AbaM Regulates Quorum Sensing, Biofilm Formation, and Virulence in Acinetobacter baumannii

Acinetobacter baumannii is a multiantibiotic-resistant pathogen of global health care importance. Understanding Acinetobacter virulence gene regulation could aid the development of novel anti-infective strategies.

regulator gene (abaR/ABUW_3774/ABUW_RS18375). Between abaR and abaI, a third gene is located which we term here abaM (ABUW_3775) (Fig. 1A). The chromosomal organization of these three QS genes is well conserved among Acinetobacter spp (see Fig. S1 in the supplemental material). Despite the location of abaM adjacent to abaI and transcribed in the same direction, AbaM has only low (ca. 20 to 30%) sequence identity to orthologs present in Pseudomonas fuscovaginae and Burkholderia spp. (Fig. 1B). However, it retains the well-conserved regions shared by other RsaM orthologues, including most of the protein secondary structural  (48) was used to create the alignment, and ESPript (49) was used to render residue similarities and generate the final figure. The red background indicates conserved residues. Red residues indicate conservative substitutions. Blue frames indicate highly conserved regions. The secondary structures in B. cenocepacia BcRsaM (PDB entry 4O2H) are displayed above the alignment. h , 3 10 -helix; a, a-helices; b, b-strands; TT, strict b-turns; TTT, strict a-turns.
AHL production is enhanced in an abaM mutant. AHL production in the opaque variants of the AB5075 wild-type and abaM::T26 and abaI::T26 mutant strains, respectively, was quantified via liquid chromatography-tandem mass spectrometry (LC-MS/ MS) during growth under static conditions, since it appears to be enhanced by surface and OHC10 (C) production throughout growth. Error bars represent the standard deviations between three biological replicates. Asterisks indicate statistically significant differences: **, P # 0.01; ***, P # 0.001. attachment in other A. baumannii strains (30). OHC12 ( Fig. 2A) was the major AHL produced by AB5075 under these conditions. Compared to the wild type, the abaM::T26 mutant produced significantly greater amounts of OHC12 at each time point sampled (a difference between 100-and 875-fold) (Fig. 2B). N-(3-Hydroxydecanoyl)-L-homoserine (OHC10) was also detected at much lower concentrations in the abaM::T26 mutant throughout growth but only in the 24-h sample in the wild type (Fig. 2C). No AHLs were detected in any of the abaI::T26 samples ( Fig. 2B and C). These data suggest that AbaM is a negative regulator of AHL production.
Contribution of QS and abaM to surface motility, biofilm formation, and virulence. The surface motility of all three strains on 0.3% Eiken agar LS-LB plates was examined. Compared to the wild type (59.6 6 0.7 mm), the abaI::T26 mutant exhibited significantly reduced surface motility (36.5 6 1.4 mm), whereas the abaM::T26 mutant was significantly more motile (76.7 6 2.4 mm) (Fig. 3A). The provision of exogenous OHC12 increased the surface motility of both wild-type and abaI mutant strains to levels similar to that of the abaM mutant (see Fig. S2A).
The ability of AB5075, abaI::T26, and abaM::T26 strains to attach to abiotic surfaces was evaluated on propylene tubes. The abaM mutant formed ;3-fold more biofilm than did the wild type (Fig. 3B). Under these growth conditions, the biofilm produced by the abaI mutant (opaque variant) (Fig. 3B) was not significantly different from the wild type but increased after the exogenous provision of OHC12 (see Fig. S2B).
