Thursday, January 8, 2015

A first look at the up-regulated and down-regulated genes in the toxin and antitoxin mutants

First Look: missing down-regulated genes, and an upregulated hsd Type I restriction modification enzyme complex

Three of the strains sequenced for the MIV time course experiments have deletions in the CRP-S operon consisting of genes HI0659 and HI0660.

HI0659 seems to function as an antitoxin that prevents the HI0660 toxin from inhibiting DNA uptake and transformation.  Cells without HI0659 are not transformable, but grow normally, while cells without HI0660 seem to grow and transform normally.

Previous RNA sequencing results showed that knocking out the antitoxin seemed to slightly decrease sxy expression and expression of competence genes (expression patterns were intermediate between wild type and sxy- strains in most cases), but increase toxin expression.  Knockout of the toxin seemed to slightly increase expression of some competence genes.  The HI0659/HI0660 deletion strain was not tested in the last RNAseq.

I have looked at the files Scott put together showing which genes are up-regulated and down-regulated significantly (with a false discovery rate <0.05) for the toxin deletion, antitoxin deletion, and toxin/antitoxin operon deletion strains.

The first thing that jumped out at me is that the antitoxin and toxin/antitoxin deletion strains don't show significant down-regulation of the deleted genes on these lists.  Since the genes have been deleted, we know they shouldn't be expressed at all, whereas during competence they are normally up-regulated over time, so we should see them appear on the list of down-regulated genes relative to wild-type at the later time points. It's possible this means that the criteria for being "significantly" down-regulated are too stringent, but it's also possible that this result is an artifact of how the genes were knocked out.

The genes were knocked out using a spec cassette that replaces the majority of the coding sequence of the gene.  In the case of the toxin HI0660, which is transcribed first in the operon, everything between the start codon and the last 7 codons is replaced, and in the antitoxin HI0659, everything between the first 3 codons and the last 7 codons is replaced.  The first 3 codons are retained because they overlap with the coding sequence of HI0660.  In the previous RNA sequencing, I suspect this inclusion of a few codons from the beginning of the gene caused HI0659 to appear expressed in the HI0659 knockout strain when the data was analysed (see yellow line in image 1 below), even though looking at the coverage of bases across the whole gene confirmed that the middle of the gene was knocked out properly.

Image 1: Expression level (number of reads normalized for gene length) of the gene HI0659 over time in the sequenced strains.  Strains were HI_0659 = antitoxin deletion, HI_0660 = toxin deletion, KW20_A and _B = two replicates of wild type, SXY-_A and B = two replicates of sxy gene knockout
The graph for HI0660 appears as it should (see green line in image 2 below).

Image 2: Expression level (number of reads normalized for gene length) of the gene HI0660 over time in the sequenced strains.  Strains were HI_0659 = antitoxin deletion, HI_0660 = toxin deletion, KW20_A and _B = two replicates of wild type, SXY-_A and B = two replicates of sxy gene knockout
I think this may be what's happening in the new data as well, since the file for genes down-regulated in the HI0660 deletion mutant includes HI0660 at the T=10 min and T=30 min time points as expected.  This issue definitely needs to be looked at before any further analysis is done since other genes knocked out using the same spec cassette recombineering method may show similar patterns depending on which part of the gene was replaced.

Even apart from this issue, I didn't see much consistency in the genes that were up- and down-regulated across the different time points in the toxin- and antitoxin-deletion strains.  For example, sxy did not appear to be downregulated in any of the antitoxin-deletion time points, nor did the toxin appear up-regulated as seen before, and nothing was down-regulated at t=30.

The one consistent result that jumped out at me is that when the whole toxin-antitoxin operon is knocked out, all three subunits of the hsd Type I restriction modification complex (genes 1285, 1286, and 1287) are upregulated across essentially all time points (only 2/3 appear upregulated at t=100, but at all other time points they're all up-regulated).  These are subunits of a "host specificity of DNA" Type I restriction complex, which in E. coli recognises a specific methylation site and can both methylate hemi-methylated host DNA strands (for example, after damage repair), and cleave non-methylated foreign DNA at a non-specific site [B Sain, NE Murray. The hsd (host specificity) genes of E. coli K12. Molecular and General Genetics MGG, 1980]. Since the number of reads for each of these genes remains roughly consistent across all the time points (and they are already up-regulated at time 0), this looks like it could be an effect of the toxin antitoxin system that is not specific to competence.  It's unclear to me at the moment why there would be an effect on these genes only when both the toxin and antitoxin are deleted though.

No comments: