Acinetobacter baumannii coordinates central metabolism, plasmid dissemination, and virulence by sensing nutrient availability

ABSTRACT Plasmid conjugation plays an important role in the dissemination of antibiotic-resistance genes. The emergence of multidrug-resistant isolates of Acinetobacter baumannii poses grave challenges in treating infections caused by this notorious nosocomial pathogen. Yet, the composition, functionality, and regulation of conjugative machinery utilized by A. baumannii remain poorly understood. Here, we found that conjugation of the major plasmid pAB3 of A. baumannii is mediated by a type IVB secretion system similar to the Dot/Icm transporter of Legionella pneumophila. Furthermore, the expression of the structural genes of the Dot/Icm-like system is co-regulated with genes involved in central metabolism by the GacS/GacA two-component system in response to various metabolites, including intermediates of the tricarboxylic acid cycle. Loss of GacS/A also severely impaired bacterial virulence. These results establish that A. baumannii coordinates metabolism with plasmid conjugation and virulence by sensing nutrient availability, which may be exploited to develop inhibitory agents for controlling the spread of drug-resistance genes and virulence factors. IMPORTANCE Plasmid conjugation is known to be an energy-expensive process, but our understanding of the molecular linkage between conjugation and metabolism is limited. Our finding reveals that Acinetobacter baumannii utilizes a two-component system to co-regulate metabolism, plasmid transfer, and virulence by sensing reaction intermediates of key metabolic pathways, which suggests that nutrient availability dictates not only bacterial proliferation but also horizontal gene transfer. The identification of Dot/Icm-like proteins as components of a conjugation system involved in the dissemination of antibiotic-resistance genes by A. baumannii has provided important targets for the development of agents capable of inhibiting virulence and the spread of anti-microbial-resistance genes in bacterial communities.

T ype IV secretion systems (T4SSs) are multi-subunit structures that span the cell envelope for inter-bacterial transfer of protein and/or DNA.The majority of T4SSs are associated with mobile genetic elements such as plasmids and transposons that function to promote the spread of genetic materials among cells (1,2).For T4SSs that are dedicated to bacterial pathogenicity by translocating virulence factors into host cells, many still retain the ability to transfer plasmids between bacterial cells, although at markedly lower frequencies.For example, the Dot/Icm system of Legionella pneumophila transfers its protein substrates at frequencies that are 1-2 magnitude higher than transferring plasmids (3,4).It is believed that T4SSs involved in bacterial virulence are adapted from dedicated conjugation systems (5,6).The emergence of multidrug-resistant (MDR) bacteria has become a grave challenge to clinicians worldwide (7).Horizontal gene transfer promoted by T4SSs, particularly those on conjugative plasmids, is one of the widely used mechanisms accounting for the spread of antibioticresistance genes among bacteria (8).
The Gram-negative Acinetobacter baumannii (Ab) is a successful pathogen that causes hospital-acquired infections, a problem that is compounded by the high prevalence of MDR isolates (9).Some Ab strains harbor multiple plasmids, including large conjugative plasmids (LCPs) and small mobilizable plasmids (SMPs) (10).SMPs contain a conserved active origin of transfer (oriT) and a relaxase gene (e.g., mobA), which facilitate their dissemination via the conjugative machinery often provided in trans by LCPs (11).Genomic analysis revealed that A. baumannii strain 17978 (Ab 17978 ) harbors at least eight genes that code for proteins with high-level similarity to components of the Dot/Icm type IV secretion system from species of Legionella and Coxiella (type IVB) (12).Yet, whether these proteins assemble into an active conjugation apparatus remains unclear.
The large plasmid in Ab strains has become a major carrier for antibiotic-resistance genes.For example, pAB3 in the Ab 17978 strain isolated in the 1950s harbors only one drug-resistance gene but that number is at least 13 in some more recent clinical isolates (10,12,13).In addition to antibiotic-resistance genes, LCPs also harbor regulatory proteins that control the expression of important traits encoded by Ab chromosome.For instance, the LCP pAB3 in Ab 17978 harbors two TetR-like proteins that repress the expression of its type VI secretion system, suggesting coordination of plasmid conjuga tion and killing of competitors by this bacterium (13).Similarly, the modern MDR strain UPAB1 has the capacity to replicate in macrophages by a mechanism that requires the LCP pAB5, which controls the expression of multiple virulence factors (14).
It has been long known that the expression of specialized protein secretion systems is regulated by two-component systems (TCSs).For example, SpiR/SsrB regulates the SPI-2 of Salmonella enterica Typhimurium and some of its effectors (15)(16)(17).The SpiR/SsrB TCS itself is regulated by at least two upstream TCSs, PhoP/PhoQ and OmpR/EnvZ, which respond to distinct environmental cues (18)(19)(20)(21).Similarly, the VirB T4SS of Agrobacterium tumefaciens is activated by the VirA/VirG TCS, which senses and responds to compounds such as acetosyringone released by wound plants (22).
Here, we found that the Dot/Icm protein orthologs encoded by pAB3 in strain 17978 are components of a T4SS that functions to promote not only its conjugation but also the transfer of SMPs such as RSF1010.We also found that the GacS/A TCS regulates the expression of at least some of the dot/icm-like genes as well as genes involved in central metabolism in response to multiple nutrients, including several reaction intermediates of the tricarboxylic acid (TCA) cycle.