The contribution of abaM and QS to AB5075 virulence was assessed using a G. mellonella larvae infection model (Fig. 3C). No differences in killing were observed between the wild-type and the abaI::T26 mutant when injecting either 2 Â 10 4 or 2 Â 10 5 CFU/larva. However, the abaM::T26 mutant was significantly less virulent than the parental strain. Larvae injected with 2 Â 10 5 CFU of the abaM::T26 mutant also showed a lower survival rate than the larvae injected with 2 Â 10 4 CFU of the wild Biofilm formation on polypropylene. For both biofilm and surface motility assays, error bars indicate the standard deviations. Asterisks indicate statistically significant differences compared to the wild-type AB5075 strain. ***, P # 0.001; ****, P # 0.0001. (C) Galleria mellonella larva killing after inoculation of approximately 2 Â 10 4 (left) or 2 Â 10 5 (right) CFU/larva. Each graph represents data from three independent biological replicates together. At least 30 larvae were used for each strain and assay. None of the control larvae died after 5 days. Asterisks indicate statistically significant differences compared to the wild-type AB5075 strain. ****, P # 0.0001. type or the abaI::T26 mutant. Exogenously supplied OHC12 did not affect the virulence of the wild type (see Fig. S2C).
Overall, these results suggest that abaM is a negative regulator of surface motility and biofilm formation and required for full virulence in G. mellonella.
Genetic complementation of the abaM mutant phenotypes. Complementation of the abaM::T26 mutant with the abaM gene in trans (pMQ_abaM) restored surface motility (see Fig. S3A) and biofilm formation (see Fig. S3B) and reduced both OHC12 and OHC10 production by approximately 50% (see Fig. S3C and D). However, complementation of the abaM mutation did not restore abaM::T26 virulence to wild-type levels (see Fig. S3E).
Transcriptomic analysis of abaI::T26 and abaM::T26. To characterize the AbaM and QS regulon of A. baumannii AB5075, we performed transcriptomic profiling of AB5075 in comparison with the abaM::T26 and abaI::T26 mutants using RNA sequencing (RNA-seq), which was then validated for two key target genes via quantitative real-time PCR. For these analyses, we used total RNA extractions from cells grown for 18 h in static conditions when maximum OHC12 levels are produced by the abaM mutant (Fig. 2).
Compared to the wild-type strain, 88 genes were upregulated and 9 were downregulated in the abaI::T26 mutant, whereas 52 were upregulated and 24 were downregulated in the abaM::T26 mutant [log 2 (fold change) $ 1] ( Fig. 4; see also Tables S3 and S4). Moreover, 21 of the upregulated genes were shared between abaI::T26 and abaM::T26 ( Fig. 4; see also Tables S3 and S4), whereas none of the downregulated genes was coregulated. Among the genes upregulated in both mutants there were all the genes of the csu operon, a putative TetR family transcriptional regulator (ABUW_1486/ABUW_RS07245) located immediately upstream of the csu operon, as well as genes coding for a flavohemoprotein, an uncharacterized transcriptional regulator, a thermonuclease, a sulfate permease, a toxic anion resistance protein, and seven hypothetical proteins (see Tables S3 and S4).
Moreover, some of the genes of the biosynthetic operon involved in the synthesis of acinetin 505, the QS transcriptional regulator abaR and the AHL synthase abaI, were both upregulated in the abaM::T26 mutant, which also differentially expressed genes encoding proteins involved in the stress response, iron acquisition, diverse metabolism and energy production, chaperones, protein folding, and antibiotic resistance, e.g., class D beta-lactamase OXA-23 involved in resistance to carbapenems (see Table S4). Similarly, differentially regulated genes in the abaI mutant included diverse metabolic and energy production-related genes, as well as diverse genes coding for transcriptional regulators, stress response-related proteins, and membrane transport proteins (see Table S3).
To validate the transcriptomic data, qPCR was performed with the same RNA samples used in the RNA-seq for the csuA/B and the ABUW_3773 (the first gene of the acinetin 505 biosynthetic operon) genes (Fig. 5). Compared to the wild type, the data obtained showed a significant increase in csuA/B expression in both the abaI::T26 and abaM::T26 mutants, whereas ABUW_3773 expression was significantly higher only in the abaM::T26 mutant. These results correlate with the data obtained from the RNA-seq experiments.
