Mechanism of targeted killing of P. aeruginosa by pyocins SX1 and SX2

Pseudomonas aeruginosa is a common cause of serious hospital-acquired infections, the leading proven cause of mortality in people with cystic fibrosis and is associated with high levels of antimicrobial resistance. Pyocins are narrow spectrum protein antibiotics produced by P. aeruginosa that kill strains of the same species and have the potential to be developed as therapeutics targeting multi-drug resistant isolates. We have identified two novel pyocins designated SX1 and SX2. Pyocin SX1 is a metal-dependent DNase while pyocin SX2 kills cells through inhibition of protein synthesis. Mapping the uptake pathways of SX1 and SX2 shows these pyocins utilize a combination of the common polysaccharide antigen (CPA) and a previously uncharacterized TonB-dependent transporter (TBDT) PA0434 to traverse the outer membrane. In addition, TonB1 and FtsH are required by both pyocins to energise their transport into cells and catalyse their translocation across the inner membrane, respectively. Expression of PA0434 was found to be specifically regulated by copper availability and we have designated PA0434 as Copper Responsive Transporter A, or CrtA. To our knowledge these are the first S-type pyocins described that utilize a TBDT that is not involved in iron uptake.


INTRODUCTION
P. aeruginosa is a major cause of serious hospital acquired infections with treatment frequently complicated by high levels of antibiotic resistance with broad-resistance to βlactams, aminoglycosides and fluoroquinolones and growing resistance to carbapenems observed globally (1). The World Health Organization (WHO) lists P. aeruginosa in the highest threat level of 'critical' for bacterial pathogens for which new antibiotics are urgently required (2).
Protein bacteriocins, such as pyocins, are narrow-spectrum antibiotics that kill bacteria closely related to the producing strain and play a key role in colonization and competition in bacterial communities(3) , (4). One potential advantage of utilizing bacteriocins as therapeutic antibiotics is the ability to target a specific pathogen while avoiding collateral damage to the microbiota. This approach may enable antibiotic therapy without common complications associated with broad-spectrum antibiotics, such Clostridium difficile infection and domination of the microbiota with broadly drug resistant pathogens that may subsequently disseminate to cause serious systemic infection (5)(6)(7). In addition, there is growing concern that alteration of the microbiota, including that induced by the use of broad-spectrum antibiotics, may increase the risk of developing a range of chronic inflammatory diseases (8).
P. aeruginosa can produce a diverse range of pyocins. The soluble or S-type pyocins are multi-domain proteins produced by P. aeruginosa to kill strains of the same species. Stype pyocins share basic characteristics including homologous cytotoxic domains with the well-studied colicins of E. coli (9). Most characterized pyocins display a nuclease activity which targets DNA (S1, S2, S3, S8, Sn and AP41), rRNA (S6) or tRNA (S4 and SD2) (10,11). In addition, pyocin S5 is a pore-forming toxin and pyocin M inhibits peptidoglycan synthesis through the degradation of lipid II (12,13). Nuclease pyocins are normally co-expressed with immunity proteins that bind tightly to the cytotoxic domain to protect the producing strain from self-intoxication (14,15).
Initial contact with the cell surface for multiple S-type pyocins, including S5, SD2 and S2 and the unrelated L-type pyocin L1, is through the common polysaccharide antigen (CPA) (11,13) (16). CPA is a homopolymer of the rare deoxyhexose D-rhamnose that is not widely distributed in nature, with L-rhamnose being predominant. The CPA is therefore a useful generic receptor for diverse pyocins to target P. aeruginosa, enabling their accumulation on the cell surface. Structural and functional studies of pyocin S5 show that the CPA binding domain is formed by the second of two kinked three-helix bundle (kTHB) domains and that the sequence of this domain is highly conserved in pyocins that target the CPA (13). The Nterminal kTHB domains of pyocin S5 mediates interaction with the cell surface FptA transporter, a TonB-dependent transporter (TBDT). Similar to other identified pyocin transporters, the normal physiological role of FptA is in iron acquisition, in this case through transport of the iron-siderophore ferripyochelin across the outer membrane (17). Parasitisation of iron uptake pathways by pyocins extends to the energization of transport by the TonB1-ExbB-ExbD complex which services a repertoire of outer membrane TBDTs to energise outer membrane transport. In the case of pyocins, this system is hijacked through the presence of an N-terminal TonB-binding motif that directly binds to the C-terminal domain of TonB1, energizing translocation directly though the lumen of the cognate TBDT (13,18). In the case of the pore-forming pyocin S5, transport to the periplasm is sufficient for the C-terminal cytotoxic domain to insert into and depolarize the inner membrane to kill the cell. For nuclease-type pyocins, an additional inner membrane translocation domain is required to mediate transport to the cytoplasm where its nucleic acid substrate is located (19). For pyocin G, inner membrane translocation has been shown to be dependent on both the AAA + ATPase/protease FtsH and TonB1 (20).
In this study, two novel pyocins, namely SX1 and SX2, were identified and characterized. Both pyocins display potent killing activity against P. aeruginosa PAO1.
However, only pyocin SX2 affords effective protection in a Galleria mellonella model of P. aeruginosa infection. In vitro experiments demonstrated that pyocin SX1 is a metal dependent DNase. In contrast, pyocin SX2 did not show any DNase activity but inhibited protein synthesis in an in vitro transcription-translation assay. Mapping the uptake pathways of SX1 and SX2 shows these pyocins utilize the common polysaccharide antigen (CPA) as a cell surface receptor and a previously uncharacterized TonB-dependent transporter (TBDT) PA0434 as a translocator to cross the outer membrane. Interestingly, expression of PA0434 was found to be regulated by copper availability hence we have designated PA0434 as Copper Responsive Transporter A, or CrtA.

