Cryptic effects of antibiotics and resistance alleles

This morning we had a meeting with members of another lab to discuss progress on a shared project. There isn't as much progress as I had hoped, but at least we now know what still needs to be done and who will do it.

The goal is to understand how gene expression changes when cells are exposed to antibiotics at concentrations so low they don't even slow growth of the cells, much less kill them ("sub-inhibitory concentrations"; abbreviated sub-MIC).

The first part of the project was to use microarray analysis to compare the amount of mRNA produced by each gene, when cells were grown with and without sub-MIC of the antibiotics rifampicin and erythromycin. Rifampicin inhibits production of mRNA by RNA polymerase, and erythromycin inhibits production of protein by ribosomes. This work was begun by a previous technician in our lab, and completed by an undergraduate working in the other lab. (The undergraduate just learned that she's been accepted into medical school here - Congratulations, Wendy!) I have a draft version of a conference poster with some results of this analysis, but we'll need to reanalyze the data in preparation for writing the paper we plan.

The complementary part of the project turns out to be still quite a long ways from completion. The plan is to compare the gene expression by normal (antibiotic-sensitive) cells growing without antibiotic to the expression by cells that carry a mutation making them resistant to the antibiotic (i.e. of a RifR strain and of an EryR strain). This will let us compare the set of gene-expression changes caused by resistance mutations to those caused by the antibiotic. We expect these two sets of changes to be quite different, because each represents a complex outcome of different adaptive and accidental responses to a different change.

Part of our lab's contribution was to isolate the necessary resistance mutations; that's done and we have sequenced the altered DNA so we know exactly what the changes are. The Rifampicin resistance mutation is in the rpoB gene; it creates a S->P amino acid substitution at position 509. An identical mutation is known to cause Rif resistance in Staphylococcus aureus. Isolating the erythromycin resistant strain was a lot more trouble, but it's sequenced too. It's in the L22 protein (part of the 'large' subunit of the ribosome; I forgot to note down the exact position).

Our collaborators have analyzed mRNA from the EryR strain, but unfortunately the cells were being grown in the presence of erythromycin rather than in antibiotic-free medium, so we can't use this analysis as we planned. I offered to make more mRNA, this time from cells grown without antibiotic, so the microarrays can be repeated.

Several other problems were discovered.

First, the array slides used for this analysis were quite old and had been stored in air at room temperature, which causes degradation of the DNA fragments spotted on them. When repeating the analysis we will need to use new arrays, and these must be ordered from the research group in London that makes them.

Second, the very expensive license (>$4000 per year) for the GeneSpring software used to analyze microarray results has expired. It belonged to another lab that had kindly let our collaborators use it. Luckily another lab in our group of labs is thought to have just purchased a new license, so we're going to approach the head of that lab to ask if we can use their software (on their computer, as it can't be copied to other computers) in exchange for a small financial consideration.

Alternatives to GeneSpring exist, and I think some are open access, but I suspect they require quite a bit more sophistication to use well. A Google search for "alternatives to GeneSpring" led me to BRB Array Tools, which is free from NIH and runs as an Excel add-in. It was developed by 'professional statisticians' (why do I not find this reassuring?). One strength of GeneSpring is its ability to integrate the array information with the genome sequence and metabolic pathways of the organism - this is especially valuable for simple compact genomes such as H. influenzae's.

A third problem is that neither research group (ours or our collaborators) has anyone with much experience with GeneSpring (much less any equivalent free software). I've used it (a few years ago), but I've forgotten most of what I learned. Luckily we don't need to use its more sophisticated abilities, just the basic analyses, but even so it's going to take a major investment of time to analyze the data once we get it all.