VIRULENCE 2020, VOL. 11, NO. 1, 315–326 https://doi.org/10.1080/21505594.2020.1748923 RESEARCH PAPER Direct interaction between RecA and a CheW-like protein is required for surface-associated motility, chemotaxis and the full virulence of Acinetobacter baumannii strain ATCC 17978 Jordi Corral, María Pérez-Varela, Jordi Barbé*, and Jesús Aranda* Departament de Genètica i Microbiologia, Facultat de Biociènces, Universitat Autònoma de Barcelona, Bellaterra, Spain ABSTRACT ARTICLE HISTORY Acinetobacter baumannii is a nosocomial pathogen that causes multi-drug resistant infections Received 22 November 2019 mainly in immunocompromised patients. Although this gram-negative species lacks flagella, it is Revised 28 February 2020 able to move over wet surfaces through a not well characterized type of movement known as Accepted 28 February 2020 surface-associated motility. In this study we demonstrate through the inactivation of the A1S_2813 KEYWORDS gene (coding a CheW-like protein) and recA (coding a DNA damage repair and recombination A. baumannii; surface- protein) that both genes are involved in the surface-associated motility and chemotaxis of associated motility; A. baumannii ATCC 17978 strain. In addition, we also point out that the lack of either RecA or chemotaxis; virulence; RecA; CheW-like proteins reduces its virulence in the Caenorhabditis elegans and the Galleria mellonella CheW-like protein animal models. Furthermore, we show through co-immunoprecipitation assays that the CheW-like protein and RecA interact and that this interaction is abolished by the introduction of the mutation S97A in one of the domains of CheW-like protein that is structurally conserved in Salmonella enterica and necessary for the RecA-CheW interaction in this bacterial species. Finally, we show that the replacement of the wild-type CheW-like protein by that presenting the S97A mutation impairs surface-associated motility, chemotaxis and virulence of A. baumannii strain ATCC 17978. Introduction The motility of bacteria is precisely regulated by Acinetobacter baumannii is an opportunistic pathogen a chemotaxis system that enables the cells to sense specific whose abilities to develop antimicrobial resistance and chemicals and then direct movement toward or away from to survive in hospital environments are responsible for them [11]. A recent report showed that inactivation of the its association with nosocomial infections worldwide putative chemotaxis CheA/Y-like regulator A1S_2811 in [1]. In many different bacterial species, virulence is A. baumannii strain ATCC 17978 impairs the surface- close related to motility [2]. This is the case in associated motility of the bacterium [12]. In Pseudomonas A. baumannii [3], in which two types of motility have aeruginosa, another nosocomial pathogen closely related to been described: (i) twitching motility, a coordinated Acinetobacter spp., CheA forms a complex with the che- multicellular movement driven by the extension, motactic receptor through the coupling protein CheW. attachment and retraction of type IV pili, and (ii) sur- This allows CheA to transfer its phosphate group to face-associated motility, an appendage-independent CheY, which then interacts with the flagellar motor to form of movement [4]. Despite direct connection switch the direction of flagellar rotation [11]. Although between twitching motility in virulence has not been Acinetobacter spp. lack flagella, genetic studies in observed for A. baumannii [2], several reports confirm P. aeruginosa, which possesses both flagella and pili, have the importance of surface-associated motility in the provided insights into a putative pili-mediated chemotaxis virulence of A. baumannii [3,5–7]. This type of motility pathway, including the homologous components ulti- has been associated to the synthesis of 1,3-diaminopro- mately involved in pili extension/retraction [11]. pane [5], quorum sensing [8], blue light [9], lipooligo- In A. baumannii strain ATCC 17978, homologues to saccharide production [10], and at least to five of the practically all genes encoding putative pili-mediated che- six known efflux pump superfamilies [7]. motaxis proteins have been identified in the operon CONTACT Jesús Aranda jesus.aranda@uab.cat *Co-senior authors The supplementary data for this article can be accessed here. © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 316 J. CORRAL ET AL. A1S_2811-2815, which also includes the gene A1S_2813, plasmid (Invitrogen), transformed in E. coli DH5α annotated as pilI, encoding a CheW-like protein [11,12]. and selected on kanamycin-containing LB plates. The We previously demonstrated that in Salmonella enterica purified plasmid was then introduced into WT the interaction of CheW with RecA, a protein involved in A. baumannii by electroporation and the resulting DNA damage repair and recombination, is necessary for transformants were selected on kanamycin-containing swarming motility, a form of flagella-dependent move- plates. Recombinant clones were confirmed by sequen- ment across surfaces that involves chemoreceptor cluster cing (Macrogen) of the PCR products obtained by formation [13]. In the present work, we show that the using the appropriate primers, indicated in direct interaction of RecA and CheW-like proteins is also Supplementary Table 1. Ten passages without selective necessary for the surface-associated motility, chemotaxis pressure and colony resuspension were carried out and full virulence of A. baumannii strain ATCC 17978. every 24 h to determine the stability of the knockout, by colony counting on LB plates with and without kanamycin at the last passage. For complementation, Materials and methods the target gene was cloned into the pVRL1 vector using Bacterial strains, plasmids and growth