Discovery and domain organization of pyocins SX1 and SX2
To identify new pyocins we performed BLAST searches using the amino acid sequence of the pyocin S2 CPA binding and inner membrane translocation domains as query sequences (11,19). Two genes encoding putative novel pyocins, designated pyocins SX1 and SX2, were identified in the genome sequences of P. aeruginosa LMG5031 and Pseudomonas sp. 2_1_26, respectively. Both putative pyocin genes are located downstream from a P-box element which is the regulatory sequence for production of many pyocins and upstream from genes encoding putative immunity proteins(9) (Table1).  (Figure 1A). A cytotoxic domain homologous to those of pyocin SX2 and pyocin G is present in carocin D and carocin S3 and from P. carotovorum and both of these bacteriocins have been reported to display DNase activity in vitro (21,22). The deduced domain structures for pyocin SX1 and SX2 are shown in Figure 1B. Based on the above analysis we predict that pyocins SX1 and SX2 utilise the CPA as their receptor and an unknown TonBdependent receptor as their transporter.

In vitro and in vivo activity of pyocins SX1 and SX2
To determine if the putative pyocins SX1 and SX2 display killing activity against P.
aeruginosa isolates and predicted DNase activity, we produced recombinant pyocins in complex with their cognate immunity proteins in E. coli. Pyocin SX1-ImSX1 and pyocin SX2-ImSX2 were isolated by Ni 2+ affinity chromatography (via a C-terminal His6-tag on their respective immunity proteins) and gel filtration chromatography (Figure 2A). Both purified pyocins displayed potent killing activity against P. aeruginosa P7, using an overlay spot plate method on LB agar, with a minimum killing concentration of 0.051 µg/ml (0.6 nM) for SX1 and 0.457 µg/ml (5.3 nM) for SX2. Interestingly, pyocin SX2 also displayed efficacy in a Galleria mellonella P. aeruginosa infection model although pyocin SX1 did not ( Figure 2B). Further analysis indicated that pyocin SX1 is relatively rapidly inactivated on administration to G. mellonella whereas SX2 retains its biological activity over an extended time period ( Figure   2C). Survival plot for groups of larvae injected with PBS or pyocins alone and groups of larvae infected with P. To determine if pyocins SX1 and SX2 displayed their predicted nuclease activity against DNA, we separated the pyocins from their respective immunity proteins and tested their DNAse activity in a plasmid nicking assay. Similar to other HNH-DNase type pyocins, pyocin SX1 displayed a metal-dependent DNase activity. Pyocin SX1 was highly active in the presence of magnesium and nickel and able to completely degrade plasmid DNA, while in the presence of zinc a lower level of activity was observed ( Figure 3A). In contrast, pyocin SX2 did not display DNase activity under any of the tested conditions ( Figure 3B). This result is surprising given the reported DNase activity of carocin D and carocin S3, which possess a cytotoxic domain homologous to that of pyocin SX2 (21,22). To probe the activity of pyocin SX2 further, we tested its ability to inhibit protein synthesis in an in vitro transcriptiontranslation assay. In this experiment, 200 ng of immunity protein-free pyocin SX2, colicin D, a tRNase used as a positive control, or pyocin S5, a pore-forming pyocin, used here as a negative control, were added to the transcription-translation assay. Pyocin SX2 and colicin D significantly lowered the production of the reporter protein, Renilla luciferase, by 50% and 97%, respectively, when compared with the untreated control. The negative control, pyocin S5, had little effect on luciferase production ( Figure 3C). In addition, the effect on luciferase production with pyocin SX2 treatment was shown to be dose dependent ( Figure 3D). These data indicate that the cytotoxic effect of pyocin SX2 is due to interference with either transcription or translation ultimately leading to reduced protein synthesis. However, the exact molecular target of pyocin SX2 remains unknown. To date, all known S-type pyocins for which import mechanisms have been determined require a specific TBDT for translocation across the P. aeruginosa outer membrane. To identify the outer membrane transporter for pyocins SX1 and SX2, spontaneous resistant mutants were isolated by incubating late-stationary phase cells of P. aeruginosa PAO1 with pyocin SX1 or SX2 and plating on LB agar. The colonies grown after incubation were then picked and re-screened for pyocin sensitivity. In total, thirty spontaneous mutants isolated after pyocin SX1 or SX2 treatment were tested using spot tests for sensitivity against 4 pyocins: SX1, SX2, SD2 and L1 at 1 mg/ml. Testing against SD2 and L1 was performed to determine which isolates were likely to carry mutations that deplete or abolish CPA synthesis because both of these pyocins utilize the CPA as an outer membrane receptor (11,16). Since both SX1 and SX2 appear to carry a CPA binding domain, it was hypothesized that mutant strains resistant or tolerant to all 4 pyocins were likely to be deficient in CPA production whereas mutant strains only resistant or tolerant to SX1/SX2 were putative SX1/SX2 transporter mutants.
