The growth-survival trade-off is hard-wired in the Lactococcus lactis gene regulation network.

Most microbes reside in oligotrophic environments for extended periods of time, requiring survival strategies that maintain proliferative capacity. We demonstrate that the non-spore-forming Lactococcus lactis KF147 progressively activates the expression of stress-associated functions in response to the declining growth rate elicited by prolonged retentostat cultivation, which coincides with up to 104 -fold increased stress tolerance. Our findings provide a quantified view of the transcription and stress-tolerance adaptations underlying the growth-survival trade-off in L. lactis, and exemplify the hard-wiring of this trade-off in the lactococcal gene regulation network.

Most aquatic and terrestrial ecosystems on earth are oligotrophic. Resident microbes have evolved a variety of strategies to maintain proliferative capacity under nutrient‐poor conditions for long periods (Egli, 2010; Hoehler and Jorgensen, 2013; Hallsworth, 2021). They adapt to such conditions by minimizing their energy expenditure and metabolic activity, ultimately resulting in a slow‐ or non‐growing state (Lennon and Jones, 2011; Boutte and Crosson, 2013; Ercan et al., 2013; Hoehler and Jorgensen, 2013; Kleerebezem et al., 2020). Distinct non‐growing phenotypes have been described, including spores, akinetes, biofilms and viable but non‐culturable states (Harrison et al., 2007; Navarro Llorens et al., 2010; Lennon and Jones, 2011; Boutte and Crosson, 2013; van Mastrigt et al., 2018; van Tatenhove‐Pel et al., 2019). Despite this phenotypic variation slow‐ or non‐growing phenotypes consistently show increased resistance to stress conditions (Dressaire et al., 2008; Lopez‐Maury et al., 2008; Lu et al., 2009; Zakrzewska et al., 2011; Boutte and Crosson, 2013; Maharjan et al., 2013; Ercan et al., 2015b; Ferenci, 2016; Biselli et al., 2020; Abram et al., 2021), reflecting the growth‐survival trade‐off. Although the existence of such trade‐off has been described, its quantitative assessment and molecular underpinning remain incomplete. Retentostat cultivation enables the study of microbes at progressively reduced growth rates that ultimately approximate zero‐growth under controlled conditions, which coincides with the reallocation of cellular energy from growth to maintenance‐related processes (Ercan et al., 2015b).

A slow‐growing, glucose‐limited anaerobic chemostat culture of the plant‐isolate Lactococcus lactis KF147 (specific growth rate 0.025 h−1) was switched to retentostat cultivation mode by effluent removal through a cross‐flow filter while maintaining medium dilution at 0.025 h−1, whereby microbial biomass is retained in the fermentor (Ercan et al., 2013; Ercan et al., 2015b). Previous retentostat studies showed that L. lactis KF147 growth gradually approximates a near‐zero specific growth rate after 2–3 weeks, while maintaining high levels of cell viability and cultivability (Ercan et al., 2013) (Supplementary Fig. S1). In parallel, cellular energy expenditure gradually switches from growth towards maintenance‐related processes (Ercan et al., 2013), coinciding with fluctuations in pyruvate dissipation between homolactic and mixed‐acid fermentation and the activation of import and utilization pathways for alternative carbon sources (Ercan et al., 2015a). Strikingly, similar studies that used a dairy isolate of L. lactis (strain FM03‐V1) demonstrated that this strain had an approximately sixfold higher energy requirement during near‐zero growth rates (van Mastrigt et al., 2018; van Mastrigt et al., 2019; Kleerebezem et al., 2020). These observations could reflect that the plant isolate KF147 is more adapted to the energy‐poor environment associated with plants, contrasting the adaptation of the dairy isolate FM03‐V1 to long‐term cultivation in the energy‐rich dairy environment (Kleerebezem et al., 2020).

