Crucial role of ppGpp in the resilience of Escherichia coli to growth disruption

Bacteria grow in constantly changing environments that can suddenly become completely deleted in essential nutrients. The stringent response, a rewiring of the cellular metabolism mediated by the alarmone (p)ppGpp, plays a crucial role in adjusting bacterial growth to the severity of the nutritional insult. The ability of (p)ppGpp to trigger a slowdown of cell growth or induce bacterial dormancy has been widely investigated. However, little is known about the role of (p)ppGpp in promoting growth recovery after severe growth inhibition. In this study, we performed a time-resolved analysis of (p)ppGpp metabolism in Escherichia coli as it recovered from a sudden slowdown in growth. Results show that E. coli recovers by itself from the growth disruption provoked by the addition of serine hydroxamate, the serine analogue that we used to induce the stringent response. Growth inhibition was accompanied by a severe disturbance of metabolic activity and more surprisingly, by a transient overflow of valine and alanine. Our data also show that ppGpp is crucial for growth recovery since in the absence of ppGpp, E. coli’s growth recovery was slower. In contrast, an increased concentration of pppGpp was found to have no significant effect on growth recovery. Interestingly, the observed decrease in intracellular ppGpp levels in the recovery phase correlated with bacterial growth and the main effect involved was identified as growth dilution rather than active degradative process. This report thus significantly expands our knowledge of (p)ppGpp metabolism in E. coli physiology. IMPORTANCE The capacity of microbes to resist and overcome environmental insults, know as resilience, allows them to survive in changing environments but also to resist antibiotic and biocide treatments, immune system responses. Although the role of the stringent response in bacterial resilience to nutritional insults has been well studied, little is known about its importance in the ability of the bacteria to not just resist but also recover from these disturbances. To address this important question, we investigated growth disruption resilience in the model bacterium Escherichia coli and its dependency on the stringent response alarmone (p)ppGpp by quantifying ppGpp and pppGpp levels as growth was disrupted and then recovered. Our findings may thus contribute to understanding how ppGpp improves E. coli’s resilience to nutritional stress and other environmental insults.

severe disturbance of metabolic activity and more surprisingly, by a transient overflow of 23 valine and alanine. Our data also show that ppGpp is crucial for growth recovery since in the 24 absence of ppGpp, E. coli's growth recovery was slower. In contrast, an increased 25 concentration of pppGpp was found to have no significant effect on growth recovery. 26 Interestingly, the observed decrease in intracellular ppGpp levels in the recovery phase 27 correlated with bacterial growth and the main effect involved was identified as growth 28 dilution rather than active degradative process. This report thus significantly expands our 29 knowledge of (p)ppGpp metabolism in E. coli physiology. 30

IMPORTANCE 31
The capacity of microbes to resist and overcome environmental insults, know as resilience, 32 allows them to survive in changing environments but also to resist antibiotic and biocide 33 treatments, immune system responses. Although the role of the stringent response in 34 bacterial resilience to nutritional insults has been well studied, little is known about its 35 importance in the ability of the bacteria to not just resist but also recover from these 36 disturbances. To address this important question, we investigated growth disruption 37 resilience in the model bacterium Escherichia coli and its dependency on the stringent 38 response alarmone (p)ppGpp by quantifying ppGpp and pppGpp levels as growth was 39 disrupted and then recovered. Our findings may thus contribute to understanding how ppGpp 40 improves E. coli's resilience to nutritional stress and other environmental insults. 41

