Determining the importance of the stringent response for methicillin-resistant Staphylococcus aureus virulence using a zebrafish model of infection

Staphylococcus aureus is a bacterial pathogen that poses a major threat to human health. The ability of this bacterium to adapt to stresses encountered in the host is essential for disease. The stringent response is a signalling pathway utilised by all bacteria to alarm cells when stressed, and has been linked to the virulence of a number of species. This signalling pathway is controlled by the nucleotide alarmones guanosine tetra-(ppGpp) and pentaphosphate (pppGpp: collectively termed (p)ppGpp), produced in S. aureus by three synthetase enzymes: Rel, RelP and RelQ. Here, we used a triple (p)ppGpp synthetase mutant ((p)ppGpp0) to examine the importance of this signalling network for the survival and virulence of S. aureus in vivo. Using an established zebrafish larval infection model, we observed that infection with (p)ppGpp0 resulted in attenuated virulence, which was not due to a reduced ability of the mutant to replicate in vivo. Of the three (p)ppGpp synthetases, Rel was established as key during infection, but roles for RelP and RelQ were also observed. Zebrafish myeloid cell depletion restored the virulence of (p)ppGpp0 during systemic infection, indicating that (p)ppGpp is important for survival within host phagocytes. Primary macrophages infection studies, followed by in vitro tolerance assays to key innate immune effectors, demonstrated that (p)ppGpp0 was more susceptible to stressors found within the intracellular macrophage environment, with roles for all three synthetases implicated. Lastly, the absence of CodY, a transcription factor linked to the stringent response, significantly increased the tolerance of S. aureus to phagolysosomal-like stressors in vitro, but had no impact in vivo. Taken together, these results define the importance of the stringent response for S. aureus infection, revealing that (p)ppGpp produced by all three synthetases is required for bacterial survival within the host environment by mediating adaptation to the phagolysosome.

and has been linked to the virulence of a number of species. This signalling pathway is 23 controlled by the nucleotide alarmones guanosine tetra-(ppGpp) and pentaphosphate (pppGpp: 24 collectively termed (p)ppGpp), produced in S. aureus by three synthetase enzymes: Rel, RelP 25 and RelQ. Here, we used a triple (p)ppGpp synthetase mutant ((p)ppGpp 0 ) to examine the 26 importance of this signalling network for the survival and virulence of S. aureus in vivo. Using 27 an established zebrafish larval infection model, we observed that infection with (p)ppGpp 0 28 resulted in attenuated virulence, which was not due to a reduced ability of the mutant to 29 replicate in vivo. Of the three (p)ppGpp synthetases, Rel was established as key during 30 infection, but roles for RelP and RelQ were also observed. Zebrafish myeloid cell depletion 31 restored the virulence of (p)ppGpp 0 during systemic infection, indicating that (p)ppGpp is 32 important for survival within host phagocytes. Primary macrophages infection studies, 33 followed by in vitro tolerance assays to key innate immune effectors, demonstrated that 34

Introduction 42
Staphylococcus aureus is a highly adaptable pathogen, with a large arsenal of virulence factors 43 that allow it to colonise diverse sites within the human host. Upon infection, bacteria are 44 subjected to harsh conditions due to changes in nutrient availability, pH and temperature, as 45 well as the presence of an immune response. When faced with stresses, bacteria induce a 46 conserved survival pathway termed the stringent response, which is coordinated by the 47 nucleotide alarmones guanosine tetraphosphate (ppGpp) and guanosine pentaphosphate 48 (pppGpp: collectively known as (p)ppGpp) [1]. (p)ppGpp is produced by the RelA/SpoT 49 Homologue (RSH) protein family [2], with S. aureus encoding three of these synthetases: the 50 long RSH enzyme Rel, which also has hydrolase activity, and the two monofunctional small 51 alarmone synthetases (SAS) RelP and RelQ (Fig. 1A) [3]. Production of (p)ppGpp by these 52 enzymes results in major changes within the bacterial cell, with alterations to numerous 53 macromolecular activities such as transcription and translation [4][5][6]. These alterations lead to 54 an inhibition of growth and a concurrent upregulation of stress adaptation and virulence factors, 55 which ultimately allows a switch from active growth to a more stationary phenotype to aid 56 bacterial survival. In S. aureus, the three (p)ppGpp synthetases respond to different 57 environmental stresses to increase (p)ppGpp levels, with Rel sensing amino acid starvation via 58 interactions with uncharged tRNA and the ribosome, while RelP and RelQ respond to cell wall 59 and pH stress [4]. 