conditions the appropriate primers (Supplementary Table 1), which included XhoI-NcoI restriction sequences [15]. The bacterial strains and plasmids used in this study are The recombinant plasmid was introduced by electro- listed in Table 1. A. baumannii ATCC 17978 and poration, firstly into E. col DH5α and then, once the Escherichia coli (strains DH5α, BL21 and OP50) were correct construct had been verified by PCR and sequen- grown at 37ºC in Luria-Bertani (LB) medium [14] with cing (Macrogen), into the corresponding A. baumannii shaking at 180 rpm. When necessary, the medium was knockout. The complemented mutants were selected on supplemented with kanamycin (50 mg/L), gentamicin kanamycin- and gentamicin-containing plates. (20 mg/L for E. coli or 40 mg/L for A. baumannii), ampicillin (50 mg/L) or chloramphenicol (34 mg/L). The plasmids pCR-BluntII-TOPO (Invitrogen, catalog Surface-associated motility assays number 451245), pVRL1 [15] and pUA1108 [16] were The assays were conducted on modified fresh LB agar used for mutant construction, complementation and plates (1% tryptone, 0.5% yeast extract, 0.5% NaCl, overexpression, respectively (Table 1). The growth of 0.5% glucose, and 0.5% Difco agar), prepared on A. baumannii was monitored using bacterial cultures in the day of the experiment. Bacterial cultures were LB broth inoculated with an overnight culture at grown to early stationary phase and a 5-µl droplet of a dilution of 1:100 and then incubated at 37ºC with the culture was placed in the center of the plate. The shaking at 180 rpm. The optical density at 600 nm inoculated plates were incubated at 30°C for 16–20 h (OD600) was measured hourly. until the WT strain had grown to reach the plate border. All assays were carried out at least three times Accession numbers in independent experiments, each in triplicate. A ChemiDocTM XRS+ system (Bio-Rad) was used to The sequences of RecA and CheW-like proteins of obtain representative images. A. baumannii strain ATCC 17978 used in this study were obtained from NCBI (National Center for Biotechnology Information), with the following accession numbers: Chemotaxis assays APP32573 (RecA; old annotation A1S_1962, current annota- A. baumannii movement toward a chemoattractant was tion AUO97_17690) and APP29846 (CheW-like protein; old assayed as previously described [18], with slight mod- annotation A1S_2813, current annotation AUO97_03055). ifications. Briefly, autoclaved, 3-cm-long capillary tubes (Microcaps, Drummond Scientific Co., catalog number 1-000-0010) were half-filled with chemotaxis buffer A. baumannii knockout construction and (CB; 10 mM potassium phosphate, 0.1 mM EDTA, complementation 1 mM MgSO4) solidified using agarose (3% w/v) with Gene inactivation was carried out as previously or without 2% methanol as chemoattractant and then described [17]. Briefly, an internal fragment of the completely filled with CB without agarose. Bacterial target gene from wild-type (WT) A. baumannii strain suspensions were prepared from cultures grown to ATCC 17978 was PCR-amplified using the appro- mid-exponential phase (OD600 of 0.5) in Tryptone priated primers (Supplementary Table 1). The resulting broth (1% tryptone and 0.5% NaCl) and then washed fragment was cloned into the pCR-BluntII-TOPO and resuspended in CB to an OD600 of 0.1. Chemotaxis VIRULENCE 317 Table 1. Bacterial strains and plasmids used in this work. Source or Strains Relevant characteristics reference Escherichia coli strains DH5α E. coli supE4 ΔlacU169 (φ80 ΔlacZ ΔM15) hsdR17, recA1, endA1, gyrA96, thi-1, relA1 Clontech BL21 (DE3) pLysS E. coli F− dcm ompT lon hsdS(rB−mB−) galλ(DE3) carrying pLysS plasmid, CmR Stratagene Acinetobacter baumannii strains ATCC 17978 Wild-type ATCC recA Strain ATCC 17978 with recA::Km, KmR [19] A1S_2813 Strain ATCC 17978 with A1S_2813::pCR-BluntII-TOPO disruption, KmR, ZeoR This work Plasmids pCR-BluntII-TOPO Cloning vector, KmR, ZeoR Invitrogen pVRL1 Complementation vector, GmR [9] pUA1108 pGEX 4 T-1 derivative plasmid without the GST fusion tag, carrying the Ptac promoter and the [10] q R lacI gene; used as overexpression vector, Ap pUA1108- recA-FLAG pUA1108 plasmid carrying ATCC 17978 recA with FLAG tag, ApR This work pUA1108-A1S_2813-6× His pUA1108 plasmid carrying ATCC 17978 A1S_2813 with 6× His tag, ApR This work pUA1108-A1S_2813#S97A-6× His pUA1108 plasmid carrying ATCC 17978 A1S_2813 Ser97Ala with 6× His tag, ApR This work pUA1108-A1S_2813#I121A-6× His pUA1108 plasmid carrying ATCC 17978 A1S_2813 Ile121Ala with 6× His tag, ApR This work R pVRL1- recA pVRL1 plasmid carrying ATCC 17978 recA, Gm This work R pVRL1-A1S_2813 pVRL1 plasmid carrying ATCC 17978 A1S_2813, Gm This work R pVRL1-A1S_2813#S97A pVRL1 plasmid carrying ATCC 17978 A1S_2813 Ser97Ala, Gm This work R pVRL1-A1S_2813#I121A pVRL1 plasmid carrying ATCC 17978 A1S_2813 Ile121Ala, Gm This work CmR, KmR, ZeoR, GmR, ApR stand for resistance to chloramphenicol, kanamycin, zeocin, gentamicin and ampicillin, respectively. chambers were prepared in a six-well plates (Sardstedt), count assay, performed as previously described [19]. with one plate for each bacterial suspension. A capillary Briefly, the low-virulence strain E. coli OP50, grown as containing either the chemoattractant or CB alone was a lawn on NGM (Nematode Growth Medium) plates, placed in each well so that it soaked in 2 ml of the was used to feed C. elegans strain N2. Physiological appropriate bacterial suspension. After a 2-h incubation synchronization was achieved by hatching the nematode in the chemotactic chambers at 30ºC, the capillaries eggs in M9 medium (0.02 M KH2PO4, 0.04 M Na2HPO4, were washed twice to remove any externally attached 0.08 M NaCl, and 0.001 M MgSO4) and then arresting cells, broken open and their contents poured