Loss of YhcB results in overactive fatty acid biosynthesis

ABSTRACT Loss of the Escherichia coli inner membrane protein YhcB results in pleomorphic cell morphology and clear growth defects. Prior work suggested that YhcB was directly involved in cell division or peptidoglycan assembly. We found that loss of YhcB is detrimental in genetic backgrounds in which lipopolysaccharide (LPS) or glycerophospholipid (GPL) synthesis is altered. The growth defect of ΔyhcB could be rescued through inactivation of the Mla pathway, a system responsible for the retrograde transport of GPLs that are mislocalized to the outer leaflet of the outer membrane. Interestingly, this rescue was dependent upon the outer membrane phospholipase PldA that cleaves GPLs at the bacterial surface. Since the freed fatty acids resulting from PldA activity serve as a signal to the cell to increase LPS synthesis, this result suggested that outer membrane lipids are imbalanced in ΔyhcB. Mutations that arose in ΔyhcB populations during two independent suppressor screens were in genes encoding subunits of the acetyl coenzyme A carboxylase complex, which initiates fatty acid biosynthesis (FAB). These mutations fully restored cell morphology and reduced GPL levels, which were increased compared to wild-type bacteria. Growth of ΔyhcB with the FAB-targeting antibiotic cerulenin also increased cellular fitness. Furthermore, genetic manipulation of FAB and lipid biosynthesis showed that decreasing FAB rescued ΔyhcB filamentation, whereas increasing LPS alone could not. Altogether, these results suggest that YhcB may play a pivotal role in regulating FAB and, in turn, impact cell envelope assembly and cell division. IMPORTANCE Synthesis of the Gram-negative cell envelope is a dynamic and complex process that entails careful coordination of many biosynthetic pathways. The inner and outer membranes are composed of molecules that are energy intensive to synthesize, and, accordingly, these synthetic pathways are under tight regulation. The robust nature of the Gram-negative outer membrane renders it naturally impermeable to many antibiotics and therefore a target of interest for antimicrobial design. Our data indicate that when the inner membrane protein YhcB is absent in Escherichia coli, the pathway for generating fatty acid substrates needed for all membrane lipid synthesis is dysregulated which leads to increased membrane material. These findings suggest a potentially novel regulatory mechanism for controlling the rate of fatty acid biosynthesis.

and cell shape maintenance.Notably, these potential YhcB interaction partners were not verified with additional experiments.
Due to a potential yet unverified interaction with RodZ, YhcB was examined as a cell division protein by visualizing Z-ring formation.Loss of YhcB resulted in mislocalized FtsZ, and the authors concluded that YhcB is involved in cell division (11).However, YhcB has never been shown to directly interact with FtsZ or co-localize with any proteins required for Z-ring assembly.Furthermore, it is important to note that disrupting the synthesis or transport of LPS or GPLs can also result in cell division defects (7,(18)(19)(20).
A recent transposon mutagenesis screen of a yhcB mutant (13) indicated that perturbation of cell division machinery or disruption of genes impacting the synthesis of LPS, enterobacterial common antigen, or peptidoglycan reduces bacterial fitness and may lead to cell death.It is apparent that the loss of YhcB results in cell-wide dysregula tion, and for that reason, several groups have concluded that YhcB must be involved in multiple biogenesis pathways (11,13).However, the striking cellular defects arising from loss of YhcB confound broad-scale analysis.
It has been established in the literature that deletion of yhcB results in filamentous cells that are highly sensitive to envelope stresses (11)(12)(13).Increased antibiotic sensitivity can be the result of membrane permeability defects, often caused by disruption of OM asymmetry.Published TnSeq data from our laboratory (7) and others (13) show that yhcB becomes important when lpxM is chromosomally deleted, suggesting that YhcB may have a role in membrane synthesis.Our data support published accounts that show yhcB mutants are pleomorphic with severe fitness defects.Here, we find that cells lacking YhcB appear to aberrantly generate fatty acids which preferentially shuttle into  found in E. coli are shown (bottom).(B) Top panel: the strain ΔlpxM ΔyhcB + P ara ::yhcB was grown under inducing conditions with 0.1% arabinose (blue) or under repressing conditions with 0.2% glucose (orange).Black arrow denotes when strains were diluted 1:100.The ΔlpxM + P ara ::empty strain (black) was used as a control.Bottom panel: wild type, ΔclsA, ΔyhcB, and ΔyhcB ΔclsA were grown for 24 hours.For both panels, cultures were grown at 37°C in a 96-well plate, and OD 600 was assessed every 30 minutes using a microplate reader.Data shown are representative of biological triplicates.
GPL biosynthesis.This flux results in an imbalanced ratio of LPS and GPLs within the cell, resulting in major cell envelope defects.By manipulating FAB through genetic or chemical means, these defects can be mitigated and fitness of the yhcB mutant can be restored.We therefore conclude that YhcB may be involved in the regulation of FAB and that it may support cell envelope assembly and cell division through this pathway.

