LpxK Is Essential for Growth of Acinetobacter baumannii ATCC 19606: Relationship to Toxic Accumulation of Lipid A Pathway Intermediates

Acinetobacter baumannii is a Gram-negative pathogen for which new therapies are needed. The lipid A biosynthetic pathway has several potential enzyme targets for the development of anti-Gram-negative agents (e.g., LpxC). However, A. baumannii ATCC 19606 can grow in the absence of LpxC and, correspondingly, of lipid A. In contrast, we show that cellular depletion of LpxK, a kinase occurring later in the pathway, inhibits growth. Growth inhibition results from toxic accumulation of lipid A pathway intermediates, since chemical inhibition of LpxC or fatty acid biosynthesis rescues cell growth upon loss of LpxK. Overall, this suggests that targets such as LpxK can be essential for growth even in those Gram-negative bacteria that do not require lipid A biosynthesis per se. This strain provides an elegant tool to derive a better understanding of the steps in a pathway that is the focus of intense interest for the development of novel antibacterials.

antibacterial targets. Since the OM is also important for protection against host immune factors, it is unlikely that cells lacking the LOS-containing OM would survive during many infections (10). Furthermore, such LPS-deficient organisms are hypersusceptible to multiple antibiotics; thus, inhibitors of lipid A biosynthesis may be expected to potentiate the activity of many antibiotics, opening the possibility of combination therapies. Although enzymes such as LpxA and LpxC are not essential in A. baumannii ATCC 19606, at least under standard laboratory conditions, other enzymes in the lipid A biosynthetic pathway may be essential for growth since several pathway intermediates, such as lipid X and UDP-2,3-diacyl-GlcN (which have critical micelle concentrations below 25 M), are detergent-like (26) and blocking certain pathways steps might cause the toxic accumulation of these intermediates. Indeed, we previously showed that the phosphodiesterase LpxH is essential for growth of A. baumannii ATCC 19606 using an isopropyl ␤-D-1-thiogalactopyranoside (IPTG)-regulated expression system (27). Depletion of LpxH also caused cell envelope defects and accumulation of detergent-like lipid A intermediates (27). Furthermore, the growth defect due to LpxH depletion was ameliorated by chemical inhibition of LpxC, upstream of LpxH, presumably by blocking the synthesis of detergent-like intermediates and thereby preventing their accumulation (27).
Quantitative modeling of lipid A biosynthesis in E. coli, where LPS biosynthesis is essential, predicted that LpxK, a kinase acting downstream of LpxH, may be the rate-limiting enzyme and as such a particularly attractive drug target within the lipid A pathway (12,13). We therefore asked whether loss (and by extension, chemical inhibition) of LpxK might also have the potential to be directly antibacterial. An IPTGinducible LpxK expression strain of A. baumannii ATCC 19606 was unable to grow in cation-adjusted Mueller-Hinton (MH) medium lacking IPTG, indicating that LpxK is essential for growth. Under LpxK depletion conditions, we observed changes in cell morphology and significant accumulation of the detergent-like lipid A intermediates DSMP and lipid X, as measured by liquid chromatography-mass spectrometry (LCMS). Consistent with this, chemical inhibition of LpxC allowed growth in the absence of IPTG and reversed the accumulation of these intermediates. Intriguingly, we also found that inhibitors of fatty acid biosynthesis could similarly restore growth and prevent the accumulation of toxic intermediates.