into capil- the growth of the worms in the first larval (L1) stage lary tubes containing saline solution. Serial dilutions of overnight at 16°C. The L1 larvae were then cultivated to the suspensions were plated in LB medium to deter- the last larval stage (L4) on NGM plates seeded with the mine the number of CFU (colony-forming units) per appropriate bacterial strain. Each L4 worm was placed ml obtained under each condition and then normalized on a PGS (peptone glucose sorbitol) plate with the between capillaries with and without chemoattractant. corresponding A. baumannii strain and incubated at The chemotaxis experiments were carried out in at least 25°C. During each of the following 3 days, the worms 9 independent experiments. were transferred to new plates seeded with the same bacterial strain. Nematode fertility was determined using a stereomicroscope (Olympus SZ51) to count the Transmission electron microscopy (TEM) progeny daily beginning 48 h after the removal of the Colonies of each strain from the motility plates were parent worm. Six independent replicates were estab- gently resuspended in 50 µl of 0.1 M phosphate buffer lished for each strain and each fertility assay was per- (PB). A 10-μl aliquot of each one was deposited onto formed in triplicate. carbon-coated film meshes supported by standard copper TEM grids and incubated for 30 min at room tempera- Galleria killing assays ture. The samples were then fixed for 10 min in a glutar- aldehyde solution (1% final concentration) prepared in The virulence of A. baumannii strains was also deter- PB, washed twice with PB and dried. Thereafter, they mined using a modified version of the Galleria mello- were negatively stained with 2% uranyl acetate for 1 min nella (wax moth) model described by Peleg et al [20]. and observed on a JEM-1400 transmission electron Briefly, ten caterpillars, each with an approximate microscope (JEOL Inc., Peabody, MA, USA). weight of 300 mg, were inoculated via the hemocoel with 10 μl of an exponentially growing (OD600 = 0.5) strain of A. baumannii, previously resuspended in Nematode fertility assays phosphate-buffered saline (PBS). The concentration of The virulence of A. baumannii strains was determined the inoculum was confirmed by colony counts on LB based on a Caenorhabditis elegans nematode progeny agar. As a negative control, 10 μl of PBS was injected 318 J. CORRAL ET AL. into the same number of caterpillars. The caterpillars appropriate oligonucleotides (Supplementary Table 1) were incubated at 37°C in dark, and their survival was and the Q5® Site-Directed Mutagenesis kit (BioLabs, checked every 12 h for a total of 96 h. All G. mellonella catalog number E0554S) according to the manufacturer’s killing experiments were performed at least twice. To instructions. Sequence sites to mutagenize were selected assess plasmid stability of complemented mutants used based on comparisons of the predicted secondary struc- in this animal model, at least 5 worms for each strain tures of the A. baumannii CheW-like and S. enterica were infected with sublethal doses of the corresponding CheW proteins (accession numbers: ABO13219 and complemented mutant (approximately 1 × 104 CFU/ NP_460877, respectively), performed with the SABLE worm) and incubated at 37ºC for 96 h. The worms were program (http://sable.cchmc.org/). The pUA1108 plas- then washed with ethanol 70% and sterile water. mids containing genes coding either RecA-FLAG or Afterward, the hemolymph was extracted and serial A1S_2813-6× His and the derivate site-specific muta- dilutions were plated on solid media containing kana- genic proteins were introduced by electroporation into mycin or kanamycin plus gentamycin. E. coli BL21 (DE3) pLysS (Promega) and selected on ampicillin- and chloramphenicol-containing plates. Additionally, all A1S_2813 genes carrying a specific mutagenized sequence were cloned from the pUA1108 Co-immunoprecipitation assays plasmid into the pVRL1 vector using the appropriate The recA and A1S_2813 genes were cloned into the primers containing XhoI-NotI restriction sequences pUA1108 overexpression vector after their amplification (Supplementary Table 1) and introduced into the by PCR using the appropriate primers (Supplementary A. baumannii A1S_2813 mutant strain. The transfor- Table 1). Recombinant plasmids were introduced by elec- mants were selected on kanamycin- and gentamicin- troporation of E. coli BL21, selected on ampicillin- containing plates. All constructs were verified by sequen- containing plates and confirmed by sequencing cing (Macrogen). (Macrogen). Interaction of the RecA and A1S_2813 pro- teins (carrying FLAG and 6× His tag sequences, respec- tively) was determined as described by Mayola et al [16]., Statistical analysis with slight modifications. Briefly, pellets obtained from Data obtained in the chemotaxis assays and the nema- IPTG-induced cultures were sonicated and the superna- tode model of virulence are presented as the mean ± tants collected after centrifugation. The cell lysates were standard deviation and were analyzed using a two- immuneseparated using Pure Proteome Protein tailed, one-way analysis of variance (ANOVA), fol- A magnetic beads (Millipore, catalog number lowed by the Tukey test for post-hoc multiple group LSKMAGA10) coated with the appropriate primary anti- comparisons. In the G. mellonella killing assay, survival bodies: rabbit anti-FLAG (Sigma-Aldrich, catalog number curves were plotted using the Kaplan-Meier method F7425) or mouse anti-6× His IgGs (Roche, catalog num- and differences in survival were calculated using the ber 11922416001), according to the manufacturer’s log-rank test. In all tests, a P value < 0.05 was consid- instructions. The samples were separated by SDS-PAGE ered to indicate