INTRODUCTION 42
As single-cell organisms, bacteria face constant changes in their direct physico-chemical and 43 nutritional environments. To overcome these disturbances, bacteria have developed adaptive 44 properties that allow them to survive, grow, and eventually evolve. Depletion of external 45 nutrients is one of the most serious insults for these organisms because they have very little 46 internal storage and the fact that their ability to rapidly modulate metabolic functions is key 47 7 nmol·gCDW −1 , lower than the value measured for the WT strain (123 ± 85 nmol·gCDW −1 ) (Fig.  161   S4B) and close to the detection limit. The presence of ppGpp in this ∆relA mutant indicates 162 that under these conditions, SpoT synthesizes low levels of ppGpp in the exponential regime. 163 As expected, we did not detect any transient accumulation of ppGpp after SHX addition. This 164 confirms that the synthetase activity of SpoT is mainly silent in this situation and that, in 165 agreement with previous studies (22,26), no other RSH is involved in this response in E. coli. 166 More importantly, this means that SHX addition induces growth inhibition by itself, without 167 (p)ppGpp. Finally, although SHX disappeared completely from the medium in less than 3 h, as 168 also observed for the WT strain (Fig. 3C), the cells had failed to fully recover their initial growth 169 rate 6 h after SHX addition (Fig. 3A). The recovery rate of the ∆relA mutant was a factor of 2 170 lower (0.0847 ± 0.0261 h −2 ) than the WT's (Table 1). These results highlight the crucial role of 171 the stringent response in E. coli's ability to overcome the growth disruption caused by SHX. 172 Based on the analysis of instantaneous growth rates, WT and the ∆relA strains had similar 173 robustness. Furthermore, alanine and valine were also found to accumulate in the culture 174 medium with the ∆relA mutant, and the accumulation of alanine was even more pronounced 175 with the mutant than it was with the WT strain (Table S1), indicating that this phenomenon is 176 not related to the stringent response. 177 pppGpp over-accumulation has no effect on growth recovery 178 The physiological role of pppGpp in the stringent response in E. coli has remained rather 179 unclear to date. To explore its effect on growth recovery, we applied our methodology to a 180 mutant deleted for the gppA gene, which encodes for the enzyme, pppGpp 5'-gamma 181 phosphohydrolase, which converts pppGpp to ppGpp. The ∆gppA mutant is known to 182 accumulate high concentrations of pppGpp after SHX addition (23). 183 As expected therefore, the intracellular levels of pppGpp in the ∆gppA mutant were higher 184 during the exponential phase than in the WT strain while intracellular ppGpp levels were 185 similar (Fig. S4B). In the ∆gppA mutant, the concentrations of pppGpp and ppGpp were thus 186 similar, as reported previously (23), with an estimated ppGpp/ppGpp ratio of 0.62 ± 0.38. 187 Although the pppGpp concentration was about one order of magnitude higher than in the WT 188 strain, the growth rates were almost identical (Fig. 3A), suggesting that pppGpp does not 189 affect the growth rate. 190 8 As in the WT strain, adding SHX triggered the accumulation of ppGpp and pppGpp. However, 191 while the ppGpp concentration varied around the same levels as measured for the WT strain, 192 the pppGpp concentration was roughly twice as high as in the WT (Fig. 3B, D). The total 193 concentration of ppGpp and pppGpp was therefore significantly higher, with the 194 pentaphosphate form predominating, contrary to what is observed in other conditions. In this 195 strain, the principal product of RelA is therefore pppGpp. The pppGpp/ppGpp ratio decreased 196 over time (Fig. 3D), tending toward the value measured before SHX addition. Although this 197 ratio is markedly different in the ∆gppA mutant, the growth recovery profile of the mutant 198 was similar to that of the WT strain (with an estimated recovery rate of 0.1941 ± 0.0482 h −2 199 for the ∆gppA mutant; Fig. 3A and Table 1). This means that the build-up of pppGpp has no 200 significant effect on growth recovery, the only difference in this strain being a slightly greater 201 robustness ( Table 1). Note that alanine and valine accumulated in the culture medium once 202 again, as observed for the WT strain (Table S1). 203 The decrease in (p)ppGpp concentration can be explained by growth 204 As shown above for the WT and ∆gppA strains, the addition of SHX leads to a rapid 205 accumulation of ppGpp and pppGpp (in a few minutes), a plateau stage that last for less than 206 one hour, and then a slow decrease in the concentrations of (p)ppGpp. It takes about 3 h in 207 the latter phase for the intracellular concentrations of ppGpp and pppGpp to drop to the levels 208 measured in the exponential phase (Fig. 2B). The question then arises whether the decrease 209 in the concentration of (p)ppGpp is the result of an active degradation process or simply due 210 to growth-driven dilution as suggested by the ppGpp and pppGpp levels being and shows that in this strain, the decrease in ppGpp and pppGpp levels is mainly due to growth 221 9 dilution rather than an active degradation process. In contrast, the measured intracellular 222 levels of ppGpp and pppGpp do not follow this line for the WT strain (Figs 4C and 4D), meaning 223 that a degradation process is involved. To estimate its contribution, we calculated what the 224 intracellular ppGpp and pppGpp concentrations would be if the degradation flux were 1 to 8 225 times the growth dilution rate (grey lines in Fig. 4). For pppGpp (Fig. 4D), most of the 226 experimental points are located between the first and the second grey lines, indicating that 227 the degradation flux-likely via GppA-is about twice the growth dilution rate. The flux of ppGpp 228 degradation is even more modest since the experimental points fall mostly between the black 229 line and the first grey line (Fig. 4C). Altogether, these results indicate that the decrease in the 230 concentration of (p)ppGpp is mostly accounted for by growth, even if GppA appears to 231 participate somewhat in the degradation of pppGpp. The surprising implication of these 232 results is that SpoT plays only a minor role in hydrolyzing ppGpp and pppGpp under these 233 conditions. 234

