The two-component system CopRS maintains femtomolar levels of free copper in the periplasm of Pseudomonas aeruginosa using a phosphatase-based mechanism

Two component systems control periplasmic Cu+ homeostasis in Gram-negative bacteria. In characterized systems such as Escherichia coli CusRS, upon Cu+ binding to the periplasmic sensing domain of CusS, a cytoplasmic phosphotransfer domain phosphorylates the response regulator CusR. This drives the expression of efflux transporters, chaperones, and redox enzymes to ameliorate metal toxic effects. Here, we show that the Pseudomonas aeruginosa two component sensor histidine kinase CopS exhibits a Cu-dependent phosphatase activity that maintains a non-phosphorylated CopR when the periplasmic Cu levels are below its activation threshold. Upon Cu+ binding to the sensor, the phosphatase activity is blocked and the phosphorylated CopR activates transcription of the CopRS regulon. Supporting the model, mutagenesis experiments revealed that the ΔcopS strain showed constitutive high expression of the CopRS regulon, lower intracellular Cu+ levels, and larger Cu tolerance when compared to wild type cells. The invariant phospho-acceptor residue His235 of CopS was not required for the phosphatase activity itself, but necessary for its Cu-dependency. To sense the metal, the periplasmic domain of CopS binds two Cu+ ions at its dimeric interface. Homology modeling of CopS based on CusS structure (four Ag+ binding sites) clearly explains the different binding stoichiometries in both systems. Interestingly, CopS binds Cu+/2+ with 30 × 10−15 M affinities, pointing to the absence of free (hydrated) Cu+/2+ in the periplasm. IMPORTANCE Copper is a micronutrient required as cofactor in redox enzymes. When free, copper is toxic, mismetallating proteins, and generating damaging free radicals. Consequently, copper overload is a strategy that eukaryotic cells use to combat pathogens. Bacteria have developed copper sensing transcription factors to control copper homeostasis. The cell envelope is the first compartment that has to cope with copper stress. Dedicated two component systems control the periplasmic response to metal overload. This manuscript shows that the copper sensing two component system present in Pseudomonadales exhibits a signal-dependent phosphatase activity controlling the activation of the response regulator, distinct from previously described periplasmic Cu sensors. Importantly, the data show that the sensor is activated by copper levels compatible with the absence of free copper in the cell periplasm. This emphasizes the diversity of molecular mechanisms that have evolved in various bacteria to manage the copper cellular distribution.

phosphorylates the response regulator CusR. This drives the expression of efflux 22 transporters, chaperones, and redox enzymes to ameliorate metal toxic effects. Here, we 23 show that the Pseudomonas aeruginosa two component sensor histidine kinase CopS 24 exhibits a Cu-dependent phosphatase activity that maintains a non-phosphorylated CopR 25 when the periplasmic Cu levels are below its activation threshold. Upon Cu + binding to 26 the sensor, the phosphatase activity is blocked and the phosphorylated CopR activates 27 transcription of the CopRS regulon. Supporting the model, mutagenesis experiments 28 revealed that the DcopS strain showed constitutive high expression of the CopRS regulon, 29 lower intracellular Cu + levels, and larger Cu tolerance when compared to wild type cells. 30 The invariant phospho-acceptor residue His235 of CopS was not required for the 31 phosphatase activity itself, but necessary for its Cu-dependency. To sense the metal, the 32 periplasmic domain of CopS binds two Cu + ions at its dimeric interface. Homology 33 modeling of CopS based on CusS structure (four Ag + binding sites) clearly explains the 34 different binding stoichiometries in both systems. Interestingly, CopS binds Cu +/2+ with 30 35 x 10 -15 M affinities, pointing to the absence of free (hydrated) Cu +/2+ in the periplasm. 36

IMPORTANCE 37
Copper is a micronutrient required as cofactor in redox enzymes. When free, copper is 38 toxic, mismetallating proteins, and generating damaging free radicals. Consequently, 39 copper overload is a strategy that eukaryotic cells use to combat pathogens. Bacteria 40 have developed copper sensing transcription factors to control copper homeostasis. The 41 cell envelope is the first compartment that has to cope with copper stress. Dedicated two 42 component systems control the periplasmic response to metal overload. This manuscript 43 shows that the copper sensing two component system present in Pseudomonadales 44