RESULTS
LpxK is required for A. baumannii ATCC 19606 growth under standard laboratory conditions. A. baumannii ATCC 19606 is capable of growth under standard laboratory conditions in the absence of LOS, and the LpxA, LpxC, and (possibly) LpxD enzymes that catalyze early steps of lipid A biosynthesis were shown to be dispensable for growth (11,28). We recently demonstrated that LpxH, which catalyzes a step in lipid A biosynthesis downstream of these enzymes, is required for growth of A. baumannii ATCC 19606 under standard laboratory conditions. The loss of LpxH function led to toxic accumulation of detergent-like pathway intermediates (27). LpxK, an integral membrane kinase responsible for the phosphorylation of DSMP, has recently been proposed as an attractive target in the lipid A pathway, based on modeling predictions suggesting that it may catalyze the rate-limiting step in lipid A biosynthesis in E. coli (wherein LPS is essential [12]). To determine whether LpxK is also essential for growth in A. baumannii ATCC 19606, we attempted to delete lpxK on the genome. Attempts to delete lpxK via homologous recombination were unsuccessful, suggesting that LpxK was essential for growth in A. baumannii ATCC 19606. To confirm this, we first constructed an IPTG-inducible strain wherein the P tac promoter and the lacI gene were inserted directly into the chromosome upstream of lpxK (strain JWK0013). However, in A. baumannii ATCC 19606, lpxK is the fourth gene within a predicted 13-gene operon containing multiple essential genes downstream of lpxK (e.g., encoding DNA polymerase III); therefore, any growth defect occurring in the absence of induction would not be attributable specifically to loss of LpxK. To deconvolute the determinants of such defects, we constructed 2 plasmids, pNOV043 and pNOV044; plasmid pNOV043 harbors a native promoter driving lpxK expression and includes all of the cotranscribed genes downstream of lpxK (and would therefore fully complement downregulation of the transcript in the absence of IPTG), and pNOV044 is identical to pNOV043 except that it lacks lpxK (and therefore harbors only the relevant genes located downstream of lpxK). Strain JWK0013(pNOV044) (depicted in Fig. 1A) thereby constitutes a specifically lpxKregulated expression strain. When cultured on medium with or without IPTG, JWK0013 (no plasmid) and JWK0013(pNOV044) required IPTG for growth, whereas growth of JWK0013(pNOV043) was not IPTG dependent (Fig. 1B). This finding strongly supports the hypothesis that lpxK is essential for the growth of A. baumannii ATCC 19606. We confirmed this result in broth culture by demonstrating that growth of JWK0013 (pNOV044) is dependent on IPTG in a dose-dependent manner (Fig. 1C).
LpxK depletion caused morphological changes to cells. To examine the effect of LpxK depletion on A. baumannii ATCC 19606 cellular morphology, we identified an appropriate inoculum of JWK0013(pNOV044) (1:8) to subculture into noninducing (ϪIPTG) medium so that progressive depletion of LpxK would reduce growth over time in liquid culture. We chose conditions (inocula) in which culture growth leveled off at an optical density at 600 nm (OD 600 ) of approximately 0.4 to ensure that growth cessation could be clearly observed while providing an adequate quantity of cells for fluorescence and transmission electron microscopy (3 h time point of the growth curve in Fig. 1D). We also collected samples from the parental strain and from an A. baumannii ΔlpxC strain for comparison. Depletion of LpxK resulted in abnormal cell morphology, with elongated and bent cells compared to the parent strain observed by the use of fluorescence microscopy (Fig. 2). This differed from the phenotype of the ΔlpxC mutant, where cells were enlarged and rounded and clumped together (Fig. 3). Electron microscopy also revealed an aberrant inner membrane, blebbing of cell envelope, and increased vesicle formation in cells depleted for LpxK (Fig. 4). These apparent envelope defects are consistent with the notion of detergent-like lipid A pathway intermediates potentially accumulating in the inner membrane and causing its deformation by inducing curvature.
Depletion of LpxK causes accumulation of LOS pathway intermediates in A. baumannii ATCC 19606. Depletion of LpxK affected bacterial growth and caused morphological changes to cells, possibly reflecting toxic accumulation of lipid A biosynthetic pathway intermediates. To test this hypothesis, we used LCMS to directly compare lipid A pathway intermediate levels for strain JWK0013(pNOV044) cultured with or without induction of LpxK expression by IPTG. For the noninduced sample, we subcultured JWK0013(pNOV044) (1:20) into noninducing (ϪIPTG) medium so that progressive depletion of LpxK would reduce growth over time in liquid culture while providing an adequate amount of cells for quantification of lipid A pathway intermediates (OD 600 ,~0.5). We observed significant accumulations of lipid A pathway intermediates, including DSMP and lipid X, in A. baumannii depleted for LpxK (Fig. 5). These findings supported our hypothesis that toxic accumulations may contribute to growth inhibition upon loss of LpxK function. Intriguingly, we also observed a small but reproducible decrease in the levels of LpxA and LpxC products (Fig. 5), despite the fact that these enzymes occur upstream of the block at LpxK. This suggested the possibility that A. baumannii ATCC 19606 may be able to decrease substrate flux into the lipid A biosynthetic pathway upon sensing certain toxic accumulations, but confirmation of this would require further exploration.