statistical significance. on a 15% polyacrylamide gel and analyzed by western blotting. Horseradish peroxidase (HRP)-coupled goat anti-rabbit (Yo proteins, catalog number 464) or anti- Results mouse IgGs antibodies (Acris, catalog number RecA and CheW-like proteins are involved in the R1253HRP) were used as secondary antibodies and the surface-associated motility and chemotaxis of immune complexes were visualized by the addition of the A. baumannii strain ATCC 17978. Luminata ForteTM western HRP substrate (Millipore, catalog number WBLUF0100). Representative membrane After 24 h of incubation on motility plates, the recA images were obtained using a ChemiDocTM XRS+ sys- and the A1S_2813 A. baumannii knock-out mutants tem (Bio-Rad). (Table 1) did not show surface-associated motility, unlike the WT parental strain A. baumannii ATCC 17978 (Figure 1). To link the impaired motility to the Site-specific mutagenesis of CheW-like protein lack of either the recA or the A1S_2813 gene, the The ORF of A. baumannii CheW-like protein was respective WT genes were introduced into the corre- cloned into the overexpression plasmid pUA1108 and sponding knockout-mutant strains. The resulting com- specific nucleotides were then replaced using the plementation restored surface-associated motility by VIRULENCE 319 Figure 1. Representative image of the surface-associated motility of the indicated A. baumannii strains. both mutants to a level similar to that observed in the The lack of either RecA or cheW-like proteins WT strain (Figure 1). By contrast, introduction of the reduces the virulence of A. baumannii strain ATCC empty expression vector pVRL1 did not modify the 17978 in two different animal models. behavior of any of the studied strains (Figure 1). Since In the C. elegans fertility assay, worms fed either the the growth curves of the recA and A1S_2813 knockout recA or the A1S_2813 mutant exhibited an increased mutants (both presenting a stability of >90%) cultured number of progeny compared to worms fed the WT in LB medium were comparable to that of the WT strain (Figure 4(a)), indicating a significant reduction strain (Fig. S1), the altered surface-associated motility in the pathogenicity of the mutants. When the worms of the mutants cannot be attributed to reductions in the were fed the complemented strains (presenting bacterial growth rate. To determine whether the lack of a plasmid stability of 109 ± 28% and 94 ± 10% for the RecA and/or CheW-like proteins affects bacterial che- complemented recA and A1S_2813 mutants, respec- motaxis, a fishing method was used in which methanol tively), the number of progeny was less than obtained served as the chemoattractant. The results showed an with the respective mutants and similar to that approximately 3-fold increase in bacterial counts of the obtained when the WT strain was used as nourishment WT strain in capillaries with methanol compared to the (Figure 4(a)). Accordingly, in the G. mellonella killing counts obtained in capillaries containing only CB (con- assay, approximately 50% of the worms inoculated with trol), whereas in the case of both the recA and either the recA or the A1S_2813 mutant survived dur- A1S_2813 knockout mutants there were no significant ing the 96-h experiment, whereas practically all larvae differences of theses mutants in cell migration toward inoculated with the WT or the complemented strains methanol compared to the control capillaries contain- died within the first 40 h (Figure 4(b)). ing only CB, showing a significant difference in the fold-change cell count in both mutants with respect to that observed in the WT parental strain (Figure 2). In vitro interaction between the A. baumannii Furthermore, the complementation of either mutant strain ATCC 17978 RecA-FLAG and CheW-like restored the WT chemotaxis phenotype (Figure 2). In -6× His protein addition, on the TEM images, both recA and A1S_2813 knockouts lacked the pilus-like structure typically pre- In silico analysis revealed that while S. enterica CheW and sent on cells of the ATCC 17978 WT strain, but in each the A. baumannii CheW-like protein had an identity of mutant complementation resulted in its recovery only 24%, their secondary structures were highly similar (Figure 3 and Fig. S2). (Figure 5(a)). In S. enterica, the RecA and CheW proteins 320 J. CORRAL ET AL. in the recovery of the opposite tagged protein (RecA- FLAG with anti-6× His and A1S_2813-6× His with anti- FLAG) and its subsequent detection in a western blot assay (Figure 5(b)). Together, these findings clearly demonstrated that RecA-FLAG and the CheW-like -6× His protein of A. baumannii specifically interact in vitro. To examine whether the CheW and RecA interaction domains in S. enterica are conserved in A. baumannii, a CheW-like-6× His derivate-tagged protein was generated by the substitution of residue Ser97Ala. This position of the CheW-like-6× His protein of A. baumannii corresponds exactly to the Asp that is the second-to-last residue of the Turn-6 region of S. enterica CheW (Figure 6(a)). In CheW, replacement of the Asp by an alanine, a non-reactive amino-acid residue [21], abolishes its interaction with RecA [13]. As a negative control, we also generated, a CheW-like-6× His derivate protein containing the sub- stitution Ile121Ala at a site corresponding to a region of the S. enterica CheW protein that is not involved in the inter- action with RecA (Figure 6(a)). It is worth noting that appropriate expression of all proteins used in this work was confirmed through western blot assay (Fig. S4). Co- immunoprecipitation assays showed that the RecA-FLAG protein was not able to bind the CheW-like-6× His protein containing the substitution Ser97Ala, whereas the derivate Ile121Ala was recovered (Figure 6(b)). These results