RESULTS
Deletion of lpxM or clsA in a yhcB mutant results in synthetic lethal and synthetic sick phenotypes, respectively Random transposon (Tn) mutagenesis in individual lpxM and clsA mutants paired with next-generation sequencing (TnSeq) suggested that loss of yhcB would be detrimental to bacterial fitness in these strains (7).Using generalized transduction and the Keio strain collection (21), we attempted to generate a strain lacking yhcB in combination with either lpxM or clsA.In the case of the lpxM mutant, attempts to delete yhcB resulted in no transductants, and changing the order of deletion had no effect.To confirm a possible synthetic lethal phenotype, a plasmid-expressing yhcB from an arabinose-indu cible promoter (P ara ::yhcB) was introduced into ΔlpxM followed by deletion of yhcB by phage transduction in the presence of arabinose.The lpxM yhcB mutant, carrying either empty vector or P ara ::yhcB, was grown in lysogeny broth (LB) with either an inducer or a repressor (glucose) (Fig. 1B).Growth arrests when YhcB is depleted in the lpxM yhcB double mutant, demonstrating synthetic lethality.In the case of clsA, we were able to generate the yhcB clsA double mutant.Compared to the yhcB single mutant, however, ΔyhcB ΔclsA exhibited a worsened growth defect compared to the parent strains, indicating a synthetic sick phenotype (Fig. 1B).These results suggested that when the membrane is already perturbed by the loss of LpxM or ClsA, the additional loss of YhcB weakens the cell envelope further.Given that LpxM and ClsA have related functions in LPS biosynthesis/transport, these phenotypes suggest a connection of YhcB to the cell envelope (7).

Characterization of the yhcB mutant in W3110
All prior work investigating YhcB was performed in E. coli strains BW25113 and MG1655.It has been reported that many MG1655 laboratory stocks possess a 1-bp deletion in glpR that encodes the repressor of the glycerol-3-phosphate regulon, a mutation predicted to abolish function (22).Loss of this regulator leads to constitu tive expression of glycerol metabolism genes, including genes involved in degrada tion of an important component used in GPL synthesis (23).The BW25113 strain, used as the parent strain for the Keio collection, possesses a Gly-to-Thr missense mutation in fabR and has been shown to act similarly to a ΔfabR mutant (21,24).FabR is a repressor of fabA and fabB and therefore has a direct impact on FAB when disrupted (24).Because these two strains have mutations impacting key GPL precursors, we chose to work in K-12 strain W3110.This strain does not contain mutations that could affect how the cell adapts to dysregulation of fatty acid or GPL synthesis as confirmed by whole-genome sequencing of our laboratory W3110 stock (22,24) (see Data Set S1 in supplemental material).
First, deletion of yhcB in W3110 was evaluated for changes in growth and cell shape.By both measuring optical density (OD 600 ) and determining colony-forming units (CFU), it was clear that ΔyhcB has a drastic growth defect (see Fig. S1A in supplemental material).Scanning electron microscopy (SEM) of ΔyhcB showed the reported pleomor phic phenotype in the W3110 background (Fig. S1B).There were cells of increased length (more than 60 µm), cells with abnormal division sites, and cells displaying increased surface volume.We observed after 24 hours of liquid culture at 37°C that the yhcB mutant had visible cell debris in the culture tube (Fig. S1C).Loss of YhcB also rendered E. coli more sensitive to vancomycin, an antibiotic that does not easily penetrate the OM, suggesting changes in OM integrity (Fig. S1D).

Suppressors that rescue synthetic lethality and growth defect
We sought to isolate suppressors that would provide insight into the function of YhcB in the cell.To do so, we used two methodologies: one that focused on what mutations would sustain viability in the synthetic lethal yhcB lpxM mutant, and the other focused on what mutations would restore normal growth in the single yhcB mutant.
Suppressors that restored viability of ΔyhcB ΔlpxM were obtained via large-scale transductions intended to isolate the infrequent successful transductants.Independent transductions were repeatedly performed for each single mutant, yhcB or lpxM, to chromosomally delete the secondary gene.Resulting colonies were PCR screened for successful deletions and then subjected to whole-genome sequencing (Fig. 2A).Multiple suppressor mutations arose within three of the four genes encoding the subunits of the acetyl-CoA carboxylase (ACC) complex, two suppressors in accC, and one in each accA and accD.The AccABCD complex initiates FAB and catalyzes the carboxylation of acetyl-CoA to form malonyl-CoA, an enzymatic product that is dedicated solely to FAB.
Suppressors that restored the growth defect of the yhcB mutant were obtained via a short-term evolution experiment.Three independent cultures were passaged daily until the cellular debris that is always present after 24 hours of growth of ΔyhcB was no longer visible (Fig. S1C).Surprisingly, all three independent populations were visually free of debris on day 7, and an aliquot of each was sent for whole-genome sequencing.The cultures were continued until day 30 when populations were whole-genome sequenced a second time (Fig. 2B).
At the time that visual cell debris had resolved, no one particular mutation had fully overtaken any of the populations.In population 1, mutations in accC, mlaD, obgE, mlaF, and mdoG were found at varying frequencies.These genes encode (i) the acetyl-CoA carboxylase subunit α (part of ACC complex), (ii) an IM subunit of the Mla (maintenance of OM lipid asymmetry) system involved in GPL transport, (iii) a GTPase implicated in stringent response, (iv) an ATPase of the Mla system, and (v) a periplasmic protein involved in the formation of osmoregulated periplasmic glucans, also known as mem brane-derived oligosaccharides (Fig. 2B).After 30 days of continuous passaging, the only mutation in evolved population 1 was the original accC mutation found at day 7. Population 2 originally had mutations in cdsA (the CDP-diglyceride synthetase required for GPL synthesis), obgE, and mlaE, which encodes another component of the Mla system.After 30 days, the cdsA mutation had become the dominant mutation in population 2, and a mutation in arcA (the transcriptional regulator of a two-component system) had arisen.The third population only had a mutation in mlaA (the OM lipopro tein subunit in the Mla transport system) when cellular debris was no longer observed, and by day 30, a mutation in accB (the biotin carboxyl carrier protein in the ACC complex) had overtaken the population.All three populations still displayed filamentous cell morphology at day 7, despite visual debris resolution (Fig. S2).The recurring mutations in genes involved in FAB or GPL biosynthesis/transport suggested that YhcB may be involved in lipid biogenesis.

