by Christoph Weigel
Making mutants I. Having studied the famous Luria-Delbrück experiment from 1943, you know, as a geneticist, how to "make" an E. coli strain resistant to the streptomycin: grow cells to late-log phase in a flask, withdraw 1 ml of the culture, spread 0.1 ml aliquots on several agar plates containing 20 µg/ml of the antibiotic, and incubate the plates overnight at 37°C. You would not be able to "see" single streptomycin-resistant cells in the crowd* but already before sunrise (and equipped with a magnifier) you would be able to spot micro-colonies, each a pile of ~106 siblings of one plated resistant cell. Later that day, before lunch, you see on the plates 2 − 6 colonies thriving in the presence of a lethal dose of streptomycin: the mutant cells you were looking for, each colony a novel strain. And from the paper by Salvador Luria & Max Delbrück, you know that, prior to imposing selection, around 10 of the ~109 cells in your 1 ml aliquot already had a mutation conferring resistance — the small number reflecting the combined precision of DNA replication and post-replication mismatch repair.
1, 2, 3 – Dead. The drama in short: 1 cells take up streptomycin easily; 2 the antibiotic compromises the translational accuracy of ribosomes; 3 the ribosomes synthesize incomplete and/or faulty proteins of every kind (structural proteins, enzymes, membrane proteins...) that accumulate in the cells with fatal consequences; Dead at streptomycin concentrations above ~2 µg/ml, E. coli cells are doomed. Point mutations in the rpsL(strA ) gene that encodes the small-subunit ribosomal protein S12 rescue E.coli cells by making their ribosomes refractory to the adverse effect of the antibiotic. Such streptomycin-resistant ribosomes translate at higher-than-normal accuracy albeit at the cost of decreased processivity; in mixed cultures, a strA mutant is easily overgrown by the wildtype parent strain. So you were lucky to have a few strA mutant cells in your culture at the time of plating! All this is no case for "induced" mutations and Charles Darwin, who speculated that variations (=mutations) were induced by stress, probably wrong.
Making Mutants II. In 1988, enter John Cairns, Julie Overbaugh & Stephan Miller with their puzzling story of "directed mutations", the observation that a nongrowing population of an E. coli lac frameshift mutant appears to specifically accumulate Lac+ revertants when starved on medium with lactose as sole carbon source. The Lamarckian malodor of this observation made quite a splash among geneticists! John Roth & coworkers made the point that Cairn's Lac+ revertants weren't so "directed" after all (see Fig. 2) and it's not generally accepted that "adaptive mutations" do in fact happen (Sorry, Charles). We know, however, that E. coli cells respond to various kinds of stress — and being deprived of dinner is stressful! — by induction of the RpoS-dependent stress response and/or the SOS response, which includes the expression of error-prone DNA polymerase(s) — that raise the mutation rate by ~4 orders of magnitude — and the stimulation of recombination-triggered replication & repair processes. Therefore, a transient increase of the mutation rate in cells growing under stress conditions cannot be ruled out, and selection of a particular phenotype may make corresponding mutations appear to be "directed". Yes, this is complicated. And in all these experiments seeking to prove/disprove "adaptive mutations", the first adapted cells in a stressed culture have always been elusive. Not so anymore!
Microhabitat Patches. Several years ago, Robert Austin's lab turned to cultivating E. coli in what they call "microhabitat patches" (MHPs). To study the evolution of mutants resistant to the antibiotic ciproflaxin (cipro), they shrank the Galapagos Islands of Darwin (as the authors put it) to a 2x2 cm array of ~1,200 interconnected hexagonal microhabitats, etched to a chip, each 200 μm in diameter, and each of which can contain up to several thousand bacteria (Fig. 3). To generate a gradient of antibiotic concentrations within the array, nutrients circulate around one half the perimeter of the device (Fig. 3 A), while nutrients+cipro circulate around the other half (Fig. 3 C). The nano-slits at the periphery (100 nm deep, 6 μm wide) allow for nutrient and nutrient+cipro flow but can't be crossed by bacteria; they are instrumental for generating emergent population gradients (even in the absence of the antibiotic) because 10 μm wide channels connecting the microhabitats allow bacteria to move freely between different populations, the inhabitants of individual microhabitats (Fig. 3 D). To prevent evaporation from the shallow microhabitat ponds, a glass coverslip coated with highly oxygen-permeable PDMS is applied as gasket to the chip, covering the entire array once it has been inocculated with ~106 cells through the center-hole inlet (Fig. 3 A). The finesse of this assembly is described in more detail here.
Within less than 5 hours, Zhang et al. observed the emergence of cipro resistance in E. coli at an external antibiotic concentration of 10 mg/ml maintained downstream of the introduction port (Fig. 3 A), a concentration that is about 200x the minimum inhibitory concentration (MIC). When bacteria were inoculated into the center of the device, chemotaxis toward increasing nutrient concentration quickly led to their accumulation at the perimeter against the nanoslits, the source of fresh nutrient supply. The already highly resistant cells, being the fittest, were the first to arrive there. Oops, this was way too fast to understand the emergence of resistance better...