60 Activation of the stringent response is reported to contribute to the pathogenicity of a 61 number of bacterial species. For example, a Salmonella enterica subspecies Typhimurium 62 (p)ppGpp-null ((p)ppGpp 0 ) mutant was unable to replicate in the mouse spleen after five days 63 [7]. Similarly, the numbers of a Mycobacterium tuberculosis rel mutant recovered from murine 64 lung and spleen tissues was 2-log lower than the wildtype over a 38-week period, implicating 65 a requirement for Rel for long-term viability and chronic infection [8]. The absence of 66 (p)ppGpp also affected the ability of Enterococcus faecalis to form biofilms in murine models 67 of catheter-associated urinary tract infections [9]. For S. aureus, a methicillin-resistant S. 68 aureus (MRSA) rel mutant formed cutaneous abscess lesions in mice that were over 13-times 69 smaller than those formed by the wildtype [10]. Rel was also required for maintaining 70 methicillin-sensitive S. aureus (MSSA) load in murine renal abscesses and for reducing mouse 71 body weight [11]. This loss of body weight was dependent on the transcription factor CodY, 72 which derepresses amino acid and virulence genes upon induction of the stringent response 73 [12]. Together, these reports indicate that the stringent response is important for the virulence 74 of a number of bacterial pathogens. 75 Zebrafish (Danio rerio) are a well-established animal model for various types of disease 76 and infection. Despite being non-mammalian, zebrafish are jawed vertebrates with high genetic 77 homology to humans -more than 80% of disease-associated genes have a human counterpart 78 [13]. Furthermore, advantages over mammalian models include its rapid embryonic 79 development, genetic manipulability and its transparency at the embryonic and larval stages, 80 which allows for live imaging of developing zebrafish [14]. With a functional innate immune 81 system by 30 hours post fertilisation (hpf) [15,16], zebrafish larvae are also useful for studying 82 host-pathogen interactions, as demonstrated by the numerous infection models that exist [17]. 83 For example, we have previously developed systemic infection models to study the 84 pathogenicity of S. aureus and E. faecalis within zebrafish [18,19]. Studies such as these 85 highlight how the zebrafish model can be used to further our knowledge of host-pathogen 86 interactions and disease pathogenesis. 87 While reports indicate that the Rel synthetase is important for the survival of S. aureus 88 in polymorphonuclear leukocytes (PMNs) [20], as well as in cutaneous abscess lesions and 89 murine renal abscess models [10,11], the importance of the entire signalling system, and the 90 contribution of the two SAS enzymes, RelP and RelQ, to virulence is much less well 91 understood. Here, we sought to use the versatility of the zebrafish model to establish the 92 importance of (p)ppGpp for S. aureus systemic infection, extending previous findings relating 93 to bacterial load in host organs to examine the contribution of the stringent response to host 94 killing. Using the zebrafish model, we have determined that all three synthetases contribute to 95 the virulence of S. aureus. The attenuated phenotype of a (p)ppGpp 0 mutant was determined to 96 be at least partly myeloid cell-dependent, as morpholino-mediated ablation of myeloid cells 97 restored virulence. Moreover, we show a requirement for (p)ppGpp for survival of S. aureus 98 within primary macrophages, with in vitro studies highlighting its importance for tolerance of 99 phagolysosomal stressors. While the overproduction of (p)ppGpp has been suggested to 100 support chronic and recurrent infections, here we observed that increased (p)ppGpp production 101 led to higher tolerance of S. aureus to stressors in vitro but actually reduced bacterial virulence. 102 Finally, deletion of the CodY transcription factor restored the survival defect of the (p)ppGpp 0 103 mutant in vitro, but did not restore virulence. Altogether, this work further defines the 104 importance of (p)ppGpp, and each of the three synthetases, for S. aureus growth, systemic 105 infection and host death. 