel pair of mutants whose mutation has been identified (200 µl of each, at about 150 ng DNA/µl).  I set 50 µl aside for the postdoc's sequencing.

More hypercompetent mutants Part 1

Well, I did submit a CIHR proposal after all.  Not on the regulation of competence as I had originally planned, but a resubmission of our last proposal (from 18 months ago) to develop the knowledge base to predict transformation in vivo.  Now that's done I'm planning some research to identify new mutations that cause hypercompetence.

One project will start right away, to identify the mutation causing hypercompetence in a strain we isolated about 10 years ago.  I wrote about this a few months ago, but we didn't do anything then.  We're now going to do sequence both the original EMS-mutagenized strain (RR735) and its backcrossed derivative RR753.  As controls we'll sequence the original and backcrossed versions of a parallel isolate whose mutation is well-characterized (RR727 and RR749).

My job is to prepare the DNA preps for sequencing.  I've streaked out the strains from the freezer stocks so I can grow up the cells - making DNA is easy.  If I inoculate the cultures this morning I can make DNA later today.  (Later: inoculated them in the afternoon; will make DNA tomorrow.)

But I'll also repeat the transformation time courses for RR735 and RR753, with wildtype cells as a control. I should try to get that done tomorrow (in between appointments) so I can get the data analyzed over the weekend, since I'll be out of town Monday to Wednesday (giving seminar at Michigan State!). (Later: too many appointments tomorrow so will set things up for Saturday.  If the colonies aren't ready to count on Sunday I'll have the grad student put them in the fridge on Monday so I can count them when I get back.)

Fortunately all the strains are NovS, so I can select for NovR (that's the easiest marker to work with because it gives a low background and doesn't need expression time).

There's a chance that this sequencing will discover something interesting about EMS mutagenesis.  As far as I know nobody has used genome sequencing to examine the effects of EMS in bacteria (I surveyed the community via Twitter but nothing was known), though the post-doc thinks this may have been done in Drosophila.

Part 2 of this post will be on strategies to isolate large numbers of new hypercompetence mutations.

Who really discovered trisomy 21? (righting an old wrong)

A few days ago a French student in my Useful Genetics Coursera course posted a link to an article in Le Monde (sorry, it's both in French and behind a paywall, but this link might get you a translation).  It reported that a Jan. 31 award ceremony for the discovery of the cause of Down syndrome, part of the 7th Human and Medical Genetics Congress  in Bordeaux, had been blocked by a Down syndrome support organization (Fondation Jerome-Lejeune).  The back story is very depressing, an egregious example of a woman scientist being denied credit for her discovery.

Photo source: Le Monde
The woman is Dr. Marthe Gautier, now 88 years old.  In 1956 she was a young physician, returning to Paris from a year's study of pediatric cardiology at Harvard.  She was given a clinical/teaching position at a local hospital, with no funds for research.  The Head of the Pediatric Unit, Raymond Turpin, was interested in mongolism (as Down syndrome was then called); years earlier he had proposed that it might be caused by a chromosome abnormality.  Human cytogenetics was not well understood, but a big breakthrough came this same year, when the true chromosome number was finally established as 46 (not 48).  When Turpin complained that nobody was investigating his hypothesis, Gautier proposed that she take this problem on, since her Harvard training had introduced her to both cell culture and histology.  Turpin agreed to provide a tissue sample from a patient.

For this work she was given a disused laboratory with a fridge, a centrifuge, and a poor quality microscope, but no funding.  And of course she still had her other responsibilities.  But she was keen and resourceful, so she took out a personal loan to buy glassware, kept a live cockerel as a source of serum, and used her own blood when she needed human serum.

By the end of 1957 she had everything working with normal human cells, and could clearly distinguish the 46 chromosomes.  So she asked Prof. Turpin for the patient sample.  After 6 months wait it arrived, and she quickly was able to prepare slides showing that it had not 46 but 47 chromosomes, with three copies of a small chromosome.  But her microscope was very poor, and she could not identify the chromosome or take the photographs of her slides that a publication would need.

