A spore quality–quantity tradeoff favors diverse sporulation strategies in Bacillus subtilis

Quality–quantity tradeoffs govern the production of propagules across taxa and can explain variability in life-history traits in higher organisms. A quality–quantity tradeoff was recently discovered in spore forming bacteria, but whether it impacts fitness is unclear. Here we show both theoretically and experimentally that the nutrient supply during spore revival determines the fitness advantage associated with different sporulation behaviors in Bacillus subtilis. By tuning sporulation rates we generate spore-yield and spore-quality strategists that compete with each other in a microscopic life-cycle assay. The quality (yield) strategist is favored when spore revival is triggered by poor (rich) nutrients. We also show that natural isolates from the gut and soil employ different life-cycle strategies that result from genomic variations in the number of rap-phr signaling systems. Taken together, our results suggest that a spore quality–quantity tradeoff contributes to the evolutionary adaptation of sporulating bacteria.


SUPPLEMENTARY TEXT
Knowledge of the molecular mechanisms that underlie the spore quality-quantity tradeoff is not required in order to investigate its effects on fitness. Nonetheless, insights into spore-quality determinants could help to identify conditions that might select for different sporulation strategies.
In a previous study [1] we provided evidence that the enzyme alanine dehydrogenase (Ald) affects spore quality. Here, we summarize previous findings and provide additional experimental evidence that Ald is a spore quality determinant. We then use this knowledge to develop a mechanistically inspired life cycle model that enables us to investigate theoretically under what conditions a spore quality-quantity tradeoff is likely to be of relevance to fitness, and use the model to guide the fitness experiments in this study.

Evidence for Ald as a determinant of B. subtilis spore quality
The sporulation strategy determines Ald levels in spores For most, if not all, organisms the underlying molecular mechanisms that give rise to a qualityquantity tradeoff are still poorly understood. In higher organisms it is typically assumed that the tradeoff arises as a consequence of limited resources available for investment in reproduction. The limiting resource can be assigned to the generation of progeny in accordance with various resource allocation schemes that control the number and quality of offspring in varying proportions. In a recent study we provided evidence that the metabolic enzyme alanine dehydrogenase Ald may be a limiting resource in starving B. subtilis cells. Ald is distributed among the derived ("daughter") spores following a phase of pre-sporulation vegetative growth during which Ald levels are diluted with each cell division that occurs prior to sporulation. The sporulation strategy determines the length of the pre-sporulation growth phase, and therefore, the amount of Ald available for allocation to the derived spores [1].
We have previously studied the function of Ald during starvation, sporulation and spore revival in time-lapse microscopy experiments. We found that transcription of ald is strongly downregulated, if not completely repressed, under starvation stress, presumably as a result of starvation for L-alanine. Consequently, the total amount of Ald stored in each cell upon entry into the starved state may be considered to be a fixed resource. With each subsequent cell division, Ald proteins will be shared among daughter cells and thus Ald levels should decay exponentially (assuming that the protein is stable). Upon sporulation, the cell divides asymmetrically and the smaller forespore compartment subsequently gives rise to the spore core. Hence, the Ald concentration in the spore should correlate with Ald concentrations in the corresponding sporulating cell.
This model makes several testable predictions. First, if Ald is a fixed resource during starvation, Ald levels in the spore should decrease exponentially with sporulation timing. We tested this prediction by correlating spore fluorescence from an mCherry-labeled Ald protein with sporulation time by re-analyzing time-lapse movies of sporulating microcolonies acquired previously (see https://av.tib.eu/media/33997 for a representative movie). As expected, we find that spore fluorescence levels decay exponentially with sporulation time (Fig. S1a). That is, spores produced earlier in time have a higher Ald levels than spores produced later in time.
Moreover, the timescale of fluorescence decay (0.14 h -1 ) matches well with the cellular division rate (0.15 h -1 ).
Another prediction of this model is that more rapid sporulation should shift the distribution of Ald concentrations in spores to higher levels, as fewer cell divisions occur prior to sporulation.
We previously showed that the average spore fluorescence from Ald-mCherry increased when sporulation was accelerated by induction of a sporulation kinase KinA driven by an IPTGinducible promoter. To further substantiate our previous finding, we determined the corresponding fluorescence distribution as a function of IPTG levels. The histograms clearly show the predicted shift in the distribution to higher fluorescence intensities (Fig. S1b). We next repeated this experiment in a liquid sporulation culture and again observed the shift of the fluorescence distribution to higher levels with increasing IPTG levels (Fig. S1c). Together, these data confirm that the levels of Ald in spores are controlled by the sporulation strategy on both solid and in liquid media.

