Defects in the First Step of Lipoprotein Maturation Underlie the Synthetic Lethality of Escherichia coli Lacking the Inner Membrane Proteins YciB and DcrB

A distinguishing feature of Gram-negative bacteria is their double-membraned cell envelope, which presents a formidable barrier against environmental stress. In E. coli, more than a quarter of the cellular proteins reside at the inner membrane, but about a third of these proteins are functionally unassigned or their function is incompletely understood.

ABSTRACT

Nearly a quarter of the Escherichia coli genome encodes inner membrane proteins, of which approximately a third have unassigned or poorly understood functions. We had previously demonstrated that the synergy between the functional roles of the inner membrane-spanning YciB and the inner membrane lipoprotein DcrB is essential in maintaining cell envelope integrity. In yciB dcrB cells, the abundant outer membrane lipoprotein Lpp mislocalizes to the inner membrane, where it forms toxic linkages to peptidoglycan. Here, we report that the aberrant localization of Lpp in this double mutant is due to inefficient lipid modification at the first step in lipoprotein maturation. Both Cpx and Rcs signaling systems are upregulated in response to the envelope stress. The phosphatidylglycerol:preprolipoprotein diacylglyceryl transferase, Lgt, catalyzes the initial step in lipoprotein maturation. Our results suggest that the attenuation in Lgt-mediated transacylation in the double mutant is not a consequence of lowered phosphatidylglycerol levels. Instead, we posit that altered membrane fluidity, perhaps due to changes in lipid homeostasis, may lead to the impairment in Lgt function. Consistent with this idea, a dcrB null mutant is not viable when grown at low temperatures, conditions that impact membrane fluidity. Like the yciB dcrB double mutant, dcrB null-mediated toxicity can be overcome in distinct ways—by increased expression of Lgt, deletion of lpp, or removal of Lpp-peptidoglycan linkages. The last of these events leads to elevated membrane vesiculation and lipid loss, which may in turn impact membrane homeostasis in the double mutant.

IMPORTANCE A distinguishing feature of Gram-negative bacteria is their double-membraned cell envelope, which presents a formidable barrier against environmental stress. In E. coli, more than a quarter of the cellular proteins reside at the inner membrane, but about a third of these proteins are functionally unassigned or their function is incompletely understood. Here, we show that the synthetic lethality underlying the inactivation of two inner membrane proteins, a small integral membrane protein (YciB) and a lipoprotein (DcrB), results from the attenuation of the first step of lipoprotein maturation at the inner membrane. We propose that these two inner membrane proteins, YciB and DcrB, play a role in membrane homeostasis in E. coli and related bacteria.

INTRODUCTION

The Gram-negative cell envelope is an essential multilayered structure that defines the bacterial cell interior. The cell envelope consists of two lipid bilayers, the inner membrane (IM) and outer membrane (OM), which are separated from one another by the periplasmic compartment that houses the peptidoglycan (PGN) cell wall. Each layer of the cell envelope participates in several essential processes, including selective solute passage, environmental sensing, and biogenesis of the extracytoplasmic proteins that directly perform many of these cellular functions (1, 2). Following their synthesis in the cytoplasm, extracytoplasmic proteins are secreted across the IM, where they may undergo various forms of postsecretion processing to reach their destination or attain their final functional states (3–5).

For proteins destined for the OM, including the integral β-barrel outer membrane proteins (OMPs) and lipoproteins, postsecretion processing occurs through distinct pathways. OMPs, following their secretion, are trafficked across the periplasm to the OM with the assistance of general periplasmic chaperones (5). This is primarily accomplished by the chaperone SurA (6, 7). Additional chaperones Skp and DegP are believed to play accessory roles in targeting OMPs that have exited from the SurA-bound pathway to the OM (7). Once at the OM, the chaperone-bound OMPs are inserted into the membrane through the assistance of the β-barrel assembly machinery, the Bam complex (5, 8, 9).

Postsecretion, all preprolipoproteins (the secreted protein with the uncleaved signal peptide) are modified to their mature triacylated forms at the outer leaflet of the IM through the sequential action of three IM-bound enzymes. In the first step, Lgt, a phosphatidylglycerol:preprolipoprotein diacylglyceryl transferase enzyme, modifies the preprolipoprotein to the prolipoprotein form. In this step, a diacylglyceryl (DAG) group is transferred from phosphatidylglycerol (PG) to the thiol side chain of an invariant cysteine residue (Cys⁺¹) housed immediately downstream of the N-terminal signal sequence of the preprolipoprotein (10–12). In the second step, LspA, a signal peptidase II, cleaves the signal sequence resulting in an apolipoprotein form (12, 13). In the third and final step, Lnt, an apolipoprotein N-acyltransferase, attaches a fatty acyl chain from phosphatidylethanolamine (PE) to the Cys⁺¹ amide, resulting in a mature triacylated lipoprotein (12, 14, 15). Lgt and LspA are both essential and broadly conserved across bacterial species (12). The presence of Lnt is limited to the proteobacterial and actinobacterial classes, although not in an essential role in all members of the former, but it is absent in firmicutes and mollicutes (12). Following their modification at the IM, the mature lipoproteins are sorted to their appropriate membrane layer. Lipoproteins containing Asp immediately following the conserved Cys⁺¹ (referred to as Asp⁺²) remain in the IM. Other mature lipoproteins are extracted from the IM by the OM lipoprotein localization ABC transport machinery, LolCDE, trafficked to the OM by the periplasmic lipoprotein chaperone LolA, and inserted into the OM by LolB, itself an OM lipoprotein (16–19). In certain mutant backgrounds, the existence of an as yet unidentified LolAB-independent trafficking route has been reported (20).