The physiological changes observed in L. lactis KF147 were in agreement with the progressive transcriptome changes observed at consecutive timepoints during retentostat cultivation (Ercan et al., 2015a). Further mining of the transcriptome data revealed that adaptation of L. lactis during retentostat cultivation includes a gradual alleviation of repression of stress‐response associated functions (Fig. 1). These functions were previously assigned to the so‐called ‘stressome’ of L. lactis and other lactic acid bacteria (Papadimitriou et al., 2016) and include canonical class I and class III heat shock proteins as well as several functions involved in cell membrane‐, acid‐ and low‐temperature stress responses, which were described to contribute to tolerance to various stress conditions such as heat‐, acid‐ and oxidative‐stress (Papadimitriou et al., 2016). Quantitative analysis showed that the specific growth rate and stressome gene expression levels displayed a strong negative correlation. The increase of expression of stressome genes was most prominent during the first 4 weeks of retentostat cultivation, closely approximating (~90%) final expression of these genes after 6 weeks of retentostat cultivation (>15‐fold induction relative to chemostat conditions), and coincided with the strongest reduction (>95%) of the relative growth rate (Fig. 1). This finding establishes that a broad panel of stressome associated genes is consistently and progressively induced, and increased expression is quantitatively correlated with reduction of the specific growth rate, providing a molecular mechanism that plausibly underlies the growth‐survival trade‐off in this species.

Fig. 1

Expression of stress‐associated genes of L. lactis KF147 and correlation with specific growth rate during retentostat cultivation. Average gene‐specific expression values (Ercan et al., 2015a) of L. Lactis KF147 stressome‐associated genes during 2, 7, 14, 21, 28, 35 and 42 days of two independent retentostat cultivations are displayed relative (in log2‐scale, corrected p‐value ≤0.05) to their expression levels during chemostat growth (day 0) (panel A). The broad spectrum of stressome expression activation is apparent from the genes associated with cell membrane (A1), heat (A2), acid (A3) and low temperature (A4) stress responses. Panel B displays the relationship of the averaged differential expression of stressome genes in comparison to chemostat conditions (average‐fold‐change; log2‐scale) and the relative growth rate during retentostat cultivation [v(i)] compared to chemostat cultivation [v(0); specific growth rate 0.025 h−1] (Ercan et al., 2013) in log2‐scale. Data points represent average ± standard deviation of measurements of two independent retentostat cultures.

To evaluate whether increased stressome transcription indeed increases the stress resistance of these energy‐restricted L. lactis KF147 cultures, stress tolerance was analysed in bacteria sampled after 14, 21 and 29 days of retentostat cultivation and compared to the tolerance levels measured for the initial chemostat culture (see supplementary methods and supplementary Fig. S2 for details). These sampling time‐points for the retentostat culture were chosen to encompass the period of the most prominent change in stressome gene expression levels (Fig. 1A). Exposure to heat (50°C) and acid (pH 2.5) stress revealed a significantly increased survival capacity of retentostat‐ compared to chemostat‐derived cells. Moreover, prolongation of retentostat cultivation led to consistent improvement of stress tolerance reaching maximum stress resistance levels in samples obtained after 29 days of retentostat cultivation (Fig. 2). Two hours of heat or acid stress exposure led to an almost 5‐log reduction in cultivability in chemostat‐derived bacteria (equivalent to approximately 0.001% survival), whereas the remaining cultivability was more than 102, 103 and 104‐fold higher in the bacteria taken from the retentostat cultivation after 14, 21 and 29 days respectively (Fig. 2, panels A and B). Exposure to oxidative stress (20 mM H2O2) was characterized by a rapid loss of cultivability (more than 90% within the first 15 min) of the cells derived from the chemostat cultivation. This level of oxidative stress was also detrimental for the culture samples obtained from the retentostat, but after 15 min of exposure to H2O2 approximately 15%, 25% and 50% of the cells remained culturable in samples obtained after 14, 21 and 29 days retentostat cultivation respectively (Fig. 2, panel C). Prolongation of oxidative stress exposure roughly sustained the magnitude of this initial difference.