Disruption of the Mla system rescues growth defect caused by loss of YhcB
Because mutations in the Mla system were found in every evolved population when debris was resolved, we investigated the role of this system in the yhcB mutant.The Mla system carries out retrograde transport of mislocalized GPLs in the outer leaflet of the OM for reinsertion into the IM, helping to maintain the asymmetrical nature of the OM (25)(26)(27).Loss of any one of the six Mla proteins abolishes its retrograde transport of GPLs (28).To evaluate whether our mla mutations were disrupting function, we deleted mlaF (encodes the ATPase subunit of the ABC transporter complex) which was mutated in evolved ΔyhcB population 1. Deletion of mlaF improved cell survivability as evidenced by an increase in CFU over time compared to the ΔyhcB parent (Fig. 3A) This rescue of the growth defect suggests that the mla suppressors across all three evolved populations were loss-of-function mutations.
In addition to the Mla system, E. coli possesses an OM phospholipase, PldA, that helps maintain asymmetry of the OM by cleaving GPLs mislocalized to the cell surface.The fatty acids released by PldA activity act as a signal to increase LPS synthesis and restore the balance of OM lipids (29).Because it is known that abolishing Mla transport causes an increase in PldA activity due to accumulated GPLs in the OM, we deleted pldA from the yhcB mlaF mutant (28).When PldA function is lost, the growth defect returns (Fig. 3A; Fig. S3).This result indicates that the fitness improvement caused by the loss of Mla function is likely mediated by PldA and suggests that the LPS to GPL ratio in the yhcB mutant may be out of balance.A change in OM asymmetry, whether caused by too many GPLs or too little LPS, may also explain the synthetic phenotypes seen with lpxM and clsA deletions.Although the growth defect was resolved in the yhcB mlaF strain, it did not have improved morphology and remained filamentous (Fig. 3B).These mutations weakly suppressed the yhcB phenotypes by rescuing only the growth defect but were eventually overtaken by the FAB-and GPL-related mutations in the evolved populations.