106 107 were embedded in 3% w/v methylcellulose on a glass slide. 1 nl of bacterial suspensions were 133 injected into the yolk sac circulation valley of the ≥ 30 embryos per condition using a pneumatic 134 micropump (World Precision Instruments PV820), a micromanipulator (WPI) and a dissecting 135 microscope. Following injection, embryos were recovered in fresh E3 and placed into individual 136 wells of a 96-well plate. After 2 dpf, embryos are referred to as larvae. The larvae were 137 monitored twice a day up to 93 hours post infection (hpi) and the number of dead larvae at each 138 timepoint recorded. To confirm bacterial numbers in each injection, the same volume was 139 ejected into 1 ml of PBS and the viable counts determined on tryptic soy agar (TSA) plates. 140 Survival curves were generated using GraphPad Prism. from human blood (day 0). Briefly, whole blood was separated by density centrifugation using 167 Ficoll Paque Plus and the buffy layer (containing PBMCs) was extracted for further processing. Platelets were removed by low-speed centrifugation and Ammonium-Chloride-Potassium 169 (ACK -Thermo Fisher) lysis buffer was used to lyse red blood cells. Isolated PBMCs were 170 resuspended in RPMI 1640 medium containing 10% new-born calf serum, 1% L-Glutamine 171 and 1% antibiotic-antimycotic solution. A cell count was performed and PBMCs were seeded 172 into tissue culture plates at 2 x10 6 cells/ml -this seeding density is estimated to provide 2 x 10 5 173 cells/ml MDMs. After 24-48 hrs, the seeding medium was removed and replaced with RPMI 174 1640 containing 10% foetal bovine serum, 1% L-Glutamine and 1% antibiotic-antimycotic 175 solution. This media was replaced every 3-4 days to promote the differentiation of MDMs. 176 MDMs were used in experiments between 12-and 14-days post isolation and the supplemented 177 RPMI 1640 media was replaced with RPMI 1640 that did not contain antibiotic-antimycotic 178 solution at least 24 hrs prior to infection. 179 180 Macrophage infection assays. PBMCs were seeded into 6-well tissue culture plates at 2 x10 6 181 cells/ml. On day 12, MDMs were washed once with Hanks balanced salt solution (HBSS) and 182 cells were dissociated by 20 min incubation with accutase at 37°C, followed by gentle cell 183 scraping. Dissociated cells were pooled, centrifuged at 400 x g for 5 mins and resuspended in 184 RPMI 1640 medium for cell counting. MDMs were seeded into 12-well tissue culture plates at 185 2 x10 5 cells/ml and returned to tissue culture incubators prior to infection. On days 13-14, 186 MDMs were washed once with HBSS and infected using frozen bacterial stocks at MoI 10. 187 Plates were centrifuged at 277 x g for 2 min to synchronise infection and then incubated for 30 188 min at 37°C. After 30 min, cells were washed twice with ice-cold PBS to remove non-adherent 189 bacteria and halt bacterial internalisation. Gentamicin was prepared in RPMI 1640 containing 190 no additional supplements at 100 µg/ml and added to infected MDMs for 30 min at 37°C to 191 kill extracellular bacteria. To measure bacterial killing, high-dose gentamicin (100 µg/ml) was 192 replaced with RPMI 1640 containing 4 µg/ml gentamicin and 0.8 µg/ml lysostaphin. Infected 193 MDMs were incubated at 37°C until 6 hpi and low-dose gentamicin/lysostaphin was removed. and Atet if required) and incubated at 37°C with aeration at 200 rpm. CFU were determinated 204 at 0.5 or 1 hr after addition of each antimicrobial compound. Experiments were repeated up to 205 ten times due to variation in survival between biological replicates. 206 207 Statistics. Statistical analyses were performed using GraphPad Prism 9.0 software. Statistical 208 differences between zebrafish larval survival experiments were evaluated using the Kaplan-209 Meier method and pairwise comparisons between survival curves were made using the log-210 rank (Mantel-Cox) test. For tolerance assays the normality of data sets were determined using 211 the Shapiro-Wilk test. Differences in tolerance were then assessed using either Mann-Whitney 212 test, or one-way ANOVA followed by Tukey's multiple comparisons test or Kruskal-Wallis 213 multiple comparison test, as indicated in the figure legends. Macrophage assays were analysed 214 by one-way ANOVA with Dunnett's multiple comparisons test. 215

(p)ppGpp is important for S. aureus virulence in a zebrafish model 217