Fig. 2

Kinetics of loss of cultivability of chemostat‐ and retentostat‐grown L. lactis KF147 cells during exposure to heat stress at 50°C (panel A), acid stress at a pH of 2.5 (panel B) and oxidative stress at 20 mM H2O2 (panel C). Remaining cultivability is displayed as the percentage (plotted in a log10 scale) of remaining cultivability relative to the original numbers of colonies obtained in the untreated samples (i.e. prior to stress exposure). The different culture samples included are chemostat‐grown cells (blue); retentostat‐grown cells on day 14 (red), 21 (green) and 29 (purple). Data points represent average ± standard deviation of measurements of two independent retentostat cultures. Notably, CFU enumerations did not decline for untreated and non‐stressed cells (i.e. the initial 100% numbers of CFU) over a time span of 2 h. Panels D and E show the relationship between the average of the loss of cultivability kinetics [expressed as the average first decimal reduction time (δ) in log2‐scale] and the averaged differential expression of stressome genes (see Fig. 1 for details) (panel D), or the relative growth rate during retentostat cultivation (see Fig. 1 for details) in log2‐scale (panel E). The curve in panel E represents the best‐fit model and data points represent average ± standard deviation of the measurements derived from two independent retentostat cultures.

Stress tolerance of the different L. lactis KF147 culture samples was quantified using the Weibull microbial survival model (den Besten et al., 2006), which showed acceptable model‐fitting performance for most of the experimental conditions (Supplementary Table ST1). Stress tolerance is represented by the δ parameter in the Weibull model, which represents the first decimal reduction time (see Supplementary Methods). The δ values obtained for heat and acid stress exposure were increasing with prolongation of the retentostat cultivation, reaching a 2.5–3‐fold higher δ after 29 days of retentostat cultivation relative to the value obtained from a chemostat‐derived culture (Supplementary Table ST2). Analogously, model fitting of oxidative stress survival confirmed a higher δ value for the retentostat‐derived samples than those from the chemostat cultivation (Supplementary Table ST2). Importantly, a strong correlation between the determined stress tolerance levels (as reflected by the δ value) and the stressome gene expression levels was observed, illustrating that the selected transcripts are adequate cellular indicators for bacterial stress tolerance (den Besten et al., 2010; Abee et al., 2011), even in these energy‐restricted cells at near‐zero growth rate (Fig. 2, panel D). Remarkably, a non‐linear relationship was revealed between stress tolerance and growth rate, indicative of an exponential stress tolerance increase upon the adaptation towards near‐zero growth rates (Fig. 2, panel E). Previous studies have argued that the shape of a trade‐off curve between growth rate and stress survival determines the response of an ecosystem to competition (Maharjan et al., 2013; Abram et al., 2021). The non‐linear shape of the growth‐survival trade‐off curve in L. lactis implies that L. lactis is evolutionarily specialized for survival at the cost of rapid growth under nutrient‐poor conditions (Maharjan et al., 2013; Abram et al., 2021).

This quantitative study exemplifies that during the approximation of a zero‐growth state, L. lactis employs intrinsic gene‐regulatory networks for progressively increased stressome expression even though cells are not exposed to environmental stress conditions. This molecular adaptation protects the bacterial cells and ensures their sustained fitness and cultivability under conditions that do not support growth. The quantitative relationships between growth rate and the level of expression of stress‐associated functions correspond with stress tolerance levels, exemplifying the hard‐wiring of the growth‐survival trade‐off within the gene regulation network of L. lactis. In Escherichia coli, a role has been suggested for small regulatory RNA molecules (sRNAs) in the adjustment of gene regulation repertoires under nutrient‐poor conditions and during biofilm formation (Andreassen et al., 2018; Abram et al., 2021). Intriguingly, in L. lactis a role of sRNAs has been described for the regulation of nutrient acquisition (van der Meulen et al., 2016), energy generation (van der Meulen et al., 2019) as well as stress response (van der Meulen et al., 2017; Wu et al., 2018; Tian et al., 2019), indicating that these sRNAs might play a role in several of the gene expression adjustments observed in L. lactis KF147 during prolonged retentostat cultivation.

Taken together, our study contributes to the understanding of the transcription and stress tolerance adaptation that underlie the growth‐survival trade‐off in L. lactis, using well‐controlled retentostat cultivation to mimic environmentally prevalent oligotrophic conditions that induce non‐growing states.

Appendix S1. Supporting Information.

Click here for additional data file.

Appendix S2. Supporting Information.

Click here for additional data file.