A1S_2811, a CheA/Y‐like hybrid two‐component regulator from Acinetobacter baumannii ATCC17978, is involved in surface motility and biofilm formation in this bacterium

Abstract Two‐component systems in Acinetobacter baumannii are associated with its virulence, drug resistance, motility, biofilm formation, and other characteristics. In this study, we used RecAb, a genetic engineering method, to investigate the function of A1S_2811 in A. baumannii strain ATCC17978. A1S_2811, a hypothetical hybrid sensor histidine kinase/response regulator, has four histidine‐containing phosphotransfer domains, a CheA‐like regulatory domain, and a CheY‐like receiver domain at its C terminus. Compared with the ATCC17978 strain, both surface motility and biofilm formation at the gas–liquid interface decreased significantly in the A1S_2811 knock‐out strain. The number of pilus‐like structures and the amount of extrapolymeric substances on the cell surface also decreased in the A1S_2811 null strain. Transcription of abaI, which encodes an N‐acylhomoserine lactone synthase in A. baumannii , decreased significantly in the A1S_2811 null strain, and supplementation with synthetic N‐(3‐oxodecanoyl) homoserine‐l‐lactone rescued the surface motility and biofilm formation phenotype in the null mutant. We speculate that A1S_2811 regulates surface motility and biofilm formation, not by regulating type IV pili‐associated genes expression, but by regulating the chaperone/usher pili‐associated csuA/ABCDE operon and the AbaI‐dependent quorum‐sensing pathway‐associated A1S_0112‐0119 operon instead.

phenotypes has been of great help in gaining better understanding this pathogen.
According to the bioinformatics analysis based on the presence of conserved amino acid motifs, structural features or limited homology, A1S_2811 (4,521 bp) in A. baumannii ATCC17978 is annotated as a cheA homolog (GenBank: NC_009085) (Smith et al., 2007). CheA in Escherichia coli and its homolog chpA in Pseudomonas aeruginosa, which are components of the chemotactic signal transduction system in these bacteria, have been investigated in detail (Baker, Wolanin, & Stock, 2006;Elowitz, Surette, Wolf, Stock, & Leibler, 1999;Li, Swanson, Simon, & Weis, 1995;Stewart, 1997;Whitchurch et al., 2004). Both of them are TCSs.
It was reported that cheA/Y in E. coli and chpA/Y in P. aeruginosa play regulatory roles in controlling bacterial motility via flagella or type IV pili (Alon et al., 1998;Baker et al., 2006;Bertrand, West, & Engel, 2010;Elowitz et al., 1999;Li et al., 1995;Whitchurch et al., 2004); however, A1S_2811 in A. baumannii has not been studied as yet.
Although it lacks flagella, A. baumannii is motile (Clemmer, Bonomo, & Rather, 2011;Mussi et al., 2010). The underlying molecular and genetic basis of motility in A. baumannii' remains ambiguous (McBride, 2010), and its motility phenotypes are diverse. Certain A. baumannii strains exhibit a phenomenon known as twitching motility, which is visualized as a kind of jerky movement on wet surfaces (Eijkelkamp et al., 2011b;Semmler, Whitchurch, & Mattick, 1999). Barker & Maxted (1975) observed movements on the wet surface of semisolid media, which they called "swarming" movement; they also noticed that when A. baumannii was inoculated by stabbing it into agar, some strains could move beneath the agar or form "ditches." A series of genes required for the surface motility of A. baumannii have been identified (Clemmer et al., 2011), and the motility was associated with type IV pili (Clemmer et al., 2011;Harding et al., 2013), quorum sensing (Clemmer et al., 2011), blue light sensing (Mussi et al., 2010), iron deficiency (Eijkelkamp, Hassan, Paulsen, & Brown, 2011a), and 1,3-diaminopropane (Skiebe et al., 2012). However, the published literature has insufficiently answered the question of how A. baumannii moves, and even the definitions of the various forms of movement are confusing. Therefore, investigating the A1S_2811 hypothetical chemotactic signal transduction system component in A. baumannii ATCC17978 has potential to contribute not only to better understanding of the function of TCSs in A. baumannii but also to elucidate its motility mechanism.

| Bacterial strains and plasmids
The bacterial strains and plasmids used in this study are listed in Table 1. Bacterial strains were routinely maintained in Luria-Bertani (LB) broth or agar.

| Identification of the operon containing A1S_2811
The organization of A1S_2811 and its surrounding genes suggests that five genes spanning A1S_2811 to A1S_2815 might belong to one operon. Primers to amplify the intergenic regions between these genes were designed (Table S1), and then synthesized by Sangon Biotech Co., (Shanghai, China). RNA was extracted from ATCC17978 and transcribed into cDNA. For the PCR amplifications, the extracted RNA and genomic DNA were set as the controls.