Regulation of abaM. To further elucidate the regulation of abaM expression, an abaM promoter-luxCDABE operon fusion was constructed. This was introduced via a miniTn7 transposon into AB5075 and both abaI::T26 and abaM::T26 mutants, and the activity of the predicted promoter was measured by luminescence output in the presence or absence of OHC12. The activity of the abaM promoter significantly varied between the strains. In the abaM::T26, luminescence was approximately 40% higher than in the wild type, whereas the abaI::T26 mutant showed a 75% reduction compared to the parental strain (Fig. 6A). Moreover, exogenous provision of OHC12 increased the abaM promoter activity in all three strains (Fig. 6B).
Similarly, Fig. 6C shows that expression of an abaI::lux promoter fusion, which is reduced in the abaI mutant compared to the wild-type strain, is strongly stimulated by OHC12. These data suggest that abaM expression is negatively autoregulated but, in common with abaI, is positively regulated by QS, which in turn is negatively controlled via AbaM (Fig. 7).

DISCUSSION
In this study, we have established that the RsaM ortholog, AbaM plays a major role in regulating QS-dependent and QS-independent genes in A. baumannii 5075. Disruption of abaM substantially increased AHL production, indicating that AbaM negatively regulates AHL biosynthesis. In A. baumannii 5075, the concentration of OHC12 produced was very low (,1 nM), suggesting that, at least under the static growth conditions used in this study, AbaM exerts tight control over QS. This raises the question of when and under what conditions QS is active in AB5075, especially since the half-maximal responses for LuxR proteins activated by long-chain AHLs is in the 5 to 10 nM range (31). In A. baumannii, the QS locus has an RXI topological arrangement previously described in other bacteria, where the X gene between the QS regulator (R) and the synthase (I) genes is a negative regulator of QS. Examples of three different classes of X include rsaL, rsaM, and mupX (32). In this context, AbaM appears to behave similarly to RsaL in the Pseudomonas aeruginosa lasR/rsaL/lasI QS system, despite their functional and structural differences. The rsaL gene, which is divergent to lasI, is positively regulated by LasR and antagonizes the LasR-mediated activation of lasI, thus counteracting the QS positive-feedback autoinduction and providing AHL homeostasis (33). This type of regulatory circuit (Fig. 7), termed an incoherent feed forward loop (IFFL) in contrast to simple feed forward loops, can display complex behaviors that include stabilization of output signals and bounded output which ensures robustness against fluctuations in the input signal levels (32,34). Hence, the large increase in AHL Error bars indicate standard deviations between three independent biological replicates. Asterisks indicate statistically significant differences compared to the wild-type AB5075 strain. ***, P # 0.001; ****, P # 0.0001. production following deletion of abaM, rsaL, and rsaM genes that results in less-virulent mutants is indicative of the importance of a stabilizing negative regulatory pathway in AHL-dependent QS systems (32,34). Consistent with this model, abaM expression was found to be negatively autoregulated but positively regulated by QS. In silico analysis of the DNA sequence between abaR and abaI by using Bprom (Softberry) and BDGP (www.fruitfly.org), as well as our RNAseq data, all predict the presence of putative 210 and 235 regions for abaI and abaM, respectively, suggesting that these genes are not cotranscribed. Similar findings have been reported for rsaM1 and rsaM2 from Burkholderia thailandensis (27). Reverse transcriptase-PCR (see Fig. S4 in the supplemental material) confirm that abaM and abaI in A. baumannii A5075 do not form an operon. Moreover, the abaM gene has a predicted lux box (CTGGTTAAATATAACAG) 68 bp upstream of the start codon and 178 bp downstream of the 235 and 210 promoter elements. This lux box is similar to those found upstream of abaI (CTGTAAATTCTTACAG) in both A. baumannii 5075 and A. nosocomialis M2. These results are consistent with an IFFL circuit, although further work will be required to fully characterize its properties and control of genes coregulated by QS and AbaM.
