Risks and benefits of workouts while fasting (1|2)

by Christoph
 

Warning: No intro today, I'm about to make one big leap right into the subject. It's about Bacillus subtilis but I will not touch what this multi-talented bacterium is so well known for: bio­film forma­tion, sporulation, motility by swimming, swarming motility, cannibalism, and "natural compe­tence" for DNA uptake (aka transformation). It's about the reaction of B. subtilis to food depriva­tion. And no, not its reaction to starvation you certainly remember from microbiology classes: sporulation (see here in STC). Leendert Hamoen's lab at the Swammerdam Institute for Life Sciences (SILS), University of Amsterdam, The Netherlands, studied the reaction of B. subtilis to starvation – 'deep starvation' as they call it – in a mutant that cannot sporulate due to a deletion of the spoEII gene that triggers the 2nd step in the sporulation pathway (for the experts: they didn't use a spo0A mu­tant because Spo0A regulates, in addition to the 1st step of sporulation, spoEII activation, a ple­tho­ra of other cellular processes). So, what was their "deep starvation regime"? Gray et al. cultured the bacteria in SMM medium (carbon-supplemented minimal medium) at 37°C with aeration for 48 hours, thus keeping the cultures in stationary phase for about a day. They collected the cells by vacuum filtration at 37°C to prevent a temperature shock because B. subtilis reacts immediately to temperature downshifts. They then resuspended the cells in starvation buffer (minimal medium sans carbon sources) and incubated them at 37°C while shaking for 14 days with periodic sam­pling.
 

Figure 1. Membrane potential levels of B. subti­lis ΔspoIIE cells after deep starvation for 0, 4, 7, 11, and 14 days. Relative membrane potential levels were de­termined based on the uptake of the membrane potential sensitive fluorescent dye 3,3'-dipropyl­thi­adi­carbocyanine iodide (Dis­c3(5)). As controls, membrane potential levels of exponentially growing cells (OD₆₀₀ 0.2), and cells treated with the ionophore gramicidin ABC (10 μgml⁻¹) were determined. Fluorescence in­tensities of approximately 100 cells were quan­tified (median in red). Two biological replicates are presented here and here. Source

When Gray et al. looked at starved ΔspoIIE cells they found them almost coccoid in shape, reduced to ~1.5 µm in length as compared to the ~2.5 µm long rods of un­starved cells, and lacking flagella (see frontispiece). The cells lost their rod shape during the fist 2 days of star­va­tion, indicating reductive cell divisions, and maintained the coccoid morphology for the monitored starvation period of 14 days. Morphological changes caused by long‑term nutrient deprivation, primarily a reduction in cell size, were found earlier for several Gram-negative Gammaproteobacteria and the rod-shaped Flavobacte­ri­um columnare (its cells curl up into what looks like do­nuts), and also for Gram-positive, non-sporulating bac­te­ria: Staphylococcus aureus cocci shrink in diameters from 0.7 µm down to 0.4 µm, and Micrococcus luteus cells re­duce their diameters from 1.3 µm down to 0.4 µm. Star­ved bacterial cells frequently slip into a viable but non-culturable state (VBNC) that more often than not de­pends on a dedicated resuscitation regime for a "restart". Not so the starved B. subtilis ΔspoIIE cells because their plating efficiency correlated well with the cell counts over the entire starvation period. Also, their recovery period (lag phase) when diluted into fresh medium for re-growth was the same as that of over-night sta­tionary phase cells. But since the starved cells showed no drop in colony forming units (CFUs) upon plating to non-selective media after exposure to several antibiotics for up to 8 hours each, they clearly have an increased resistance to antibiotics. Therefore, the authors assumed that they were dormant, that is, have a maximally reduced metabolism during starvation. A reliable indi­ca­tor for such a "maximally reduced metabolism" would be the maintenance of the membrane po­tential (or proton motive force) that is essential, for example, for the cell's capacity to fuel its me­ta­bolism by ATP production. And indeed, when the authors measured the membrane potential in individual cells using the voltage-sensitive dye DiSC3(5), the deep-starved cells had membrane po­tential levels almost comparable to those found in exponentially growing cells except for day 0 when the cells were probably still in some kind of "metabolic imbroglio" (Figure 1).
 

An aside: I like this figure a lot. Maybe because I'm so familiar with looking at many bacterial cells simultaneously in the microscope? I'm fluent in "reading" bar charts of the box-and-whisker plot type with indications of the mean, the standard deviation, etcetera, but I've rarely seen such an intuitively readable presentation of statistical data. Gray et al. mark the median as "visual handle", sort of, but refrain from indicating upper/lower quartils, and min/max values and thus avoid clut­tering of their figure. But that's not the point. What's important is that one can see immediately the type of data distribution, and, concurrently, this presentation conveys to the reader a sense for the inherent noise of fluorescence measurements of single bacterial cells by microscopy. "Noi­sy" because of the technical challenges of these measurements and because the cells themselves don't "behave" uniformly, the starved cells even less so. In addition, one can hardly miss to see that, in all samples from day 4 onwards, about 10% of the cells struggle or fail to maintain the mem­brane po­tential (confirmed in the biological replicates, see here and here). This indicates the imminent death of this 10% fraction as it is known that B. subtilis cells are prone to lysis as a con­sequence of loosing their membrane potential.
 