Consistent with this strategy, genome sequencing of 4 putative CPA production mutants showed these strains carried mutations in either the gmd gene encoding GDPmannose 4,6-dehydratase (GMD) or PA5455 coding for a putative glycosyltransferase (Table   S1). Previous studies have shown that GMD is involved in CPA biosynthesis and gmd is located in the CPA O-antigen gene cluster together with other CPA biosynthesis genes such as wzt and wzm (23). PA5455 is located in a conserved gene cluster adjacent to the CPA Oantigen gene cluster. The protein produced from this gene is predicted to be a glycosyltransferase which is proposed to be involved in CPA biosynthesis and modification (24). These data confirm that pyocins SX1 and SX2 utilise CPA as a receptor.
The genomes of 4 putative transporter mutants (resistant to both SX1 and SX2 but still sensitive to SD2 and L1) were sequenced and analyzed by comparing to the genome of the parent strain. All putative transporter mutants were found to carry nonsense mutations in the gene encoding the uncharacterized TBDT, PA0434. To confirm that PA0434 is the transporter for pyocins SX1 and SX2, both pyocins were tested against a P. aeruginosa strain (PAO1ΔPA0434) which contains a transposon insertion in PA0434. Both pyocins SX1 and SX2 were inactive against PAO1ΔPA0434; cell killing was restored after plasmid-based complementation indicating that PA0434 is the transporter for both pyocin SX1 and SX2 ( Figure 4). Generally, the uptake of substrates via TBDTs requires an energy-transducing TonB complex, consisting of TonB, ExbB, and ExbD proteins in the inner membrane. To determine if transport of pyocins SX1 and SX2 is Ton-dependent we tested their activity against strains lacking one of the three TonB proteins encoded by the P. aeruginosa genome. PAO1∆tonB1 was resistant to both pyocins SX1 and SX2 while PAO1∆tonB2 and PAO1∆tonB3 remained sensitive suggesting that pyocins SX1 and SX2 are TonB1 dependent ( Figure S1). In addition, previous studies have shown that an inner-membrane protein FtsH is required for killing of nuclease-type colicins as well as pyocin G. Both pyocins SX1 and SX2 were inactive against PAO1∆ftsH indicating that these pyocins are also FtsH dependent ( Figure S1). Killing by pyocins such as S5, SD2 and S2, which exploit iron siderophore transporters, is enhanced under iron limiting conditions, where expression of the transporter is upregulated (25). To examine the effect of Fe(III) concentration on PA0434 expression, P. aeruginosa PAO1 was treated with pyocins SX1, SX2 or SD2, which utilizes the ferripyoverdine transporter FpvAI, under iron-rich (with 50 µM FeCl3) and iron-limiting (with 200 µM 2,2'bipyridine) conditions. Iron availability had little effect on sensitivity of PAO1 to pyocins SX1 and SX2 whereas sensitivity to pyocin SD2 was as expected inhibited by FeCl3 and improved by bipyridine reflecting expression of the FpvAI transporter under these conditions ( Figure   S1). These results suggest that the expression of PA0434 is independent of iron availability and so PA0434 likely does not play a role in iron uptake. Consistent with this, PA0434 lacks an upstream Fur box (5'-GATAATGATAATCATTATC-3') which acts as the recognition sequence for the ferric uptake regulator (Fur) (26). In P. aeruginosa, Fur controls both metabolism and virulence in response to iron availability including iron uptake via ironsiderophores (27). Thus, PA0434 may be involved in the uptake of a molecule other than an iron-siderophore. To determine if pyocin sensitivity can be suppressed by other metal ions, reflecting repression of PA0434 expression, the effect of Zn(II), Mn(II), Ni(II) and Cu(II) ions on sensitivity of P. aeruginosa PAO1 to pyocins SX1 and SX2 was determined using the overlay spot plate assay. From the metal ions tested, only the presence of CuSO4 decreased sensitivity of PAO1 to pyocins SX1 and SX2 while the addition of other metals did not ( Figure   5A).