Phenotypic characterization of the A. baumannii abaM mutant revealed enhanced surface motility and biofilm formation but reduced virulence compared to the wild type. Previous studies have shown that rsaM orthologues are required for full virulence in plants (24,26), but this is, to the best of our knowledge, the first time that an rsaMlike gene has been reported to be required for full virulence in a human pathogen albeit in an insect infection model. The contribution of BcRsaM in B. cenocepacia to swarming motility and biofilm formation has also been reported (28). However, deletion of BcRsaM reduced both swarming and surface attachment, the opposite to that observed for the abaM mutant. Interestingly, both biofilm formation and surface motility in Acinetobacter have been associated with increased virulence (10,35).
Genetic complementation of the abaM mutant was achieved for surface motility, biofilm formation, and AHL production but not for virulence in Galleria mellonella. Similar observations have been previously reported for other abaM orthologues, most notably tofM (26) and BcRsaM (28), leading to the suggestion that RsaM-like proteins may be cis-acting regulators (23).
To further define the role of QS in A. baumannii 5075, phenotypic characterization of the AHL synthase mutant, abaI::T26, was performed. The mutant did not produce any detectable AHLs, consistent with previous studies and bioinformatic analysis indicating that A. baumannii 5075 possesses a single QS locus that is responsible for AHL production (18,36). Similarly, disruption of abaI negatively affected surface motility and responded to exogenous 3OHC12, as previously noted for other Acinetobacter strains/species (20). In AB5075, biofilm formation in the opaque variant of the abaI::T26 mutant was not significantly different from with the wild type. However, it increased well above the wild type in response to exogenous OHC12, consistent with other work on the Acinetobacter AHL synthase (18,37). Previous studies on QS and biofilm formation in Acinetobacter have been performed with strains that were, in contrast to AB5075, either not phase variable or not known to be phase variable. Since the experiments performed in this study were all carried out with the opaque Acinetobacter variant, we also investigated biofilm formation by the translucent variant. Figure S5 shows that the abaI translucent variant produced less biofilm than the wild type. However, biofilm formation increased for both opaque and translucent abaI variants in response to exogenous OHC12. Furthermore, our results suggest that QS does not play an important role in virulence in the G. mellonella. While this is not unprecedented (38), the role of Acinetobacter QS in virulence is still not well defined, and other studies suggest that QS may play an important role, depending on the strain and infection model used (21,22). Overall, our data suggest that for strain AB5075 QS is important in surface motility and biofilm formation but not virulence. However, further work is required to fully elucidate the role of QS in the pathogenesis of Acinetobacter infection and any cross talk with other regulatory networks.
Here, we performed RNA-seq on A. baumannii AB5075 grown in static conditions where AHL production was elevated in order to identify genes regulated via AbaM and QS and likely to be involved in surface attachment and biofilm formation. A comparison of the genes differentially expressed in the abaM and abaI mutants with the wild type revealed that AbaM has both QS-dependent (;22% of the QS regulon) and QS-independent gene targets. A similar overlap has been noted for the RsaM regulon and its cognate QS system in P. fuscovaginae (25). Among the most upregulated genes in both abaI and abaM mutants compared to the wild type were those belonging to the csu operon (ABUW_1487-ABUW_1492/ABUW_RS07250-ABUW_RS07275). This operon encodes the proteins responsible for the synthesis of the Csu pilus, a type I chaperoneusher pilus involved in attachment and biofilm formation (39)(40)(41). Moreover, the abaM::T26 mutant also showed higher expression of some genes of the acinetin 505 biosynthetic operon, which has also been linked to biofilm formation in A. baumannii ATCC 17978 (42).