Reducing fatty acid biosynthesis rescues yhcB phenotypes
Multiple suppressor mutations in genes encoding the subunits of the ACC complex arose in the yhcB lpxM double mutant, allowing viability.To examine these mutations solely when YhcB function is lost, deletions were made in genes close to accA and accD (yaeR (B) Summary of mutations that occurred during a short-term evolution experiment of three independent yhcB mutant populations.Bacteria were subjected to whole-genome sequencing at the time cellular debris was no longer observed (7 days) and at the end time point (30 days) of the experiment.DPI, days post inoculation.Created in part with BioRender.com.and purF, respectively).Taking advantage of the close genetic linkage, we then used P1 phage-mediated transduction to transfer the mutant alleles of accA and accD from the suppressor mutants into wild-type W3110 and ΔyhcB.These mutants were then assessed visually via microscopy.W3110 WT carrying either the accA R175L or accD L213P allele appeared slightly smaller than wild type, an expected phenotype if fatty acid production has decreased (see Fig. S4A and B).Introduction of accA R175L into ΔyhcB resulted in strong suppression of the gross morphological defects seen in the parent.The yhcB accD L213P strain had a partial recovery of the filamentous yhcB phenotype (Fig. 4A and Fig. S4).It is important to note deletion of either yaeR or purF for introduction of the mutant acc alleles had no effect on cell morphology (Fig. S4C).
Because both suppressor screens had suggested that FAB was connected to the yhcB defects, overall lipid content was analyzed from cells grown in the presence of 32 P i .The freed lipid A domain of LPS and GPLs were isolated in the same extraction, and the lipid species were separated by thin-layer chromatography (TLC).In this solvent system, the major GPLs migrate faster than lipid A species, with PG and CL migrating to the same position.The yhcB mutant exhibits a growth defect that worsens in stationary phase (Fig. S1; Fig. 3A); therefore, lipids were extracted from cultures at that phase of growth to better assess what defects were contributing to the growth phenotype.As a control, we also included W3110 expressing a truncated variant of YejM (yejM 569 ).LPS biosynthesis is negatively regulated by YejM, and LPS (lipid A) synthesis in this mutant is greatly reduced (30), resulting in a significant increase in the GPL:lipid A ratio (Fig. 4B and C).Similar to the yejM mutant, loss of YhcB resulted in a clear increase in the GPL:lipid A ratio.However, because this quantification measures the ratio of all GPL species compared to all lipid A species, reduced LPS or increased GPLs can result in an increased GPL:lipid A ratio as seen in both the yejM and the yhcB mutants.
To evaluate whether the increased GPL:lipid A (LPS) ratio in the yhcB mutant was due to a reduction in LPS synthesis, proteinase K-treated whole-cell lysates were subjected to SDS-PAGE, and the LPS was visualized by staining with Pro-Q Emerald carbohydrate dye.Deletion of yhcB did not result in a significant change in LPS production, whereas the yejM 569 control showed the expected reduction in LPS levels (Fig. 4D; Fig. S5A).Thus, the altered GPL:lipid A ratio observed in ΔyhcB must arise from an increase in GPLs.To answer whether the accA R175L allele was fixing morphology by restoring lipid balance, total lipids were again extracted.The increased GPL:lipid A ratio of the yhcB mutant is abolished when the accA R175L allele is introduced (Fig. 4C; Fig. S5B and C).Mirroring the cell morphology data, the accD L213P allele only partially restored lipid levels (Fig. S5B and  C).When the acc alleles are introduced, the yhcB phenotype is suppressed, suggesting that these mutations are epistatic to yhcB and that this rescue is upstream of the yhcB defect.
Because mutations that rescued viability of the yhcB lpxM double mutants suggested a connection to FAB, we next assessed how chemical inhibition of FAB would affect the yhcB mutant.Cerulenin is an inhibitor that prevents FAB elongation by irreversibly binding FabB and FabF (31)(32)(33).Several groups have suggested that the defects of yhcB are due to a potential role in cell division, so we included an envC knockout mutant as a control (11,12).EnvC activates key peptidoglycan amidases that are required for septal division and exhibits a filamentation defect visually similar to the yhcB mutant (34).
As expected, growth of wild-type W3110 is inhibited at 60 µg/mL of cerulenin.Cells lacking EnvC showed increased sensitivity to the antibiotic, and inhibition of FAB had no impact on ΔenvC morphology.Strikingly, the yhcB mutant grows well in the presence of cerulenin and better than wild type (Fig. 5A).Furthermore, the presence of the antibiotic restored cell morphology of ΔyhcB (Fig. 5B) and ameliorated the growth defect, acting as a strong suppressor of the yhcB phenotypes.The fact that cerulenin could not rescue a generalized cell division defect typified by ΔenvC yet rescues the morphology of ΔyhcB suggests that the cause of the ΔyhcB morphological defect is related to FAB.

The phenotypes associated with yhcB are due to excess fatty acids
Although the yhcB mutant has excess GPLs and its fitness benefits from reduction of FAB, it was still unclear whether the mutant had excess flux through FAB or through GPL synthesis.Since the products of FAB are disproportionally shunted into GPL synthesis as compared to LPS biogenesis (Fig. 6A), two hypotheses emerged.One, that the yhcB mutant exhibits overactive GPL biosynthesis, leading to increased FAB flux due to rapidly used end products.Alternatively, it is FAB that is hyperactive in ΔyhcB, and the excess products feed into GPL biosynthesis, resulting in increased GPLs.In short, the origin of the problem could be dysregulated GPL synthesis or dysregulated FAB.To better understand whether the ΔyhcB phenotypes were caused by overactive GPL synthesis or overactive FAB, we assessed how manipulation of LPS biogenesis and acyl-ACP degradation would affect the yhcB mutant.
LpxC catalyzes the first committed step of LPS biosynthesis, and, for this reason, the cellular levels of enzyme are tightly controlled.We expressed lpxC using the low-copynumber, isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible vector pWSKI (pW in Fig. 6) (35).As previously reported (36), increasing LpxC levels in a wild-type cell was detrimental to fitness as determined by CFU counts after 6 hours of growth.Much like what we saw with our mla suppressors, increasing LpxC in the yhcB mutant was a weak suppressor that rescued growth but not cell morphology (Fig. 6).TesA is a periplasmic esterase that can act on a variety of acyl-CoA and acyl-ACP molecules; truncation of the gene results in localization of TesA in the cytoplasm (TesA cyto ) where it can still function and has been demonstrated to generate free fatty acids (37)(38)(39)(40).Plasmid expression of TesA cyto in wild-type W3110 resulted in reduced CFU counts at hour 6.Expression of TesA cyto in the yhcB mutant, however, strongly suppressed the filamentous morphology and growth defect, suggesting that the degradation of acyl-CoA and acyl-ACP molecules in the cytoplasm is beneficial to cells lacking YhcB (Fig. 6B and C).
These data suggest that increased LPS can restore the asymmetry in the OM and rescue cell growth, but increased LPS does not alleviate the original issue of excess lipids.However, degrading acyl-ACP molecules arising from FAB is highly beneficial in the absence of YhcB.It is only by modulating fatty acids and not LPS synthesis that all yhcB phenotypes can be fixed, suggesting that FAB flux in the yhcB mutant is abnormally increased.