RESULTS
The plasmid pAB3 encodes multiple proteins similar to components of the Dot/Icm system of L. pneumophila Earlier genomic analysis revealed that strain Ab 17978 codes for eight proteins that are homologous to components of the Dot/Icm system of L. pneumophila (12).More recent studies found that these genes are located on the large plasmid pAB3 (13).To determine the potential role of these proteins in the biology of A. baumannii, we first more carefully examined the genes on pAB3 and found that 11 instead of 8 of them encode proteins with high-level similarity to Dot/Icm proteins.Similar to dot/icm genes in L. pneumophila (23), these genes are clustered in two regions on the plasmid, and their organization in each case is almost identical to the respective dot/icm genes of L. pneumophila.Furthermore, dotDCB and dotIHGF each forms an operon-like structure (Fig. 1A).With the exception of DotG, which is considerably shorter than its counterpart in the L. pneumophila system (534 residues vs 1,048 residues), most of the ortholog pairs have similar lengths (Fig. 1B).

The Dot/Icm protein orthologs are parts of an active T4SS
Although the Dot/Icm system of L. pneumophila is a dedicated protein translocation apparatus, it retains the ability to transfer plasmids, despite at much lower frequencies (3,4).We thus set out to determine whether these Dot/Icm homologs are components of a T4SS that promotes plasmid dissemination.To this end, we first created the mutant ∆dotG and examined its ability to transfer pAB3 into strain WT -R , a derivative of Ab 17978 , which had been cured of pAB3 (24).Strain Ab 17978 transfers pAB3 to WT -R at a frequency of approximately 1 × 10 −5 per donor cell, and deletion of dotG rendered the transfer undetectable.The conjugation defect can be fully complemented by expressing DotG on a plasmid (Fig. 2A).Interestingly, we did not observe the transfer of pAB3 when E. coli strain DH5α was used as a recipient, probably because this plasmid cannot replicate in this bacterium (Fig. S1).To determine the requirement of the remaining Dot/Icm orthologs, we constructed a series of deletion mutants lacking one of these genes and examined them as the donor for pAB3 conjugation.With the exception of dotC, deletion of which reduced the transfer efficiency by appropriately 3 orders of magnitude, mutations in all other genes completely abolished the mobilization of pAB3 between Ab strains (Fig. 2B).
We also examined the role of dotG in the mobilization of the IncQ plasmid RSF1010 (4), which is capable of replicating in Ab (25).Similar to its importance in the trans fer of pAB3, DotG is required for mobilizing pKB5-Gm, which was modified from the RSF1010 derivative pKB5 (23) between Ab strains (Fig. 2C).Furthermore, akin to their role in pAB3 conjugation, with the exception of dotC, which is partially required for transferring pKB5-Gm, deletion of other dot/icm ortholog genes abolished the ability of pAB3 to mobilize the plasmid (Fig. 2D).Finally, mobilization of pKB5-Gm requires the cis-acting origin of transfer (oriT) because pJB908-Gm derived from pJB908 lacking this DNA element (26) cannot be mobilized under similar conditions (Fig. 2E).Similar results were obtained when E. coli strain DH5α was used as the recipient in experiments aiming at measuring the transfer of pKB5-Gm and its oriT-defective variant pJB908-Gm (Fig. 2F through H).Taken together, these results indicate that the Dot/Icm orthologs are essential components of a T4SS that promotes not only the conjugation of pAB3 but also SMPs recognizable by the machinery.

Nutrients promote pAB3 conjugation by inducing dot-like gene expression
Plasmid dissemination via type IV secretion systems among bacteria is an energy-expen sive process (27).It is thus possible that genes coding for components of conjugation machineries may be induced by certain nutrients.To test this hypothesis, we attemp ted to determine whether compounds that can be readily consumed by A. baumannii induce the expression of the dot/icm-like genes.To this end, several amino acids and intermediates of the TCA cycle were individually added to cultures of an A. baumannii strain harboring a plasmid carrying a reporter in which the expression of the FACSoptimized gfp variant (28) was driven by the promoter of the dotDCB gene cluster.Cells from saturated cultures of this strain grown in LB broth were diluted into a synthetic medium in which pyruvate was the sole carbon source.Each of the tested compounds was added to the medium at a final concentration of 10 mM and the expression of gfp was measured after 2-h incubation.We found that succinate, fuma rate, malate, and oxaloacetate significantly induced expression from the dotD promoter (Fig. 3A).Importantly, none of these metabolites detectably affected the growth of the bacteria within the experimental durations (Fig. S2).Similar induction was observed when asparagine, aspartic acids, or phenylalanine was used.Tyrosine and glutamate also induced the expression but at detectably lower levels.In contrast, methionine, valine, or glycine did not exert detectable induction (Fig. 3A).We also examined the potential dose-dependent response of the reporter to succinate and fumarate, the two most potent inducers.For succinate, induction was observed when the compound was added at 1 mM and expression plateaued at 5 mM (Fig. 3B).Interestingly, the reporter responded to fumarate differently, the expression almost reached the maximal level when 0.05 mM of this metabolite was added (Fig. 3B).
To confirm the induction observed using the gfp reporter, we constructed an A. baumannii strain in which the dotD gene on pAB3 was replaced with a dotD-flag fusion, thus allowing detection of endogenously expressed DotD by immunoblotting with the Flagspecific antibody.Consistent with results from the gfp reporter, succinate, fumarate, malate, and oxaloacetate each significantly induced the expression of dotD.Similar induction occurred in cultures receiving the testing amino acids (Fig. 3C).
To determine whether the induction of dot/icm-like genes impacts plasmid conjuga tion, we examined the effects of succinate and fumarate on the transfer of pAB3.Donor cells grown in synthetic pyruvate medium supplemented with succinate or fumarate for 2 h increased the conjugation frequency by approximately seven-and fivefold, respectively (Fig. 3D).

