Timing is everything – impact of naturally occurring 1 Staphylococcus aureus AgrC cytoplasmic domain adaptive 2 mutations on auto-induction 3

Mutations in the polymorphic Staphylococcus aureus agr locus responsible for quorum sensing (QS) dependent virulence gene regulation occur frequently during host adaptation. 19 In two genomically closely related S. aureus clinical isolates exhibiting marked differences in 20 Pantone-Valentine leukocidin production, a mutation conferring an N267I substitution was 21 identified in the cytoplasmic domain of the QS sensor kinase, AgrC. This natural mutation 22 delayed the onset and accumulation of auto-inducing peptide (AIP) and showed reduced 23 responsiveness to exogenous AIPs. Other S. aureus strains harbouring naturally occurring 24 AgrC cytoplasmic domain mutations were identified including T247I, I311T, A343T, L245S 25 and F264C. These mutations were associated with reduced cytotoxicity, delayed/reduced 26 AIP production and impaired sensitivity to exogenous AIP. Molecular dynamics simulations 27 were used to model the AgrC cytoplasmic domain conformational changes arising. While 28 mutations were localised in different parts of the C-terminal domain, their impact on 29 molecular structure was manifested by twisting of the leading helical hairpin  1-  2, 30 accompanied by repositioning of the H-box and G-box along with closure of the flexible loop 31 connecting the two and occlusion of the ATP-binding site. Such conformational 32 rearrangements of key functional subdomains in these mutants highlight the cooperative 33 response of molecular structure involving dimerization, ATP binding and phosphorylation, as 34 well as the binding site for the downstream response element AgrA. These appear to increase the threshold for agr activation via AIP-dependent autoinduction so reducing virulence and maintaining S. aureus in an agr -down-regulated ‘colonization’ mode. activation which an integral part the quorum sensing system. response cytoplasmic domain are consistent with repositioning of key functional domains, impairing dimerization and 45 restricting access to the ATP binding pocket. Strains harbouring specific AgrC-cyt mutations 46 exhibit reduced AIP auto-induction efficiency and a timing-dependent attenuation of 47 cytotoxicity which may confer a survival advantage during established infection by promoting 48 colonization while restricting unnecessary overproduction of exotoxins. that switched the virulence phenotype from cytotoxic to colonizing but was reversed on mutating back to C223Y. However not all naturally occurring agr mutations are likely to be inactivating but may modify the timing and 119 strength of agr induction. Here we explore the experimental impact of naturally occurring 120 AgrC cytoplasmic domain substitutions on exotoxin (PVL) and AIP production focusing on 121 the dynamics of agr activation and their impact on the conformation of the AgrC cytoplasmic 122 domain using molecular dynamics simulations. The findings presented have potential implications for the adaptation of S. aureus during infection and the clinical potential of agr 124 antagonists as anti-virulence agents. agr agr

116 cytoplasmic domain substitution ( Fig. 1(b)) that switched the virulence phenotype from 117 cytotoxic to colonizing but was reversed on mutating back to C223Y. However not all 118 naturally occurring agr mutations are likely to be inactivating but may modify the timing and common when compared to an ST22 reference strain).
To determine the relative 138 expression levels of the PVL genes in TS13 and TS14, quantitative RT-PCR was carried out 139 using lukF-PV and found to be 59.7 (95% CI 37.8 to 94.1, p<0.01) fold higher in TS14 ( Fig.   140 2(a)). Since differences in agr expression may account for the observed variation, we also 141 examined agr transcription by measuring RNAIII levels and found an 11.3 (95% CI 7.0 to 142 18.4, p<0.05) fold difference between the two strains ( Fig. 2(a)). This defect in PVL 143 production in TS13 could be restored to TS14-like levels by provision of exogenous AIP-1 at 144 the time of inoculation (0 h) ( Fig. 3(a) and 3(c)). Similar results were also obtained with the 145 low level PVL producer, agr group 3 strain TS12 in response to exogenous AIP-3 ( Fig. 3(a)).

146
For TS14 but not TS13, PVL could be detected at the end of log phase some 5.5 h after 147 inoculation ( Fig. 3(b)). However, when AIP-1 was added to TS13 after 5.5 h growth, PVL 148 was not restored (Fig. 3(c)). For TS14, PVL production was inhibited by the cross group agr 149 inhibitor (Ala 5 )AIP-1 when added at 0 or 2 h after inoculation but not after 4 or 6 h of growth 150 ( Fig. 3(d)). These data confirm the agr-dependent nature of PVL production in these strains

156
Since PVL-production in TS13 responds to exogenous AIP-1, we investigated whether the 157 AIP production profiles differed between TS13 and TS14 as a function of growth phase.

198
I311T AgrC substitution with a subset of these also having an A343T substitution. Since 199 strain HU24 (I311T and A343T) was substantially less cytotoxic than strain IU12 (I311T) and 200 the only difference in the agr locus was the A343T substitution, we profiled AIP production 201 profiles as a function of growth for both strains. Fig. 5

219
Cytoplasmic domain AgrC substitutions in the same region as N267I which have previously 220 been reported to affect exotoxin production include L245S (18) and F264C (12). We  Table S5.

227
How do the AgrC cytoplasmic domain substitutions impact on structure?

229
The structure of the AgrC cytoplasmic domain (residues 201 to 430) divided into DHp

230
(dimerization and histidine phosphorylation) and CA (catalytic and ATP binding) sub-231 domains is illustrated in Fig. 1(b) which also indicates the locations of the naturally occurring 232 AgrC mutations described above.