INTRODUCTION 50
Cooper is a cellular micronutrient required for redox enzymatic functions (1, 2). However, 51 free Cu undergoes deleterious Fenton reactions, metallates noncognate binding sites, 52 and promotes disassemble of Fe-S centers (3,4). Early studies in the field took advantage 53 of Cu toxicity to identify widely distributed proteins conferring metal tolerance; namely, 54 metal sensing transcriptional regulators and efflux transporters (1,(4)(5)(6)(7). Recent studies 55 have, however, started to uncover regulated distributions systems that move the metal 56 among cellular compartments and target Cu + to cognate metalloproteins while 57 maintaining the required homeostasis (8-15). These include Cu + sensing transcriptional 58 regulators, influx and efflux transmembrane transporters, chaperones, and storage 59 molecules. In this context, bacterial cells prevent Cu toxicity by expressing some of these 60 molecules in response to high intracellular metal conditions. The cytoplasmic response 61 to Cu + excess has been characterized in numerous Gram-positive and Gram-negative 62 bacteria (11,(16)(17)(18)(19). Nevertheless, periplasmic components involved in Cu + homeostasis 63 have received much less attention. A simple consideration of the Gram-negative bacterial 64 architecture points out that periplasmic dyshomeostasis is likely to precede the 65 cytoplasmic response to a surge of Cu + influx. Supporting this idea, mathematical 66 simulations based on Cu + uptake experiments in Pseudomonas aeruginosa under 67 kinase and phosphatase activities determines the RR~P levels, modulating the output 91 response (38). In the archetypical E. coli CusRS TCS, Cu + binding to the periplasmic loop 92 of CusS promotes its autophosphorylation, and the subsequent phosphorylation of the 93 transcriptional regulator CusR (Fig. 1A). Supporting this model, deletion of either the SHK 94 CusS or the RR CusR leads to a reduced tolerance to external Cu 2+ , increased 95 intracellular Cu + , and lack of transcriptional activation of regulated genes (e.g. cusC) (24-96 27). 97 The regulons controlled by the canonical Cu + responsive TCS are limited to genes coding 98 for the Resistance-Nodulation-Division (RND) transporter CusCFBA (26), the 99 PcoABCDRSE (27), and CopABCDRS systems (34,35). However, Cu + homeostatic 100 pathways do not behave as evolutionary units. Instead, distinct species assemble 101 different repertoires of metal handling proteins to achieve periplasmic Cu + homeostasis 102 (21). In particular, the P. aeruginosa CopRS regulon includes genes coding for an outer 103 membrane transporter (PcoB), a multicopper oxidase (PcoA), and auxiliary proteins (PtrA, 104 PA2807, QueF) whose role in periplasmic Cu + distribution is still unclear (40-42) (Fig. 1B). 105 Interesting, the P. aeruginosa CusCBA transporter is not part of the CopRS regulon but 106 is rather controlled by the cytoplasmic Cu + sensor CueR (9). Given the distinct 107 architecture of the CopRS regulon, could it be expected a distinct sensing/activating 108 mechanism for the control of periplasmic Cu + homeostasis in P. aeruginosa? 109 The structure of the isolated periplasmic domain of E. coli CusS shows four Ag + (acting 110 as Cu + analog) binding sites per dimer (43). Two sites are symmetrically located at the 111 dimer interface, and two are situated in outer loops of separated monomers. Reported 112 estimates of metal-sensor affinities are limited and quite different. The E. coli CusS 113 interacts with Ag + with an affinity in the µM range (44), while Synechocystis CopS binds 114 Cu 2+ with high sub-attomolar affinity (32). Then, significant aspects of sensor activation 115 appear undefined. Consider that selectivity and sensitivity will determine the level of free 116 metal in the periplasm and provide evidence for the Cu redox status. This is, can both 117 Cu + /Cu 2+ bind the sensor? What is the affinity of the sensor for these ions? 118 Here, we report that the transcriptional control of the CopRS regulon in P. aeruginosa 119 relies on the Cu-dependent phosphatase activity of CopS, rather than on its kinase 120 activity. The RR CopR appears constitutively phosphorylated. However, in the absence 121 of Cu, CopS dephosphorylates CopR shutting down the transcriptional response to Cu + . 122 When the periplasmic Cu + level rises, the phosphatase activity of CopS is blocked, 123 allowing the accumulation of phosphorylated CopR (CopR~P) which promotes the 124 expression of the periplasmic Cu + -homeostasis network. Finally, CopS binds both Cu + 125 and Cu 2+ with similar high affinities ensuring the absence of free Cu in the periplasm. 126