Expression of the dotDCB and dotIHGF gene clusters is regulated by the GacS/A two-component system
Two-component systems are the most widely used regulatory mechanism in bacteria, which control gene expression in response to diverse signals, including compounds derived from hosts, nutrients, and osmotic fluctuations (29).Among these, the GacS/A (TCS) is known to control diverse biological processes such as phenylacetic acid catabolism, biofilm, and oxidative stress in a large variety of bacteria (30,31).We thus examined how GacS/A impacts the expression of these genes by comparing the mRNA levels of these genes between the wild-type strain and a mutant lacking gacA (∆gacA) by qPCR.Deletion of gacA indeed resulted in a significant reduction in the expression of all examined dot/icm-like genes (Fig. 4A).
We further investigated the expression of these two gene clusters by constructing reporter strains in which an FACS-optimized gfp variant (28) was fused to the promoter of dotD and dotI, respectively.Consistent with the results obtained by qPCR, the intensity of GFP signals for each fusion in the ∆gacA mutant was markedly lower than that of wild type, and the defects can be fully complemented by expressing GacA from a plasmid (Fig. 4B).
To determine whether the impact of GacS/A on transcription of these two gene clusters results in changes in protein abundance, we created bacterial strains from the wild type and the ∆gacA mutant in which dotD and dotH in pAB3 were replaced with alleles that contained a Flag tag fused to the 3′ end of the genes, respectively.The level of DotD-Flag and DotH-Flag between wild type and the ∆gacA mutant was probed by immunoblotting with the Flagspecific antibody.Whereas the fusions were readily detected in strains derived from the wild-type strain, they were barely detectable in the ∆gacA mutant.Again, expression of GacA in the mutant restored the protein levels to those of the wild-type strain (Fig. 4C).Thus, GacS/A regulates the expression of at least a subset of the T4SS structural genes, particularly those in the two clusters that form operon-like structures.

The GacS/A TCS is required for efficient plasmid transfer by the T4SS on pAB3
The requirement of the GacS/A for optimal expression of a subset dot/icm-like genes essential for the transfer of pAB3 suggests that this regulatory circuit plays an important role in its conjugation and mobilization of SMPs.We thus examined the transfer of pAB3 and pKB5-Gm by mutants lacking gacS and gacA, respectively.The ∆gacS and ∆gacA mutants transferred pAB3 at efficiencies that were approximately 2 orders of magnitude lower than those of the wild-type strain.Such defects can be fully complemented by expressing the corresponding gene in the mutants (Fig. 5A).When the transfer of pKB5-Gm was measured, the mutants displayed 10-to 100-fold defects in experiments using Ab or E. coli as recipients (Fig. 5B and C).In each case, the reduction in mobilizing the plasmid can be fully restored by complementation constructs carrying the affected genes (Fig. 5B and C).The dotD and dotH in wild-type, the gacA mutant, or its complementation strain were replaced with alleles that contained a Flag tag fused to the 3′ end of the genes, respectively.Cells of overnight cultures diluted at 1:50 into a fresh medium were incubated for 4 h prior to being lysed to prepare total cell lysates.The levels of DotD-Flag and DotH-Flag were probed by immunoblotting using the Flagspecific antibody (lower panel).An antibody-detected nonspecific signal of approximately 55 kDa was used as a loading control.The levels of DotD-Flag and DotH-Flag were quantitated by measuring band intensity relative to that of the nonspecific band (upper panel).The data shown were from three independent experiments.

GacS/A controls the expression of key metabolic enzymes
The observation that some intermediates of the TCA cycle induced the expression of the dot/icm-like genes prompted us to examine whether the expression of metabolic genes, particularly those involved in connecting the TCA cycle with pathways such as gluconeogenesis, is responding to these compounds (Fig. 6A).Examination of several such genes by qPCR revealed that succinate and malate significantly induced the transcription of pckG and sdhB (succinate dehydrogenase b subunit) and mdh (malate dehydrogenase), while repressing the pdcE1a gene (Fig. 6B).The genes of pckG and pdcE1a code for the phosphoenolpyruvate carboxykinase (PEPCK) and a subunit of the pyruvate dehydrogenase complex (PDC), respectively.PEPCK catalyzes the conversion of oxaloacetate into phosphoenolpyruvate through a guanosine mononucleotide-depend ent mechanism (32), whereas the PDC is involved in the catalytic process that irreversibly converts pyruvate into acetyl-coenzyme A concomitant with the reduction of NAD + (33,34).Both of these enzyme complexes are critical for the central carbon metabolism in bacteria (35,36).
Next, we examined the impact of the GacA/S system on the expression of these genes by qPCR and GFP fusion reporter assays.Deletion of gacA led to a significant reduction in the expression of both pckG and pdcE1a genes, and the defects can be fully complemented by expressing GacA from a plasmid (Fig. 6C and D; Fig. S4A).Next, we evaluated whether the influence of GacS/A on the transcription of pckG results in changes in protein level by preparing PEPCKspecific antibodies.Indeed, the deletion of gacA significantly reduced the protein level of PEPCK, a defect that can be fully complemented by expressing GacA in the ∆gacA mutant (Fig. 6E).
We also measured the transcription kinetics and the transcriptional activity of dotD and pckG under different nutrient conditions using the GFP reporter strains.GFP signals for each fusion exhibited a robust increase from 0 to 2 h and plateaued and stabilized in the next 6 h (Fig. S3A).Consistent with the transcription of dotD (Fig. 4B), the pckG in the ∆gacA mutant was markedly lower than that of the wild type (Fig. S3A).In addition to the induction of dotD promoter (Fig. 3A), we observed significant induction of the pckG promoter by intermediates of the TCA cycle, including succinate, fumarate, and malate, as well as certain amino acids such as glutamate and asparagine (Fig. S3B).
The absence of pckG or the PDC operon is predicted to prevent the use of reaction intermediates of the TCA cycle as the sole carbon source by Ab strain (Fig. 6A, left panel).To test this hypothesis, we created the ∆pckG and ∆PDC mutants and measured their growth in minimal media supplemented with succinate, malate, or citrate as the sole carbon source.The ∆pckG mutant was unable to grow in a medium containing any of these compounds and the growth defect can be restored by a plasmid that directs the expression of PEPCK (Fig. 6F; Fig. S4B).In contrast, the ∆PDC mutant grew at rates indistinguishable from those of the wild-type strain in a medium supplemented with succinate as the sole carbon source (Fig. 6F).The lack of growth defect of the PDC mutant may be due to the presence of genes that code for the components of this enzyme complex.Indeed, bioinformatics analysis revealed that A1S_3327 and A1S_2717 are highly similar to the two components (A1S_1701 and A1S_1702) of the PDC complex (Fig. S5A and B) (12).
The expression of PEPCK depends upon the GacS/A TCS predicts that this system is required for the use of intermediates of the TCA cycles as the sole carbon source.Indeed, neither the ∆gacS nor the ∆gacA mutant can grow in minimal media contain ing succinate, malate, or citrate as the sole carbon source.Such defects can be fully complemented with plasmids expressing the affected genes (Fig. 6G; Fig. S4C).It is worth noting that the growth of gacS and gacA mutants in LB broth was indistinguishable from that of the wild-type strain (Fig. S3C).Interestingly, overexpression of PEPCK in the ∆gacA mutant cannot rescue the growth defects (Fig. S4D), suggesting that other pathways controlled by the GacS/A regulatory circuit are involved in the utilization of these compounds as the sole carbon source.