Several substrates of each of the two OM protein transport systems are essential for survival, and therefore E. coli has evolved elaborate mechanisms for sensing and responding to their functional perturbations (19, 21–24). Recent work has uncovered the importance of the two-component Cpx envelope stress response (ESR) system in coping with defects associated with lipoprotein trafficking to the OM (25, 26). The OM lipoprotein NlpE stalls at the IM due to inhibited Lol transport and takes part in a direct interaction with the sensor kinase, CpxA, which relays a signal to CpxR, thereby activating the cpxR regulon (25, 26). A similar phenomenon is observed for the Rcs ESR that is responsible for sensing OM and PGN perturbations (22, 27, 28). In this case, perturbations to OM lipoprotein trafficking trigger strong induction of the Rcs regulon through the interaction of the OM lipoprotein RcsF with an IM Rcs phosphorelay component, thus liberating it from repression (24).

Incomplete maturation or aberrant trafficking of lipoproteins to their appropriate destinations leads to a variety of cell envelope integrity defects. In this work, we address one such instance, the conditional lethality of an E. coli double deletion of yciB, encoding a predicted polytopic IM protein, and dcrB, an IM lipoprotein gene, in low-salt medium (29). Prior evidence revealed that yciB dcrB synthetic lethality is primarily due to the accumulation of the major OM lipoprotein Lpp, which normally links PGN to the OM, at the IM, resulting in abnormal and toxic PGN-IM linkages (29, 30). Previous results also pointed to mislocalization of OMPs at the IM of the yciB dcrB double mutant, and intriguingly, the viability of the yciB dcrB double mutant was restored by the deletion of the periplasmic chaperone gene, skp (29). Here, we show that the molecular basis of skp inactivation-mediated suppression of the double mutant lethality is indirect and mediated through the σ^E-MicL-Lpp regulatory loop to downregulate the synthesis of Lpp. The σ^E ESR system is activated in response to misfolded proteins and upregulates the small RNA MicL to specifically repress Lpp synthesis (31). Further examination of the nature of the IM-mislocalized Lpp-related toxicity and the resultant activation of the Cpx and Rcs ESR systems led us to identify defects in lipoprotein maturation as responsible for the loss of viability of yciB dcrB cells. Specifically, DAG transfer from PG to preprolipoproteins by Lgt is inefficient, which leads to aberrant localization of Lpp at the IM, where it lethally links the IM to the PGN. Other OM lipoproteins, including the stress sensor RcsF, and constituents of the Bam complex, BamC and BamE, also stall at the IM of the double mutant, supporting widespread effects on lipoprotein maturation in these cells. We show that yciB dcrB lethality can be overcome by two distinct mechanisms at the very least—the overexpression of Lgt leading to more efficient lipoprotein biogenesis or the deletion of lpp itself or Lpp-mediated linkages to the peptidoglycan. Furthermore, yciB dcrB null cells display a reduction in membrane fluidity. Taken together, we suggest that membrane fluidity, perhaps due to an imbalance in lipid homeostasis, is a factor in delaying lipoprotein maturation in yciB dcrB mutant cells. Finally, we demonstrate that dcrB is essential when cells are grown under low-salt and low-temperature conditions, and viability can be restored by the same set of genetic suppressors of yciB dcrB lethality. This result is consistent with an essential role for DcrB in lipoprotein maturation under conditions when membrane homeostasis may be altered.

RESULTS

Suppression of yciB dcrB mutant lethality by skp deletion is micL dependent.

We previously found that deletion of the periplasmic chaperone gene, skp, suppressed yciB dcrB synthetic lethality (29). To gain further insight into the mechanism of suppression by skp, we examined the impact of a skp deletion on ESR systems in the yciB dcrB mutant. We employed chromosomal promoter-lacZ fusions to proxy genes (rprA, cpxP, spy, pspA, and rpoH) representing the ESR pathways and measured LacZ expression by β-galactosidase assays. An approximately 3- to 4-fold increase in σ^E activity in the yciB dcrB double and skp single mutant cells compared to the wild type (WT) was noted (Fig. 1A). The yciB dcrB skp triple mutant displayed a 4-fold increase in transcriptional activity compared to that of the yciB dcrB double or the skp null mutants (Fig. 1A). Apart from a 2-fold reduction in the Bae system, no substantial changes in transcriptional activity were noted from representative Cpx, Rcs, and Psp promoters (see Fig. S1A in the supplemental material).

FIG 1.

Suppression of yciB dcrB mutant lethality by skp deletion is micL dependent. (A) A chromosomal lacZ fusion to a target gene promoter of the RpoE ESR, rpoH, was assayed for β-galactosidase activity in the wild type (WT) (AM1251) and the derivative mutants skp (AM1825), yciB dcrB (AM1266), and yciB dcrB skp (AM1680). Overnight cultures grown under permissive conditions were subcultured 1:100 in LB (0% NaCl) and grown for 2.5 h at 37°C. Transcriptional activity was determined as described in Materials and Methods. Mean values and standard deviations are derived from triplicate readings of 3 or more independent cultures. (B) Spot viability of the WT and derivative mutants, yciB dcrB (AM519), yciB dcrB skp (AM866), yciB dcrB skp micL (AM1650), yciB dcrB surA (AM975), and yciB dcrB surA micL (AM1765) mutants. Strains grown under permissive conditions overnight were normalized to an OD₆₀₀ of 1.0 and serially diluted, and 4 μl of each dilution was spotted on LB (0% NaCl) agar and incubated at 37°C overnight prior to imaging. (C) A schematic depicting the mechanism of periplasmic chaperone gene skp or surA deletion-mediated activation of the RpoE ESR and the resultant downregulation of Lpp synthesis. Under normal conditions, Skp and SurA chaperones (green) contribute to the folding and transport of OMPs. Deletion of skp or surA inhibits proper assembly of OMPs (uOMPs), which leads to the activation of rpoE. As part of the RpoE regulon, the small RNA micL is upregulated, which in turn represses the translation of Lpp. IM, inner membrane; PGN, peptidoglycan; OM, outer membrane; uOMP, unfolded OMP.