| Creating gene knockouts with the Rec Ab system
Recombination-mediated chromosomal gene inactivation was performed as previously described (Tucker et al., 2014). A1S_2811 and A1S_2213 (csuE) were knocked out in ATCC17978. To knock out the entire 4,521 bp sequence of the A1S_2811 gene, we increased the length of the homologous regions to promote the recombination efficiency. By fusion PCR, we constructed a 2,047 bp DNA fragment containing 378 bases upstream and 330 bases downstream of the A1S_2811 coding sequence (CDS), flanking the kanamycin resistance cassette, which was amplified from the PKD4 plasmid. To knock out csuE, oligonucleotides containing 112 bases flanking the CDS of the csuE gene were synthesized by Sangon Biotech Co. A. baumannii carrying Rec Ab on pMMB67EH (pAT01) was inoculated into LB media, which contained carbenicillin (100 μg/mL) to maintain the plasmid. Isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 2 mmol/L, and the bacteria were grown at 37°C, 200 rpm for 3 h.
After three washes with ice-cold 10% glycerol, 100 μl of bacteria (~10 10 bacteria) was mixed with 5 μg of the PCR products and then electroporated in a 2-mm cuvette at 1.8 kV. After culturing in 4 ml of LB medium containing 2 mmol/L IPTG, the transformants were selected on kanamycin agar to identify colonies whose targeted genes were replaced by the kanamycin cassette. The selected colonies were verified by PCR and DNA sequencing, and then transferred to carbenicillin-negative agar to cure the plasmid pAT01. Then, another plasmid containing the FLP/FRT recombinase system (pAT02) was electroporated into the selected colonies. After FLP/FRT recombination, the kanamycin resistance cassettes were replaced by the FLP recognition target (FRT) loci.

| Complementation of mutants
Complementation vectors for the Δ2811::FRT and ΔcsuE::FRT strains were constructed using the primer sets listed in Table S1. Plasmid pBAD18Kan-ori (Choi, Slamti, Avci, Pier, & Maira-Litran, 2009), provided by Professor Xilin Zhao of Xiamen University, China, was used to construct the A1S_2811 complementary strain. A1S_2811 gene was cloned into the multiple cloning site (MCS) in pBAD18Kanori and electroporated into the Δ2811::FRT mutant. The csuE complementation was conducted by amplifying the full-length gene with a primer set containing the Shine-Dalgarno AGGAGG sequence (Table   S1). Next, the PCR product was cloned into pABBR_MCS and electroporated into the ΔcsuE::FRT mutant (Tucker et al., 2014).
Then, the overnight culture was diluted 1:100 with LB. For each strain, 18 replicates of 100 μL aliquots of diluted culture were placed into each well of a polystyrene 96-well cell culture plate and then grown without shaking for 24 h at 30°C (Tucker et al., 2014). Nine wells were used to determine the optical density (OD) 600 to estimate the total cell biomass. The liquid from the other nine wells was aspirated carefully and the remaining biofilms were rinsed with distilled water. The biofilm walls were then stained with 0.1% crystal violet and solubilized with ethanol-acetone (O'Toole et al., 1999). The OD 580 of the processed solution was determined and the OD 580 /OD 600 ratio was used to measure the biofilm amounts (Tomaras et al., 2003). All assays were performed twice using fresh samples each time.

| Motility assays
First, the strains were cultured on LB plates for two passages after recovery from the glycerol stocks. A single colony was then inoculated into 5 ml of LB broth and incubated for 24 h at 30°C without shaking prior to performing the assay. Next, the samples were adjusted to an OD 600 of 0.6 with LB broth, and 2 μl of the bacterial culture was placed on the surface of the motility assay plates. Motility was investigated on motility plates after bacterial incubation at 37°C for 18 h. The motility plates were prepared with 10 g/L tryptone, 10 g/L NaCl and 5 g/L yeast extract, and the addition of 0.5% noble agar (Becton Dickinson, Sparks, MD, USA).