A previous comparison of the transcriptomes of the multidrug resistant clinical A. baumannii strain 863 with an isogenic abaI deletion mutant highlighted the differential regulation of 352 genes involved in carbon source metabolism, energy production, stress response, and translation (43). However, apart from abaI, no other common differentially regulated genes could be identified when the A5075 abaI and the 863 abaI mutant transcriptomes are compared. This may be because of the different strains and growth conditions and sampling times used. In addition, the abaI mutant reported by Ng et al. (43) exhibited a growth defect. This raises the possibility of a secondary mutation contributing to the transcriptome data, which was not validated by chemical or genetic complementation with OHC12 or abaI, respectively.
In the transcriptomic experiments presented here, the abaI transcripts were found in larger amounts in the abaI mutant in contrast to the abaI::lux promoter fusion data where, as expected, abaI expression was lower in the abaI mutant and stimulated by provision of exogenous OHC12. However, the regulation of A. baumannii QS is complex, especially given the nature of incoherent feed-forward loops, our lack of understanding of the mode of action of AbaM and the impact of other regulatory factors on the system. In addition, the promoter fusion assays were carried out over the entire growth curve in 96-well plates, whereas the RNA was prepared from cells at a single time point grown in larger volumes. In these experiments, the peak of abaI promoter activity was in mid/late log phase, whereas the RNA-seq samples were prepared from late stationary-phase cultures. It is also possible that the transposon insertion in abaI impacts the amount and stability of the transcript.
For abaM, we found that promoter activity and transcript levels mutant were similar. While the activity of the promoter increased by ;50% in the abaM mutant compared to the wild type, we found a log 2 fold change of ;0.9 in the abaM mutant. This was just below the cutoff applied to our data (see Table S4). In the abaI mutant, abaM was not differentially expressed at the late stationary time point chosen for the RNA-seq. Consequently, future work will be required to unravel these observations with respect to abaI and abaM regulation, particularly in the context of growth environment.
Overall, the work described here establishes that AbaM plays a central role in regulating QS, surface motility, biofilm formation, and virulence. The apparently contradictory regulatory impact of abaM and abaI mutations that result in either increased or no AHL production, respectively, on the expression of genes such as the csu cluster can be explained as follows. In an abaI mutant (no AHLs), abaM expression is reduced and hence csu expression is increased. In an abaM mutant csu expression is also increased since AbaM is absent (Fig. 7). Further work will be required to elucidate the biochemical function and mechanism of action of AbaM and the RsaM protein family in general.

MATERIALS AND METHODS
Strains and growth conditions. The strains and plasmids used are listed in Table S1 in the supplemental material. A. baumannii AB5075 (29) and the isogenic abaI::T26 and abaM::T26 mutants were obtained from the transposon library available from the University of Washington (44). A. baumannii was routinely grown in low-sodium chloride (5 g/liter) lysogeny broth (LS-LB). OHC12 was synthesized as described previously (45). The opaque and translucent phases of the wild type and mutants were separated as described by Tipton et al. (11) after growth on phase-observation LB (PO-LB) plates and observation of colonies under light microscopy using oblique indirect illumination.
Construction of a genetically complemented abaM::T26 strain. Plasmid pMQ557M (see Table S1) was obtained by digesting pMQ557 with PmlI (to remove the genes required for yeast replication) and religating the resulting large linear product. The abaM gene plus 768 bp from its upstream region (containing the predicted native promoter) were amplified by PCR using the primers listed in Table S2. The PCR fragments were digested with BamHI and KpnI and ligated in the multiple cloning site (MCS) of both pMQpMQ557M and introduced into abaM::T26 by electroporation. The stability of the vector pMQ557M and abaM complementing plasmid pMQ_abaM in both A5075 and the abaM mutant were confirmed by repeated daily subculture and plating out on LB agar with or without hygromycin (125 mg/ml) to determine viable counts as CFU/ml (see Fig. S6).