Ald is a spore quality determinant
Since sporulation timing controls both the number of spores produced from a single progenitor cell and the corresponding distribution of Ald levels in the spores, a quality-quantity tradeoff emerges -provided that Ald controls 'spore quality'. The proposition that Ald is a spore quality determinant is supported by several experimental findings [1]: 1) the spore revival frequency upon adding the nutrient-germinant L-alanine to spores on agarose pads containing (exhausted) sporulation media is correlated with Ald levels in the spores, 2) spores of an Δald strain germinated normally but were blocked in spore outgrowth under the same experimental conditions, and 3) the spore outgrowth frequency could be increased beyond wt levels by overexpressing Ald in progenitor cells.
Ald is an enzyme that metabolizes L-alanine to pyruvate, and has been implicated in metabolization of the nutrient germinant L-alanine in germinated spores [2]. We thus hypothesized that the enzyme plays a role in bootstrapping metabolism in the reviving spore.
We tested this by monitoring the onset of metabolic activity in spore suspensions with the help of a colorimetric assay using the redox dye Alamar Blue (resazurin) [3,4]. The dye is reduced as germinated spores regain their metabolic activity (and replenish their NADH and NADPH pools [5]), shifting its absorption maximum from 605 nm to 575 nm. This color change can be detected by a photospectrometer by measuring the ratio R = A575 nm/ A605 nm of absorbance A (or optical density (OD)) at 575 nm and 605 nm, respectively of a germinating spore suspension ( Fig. S1d, see Supplementary Methods for details). For Δald spores the color change indeed occurred at a slower rate than for wt when induced with L-alanine (Fig. S1d, left panel). For stimulation with rich media (LB plus nutrient germinants L-alanine and AGFK) the color change occurred more rapidly but the rates were comparable in both strains (Fig. 1d,right panel). No response occurred when water was added, which excludes unspecific conversion of the dye (Fig. S1d, dotted lines) while revival without a dye increased R to a higher but constant value (Fig. 1d, Inset). Since the overall color change is affected by the metabolic activity of germinating spores and their numbers, we confirmed that Δald spores germinate like wt [1]  Together these data further support the idea that Ald acts as a spore quality determinant, which affects the capacity of spores to metabolize the nutrient germinant L-alanine. Notably, in rich nutrient media, other pathways contribute to the metabolism of the germinated spore. As a result, the Ald pathway may become dispensable for outgrowth.