Inhibition of LpxC rescues growth of cells depleted for LpxK and alleviates toxic intermediate accumulation.
If the failure of JWK0013(pNOV044) to grow upon depletion of LpxK was indeed caused by the toxic accumulation of lipid A pathway intermediates, we would expect the growth of the strain to be restored by inhibiting LpxC. Supporting this idea, growth of JWK0013(pNOV044) was no longer IPTG dependent when the LpxC inhibitor CHIR-090 (29)(30)(31)(32) was included in either solid growth medium (Fig. 6A, left side of the panel) or liquid cultures (Fig. 6B). Growth rescue was dose dependent, and it occurred at CHIR-090 concentrations shown previously (27) to dramatically reduce LOS production in A. baumannii ATCC 19606 ( Fig. 6B and C). Similar growth rescue by CHIR-090 was seen when tryptic soy agar plates were used (data not shown). Furthermore, JWK0013(pNOV044) subcultured in the absence of IPTG under conditions (1:50 dilution) that required CHIR-090 (8 g/ml) to grow and reach a culture OD 600 of 0.5 accumulated much less DSMP and lipid X than the same strain grown under noninducing conditions using a 1:20 subculture dilution in the absence of CHIR-090, which reached an OD 600 of 0.5 (Fig. 7). Therefore, the decrease in accumulation of detergent-like pathway intermediates occurred concomitantly with A. baumannii growth restoration mediated by inhibition of LpxC.

Inhibitors of fatty acid biosynthesis rescue growth of cells depleted for LpxK and reduce accumulation of DSMP and lipid X.
A substantial body of work has demonstrated that there are important relationships between the fatty acid and lipid A biosynthetic pathways in Gram-negative bacteria. For example, in E. coli, fabZ mutations decrease susceptibility to LpxC inhibitors, and accD and fabH mutations can rescue the lethality of defects in the LPS transporter proteins LptF and LptG. We therefore tested whether the fatty acid biosynthesis (␤-ketoacyl-acyl carrier protein synthase) inhibitor cerulenin (33) and the recently reported acetyl-CoA synthesis (biotin carboxylase, acetyl-CoA carboxylase [Acc]) inhibitor pyridopyrimidine (34) could rescue the growth of JWK0013(pNOV044) under noninducing conditions. Interestingly, growth of LpxK-depleted cells (ϪIPTG) was rescued by exposure to either compound, and the rescue effects were dose dependent, albeit they were not as robust as those observed for LpxC inhibition by CHIR-090 (Fig. 6A, right side of the panel). Similar growth rescue by these compounds was seen when tryptic soy agar plates were used (data not shown). Furthermore, JWK0013(pNOV044) subcultured at a dilution (1:50) that required cerulenin (8 g/ml) or pyridopyrimidine (16 g/ml) treatment to reach an OD 600 of 0.5 had a significantly lower accumulation of DSMP and lipid X than the same strain under noninducing conditions using a 1:20 subculture dilution in the absence of either compound, which reached an OD 600 of 0.5 (Fig. 7). The relative decrease in the levels  of these metabolites was larger in CHIR-090-treated cells and smaller in cells treated with cerulenin or pyridopyrimidine. This suggests that inhibition of fatty acid biosynthesis by either of these compounds might also serve to reduce the toxic accumulation of pathway intermediates caused by loss of LpxK, restoring growth. It is of interest that when growth rescue is accomplished by blocking the early stages of lipid A biosynthesis with LpxC inhibitor CHIR-090, as described above, the growth-rescued cells would be expected to lack LOS and to not have an intact OM. Supporting this expectation, CHIR-090 could rescue cells on Mueller-Hinton broth (MHB) agar but not on MacConkey agar, which contains toxic bile salts, presumably reflecting a defect (lack) of an OM permeability barrier (Fig. 6A, bottom panel). Interestingly, growth rescue by cerulenin or pyridopyrimidine also occurred on MHB agar but not on MacConkey agar (Fig. 6A), indicating that although growth per se could be restored, the cells apparently still exhibited a permeability defect. Possibly explaining this, cellassociated LOS in the chemically rescued cells was not restored to levels comparable to those seen with wild-type ATCC 19606 or the LpxK-regulated strain [JWK0013 (pNOV044)] grown in the presence of IPTG, further suggesting that the OM was still defective in the chemically rescued cells (Fig. 6C).  