«PemIK (PemK/PemI) type II TA system from Klebsiella pneumoniae clinical strains inhibits lytic phage»

Since their discovery, toxin-antitoxin systems have captivated many scientists. Recent studies demonstrated that a key role of TA systems is phage inhibition. Therefore, the aim of this study was to investigate the role of the PemIK (PemK/PemI) type II TA system in phage inhibition by its intrinsic expression in clinical strains of Klebsiella pneumoniae carrying OXA-48 carbapenemase and by induced its expression in an IPTG-inducible plasmid in a reference strain of K. pneumoniae ATCC®10031™. qRT-PCR revealed that pemK toxin in clinical strain ST16-OXA48 was induced when phage did not infect the strain, whereas when phage infection was successful pemK toxin was not induced. In addition, induced expression of the whole system did not inhibit phage infection, whereas overexpression of the pemK toxin prevented infection during the rst hours. To investigate the molecular mechanism involved in the PemK toxin-mediated inhibition of phage infection, an assay measuring metabolic activity was performed, which revealed that production of toxin PemK led to the dormancy of the bacteria. Thus, we demonstrate that the PemK/PemI TA system plays a role in phage infection, and that the action of the free toxin causes the cells to go into a dormant state resulting in inhibition of phage infections. Phage infectivity assay in solid medium was carried out by the spot test technique 44 , in the collection of clinical strain of K. pneumoniae. Briey, 200 µL of an overnight culture was mixed with 4 mL soft agar and poured onto TA agar plates. Once the soft medium had solidied, 15 µL drops of high titer phages were added to the plates. For each strain, a negative control consisting of SM buffer was included for each plate. All the determinations were performed for triplicate. The criteria used to determine the phage infectivity was: - of spot, + of clear spot and +/- of turbid


Introduction
Bacteriophages, also known as "phages", are viruses that infect bacteria. These prokaryotic viruses are considered the most abundant biological entities on Earth with 10 31 viral particles, found in all environmental niches colonized by bacteria 1,2 . Traditionally, phages have been divided based on their life cycle into either lytic or lysogenic phages. Most lytic phages, after infecting their host, use the bacterial machinery to replicate, transcribe, and/or translate their nucleic acid to nally lyse the bacterial cell, releasing many viral particles, while lysogenic phages can integrate their genome into the bacterial chromosome or follow the lytic cycle, being known as prophages or temperate phages 3 . Temperate phages are believed to be responsible for introducing a large number of genes that provide different functions to their bacterial hosts. Accordingly, the phage packaging system can act as mobile elements for horizontal gene transfer, even modulating the behavior of these bacteria, through virulence and defense genes 1 . The continuous war between phages and bacteria has led to the coevolution of both entities, so bacteria developed defenses to protect themselves from the phages while the phages in turn developed counterstrategies to evade those defenses 4 . Examples of defense mechanisms against the continuous attacks of phages are: (i) surface alterations to avoid phage adsorption, (ii) prevention of phage DNA injection, (iii) restriction of incoming DNA, (iv) acquiring phage-speci c immunity through clustered regularly interspaced short palindromic repeats (CRISPR) and (v) toxin-antitoxin (TA) systems 5 .
TA systems are widely distributed in bacterial strains 6 , located on bacterial chromosomes, plasmids, and in phages 7 . The wide range of TA systems have been classi ed into eight groups (type I-VIII) based on the nature and mechanism of action of the antitoxin 8 .These systems are encoded by adjacent genes, generally consisting of two components: a stable toxin, and an unstable antitoxin, which is degraded under stress conditions by protease systems 9 , leading to toxin activation, often resulting in reduced bacterial metabolism 8 . The most prevalent kind of TA systems is the type II TA system, where both toxin and antitoxin are proteins 10 , encoded in the same operon and co-expressed 11 , and where the antitoxin neutralizes the toxicity of the toxin by direct protein-protein interactions. Toxin targets vary, but most inhibit a central cellular process, such as translation or DNA replication 12 . Since their discovery in a plasmid in 1983 by Ogura and Sota 13 , TA systems have captivated the minds of many scientists, who have attributed to them multiple functions in cell physiology such as plasmid maintenance 13,14 which play a crucial role in the dissemination and evolution of antibiotic resistance, such as maintaining multiresistant plasmids 7,15 , bacterial persistence [16][17][18][19] , bio lm formation 20,21 , general stress response 22 , and phage inhibition 5,23−26 . However, today there are still many unanswered questions in regard to their functions in cell physiology. Recently, we proposed that the main physiological role of TA systems is the inhibition of phages 8 .
However, the relevance of TA systems to phage inhibition has been described by relatively few reports. The rst involvement was demonstrated by the type I TA system Hok/SoK from the R1 plasmid, which inhibited T4 phage infection, most likely due to activation of the toxin after global transcription reduction by the lytic phage 23 . Additional evidence of the role of TA systems was provided years later, when the type II TA systems MazF/MazE and RnlA/RnlB showed inhibition of infection by phage P1 and T4, respectively 24,25,27 . More recently the discovery of the type III TA systems ToxN/ToxI, from plasmid pECA1039 of Erwinia carotovora, revealed inhibition of phages phiA2 and phiM1 5 . Finally, the well-known abortive infection AbiEii/AbiEi system, from plasmid pNP40, that inhibits the 936 phages family, was suggested to be a type IV TA system 26 .
Based on the ndings, the aim of this work was to study the role of the type II TA system PemIK (PemK/PemI) harboring by a plasmid, previously isolated and characterized by our group 28 , by intrinsic expression of this system in clinical strains of Klebsiella pneumoniae carrying the OXA-48 carbapenemase and by induced expression of this system in an IPTG-inducible expression plasmid in a reference strain of K. pneumoniae ATCC®10031™.