DISCUSSION
E. coli YhcB has been suggested to play a role in cell division, peptidoglycan synthesis, and even LPS regulation (11,13,41).Based on a previous Tn-Seq screen, Goodall and coauthors (13) predicted that yhcB may have a synthetically lethal phenotype with 87 genes, with the vast majority of these genes associated with cell envelope biogenesis.Although this work revealed interesting genetic interactions, cells lacking YhcB are both pleiomorphic and show a drastic reduction in bacterial fitness.Therefore, it is not unexpected that the disruption of additional cell envelope pathways would cause cell death, and these synthetic phenotypes may not necessarily indicate a direct relationship with YhcB.Here, we have shown that modulating lipids in the cell directly benefit a yhcB mutant and have provided data that link YhcB to FAB.
Through a short-term evolution experiment, we found that the loss of the Mla system can weakly suppress the ΔyhcB growth defect but cannot rescue the original defect (demonstrated by the unchanged filamentation).Goodall et al. ( 13) also reported that any disruption of the Mla system rescued both SDS-EDTA and vancomycin sensitivity in the yhcB mutant, though they did not report any cell morphology changes.When Mla is disrupted, GPLs accumulate in the outer leaflet of the OM, leading to increased PldA activity.We show here that rescue of the yhcB growth defect conferred by the loss of the Mla system is directly tied to PldA activity (Fig. 3).PldA cleaves GPLs, the by-products of which stimulate LPS biosynthesis.These results also indicate that the defect is not related to a deficiency of GPL transport to the OM since PldA-mediated rescue would only occur if there were too many GPLs in the outer leaflet of the OM.This dependency on PldA is also supported by our finding that the yhcB mutant synthesizes an increased number of GPLs (Fig. 4).
Although increasing degradation of mislocalized GPLs and signaling to increase LPS via PldA prohibit the growth defect, these mutations were eventually overtaken by mutations in FAB and GPL synthesis in yhcB mutants during a short-term evolution experiment.Of the three populations, the one harboring a mutation in cdsA never recovered wild-type cell morphology (Fig. 2B and Fig. S2), suggesting that the defect in the yhcB mutant is further upstream.It is interesting to note that these suppressor screens did not uncover any mutations in peptidoglycan or LPS biosynthesis pathways or proteins specific to cell division.Instead, it was mutations in the genes encoding the various subunits of the ACC complex, AccA and AccD, that recovered cellular fitness.The accA mutation not only suppressed morphological defects but also stabilized the GPL:lipid A (LPS) ratio within the cell envelope, presumably by reducing flux through FAB (Fig. 4; Fig. S4A and Fig. S5).However, a recent publication by Som and Reddy (42) exploring the crosstalk between FA and peptidoglycan synthesis may explain the apparent peptidoglycan and cell division defects suggested by the filamentous yhcB cell.This work demonstrated that MepS, a major endopeptidase involved in peptidoglycan growth, is stabilized in genetic backgrounds with more fatty acid synthesis, including a yhcB mutant.These results not only demonstrate a reasonable pathway through which peptidoglycan synthesis responds to the availability of fatty acids, these data also suggest that due to the increased flux through FAB in the absence of YhcB, abundance of fatty acids leads to a concomitant increase in peptidoglycan synthesis, resulting in the filamentous cells.
These genetic findings were recapitulated using a chemical biology approach by inhibiting FAB via cerulenin (Fig. 5).Noga and co-authors (43) present mass spectrom etry data that during growth inhibition by cerulenin, longer-chain acyl-ACPs that are preferred substrates for GPLs (C16 and C18 acyl chains) drop rapidly, while levels of acyl-ACPs preferred for LPS increase (C12 and C14 acyl chains).Because cerulenin is an irreversible inhibitor, overall FAB flux would dramatically reduce.Thus, GPL synthesis would decrease at a faster rate compared to LPS synthesis, restoring the balance of OM lipids and the fitness of cells lacking YhcB.The phenotypes of the yhcB mutant appear specific to stationary phase, and the presence of cerulenin strongly suppresses both cell morphology and growth defects (Fig. 5A).Fatty acid biosynthesis slows during stationary phase in a normal cell, so perhaps YhcB is involved in reducing flux through the system during this transition.Supporting this hypothesis, two of the ΔyhcB populations in our short-term evolution experiment developed mutations in obgE, which encodes a GTPase involved in the stringent stress response (44).