All this time Prof. Turpin had never visited her lab, but she'd had frequent visits from a protege of his, Jerome Lejeune.  When she showed Lejeune her discovery, he offered to take the slides to another laboratory where they could be photographed.  She never saw the slides again, but the photographs appeared in Montreal two months later (August 1958), where Lejeune announced to the International Conference of Human Genetics in Montreal that he had discovered the cause of Down syndrome!  Lejeune and Turpin quickly wrote up 'their' discovery, with Gautier as middle author, but Gautier only learned about this publication the day before it appeared in print.

These were tough times for a woman scientist in France, and Gautier decided not to fight for the credit for her discovery, instead returning to her clinical and teaching work on congenital heart diseases.

Lejeune became not just a renowned researcher but the darling of the French Catholic right-to-life movement.  You can read long flattering Wikipedia biographies in both French and English. He was showered with awards and given a prestigious Chair of Human Genetics at the Paris School of Medicine, bypassing the usual competition.

When prenatal diagnosis became available Lejeune campaigned against it on religious grounds. He became a friend of Pope John Paul II and was appointed President of the Pontifical Academy for Life (Wikipedia), the Catholic think-tank for medical ethics.  He died in 1994.  The Fondation Jerome-Lejeune was established in his honour; there's an American branch too.  This foundation provides funds for research into Down syndrome and support for families and patients, but only in the context of very strong opposition to abortion.  They're also campaigning to have Lejeune beatified by the Vatican.

But Gautier's role in the discovery of trisomy 21 was not totally forgotten.  It has been very well described in a 2009 article in the journal Human Genetics by Gautier and Peter Harper, the author of a major history of cytogenetics (paywalled but try this link to a pdf), and in a 2013 interview.  There's also a French Wikipedia page about her.  But few people in the field know about this injustice, and cytogenetics textbooks and courses still credit Lejeune for the discovery.  Gautier has no English Wikipedia page, and the Wikipedia pages on Lejeune describe her contribution as follows:
"Using a new tissue culture technique brought back from the United States by his colleague Marthe Gautier, Lejeune began working with her to count the number of chromosomes in children with Down syndrome. The laboratory notebook begun by Dr. Lejeune on July 10, 1957 indicates that on May 22, 1958, he succeeded in showing, for the first time, the presence of 47 chromosomes in a child with Down syndrome."
Not surprisingly, the Fondation Jerome-Lejeune strongly opposes any correction of the scientific record, since this would reveal the intellectual theft at the base of their hero's reputation.  That's why they sent in the bailiffs to record her award ceremony.

I'm of course outraged to learn about this situation, and this post is one attempt to set the record straight.  My other venue is Wikipedia, which I've been learning to edit.  So far I've added sentences crediting Gautier with the discovery to the Wikipedia entries on Down Syndrome and on Jerome Lejeune.  Someone else had added a mention of the dispute to his French entry.  I'm going to expand these each into a paragraph.  I've also created an empty page for Marthe Gautier and requested that the Wikipedia translation people fill it from her French entry.

So please spread the word.  Marthe Gautier discovered that trisomy 21 is the cause of Down syndrome, and Jerome Lejeune's saintly reputation is based on scientific fraud.

Later:  I've corrected an error: Gautier could not identify the trisomic chromosome with her poor microscope.  And here are links to news articles about this controversy, in Nature and Science.

Plans for RNA-seq analyses

I should have posted this after last week's lab meeting but am only now getting to it.  I sensibly took snapshots of the whiteboard at the end of that lab meeting, so I could check what we'd decided.

The issues:  We have several Haemophilus influenzae mutants whose gene-expression profiles we want to examine, either during competence development in the MIV starvation medium or during normal growth in the rich culture medium sBHI.  For most of these (i) we want samples from several timepoints over a few hours, (ii)we want wildtype controls done in the same experiment, and (iii) we want three replicate samples from experiments done on different days.  And it would be nice to have multiples of 24 samples, since the kits and sequencing are most efficient with that.