Next, we considered whether increased σ^E activity in the yciB dcrB skp deletion mutant was responsible for the suppression of yciB dcrB synthetic lethality. The primary role of the σ^E ESR is to serve as a quality control mechanism of OMP assembly (26, 27). This is achieved, in part, by the expression of small RNAs (sRNAs) that bind to and inhibit translation of the mRNAs encoding several OM proteins, including Lpp, which links PGN to the OM (31–34). Since lethality of the yciB dcrB double mutant is primarily due to inappropriate PGN-IM linkages mediated by Lpp, it was possible that the skp deletion served as a mutational suppressor of yciB dcrB by increasing σ^E activity and therefore enhancing MicL sRNA amounts, which in turn lead to decreased Lpp levels, as lpp mRNA is the sole target of interference of the micL sRNA (31, 35). Hence, we generated a yciB dcrB skp micL quadruple-deletion mutant to test the prediction that skp suppression was MicL dependent. Indeed, micL was required for restoration of viability in the yciB dcrB skp strain (Fig. 1B and C). The micL deletion did not adversely affect yciB dcrB mutant fitness (Fig. S1B). These results point to the observed yciB dcrB mutant suppression by a skp null mutation being an indirect effect of σ^E activation, which in turn leads to repression of Lpp translation through MicL. Corroborating these results, a deletion of the major periplasmic chaperone gene surA, which is also known to dramatically increase σ^E activity under standard laboratory growth conditions, served as a potent suppressor of yciB dcrB lethality in an micL-dependent fashion (36) (Fig. 1B and C). Thus, all reported suppressors of yciB dcrB synthetic lethality appear to act through reduction in Lpp levels or loss of activity, including previously reported null mutations in lpp itself and in ldtB, a gene encoding a l,d-transpeptidase considered to be a major factor in linking Lpp to the PGN, or lpp^ΔK58, which is unable to connect to the PGN (29).

Lpp-related toxicity in yciB dcrB is not associated with lipoprotein trafficking defects.

Our earlier work suggested that the essential OM lipoprotein BamD might also be present at the IM in yciB dcrB cells, suggesting that some step in lipoprotein biogenesis may be defective in the double mutant (29). To determine the identity of the step that may be affected, we first investigated whether lipoprotein trafficking to the OM was impaired in these mutant cells. Recent studies have highlighted the importance of Cpx in coping with OM lipoprotein trafficking defects and identified the OM sentry lipoprotein NlpE as critical for activating the Cpx ESR in response to lipoprotein trafficking stress (25, 26). Previously we showed that the Cpx ESR is significantly activated in yciB dcrB cells and, under permissive conditions of growth (LB [1% NaCl]), CpxR is required for growth of the double mutant (29) (Fig. 2A). To determine whether NlpE mediates the increased Cpx activation observed in yciB dcrB cells, we introduced an nlpE deletion into this background. We observed that a yciB dcrB nlpE triple deletion mutant did not phenocopy a yciB dcrB cpxR triple mutant in terms of reduced viability under permissive conditions of growth (Fig. 2A). Furthermore, deletion of nlpE did not significantly alter cpx activation levels in the yciB dcrB double mutant (Fig. 2B). Thus, NlpE is not required for Cpx upregulation in yciB dcrB cells. These results are consistent with previous findings wherein overexpression of the LolCDE transporter in yciB dcrB cells did not rescue viability, suggesting Lpp trafficking is not affected (29). Since NlpE was not required for Cpx activation, we analyzed a possible role for YafY, an IM lipoprotein of unknown function whose overexpression is known to induce the Cpx system (37). The deletion of yafY did not reduce viability of the yciB dcrB double mutant, suggesting that yafY does not mediate the observed cpx activation in the mutant background (see Fig. S2 in the supplemental material).

FIG 2.

cpxR activation in a yciB dcrB mutant is largely NlpE independent. (A) Spot viability of the WT and derivative yciB dcrB (AM519), yciB dcrB cpxR (AM1646), and yciB dcrB nlpE (AM974) mutants. Strains grown under permissive conditions overnight were normalized to an OD₆₀₀ of 1.0 and serially diluted, and 4 μl of each dilution was spotted on LB (1% NaCl) agar and incubated at 37°C overnight prior to imaging. (B) A chromosomal lacZ fusion to a target gene promoter of the Cpx ESR, cpxP, was assayed for β-galactosidase activity in the WT (AM1245) and derivative yciB dcrB (AM1263) and yciB dcrB nlpE (AM1665) mutants. Overnight cultures grown under permissive conditions were subcultured 1:150 under LB (1% NaCl) or restrictive LB (0% NaCl) conditions and grown at 37°C for 2.5 h prior to isolating cells for analysis. Transcriptional activity was determined as described in Materials and Methods. Mean values and standard deviations are derived from triplicate readings of 3 or more independent cultures.

Expression of Lgt in trans restores viability to the yciB dcrB double mutant.

We next probed whether lipoprotein maturation defects were the cause for Lpp, and perhaps other OM lipoproteins, to be mislocalized to the IM in yciB dcrB cells. An important clue came from the striking morphological similarities of lgt-depleted and yciB dcrB mutant strains (38) (Fig. 3A). As mentioned earlier, Lgt, a PG-preprolipoprotein diacylgyceryl transferase, catalyzes the essential first step in lipoprotein maturation at the IM, where it adds a DAG group derived from a PG substrate to preprolipoprotein substrates. Both mutants display club-shaped cells, dramatic cytoplasmic shrinkage with concomitant large phase-light regions at or near cell poles; as reported previously, the morphological defects in yciB dcrB cells are observable even in the presence of salt (29) (Fig. 3A).

FIG 3.