| Transcriptome analysis
Total RNA was isolated from ATCC17978 and the Δ2811::FRT mutant, both of which were previously grown on motility plates, using a Pure Link ™ RNA Mini Kit (Invitrogen, Carlsbad, CA, USA). The RNA concentration and quality of each sample were determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Rockford, IL, USA). In total, 3 μg of RNA per sample was used as the input material for RNASeq library preparation. Strand-specific transcriptome sequencing was performed by Novogene Bioinformatics Technology Co., Ltd. (Beijing, China). HTSeq v0.6.1 was used to count the number of reads mapped to each gene (Anders & Huber, 2010).
The mapped fragments per kilo-base of gene model per million reads associated with each gene was calculated based on the length of the gene and counts for the reads mapped to this gene (Trapnell, Pachter, & Salzberg, 2009). Prior to differential gene expression analysis, the read counts were adjusted using the edgeR program package using one scaling normalized factor for each sequenced library (Robinson, McCarthy, & Smyth, 2010). Differential expression analysis of two conditions was performed using the DEGSeq R package (1.20.0) (Wang, Feng, Wang, Wang, & Zhang, 2010). The p values were adjusted using the Benjamini and Hochberg method (Hochberg & Benjamini, 1990). A corrected p value (q value) of 0.005 and a log 2 of 1 (fold change) was set as the threshold for significant transcription variation.

| Reverse transcriptase-PCR (RT-PCR)
RT-PCR was performed to verify parts of the transcriptome results (the primers used are shown in Table S1). All RNA samples were extracted from ATCC17978, Δ2811::FRT mutant and the complementation strain, which were previously grown on motility plates, using a following the manufacturer-supplied protocols. Expression was quantified relatively by comparison to that of 16S rRNA.

| Identification of the operon containing A1S_2811
The genome annotation of ATCC17978 suggests that A1S_2811 11-13). The templates used were cDNA (lanes 2, 5, 8, 11), genomic DNA from ATCC17978 (lanes 3, 6, 9, 12), and genomic RNA as a negative control (lanes 4, 7, 10, 13). Lane 1, 2,000-kb DNA marker terminus (GenBank: NC_009085) (Smith et al., 2007). To investigate whether the five genes spanning A1S_2811 to A1S_2815 are parts of a single operon, primers were designed to amplify the intergenic regions between these genes. ATCC17978 cDNA was used as a PCR template, and genomic DNA was set as the control. The results demonstrated that A1S_2811 is cotranscribed with the following four upstream genes: A1S_2812 (pilJ), A1S_2813 (pilI), A1S_2814 (pilH), and A1S_2815 (pilG) (Figure 1). Because A1S_2811 is the last gene in this five-gene operon and the transcriptional direction of this operon is opposite to the adjacent gene A1S_2810, the probability of a polarity effect after A1S_2811 gene knockout should be relatively low.

| Phenotypes of ATCC17978 and ∆2811::FRT
We applied a new recombination-mediated knock-out system (Rec Ab system) to delete A1S_2811 from ATCC17978 and construct the Δ2811::FRT mutant. By PCR and sequencing tests, we confirmed that full-length A1S_2811 was deleted and replaced by 91-bp FRT loci. To investigate whether deleting A1S_2811 would affect the in vitro growth of A. baumannii, we tested the growth rates of the ∆2811::FRT and wild-type (WT) strains in LB media (Mussi, Relling, Limansky, & Viale, 2007). There was no significant difference between them when grown on LB medium ( Figure S1). Therefore, depleting A1S_2811 does not affect the in vitro growth of A. baumannii on LB medium.
We then tested the motility of ATCC17978 and Δ2811::FRT on motility plates. On the 0.5% noble agar motility plate, we found that ATCC17978 could not move at the interface between the bottom of the plate and the medium, and only surface motility on the motility plate was observed. Therefore, hereafter, when we talk about motility in this study, we are referring to "surface motility." ATCC17978 formed

| Gene expression changes caused by deletion of A1S_2811
Transcriptome sequencing showed that after deletion of A1S_2811, the expression of 117 genes was significantly downregulated (log 2 . Fold_change <−1), whereas 80 genes were upregulated (log 2 .  (Kanehisa, Sato, Kawashima, Furumichi, & Tanabe, 2016) and previous studies also (Clemmer et al., 2011;Giles, Stroeher, Eijkelkamp, & Brown, 2015). The high-throughput transcriptome results were verified by RT-PCR using four selected genes associated with motility and biofilm ( Figure 5). All the genes we tested were downregulated in both Δ2811::FRT and Δ2811-pBAD18Kan-ori as expected, whereas the transcriptional levels of these genes were restored in the complementation strain ( Figure 5). In addition to these four genes, we also compared the transcriptional profiles of 23 other genes related to type IV pili. The RT-PCR results for these four genes verified the reliability of the RNASeq transcriptome results; the RT-PCR results for the other tested genes are shown in Table S4.