To determine the requirement of a functional stringent response for S. aureus virulence, 218 zebrafish embryos were infected with either the community-acquired MRSA strain JE2, or a 219 JE2 (p)ppGpp 0 mutant [23]. Bacteria were injected into the bloodstream via the yolk sac 220 circulation valley (Fig. 1A). Bacteria injected here enter the heart before more widespread 221 dissemination throughout the bloodstream, culminating in bacteraemia [18]. A dose of 222 approximately 3000 -4000 CFU of wildtype JE2 led to 50% zebrafish mortality (Fig. 1). In 223 contrast, the (p)ppGpp 0 mutant killed significantly fewer larvae, which occurred with both 2A, 2B). At 21 hpi, bacterial loads had increased from the initial inoculum of 10 3 to between 234 10 5 -10 7 for both JE2 and the (p)ppGpp 0 mutant. This demonstrates that both were able to 235 replicate within the larvae, although there were more dead larvae in the JE2-infected 236 population. The (p)ppGpp 0 mutant was isolated from larvae at numbers higher than 10 3 from 237 timepoints up to 69 hpi, suggesting that the (p)ppGpp 0 mutant is also able to replicate later on 238 during infection (Fig. 2B). Altogether, this suggests that while the (p)ppGpp 0 mutant strain has 239 attenuated virulence in vivo, it is still able to replicate in the host. 240

Rel, RelP and RelQ all contribute to virulence of S. aureus 242
In S. aureus, (p)ppGpp is produced by the long bifunctional RSH enzyme Rel, as well as from 243 the two SAS enzymes RelP and RelQ in response to different stresses [3]. To understand the 244 contribution of the RSH versus the SAS enzymes to S. aureus infections, the virulence of JE2 245 and the (p)ppGpp 0 mutant were first compared to JE2 ∆relQP, a strain with in-frame deletions 246 of both SAS enzymes. Survival curves revealed that the ∆relQP mutant was able to kill larvae 247 similarly to JE2 (Fig. 3A, 3E), suggesting that the presence of Rel is sufficient for virulence in 248 this model. To confirm this, the (p)ppGpp 0 mutant was complemented with full-length rel from 249 the Atet-inducible integrative vector pCL55iTETr862 (iTET). Expression of Rel restored 250 killing of the larvae to wildtype levels (Fig. 3B, 3E), while complementation with the single 251 SAS enzyme relP did not (Fig. 3C, 3E). This confirms the importance of the Rel synthetase in 252 vivo, as has been reported previously [10,11]. 253 While the above data indicate that the presence of Rel alone is sufficient for S. aureus 254 virulence, we wanted to determine whether a strain containing the two SAS enzymes alone in 255 the absence of Rel had a virulence defect. To establish this, we used a Rel mutant strain in 256 which three conserved amino acids in the synthetase domain (Y308, Q309 and S310) are 257 deleted, rendering it unable to produce (p)ppGpp. This leaves the hydrolase function intact, 258 which is essential in strains encoding RelP and RelQ to prevent toxic accumulation of (p)ppGpp 259 [11,24]. This mutant was available in the LAC* background, a strain identical to JE2, except 260 JE2 has been cured of the cryptic plasmid p01 [25]. A comparison of the virulence of this 261 mutant to the wildtype LAC* revealed no difference in killing (Fig. 3D, 3E). This would 262 suggest that while the presence of Rel alone is sufficient for infection with S. aureus, the 263 combined level of (p)ppGpp produced by RelP and RelQ in LAC* relsyn is enough to 264 macrophages were incubated at 37°C for a further 6 hrs before surviving numbers were 291 determined. In comparison to the wildtype, the (p)ppGpp 0 mutant was significantly less able 292 to survive the intracellular environment within macrophages (Fig. 4B). 293 To understand more about the individual contribution of Rel versus the SAS enzymes 294 to this phenotype, we monitored the survival of both the ∆relQP and the relsyn mutants. A strain 295 lacking RelP and RelQ survived just as well as the wildtype, indicating that Rel alone is 296 sufficient to promote bacterial survival (Fig. 4B). This complements the zebrafish data, where 297 Rel alone is sufficient for virulence (Fig. 3). The LAC* relsyn mutant, where the synthetase 298 domain of Rel is inactivated, had a lower, but not statistically significant, survival rate than the 299 wildtype, and was not killed as efficiently as the (p)ppGpp 0 mutant, suggesting that while Rel 300 plays a key role in this niche, RelP and RelQ also aid survival in its absence (Fig. 4C). 