Mathematical life cycle model for sporulation and revival
Sporulation. We describe the kinetics of sporulation by a minimal model introduced in Ref.
[1] and depicted in Fig. S2a. At time = 0, a single bacterium in a naïve vegetative state is placed in nutrient-depleted sporulation medium. As the seed bacterium and its progeny progressively consume the available nutrients and/or intracellular reserves, they transition to starvation. This process is modelled as a passage through a series of vegetative states , . . . , , with transition rate . In all vegetative states bacteria grow with unchanged rate ; however, in the final starved state , cell death occurs at a rate that exceeds the growth rate, > .
From all non-naïve vegetative states , . . . , , bacteria can also differentiate into spores at a constant rate . We note here that sporulation from stages < is necessary to reproduce a broad distribution of spore-formation times, as generally observed in our experiments. We denote the number of cells in state at time by ( ).
In response to starvation at = 0, bacteria switch off production of enzymes required for bootstrapping metabolism upon revival; this was shown to be the case for alanine dehydrogenase (Ald) in Ref. [1] with additional evidence provided in Fig. S1, but other such quality-marker enzymes are likely to exist. For concreteness we focus on Ald here. As cells continue to grow, the Ald amount in each cell decreases as ( ) = (see Fig. S1a), while the total amount of Ald in the clone remains constant (at ). At the moment when a spore is produced, the amount of Ald present in the cell, ( ), is encapsulated into the spore, which thereby acquires the spore quality = ( ). We denote the number of spores with quality at time by ( ).
The resulting kinetic equations are: Spore-quality density and yield. The process of starvation and sporulation produces spores of variable quality. We define the spore-quality density ( ) as the number density of spores generated with quality . The spore-quality density ( ) can be determined from the solution of Eqs. (1) - (4). We obtain: The number of spores within a given quality range is then obtained by integrating over ( ).
The total spore yield (per initial cell) is obtained by summing the spore quality density over all qualities as Eq. (6) Revival. The probability of a spore of quality to revive in an outgrowth medium with nutrient level is denoted ( ). The product ( ) ( ) then defines the density of revived spores.
Thus, the average total average number of revived spores per initial cell is given by: Eq. (7) and determines the revival success in a given revival environment.
The revival fitness is then defined as the natural logarithm of the revival success = ln( ).
Eq. (8) It provides a measure for the relative increase or decrease of the population size per starvationrevival cycle and is therefore analogous to the logarithmic growth rate, which measures relative changes in population size per time. In essence, revival fitness is determined by the fraction of spores that are of sufficient quality to grow out in the given outgrowth medium. This fraction depends on both the sporulation process (via ) and on the revival process in the given environment (via ). In this way, the minimal life-cycle model presented here links sporulation kinetics and revival fitness of spores.
Quality-outgrowth relation. In the absence of direct measurements, we resort to describing the quality-outgrowth relationship ( ) by a simple model that is motivated by the basic physiology of spore revival (Fig.S3a). We assume that revival is gated by a bottleneck in an overall metabolic flux required within the germinated spore to bootstrap vegetative metabolism. In other words, the spore will regrow with high probability if the metabolic flux exceeds some universal, medium-independent threshold value * , but for decreasing , its chance of reviving progressively decreases.
The quality markerthe enzyme alanine dehydrogenase Aldmetabolizes L-alanine in germinated spores [2] and is required for alanine-dependent outgrowth [1], but is required in much lower amounts or even not at all in rich medium [1] (Fig. S1d, e). We tentatively attribute this fact to the presence of alternative nutrient sources in the medium, made available through alternative enzymatic pathways, which also contribute to the metabolic flux . The total flux is then composed of an alanine-dependent, Ald-mediated and therefore spore quality-dependent flux and an alanine-independent, alternative flux maintained by other nutrients present in concentration in the regrowth medium. That is, (9) In the following, for simplicity, we consider the alanine content in all outgrowth media to be equal and non-limiting, so that ( ) = Eq. (10) is proportional to the spore quality alone. Here, the proportionality constant is the flux per unit amount of Ald, i.e. the Ald efficiency. By contrast, the alternative flux strongly depends on the chosen outgrowth medium; it is a measure of the richness of the medium. We take to be independent of spore quality, which amounts to the model assumption that the alternative enzymes carrying the alternative flux are not subject to dilution. This assumption is valid if their production continues during starvation, or if these enzymes are diluted during starvation but always remain at high enough concentration to be non-limiting for .
To complete the model, we take the outgrowth probability to be a saturating function of the total flux. Specifically, Eq. (11) where * sets the typical flux needed for revival. Note that within this complementary flux model, any spore can grow out if the medium is rich enough, i.e. when > * . Because the Ald efficiency acts as a scaling factor for the fluxes, the outgrowth probability only depends on the rescaled fluxes / and thus Eq. (11) can be rewritten as: Eq. (12) Here, * = * / defines an apparent spore quality threshold, which is the effective spore quality required for reviving with 50% probability on poor media by utilizing alanine; = ( )/ is the apparent gain in quality that a rich medium supplies to the spore, compared to alanine only. This implies that when ≫ * (or equivalently ≫ * ) in very rich media, revival is dominated by the metabolic flux provided by the alternative nutrients and thus any spore will revive, regardless of its quality.