RESULTS 127
CopRS controls P. aeruginosa periplasmic Cu + homeostasis (9). Notably, there are 128 significant differences between the CopRS regulon and those of other characterized Cu + 129 sensing TCSs, e.g. E. coli CusRS. The likely presence of additional mechanistic and 130 molecular differences warranted a closer examination of CopRS function. 131 Mutation of CopS leads to Cu tolerance. We initiated our studies by looking at the 132 growth rate of DcopS and DcopR mutant strains in the presence of external metal. Based 133 on the mechanism of described TCS (Fig. 1A), it was expected that the lack of either 134 CopS or CopR would lower the cellular tolerance to external Cu 2+ . As anticipated, the 135 DcopR strain was more susceptible to Cu 2+ than the WT strain (Fig. 2). In contrast, two 136 independent copS transposon mutants, PW5705 and PW5706 (Fig. S1), were 137 surprisingly much more tolerant to external Cu 2+ than the WT strain. As these phenotypes 138 were reversed by complementation with the corresponding genes all subsequent 139 experiments were performed with the DcopS PW5706 strain. For comparison, in addition 140 to the WT strain, the well characterized Cu + sensitive DcopA1 mutant strain was also 141 included as control in this initial phenotypical characterization (8). 142 Importantly, these growth phenotypes were the consequence of significantly different 143 levels of intracellular Cu + upon exposure to CuSO4 (Fig. 3). Thus, the DcopR mutant strain 144 accumulated more Cu + , while the DcopS cells stored less metal than the WT strain. Again, 145 alterations in Cu + levels were reversed by gene complementation of the mutant strains. 146 These differences in Cu tolerance and cellular metal levels observed for the ΔcopR and 147 ΔcopS mutant strains cannot be explained by the currently accepted model for the E. coli 148 TCS CusRS (Fig. 1A) and suggest an alternative mechanism for coupling periplasmic Cu + 149 sensing and gene expression in P. aeruginosa. 150 The CopRS regulon is constitutively expressed in the DcopS mutant strain 151 independently of Cu + levels. Toward understanding the increased Cu tolerance and 152 intracellular levels in the DcopS strain, we investigated the transcriptional response to 153 Cu 2+ exposure of the CopRS regulon in the DcopR and DcopS mutant strains. We have 154 described that CopRS controls the expression of pcoA, pcoB, ptrA, queF, and PA2807 155 coding for periplasmic and outer membrane proteins (Fig. 1B) (9). As previously observed 156 in the WT strain, genes of the CopRS regulon are induced in response to external Cu 2+ 157 exposure (Fig. 4). As expected, their Cu-induced expression was abolished in the DcopR 158 mutant. Quite the opposite, the DcopS mutant strain showed a constitutive activation of 159 all the genes of the CopRS regulon, even in the absence of the Cu 2+ stimulus. In the 160 DcopS background, high expression levels of these genes were similar either under the 161 absence of added Cu 2+ , low, non-deleterious 0.5 mM Cu 2+ levels, intermedium toxic 2 mM 162 Cu 2+ , and high lethal 4 mM Cu 2+ (Fig. S2). This suggests that CopS is not necessary to 163 activate, i.e. phosphorylate, CopR. Furthermore, it implies that CopR is phosphorylated 164 independently of CopS. The activation of CopR in the DcopS mutant in the absence of 165 supplemented Cu 2+ points to a mechanism where the phosphatase activity of CopS 166 maintains low levels of CopR~P under noninducing conditions. This defect in the DcopS 167 strain to maintain the system off in absence of Cu + was reversed in the complemented 168 strain (Fig. 4). The transcriptional analysis also showed that the expression of the copRS 169 operon is not autoregulated by CopRS (Fig. S3). This is, even though copRS expression 170 is induced in response to Cu + , no defects on its expression were observed in the DcopR 171 or the DcopS mutant strains. Noticeably, the repressed transcription of oprC, codding for 172 the outer membrane Cu importer (9, 45), was further repressed on the DcopS mutant 173 strain, consistent with the Cu + tolerant phenotype exhibited by this strain (Fig. S4A). 