Lgt overproduction restores viability to a yciB dcrB mutant. (A) Representative microscopy images showing similar cell death trajectories of yciB dcrB deletion (AM519; i and iii), and lgt depletion (PAP9403; ii and iv) strains. Cells were grown under permissive conditions (LB [1% NaCl] with 0.02% l-arabinose) overnight, subcultured 1:150 in LB (1% NaCl), and grown at 37°C. An aliquot of the cultures was visualized by phase-contrast microscopy at the following time points of growth: (i) 3 h, (ii) 2.5 h, (iii) 5 h, and (iv) 3 h. Arrows point to “void” regions resulting from cytoplasmic shrinkage in the cells and arrowheads point to cell lysis. Bar, 5 μm. (B) Spot viability of the WT bearing pTrc99a empty vector or with lgt (AM1774) and isogenic yciB dcrB mutant with vector alone or with lgt (AM1676). Strains were grown under permissive conditions overnight, normalized to an OD₆₀₀ of 1.0, and serially diluted, and 4 μl of each dilution was spotted on LB (0% NaCl) agar plates containing ampicillin and β‐d‐1‐thiogalactopyranoside (IPTG) at indicated concentrations. Plates were incubated overnight at 37°C prior to being imaged. (C) Spot viability of the WT and isogenic yciB, dcrB, and yciB dcrB mutants with either pDSW206 empty vector or with lgt (AM1441, AM1442, and AM1436). Strains were grown under permissive conditions overnight, normalized to an OD₆₀₀ of 1.0 and serially diluted, and 4 μl of each dilution was spotted on LB (1% NaCl) agar plates with ampicillin, IPTG (100 μM), and various concentrations of CuCl₂ as indicated. Plates were incubated overnight at 37°C and imaged the next day.

We determined whether the cell envelope defects exhibited by the yciB dcrB mutant were a result of impaired Lgt activity. Overexpression of Lgt in trans restored growth to the yciB dcrB mutant, but high levels of Lgt were toxic to WT cells, implying that lipoprotein maturation may indeed be defective in the mutant (Fig. 3B). Acidic phospholipids (PL) serve as donor substrates for the Lgt-mediated step in lipoprotein maturation (38, 39). To rule out the possibility of a deleterious flux of PLs toward pathways that would lead to alterations in PL substrate levels and therefore in Lgt activity, we overexpressed pgsA and pssA from low-copy-number vectors in the yciB dcrB background and checked for changes to viability under restrictive growth conditions. PgsA (phosphatidylglycerophosphate synthase) and PssA (phosphatidylserine synthase) control the metabolic branch point in utilizing cytidine diphosphate diacylglycerol (CDP)-DAG as a substrate for bulk PL synthesis by controlling the flux toward PG and PE synthesis, respectively (40) (see Fig. S3A in the supplemental material). Increased expression of either of these proteins did not alter yciB dcrB mutant viability (Fig. S3B). A control test indicated that the levels of plasmid-borne pgsA expression were sufficient to suppress increased vancomycin resistance of pgsA444, a point mutant with reduced levels of PG (41) (Fig. S3C). Furthermore, genetic mutations which eliminate the synthesis of osmoregulated periplasmic glucans (OPGs), a source of PG consumption under low-salt conditions, did not suppress yciB dcrB lethality in LB (0% NaCl) (42) (Fig. S3A and D). Collectively, these data add to our previous thin-layer chromatography (TLC) finding that alterations in the abundance of major PLs are not the prominent factor underlying yciB dcrB synthetic lethality (29).

Lipoprotein maturation is sensitive to copper, which is posited to inhibit Lgt catalysis by binding to the Cys⁺¹ residue of lipoproteins (25, 43). Indeed, the yciB dcrB mutant displayed a marked sensitivity to copper ions at concentrations as low as 1 mM (Fig. 3C). The yciB and dcrB single deletions were also mildly sensitive to Cu²⁺ with MICs at 2.25 mM and 2.75 mM copper, respectively, compared to WT cells. Overexpression of Lgt restored growth to the single and double mutants, suggesting that the copper sensitivity displayed by the mutants could be a result of inhibited Lgt function (43) (Fig. 3C). These data support the hypothesis that yciB dcrB lethality under nonpermissive conditions is due to defective lipoprotein maturation at the first step of diacylglyceryl transfer to preprolipoproteins.

A yciB dcrB double mutant accumulates additional forms of Lpp.

Next, we investigated the possible impairment of lipoprotein maturation by probing for incompletely modified forms of Lpp in yciB, dcrB, and yciB dcrB mutants. Cells depleted of Lgt were seen to accumulate unmodified, partially modified, and PGN-linked Lpp forms in addition to the mature Lpp (38, 39) (Fig. 4A). Notably, the yciB dcrB cells accumulated detectable levels of additional Lpp forms, even under permissive conditions of growth, whose sizes and relative electrophoretic mobilities were reminiscent of those in the lgt depletion mutant (Fig. 4B and C). These Lpp forms were not observed in the yciB or the dcrB single mutants (see Fig. S4A in the supplemental material). Supporting evidence came from examining the lpp^ΔK58 allele, which produces an Lpp variant that is unable to be covalently linked to the PGN through its C-terminal lysine (K58) residue, in lgt and yciB dcrB backgrounds (44, 45). In both lgt lpp^ΔK58 and yciB dcrB lpp^ΔK58 mutants, the slow-migrating band (“a”) was not detectable, but the one above (“b”) was, indicating that the former band represents a PGN-bound form of Lpp (Fig. 4B and Fig. S4B).

FIG 4.