301 Altogether, these data show that the reduced virulence observed in zebrafish is likely due to 302 the inability of stringent response mutants to survive within macrophages and highlights a role 303 for both the long RSH enzyme Rel and the SAS enzymes RelP and RelQ for responding to 304 stress signals and producing sufficient (p)ppGpp to allow survival within these cells. contributor to low pH is the proton-pumping v-ATPase, present on the membrane of 312 phagolysosomes, though metabolites such as itaconic acid also contribute to this. Itaconic acid 313 is produced in phagocyte mitochondria by aconitate decarboxylase, an enzyme that converts 314 aconitic acid, a by-product of the Krebs cycle, to itaconic acid [31]. 315 Previously, a methicillin sensitive S. aureus (MSSA) (p)ppGpp 0 mutant exhibited 316 susceptibility to H2O2 [5], while an MRSA (p)ppGpp 0 mutant displayed reduced tolerance to 317 HOCl [32]. These studies suggest that the stringent response is important for surviving stresses 318 within the phagolysosome. To confirm this, JE2 and the (p)ppGpp 0 mutant were exposed to 319 HOCl, H2O2 and itaconic acid, and tolerance quantified (Fig. 5A-C). In keeping with previous 320 observations, the (p)ppGpp 0 mutant was 1-2 log more susceptible to HOCl and H2O2, and 321 additionally showed a 1-log decreased tolerance to itaconic acid. To examine roles for each 322 synthetase in combatting external stressors, the (p)ppGpp 0 mutant was complemented with 323 either the RSH enzyme Rel or the SAS enzyme RelP. Expression of RelP, while alone was 324 unable to restore virulence in zebrafish (Fig. 3C), was sufficient to restore tolerance to both 325 ROS stress and pH stress in vitro (Fig. 5D, 5E), while expression of Rel conferred tolerance to 326 ROS stress only (Fig. 5D). These data are in keeping with previous reports suggesting roles for 327 SAS enzymes in responding to pH stress [3,33], as well as work indicating that expression of 328 Rel is sufficient to combat ROS stress [32]. They also support the idea that the different classes 329 of synthetase produce (p)ppGpp in response to different environmental stresses. 330 331

Overproduction of (p)ppGpp confers tolerance to stress conditions in vitro but reduces 332 virulence 333
The overproduction of (p)ppGpp in a clinical S. aureus strain has been associated with a 334 persistent infection that did not respond well to antibiotic therapy [34,35]. As (p)ppGpp acts 335 to aid bacteria in surviving stresses, excess (p)ppGpp may serve to provide enhanced 336 protection. Thus, we investigated how (p)ppGpp overproduction affects the survival of S. 337 aureus in the presence of H2O2, HOCl and itaconic acid. Here, we introduced the Atet-inducible 338 multi-copy plasmid pALC2073-relQ into JE2 and measured bacterial survival upon 339 overproduction of (p)ppGpp. In the presence of all three stressors, excess (p)ppGpp led to an 340 increase in survival in comparison to the JE2 empty vector strain (Fig. 6A-C). These results 341 indicate that surplus (p)ppGpp has a protective effect in vitro and could contribute to the 342 survival of S. aureus in stress conditions found within a macrophage. 343 Following this, we were curious to examine how overproduction of (p)ppGpp would 344 impact the virulence of S. aureus. The (p)ppGpp overproduction strain JE2 iTET-rel, where a 345 second copy of rel is integrated into the S. aureus genome, was injected into zebrafish embryos 346 alongside the empty vector-containing controls. Here, overproduction of (p)ppGpp killed fewer 347 larvae compared to the wildtype (Fig. 6D, 6E). While the in vitro data show that excess 348 Under nutrient-rich conditions, the CodY transcription factor represses genes related to nutrient 354 acquisition and stress, including those involved in nitrogen and amino acid metabolism, as well 355 as some virulence-associated genes [12]. In S. aureus, this repression requires GTP and 356 branched-chain amino acids as CodY cofactors. During the stringent response, (p)ppGpp levels 357 rise and cellular GTP levels fall. This is due to the consumption of GTP during the production 358 of (p)ppGpp, in addition to the active inhibition of enzymes in the GTP synthesis pathway by 359 (p)ppGpp [4]. This leads to the derepression of CodY and thus, the expression of stress-related 360 genes in order to cope with the change in environment. As GTP levels in S. aureus are increased 361 in strains lacking (p)ppGpp [23], we hypothesised that the continued repression of the CodY 362 regulon by GTP-bound CodY in the (p)ppGpp 0 mutant could be responsible for the decreased 363 virulence phenotype observed in zebrafish larvae. 