Model Parameters
Introduction of non-dimensional parameters (parameter reduction) The dynamic equation Eq. (1) to Eq. (4)  To remove this irrelevant scale, we set = 1 in all equations. In effect, we measure quality (including * and ) relative to the maximal quality .

Parameter values
For our minimal model, we consider a sporulation model with = 3 non-naïve vegetative states and a starvation rate of = 1.3, which indicates that on average, 3/(1.3 × ln 2) = 3.3 doubling times are needed to progress to the fully starved state . In the fully starved state, death is = 1.5 times more likely than another division.
To model different life cycle strategies we vary the sporulation rate . To give a specific example, for fast sporulation with = 3, sporulation happens three times faster than replication, while for slow sporulation with = 0.25, sporulation is four times slower than replication. Fig. S2b shows the corresponding spore-quality densities resulting in a quality strategy (few spores of high quality) and a yield strategy (more spores but of lower quality), respectively.
To enable spore quality to be limiting for spore revival, we take * = 1, that is, the apparent quality threshold for revival matches the maximal spore quality.
To model different revival environments we vary the apparent gain in quality that is supplied by increasing the "richness" of the revival medium. Specifically, = 0 is a nutrient-poor environment, which thus selects for high quality spores. In contrast, in very nutrient-rich environment ≫ 1, any spore revives, regardless of its quality. Fig. S3b shows the resulting density profiles of revived spores vs. spore quality for the two life cycle strategies described in a poor environment = 0 and a moderately rich environment with = 1, respectively.
Effects of the spore quality-quantity tradeoff on revival fitness Fig. S4 shows the revival fitness (and the revival success ) as a function of the sporulation rate for different environments. Clearly, when all spores are able to revive regardless of their intrinsic spore quality due to the richness of the medium ( ≫ * ), a yield-optimizing strategy will optimize the revival fitness. Thus, in very rich media, revival success equals yield, → , so that slow sporulation becomes optimal. (In passing we remark that the optimal sporulation rate never approaches 0 in rich media, since in this limit, cells will die of starvation nonproductively, reducing at very low .) Conversely, in a poor medium where ≃ 0, only spores of sufficiently high quality have a good chance of revival, and a quality strategy may pay off. Whether the increased quality over-compensates for the decrease in yield depends on the details of the spore-quality profile and revival probability.
As Fig. S4 shows, a swap of optimal strategies indeed occurs in the present model: While the fitness decreases in a nutrient-poor environment for any sporulation strategy, the decrease is minimized when spore quality is optimized. The dashed line in Fig. S4 shows how the optimal sporulation shifts from slow sporulation to fast sporulation as the richness of the medium decreases. In effect, the optimal sporulation rate is slow (corresponding to a yield strategy) for rich media and fast (quality strategy) for poor media. Interestingly, in an intermediate range of medium richness, the choice of sporulation strategy has almost no bearing on fitness.