Lipoprotein maturation is impaired in a yciB dcrB mutant. (A) Lipoprotein maturation pathway. Preprolipoproteins (Pre-pro) are secreted through the Sec or Tat translocons. Lgt modifies the protein by transferring a diacylgyceryl moiety from PG to an invariable Cys residue in the lipobox to form the prolipoprotein (Pro). In the second step, Lsp cleaves the signal peptide, leaving the apolipoprotein (Apo) embedded at the IM. In some bacteria, including E. coli, the apolipoprotein is further processed by Lnt, which adds another acyl chain to form a mature triacylated lipoprotein (Mature) that can be either retained at the IM or transported to the OM. (B) Additional forms of the abundant OM lipoprotein Lpp are present in the yciB dcrB double mutant. Immunoblots from cell lysates of the WT and the following mutants: lgt (PAP9403), yciB dcrB (AM562), and yciB dcrB lpp^ΔK58 (AM1715). Cells were grown overnight under permissive conditions (LB 1% NaCl with 0.02% l-arabinose) and diluted 1:150 in LB (1% NaCl) and grown for 3 h at 37°C. Lysates were prepared and analyzed by immunoblotting with an anti-Lpp polyclonal antibody, as described in Materials and Methods. Slow-migrating additional forms of Lpp (marked “a” and “b”) of similar sizes and relative mobilities are present in in both lgt depletion and yciB dcrB mutants. The middle band (a) is barely detectable in the yciB dcrB lpp^ΔK58 strain, suggesting that it is a PGN-bound form, and the top band (b), present in both mutant backgrounds, is probably an immature variant that is not linked to PGN. (C) Immunoblot result from lysates of the WT and mutant yciB dcrB with pTrc99a empty vector or expressing lgt (AM1711). Cells were grown as described for panel A; however, ampicillin and IPTG (10 μM) were included to maintain expression of lgt. Immunoblotting was performed as described above. (D and E) Chromosomal lacZ fusions to target gene promoters of the Rcs and Cpx ESR, rprA and cpxP, were assayed for β-galactosidase activity in WT (TB28) strains with empty pDSW206 vector or with lgt (AM1249, AM1245, AM1755, and AM1754) and yciB (AM1255, AM1253, AM1757, and AM1756), dcrB (AM1260, AM1258, AM1759, and AM1758), and yciB dcrB (AM1265, AM1263, AM1761, and AM1760) mutant derivatives. Transcriptional activity of rprA and cpxP was also monitored in the yciB dcrB lpp triple mutant background with empty pDSW206 vector or with lgt (AM1779, AM1815, AM1740, and AM1816). Overnight cultures grown under permissive conditions were subcultured 1:150 in LB (1% NaCl) and grown at 37°C until an OD₆₀₀ of 0.25 to 0.35. Ampicillin and IPTG (100 μM) were included for the plasmid-bearing strains. Transcriptional activity was determined as described in Materials and Methods. Mean values and standard deviations are derived from triplicate readings of 3 or more independent cultures.

Prior reports have suggested that the middle and top slower-migrating bands in lgt depletion cells are likely to represent the prepro-Lpp and prepro-Lpp–PGN-linked forms, respectively (38, 39). We utilized the same lgt depletion strain as in prior studies, but our present deployment of the lpp^ΔK58 allele has led us to assign the middle band (a) to a PGN-linked form, and the one above it (b) to a partially mature form of Lpp that is not linked to PGN, an interpretation that is at odds with earlier descriptions (25, 38) (Fig. 4B and C and Fig. S4A and B). The precise biochemical identities of the additional Lpp variants present in these strain backgrounds will require detailed mass spectrometry analysis. Regardless of the specific chemical nature of the bands, our results confirm that yciB dcrB cells accumulate Lpp forms similar in sizes and relative electrophoretic mobilities to those in Lgt-depleted cells.

Next, we examined whether expression of lgt in trans, shown to restore growth to a yciB dcrB mutant, would have an impact on the accumulation of the additional Lpp forms described above. Indeed, unmodified or PGN-linked Lpp forms were not detectable in a yciB dcrB mutant overexpressing Lgt (Fig. 4C). Together, these results suggest that the accumulation of additional Lpp forms in a yciB dcrB mutant is due to reduced Lgt-catalyzed modification of Lpp. We confirmed that the attenuation in Lgt function was not a consequence of reduced transcription or protein levels (Fig. S4C and D).

Increased expression of lgt reduces Rcs but not Cpx activation to WT levels in a yciB dcrB mutant.

A reduction in the diacylglyceryl modification of lipoproteins is expected to result in the accumulation of additional OM lipoproteins, including RcsF, in the IM (46). We had previously reported increased Rcs activity and the appearance of a capsule-linked lipopolysaccharide (LPS) form, known as M_LPS, in the yciB dcrB double mutant (29). We hypothesized that these phenotypes were likely associated with the accumulation of OM lipoproteins, including RcsF at the IM, as a result of reduced Lgt activity. To test this idea, we monitored the activity of a chromosomal rprA-lacZ reporter in the context of overexpression of Lgt in strains with deletions in yciB, dcrB, or both. The transcriptional activity analyses for Rcs (and Cpx reported below) were conducted in medium containing salt, in which, as discussed previously, Lpp maturation defects are evident (29).

Rcs activation was comparable to WT levels in yciB and dcrB single mutants but elevated by more than 10-fold in the yciB dcrB double mutant (Fig. 4D). The Rcs upregulation in the latter was largely dependent on the stress sensor RcsF (Fig. S4E). Indeed RcsF, and lipoproteins BamC and BamE, which are constituents of the OM Bam complex (5), showed a broad distribution in the membrane gradient for yciB dcrB lpp^ΔK58 compared to the WT (Fig. S4F). As a control, we confirmed that Lpp^ΔK58 is mislocalized to IM fractions of the gradient in the yciB dcrB lpp^ΔK58 triple mutant (Fig. S4F). Upon expression of Lgt in trans, Rcs activity was found to be reduced to near WT levels in the double mutant, suggesting that increased Lgt levels reduced the improper localization of RcsF at the IM (Fig. 4D). Next, we tested the impact of the deletion of lpp on Rcs activation in the yciB dcrB double mutant; an lpp null mutation restores growth to the yciB dcrB mutant (29). While the yciB dcrB lpp triple mutant displayed reduced Rcs activation compared to that of the yciB dcrB mutant, Rcs levels were found to be ∼2-fold higher than those of yciB dcrB cells overexpressing Lgt (Fig. 4D). Similar levels of Rcs activity could be recovered upon increased expression of lgt in the triple mutant background (Fig. 4D). These data together suggest that OM localization of both stress-related and non-stress-related lipoproteins is altered in the yciB dcrB double mutant. While removal of the abundant cargo Lpp reduces the lipoprotein maturation stress, lipoprotein biogenesis problems persist, as revealed by elevated Rcs levels.