GacA recognizes DNA elements in promoters of genes involved in plasmid conjugation and central metabolism
The observation that GacS/A controls the expression of genes of distinct functions suggests that promoter regions of these genes contain common DNA elements recognizable by the response regulator of this system.We first searched for potential common motifs in these promoters using the XSTREME algorithm (37).These efforts identified three potential operators with considerable conservation from the promoter regions of dotDCB, dotIHGF, pckG, and pdcE1a, with the first motif having a high confidence (Fig. S6).
To confirm the results obtained by in silico analysis, we determined the binding affinity of the response regulator GacA to DNA fragments containing the predicted operators by electrophoretic mobility shift assays (EMSAs).To this end, we purified recombinant His 6 -GacA and added it to reactions containing IR700-labeled DNA probes expression of the genes of interest was determined by qPCR with specific primer sets.The fold change in transcript levels (vs unstimulated WT strain) was calculated.Results shown are the mean of the fold change in transcript levels (vs unstimulated WT strain) ± standard deviation (n = 3) from one representative of three independent experiments with similar results.(C and D) GacA is required for the optimal expression of pckG and pdcE1a.Cells grown as described in Fig. 4A were used to measure gene expression by qPCR with primers specific for the indicated genes.(C) The results shown are the mean of the fold change in transcript levels relative to the levels in the ∆gacA mutant ± standard deviation (n = 3) from one representative of three independent experiments with similar results.
Plasmids carrying the GFP reporter fused to the promoter of the indicated genes were introduced into the indicated strains and the intensity of the GFP signals was measured using a fluorometer.(D)The results shown are from one representative of three independent experiments with similar results.(E) GacA is required for the expression of PEPCK.The protein level of PEPCK in the indicated strains was probed with antibodies specific to the protein (lower panel).The metabolic enzyme isocitrate dehydrogenase (ICDH) was probed as a loading control.The relative intensity of the band representing PEPCK was quantitated using the bands of ICDH as reference (upper level).Similar results were obtained in three independent experiments.(F and G) PEPCK, GacS, and GacA but not PDC are required for the use of succinate as the sole carbon source.Overnight cultures of the indicated bacterial strains grown in LB were diluted at 1:1,000 into a synthetic medium containing succinate as the sole carbon source.Bacterial growth was measured by determining the values of OD 600 after incubation on a shaker for 14-16 h at 37°C.The results shown are from one representative of three independent experiments each done in triplicate.
amplified from the promoter region of dotD, dotI, pckG, or PDC.Protein-DNA complexes formed between His 6 -GacA and each of these DNA probes were readily detectable at two different protein concentrations.Furthermore, increasing the amounts of protein in the reactions led to the formation of complexes of higher molecular weight (Fig. 7A).In each case, the addition of unlabeled DNA to the reactions abolished the complexes formed by His 6 -GacA and the fluorescencelabeled probes (Fig. 7B).Thus, GacA regulates the expression of these functionally distinct gene sets by recognizing DNA elements embedded in their promoter regions.

The GacS/A is required for virulence in the Galleria mellonella infection model
The GacSA two-component system of A. baumannii has been reported to regulate multiple biological processes, including biofilm formation, phenylacetic acid metabo lism, bacterial motility, and immune response (30,38).The finding that this regulatory circuit plays an important role in plasmid mobilization and central carbon metabolism prompted us to evaluate its role in bacterial virulence with G. mellonella, which has been successfully used as a host to evaluate the virulence of diverse pathogens (39), including Staphylococcus aureus (40), Pseudomonas aeruginosa (41), and A. baumannii (42).When infected at an inoculum of 5 × 10 6 CFU, the wild-type strain caused approximately 85% mortality in the larvae, while the gacA and gacS mutants showed significantly lower mortality rates of 70% and 55%, respectively (Fig. 8A).Interestingly, compared to the wild type, the complemented strains ∆gacA::pGacA and ∆gacS::pGacS fully restored the virulence against the larvae.These complementation strains were detectably more virulent than the wild-type strain, particularly in experiments with higher inocula (1 × 10 7 CFU) (Fig. 8B).We also examined the potential role of pAB3 in the virulence of strain Ab 17978 towards G. mellonella and found that under our experimental conditions the strain cured of the plasmid displayed undiscernible defects in killing insect larvae (Fig. S7).