Strain construction
The Δald knockout strains (BIB1396 and BIB1397) were constructed via transformation of BIB1392 with a ald::mls R fusion product. ald-up, ald-down fragments were amplified from the B. subtilis 168 1A700 genome with MA51/MA52 and MA53/MA54 primer pairs, respectively.
The mls R cassette was amplified from pDG647 [7] with primers DW34 and DW35. The resulting fragments were fused by overlap-extension PCR using the end primers MA51 and MA54. Transformants were checked via colony PCR with the primers KN1, KN2, KN3 and LA61.
Colorimetric measurement of metabolic activity (Alamar Blue Assay) The Alamar Blue assay was adapted from (Nagler, 2014) [4]. The titer of purified spores was assessed by microscopy and spore concentration was adjusted to 5*10 8 spores ml - intervals over 3 hours. The absorbance signal A (t) was determined by averaging results from three technical replicates. The ratio R = A575nm /A605nm was then used as a measure for reporting on the color change. A control without the dye was performed for each sample to monitor for changes in optical density OD605 nm due to spore refractivity loss upon germination, and outgrowth.

Fluorescence-based spore analysis
To determine spore fluorescence levels, spores were segmented from bright-field images using Fiji software or a custom-written Matlab program, respectively. The fluorescence intensity was determined by subtracting the background fluorescence. Where applicable, individual spores were classified into groups according to specific criteria as described in the text (e.g. higher or lower intensity than median fluorescence of the spore population).

Fitness assays using spore mixtures produced in well-mixed liquid sporulation medium
Sporulation: Cells grown overnight in selective CH medium were re-inoculated in fresh CH medium (without selection) and grown to an OD600nm of 0.8 -1.0. Cells were then re-suspended in SM A medium (SM medium + 1 mM L-alanine) (final OD600nm = 0.5). Wells of a FlowerPlate (m2p-labs) were each filled with 1.2 ml of the cell suspension. Where indicated, IPTG was added to the wells. The culture was incubated for 4 days at 37°C with rigorous shaking (900 rpm) and air humidity was maintained at 75% in an Infors Multitron Pro 3mm micro-titer plate shaker. Spores were harvested and washed twice with MiliQ water and stored at 4°C before being assayed for revival.
Revival fitness: For revival assays, 4 µl spore suspension was spotted on agarose pads containing SM* media without L-alanine supplementation. These modified pads do not support spore revival without addition of further nutrients. Revival was induced by adding 10 µl of either 100 µM L-alanine solution or LB-germinant mix (33% LB, 33% AGFK, 33 mM Lalanine) on top of the pads, and the response was monitored by time-lapse microscopy. Since outgrowing spores will initiate the formation of microcolonies, the number of (micro)colony forming units (mCFU) was approximated by manually counting the number of outgrowing spores in each movie.