To address if increased lgt expression in yciB dcrB cells reduces activation of Cpx, we monitored cpxP-lacZ activity in the yciB and dcrB single and double mutant strains. The double mutant displayed an approximately 5-fold increase in Cpx activation (Fig. 4E). In contrast to the strong mitigatory effect on the Rcs response, overexpression of lgt only reduced Cpx activation in yciB dcrB cells by one-third (Fig. 4E). A yciB single mutant alone showed a 3-fold increase in Cpx activation, whereas a dcrB null mutant did not reveal significant differences in Cpx upregulation compared to the WT (29) (Fig. 4E). Notably, cpx activation was reduced on average by a quarter in the yciB null mutant upon Lgt expression (Fig. 4E). Since a yciB mutant does not display detectable levels of immature or atypical forms of Lpp, nor does it exhibit significantly increased Rcs activation, the major stress signal sensed by Cpx in a yciB deletion is likely unrelated to lipoprotein maturation defects. In the yciB dcrB lpp triple mutant, with or without Lgt expression in trans, Cpx upregulation was partially reduced and resembled levels similar to those in the yciB single mutant expressing Lgt (Fig. 4E). This suggests that the Cpx upregulation observed in the yciB dcrB mutant is only in part attributable to the stress associated with lipoprotein maturation. An additional unidentified stress stimulant, likely contributed by the deletion of yciB, triggers Cpx activity in the double mutant.

DcrB is essential for lipoprotein maturation under conditions where membrane fluidity may be altered.

Towards defining the individual contributions of yciB and dcrB in the observed synthetic lethality, we examined whether mutational or multicopy suppressors of the yciB dcrB double mutant would rescue growth of the single mutants under conditions in which each exhibits reduced growth. Previously, we had reported that a yciB deletion shows reduced growth in the presence of the hydrophilic antibiotic vancomycin or the detergent SDS in the presence of the chelator EDTA, both indicative of OM integrity defects (29). We tested whether deletions in lpp or skp or increased expression of lgt would rescue the yciB null mutant from its sensitivity to vancomycin or SDS-EDTA. Restoration of robust viability to the yciB null mutant was not observed, suggesting that the yciB mutant-related OM permeability defects could not be suppressed in these backgrounds (Fig. S5A and B).

During the course of our studies, we observed poor viability of the yciB dcrB mutant when grown at lower temperatures, even in the presence of salt (LB [1% NaCl]; 20°C) (Fig. 5A). Strikingly, the dcrB null mutant alone showed significant reduction in growth compared to that of the WT or a yciB mutant when incubated at 20°C or below in LB (0% NaCl) (Fig. 5A). This suggested that dcrB is essential for E. coli viability under salt stress at low temperatures. Next, we asked whether yciB dcrB genetic suppressors, described above, could restore growth of the dcrB null mutant under these restrictive conditions. Indeed, deletion of lpp, lpp^ΔK58, or skp restored the viability of the dcrB mutant, as did overexpression of lgt (Fig. 5A). The observation that the lethality of a dcrB mutant can be rescued by the same genetic suppressors of a yciB dcrB double mutant suggests that the underlying defects in a dcrB single mutant mimic those of the yciB dcrB double mutant. Therefore, it is likely that the dcrB mutant is unable to grow under restrictive conditions because of IM-localized Lpp-mediated toxic linkages to the PGN resulting from suboptimal Lgt activity. E. coli cells grow as mucoid colonies at low temperatures due to elevated capsular synthesis; the transcriptional regulator RcsB controls the expression of capsular polysaccharide synthesis (47) (Fig. 5B). Interestingly, overexpression of Lgt in WT E. coli reduced the mucoid appearance, suggesting that lipoprotein maturation may, in general, be impaired when E. coli is grown under low temperature and salt stress (Fig. 5B). Together, these data suggest that dcrB is essential for growth under the assayed stress conditions due to its role in lipoprotein maturation.

FIG 5.

Genetic suppressors of yciB dcrB lethality can restore dcrB null viability under restrictive conditions of low-salt and low-temperature growth. (A) Spot viability of the WT and derivative mutants yciB (AM134), dcrB (AM135), yciB dcrB (AM519), dcrB skp (AM871), dcrB lpp (AM814), dcrB lpp^ΔK58 (AM1714), and dcrB with lgt (AM1772). Strains were grown under permissive conditions overnight, normalized to an OD₆₀₀ of 1.0, and serially diluted, and 4 μl of each dilution was spotted on LB (0% NaCl) and LB (1% NaCl) agar plates. Ampicillin and IPTG (100 μM) were added to maintain the expression of lgt-bearing strains. Plates were incubated at 20°C for approximately 72 h prior to being imaged. (B) Lipoprotein maturation appears to be inefficient in WT cells grown under low-salt, low-temperature conditions (LB [0% NaCl]; 20°C). Shown are the WT with empty pTrc99a vector or with lgt (AM1774) and an isogenic rcsB null mutant with the empty pTrc99a (AM1780). Strains were grown under permissive conditions overnight, and serially diluted, and 100 μl of cells was spread on LB (0% NaCl) agar containing ampicillin and IPTG (100 μM). The plates were incubated at 20°C for 72 h prior to imaging the colonies (Gel-Doc system; Syngene). (C) Laurdan generalized polarization (GP) measurements in the WT and isogenic mutants yciB (AM134), dcrB (AM135), and yciB dcrB (AM519). Cells were grown and treated as described in Materials and Methods. The average GP values and standard deviations are plotted for each strain. The circles represent mean values of triplicate technical repeats of 3 or 4 independent biological cultures. WT cells treated with the membrane fluidizer benzyl alcohol (BA) represent a positive control for the assay. Laurdan GP values are inversely correlated with membrane fluidity.