Expression of the thioesterase TesA in the cytoplasm strongly suppresses the growth defect and cell morphology in the yhcB mutant (Fig. 6).It has been demonstrated that expression of TesA cyto results in an approximately threefold increase in free fatty acids (FFAs) during stationary phase, with excess FFAs even being found in the culture medium.Although the specific role for TesA is still unidentified, analysis of the FFAs generated by TesA cyto activity indicates a substrate preference for long-chain fatty acids such as C16 and C18, which act as acyl donors for GPLs.Some cleavage of acyl donors preferred for LPS synthesis was detected, though in lower amounts compared to the long-chain fatty acids (37).In the yhcB mutant, we hypothesize that TesA cyto is degrading the excess acyl-ACPs with preference for acyl donors destined for GPL synthesis, restoring overall fitness.Supporting our work, expression of TesA cyto also rectifies phenotypes associated with overexpression of the regulator FadR which results in increased FAB and GPL synthesis similar to ΔyhcB (20).In both scenarios, these results suggest that TesA cyto is generating FFAs from the excess acyl-ACP pool, which can then be degraded through β-oxidation and/or transported into the culture medium through mechanisms that have not been identified.
We expected that if the problem were solely excess GPL flux, then increasing LPS to rebalance lipids in the cell would be beneficial.Indeed, we saw that overexpression of lpxC in the yhcB mutant mediated the growth defect (Fig. 6C), similar to the PldA-medi ated, weak suppression seen with Mla system inactivation (Fig. 3).In support of this finding, Wieczorek et al. (41) demonstrate that LpxC protein levels are stabilized in a yhcB mutant.Goodall and co-authors (13) also report that a yhcB ftsH double mutant is viable without suppressors.FtsH is an essential protease that proteolytically controls the level of LpxC (and therefore LPS) in the cell.These results are likely due to the cell attempting to rebalance the GPL:lipid A (LPS) ratio to compensate for increased GPL synthesis due to acyl-ACP accumulation.We report that increasing LPS alone cannot fully restore the yhcB mutant, evidenced by the filamentous cell morphology (Fig. 6B), because the origin of the problem may be excess flux through FAB, not GPLs.In the context of this information, the synthetic lethality with lpxM and the synthetic sick phenotype with clsA (Fig. 1) is clear because both of these mutations impact LPS levels and transport (7).Increasing LPS alone cannot fix the yhcB cell morphology phenotype, but degradation of acyl-ACPs in the cytoplasm can.These results make sense with the conclusion that FAB is dysregulated and that flux through the biosynthetic pathway is aberrantly increased when YhcB is absent.
However, it is difficult to separate GPL and FAB flux due to the connectivity of these two pathways.Increased FAB flux leads to increased GPL flux, and the opposite is also true (20,45).For example, increased expression of FadR results in increased production of enzymes involved in FAB, leading to elevated FAB flux and more GPLs (20,46).Whereas increasing activity of the enzyme that catalyzes the rate-controlling step in GPL synthesis, PlsB, lowers the pool of long-chain ACP species and reduces feedback inhibition of the ACC complex.This elevated activity thereby increases FAB flux in response to raised GPL flux (45).Given the tight connection between the two pathways, experimentally modulating one without impacting the other is impossible.The mutations that evolved in the yhcB populations were not in plsB but instead in genes encoding subunits of the ACC complex that together determine the rate-controlling step for FAB.The context of these mutations suggests that it is FAB, not GPL synthesis, that is aberrantly increased in the yhcB mutant.
The ACC complex (comprised AccABCD) initiates FAB by converting acetyl-CoA to malonyl-CoA in two half-reactions.In the first, AccC carboxylates the biotin attached to AccB (using Mg 2+ ions as cofactor and ATP), and in the second, AccA and AccD transfer the carboxyl group from biotin to acetyl-CoA, generating malonyl-CoA (47).Thorough understanding of the regulation of the ACC complex has remained elusive, but limited in vitro evidence indicates that acyl-ACP molecules (C6 to C20) can inhibit the complex (48,49).Still, these data have the caveat that the kinetic assays were performed with both holo-and apo-AccB present (biotinylated and unbiotinylated), the latter of which is chemically inert and could prevent ACC complex activity (50).The delicate balance of energy consumption and the need for membrane lipid demands tight control of FAB in bacteria.Here, we have presented evidence that a yhcB mutant has lost regulation of this system to deleterious effects.That the best and most repeated response to the loss of YhcB is to mutate any of the subunits of the essential ACC complex suggests that YhcB may play a role in throttling FAB, perhaps at the initiating enzymatic step.Several groups (11,16,41) have published data from protein-protein interaction experiments, and only one (41) has reported interactions with phospholipid and fatty acid biosynthe sis enzymes (PssA, AccD, and FabBFHY), though direct analysis is required, given the complexity and remaining unknowns of these potential interactions.It is an interesting hypothesis that YhcB may be regulating FAB, through direct or indirect means, and merits further investigation.