DISCUSSION
The LCPs found in A. baumannii strains are vehicles carrying important genes for bacterial physiology such as antibiotic resistance and expression of type VI secretion (10) critical for its survival in niches occupied by other microorganisms, including fungi (24).Here, we found that pAB3 utilizes a type IVB system similar to the Dot/Icm system of L. pneumophila for its dissemination.With the exception of the DotC homolog, which is important but not essential, all the other 11 Dot/Icm-like proteins are essential for its conjugation and for the spread of mobilizable plasmids (Fig. 1 and 2).Interestingly, 11 of the proteins important for conjugation are Dot/Icm orthologs on IncI plasmids ColIb-P9 and R64 (43).The Dot/Icm systems of both L. pneumophila (4) and Coxiella burnetii (44,45) are essential for virulence by transferring effectors into host cells.Some recent A. baumannii isolates have been shown to grow intracellularly in a manner that requires their large conjugative plasmids (14).It is known that these plasmids encode multiple regulatory proteins that control chromosomal genes (14), whether these plasmids also contribute to bacterial pathogenicity by using the Dot/Icm-like apparatus to transfer virulence factors into host cells needs further investigation.
In most cases, the expression of genes involved in the conjugation of drug-resistance plasmids appears constitutive (46).Yet, conjugation machineries are made of at least close to a dozen proteins and DNA transfer is an energy-consuming process (47).The number of proteins for building Dot/Icm-like apparatuses often exceeds 20, implying a necessity to coordinate metabolism and plasmid transfer.The TCA cycle is one branch of the bacterial central metabolism essential for the production of energy and cell-building materials.It also provides the entry points for carbon sources derived from compounds such as amino acids (48).Thus, the availability of nutrients that support the robust operation of the TCA cycle can be sensed by bacteria as signals that the environment is suitable for their growth and proliferation.Our findings that some TCA cycle intermedi ates significantly induce the expression of dot/icm-like genes indicate that metabolically favorable conditions promote the dissemination of this plasmid, which confers traits such as virulence, drug resistance, and metabolic capacity to ensure the survival of the offspring in the specific niches.Induction by compounds of central metabolic pathways also suggests that conjugation is more likely to occur when bacteria have access to these nutrients, which are more likely to present in niches with lysed cells (bacteria or eukaryotes).Given the role of GacA/S in essential cellular processes, it is not surprising it is required for optimal virulence against various hosts, including mouse, fish, and fungi (30,38).Using the G. mellonella larvae infection model, our results also demonstrate the importance of the coordination of distinct cellular processes in the fitness of A. baumannii during its interactions with hosts.Because contents of cells lysed by infection promote plasmid transfer, we speculate that conjugation may occur when multiple A. baumannii strains are associated with their hosts.The expression of T3SS expression in P. aeruginosa has been shown to be strongly affected by mutations in enzymes of the TCA cycle (49).Similarly, the TCA cycle appears to signal the switch between a pathogenic state and a mutualistic state when Photorhab dus luminescens changes hosts (50).Yet, the molecular mechanism of these responses remains elusive.Importantly, we found that A. baumannii responds directly or indirectly to intermediates of the TCA cycle via signaling mediated by the versatile GacS/A TCS system, which is known to regulate a wide range of genes involved in diverse cellular processes, including the phenylacetic acid catabolic pathway (30).Our results indicate that the promoter regions of plasmid-borne dot/icm-like genes and the TCA cycle genes responsive to GacS/A harbor common DNA elements recognizable by GacA (Fig. S6  and S7).Despite these progresses, how the signal sensor histidine kinase (HK) GacS recognizes these molecules remains largely unknown.HKs are known to respond to structurally and physiologically diverse cues, including nutrients, metal ions, fluctuations in temperature, and redox states (51).The finding that GacS/A is required for catabolism of TCA cycle intermediates such as succinate and malate by controlling the expression of genes coding for key metabolic enzymes (Fig. 6) suggests that GacA activated by diverse signals binds to the promoter of functionally diverse genes to coordinate their expression for the bacterium to better colonize specific niches (Fig. 9).
In P. aeruginosa, despite extensive genetic evidence of GacS/A-mediated regulation of central carbon metabolism via a number of sRNAs, the periplasmic detector domain of GacS did not detectably interact with any of the TCA cycle intermediates or several examined amino acids or metal ions (52).Thus, although the hypothesis that the GacS family of kinases sense these agonists by direct binding is still plausible, an alternative model is that these compounds and other stimuli may induce a yet unrecognized cellular condition that is sensed by the kinases.
Identification of small molecules that target plasmid conjugation has been consid ered an effective strategy to block the spread of antibiotic-resistance genes (53), and such efforts have gained some success.For example, 2-alkynoic fatty acids have been shown to inhibit the conjugation of the plasmid R388 by targeting the ATPase of TrwD (54), whereas B8I inhibits the transfer of plasmid pKM101 by binding TraE with a high affinity (55).Similarly, inhibition of TCSs such as the GacS/A system may be more effective in treating infections as these regulatory circuits often regulate multiple targets, ranging from virulence, metabolism, and plasmid dissemination (56).In this regard, 2-aminoimidazole compounds have been shown to be effective against PmrA (57) and BfmR (58) in A. baumannii.The discovery of a type IVB machinery responsible for plasmid conjugation in A. baumannii offers an opportunity to interfere with the spread of pAB3 and its family members, which not only are major carriers of antibiotic-resistance genes (10) but also facilitators of SMPs dissemination as we demonstrated here.In addition, given the relatively high promiscuity of type IVB systems in recognizing protein substrates as exemplified by the Dot/Icm of L. pneumophila (59,60), the possibility that the transporter on pAB3 and its relatives may directly contribute to A. baumannii virulence may open up a new exciting research avenue.