Making Mutants III. Therefore, in a follow-up study Bos et al. tried to slow down this race by studying the response of wildtype E. coli cells to sub-lethal cipro concentrations (0.125x MIC) in MHPs. But also here came a quick response by the cells: within 3 hours the colonized microhabitats were full of filaments, cells that continued growing but had stopped dividing (Fig. 4 A + B). Elio concluded in an earlier STC post on bacterial filamentation: "Be that as it may, filamentation is to bacteria what fever is to children." In this case, the trigger prompting the bacteria to respond with fever/filamentation was ciproflaxin, which is known to specifically block the cell's topoisomerases (enzymes that control the degree supercoiling of the chromosomal DNA). Blocking the topoisomerases leads to replication fork stalling and, eventually, double-strand breaks in the DNA. Cells respond *immediately* to such lesions by mobilizing the SOS-response (see above, Making Mutants II) and by shutting down cell division. In brief:: 1 RecA protein kicks off a chain reaction by forcing the degradation of the LexA repressor, 2 LexA controls a number of stress-response genes, among them the sulA gene, 3 increaed expression of SulA inhibits FtsZ, which is responsible for septum formation, 4 cell division is blocked. In a reasonable percentage, however, normal-sized cells budded off from these filaments (Fig. 4 B + C) and proceeded on their way through the microhabitats to the perimeter of the MHP in search of nutrients. Such normal-sized cells could be subcultured and showed the low-level resistance to cipro that had allowed them to crawl against the cipro gradient in the MHP. Sequencing revealed that some of them had mutations in the gyrA gene encoding a topoisomerase subunit know to be affected by cipro, others had small deletions in the marR gene, which encodes a repressor of genes encoding efflux pumps. Lack of the the MarR repressor presumably led to an increase in efflux pump synthesis, which helped to overcome the cell division block induced by cipro in the filaments.
When a SulA-induced division block is overcome, septation and cell division usually occurs over the entire length of a filament and leads to the simultaneous 'release' of several normal-sized progeny cells. One of the more unexpected observations of Bos et al. was the budding-off of progeny cells from the tips of filaments, among them the low-cipro resistant cells (see Figs. 4 C + 5 A, and a stunning time-lapse video here). The authors speculate that "...either better-adapted chromosomes are generated near the tips or that better-adapted chromosomes generated internally migrate toward the tips, where they are then budded off by formation of a single septum." (Fig. 5 B) It appears more likely, though, that better-adapted chromosomes are generated in a random fashion throughout the filament. Also, a mechanism that preferentially transports such better-adapted chromosomes to the filament tip is hard to envision. Yet, newly made membrane proteins tend to be incorporated into membranes close to their place of synthesis — which in turn isn't far from that part of a chromosome where the respective genes are transcribed to mRNA — and are unlikely to diffuse throughout the cytoplasm of the entire filament. Therefore, marR mutants with elevated expression of efflux pumps in chromosomes near the filament tips could lower the local antibiotic concentration more effectively here than would expression in chromosomes at midcell where ciprofloxacin can "re-fill" the gradient from both sides. There is thus a slightly better chance to have the SulA-mediated division block relieved at the tips and budding taking place here. Is this reasonable? Let's see what forthcoming experiments tell...
Outlook (Bright). Of course, curious bacteriologists have for a long time been able to observe the growth of live cells on slides under the microscope but with restrictions: the necessity to use glass cover slips easily creates anoxic conditions, while media supply is hard to control, let alone maintenance of a constant temperature (unless you take your microscope to the 37°C incubator room that some microbiology institutes have). Also, monitoring a statistically relevant numbers of cells under the microscope is difficult, close to impossible. The novel cultivation technique for bacteria in microhabitat patches developed by the Austin lab and other biophysicists experimenting with microfluidics — one of the choices in STC's recent poll on innovative experimental techniques — pave the way for studying bacterial populations in controlled environments that can simulate their natural habitats far better than is possible by swirling them vigorously in Erlenmeyer flasks or 'nailing' them down to agar surfaces in Petri dishes.
Yet a 'quantum leap' in understanding evolution by studying bacteria in "microhabitat patches" may not come from technical refinements in cultivation but from exposing a preconception in the experimenters' brains. Being able to watch bacteria actively choosing the microhabitat that suits them best by swimming from one micro-pond to another will subtly erode the common understanding of natural selection, a key mechanism in evolution, as being something that is imposed on living things like, say, a physico-chemical equivalent of 'destiny'. Living things, large and small alike, always take the active part — sometimes successfully, sometimes less so.
* By using the clever techique developed by the Radman lab you could in fact see all the mutants in a starting culture but you cannot pinpoint the streptomycin-resistant cells among them.
Christoph Weigel is lecturer at the Life Science Engineering faculty of HTW, Berlin’s University for Applied Sciences and an Associate Blogger for STC