Results
Isolation, propagation and electron microscopy of phages The ten phages used in this study, named according to the accepted practices 29 (Fig. 1), were obtained from residual water samples, using as host the reference strain of K. pneumoniae ATCC®10031™. Focusing on the morphology of the plaques produced by the different phages, we observed that the size of the plaques oscillated from 2.5 to 7.8 mm in diameter on the lawn of ATCC®10031™ cells (Fig. 1). In addition, all of them were surrounded by a halo that is often interpreted as an indicator of depolymerasemediated digestion of bacterial capsules 30 . Transmission electron microscopy images revealed that all the phages belonged to the order Caudovirales, i.e. tailed phages of double-stranded DNA (dsDNA). All of them present an icosahedral head, however most had a long and exible tail characteristic to the Siphoviridae family, while one phage, vB_KpnP-VAC1, presented a small tail, characteristic to the Podoviridae family (Fig. 1).

Phage genome sequencing
Genome sequencing revealed that all phages under study, available in Genbank BioProject PRJNA739095 (http://www.ncbi.nlm.nih.gov/bioproject/739095) (Table 1), were lytic Caudovirales phages, lacking lysogenic genes such as integrases, recombinase, repressor and excisionase. In addition, sequencing corroborated the TEM results by con rming that phage vB_KpnP-VAC1 was a member of the Podoviridae family, while the other nine phages were members of the Siphoviridae family. Furthermore, it allowed them to be classi ed by genus by sequence homology with the other sequences available in the NCBI database, which revealed that phage vB_KpnP-VAC1 was a member of the genus Teetrevirus, while the other phages were members of the genus Webervirus. Phage size ranged from 39,498 to 53,113 bp, and guanine-cytosine content ranged from 47.86 % to 51.63 % ( Table 1). Sequence analysis revealed that the proteins were organized into functional modules within the genome, grouping them into genes related to structure, packaging, lysis, transcription and regulation. Focusing on the lysis genes, all phages had endolysins and holin, two proteins responsible of the degradation of the bacterial cell wall during the infection of the host; however, a difference was observed in terms of spanin (protein involved in the lysis process in gram-negative host) depending on the family to which the phage belonged, i.e. the Siphoviridae phages had a unimolecular spanin (U-Spanin) while the Podoviridae phage had an heterodimer molecule spanin (I-Spanin). Finally, regarding host capsid degradation genes during virus entry, ve of the phages were found to have depolymerase (vB_KpnS-VAC2, vB_KpnS-VAC4, vB_KpnS-VAC5, vB_KpnS-VAC6 and vB_KpnS-VAC10), and three of them (vB_KpnS-VAC2, vB_KpnS-VAC4 and vB_KpnS-VAC6) had two different depolymerases. Phage infectivity assays in solid medium The phages infectivity assay in solid medium was performed using the spot test technique on the collection of clinical strains of K. pneumoniae to see the susceptibility of the strains to phages ( Fig. 2A). All strains in the collection harbor carbapenemase OXA-48 and the PemK/PemI TA system, and present different capsular types. The results obtained showed that the clinical strain ST16-OXA48 was the most susceptible strain to phages, as it was infected by 10 different phages. In contrast, the clinical strains ST405-OXA48 and ST15-OXA48 c were not infected by any of the phages assayed. Based on these results, we selected the susceptible clinical strain ST16-OXA48 and two lytic phages according to their infectivity capacity: vB_KpnP-VAC1 with no infectivity and vB_KpnS-VAC7 with high infectivity (Fig. 2B), for the remaining experiments.
Phage adsorption and One-Step growth curve The adsorption assay was performed to study the adsorption of the phage to the bacterial surface receptors with the strain and phages previously selected. Phage vB_KpnP-VAC1 showed a slight adsorption with 27.08 % of phage adsorbed at 10 min on strain ST16-OXA48. While phage vB_KpnS-VAC7 showed high percentage of adsorption with 88.69 % phage adsorbed at 6 minutes ( Fig. 2C).
Accordingly, a one-step growth curve was performed with phage vB_KpnS-VAC7 to determine the latency time and burst size, which were respectively 6 min and 15 ± 2 PFU (Fig. 2D). In the case of phage vB_KpnP-VAC1, the one-step growth curve was not performed because the phage was not able to produce a successful infection of the strain.
The relative expression of the pemK toxin gene with respect to the pemI antitoxin gene in the clinical strain ST16-OXA48 after 15 min of phage infection was signi cantly different according to the phage used for infection (Fig. 2E). Indeed, in the case of inhibited infection for the phage vB_KpnP-VAC1, the pemK toxin was found to be over-expressed (1.5-fold) with respect to the antitoxin. Whereas, in the case of infection with phage vB_KpnS-VAC7, the toxin was not found to be over-expressed relative to the antitoxin (0.7-fold). Both patterns are signi cantly different from the expression of this system in the absence of infection (control).