Strains and growth conditions
All strains, plasmids, and primers used in this study are provided in Data Set S1 in supplemental material.Unless otherwise stated, strains were grown in lysogeny broth or on LB agar at 37°C.Where indicated, growth media were supplemented with ampicillin (100 µg/mL), kanamycin (30 µg/mL), vancomycin (50 µg/mL), cerulenin (60 µg/mL), L-arabinose (0.05% or 0.1% [wt/vol]) as indicated in figure legends, D-glucose (0.2% glucose [wt/vol]), and/or IPTG (1 mM or 5 µM) as indicated in figure legends.Growth curves were performed both in 0.2 mL LB using the BioTek Epoch 2 plate reader in a polystyrene 96-well plate and in 5 mL LB in glass test tubes.

Strain construction
Chromosomal deletions were generated in the E. coli K-12 strain W3110 via general ized transduction and the Keio collection (21).The flippase recognition target-flanked kanamycin cassette was removed by flippase site-specific recombination as previously described (51).A plasmid, pCP20, which expresses the flippase from a temperature-sen sitive promoter, was transformed into the strains possessing the kanamycin resistance cassette.Transformants were then grown at 30°C overnight and then passaged at 37°C.Colonies were screened for loss of both the kanamycin and ampicillin resistance, and the removal of the kanamycin cassette was confirmed via PCR amplification of the appropriate region (51).The strain lpxM yhcB::kan + P ara ::yhcB was maintained in the presence of arabinose unless otherwise stated.The mutant acc alleles were transferred into relevant strains by linking the mutant allele with a nearby kanamycin cassette.To link these genetic elements, a nearby gene (yaeR for accA and purF for accD) with a predicted unrelated function was deleted using generalized transduction, and phage lysate was made from the resulting strain.This lysate was then used to transfer the mutant acc alleles with high frequency by selecting for kanamycin resistance in the resulting recombinants.Colonies were then PCR verified for the deletion of yaeR or purF, and then accA or accD was Sanger sequenced to verify the point mutation (Data Set S1).

Scanning electron microscopy
Cultures were grown to OD = 0.5 in 5 mL LB and harvested by centrifugation.Cell pellets were fixed in 2% glutaraldehyde 0.1 M cacodylate buffer for 2 hours at room temperature.Cells were then washed in 0.1 M cacodylate buffer and post-fixed in 1% OsO 4 0.1 M cacodylate buffer for 1 hour at room temperature.After washing, samples were dehydrated in an ethanol series (25%, 50%, 75%, and 100% ethanol) for 10 minutes per step.Samples were then placed on poly-L-lysine-coated glass cover slips and critical point dried.Finally, samples were mounted on aluminum SEM stubs and sputter coated with gold-palladium.Images were taken using a Teneo FE-SEM.

Selecting for ΔyhcB suppressors
Suppressors that restored viability of the synthetically lethal yhcB lpxM double mutant were obtained via large-scale transductions.Independent cultures of yhcB and lpxM single mutants were used for separate transductions.Resulting colonies were screened internally and externally via PCR for yhcB and lpxM and whole-genome sequenced (Data Set S1).Suppressors that rescued formation of cellular debris in the single yhcB mutant were obtained from a short-term evolution experiment.Three independent colonies were inoculated in 5 mL LB and passaged daily into fresh 5 mL LB.Aliquots were sent for whole-genome sequencing upon visible resolution of cellular debris (day 7), and additional aliquots were sent after 30 days of passaging.Analysis of the whole-genome sequencing for both suppressor experiments was performed using CLC Genomics Workbench (QIAGEN) and the E. coli W3110 genome (GenBank accession number NC_007779).

Phase-contrast microscopy and cell measurements
Cultures were grown for 5 hours and spotted on 3% agarose pads.Images were captured with the Olympus CX43 equipped with an Olympus Infinity 355 camera in phase contrast.Measurements were performed on >5 fields of view and 298 cells per strain with MicrobeJ and analyzed in GraphPad Prism 10.In the case of the yhcB mutant, manual measurements were required for the extremely filamentous cells that were not detectable by MicrobeJ.These cells (<20 cells) were manually measured using the segmented line tool function in ImageJ.A one-way analysis of variance test was performed with Brown-Forsythe and Welch tests, assuming that standard deviations were not equal, and significant differences were assessed using a Games-Howell test.

Spot plates
Cultures were normalized to an OD = 0.05 and grown in 5 mL LB at 37°C.At each time point, cultures were serially diluted and spotted (5 µL) on LB agar.Plates were incubated overnight at 37°C.A two-tailed, unpaired t-test was performed to determine significant differences between samples at specific time points.

Lipid extractions and visualization
Strains were normalized to an OD of 0.05 in 5 mL LB containing 2.5 µCi/mL of inor ganic radioactive phosphate ( 32 P i ) and grown at 37°C for 6 hours.At hour 6, cells were harvested by centrifugation, and total lipids were isolated as previously described (18).Briefly, cell pellets were washed once with 1× PBS, resuspended in a solution of 50 mM sodium acetate (pH 4.35) containing 1% SDS, and incubated at 100°C for 30 minutes.This incubation hydrolyzes the lipid A domain from the polysaccharide portion of LPS and has no effect on GPL structure.The lipid A and GPL species were isolated via Bligh-Dyer extraction, and the lipids were separated using TLC in a solvent system of chloroform, pyridine, 88% formic acid, and water (50:50:16:5, vol/vol).Lipids were visualized by phosphorimaging analysis using the Amersham Typhoon Biomolecular Imager (Cytiva), and densitometry was performed to determine the GPL:lipid A ratios.Images shown in figures have been adjusted for visibility of all lipid species, and an unsaturated image was used for all densitometry calculations.A two-tailed, unpaired t-test was performed to determine significant differences between samples.