The following day, cells were washed repeatedly and resuspended to an OD 600 of 0.01, and a 100-l volume was plated on MHB agar or MacConkey agar plates without IPTG. Sterile filter discs spotted with IPTG, DMSO, CHIR-090, pyridopyrimidine, or cerulenin were placed on the plates, which were then incubated at 37°C for 24 h (cerulenin was incubated for 72 h). Growth of JWK0013(pNOV044) was restored in the presence of IPTG on both media. Growth of JWK0013(pNOV044) was not observed under noninducing conditions (minus IPTG and DMSO). JWK0013(pNOV044) grew under noninducing conditions in the presence of CHIR-090 (LpxC inhibitor), pyridopyrimidine (acetyl-CoA-carboxylase inhibitor), or cerulenin (␤-ketoacyl-acyl carrier protein synthase inhibitor) on MHA but not MacConkey agar. (B) An overnight culture of JWK0013(pNOV044) grown under inducing conditions (ϩIPTG) was diluted to an OD 600 of 0.1 and then was diluted 100-fold into MHB containing 10% alamarBlue. Next, 100 l of the inoculum was added to the wells of a 96-well plate containing CHIR-090, pyridopyrimidine, or cerulenin to final assay concentrations ranging from 0 to 128 g/ml. The plate was incubated for 6 h at 37°C before fluorescence (excitation, 545 nm; emission, 590 nm) was read on a SpectraMax microplate reader, and data were processed with Softmax Pro software v 5.

DISCUSSION
Identification of bacterial proteins whose functional inhibition can cause toxic accumulation of pathway intermediates could inform selection of antibacterial targets. For example, although lipid A biosynthesis itself is essential per se in organisms such as E. coli, an enzyme such as LpxD, within the lipid A biosynthetic pathway, has been implicated as a particularly good candidate antibacterial target since its substrate is detergent-like and therefore cells may be particularly sensitive to inhibition of LpxD (35). However, whether toxic accumulations are playing a role in growth inhibition is difficult to fully explore when the product of the pathway itself (e.g., lipid A) is essential for growth. The ability of A. baumannii ATCC 19606 to grow without lipid A biosynthesis, which can be caused by inactivation of genes encoding enzymes catalyzing early steps of the pathway (e.g., lpxC), provides an elegant platform with which to explore the potential role of toxic intermediate accumulation in mediating the essentiality of certain catalytic steps that occur later in the lipid A biosynthetic pathway. We previously demonstrated, using a strain with controlled expression of lpxH, that LpxH was essential for growth in A. baumannii ATCC 19606 due to this phenomenon. Here we extend this finding to include LpxK, which acts later in the pathway than LpxH (see Fig. S4 in the supplemental material). LpxK has also drawn recent interest as a potential antibacterial target since it may be represent a rate-limiting step in lipid A biosynthesis (12,13). Here we show that A. baumannii ATCC 19606 cells depleted for LpxK fail to grow, suggesting that inhibitors of LpxK may be antibacterial for strains where LOS is not essential, such as A. baumannii ATCC 19606. Therefore, the antibacterial potency of LpxK inhibitors should be measurable using routine antimicrobial susceptibility testing protocols, such as those outlined by the Clinical and Laboratory Standards Institute (CLSI; M100, 27th edition).
The accumulation of DSMP and lipid X observed upon depletion of LpxK strongly indicates that toxic accumulation(s) plays a role in the growth phenotypes of depleted strains. The ability of the LpxC inhibitor CHIR-090 to rescue growth of cells downregulated for LpxK strongly supports this hypothesis. In the case of A. baumannii ATCC 19606, our observations explain why LpxK (and, previously, LpxH) is essential for growth. In Gram-negative pathogens where LPS itself is essential (e.g., E. coli or P. aeruginosa), such toxic accumulations could mean that bacteria could be particularly sensitive to inhibition of targets such as LpxK due to the contribution of both inhibition of lipid A synthesis and toxic precursor accumulation.