174 Conversely, the increased transcription of genes in the CueR regulon (copA1 and cusA) 175 in response to Cu + , was not altered neither in DcopR nor DcopS mutant strains (Fig. S4B). 176 This confirms that the lack of transcriptional control observed in the DcopR and DcopS 177 mutant strains is limited to genes of the CopRS regulon. 178 His235 acts as a switch to turn on/off the CopS signaling pathway. The cytoplasmic 179 region of the SHK sensory proteins contains the catalytic phosphotransfer domain able 180 to switch between kinase and phosphatase activities in a signal-dependent manner (39, 181 46   The CopS periplasmic sensor binds Cu +/2+ with femtomolar affinities. By analogy on 237 how cytoplasmic sensors metal affinities are tuned to maintain free metal levels (48,49), 238 the affinity of CopS for Cu + ions will certainly have determinant effects on free (hydrated) 239 Cu + ions levels in the periplasm. Exploring the binding of Cu + to CopS, we measured the 240 sensor metal binding affinity using competing ligands. The ligands were present in excess 241 to ensure effective competition. In all cases, the determinations were performed assuming 242 that both Cu sites at the CopS dimer interface were functionally independent and 243 thermodynamically indistinguishable. Initial determinations of CopS(34-151) affinity for Cu + 244 using the bathocuproine disulfonate (BCS) as competitor, showed a limited but 245 measurable decrease in the absorbance of the [Cu I (BCS)2 3-] complex, corresponding to 246 a KD value of CopS(34-151) for Cu + of 22 x 10 -15 M. However, it was apparent that CopS 247 was not an effective competitor with BCS for the metal. Instead 2,2′-bicinchoninic acid 248 (BCA), with a lower affinity for copper compared to BCS, appeared more appropriate to 249 measure affinities in the femtomolar range (50). Using BCA as the competing ligand and 250 fitting titration curves to Eq. 2, a CopS(34-151)-Cu + KD = 27.7 ± 0.7 x 10 -15 M was obtained 251 (Fig. 8A). This appears within the range of affinities observed for many other Cu + binding 252 molecules (11,50,51). 253 On the other hand, Synechocystis CopS binds Cu 2+ with high sub-attomolar affinity (Cu + 254 binding stoichiometry was not reported) (32). Exploring the possibility of high affinity Cu Salmonella, that has a distinct Cu + balance mechanisms (6) The Cu resistance phenotype of the P. aeruginosa DcopS strain, is supported by the 288 constitutive expression of the CopRS regulon and the consequent reduced whole-cell Cu + 289 content. The simplest explanation for these observations is a mechanism where in 290 absence of Cu + , the CopS phosphatase activity abrogates the induction of the CopRS 291 regulon by maintaining low levels of CopR~P ( Fig. 9). When CopS detects periplasmic 292 Cu overload, its phosphatase activity is blocked allowing the accumulation of CopR~P, 293 which promotes the expression of the periplasmic Cu-homeostasis network. 294 Signal transduction by archetypical TCSs relies on bifunctional kinase/phosphatase 295 SHKs (56). A positive action results from sensor autokinase activity and phosphotransfer 296 to the RR; while, negative regulation involves the sensor phosphatase activity (46). The 297 ultimate determining factor of the cascade activation is the phosphorylation status of the 298 RR. Accumulation of a RR~P is the consequence of a signal-dependent stimulation of the 299 sensor-kinase activity or a signal-dependent blockage of the sensor-phosphatase activity. 300 Our data suggests that CopS harbors autokinase and phosphatase activities. The signal-301 independent activation of the CopRS regulon in the DcopS background evidences the 302 requirement of the CopS phosphatase activity to maintain low levels of CopR~P in the 303 absence of Cu. It is also apparent that CopS is not required for the phosphorylation of 304 CopR, implying that an alternative mechanism for the phosphorylation of CopR should 305 exist. There is extensive evidence that RRs can be phosphorylated by endogenous 306 phosphor-donors like acetyl phosphate (57). Alternative mechanisms for RR 307 phosphorylation (cross-phosphorylation) known as many-to-one or one-to-many, where 308 many SHKs phosphorylate a given RR or a single SHK phosphorylates multiple RRs have 309 been proposed (38, 56). It could then be argued that CopR phosphorylation might be the 310 consequence of an unspecific crosstalk with a non-cognate SHK that occurs only in the 311 absence of CopS. However, such crosstalk has been only observed when both, the 312 reciprocal RR and the cognate SHK, were absent (58). These conditions are distinct from 313 those in our experiments. 