That's a lot of constraints , but we came up with a plan that meets them all:  The first two days are the samples that have already been prepared and sequenced; the other 6 days are for me to generate the samples.

The samples will consist of viable cells frozen in glycerol (one or two 1.5 ml tubes), and duplicate pellets of cells that have been briefly incubated with a RNA-protection reagent to stabilize their RNA.  Later the frozen viable cells will be thawed and transformed to check that they have the expected level of competence - I can probably do all of them in a few days.

After all the samples have been collected, the RNA-prep pellets will be thawed and the RNA isolated using a kit.  After a quick check of RNA concentration and size, the contaminating DNA is removed by treatment with Turbo-DNase.

All the samples are then checked for concentration and quality using a special something (the post-doc recommends, using equipment in another lab), then treated to remove the bulk of the ribosomal RNA (this kit costs a lot, I think $200 per sample), then rechecked for concentration and quality using the special something.

Finally the samples are ready to be made into multiplexed sequencing libraries (expensive service on campus) and then sequenced (another service on campus).

To order (first check what we have on hand):
  •  RNA protection reagent for 72 samples
  • RNA prep kits for 72 samples
  • Turbo-DNase for 72 samples
  • Does the 'special something' require special reagents?
  • rRNA-removing kits for 72 samples
The reasons for doing these experiments are described briefly in this post.  

Because my hydroxyurea experiments found that cell doubling is substantially faster when cultures are at very low cell density, for the first samples in experiments F. G and H I'll use large volumes of cells at a lower cell density than the usual OD600=0.2.  Getting the right density will be a bit tricky since we can't use OD600 to check densities of very dilute cultures.  In the table above I said I'd use an OD600 of 0.02, but I'll start by growing cells from a lower density to about this OD and then diluting them 16-fold so they'll have another four doublings before I sample them.  The hydroxyurea-experiment cultures were much more dilute than this, but I need to balance culture density and culture volume so I'll have enough cells for the RNA extractions.  With this plan I'll be collecting cells from about 2 x 250 ml of culture for the first samples, and combining the remaining volumes in the two flasks for the later samples.

Other experiments to do:  We need to sequence the genome of strain RR753, to identify candidate mutations that cause its hypercompetence.  Before doing this I should recheck it and the backcross strain I made - if they both are hypercompetent we'll want to sequence both.

But it's MY figure!

The postdoc and I have a minireview coming out in an ASM journal, and we're at the 'permissions' stage.

One of the figures is an explanatory diagram I drew for this article (figure on the left below).  It's similar to an explanatory diagram I drew for an article we published in Genetics a few years ago (on the right below).  Because one of the reviewers asked if our minireview figure was adapted from the published figure, in the Figure Legend we wrote '(adapted from (91)' even though it wasn't. 

Now the Production Editor is asking us for the formal permission from Genetics to republish this figure. I explained that it wasn't really adapted from the published one, but because the figures are similar she recommends that we request permission.


So Genetics has a 'Get Permissions' link beside each article, and that sends me to the Copyright Clearance Center, where I specify that I want permission to republish their content in a journal.  I click on Price & Order (not a good sign). 

That takes me to another form, asking me questions I don't understand ('Duration of use'?  'Lifetime unit quantity'?).  Whatever.

I click on 'Get Price'.

I click on 'Continue'.  This takes me to a 'Login' page.

 Of course I don't have an account.  I create an account.

Now I log in with my account, but I don't see any record that they remember what I was asking for.  So I do it all again.

Several more pages.

 I click 'Add to Cart'.

My order is not complete until I click 'Checkout'.

My order is not complete until I accept the Terms and Conditions.

OK, my order is complete.  They're sending me an email confirmation.  They say it could take 15 days....

(Added later:  In fact it took less than 24 hr!)

On the other hand, I  just also requested permission to use a figure from our Nucleic Acids Research paper.  Very quick and easy - permission granted immediately.