234
To gain further insights into their impact on domain structure, we generated homology 235 models of the cytoplasmic domains of the AgrC wild type (WT) and F251Y, N267I, I311T,

250
The double substitution F251Y-N267I enhances the kink in the trailing helix 2 of the N-

251
terminal hairpin, as observed also in the single mutation N267I, which leads to a very similar 252 conformational change in the C-terminal catalytic domain as for N267I alone (Fig. 6) that 253 resulted in a 3-fold increase in the EC 50 ( Fig. 4(a)). Both mutations are within the putative 254 AgrA-binding region and considering the similarity in conformational change in the CA 255 subdomain, the observed change in EC 50 can be attributed to alteration of the binding site 256 for AgrA resulting from the enhanced kink in helix-2.

259
I311T is located within the catalytic end of the subdomain, A343T is located directly adjacent 260 to the ATP-binding site (cf. Fig S2). The conformation of helical hairpin 1, 2 remains almost 261 unperturbed while a conformational rearrangement is observed in the CA region including

275
The double mutation T247I-I311T is associated with a collapse of the helical hairpin onto the 276 CA-subdomain (Fig. 6). This occurs at the end of helix 2 and in the vicinity of I230 within 277 helix 1 where even in the WT, we observe an inherent weakness seen as slight uncoiling 278 and bending in helix 1. The location of I230 within helix 1 marks the beginning of the H-box 279 ( Fig. 1(b)). Despite this dramatic domain rearrangement and collapse of the helical hairpin

284
Mutations T247I, F251Y and N267I are located near the turn between helices 1 and 2 285 directly in the putative AgrA-binding region. Substitutions T247I and N267I kink the long 286 helices 1 and 2, as well as altering their mutual orientation and interactions (Fig. 6)

295
The observed changes in domain organisation within the cytoplasmic domain of AgrC can 296 affect both global and local motional freedom in the polypeptide chain. Changes in local 297 segmental mobility at individual amino acid residues that result from specific mutations, were 298 followed via the root mean square deviation (RMSD) along the trajectory (Fig. S1).

299
Cumulative average excursions over the entire trajectory are summarised in Table S6.

300
Overall, the cytoplasmic domain remains well-structured with little mobility over the majority

303
binding site, as well as the turn between helices 1 and 2 E250-L260.

305
The local mobility of loop F386-L395 in A343T and I311T-A343T is reduced compared to 306 WT, while I311T alone significantly increases loop mobility. This shows that A343T alone is 307 the deciding factor in restricting local loop dynamics that is sufficient to counter the increase 308 in motional freedom resulting from substitution I311T (Fig. S1). Specifically, flexibility within 309 the WT loop is focussed in GLG around L395, while in all other cases an increase in loop 310 mobility is observed over a much longer stretch of the chain. Since this loop forms a major 311 part of the ATP-binding pocket, changes in flexibility will likely affect both access of ATP to 312 the binding site and ATPase kinetics.

314
Mutations N267I just after and T247I just before the helical hairpin turn (Fig. 6) (Fig. 6). Within the AgrC 321 mutations investigated this was the only case in which a major structural rearrangement of 322 the cytoplasmic domain is observed.

324
The mutations identified within the cytoplasmic domain of AgrC have a pronounced effect on 325 the mobility of loop E384-G394 located within the G-box ( Fig. 1(b)) that correlates with

330
To assess the impact of these mutations on loop conformation we used the pairwise radial 331 distribution function, RDF, g(r) determined between reporter residues K389 in the middle of 332 the loop and E342 located on helix 4 ( Fig. 1(b) and Fig. 7). The RDF provides a statistical 333 snapshot of the distance between these residues weighted by the time (number of trajectory 334 frames) the two residues spend at that particular distance (Fig. 7). The RDF in the WT 335 simulation shows a bimodal trajectory occupation with prevalent inter-residue distance near 336 20 Å corresponding to a more open loop conformation and a minor population centred near 337 9 Å corresponding to a closed loop conformation (Fig. 7). Also a bimodal trajectory 338 occupation is observed in I311T but in the reverse occupation with a major closed cluster at

372
The key AgrC substitution in TS13, i.e. N267I, distorted packing of the leading alpha-helices 373 in the dimerization domain independently of the presence of Y251F, which alone also 374 changed orientation of the catalytic domain with respect to the helical hairpin dimerization 375 interface (Fig. 6). This mutation may either impair dimerization or impede auto-

449
This study demonstrates how naturally occurring AgrC mutations impact on agr activation 450 kinetics and exotoxin production with reference to the structure and function of the AgrC.

451
Future studies will need to address how the tempo-spatial and environmental factors present The bacterial strains and plasmids used in this study are described in Table S1. S. aureus 463 strains were grown aerobically in brain heart infusion (BHI) or CYGP broth (24) at 37 ºC with 464 shaking at 250 rpm. Where required growth media were supplemented with erythromycin 465 (10 µg/ml), chloramphenicol (10 µg/ml) or tetracycline (10 µg/ml). Oligonucleotides used in 466 the study are listed in Table S2. AIP-1, AIP-3 and (Ala 5 )AIP-1) were synthesized as

557
The models were in good agreement and were subsequently annealed by all atom molecular 558 dynamics for 60 ns using NAMD (20)

779
on March 19, 2020 by guest