Construction of abaM::luxCDABE and abaI::luxCDABE promoter fusions. The abaR gene and the intergenic region between abaR and abaM (for the abaM fusion) or the region between abaR and the abaI (for the abaI fusion) were amplified by PCR and ligated in pGEM-T Easy using the pGEM-T Easy Vector System (Promega). The resulting plasmids and the promoterless luxCDABE operon were digested with KpnI and BamHI and ligated in order to introduce the lux operon downstream of the predicted promoter of abaM or abaI. These constructs were transferred into the MCS of the miniTn7T in pUC18T-miniTn7T-Hyg R plasmid (see Table S1) after digestion with NotI and PstI and ligation of the corresponding fragments.
The miniTn7 transposon (46) carrying the abaM promoter-lux operon fusion was inserted into A. baumannii AB5075 and the isogenic abaI::T26 and abaM::T26 mutants, whereas the abaI promoter-lux fusion was inserted into AB5075 and the isogenic abaI:T26 mutant. The bioluminescence output from the reporter fusions as a function of bacterial growth was quantified using an Infinite 200 PRO (Tecan Diagnostics) plate-reader over 24 h, and the optical density at 600 nm (OD 600 ) and relative light units (RLUs) were recorded every 30 min. When required, OHC12 was added at 200 nM unless otherwise stated.
Biofilm assays. Strains to be tested were inoculated into 1.5-ml polypropylene microcentrifuge tubes in LS-LB with or without OHC12, followed by incubation under static conditions at 37°C for 24 h. Biofilms were quantified by staining with 0.25% crystal violet and extraction with ethanol, and the absorbance (A 585 ) was recorded.
Surface motility assays. Surface motility was quantified as previously described (11) on LS-LB plates with or without OHC12 and containing 0.3% Eiken agar. Plates were incubated at 30°C for 16 h.
AHL extraction and detection. Cell-free supernatants from cultures grown in LS-LB under static conditions at 37°C were sterile filtered and extracted with acidified ethyl acetate. Extracts were evaporated to dryness and subjected to LC-MS/MS as previously described (45).
G. mellonella killing assays. G. mellonella larvae (Trularv) were obtained from BioSystems Technology, Ltd., Devon, United Kingdom. Assays were performed as described previously (11). Briefly, 2 Â 10 4 or 2 Â 10 5 CFU of Acinetobacter were injected into the larval hemolymph and incubated at 37°C, and the larvae were monitored for viability. At least 10 larvae were used for each strain and assay.
Total RNA extraction and RNA-seq. Bacteria were cultured in LS-LB under static conditions at 37°C for 18 h. The cells were resuspended in RNAprotect (Qiagen) prior to extracting total RNA using an RNeasy minikit (Qiagen). After treatment with DNA-free (Invitrogen), the absence of DNA contamination was confirmed using PCR, and the quality and quantity of the RNA samples was established using a 2100 Bioanalyzer (Agilent). Samples were sent for 150-bp paired-end sequencing via an Illumina platform and bioinformatic analysis to NovoGene (Hong Kong, China).
Quantitative real-time PCR. Complementary DNA (cDNA) synthesis and qPCR were carried out using LunaScript RT Supermix and Luna Universal qPCR Master Mix (New England BioLabs), respectively. The oligonucleotides used for qPCR are listed in Table S2 and qPCR was carried out in triplicate using a 7500 real-time PCR system (Thermo Fisher). Negative controls lacking template or RNA incubated without reverse transcriptase were included. The housekeeping gene rpoB was used as endogenous control for normalization.
Reverse transcription-PCR. cDNA was amplified using Q5 high-fidelity polymerase (New England Biolabs) with specific primers annealing in the coding region of each gene. Genomic DNA, extracted with a DNeasy blood and tissue kit (Qiagen), was used as a positive control. The PCR products were run in a 1.5% agarose electrophoresis gel before imaging under UV light using a Gel Doc XR1 Imager (Bio-Rad).
Data availability. Bacterial sequencing data have been deposited in NCBI's Gene Expression Omnibus (47) and are accessible through GEO series accession number GSE151925.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1 MB.