364 To investigate this, a codY mutant was introduced into both the wildtype and (p)ppGpp 0 365 backgrounds, and the strains first exposed to itaconic acid and H2O2. In both the wildtype and 366 the (p)ppGpp 0 backgrounds, deleting codY rendered cells considerably more tolerant to stress 367  demonstrated by the attenuation of virulence for the (p)ppGpp 0 strain (Fig. 1). During nutrient 387 starvation, (p)ppGpp production and subsequent GTP depletion are required for the 388 derepression of amino acid transport and synthesis genes via the transcription factor CodY [12]. 389 This led us to hypothesise that the absence of (p)ppGpp could lead to a lack of nutrient 390 acquisition and a subsequent growth defect in vivo. However, this was not the case, as both the 391 wildtype and the (p)ppGpp 0 mutant were able to replicate in vivo (Fig. 2). In agreement with 392 this, cutaneous abscess formation by an S. aureus rel mutant was previously observed to be 393 diminished, however the CFU per abscess was similar to the wildtype [10]. This is in contrast 394 to the lower bacterial load of an MSSA relsyn mutant recovered from murine kidneys [11], 395 indicating that differences in bacterial load may occur at specific tissue sites. 396 The ability of the S. aureus (p)ppGpp 0 mutant to replicate in vivo (Fig. 2), coupled with 397 the observation that deleting the CodY repressor does not restore killing (Fig. 7), suggests that 398 the absence of (p)ppGpp result in a survival or virulence defect, rather than a growth defect 399 due to a lack of nutrient acquisition. This hypothesis is supported by multiple studies 400 demonstrating an impact of the stringent response on virulence [9,[41][42][43]. By transiently 401 depleting zebrafish embryos of myeloid cells, we demonstrated that the virulence of the 402 (p)ppGpp 0 mutant could be restored (Fig. 4A), supporting the idea that phagocytes are required 403 for controlling S. aureus infection. As neutrophils are the most abundant circulating phagocyte 404 [44], and thus are often the first immune cells to infiltrate a site of infection, further studies 405 using this cell type are needed to examine the broader importance of the stringent response for 406

infection. 407
The requirement of (p)ppGpp to survive ROS stress has been noted previously [32]. We 408 extend this by showing that the stringent response is also necessary for tolerating itaconic acid, 409 which may contribute to survival within macrophages. In vivo, itaconic acid has functions in 410 addition to modulating the pH. It can have antimicrobial effects by inhibiting isocitrate lyase, This study demonstrates that all three (p)ppGpp synthetases contribute to the virulence 427 of S. aureus. The importance of Rel, revealed by Geiger and colleagues [11,20], is corroborated 428 by our studies where we show that both the expression of rel in a (p)ppGpp 0 mutant, and the 429 presence of rel alone in a ∆relQP mutant, is sufficient to maintain wildtype levels of virulence 430 (Fig. 3). This is then extended to show that both RelP and RelQ were sufficient for virulence 431 (Fig. 3) and partial survival within macrophages in the absence of Rel (Fig. 4C). Due to the 432 roles of RelP and RelQ in responding to cell wall and pH stress [4], and the fact that low pH is 433 a condition encountered by pathogens following phagocytosis, it is not surprising that RelP and 434 RelQ may play an important role in producing the (p)ppGpp required for the survival of S. increased antibiotic tolerance. Here, we observed that the overproduction of (p)ppGpp 441 increased the tolerance of S. aureus to acid and ROS stress in vitro ( Fig. 6A-C), however excess 442 (p)ppGpp reduced S. aureus virulence (Fig. 6D). Likewise, Gao and colleagues found that a 443 Rel hydrolase domain mutation led to a permanently activated stringent response in a clinical 444 S. aureus strain isolated from a persistent infection. When the mutation was recapitulated in 445 the laboratory, the strain also displayed attenuated virulence in a Galleria mellonella model 446 [35]. This highlights the importance of regulating levels of (p)ppGpp within bacteria, and 447 supports the idea that while (p)ppGpp overproduction may increase long-term persistence in 448 vivo [35], it has a negative impact on virulence. Infection models lasting longer that 93 hpi 449 would be required to delve into the role of (p)ppGpp in chronic infection further [36,37]. 450