PA0434 is a copper responsive transporter
To confirm the specificity of the inhibitory effect of Cu(II), sensitivity of PAO1 to pyocins SX1 and SX2 was tested under a number of conditions. Pyocin SD2 and pyocin AP41 were selected as controls since pyocin SD2 utilizes the iron transporter FpvAI and sensitivity to pyocin AP41 is largely independent of iron availability (11,25). In these experiments, sensitivity to pyocins SX1 and SX2 was reduced when 1 mM CuSO4 was added to the medium.
Conversely, the killing activity of both pyocins was increased in the presence of triethylenetetramine (TETA), a high-affinity Cu(II) chelator ( Figure 5BC) Figure 5C). In addition, sensitivity to pyocin SD2 was decreased by the addition of FeCl3 and increased by the addition of bipyridine. The addition of CuSO4 and ZnCl2 did not affect sensitivity to pyocin SD2, reflecting the highly specific effect of iron availability on the expression of the iron-siderophore transporter FpvAI (Figure 5C).
The killing activity of pyocin SD2 was modestly increased in the presence of TETA or TPEN.
TETA and TPEN are recognized as selective copper and zinc chelators, respectively, although both chelators bind iron with a lower affinity (28,29). Sensitivity to pyocin AP41 was not affected by the presence of any metal ions or metal ion chelators ( Figure 5C). The pyocin AP41 transporter has not been identified but sensitivity to the pyocin is known to be independent of iron availability(25).   and TPEN also triggered increased fptA expression, but at a lower level than bipyridine. These results indicate the effect of non-specific chelation by the metal chelators ( Figure 6).
Altogether, these results suggested that PA0434 is a Cu(II) responsive transporter and may play a role in Cu(II) transport. Thus, we named this transporter as Copper Responsive Transporter A or CrtA. Error bar indicates standard error of the mean. * indicates significantly difference compared to the untreated control (T-test, P < 0.010).