Induced expression of TA system
The ability of the PemK/PemI TA system to inhibit phage during induced expression of both genes was tested by monitoring infection with the 10 lytic phages at an MOI of 0.1 in the transformed strains: In order to rule out that the effect observed in the different infection curves was due solely to IPTG induction of toxin gene pemK, control of induction without infection was performed ( Fig. 4A and B). As a result, the overexpression of pemK in the strain ATCC®10031 TM /pCA24N (pemK) inhibited bacterial growth during the rst three hours until the strain regrew to double the OD 600nm at 6 h after induction, while the overexpression of the whole pemK/pemI TA system in the strain ATCC®10031 TM /pCA24N (pemK/pemI) led to normal growth, with no signi cant differences with the strain ATCC®10031 TM /pCA24N (Fig. 4C).
Enzymatic assay using the cell proliferation reagent WST-1 The assay, which measures the ubiquitous reducing agents NADH and NADPH as biochemical markers to assess the metabolic activity of the cell, revealed that the transformed strains ATCC®10031 TM /pCA24N as well as ATCC®10031 TM /pCA24N (pemK/pemI) without phage infection showed a metabolic activity reaching a OD 480nm = 0.337 and 0.178, respectively, after 2 h of induction with IPTG (Fig. 4D). In contrast, as expected, the strains ATCC®10031 TM /pCA24N and ATCC®10031 TM /pCA24N (pemK/pemI) infected by the phage vB_KpnP-VAC1 lacked metabolic activity (OD 480nm < 0.1), indicating cell death due to phage infection. However, for strain ATCC®10031 TM /pCA24N (pemK), we observed that the strain lacked metabolic activity both with and without phage infection (OD 480nm < 0.1). This data indicates that overexpression of the toxin leads to a dormant state of the cell, which prevents phage infection.