Figure 2. Growth under deep starvation con­di­tions. B. subtilis ΔspoIIE cells (strain DG001) were incubated for 14 days in starvation buffer. At re­gu­lar time intervals samples were withdrawn and incubated with the cell division inhibitor 3­methoxybenzamide (3-MBA) for 48 h. a Cells were stained with the fluorescent membrane dye FM5‑95 in order to determine cell length before and after 3-BMA treatment. Scale bar is 2 μm. b Average change in cell length was cal­culated for ap­proximately 100 individual cells for each time point. Bar diagram depicts the average and standard deviation of three inde­pendent experiments. Source

But how dormant is "dormant"? Gray et al. went to great lengths to figure this out for deep-starved B. subtilis Δspo­IIE cells. By tagging the cells with an IPTG-inducible gfp gene (green fluorescent protein) and inducing such cells for 4 hours after 7 and 14 days of starvation, respectively, they measured high and rather uniform levels of fluores­cence in >75% of the cell population. Much like sleep­wal­kers running off with their eyes closed when they're nud­ged, the "dormant" cells are obviously able to immedia­te­ly synthesize large amounts of protein upon induction. In a different approach, the authors found, by using a Div­IVA-GFP reporter fusion, that the cell division protein Div­IVA localized at midcell in ~20% of deep-starved cells at days 4 through 14, indicating ongoing rounds of cell di­vi­sion (see frontispiece, legend see below). Yet, even coc­coid cells have to grow – at least a bit, and to replicate their chromosomes – before they divide. To detect signs of cell growth, the authors took samples of their deep-starved cells at different time points and incubated them for 48 hours with 3-meth­oxybenzamide (3-MBA). This drug inhibits the key cell division protein FtsZ of B. subtilis, which leads to a block of cell division and, consequently, results in further elon­ga­tion of growing cells. All cells taken between days 2 and 14 and treated with 3-MBA roughly doubled in length during the incubation period of 48 hours (Figure 2). A neat internal control: cells sampled at day 0 and exposed to the "division blocker" 3‑MBA could not undergo the reductive divisions typical for such "young" starved cells. Gray et al. could thus show that the deep-starved cells grow continuously albeit at a snail's pace, that is, with an estimated doubling time of ~4 days (doubling time of exponentially growing B. subtilis cells in carbon-supplemented minimal medium: ~90 min).
 

The small coccoid, starved B. subtilis ΔspoIIE cells and 'classic' stationary phase B. subtilis cells have in common that they show tolerance to antibiotics. On the other hand, the small cells are capable of synthesizing high levels of protein much like exponentially growing B. subtilis cells. So, how ma­ny and which genes are up- or down-regulated to maintain a "maximally reduced metabolism" du­ring deep starvation of B. subtilis ΔspoIIE cells? Gray et al. performed a transcriptome analysis of mRNA obtained from 14 days-old starved cells, and, for comparison, of exponentially growing cells and of cells grown into stationary phase for 20 hours. The heat map of the averaged profiles of their RNAseq data shows (Figure 3b) that all three growth conditions have a distinctly different pat­tern of up- and down-regulated genes. The Venn diagram shown in Figure 3c let the authors quantify these differences: while a small set of 30 genes is uniquely up-regulated in the starved cells, 165 genes are at least 2-fold up-regulated when compared to the exponentially growing cells, and 145 different genes are up-regulated when compared to the stationary cells. The majority of the 30 genes uniquely up-regulated in the starved cells are involved in nutrient acquisition and utilization. The 21 genes uniquely down-regulated in the 14-day old starved cells are mainly in­volved in biofilm formation or development of competence.
 

Figure 3. Transcriptomes (RNA-seq) of B. subtilis ΔspoIIE were analyzed for cells incubated 14 d in starvation buffer (in triplicate), and for wild­type cells grown in Spizizen minimal medium (SMM) and harvested in exponential or statio­nary growth phase, respectively (each in dupli­cate). b Heat map of the averaged transcript­ome profiles. Expression levels were trans­for­med to Z-scores. Green indicates low expres­sion (Z-score =−3.5), yellow indicates average expression (Z-score = 0), and red indicates high expression (Z-score=+3). c Venn diagram indi­ca­ting the number of up-regulated and down-re­gulated genes in 14-day-starved cells compared to exponential growth phase cells (left) and sta­tionary growth phase cells (right). Strain used: B. subtilis PG344. Source