LPS quantification
Strains were grown for 6 hours from a starting OD of 0.5 and harvested by centrifugation.Samples were normalized by protein concentration, and the cell pellet was resuspended in lithium dodecyl sulfate loading buffer and treated with proteinase K. Samples were analyzed by SDS-PAGE using a 4%-12% Bis-Tris gel (Fisher).Following the manufacturer's instructions, LPS was stained using the Pro-Q Emerald Kit.Gels were visualized using the Bio-Rad ChemiDoc MP Imaging System, and relative LPS values were calculated using densitometry of the stained bands.A two-tailed, unpaired t-test was performed to determine significant differences between samples.

FIG 1
FIG 1 Depletion of YhcB in an lpxM yhcB mutant is lethal and a yhcB clsA mutant is synthetically sick.(A) Chemical structure of the lipid A domain of LPS (top).

FIG 2
FIG 2 Summary of spontaneous suppressor mutations.(A) Suppressor mutations that restored viability of the synthetically lethal yhcB lpxM double deletion.

FIG 3
FIG 3 Loss of the Mla retrograde lipid transport system rescues the growth defect but not cell morphology of ΔyhcB in a PldA-dependent manner.(A) Growth was assessed in 5 mL LB and measured by both OD 600 (bottom) and serial dilutions (top) to calculate CFU/mL.Data shown are representative of biological triplicates.(B) Bacteria represented in the growth curves were imaged using phase-contrast microscopy at ×1,000 magnification with a 10-µm scale bar that is indicated in the top left panel.Additional controls can be found in Fig. S3.

FIG 4
FIG 4 Mutations in the ACC complex rescue ΔyhcB cell morphology and restore the GPL:lipid A (LPS) ratio.(A) Alleles of acc (accA R175L and accD L213P ) that restored viability of a yhcB, lpxM double mutant were transduced into a single yhcB mutant and evaluated by phase contrast microscopy at ×1,000 magnification (10 µm scale bar).(B) Thin-layer chromatography (TLC) of total 32 P-labeled lipids (both lipid A and GPLs) extracted from wild type, the yhcB mutant, and wild type with a truncated yejM allele (yejM 569 ).The latter strain has greatly reduced LPS levels and serves as a control for an altered GPL:lipid A (LPS) ratio.Major species of GPLs are indicated on the right side of the TLC and major lipid A species on the left.Image shown has been adjusted for visibility of all lipid species; an unsaturated image was used for densitometry calculations.(C) Bar graph representing the GPL:lipid A (LPS) ratios of wild type, ΔyhcB with wild-type accA, and ΔyhcB with the accA R175L allele.Again, yejM 569 strain is included as a control.Data shown are representative of biological triplicates.Significance was assessed using two-tailed, unpaired t-tests.(D) Bar graph representing relative amounts of LPS of wild type, ΔyhcB, and yejM 569 quantified using densitometry.Data are representative of biological triplicates.Significance was assessed using two-tailed, unpaired t-tests.NS indicates not significant; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.

FIG 5
FIG 5 Chemical inhibition of FAB rescues ΔyhcB growth defect and cell morphology.(A) Growth was assessed by measuring OD 600 (top panels) and by serial dilutions (bottom panels) to calculate CFU/mL every hour.Wild type, ΔyhcB, and ΔenvC (control strain) were grown in LB with and without cerulenin (60 µg/mL).(B) Bacteria represented in the growth curves were imaged at hour 5 using phase-contrast microscopy at ×1,000 magnification with a 10-µm scale bar for all panels.Data shown are representative of biological triplicates.

FIG 6
FIG 6 ΔyhcB phenotypes originate with excess fatty acids.(A) Schematic outlining how overexpression of TesA cyto and LpxC from the low-copy-number, isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible plasmid pWSKI (pW) is predicted to affect the yhcB mutant.In the first panel depicting a wild-type cell, acyl-ACPs generated by FAB disproportionately shunt into GPL synthesis.In the second panel, a yhcB mutant has increased flux through FAB and through GPL synthesis, leading to excess GPLs.In the third panel, increased expression of LpxC increases LPS biosynthesis and helps to restore the misbalanced GPL:LPS ratio.In the fourth panel, expression of TesA cyto in the yhcB mutant leads to the degradation of FAB end products and reduced GPL and, to a lesser extent, LPS flux.(B) Phase-contrast microscopy (×1,000 magnification, 10 µm scale bar) of cells with increased expression of LpxC or a cytoplasmic version of TesA after 6 hours of growth.IPTG was added at initial inoculation, 1 mM IPTG for TesA cyto and 5 µM IPTG for LpxC.(C) CFU/mL values determined by serial dilutions of cultures described in panel B. Data shown are representative of biological triplicates.Significance was assessed using two-tailed, unpaired t-tests.NS indicates not significant; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001.