Intriguingly, we also observed that exposure of LpxK-depleted A. baumanni ATCC 19606 to inhibitors of fatty acid biosynthesis (cerulenin and pyridopyrimidine) rescued the growth defect caused by loss of LpxK and reduced the accumulation of DSMP and lipid X. The fatty acid biosynthetic pathway is necessary for biosynthesis of both the inner membrane (phospholipid bilayer) and the OM (phospholipid inner leaflet and LPS outer leaflet) and in organisms such as E. coli, where the balance of the levels of biogenesis of these membranes is crucial for bacterial growth. This is further reflected in the observation that mutations in fatty acid genes (fabZ in E. coli [36] and fabG in P. aeruginosa [37]) decreased susceptibility to LpxC inhibitors, which was proposed to result from increased substrate flux into the lipid A pathway (38). A reduction in LpxC levels was observed in E. coli fabZ mutants, suggesting a regulatory relationship between the pathways (36). Presumably, strains like A. baumannii ATCC 19606 would also need to balance phospholipid (PL) and LOS levels under normal conditions when LOS is being made, so as to generate a proper cell envelope and to maintain the asymmetric permeability barrier of the OM. The absence of PBP1A and the corresponding upregulation of several surface-associated lipoproteins were shown to underlie the ability of strains like A. baumannii ATCC 19606 to tolerate complete loss of lipid A (25), but how the cells respond to lipid A loss in terms of fatty acid biosynthesis remains to be elucidated (11,39,40). The impact of inhibiting fatty acid synthesis in reversing the LpxK dependency of A. baumannii ATCC 19606, demonstrated here, may mainly reflect relief of toxic lipid A precursor accumulation, since rescued cells show a significant reduction of DSMP and lipid X accumulation concomitant with growth restoration, but this remains to be conclusively shown. However, growth rescue of LpxK-depleted cells via inhibition of either fatty acid biosynthesis or LpxC was successful only when cells were grown on typical laboratory media such as Mueller-Hinton agar (MHA) but not MacConkey agar, suggesting that the rescued cells lack a functional OM permeability barrier. Supporting this, cell-associated LOS levels were not restored to normal during chemical rescue. This is expected in the case of LpxC inhibition, which blocks lipid A synthesis at an early stage, making it highly unlikely that a normal OM could be reestablished in the growth-rescued cells. The inability to grow on MacConkey agar and reduced cell-associated LOS levels seen in cells rescued by inhibition of fatty acid biosynthesis suggest that growth rescue in that case may reflect a reduction of substrate flow into lipid A biosynthesis. Intriguingly, accD and fabH mutations were able to suppress the lethality of an E. coli strain with mutations affecting LptF and LptG (41), which mediate transport of LPS to the OM. The suppression of lethality of the E. coli lptFG mutation by accD and fabH mutations might also involve a reduction of toxic intermediates. Since E. coli requires LPS for growth, these mutations may reduce substrate flow into the lipid A pathway, first relieving (preventing) the toxic accumulation and ultimately reestablishing the balance between PL and LPS levels according to the residual rate of LPS transport mediated by the partially defective LptFG transport proteins. In contrast to what we have observed here, those suppressor mutants can have an intact OM permeability barrier, suggesting proper restoration of an LPS-containing OM (41). The exact mechanism by which inhibition of fatty acid biosynthesis mediates growth rescue in A. baumannii remains to be fully elucidated, but our data suggest that reduction of the accumulation of toxic intermediates is at least partly responsible.
Finally, it may be useful to identify nodes of antibacterial synergy within essential biosynthetic pathways, as exemplified by the synergy seen for trimethoprimsulfamethoxazole treatments, which inhibit the dihydrofolate reductase and dihydropteroate synthetase enzymes in the folate biosynthesis pathway (42). Here we have explored the potential of toxic intermediate accumulation in the context of inhibiting individual targets within the lipid A biosynthetic pathway. However, pathways having potentially toxic intermediates may also be intrinsically subject to growth rescue by relief of this accumulation, as shown here and elsewhere, which could be observed as antagonism between inhibitors of certain individual targets. In the case of A. baumannii ATCC 19606, this may be less of an issue for the lipid A pathway, since growth-rescued cells appear to have highly defective envelopes, but this revives the notion that the characterization of potential synergies in biosynthetic pathways must be carefully and systematically undertaken.