314 The evidence indicates that Cu-dependent CopS autokinase activity is required for the 315 inhibition of the CopS-phosphatase activity. Replacement His235Ala, leads to constitutive 316 phosphatase activity regardless the periplasmic Cu + levels. While this points out that 317 His235 is not required for the CopS phosphatase activity, it implies that Cu-dependent 318 CopS autophosphorylation turns off the CopS phosphatase activity, leading to 319 accumulation of CopR~P. This is, as described, the dephosphorylated SHKs has 320 phosphatase activity (39,46). 321 CopS binds Cu 2+ with sub-attomolar affinity (32). It would be quite speculative to compare 333 such dissimilar determinations. However, it might be instructive to consider the observed 334 10 -19 -10 -21 M affinities of cytoplasmic copper sensors in general (51, 59); and those 335 determined for the cytoplasmic triad CopZ2/CueR/CopZ1 of P. aeruginosa, with relative 336 affinities for Cu + ranging between 10 -15 -10 -17 M (9, 11). The weaker affinity of CopS 337 compared to the cytoplasmic regulators and chaperones, is likely the consequence of a 338 metal binding site formed by His rather than Cys residues. This is a logic arrangement, 339

CopS binds two Cu
given the possible oxidation of proximal Cys under periplasmic redox stress. Importantly, 340 a femtomolar affinity still supports the idea that there would not be free Cu +/2+ in the cell 341 periplasm, as shown for the cytoplasm (51, 60). However, the relative binding strength of 342 CopS is likely to be linked to those of periplasmic Cu + chaperones that exchange metal 343 with the sensor. This is, the proteins should be able to exchange the metal. Although, as 344 shown with cytoplasmic chaperone/sensor partners, the protein-protein binding affinity 345 will have a significant effect in the final exchange constant (11). 346 CopS binds both, Cu + and Cu 2+ , with similar high affinity. It is accepted that cytoplasmic 347 transporters and chaperones, bind and distribute Cu + . However, the periplasm is a more 348 oxidizing compartment (61, 62). PcoA is a multicopper oxidase present in the P. 349 aeruginosa periplasm (63). It has been proposed that periplasmic enzymes might 350 catalyze Cu + oxidation to the assumed less toxic Cu 2+ (64). However, free (hydrated) Cu + 351 would be spontaneously oxidized by O2 in an aerobic environment. Then, the redox status 352 of periplasmic Cu is unclear. Beyond the goals of this report, we presume that Cu 353 oxidation state will depend on the molecule interacting and delivering Cu to CopS. In any 354 case, the capability to bind Cu +/2+ might serve CopS to sense the metal under redox 355

stress. 356
The distinct CopRS mechanism is in line with the singular architecture of P. 357 aeruginosa Cu homeostasis system. E. coli and Salmonella are the frequent models 358 to explore transition metal homeostasis in Gram-negative bacteria. However, recent 359 studies of P. aeruginosa have begun to show different novel molecular strategies to 360 sense, buffer and distribute Cu + (8-10, 65). For instance, consider how the regulons of 361 both compartmental sensors, CopRS and CueR, differ among these three organisms (6, 362 9, 24, 66, 67). Also, analyze the multiple functionally distinct homologous Cu + ATPases 363 present in Salmonella and Pseudomonas and how these three Gram-negative bacteria 364 have solved cytoplasmic Cu + -chaperoning using alternative strategies (6,11,68). Along 365 these observations, the relevance of periplasmic Cu + sensing, storage and transport has 366 become more apparent. Then, it is not surprising that these model systems solve 367 periplasmic Cu + sensing either via a kinase sensor (CusRS, E. coli), an integration of a 368 cytoplasmic Cu sensor with a general envelope stress response TCS (CueR-CpxRS, 369 Salmonella (69)) or a phosphatase sensor (CopRS, P. aeruginosa). The evolutive and 370 ecological advantages of these systems are still to be discovered and will be the subject 371 of future enquires in the field. 372