Bacterial transformation
Electrocompetent cells of A. baumannii were prepared as previously described (63).Briefly, A. baumannii grown in LB broth at 37°C for 18 h were diluted 1:100 into 100 mL LB broth, and the culture was grown in a shaker for another 24 h at 37°C.Cells were collected by centrifugation (5,000 g) for 3 min and washed twice with glycerol (10%) at 4°C. Cell suspensions in 1.5 mL glycerol (10%) were aliquoted into 100 µL and stored at −80°C.For electroporation, 500 ng (3-5 μL) plasmid DNA was mixed with 100 µL competent cells in a 0.2 cm cuvette, and electroporation was performed on a Gene Pulser Xcell Electroporation System (Bio-Rad, USA) set at 25 µF, 2.5 kV/cm, 200 Ω.After incubation at 37°C for 1 h in 1 mL warm LB, cells were plated onto LB agar supplemented with appropriate antibiotics to select transformants.
For tri-parental matings, 500 µL cells of the donor, recipient, and the helper strain MT607 containing the plasmid pRK600 (64) mixed in 1.5 mL microtubes were collected by centrifugation (5,000 g) for 3 min, and the cells were washed twice with 800 µL fresh LB broth prior to being placed onto 0.2 µm nitrocellulose membranes on LB agar.The mating was allowed to proceed for 4 h at 37°C and cells removed from membranes were serially diluted in sterile phosphatebuffered saline (PBS) and plated on a selective medium to obtain transconjugants.

DNA manipulation and plasmid construction
Plasmids and primers used in this study are listed in Table S2.DNA mini kit (TIANGEN, Cat# DP302-02) was used to isolate bacterial genomic DNA, and TIANGEN Plasmid Extraction Kits (TIANGEN, Cat# DP103-02) were used to extract plasmid DNA.Pfu-DNA polymerase (Transgen, TransStart Fast Pfu DNA Polymerase, Cat# AP221-03) was used for PCR reactions.T4 DNA ligase (Cat# M0202S) and restriction enzymes were purchased from NEB unless otherwise noted.Plasmids pJL05, pKB5-GmR, and pJB908-Gm suitable for protein expression or conjugation in A. baumannii were constructed as follows.pJL05, a derivate of pJL02 (25), which allows P TAC -driven expression of HA-tagged protein, was made by amplifying the promoter region of pJL02 with primers containing the HA tag sequence.The PCR product was inserted into pJL02 as a PciI/BamHI fragment.To make pKB5-GmR and its oriT-defective variant pJB908-Gm, the cassette harboring the gentamicin-resistance gene and its promoter was amplified with appropriate primers from pBBR1-MCS-5 (65) and was inserted into pKB5 (23) and pJB908 (26) as a SalI/SphI fragment, respectively.
For complementation and protein expression, the gene of interest amplified from the genomic DNA of A. baumannii strain 17978 was inserted into pJL05 or pKB5-GmR as BamHI/SalI fragments.

Construction of A. baumannii mutants
A spontaneous streptomycin-resistant mutant of Ab 17978 was obtained by plating 200 µL saturated culture (~2 × 10 9 cells) onto LB plates supplemented with 100 µg/mL streptomycin, and the randomly isolated strain AcbS R (25) was used for subsequent experiments.
To construct plasmids for gene deletion, an approximately 1.0 kilobase DNA fragment was amplified from upstream and downstream of the gene of interest, respectively.In each case, the primers were designed in a way such that the gene to be deleted would be replaced with an open reading frame consisting of the first and last 15 codons linked by the sequence of the restriction enzyme used to fuse the two fragments.The DNA fragments were inserted into appropriate pSR47s, a R6K-based suicide vector harboring sacB (66).
Each of the deletion plasmids was introduced into Acb strain AcbS R and the transconjugants were selected on LB plates containing streptomycin and kanamycin.Cells of transconjugants were struck onto LB plates containing 5% sucrose to allow looping out of the integrated plasmid by recombination.Mutants were identified by colony PCR using primers corresponding to the 5′ end of the upstream fragment and the 3′ end of the downstream fragment.

Plasmid conjugation assays
To evaluate the conjugation efficiency of pAB3 between the wild-type strain Ab 17978 , mutants lacking the indicated dot/icm genes, gacS, or gacA were used as donors.We first constructed an A. baumannii strain suitable for being used as the recipient by introduc ing the mariner transposon carrying a kanamycin-resistance gene (67) into strain WT -R (24) to generate the kanamycin-resistant variant WT -R (Km R ) (Table S1).Conjugation was performed as follows: cells of donor and recipient strains from overnight cultures were washed 3× with fresh LB broth and were then mixed at a 1:5 ratio in 200 µL LB broth.A total of 50 µL of mixed cells were spotted onto LB agar and the mating was allowed to proceed for 2 h prior to being resuspended in PBS.Serially diluted cells were plated onto selective LB agar containing kanamycin, sulfamethoxazole, and trimethoprim to obtain and quantitate transconjugates.The efficiency of plasmid transfer was determined by dividing the number of transconjugates by the number of donor cells.
To determine the efficiency of transferring the RSF1010 plasmid by the T4SS of A. baumannii, pKB5-Gm was first introduced into each of the testing donor strains.Strains WT -R (Km R ) and DH5α (pET28a), both resistant to kanamycin, were used as the recipient for A. baumannii and E. coli, respectively.The efficiency of plasmid transfer was determined by a procedure similar to that used for determining pAB3 conjugation with transconjugates being selected on LB agar containing kanamycin and gentamicin.