Sporulation of natural isolates in well mixed cultures
For the initial set of experiments (Fig. S8), spores were sporulated in culture tubes. Cells grown overnight in selective CH medium were re-inoculated in fresh CH medium (without selection) to grow to an OD600nm of 0.8 -1.0. Cells were then re-suspended in SM medium (final OD600nm = 0.5) and incubated for 4 days in a culture tube with shaking at 37 °C. Next, the spores were washed twice with phosphate-buffered saline (PBS) and stored at room temperature until being assayed for revival.
For all other experiments (Fig. S8 & Fig. S9) spores were produced in microtiter plates using a micro-plate shaker, as described in the fitness experiments. Spores were harvested after 4 days, if not otherwise indicated. Spores were washed twice with MiliQ water directly after harvesting.
Spores of the natural isolates (strains BIB1249, BIB1282, BIB2115 and BIB2117) were heat inactivated at 80°C for 10 minutes to kill any residual vegetative cells and were washed twice with MiliQ water immediately before the revival assay. Δald spores (green). The dye Alamar blue is reduced by redox active compounds produced as germinating spores initiate metabolism, which shifts its absorption maximum from 605 nm to 575 nm.
The ratio R = A575 nm/A605 nm of absorbance A at 605 nm and 575 nm, respectively is plotted as a function  other rates are now expressed relative to . Bottom: As sporulation is delayed by time , the quality marker is diluted from its initial value by cell growth with rate (left). Upon sporulation, the enzyme is stored in the spore. Thus, the spore quality q decreases exponentially with increasing delay time (right). The parameter sets the overall scales for quality and denotes the maximal spore quality = that can be achieved upon immediate sporulation. We thus normalize all quality related quantities to = . (b) Spore-quality density ( ), defined as the number density of generated spores with quality q resulting from the sporulation model with = 3 non-naïve vegetative states, at fast and slow sporulation rates (magenta and cyan, respectively). Quality q is expressed relative to and thus  The probability for a spore to revive depends on and thus on spore quality q. We introduce * = * / , which denotes an apparent spore quality threshold that is required for reviving with 50% probability by utilizing alanine only. The "richness" of the medium supplied during nutrient upshift modulates the strength of the alternative flux by providing other substrates. In effect = ( )/ is the apparent gain in quality that a rich medium supplies to the spore, compared to alanine only. (b) Number density ( ) ( ) of revived spores with quality q, for fast (magenta) and slow (cyan) sporulation rates assuming * = 1. Thus, only spores with an overall apparent quality close to = 1 revive with high probability. Top panel: In rich medium ( = 1), the quality-independent flux dominates, and thus the density of revived spores mirrors the spore-quality density for both fast (magenta) and slow (cyan) sporulation. Bottom panel: In poor medium ( = 0), outgrowth of high-quality spores is strongly favored. These are present for fast sporulation (magenta) but are missing in slow sporulation (cyan). All other parameters as in Fig. S1b.     and PS216 (cyan) were transformed with the spore history reporter (PrapA-mCherry). The rapA promoter is activated in cells that delay sporulation, therefore late-generated spores have a higher fluorescence.
Spores were generated in a shaken culture with SM media. For each strain, the median in spore fluorescence was used to group spores into early-generated (low fluorescent) spores and late-generated (high fluorescent) spores, respectively, in order to test whether sporulation timing affects their revival capacity. (b) Box-plot of spore revival frequency for early and late spores as defined in a. Spores were dispersed on a SM-pad and stimulated with L-alanine and the revival frequency was measured by timelapse microscopy. Each data-point corresponds to data from a single movie with on average more  independently of spore age (time that spores have spent in the sporulation medium). The median spore fluorescence was used to classify spores into early and late spores. Spore samples were heat-treated and dispersed on SM pads and stimulated with L-alanine. The revival frequencies for both groups (early and late) were measured by time-lapse microscopy. Movies in which less than 2% of the total spore population grew out were discarded from the analysis. Each datapoint corresponds to data from a single movie with > 60 spores in total. To account for technical variations in measurements of the overall revival frequency, the revival frequencies of early and late spores are reported relative to the overall population mean in each movie. Statistics: N > 8 movies. For both strains, early spores exhibited a significantly higher relative revival frequency than late spores, independently of spore age. Statistical test: Two-sided Wilcoxon signed rank test. * p < 0.05, ** p < 0.01, *** p < 0.001. Figure S10. Ald may act as a spore quality determinant in natural isolates. Results are shown for natural isolates BSP1 (left) and PS216 (right) carrying an ectopically integrated fluorescence reporter of Ald protein levels (Pald-ald-mCherry). Spores were generated in a well-mixed liquid culture. The insets show the corresponding histograms of spore fluorescence for each strain, which indicate that spores from both strains contain Ald. The median spore fluorescence was used to group spores into "dim" and "bright" classes, which correspond to spores with low and high Ald levels, respectively. Boxplots of the revival frequencies of dim and bright spores are depicted in the main panels. Spores were dispersed on a SM pad and stimulated with L-alanine, and the revival frequency was measured by timelapse microscopy. Each data-point corresponds to a single movie with > 70 spores in total. Statistics: N > 7 movies. For both strains, bright spores had a significantly higher relative revival frequency than dim spores. Statistical test: Two-sided Wilcoxon signed rank test. **p < 0.01.