Discussion
Recent studies have revealed that the key role of TA systems in bacterial cell physiology could be related to phage inhibition. In this regard, here we demonstrate the involvement of the type II TA system PemK/PemI in the reversible inhibition of phage infection caused by cell dormancy due to the effect of the free toxin.
First, focusing on the phages studied, TEM and sequencing revealed that they were all lytic phages belonging to the order Caudovirales, i.e. dsDNA-tailed bacteriophages, most of them being members of the genus Webervirus belonging to the family Siphoviridae and one member of the genus Teetrevirus belonging to the family Podoviridae.
Regarding the intrinsic study, the phage infectivity assay in solid medium revealed that a high percentage of strains were either not infected by the phages (66 %) or were only weakly infected (15 %). Therefore, given that all clinical strains studied harbor the PemK/PemI TA system, we reasoned this type II TA system could be probably involved in the inhibition of phage infection. For this purpose, the intrinsic study focused on two lytic phages that, despite being able to bind to the bacterial surface receptors of clinical strain ST16-OXA48, showed different patterns of infectivity in the solid medium infectivity assay: To better understand the role of the PemK toxin in phage inhibition, we performed an induced expression study of the TA system using an IPTG-induced expression plasmid. The study revealed that overexpression of the pemK toxin led to inhibition of phage infection during the rst hours. In contrast, overexpression of the complete pemK/pemI TA system, with the PemK toxin blocked by the antitoxin PemI, did not confer any protection against phage infection (similar to the empty plasmid control). It is widely described that toxins of type II TA systems lead to down-regulation of cell metabolism as gene transcription is inhibited 19,28 . In this case the metabolic measurement by the enzymatic assay with the WST-1 reagent revealed that overexpression of pemK toxin led to the dormancy of bacterial cells after 2 h of induction with IPTG. However, bacteriophages are viruses that require the machinery of their bacterial hosts to replicate 34 . Therefore, as the cell is in a dormant state, the bacteriophage cannot replicate and is therefore unable to proliferate. This was observed by enumerating the PFU/mL of the ATCC®10031 TM /pCA24N (pemK) strain, where the phage counts were signi cantly lower than in the strains ATCC®10031 TM /pCA24N and ATCC®10031 TM /pCA24N (pemK/pemI). These results con rm that the presence of free toxin 33 autoregulation of the toxin through its own endoribonuclease activity 37,38 as was previously described in the homologous system MazF/MazE type II TA system in which, in the presence of stress, MazF toxin cleaves its own mRNA, thus autoregulating its expression 39 . Therefore, the bacterium returns to a metabolically active stage, allowing infection by the remaining phages present in the medium.

Conclusion
This is the rst study about the role of the plasmidic type II TA system PemK/PemI in the phage inhibition. The results obtained throughout this study demonstrate that the dormancy of bacterial cells due to the action of pemK toxin leads to inhibition of phage infection. This captivating eld is in its infancy; hence, it requires further analyses to improve the understanding of the relationship between TA systems and phage infection.

Bacterial strains and plasmid
A collection of clinical strains of K. pneumoniae, harboring the OXA-48 carbapenemase and the PemK/PemI TA system, was used to study the intrinsic role of this TA system in phage infection (  40 , inducible by IPTG and constructed by our group, harboring the complete pemK/pemI TA system and the pemK toxin alone 28 . All the strains were grown in Luria-Bertani (LB) medium, in the case of the transformed strains the media was supplemented with chloramphenicol (30 µg/ml) (LB-CM) to maintain the plasmid. Isolation and propagation of lytic phage Ten lytic phages isolated from environmental water samples were used in this study. Brie y, 50 mL of water were taken near sewage plants and kept at room temperature until processing in the laboratory. Once in the laboratory, the samples were vortexed and centrifuged 4000 × g 10 min. The supernatant was recovered and ltered through with 0.45 µm and 0.22 µm lters, to remove the cells and debris. Then, 1 mL of the ltered samples was added to 500 µL of K. pneumoniae ATCC®10031™ in 4 mL soft agar (0.5 % NaCl, 1 % tryptone and 0.4 % agar; supplemented with 1 mM CaCl2) and poured onto TA agar plates (0.5 % NaCl, 1 % tryptone and 1.5 % agar; supplemented with 1 mM CaCl 2 ); i.e., the double-layer method.
Plates were incubated at 37 ºC. Isolated plaques of different morphology (i.e., plaque size and presence of a surrounding halo) were then recovered by picking with a micropipette and stored at -70 ºC. In order to check the isolated plaques and purify them, two additional plaque assays and plaque picking steps were performed. Sequences of 300 bp paired-end reads of each isolate were assembled "de novo" with Spades V.3.15.2 43 . All assembly were annotated using Patric 3.6.9 (http://www.patricbrc.org), Blastx (http://blast.ncbi.nlm.nih.gov), HHmer (http://hmmer.org) and HHpred (https://toolkit.tuebingen.mpg.de/tools/hhpred). The determination of the family and genus of the different phages was performed by sequence homology with the phage sequences available in the NCBI database. Complete genome sequences were included in GenBank BioProject PRJNA739095

Phage infectivity assay in solid medium
Phage infectivity assay in solid medium was carried out by the spot test technique 44 Table 3. For statistical analysis of the data from these qRT-PCR experiments, a 1.5-fold cut-off value was applied to identify differentially expressed TA genes according to other work 47 . Thus, TA gene expression was considered signi cantly up or downregulated if p-value < 0.05 and fold change > 1.5 or < 1.5, respectively. Enzymatic assay using the Cell proliferation Reagent WST-1.