Bacterial RNA isolation and reverse transcription quantitative PCR
To isolate bacterial RNA, overnight cultures of the testing A. baumannii strains diluted in cells from 1 mL culture collected by 3-min centrifugation at 5,000 g were used for RNA isolation using a bacterial total RNA rapid extraction kit (Sangon Biotech, Cat# B518625).
For reverse transcription PCR, residual DNA was eliminated with DNase I (MeiLunbio, Cat# MB3069-1), and the first strand of cDNA was synthesized from 1 µg of total RNA using the high synthesis efficiency and high amplification efficiency RNA-to-cDNA kit (PerfectStart Uni RT and qPCR Kit, Transgen, Cat# AUQ-01) per manufacturer's instruc tions.For each reaction, 1 µL of cDNA with fivefold dilution and 50 nM primers were used for quantitative PCR (qPCR) using the SYBR Green master mix (KTSM, Cat# KTSM1401S) on a real-time PCR machine (Bio-Rad, model No. CFX Connect Optics Module) following a protocol suggested by the manufacturer.The rpoC gene (68) of A. baumannii was used as an internal reference.The software Primer3Plus (69) was used to design genespecific primers used for qPCR.All primers are listed in Table S2.The ∆∆C T method (70) was used to calculate fold changes and log 2 (fold changes) from threshold cycle (C T ) values normalized to rpoC.

GFP reporter assays
For GFP translational fusion, gfpmut3 amplified from pFV25 [Addgene plasmid# 20667 (28)] was inserted into pVRL1 (71) as a BamHI/SalI fragment to give pZF927.The promoter regions of dotD, dotI, pckG (A1S_2668), and pdcE1a (A1S_1699) were cloned into the upstream of gfpmut3 individually as an AatII/BamHI fragment, respectively (Table S1). A. baumannii strains harboring reporter plasmids were grown to early post-exponential phase in LB broth supplemented with gentamicin and kanamycin.A total of 2 × 10 9 cells washed with PBS three times were serially diluted at fivefold and transferred to 96-well plates (Coster 3631).The fluorescence signals (excitation at 480 nm/emission at 515 nm) and values of OD 600 were simultaneously measured with a Biotek Synergy H1 Multi-Mode Microplate Reader.Fluorescence units (RFU) were normalized, and the background corrected reporter fluorescence values were calculated by subtracting the RFU/OD 600 (fluorescence = fluorescence units/OD 600 ).

Antibodies and immunoblotting
The mouse-derived anti-Flag antibody, purchased from Sigma (Cat# F1804), was used to detect the plasmid of pAB3-encoded DotD-Flag and DotI-Flag.Testing strains of A. baumannii were grown as described for the GFP reporter assays.Cells corresponding to 1.0 OD 600 unit (approximately 1 × 10 9 cells) resuspended in 50 µL SDS sample buffer were boiled for 5 min.Lysates cleared by centrifugation at 6,000 g for 5 min were separated by SDS-PAGE.Proteins were transferred onto nitrocellulose membranes, which were first blocked in 5% nonfat milk for 12 h prior to being incubated in PBS containing the Flagspecific antibody (1:500 dilution) for 12 h at 4°C.Membranes washed 3× with PBS were incubated with secondary mouse IRDye 680-conjugated antibody following the protocol provided by the manufacturer (Li-Cor).The signals were detected and quantitated using an Odyssey CLx Imaging System (Li-Cor).

Protein purification and EMSAs
To produce GacA, its coding region amplified from genomic DNA of A. baumannii was inserted into pET28a (Novagen) to give pET28a-GacA, which was introduced into E. coli strain BL21(DE3) and the resulting strain was used for protein expression and purification.Briefly, overnight bacterial cultures were diluted at 1:100 into fresh LB broth containing kanamycin (30 µg/mL) and were grown on a shaker at 37°C.When OD 600 of the new cultures had reached 0.6-0.8,isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 200 µM to induce protein production.Bacterial cultures were further incubated in a rotary shaker (220 rpm/min) at 18°C for 18 h.Cells collected from 1,000 mL cultures were resuspended in 40 mL buffer (300 mM NaCl, 50 mM NaH 2 PO 4 , 10 mM imidazole, pH 8.0) and lysed using a Mini Low Temperature High Pressure Flow Cell Disrupter (JNBIO, Guangzhou, China).The lysates were cleared by centrifugation at 12,000 g for 30 min, and the soluble portion including the protein of interest was mixed with 800 µL Ni 2+ -NTA agarose beads (Qiagen, Cat# 30250) by rotation at 4°C for 1 h.The beads were loaded into a 50 mL tube equipped with a flow control device, and unbound proteins were removed by 5× washes with a washing buffer (300 mM NaCl, 50 mM NaH 2 PO 4 , 40 mM imidazole, pH 8.0).His 6 -GacA was eluted with the elution buffer (300 mM NaCl, 50 mM NaH 2 PO 4 , 250 mM imidazole, pH 8.0).Pooled fractions containing the target proteins were dialyzed against a buffer containing 25 mM Tris-HCl (pH 7.5), 200 mM NaCl, and 10% glycerol.Protein concentrations were determined by the Bradford method using a standard curve generated with serially diluted bovine serum albumin (BSA) of known concentrations (72).
For EMSA, DNA fragments containing the putative GacA binding sites for the genes of interest (approximately −300 to +10 bp relative to the predicted transcriptional start site) were amplified by PCR from the genomic DNA of A. baumannii using IR700 labeled primers (Viagene, ChangZhou, China) (Table S2).DNA probes were purified using a PCR purification mini kit (Qiagen, Cat# 28106).For each 20 µL binding reaction, 0, 4, 8, and 12 µM GacA were incubated with ~10 nM DNA probes in a binding buffer [20 mM Tris (PH 8.0), 150 mM NaCl, 10 mM EDTA, 100 µg/mL BSA, 20 mM acetyl phosphate, 10% glycerin, and 1 mM DTT] for 20 min at room temperature.For competitive EMSA, unlabeled DNA probes were added to the reactions at ratios indicated for the specific experiments.The reaction mixture was loaded onto native nondenaturing 6% polyacry lamide gels and run at 150 V for 3 h.Signals were detected using an Odyssey CLx Imaging System (Li-Cor).