Construction of IPTG-inducible LpxK strain JWK0013(pNOV044) in A. baumannii ATCC 19606.
To make the lpxK gene on the chromosome of A. baumannii ATCC 19606 regulated by IPTG, the P tac promoter and lacI were inserted in front of the lpxK gene on the chromosome to generate strain JWK0013. The upstream region of lpxK was amplified from strain A. baumannii ATCC 19606 genomic DNA using primers KTT433 and KTT434; KTT434 attaches a linker to the aph(Km r ) (aph kanamycin resistance) cassette on the 3= end of the fragment. The aph(Km r ) cassette was amplified using primers KTT85 and KTT86 from plasmid pC009. The lacI gene and P tac promoter were amplified from pNOV002 using primers KTT238 and KTT239; KTT239 attaches a linker to the aph(Km r ) cassette on the 5= end of the fragment. The lpxK gene was amplified from A. baumannii ATCC 19606 genomic DNA using primers KTT435, which incorporates a linker to the P tac promoter on the 5= end of the fragment, and KTT436. The final construct was generated by overlap extension PCR (43) to ligate the upstream fragment, the aph(Km r ) cassette, the lacI gene and P tac fragment, and the downstream fragment. The resultant DNA constructs were then transformed into A. baumannii ATCC 19606 via electroporation. One of the resulting transformants, JWK0013, was confirmed for correct integration of the constructs by PCR and sequencing using primers cPCR lpxK F and cPCR lpxK R. Plasmid pNOV043 was generated by cloning the promoter of the putative operon containing the lpxK gene (lpxK is the fourth gene of the operon) and the genes downstream of lpxK into the backbone generated from pNOV018, which contains the pBR322 ori, pWH1266, and P Km ::lacI and the aacC1 gentamicin resistance cassette. Plasmid pNOV044 is similar to pNOV043 except that it lacks the lpxK gene. Both plasmids are deposited in GenBank (see below). The primers used to construct pNOV043 and pNOV044 are listed in Table S1 in the supplemental material. The vectors were verified by sequencing and were electroporated into JWK0013 to generate JWK0013(pNOV043) and JWK0013(pNOV044) (Fig. 1B).
Growth curve of JWK0013(pNOV044) with different concentrations of IPTG. JWK0013(pNOV044) (Ϫ80°C) was streaked onto an LB plate with gentamicin and IPTG. The following day, a bacterial suspension was prepared using a BBL prompt inoculation system (Becton, Dickinson and Company, Franklin Lakes, NJ) and further diluted 1:1,000 in MHB. In a 96-well flat-bottom plate, IPTG was serially diluted in MHB and the diluted bacterial suspension was applied to the serially diluted IPTG solution. The plate was incubated in a Spectramax detection platform (Softmax Pro software v 5.4.1) at 37°C with shaking, and the contents were measured every minute. The results were then analyzed and plotted in Excel (Fig. 1C).