indi- Figure 2. The fold-change between viable bacteria (CFU) of the cate that in the CheW-like-6× His protein Ser97 but not indicated strains, calculated as the number of CFU in capillaries Ile121 is essential for the interaction with RecA in containing chemoattractant (2% methanol) divided by the A. baumannii. Thus, at least one of the essential domains number in control capillaries containing only CB (chemotaxis involved in the interaction of CheW with RecA in buffer). Error bars represent the standard deviation (SD) of the S. enterica is also required for the interaction of the corre- means. *P < 0.05 in a comparison with the A. baumannii parental strain (ATCC 17978) and based on at least nine inde- sponding homologues in A. baumannii. pendent experiments. In vivo interaction between the A. baumannii ATCC 17978 RecA and CheW-like protein interact [16]. To determine whether this was also the case for the RecA and CheW-like (A1S_2813) proteins of To determine whether the RecA-A1S_2813 interaction is A. baumannii, we conducted an in silico modeling of involved in surface-associated motility and virulence, interaction (Fig. S3) and a co-immunoprecipitation assay mutant A1S_2813 genes encoding the proteins that using the overexpressed tagged proteins RecA-FLAG and included the Ser97Ala or Ile121Ala mutations were cloned A1S_2813-6× His (Figure 5(b)). When the tagged proteins into the pVRL1 plasmid and introduced into the were mixed with their corresponding antibody-coated A1S_2813 mutant strain. Motility and chemotaxis assays beads (RecA-FLAG with anti-FLAG or A1S_2813- showed that expression in the A1S_2813 knockout of the 6× His with anti-6× His), the matching protein was mutagenized A1S_2813 gene whose product included the detected in the recovered supernatant (Figure 5(b)). Ile121Ala substitution completely restored the parental When the tagged proteins were mixed with the opposite phenotypes (Figure 7). By contrast, the A1S_2813 knock- antibody-coated beads (RecA-FLAG with anti-6× His or out expressing the A1S_2813 protein containing the A1S_2813-6× His with anti-FLAG), no proteins were Ser97Ala mutation and thus unable to interact with detected. These results confirmed the specificity of the RecA in vitro, did not restore either surface-associated assay (Figure 5(b)). We then added a mixture of the motility or chemotaxis (Figure 7). Finally, the pathogeni- tagged proteins (RecA-FLAG and A1S_2813-6× His) to city of both strains was tested in the C. elegans fertility each suspension of antibody-coated beads, which resulted and G. mellonella killing assays (Figure 8). As expected, VIRULENCE 321 Figure 3. Transmission electron microscopy of the indicated A. baumannii strains. The images were obtained at 4,000 × magnification. Figure 4. (a) Representative results of a C. elegans fertility assay using the indicated strains of A. baumannii. *P < 0.05 compared to the parental ATCC 17978 strain. E. coli strain OP50 was included as a low virulent control. (b) Representative G. mellonella killing assay of the specified strains. Larvae (n = 10 per group) were inoculated with either ~106 CFU of the indicated strain or PBS (as a negative control). Error bars represent the SDs of the means. *P < 0.05 in a comparison with the A. baumannii parental strain (ATCC 17978). the virulence of the cheW-like mutant expressing the virulence of the A1S_2813 mutant strain was fully A1S_2813 gene encoding Ser97Ala mutant protein was restored by expression of the gene encoding a CheW- impaired in both animal models and was similar to the like protein containing the Ile121Ala substitution, as A1S_2813 knockout strain (Figure 8). However, the occurred when the A1S_2813 strain was complemented 322 J. CORRAL ET AL. Figure 5. (a) Predicted secondary structure of the S. enterica CheW and A. baumannii CheW-like proteins. The green boxes indicate the S. enterica CheW domains involved in the interaction with RecA. (b) Results of a co-immunoprecipitation assay between the A. baumannii RecA-FLAG and A1S_2813-6× His proteins. The supernatants were separated by SDS-PAGE and assessed by western blotting. The image is representative of three independent experiments using three different lysates for each protein. The presence (+) or absence (−) of the RecA-FLAG or A1S_2813-6× His proteins in the corresponding lysate mixture is indicated, as is that of the antibody-coated beads used in each mixture. The RecA-FLAG and A1S_2813-6× His bands detected on a western blot are shown. M: molecular mass marker. with the corresponding WT gene (Figure 8). These results solid medium (twitching motility), both motilities or demonstrated that the interaction between the RecA and neither [22]. Strain ATCC 17978, one of the most well CheW-like proteins in vivo is necessary for the surface- studied clinical isolates of A. baumannii, is capable of associated motility and full virulence of A. baumannii. surface-associated motility but not twitching [22]. In this strain, surface-associated motility was recently linked to the CheA/Y-like protein A1S_2811, Discussion a component of a hypothetical chemotactic signal The surface-associated motility characteristic of several transduction system whose inactivation causes A. baumannii clinical isolates allows their movement a significant decrease in this type of movement [12]. over wet surfaces independent of the bacterial expression Our group recently reported that in S. enterica the of type IV pili [2]. In A. baumannii, a non-flagellated interaction between the chemotaxis system and the bacteria, the molecular and genetic basis of surface- RecA protein is necessary for the swarming motility associated motility is unclear and the terminology used of this species [13]. Specifically, RecA interacts with in the