MOST OF OUR study of gene networks has been done by comparison of related species to reconstruct network evolution and by knocking out specific genes to determine what the effects of their absence are. In a new paper Isalan and coworkers try something new, reprogramming genetic networks in Escherichia coli and examining the mutants to detect viability and any possible benefits to genetic pathway modification.
Transcriptional regulation is achieved by the use of promoter regions, DNA with a particular sequence that appears upstream of a gene. Transcription factors recognize the promoter sequence and bind to it. In prokaryotes, a σ-factor can associate with RNA polymerase and direct it to specific promoters in response to environmental cues. Successful binding of transcription factors and σ-factors result in RNA polymerase producing a RNA transcript of the gene. Promoters can be chained, so that one gene’s product upregulates the next gene, which upregulates a third, and so on down to a final gene product (downregulation also occurs, but was not studied in Isalan’s paper). The result is a web of gene regulation where a few key promoters control many downstream genes. Lower down in the network there are some hubs in which a promoter controls only a few genes. Usually it has been thought that modification of the gene network is most likely to be damaging when it influences one of the key hubs. Isalan set out to test this idea.
Isalan and coworkers made 598 modified gene networks. The modifications were done on various levels, from modifying a key hub to modifying hubs much further downstream. Seven master transcription factors, seven σ-factors, and eight downstream transcription factors were examined. The diagram (reprinted by permission from Macmillan Publishers Ltd: Nature 452(17), 840-846, copyright 2008) shows two of these modifications. For example, csgD is a downstream transcription factor that controls the expression of only six genes. In one mutant, csgD was hooked up to crp, a master transcription factor that controls the expresson of 416 genes. Since csgD is at the end of a chain of regulators, placing csgD in control of crp also places csgD‘s upstream regulators rpoS, ompR, and crp itself in control of crp regulation. A second example in the diagram is a less drastic change, with fliA regulating flhD. Both of these examples create positive feedback loops, so we might expect constant high levels of these promoters.
Rather surprisingly, Isalan found that 95% of the gene network modifications made were viable. Not only were the mutants viable, but in most cases their growth was not significantly modified. Out of the strains with abnormal growth, those with ihf A+B-promoted open reading frames grew faster, probably because IHF controls transition from logarithmic to stationary growth phases.
The next phase of the study examined the potential for evolution in these modified gene networks. The bacteria were cultured and exposed to a variety of insults, such as heat shock and prolonged culture. In the more permissive conditions of repeated passaging strains with flhD-promoted open reading frame combinations grew more successfully. Since flhD normally promotes genes that produce flagella (an energetically costly process), and flhD loss-of-function mutants lack flagella and grow very rapidly, the authors hypothesize that this is due to the abnormal network position of flhD suppressing flagella growth. In repeated passaging the bacteria are essentially able to grow as rapidly as possible without the danger of starvation or poisoning by metabolic products, so a mutant strain without flagella would succeed in outpacing those that preserve flagella.
The longevity and heat-shock experiments showed strains with rpoS–ompR survived much more successfully than other strains, and beat out the wild-type bacteria. Strains mutant in one or the other promoter did not succeed, so in this case both were needed. An initial gene expression analysis with comparison to wild-type shows only 13 out of 4,000 genes were expressed significantly differently in rpoS–ompR mutants, and these included upregulated heat shock genes and chaperone genes. Repeated heat shock and gene expression analysis showed a change in gene expression in passaged rpoS–ompR strains, with an increase to 39 differently expressed genes, now including permease downregulation. The combination of rpoS–ompR allowed evolution of a very heat-resistant strain.
Since most strains succeeded in growing with comparable effectiveness to wild type it is apparent that bacterial gene networks are more robust than thought previously. Even those mutants affecting master transcription regulators tend not to be detrimental. Not only were these mutations not detrimental, but they also provided new variation for evolution under varying conditions, leading to phenotypes more successful than the wild-type.
Isalan, M., Lemerle, C., Michalodimitrakis, K., Horn, C., Beltrao, P., Raineri, E., Garriga-Canut, M., Serrano, L. (2008). Evolvability and hierarchy in rewired bacterial gene networks. Nature, 452(7189), 840-845. DOI: 10.1038/nature06847