Monitoring cellular morphology under conditions of LpxK depletion. The A. baumannii ATCC 19606 parental strain, the A. baumannii ΔlpxC mutant, and JWK0013(pNOV044) were grown overnight in MHB. Strain JWK0013(pNOV044) was grown overnight in MHB with 50 g/ml gentamicin and 1 mM IPTG for induction of LpxK expression. The cells were diluted 1:100 the next day in 50 ml MHB (A. baumannii ATCC 19606 or A. baumannii ⌬lpxC) or 50 ml MHB-1 mM IPTG [JWK0013(pNOV044)] and grown at 37°C with shaking. When the culture reached an OD 600 of~0.5, cells were collected by centrifugation, washed twice with MHB, suspended in same volume of MHB, and then diluted to 1:8 in 300 ml MHB in a 1-liter flask. These cultures were grown at 37°C with shaking, and samples were collected every 30 min for OD 600 measurement, CFU counting, and microscopy. Figure 1D shows the representative growth curve determined by OD 600 measurement. For fluorescence microscopy, the cells were collected and 25% glutaraldehyde (Sigma G6257) was added to reach a final concentration of 2%. The cells were fixed on 1.2% agar on the slide, stained with either 100 g/ml 4=,6-diamidino-2-phenylindole (DAPI) (catalog no. 62248; Thermo Scientific) or 10 g/ml FM1-43fx (catalog no. F35355; Life Technologies, Inc.), and observed using a Nikon Eclipse Ti inverted microscope with a Nikon halogen illuminator (D-LH/LC), a Sola light engine (Lumencor, Beaverton, OR), and a Clara Interline charge-coupled-device (CCD) camera (Andor, South Windsor, CT). A Nikon CFI Plan Apo Lamda DM ϫ100 oil objective lens (1.45 numerical aperture [NA]) was used for phase-contrast and fluorescence imaging. For FM 1-43fx images, we used a fluorescein isothiocyanate (FITC)-5050ANTE-ZERO filter set (Semrock, Rochester, NY). The DAPI images were taken by using a BFP-A-Basic-NTE filter set (Semrock). The exposure times for DAPI images and green fluorescent protein (GFP) images were 500 ms and 100 ms, respectively. Images were captured by using Nikon Elements software and exported for figure preparation in ImageJ (44). For thin-section transmission electron microscopy, cells were pelleted by centrifugation at 4,000 ϫ g for 10 min at room temperature and then resuspended and fixed with Tousimis fixative (1.5% glutaraldehyde-1% formic acid-0.12 M Sorensen's buffer) from Tousimis Research Corporation. The samples were submitted for embedding and thin-section preparation to the Center for Biophotonics Science & Technology of the University of California, Davis (Sacramento, CA, USA).
Liquid chromatography-mass spectrometry (LCMS) detection of lipid A precursors. A. baumannii IPTG-regulated lpxK strain JWK0013(pNOV044) was grown overnight in MHB supplemented with 1 mM IPTG to induce LpxK expression. The following day, the cells were diluted to an OD 600 of 0.05 in 50 ml of MHB-1 mM IPTG and grown at 37°C with shaking until an OD 600 of 0.5 was reached. Cells were then collected by centrifugation, washed twice with MHB, and suspended in MHB at the original volume of 50 ml. The cell suspension was then diluted 1:20 in 50 ml of fresh MHB with or without IPTG in 250-ml flasks. The cultures were grown at 37°C with shaking, and samples were collected every hour for OD 600 measurement. When the cultures reached an OD 600 of 0.5, a 5-ml volume was removed and frozen at Ϫ80°C. For the rescue experiments, A. baumannii JWK0013(pNOV044) was again grown overnight in MHB supplemented with 1 mM IPTG for induction of LpxK expression. The cells were diluted in 50 ml MHB-1 mM IPTG to an OD 600 of 0.05 the next day and grown at 37°C with shaking until the OD 600 reached 0.5. When the culture reached an OD 600 of~0.5, cells were collected by centrifugation, washed twice with MHB, and suspended in same volume of MHB and then diluted to 1:50 in MHB, and 10 ml was added to a 50-ml conical tube with 1 mM IPTG, 8 g/ml CHIR-090, 16 g/ml pyridopyrimidine, or 8 g/ml cerulenin. These cultures were grown at 37°C with shaking (220 rpm), and samples were collected for OD 600 measurement. Once the cultures reached an OD 600 of~0.5 (after~2 h 30 min for IPTG,~3 h 15 min for CHIR-090,~5 h for pyridopyrimidine, and~7 h for cerulenin), the sample OD 600 was adjusted to 0.5, a 5-ml volume was removed and placed in a 15-ml Falcon tube and centrifuged at 10,000 rpm, and the supernatant was removed and the pellet placed at Ϫ80°C until LCMS could be performed. Each experiment was performed in triplicate at least three times. LCMS analyses were done as described previously (27,45), and data were analyzed in Skyline (46). The two-tailed Student's t test was used for statistical analysis. The predicted lipid A biosynthetic pathway in A. baumannii ATCC 19606 with corresponding m/z values is shown in Fig. S1.