Discussion
In this work, we describe the identification and characterization of two novel pyocins, SX1 and SX2. Pyocin SX1 was shown to be an active metal dependent nuclease that targets DNA. In contrast, pyocin SX2 did not display any DNase activity but inhibited protein synthesis in an in vitro transcription-translation assay. This is a surprising result since carocin D and S3, which carry cytotoxic domains homologous to pyocin SX2 have been reported to display DNase activity. Well characterized bacteriocins, including colicin E3 are known to know kill through a targeted nuclease activity against the 16S rRNA and others, such as colicin D and E5, through cleavage of specific tRNAs. Both these activities ultimately act to inhibit protein synthesis and it remains to be determined if pyocin SX2 also targets rRNA, tRNA or acts against a different component of the RNA or protein synthesis machinery.
Both pyocin SX1 and SX2 were found to utilize CPA as a cellular receptor and a novel TBDT, CrtA (PA0434) to target P. aeruginosa and cross the outer membrane. Interestingly, expression of CrtA was found to be regulated by the availability of copper, suggesting that in contrast to other identified pyocin transporters, which invariably target iron-containing siderophores or heme, CrtA likely plays a role in copper uptake. Based on these results and previous studies on pyocins S2, S5 and G, a model for the transport of pyocins SX1/SX2 across the P. aeruginosa cell envelope is proposed (Figure 7). To illustrate this mechanism, the structure of the pyocin SX2-ImSX2 complex was predicted using AlphaFoldmultimer (30,31). The predicted structure shows a highly elongated complex with the immunity protein bound to the cytotoxic domain, features typically associated with nuclease-type bacteriocins ( Figure 7A). For pyocin SX2 the transporter and CPA binding domains consist of tandemly repeated kinked 3-helix bundles, as is observed in pyocin S5, and these lie Nterminal to an inner membrane translocation and cytotoxic domains. The predicted structure of the cytotoxic domain shares no similarity with other bacteriocin cytotoxic domains for which the structures are known. The mechanism we propose for pyocin SX1 and SX2 ( Figure 7B) is a composite of the findings of this research and current knowledge derived from a variety of bacteriocins including pyocins S2, S5, G and the nuclease-type colicins (11,13,(18)(19)(20)32,33). is likely cleaved during translocation and only the cytotoxic domain is imported into the cytoplasm, as observed for E. coli colicins (32,33). In addition, a recent study on pyocin G transportation across the inner membrane demonstrated that TonB1 is also required for pyocin G import into cytoplasm (19) (Figure 7B). TonB1 interaction may be to localize the pyocin close to the inner membrane from where the inner membrane translocation domain either interacts directly with the membrane or even with FtsH itself for transfer across the membrane and proteolytic processing (19). CrtA is a novel pyocin transporter that has not been characterized previously. Several studies have demonstrated that P. aeruginosa is able to utilize different siderophores produced by other microorganisms called xenosiderophores (34). A recent study on siderophore piracy showed that transcription of crtA gene is increased when P. aeruginosa was grown in metal-limited media, suggesting that CrtA likely plays a role in metal homeostasis (34). Furthermore, this study also showed the induction of crtA gene expression by two xenosiderophores including ferric-vibriobactin and ferric-yersiniabactin produced by xenosiderophores may share some structural characteristics with the genuine target and partially trigger the expression of crtA gene. In addition, a study on the uptake of siderophoredrug conjugates indicated that the overexpression of CrtA in P. aeruginosa PAO1 increase susceptibility to several siderophore-drug conjugates including BAL030072, MC-1 and cefiderocol (35). Thus, the results obtained from this study and from previous reports suggest that the target of CrtA is likely to be a xenosiderophore involved in Cu(II) transport.
Due to the spread of MDR P. aeruginosa, the development of novel therapeutic approaches to treat P. aeruginosa infection has become essential. This work expands the repertoire of candidate pyocins for antipseudomonal therapeutic development and generates more understanding on how these proteins target and kill P. aeruginosa.

Bacterial strains and media
P. aeruginosa and E. coli isolates were grown in LB medium (10 g/L tryptone, 10 g/L NaCl, 5 g/L yeast extract (pH 7.2). All strains used in this study are summarized in Table S2.
Strains were routinely grown at 37°C, with shaking at 180 rpm. Plasmids were propagated in when required. Bacterial strains used in this study were stored in 50% glycerol at -80°C until used.

Cloning, expression and purification of pyocin SX1 and SX2
The annotated sequences of pyocins SX1 and SX2 (Table 1)  aeruginosa P7 cell lawn on a LB plate to determine pyocin killing activity.

DNase activity assay
Immunity proteins were removed from their pyocin-immunity protein complexes by treatment with 6 M guanidine. Briefly, purified pyocin-immunity protein complex dissolved in

In vitro transcription and translation assay
The effect of pyocin SX2 on protein synthesis was assessed by an in vitro transcription-  rpsL for normalization (39,40).

Data availability statement
GenBank records ON716475 and ON716476 contain the coding sequences for pyocin SX1 and ImSX1 from the plasmid pSX1ImSX1 and for pyocin SX2 and ImSX2 from the plasmid pSX2ImSX2, respectively.

Funding
This work was supported by the Chulabhorn Royal Academy (Thailand) and a Wellcome The pyocin sensitivity was determined by size of clear zone produced by spot plate assay. All pyocins were tested at the concentration of 1 mg ml -1 . S = sensitive; R = resistant and T = tolerant. ND indicates not detected. * The number indicates the nucleotide position of the gene, + indicates upstream position from the gene; del = deletion; in = insertion; → = base substitution.  Figure S1 The killing activity of pyocins SX1/SX2 depend on TonB1 and FtsH. All pyocins were tested at the concentration of 1 mg ml -1 . Figure S2. Overlay spot plate assay of pyocins SX1/SX2 and SD2 under iron-rich and iron-limited conditions. The pyocins were serially diluted from 1000 to 0.051 µg ml -1 (3X dilution) and 5 µl were spotted on the agar plates.