| Phenotypes of ATCC17978 and ∆ csuE::FRT
We found that the csuA/BABCDE operon is downregulated from the transcriptome analysis results of the A1S_2811 null mutant.
Therefore, to investigate the role of the csuA/BABCDE operon in terms of motility and biofilm formation, we knocked out csuE (the last gene in the csu operon) while taking into consideration the necessity of avoiding the polarity effect, and constructed mutant ∆ csuE::FRT using the Rec Ab system. By PCR and DNA sequencing, we confirmed that the full-length csuE was deleted and replaced by 91-bp FRT loci.
We tested the growth rates of the ∆ csuE::FRT and WT strain in LB medium (Mussi et al., 2007). There was no significant growth difference between them in this medium ( Figure S2). Therefore, depleting csuE does not affect the in vitro growth of A. baumannii in the tested medium.

| Supplementation with synthetic N-(3oxodecanoyl) homoserinel-lactone restored the phenotype of Δ2811::FRT
Among the genes in Table 2, A1S_0109 (abaI) was reported as being the only autoinducer synthase encoded in the A. baumannii genome (Niu, Clemmer, Bonomo, & Rather, 2008). To investigate the possible role of bacterial quorum sensing and the regulation of A1S-2811 in ATCC17978, we supplied 100 μM synthetic F I G U R E 5 Transcriptome sequencing validation results for selected genes. Four genes (A1S_0116, A1S_0109 (abaI), A1S_2213 (csuE), and A1S_2811) were selected from Table 2. The expression ratio of each gene was calculated as the transcriptional level in each stain divided by the transcriptional level of the WT ATCC17978 strain. In addition to Δ2811::FRT, a WT strain carrying pBAD18Kan-ori, a mutant carrying pBAD18Kan-ori and the complementation strain were tested. Asterisks denote significant differences in the transcriptional levels (t test; *p < 0.0001; n = 3) N-(3-Oxodecanoyl)l-homoserine lactone (3-oxo-C10 HSL) to the mutant ∆2811::FRT when performing biofilm and motility tests. As shown in Figure 8, the motility and biofilm defects were rescued completely with 3-oxo-C10 HSL.