literature to describe it is often confusing. For CheW to allow chemoreceptor cluster formation, example, in several publications surface-associated moti- which is ultimately involved in flagellar rotation [13]. lity is referred to as swarming, which is, however, fla- In previous work, we constructed an A. baumannii gella-dependent. The surface-associated motility of mutant lacking RecA, the major enzyme involved in A. baumannii seems to be closely related to the patho- recombination, to show the protein’s involvement in genicity of the bacterium, as an increase in motility is virulence as well [23]. The results of the present study associated with an increase in virulence [6] whereas demonstrate that inactivation of the A. baumannii recA A. baumannii mutants unable to conduct surface- gene also impairs motility (Figure 1). Since a putative associated motility exhibit an attenuated virulence [3,7]. operon involved in A. baumannii chemotaxis has been The motility phenotypes in A. baumannii are diverse recently described, we proceeded to inactivate its A1S_ and include motility on semi-solid media (surface- 2813 gene (annotated as pilI in ATCC 17978 strain associated motility), migration conducted by type IV [12]), which encodes a protein homologous to the pili, such as occurs at the medium-plastic interface of CheW-like protein of P. aeruginosa [11]. As in the VIRULENCE 323 Figure 6. (a) Detail of a selected conserved region in the predicted secondary structure of the S. enterica CheW and A. baumannii CheW-like proteins. The large and small green boxes indicate, respectively, a known domain and a residue of S. enterica CheW protein involved in the interaction with RecA. The black circles indicate the subsequently mutagenized residues from the A. baumannii CheW-like protein. (b) Co-immunoprecipitation of the RecA-FLAG and A1S_2813-6× His proteins and the derivative A1S_2813-6× His site-specific mutagenized proteins. The supernatants were separated by SDS-PAGE and assessed by western blotting. The images are representative of those from three independent assays. The presence of the RecA-FLAG and A1S_2813- 6× His proteins (WT) and the absence of (-) or residue change (Ser97Ala or Ile121Ala) in A1S_2813-6× His in the corresponding mixtures are indicated. The western blot shows the RecA-FLAG and A1S_2813-6× His protein bands revealed following incubation of the lysates with anti-FLAG coated beads. M: molecular mass marker. Figure 7. (a) Representative image of the surface-associated motility of the indicated A. baumannii strains. (b) The fold-change in the number of viable bacteria (CFU) of the indicated strains, based on the counts in capillaries containing chemoattractant (2% methanol) divided by those in control capillaries containing only CB. Error bars represent the SDs of the means. *P < 0.05 in a comparison with the A. baumannii parental strain (ATCC 17978) as determined in at least nine independent experiments. 324 J. CORRAL ET AL. Figure 8. (a) Representative results of a C. elegans fertility assay with the indicated strains. *P < 0.05 in a comparison with the A. baumannii parental strain (ATCC 17,978). E. coli strain OP50 was included as a low virulent control. (b) Representative results of a G. mellonella killing assay of the specified strains. Larvae (n = 10 per group) were inoculated with ~106 CFU of the indicated strain or PBS (as a negative control). Error bars represent the SDs of the means. *P < 0.05 in a comparison with the A. baumannii parental strain (ATCC 17978). case of the recA mutant, the absence of the gene encod- of the cell surface appendages necessary for type-IV-pili- ing the CheW-like protein of A. baumannii resulted in independent motility [26]. By contrast, in A. baumannii the abolishment of surface-associated motility (Figure strain ATCC 17978 only one chemosensory system (the 1), indicating that both RecA and the hypothetical A1S_2811-2815 operon) has been identified so far. It chemotaxis operon A1S_2811-2815 are involved in the should be implicated in the chemotactic response surface-associated motility of A. baumannii strain through the regulation of appendages that are distinct ATCC 17978. from flagella and type IV pili and/or through appen- The A1S_2811 gene (encoding a CheA/Y-like protein) dage-independent mechanisms. and the A1S_2813 gene (encoding a CheW-like protein) In P. aeruginosa, a pili-mediated chemotaxis pathway belong to the same hypothetical chemotactic operon. has been described in which the CheW-like protein is the However, while the importance of this operon in sur- coupling protein between the chemotactic receptor and the face-associated motility of A. baumannii has been CheA/Y-like two component system. The loss of the unequivocally demonstrated, to our knowledge, its role CheW-like protein of A. baumannii strain ATCC 17978 in the chemotaxis of this bacterium was previously reduces surface-associated motility as well as the number of unknown. The present work is the first to demonstrate pilus-like structures and the amount of cell-surface extra- that both the CheW-like protein and RecA are involved polymeric substances [11,12]. The present study similarly in the chemotaxis of A. baumannii (Figure 2). The showed a dramatic reduction of both pilus-like structures molecular machinery that controls chemotaxis in bac- and secreted substances in the recA and cheW-like mutant teria has been grouped in three different functional derivatives of A. baumannii strain ATCC 17978 (Figure 3). classes: flagella, type IV pili, and alternative cellular This finding is consistent with our previous work, in which functions, including those that are acquired by horizon- we showed that the inactivation of transporters belonging tal gene transfer or duplication [24]. For