The membranes of E. coli cells are known to have reduced fluidity when grown at low temperatures (48). The altered fluidity may in turn influence membrane-associated processes such as lipoprotein maturation. To understand whether the yciB dcrB mutant displays changes to membrane fluidity, we conducted a generalized polarization (GP) assay using the fluorescent dye Laurdan, a known reporter of lipid packing in cells (49, 50). An increase in GP, which is inversely correlated with fluidity, was noted in yciB dcrB cells compared to the WT or the single mutants (Fig. 5C). Additionally, an expected decrease in GP was observed in WT cells treated with benzyl alcohol, a known membrane fluidizer, and served as a control for the assay (Fig. 5C). The Laurdan spectroscopy results suggest that there is an overall increase in lipid ordering in cells lacking both YciB and DcrB.

DISCUSSION

We had previously discovered an essential synergistic relationship of yciB, encoding a predicted polytopic IM protein, with dcrB, an IM lipoprotein gene, in maintaining the integrity of the cell envelope (29). Building on that discovery, we demonstrate here that the conditional lethality of yciB dcrB cells results from lipoprotein maturation defects. We attribute these effects to a reduction in the Lgt-catalyzed DAG transfer to preprolipoproteins and the resultant mislocalization of several OM lipoproteins, including Lpp, at the IM, where Lpp mediates toxic IM-PGN associations (29). Both Rcs and Cpx ESR systems are upregulated in yciB dcrB cells, signaling defects in lipoprotein maturation. While Rcs activation is predominantly mediated by the accumulation of RcsF at the IM, Cpx upregulation occurs independent of NlpE. The inefficiency in Lgt function could be the result of altered membrane fluidity observed in the double mutant, perhaps a consequence of altered fatty acid composition. Consistent with this hypothesis, dcrB serves an essential role in lipoprotein maturation under conditions known to affect membrane homeostasis.

Deletion of YciB and DcrB leads to defects in lipoprotein maturation.

Several lines of evidence suggest that the deletion of yciB and dcrB leads to inefficient lipoprotein maturation at the level of the Lgt-catalyzed DAG transfer to preprolipoproteins. These include the following: (i) the morphological similarities in cell death trajectories exhibited by the yciB dcrB null mutant with that of an lgt depletion mutant; (ii) the accumulation of unmodified or partially modified forms of Lpp in the yciB dcrB mutant whose sizes and relative gel mobilities resemble those in Lgt-depleted cells; (iii) the suppression of yciB dcrB lethality as well as of the accumulated additional Lpp forms in the double mutant by increased expression of lgt; and (iv) the mitigation of Rcs activation in yciB dcrB cells to near-WT levels upon lgt expression in trans.

While additional experiments are required to uncover the precise mechanisms by which Lgt activity is impaired in the yciB dcrB double mutant, existing observations provide some broad insight in this regard. We find no significant differences in Lgt protein levels in yciB dcrB cells grown under restrictive conditions to those in WT cells (see Fig. S4D in the supplemental material). But elevated Lgt expression does bypass the requirement of YciB and DcrB under these nonpermissive growth conditions. It is therefore possible that the enzymatic properties of Lgt are compromised under these stress conditions, perhaps through the requirement of additional effectors for optimal function. This is in apparent contradiction to prior reports that state that Lgt levels are not directly correlated with functionality or that Lgt activity does not require complex formation with other proteins (11, 38). However, it is crucial to note that these studies did not test for Lgt functionality under the restrictive growth conditions of this work.

Lgt catalyzes a complex reaction at the IM, where the enzyme, the protein substrate, and the PG substrate are all housed within the membrane environment. It is therefore plausible that changes to lipid homeostasis, as reflected by the altered overall membrane fluidity in yciB dcrB cells, may result in reduced Lgt activity. Remarkably, the low-salt growth sensitivity phenotype of the yciB dcrB double mutant, along with the genetic suppressors of toxicity, namely, lpp null and overexpression of lgt, parallels those of a pgsA null mutant that has an abrogated phosphatidylglycerophosphate synthase activity and the resultant lack of acidic PLs, PG, and cardiolipin (51). However, our earlier results showed that a yciB dcrB double mutant has no significant differences in the overall abundance of major PLs compared to the WT (29), nor does expression of pgsA in trans rescue the growth of the yciB dcrB strain (see Fig. S3B in the supplemental material). Together, these results suggest that limiting amounts of PG substrate is unlikely to be a major contributor to the lipoprotein maturation defect in yciB dcrB cells. Interestingly, in vitro biochemical studies on PL substrate specificity using purified Lgt reveal that the glycerol head group and composition of fatty acid chains do impact the activity of the protein (11). Our earlier TLC analysis does not resolve the effects of these changes to the PL composition (29). It is well established that key cellular parameters such as membrane fluidity and fatty acyl chain homeostasis such as saturation are altered in response to temperature stress (48). Coincidentally, WT E. coli cells form mucoid colonies under low-salt and low-temperature stress, and this mucoid phenotype can be suppressed by Lgt overexpression (Fig. 5B). This suggests that Lgt activity, and in turn the fidelity of the first step of lipoprotein maturation, may in fact be fine-tuned to the physicochemical properties of the membrane. The possible roles of YciB and DcrB in affecting Lgt activity under these stress conditions are discussed further below.

Defects in early lipoprotein maturation steps lead to Cpx activation in a largely NlpE-independent manner.

Both Cpx and Rcs ESR systems are upregulated in a yciB dcrB mutant. However, NlpE is not the primary sensor for Cpx activation in the yciB dcrB mutant (Fig. 2B). This observation is compatible with earlier reports of NlpE-independent Cpx activation either in response to defects during the early steps of lipoprotein biogenesis (catalyzed by LspA or Lgt) or as a result of the reduced availability of PG or PE, which are substrates for Lgt or Lnt-mediated lipid modification of lipoproteins at the IM (25, 26, 52, 53). The precise identity of the stimulus for Cpx in each of these instances is unknown. The high levels of Rcs activation in the yciB dcrB mutant are stimulated by RcsF mislocalized at the IM and can be restored to near-WT levels by increased lgt expression (Fig. 4D). This supports the interpretation that lipoprotein maturation is defective in yciB dcrB mutant cells (Fig. 4D). Our current results and an earlier report for a pgsA null mutant both point to the ability of RcsF at the IM to transmit lipoprotein maturation stress signal to the Rcs system (Fig. S4F) (46). However, in the case of NlpE, it is possible that the mature form of the protein or sufficient amounts of NlpE accumulation at the IM are necessary for Cpx induction.