Growth rescue of JWK0013(pNOV044) by the LpxC and fatty acid biosynthesis inhibitors under noninducing conditions. To determine if inhibiting LpxC or fatty acid biosynthesis could rescue growth during LpxK depletion, JWK0013(pNOV044) was grown overnight at 37°C on MHA supplemented with 1 mM IPTG (Calbiochem). The following day, cells were suspended in 1 ml of MHB (1.5-ml microcentrifuge tube), collected by centrifugation, and suspended in fresh MHB for a total of 3 washes to remove trace amounts of IPTG. After the final wash, cells were suspended in 5 ml of MHB and the OD 600 was adjusted to 0.01. Next, 100 l of the cell suspension was spread on a fresh MHA plates and allowed to dry. Sterile paper disks (Remel; catalog no. R55054) were added to the center of the plates and inoculated with 10 l of dimethyl sulfoxide (DMSO) (Sigma), CHIR-090 (LpxC inhibitor [29]; 12.8 mg/ml), cerulenin (␤-ketoacylacyl carrier protein synthase inhibitor [33]; 12.8 mg/ml), or pyridopyrimidine (acetyl-CoA-carboxylase inhibitor [34]; 12.8 mg/ml). The plates were incubated at 37°C for 24 to 72 h before images were taken. Images were taken by the use of a Bio-Rad Universal Hood III system with Image Lab v 5.1 software. This procedure was repeated on MacConkey agar.
To establish the absolute concentrations of CHIR-090, pyridopyrimidine, and cerulenin that rescue growth of JWK0013(pNOV044), a broth-based assay was used. Compounds were dissolved in DMSO at 12.8 mg/ml (a concentration 100-fold higher than the final assay top concentration of 128 g/ml). In a 96-well plate, sequential 2-fold serial dilutions of the content of wells 3 to 11 were made in DMSO (corresponding final assay concentrations of 0.5 to 128 g/ml), leaving well 1 as the medium-only control, well 2 as the DMSO vehicle control, and well 12 for the IPTG (1 mM final concentration) growth control. Using a 12-channel electronic pipette, 1 l of each 100ϫ drug concentration, including the DMSO-only control, was transferred into a new 96-well U-bottom plate (Greiner Bio-One; catalog no. 650162). Next, using an 8-channel electronic pipette, the drug dilutions were then diluted 100-fold with the overnight inoculum for 6 to 8 h at 37°C before fluorescence reading was performed (excitation, 545 nm; emission, 590 nm) on a SpectraMax detection platform with Softmax Pro software v 5.4.1. To generate the overnight culture, JWK0013(pNOV044) was grown overnight in MHB with 1 mM IPTG for induction of LpxK expression. The following day, the cells were diluted to an OD 600 of 0.1 and a further 100-fold in fresh MHB media with 10% alamarBlue (Bio-Rad). The experiment was done in triplicate.
LOS quantity determined by gel electrophoresis. For sample collection, A. baumannii JWK0013(pNOV044) was grown overnight in MHB-1 mM IPTG for induction of LpxK expression. The cells were diluted the next day to an OD 600 of 0.05 in 50 ml MHB-1 mM IPTG and grown at 37°C with shaking for 2.5 h in a 250-ml flask until the OD 600 reached 0.5. When the culture reached an OD 600 of~0.5, cells were collected by centrifugation, washed twice with MHB, and suspended in same volume of MHB and then diluted to 1:50 in 5 ml MHB in a 50-ml conical tube with or without IPTG and 8 g/ml CHIR-090, 16 g/ml pyridopyrimidine, or 8 g/ml cerulenin. To obtain the same cell mass, the culture without IPTG alone (without rescuing chemicals) was diluted to 1:20. These cultures were grown at 37°C with shaking (220 rpm), and samples were collected for OD 600 measurement. Once the cultures reached an OD 600 of 0.5, the samples were collected in a microcentrifuge tube. The tube was spun down at 10,000 rpm, the supernatant was removed, and the pellet was placed at Ϫ80°C until an LOS gel experiment could be performed as previously described (45).
Accession number(s). The sequences of the pNOV043 and pNOV044 plasmids were deposited in GenBank under accession numbers KY933087 (pNOV043) and KY933088 (pNOV044).

ACKNOWLEDGMENT
We thank David A. Six for helpful discussions.