DISCUSSION 235
In this study, we investigated the dynamic response of E. coli to severe growth disruption and 236 the role of the stringent response in this bacterium's ability to recover growth. To this aim, we 237 monitored the growth and quantified consumed and excreted metabolites and intracellular 238 levels of ppGpp and pppGpp in E. coli cultures before and after adding SHX. 239 The results demonstrate first that the K-12 WT strain of E. coli is resilient to SHX-induced 240 growth disruption since its growth rate returned to pre-SHX levels a few hours after the 241 perturbation. This recovery was first shown in the pioneering work of Tosa and Pizer (28), 242 where growth inhibition was released by adding serine. In our work, E. coli appeared to 243 recover growth by itself. Intriguingly, we observed that SHX disappeared rapidly from the 244 medium and identified the cause as being a cell-related process, but it remains unclear 245 whether SHX was degraded or simply internalized into the cells. The results of the experiment 246 with the ∆relA mutant of K-12 E. coli indicate that resumption of growth is conditioned on the 247 stringent response. Although just as with the WT, SHX disappeared from the medium, the 248 ∆relA mutant failed to fully recover its pre-SHX growth rate, indicating that the stringent 249 response is major determinant of E. coli 's resilience to growth disruption. We also observed 250 that this resilience is not affected by an over-accumulation of pppGpp. By eliminating GppA, 251 10 we inverted the pppGpp/ppGpp concentration ratio but the ∆gppA mutant's recovery from 252 SHX addition was nevertheless similar to that of the WT. Although the robustness of the 253 ∆gppA mutant was slightly higher, these results indicate that pppGpp does not play a 254 significant role in growth recovery. This is in keeping Mechold et al.'s conclusion that pppGpp 255 is a less potent growth regulator than ppGpp (22). 256 The concentrations of pppGpp and ppGpp both peaked rapidly after the addition of SHX (in 257 less than 10 min), which is in line with an earlier study of a different bacterium (38). As 258 mentioned above, in the WT strain, the concentrations of ppGpp were higher than those of 259 pppGpp. This is in agreement with a previous report (23)

Bacterial strains and growth conditions 293
All strains were derived from E. coli strain K-12 MG1655. The ∆relA and ∆gppA strains were 294 constructed by P1 transduction of gene deletions marked with a kanamycin resistance 295 cassette from the Keio collection (42). The kanamycin resistance cassette was removed using 296 FLP recombinase from pCP20 plasmid (43). All strains, plasmids and primers are listed in Table  297 S2 and the genetic modifications were checked by PCR. Brunswick, NJ, USA). Cells were harvested during the exponential growth phase by 307 centrifugation for 10 min at 10,000 g at room temperature with a Sigma 3-18K centrifuge 308 (Sigma Aldrich, Seelze, Germany), washed with the same volume of fresh medium (without 309 glucose or thiamine), and used to inoculate 500 mL bioreactors (Multifors, Infors HT, 310 12 at OD600nm = 0.15. The temperature was set to 37°C and the pH was maintained at 7 by 312 automatically adding 14% (g/g) ammonia or 11 % (g/g) phosphoric acid. Aeration and the 313 stirrer speed were controlled to maintain adequate aeration (DOT > 30% saturation). Cell 314 growth was monitored by measuring the optical density at 600 nm with a Genesys 6 315 spectrophotometer (Thermo, Carlsbad, CA, USA) The percentages of O2, CO2 and N2 316 concentrations were measured in the gas output during the culture process using a Dycor 317 ProLine Process mass spectrometer (Ametek, Berwyn, PA, USA), and the data obtained was 318 used to calculate the oxygen uptake rate (OUR) and carbon dioxide emission rate (CER). 319 Stringent response was triggered by adding SHX at 0.8 mM to the culture when the OD 320 reached 3.5. 321

Calculation of the instantaneous growth rate 322
The instantaneous growth rate (µ(t)) was determined by fitting the time evolution of the 323 biomass concentration i) to an exponential function prior to SHX addition and ii) to a 324 parametric function after SHX addition, from which µ(t) was calculated as µ(t) = dX/(X·dt). 325

Sampling and (p)ppGpp extraction 326
Culture medium (400 µL) was withdrawn from the bioreactor and vigorously mixed with 4.5 327 mL of a pre-cooled acetonitrile/methanol/H2O (4:4:2) solution at -40°C to rapidly quench 328 metabolic activity (36). Immediately thereafter, 100 µL of 13 C labeled metabolites were added 329 to the latter mixture as internal standards. The tubes were then placed in a cooling bath of  The black lines were calculated with JF=JD, the grey lines with (JF-JD)=a.(µ(t).(p)ppGpp) with a 402 ranging from 1 to 8, using as initial conditions the mean values of the (p)ppGpp concentrations 403 and µ(t) calculated from the three biological replicates of each strain. µ(t) was calculated from 404 the recovery rates determined for each strain, as listed in Table 1  Results are presented as mean ± standard deviation (n = 3). CDW, cell dry weight. 413 a The robustness and recovered steady-state are expressed relative to the initial growth rate 414 b Not determined (only one of the three biological replicates had recovered its initial growth 415 rate 5.5 h after SHX addition).  Fig. 1 are shown in blue, repeat #1, in green, and repeat #2, in red; except for the OUR (Fig. 1 data, pink; repeat #1, green; repeat #2, red) and the CER (Fig. 1 data, blue; repeat #1, yellow; repeat #2, dark red).   T-test p-values: * p < 0.05, **p < 0.02, ***p < 0.005.