| DISCUSSION
In this study, we have shown that deleting A1S_2811 decreased the surface motility and biofilm formation of A. baumannii ATCC17978.
Biofilm, a structure of connected cells surrounded by a matrix of extracellular polysaccharides (Moonmangmee et al., 2002;Yamamoto, Arai, Ishii, & Igarashi, 2012), is associated with multidrug resistance in A. baumannii (Badave & Kulkarni, 2015;Rao et al., 2008). The pellicle is a special form of biofilm localized in the air-liquid interface (Branda, Vik, Friedman, & Kolter, 2005). In our study, we found that the motility and biofilm formation phenotypes of ATCC17978 were closely related. Culture liquid incubated at 30°C for about 24 h without shaking produced pellicle biofilms and the bacteria showed a motility phenotype on motility plates. When ATCC17978 was preincubated with shaking, no gas-liquid interface biofilm formed and the strain showed no motility on the motility plates. We have no explanation for this as yet. A previous study also found that surface film-forming strains were motile (Giles et al., 2015). In contrast, another study found that clinical respiratory isolates frequently formed more biofilm and were less motile than nonclinical strains (Vijayakumar et al., 2016). The association between motility and biofilm formation in A. baumannii remains ambiguous; this lack of clarity may be related to different genetic backgrounds in the strains of this species and requires further investigation. Notably, although both ∆2811::FRT and ∆csuE::FRT were nonmotile on the motility plates (Figures 2 and 6), they might not be completely defective in motility. When grown on plates with only 0.3% agar, they also formed large colonies. However, their colonies were far smaller than those of the WT strain. This implies that bacterial regulatory mechanisms and mechanisms involved in motility are complex.
Although a pilK homolog is not found in the corresponding location in A. baumannii, the structure of these two operons is almost F I G U R E 6 Surface motility assay for A. baumannii ATCC17978, ΔcsuE::FRT, ΔcsuE::FRT-c, 17978-pABBR_MCS, and ΔcsuE-pABBR_MCS identical. chpA was found to be associated with type IV pili assembly and/or retraction as well as expression of the pilin subunit gene pilA (Whitchurch et al., 2004). ChpA functions upstream of PilH and PilT and the histidine kinase domain of ChpA, and the phosphoacceptor sites of both PilG and PilH are required for type IV pili function (Bertrand et al., 2010).
Previous studies have also shown that motility in A. baumannii is associated with type IV pili (Clemmer et al., 2011;Harding et al., 2013). However, in our study expression of the type IV pili genes related to twitching motility and type IV pilus assembly were not affected by deleting A1S_2811, as confirmed by RT-PCR (Table S4) Asterisks denote significant differences in biofilm formation (t test; *p < 0.0001; n = 9). (b) Surface motility of ATCC17978 and Δ2811::FRT (plus or minus 3-oxo-C10-HSL) on 0.5% noble agar motility plates As we did not observe significant transcriptional variations for type IV pili in the Δ2811::FRT mutant in this study, we speculate that A1S_2811-related surface motility and biofilm formation might be independent of type IV pili. Our study suggests that A1S_2811mediated surface motility and biofilm formation might be associated with chaperone/usher (CU) pili instead. CU pilus is a type of nonflagellar appendage assembled on an outer membrane assembly platform called the usher where the periplasmic chaperone-bound pilus subunits are polymerized in an orderly fashion (Sauer, Remaut, Hultgren, & Waksman, 2004). CU pili can be found in diverse gramnegative bacteria, including important human and animal pathogens (Sauer et al., 2004). In the list of genes downregulated in the absence of A1S_2811, the csuA/BABCDE operon, which is responsible for CU pili assembly (Nait Chabane et al., 2014;Tomaras et al., 2003), is significantly downregulated ( Table 2). The csuA/BABCDE operon is required for biofilm formation on solid surfaces, and knocking-out csuE in ATCC19606 resulted in a biofilm-deficient phenotype and pili disappearance (Tomaras et al., 2003(Tomaras et al., , 2008. Little is known about the relationship between the csuA/BABCDE operon and motility in A. baumannii, except that one study found that deleting csuD in the A. baumannii M2 strain did not affect its motility (Clemmer et al., 2011), but as reported before for csuA mutant and csuE mutant (Tomaras et al., 2003), the ΔcsuD mutant exhibits a biofilm formation defect (Harding et al., 2013). To further investigate the role of CU pili in biofilm formation and motility, we constructed another mutant, ΔcsuE::FRT, and found that csuE was associated with motility and biofilm formation in ATCC17978.
Differences in the results we obtained might be related to the different strains we used and the different genes in the csuA/BABCDE operon we investigated.
Transcriptome analysis of the Δ2811 null mutant in comparison with ATCC17978 also showed that the transcriptional level of another operon (A1S_0112-0119) decreased significantly. This operon is annotated as being responsible for the nonribosomal production of a lipopeptide that possibly acts as a surfactant to aid motility. Previous studies have shown that the A1S_0112-0119 operon is essential for pellicle formation (Giles et al., 2015) and motility (Clemmer et al., 2011) in A. baumannii. The motility and pellicle phenotypes of A. baumannii might be linked via the expression of cAMP and the A1S_0112-0119 operon (Giles et al., 2015).
After knocking out A1S_2811, the transcriptional level of A1S_0109 (abaI) also decreased significantly; this gene is annotated as the only autoinducer synthase encoded in the A. baumannii genome (Niu et al., 2008). An abaI::Km mutant failed to produce any detectable AHL (N-acylhomoserine lactone) signals and was impaired in biofilm development in the A. baumannii M2 strain (Niu et al., 2008). Additionally, Luo et al. (2015) reported that non-native acylhomoserine lactone could enhance pili assembly and biofilm formation in A. baumannii ATCC19606. In our study, the Δ2811::FRT phenotype was rescuable by supplementation with synthetic 3-oxo-C10 HSL (one of the quorum-sensing AHLs). Collectively, these studies confirm that autoinducer-dependent quorum sensing plays a vital role in regulating motility and biofilm formation in A. baumannii. The A1S_0112-0119 operon was previously reported to be activated by quorum-sensing signals A1S_0109 (abaI) (Clemmer et al., 2011;Giles et al., 2015).
On the basis of the published scientific literature and our own results, we speculate that A1S_2811 is part of a TCS that regulates the A1S_0112-0119 operon via the AbaI-dependent quorum-sensing pathway in ATCC17978.
In this study, we confirmed that A1S_2811, a CheA/Y-like hybrid, two-component regulator in A. baumannii ATCC17978, is involved in this bacterium's surface motility and biofilm formation phenotypes.
The motility of ATCC17978 seemed to be not associated with the retraction of type IV pili, but was instead related to CU pili, a lipopeptide encoded by the A1S_0112-0119 operon and to the AbaI-dependent quorum-sensing pathway. A1S_2811 might regulate surface motility and biofilm formation via regulating the csuA/BABCDE operon associated with CU pili and the AbaI-dependent quorum-sensing pathwayassociated A1S_0112-0119 operon. However, the detailed regulation networks governing the exact mechanisms of interaction between the csuA/BABCDE operon, the A1S_0112-0119 operon and A1S_0109 (abaI) await further investigation.