Galleria mellonella infection
Overnight cultures of the wild-type, mutants, or their corresponding complementation strains diluted in 2 mL LB broth at a ratio of 1:100 were grown for 4 h at 37°C on a shaker.Bacterial cells equivalent to 2.0 OD collected by centrifugation were washed with filtersterilized PBS and resuspended to 1.0 OD/mL in PBS (approximately 1 × 10 9 cells).The bacteria were serially diluted for infection experiments with different inocula.Appropriately diluted bacteria were used to infect G. mellonella larvae by injection with 10 μL of PBS, 5 × 10 6 or 1 × 10 7 CFU.Five groups of 15 larvae were injected per experimental group.The larvae were scored as live/dead depending on their response to physical stimulus approximately every 6 h (42,73).Each G. mellonella killing assay was performed three times.Representative experiments are presented.

Statistical analyses
Quantitative data were processed and analyzed by GraphPad Prism 9 software (Graph Pad Prism, San Diego, CA, USA).Student's t-test was used to compare the mean levels between two groups each with at least three independent samples.

FIG 2 5 FIG 3
FIG 2 Dot/Icm homologs are required for plasmid conjugation.(A) The DotG homolog is essential for the conjugation of pAB3.Donor and recipient cells from saturated cultures were mixed at a 1:5 ratio for 2 h, and serially diluted samples were plated on a selective medium to obtain transconjugates.(B) Deletion of dot/icm-like genes affected the transfer of pAB3.Mutants of the indicated genes were used as donor cells in conjugation experiments described in panel A to transfer pAB3 into the recipient strain WT -R .(C and D) Wild-type, the ∆dotG mutant, and its complementation strain (C) or mutants lacking the indicated genes (D) were used as donor cells to measure the transfer of pKB5-Gm between strains of A. baumannii as described in panel A. (E) The transfer of an RSF1010 plasmid by pAB3 requires its origin of transfer (oriT).Wild-type A. baumannii harboring pKB5-Gm with or without an intact oriT were used to measure plasmid transfer as described in panel A. (F-H) The indicated A. baumannii strains were used to measure the transfer of the RSF1010-derived pKB5-Gm into E. coli strain DH5α.Experiments were performed as described in panel A. In each case, transfer efficiency was determined by dividing the number of transconjugates by the number of donor cells.Results shown in panels A-H are from one representative of three independent experiments done in triplicate with similar results.

FIG 4
FIG 4 Expression of the dotDCB and dotIHGF gene clusters is regulated by GacS/A.(A) Cells of the indicated bacterial strains from saturated cultures were diluted in fresh medium at 1:50 and the subcultures were grown for 4 h prior to being used for RNA isolation.The expression of the indicated genes was determined by qPCR with primers specific for the indicated genes.Results shown are the mean of the fold change in transcript levels (vs ∆gacA) ± standard deviation (n = 3) from one representative of three independent experiments with similar results.(B) GacA is required for the expression of dotD and dotI.Plasmids carrying the dotD-gfp or dotI-gfp fusion were introduced into the indicated A. baumannii strains and the resulting strains were used to measure the expression of the GFP reporter.Bacterial cultures were prepared as described in panel A and cells were measured for the intensity of GFP signals using a fluorometer.Results shown are from one representative of three independent experiments each done in triplicate.(C) Expression of endogenous DotD and DotH requires the GacS/A system.

FIG 5 FIG 6
FIG5 The GacS/A regulatory circuit is required for optimal pAB3 conjugation (A-C).A. baumannii strains harboring pAB3 were used as donors to measure its transfer (A), the mobilization of pKB5-Gm into A. baumannii (B), or E. coli strain DH5α (C).Cells of saturated cultures of the tested strains were mixed at a 1:5 ratio with recipient cells for 2 h.Diluted samples were plated onto a selective medium to obtain transconjugants.Conjugation efficiency was expressed as the numbers of transconjugants obtained per input donor cell.Results shown are from one representative of three independent experiments each done in triplicate.

12 FIG 7
FIG7 GacA interacts with the promoters of genes of plasmid conjugation and central metabolism.(A and B) DNA fragments encompassing the promoter region of the indicated genes were labeled with IR700 and the probes were individually incubated with purified recombinant His 6 -GacA for 20 min at room temperature, the binding mixture was resolved by native PAGE gels and the signals were detected using an Odyssey imaging system (Li-COR).In each case, inclusion of increasing amounts of GacA led to the formation of more protein-DNA complexes (A) and the protein-DNA complex formed with the IR700-labeled probe can be abolished with unlabeled DNA containing the promoter elements (B), indicative of specific binding between GacA and these DNA fragments.Arrows at the top of the panel indicate the position of GacA-probe complexes.Arrows at the bottom of the panel indicate the position of free IR700-labeled DNA.Results shown are from one representative of three independent experiments with similar results.

FIG 8
FIG 8The GacS/A TCS is required for virulence in G. mellonella larvae.(A and B) Groups of G.mellonella larvae were injected with 10 µL of the wild-type strain, mutants, or their corresponding complementation strains at doses of 5 × 10 6 CFU (A) or 10 7 CFU.(B) The viability of the larvae was assessed by monitoring melanin accumulation and motility at 6-h intervals.Survival curves were determined to be statistically significant using the Mantel-Cox test.n = 15.

FIG 9 A
FIG 9 A model for the regulation of plasmid conjugation and metabolism by the GacS/A TCS system in A. baumannii.Nutrients and potential other signals activate the GasS/A system, leading to the induction of the expression of components of the T4SS system and genes involved in metabolism.The of nutrients favors both bacterial proliferation and plasmid spread.