The frugal solitary lifestyle of the B. subtilis ΔspoIIE cells is well reflected in their transcriptome and, taken together, the global mRNA analysis allows to conclude that their "dormant" state is not a mere variant of either of the other two states tested, exponential and (early) statio­na­ry. It deserves to be treated separately, and Gray et al. suggest to call it the "oligotrophic growth state". To give you a more quantitative notion of what "oligotrophic" means here: the authors filtrated ΔspoIIE cells that had been incubated in starvation buffer for 7 days – they were thus fully adapted – and resuspended them for further incubation in buffered solutions containing 0.5 μg/ml peptone or tryptone, or the filtrate of the 7 day culture, or fresh starvation buffer (zero carbon). The co­lony counts (CFUs) remained constant during further in­cubation for 14 days in all but one of the samples: cells grown in fresh starvation buffer (zero carbon) died-off at a constant rate and were down by 7 logs at day 14. Thus, with just 0.5 μg/ml peptone in the growth medium B. sub­tilis ΔspoIIE cells can sustain their "oligotrophic growth state". Compare this to 10 mg/ml tryptone plus 5 mg/ml peptone in "rich" media like LB that are commonly used to achieve maximal growth rates of B. subtilis. 10.000-fold diluted rich medium, that's "frugal" by all means! You're not so much into drama? Well, the carbon concentrations in Spizizen minimal medium (SMM) are roughly 10-fold lower than in LB, so their 0.5 μg/ml peptone solution compares to ~1.000-fold di­luted SMM. Still "frugal".
 

Some theoretical musings (cerebral workout). On the population level, the oligotrophic growth state is probably advantageous for Bacillus subtilis because this adds to its repertoire another long-term survival strategy, an alternative to sporulation. No big surprise there! In the intro, I cal­led B. subtilis "multi-talented" for a reason: this bacterium has, under all known circumstances, more than one card up its sleeve. And, contrary to what you might assume, spores are not a "safe bank" since spore germination is, to the best of our knowledge, driven by both external cues and by stochastic processes within the spores, which inevitably leads, again and again, to "untimely" germinations (untimely = no food available) and thus to a slow but steady decline of the capacity for long-term survival of the species. On the level of the individual cells, however, "workouts while fasting" are risky. The chances of individual cells for long-term survival are marginally worse than 50%. How come? The oligotrophic growth state leaves them in a highly precarious situation even when they have successfully managed to survive the initial bottleneck (that is, the initial drop by two logs in CFUs after the culture enters the stationary phase). As Gray et al. have shown, deep-starved cells grow and divide continuously, albeit slowly, and they depend on that what leaks from deceased and lysed siblings for nutrition after exhaustion of the few carbons in the medium (I de­liberately say 'carbons' as it's not just 'low carb', carbohydrates). A key hint for this may come from the ~10% fraction of the cells that obviously struggle to maintain their membrane potential, which leaves them prone to lysis (Figure 1). Gray et al. found respectable cell numbers (~10^4 CFU/ml) for the ΔspoIIE mutant after 120 days of incubation under deep-starvation conditions, a number that did not change significantly after having stabilized at around day 60 (from an initial ~10^8 CFU/ml at day 0). This "stabilization" indicates that the doubling rate equals the death rate, in a first approxi­ma­tion. From "oligotrophism" to "saprotrophism", if you wish (also "necrotrophism" would be to-the-point but sounds somehow decadent). This situation in B. subtilis is strikingly si­milar to that of Escherichia coli exposed to long-term starvation, which was studied in great detail by Mechas Zam­brano, Roberto Kolter, and their collaborators more than 20 years ago (Disclosure: it wasn't neces­sary for Roberto to point this out to me, and in fact he didn't). However, an ad hoc comparison of both systems is not possible because transcriptome analyses weren't feasible for E. coli in the 1990s, and Gray et al. have transcriptome data for early stationary cells but not for later time points. Also, we do not yet know whether B. subtilis accumulates fitness-increasing mutations du­ring long-term starvation as does E. coli, the well-known GASP mutations (growth advantage in stationary phase).
 

 
 

Frontispiece: Presence of cell division sites. B. subtilis ΔspoIIE cells encoding divIVA-GFP reporter were incuba­ted for 14 days in starvation buffer. a At different time intervals samples were taken for microscopic analysis. Cells were stained with the fluorescent membrane dye FM5-95. Scale = 2 μm. b (see here) To avoid bias, the captured phase contrast images were used to select approximately 100 cells per sample, and the number of cells containing a midcell DivIVA-GFP signal, indicative of cell division, were counted. As a control, the number of division sites present in both exponentially growing cells (OD₆₀₀ ~0.2), and stationary growth phase cells (overnight) were determined. Bar diagram represents average and standard deviation of three independent experiments. Strain used: B.subtilis DG001 (spoIIE::erm divIVA:divIVA-GFP(Cm)). Source