instance, to practically all families of A. baumannii efflux pumps Myxococcus xanthus, a gram-negative soil bacterium drastically reduces the surface-associated motility of the with a complex life cycle, lacks flagella but makes use bacterium [7]. The reduction might be due to the decreased of two other types of motility, one dependent and the translocation of molecules related to this type of motility, other independent of type IV pili. It also has at least such as surfactants or quorum-sensing signals [5,8,10]. eight chemosensory systems, most of which play various Accordingly, one of the most strongly down-regulated roles in controlling either one or the other type of genes of an A. baumannii cheA/Y-like mutant belongs to motility [25]. Furthermore, M. xanthus possesses the operon A1S_0112-0119, involved in the production of a CheW homologue that is required for the biogenesis acinetin 505 and its extrusion via an RND efflux pump VIRULENCE 325 [12,27]. Acinetin 505 is a lipopeptide that may act as In S. enterica the interaction between RecA and a surfactant in the promotion of surface-associated motility CheW plays a key role in swarming motility [13]. [27]. Specifically, the adaptor CheW couples the chemorecep- The impact of CheW-like and RecA proteins on the tor protein to CheA, a histidine kinase that transfers the virulence of A. baumannii was tested in two different signal to the CheY response regulator, which acts on the animal models, based on C. elegans and G. mellonella. flagellar motor by switching flagellar rotation [28]. RecA Both showed the significantly decreased virulence of contributes to altering the distribution of CheW at the the recA and cheW-like knockout mutants (Figure 4). cell poles, at sites where flagella are distributed, to mod- The results were in agreement with those obtained ulate swarming motility [13]. In P. aeruginosa, a pili- using a RecA mutant in a mouse model [23]. mediated chemotaxis pathway has been proposed that Together, these studies support a clear relationship contains homologous components ultimately involved in between the surface-associated motility and the patho- pili extension/retraction [11]. Homologies of most of genesis of A. baumannii. Remarkably, all ATCC 17978 these proteins are also present in A. baumannii strain mutant derivatives analyzed so far presenting ATCC 17978, including a CheA/Y-like protein involved a decreased surface-associated motility also show in surface-associated motility [12] and, as demonstrated a reduction in their virulence [3,7]. On the other herein, the CheW-like protein. Thus far, a role for RecA hand, the A. baumannii strain lacking the transcrip- in chemotaxis and motility has been reported only in tional repressor H-NS is not only hypermotile but also S. enterica, as described above; however, our results shows an increased virulence [6]. In both types of evidence the importance of the protein in the surface- mutants, efflux pumps are involved in the observed associated motility of A. baumannii strain ATCC 17978. phenotype, albeit in an opposite manner. In mutants Since the latter bacterium lacks flagella and mutants lacking surface-associated motility, the expression of deficient in functional type IV pili are still able to con- transporters involved in the extrusion of components duct surface-associated motility [29], then the interac- relevant to this type of motility are absent or reduced tion between the A. baumannii RecA and CheW-like [3,7]. In the hypermotile mutant, the expression of proteins must control appendages different from either genes encoding quorum-sensing signals, a novel type flagella or type IV pili. Alternatively, or in addition, the I pilus and several efflux pumps, including those of the two proteins may control the expression of genes related aforementioned A1S_0112-0119 operon, is significa- to surface-associated motility, perhaps through a non- tively increased [6]. traditional chemosensory system such as that described In flagella-bearing S. enterica, the interaction of its above for M. xanthus [25]. RecA and CheW-like homologues is necessary for che- In summary, our study reveals a novel role for the motaxis and swarming motility [13]. We therefore asked RecA protein in A. baumannii strain ATCC 17978 and whether a similar interaction was required in demonstrates that its interaction with the CheW-like A. baumannii ATCC 17978, in which the two proteins protein of the chemotactic system is necessary for sur- were shown to be necessary for surface-associated moti- face-associated motility, chemotaxis and the full viru- lity, chemotaxis and virulence. Co-immunoprecipitation lence of the bacterium. assays using the A. baumannii RecA and CheW-like proteins revealed the highly conserved secondary struc- ture of the latter with respect to its S. enterica homo- Acknowledgments logue CheW (Figure 5(a)) and the specific interaction of A. baumannii RecA and CheW-like proteins (Figure 5 We thank Joan Ruiz (UAB) and Susana Escribano (UAB) for their excellent technical assistance. We also thank Paolo (b)). In addition, interaction of the CheW-like protein Visca (University Roma Tre, Rome, Italy) for kindly provid- with A. baumannii RecA was abolished by the introduc- ing plasmid pVRL1. This study was supported by a grant tion of a mutation into one of the former’s protein (BIO2016-77011-R) from the Ministerio de Economía domains that is structurally conserved and whose homo- y Competitividad. The funders had no role in the study logue in S. enterica is involved in the RecA-CheW inter- design, data collection or data interpretation, nor the decision action (Figure 6(a,b)). The mutation that abolished this to submit the work for publication. J.A. is a Serra Húnter Fellow, Generalitat de Catalunya, Barcelona, Spain. interaction was the same that led to impaired surface- associated motility, chemotaxis and virulence (Figures 7 and 8), which clearly demonstrated that the interaction between RecA and CheW-like proteins is necessary for Disclosure statement these functions in A. baumannii strain ATCC 17978. No potential conflict of interest was reported by the authors. 326 J. CORRAL ET AL. Funding [15] Lucidi M, Runci F, Rampioni G, et al. New shuttle vectors for gene cloning and expression in This work was supported by the Ministerio de Economía multidrug-resistant Acinetobacter species. Antimicrob y Competitividad [BIO2016-77011-R]. Agents Chemother. 2018;62:e02480–17. doi:10.1128/ AAC.02480-17. [16] Mayola A, Irazoki O, Martínez IA, et al. RecA protein References plays a role in the chemotactic response and chemor- eceptor clustering of Salmonella enterica. PLoS One. [1] Wong D, Nielsen TB, Bonomo RA, et al. Clinical and 2014;9(8):e105578. pathophysiological overview of Acinetobacter [17] Aranda J, Poza M, Pardo BG, et al. A rapid and simple Infections: a century of challenges. Clin Microbiol method for constructing stable mutants of Acinetobacter Rev. 2017;30(1):409–447. baumannii. BMC Microbiol. 2010;10(1):279. [2] Harding CM, Hennon SW, Feldman MF. Uncovering [18] Tola YH, Fujitani Y, Tani A. Bacteria with natural the mechanisms of Acinetobacter baumannii virulence. chemotaxis towards methanol revealed by chemotaxis Nat Rev Microbiol. 2018;16(2):91–102. fishing technique. Biosci Biotechnol Biochem. 2019;83 [3] Pérez-Varela M, Corral J, Andrés Vallejo J, et al. (11):2163–2171. Mutants in the β-subunit of the RNA polymerase [19] Vallejo JA, Beceiro A, Rumbo-Feal S, et al. Optimisation of impairing the surface-associated motility and virulence the Caenorhabditis elegans model for studying the patho- of Acinetobacter baumannii. Infect Immun. 2017;85:8. genesis of opportunistic Acinetobacter baumannii. [4] Biswas I, Machen A, Mettlach J. In vitro motility assays Int J Antimicrob Agents. 2015;pii: S0924-8579(15)00241. for Acinetobacter species. Methods Mol Biol. doi:10.1016/j.ijantimicag.2015.05.021 2019;1946:177–187. [20] Peleg AY, Jara S, Monga D, et al. Galleria mellonella as [5] Skiebe E, de Berardinis V, Morczinek P, et al. a model system to study Acinetobacter baumannii Surface-associated motility, a common trait of clin- pathogenesis and therapeutics. Antimicrob Agents ical isolates of Acinetobacter baumannii, depends on Chemother. 2009;53(6):2605–2609. 1,3-diaminopropane. Int J Med Microbiol. 2012;302 [21] Cunningham BC, Wells JA. High-resolution epitope map- (3):117–128. ping of hGH-receptor interactions by alanine-scanning [6] Eijkelkamp BA, Stroeher UH, Hassan KA, et al. H-NS mutagenesis. Science. 1989;244(4908):1081–1085. plays a role in expression of Acinetobacter baumannii [22] Eijkelkamp BA, Stroeher UH, Hassan KA, et al. virulence features. Infect Immun. 2013;81(7):2574–2583. Adherence and motility characteristics of clinical [7] Pérez-Varela M, Corral J, Aranda J, et al. Roles of Acinetobacter baumannii isolates. FEMS Microbiol efflux pumps from different superfamilies in the Lett. 2011;323(1):44–51. surface-associated motility and virulence of [23] Aranda J, Bardina C, Beceiro A, et al. Acinetobacter Acinetobacter baumannii ATCC 17978. Antimicrob baumannii RecA protein in repair of DNA damage, Agents Chemother. 2019;63(3). DOI:10.1128/ antimicrobial resistance, general stress response, and AAC.02190-18 virulence. J Bacteriol. 2011;193(15):3740–3747. [8] Clemmer KM, Bonomo RA, Rather PN. Genetic ana- [24] Wuichet K, Zhulin IB. Origins and diversification of lysis of surface motility in Acinetobacter baumannii. a complex signal transduction system in prokaryotes. Microbiology. 2011;157(9):2534–2544. Sci Signal. 2010;3(128):ra50. [9] Mussi MA, Gaddy JA, Cabruja M, et al. The opportu- [25] Zusman DR, Scott AE, Yang Z, et al. Chemosensory nistic human pathogen Acinetobacter baumannii senses pathways, motility and development in Myxococcus and responds to light. J Bacteriol. 2010;192 xanthus. Nat Rev Microbiol. 2007;5(11):862–872. (24):6336–6345. [26] Bellenger K, Ma X, Shi W, et al. A CheW homologue is [10] McQueary CN, Kirkup BC, Si Y, et al. Extracellular required for Myxococcus xanthus fruiting body devel- stress and lipopolysaccharide modulate Acinetobacter opment, social gliding motility, and fibril biogenesis. baumannii surface-associated motility. J Microbiol. J Bacteriol. 2002;184(20):5654–5660. 2012;50(3):434–443. [11] Sampedro I, Parales RE, Krell T, et al. Pseudomonas [27] Rumbo-Feal S, Pérez A, Ramelot TA, et al. chemotaxis. FEMS Microbiol Rev. 2015;39(1):17–46. Contribution of the A. baumannii A1S_0114 gene to [12] Chen R, Lv R, Xiao L, et al. A1S_2811, a CheA/Y-like hybrid the interaction with eukaryotic cells and virulence. two-component regulator from Acinetobacter baumannii Front Cell Infect Microbiol. 2017;7:108. ATCC17978, is involved in surface motility and biofilm forma- [28] Yao J, Allen C. Chemotaxis is required for virulence tion in this bacterium. Microbiologyopen. 2017;6(5):e00510. and competitive fitness of the bacterial wilt pathogen [13] Irazoki O, Aranda J, Zimmermann T, et al. Molecular Ralstonia solanacearum. J Bacteriol. 2006;188 interaction and cellular location of RecA and CheW (10):3697–3708. proteins in Salmonella enterica during SOS response [29] Harding CM, Tracy EN, Carruthers MD, et al. and their implication in swarming. Front Microbiol. Acinetobacter baumannii strain M2 produces type IV 2016;7:1560. pili which play a role in natural transformation and [14] Miller JH. Experiments in molecular genetics. Cold twitching motility but not surface-associated motility. Spring Harbor Laboratory, NY. 1972. MBio. 2013;4(4). DOI:10.1128/mBio.00360-13