A comparison of cpx activation levels in the yciB and dcrB single and yciB dcrB double mutants revealed that lipoprotein maturation defects are not the sole factor leading to Cpx activation in yciB and yciB dcrB mutants. While Cpx transcriptional activity is already high in the yciB single mutant and further elevated in the yciB dcrB double mutant, the expression of Lgt in trans reduces cpx levels in the double mutant, but only to levels observed in the yciB null mutant (Fig. 4E). Furthermore, the yciB single mutant does not accumulate detectable levels of incompletely modified or additional forms of Lpp or display high Rcs activity (Fig. S4A and D). This indicates that lipoprotein maturation defects are not a prominent stress factor in this single mutant. Indeed, the loss of yciB provokes a Cpx response; however, the precise nature of this stress is currently unknown.

Roles for YciB and DcrB in lipid homeostasis.

In an earlier working model, we postulated that the periplasmic chaperone Skp contributes to yciB dcrB mutant lethality through the incorrect insertion of porins into the IM (29, 54). This conjecture was based on the observation that both a yciB null mutant and a yciB dcrB double mutant showed a reduction in the proton motive force (PMF), and the inactivation of skp partially restored PMF in the double mutant (29). Here, we discovered that the leading cause of skp deletion-mediated suppression of yciB dcrB lethality is dependent on MicL, the sRNA inhibitor of Lpp translation, and is in fact due to a reduction in Lpp synthesis (Fig. 1). The dissipation of PMF in the double mutant, therefore, does not appear to be the central effect on viability under the assayed conditions and was not considered further in this study. Since the genetic suppressors of yciB dcrB mutant lethality isolated thus far suggest the loss of Lpp, the untethering of Lpp-PGN links, and the increased expression of Lgt as major routes of rescue, we have updated our working model accordingly (Fig. 6).

FIG 6.

Working model for YciB and DcrB roles in membrane homeostasis. The IM proteins YciB and DcrB influence membrane fluidity, perhaps by influencing lipid composition in the cell. In the yciB dcrB double mutant, these processes are impaired and the Lgt-catalyzed step of lipoprotein maturation slows down, leading to the accumulation of OM lipoproteins at the IM, including that of the abundant OM lipoprotein Lpp. The incompletely modified forms of Lpp at the IM covalently tether the IM to the PGN, leading to cell death (for ease of representation, only the prepro-Lpp form is shown here). Both Cpx and Rcs ESR systems are activated in response to the lipoprotein maturation stress, the latter mainly due to the presence of the OM lipoprotein RcsF at the IM. The modes of suppression of yciB dcrB toxicity and the dcrB defects at low temperatures are consistent with roles for YciB and DcrB in membrane homeostasis. Overproduction of Lgt likely improves the efficiency of acyl chain transfer to preprolipoprotein substrates such that Lpp is targeted to its proper OM destination, where it mediates PGN-OM linkages. The horizontal and vertical arrows represent the later steps of lipoprotein maturation (Lsp and Lnt mediated) and trafficking to the OM, respectively. The unlinking of PGN from Lpp (either via Lpp^ΔK58 or in an ldtB mutant where Lpp^ΔK58 or Lpp, respectively, is presumably still consuming acyl chains at levels similar to those of Lpp in the yciB dcrB mutant) restores viability to the double mutant. While the primary cause of lethality stems from IM-localized Lpp-mediated IM-PGN linkages, the increased OM vesiculation events, lipid loss, and altered mechanical properties associated with the unlinking of Lpp from the PGN may have an impact on membrane homeostasis. IM, inner membrane; PGN, peptidoglycan; OM, outer membrane; OMV, outer membrane vesiculation.

Recently, it was proposed that OM vesiculation events and concomitant loss of LPS may serve as an adaptive response to elevated LPS levels (55). The prevention of OM-PGN linkages, mediated via Lpp or other OM proteins, namely, OmpA or Pal, was considered to suppress toxic LPS overproduction due to loss of LPS through elevated membrane vesiculation (55, 56). Our earlier work demonstrated that while yciB dcrB double mutant cells do exhibit elevated LPS production and display significantly enhanced vesiculation, increased LPS is not the primary cause of cell death in the double mutant (29). Furthermore, deletions in ompA or pal do not restore viability to yciB dcrB cells (29); therefore, increased OM vesiculation and lipid loss alone are not sufficient to overcome toxicity. Instead, the ability of lpp^ΔK58 or ldtB mutant to restore viability to yciB dcrB cells reveals that cells can bear the burden of OM lipoproteins being present at the IM, as long as Lpp exists in a form that is incapable of linking to the PGN (29). Our work here also demonstrates that the overproduction of Lgt suppresses yciB dcrB mutant defects. This suggests that an increased efficiency of the Lgt-catalyzed lipoprotein maturation prevents the stalling of Lpp at the IM in yciB dcrB cells and that PGN-OM linkages are not harmful to the double mutant. More than 50 years ago, the abundant Lpp was discovered as the first bacterial lipoprotein (57). More recently, the importance of Lpp-mediated PGN-OM linkages in maintaining periplasmic distance, optimal signal communication between the two membranes, and the stiffness of the cell envelope was established (30, 58). Our present work, in combination with previous reports (51, 55), reveals that under specific instances, perhaps when lipid homeostasis is altered, severing Lpp-mediated PGN linkages either from the outside in (PGN-OM) or from the inside out (IM-PGN) is beneficial to the cell.

The reason for impaired DAG transfer to lipoproteins in the dcrB null mutant and the yciB dcrB double mutant strains is not immediately clear. One possibility is that DcrB and YciB play a direct role in altering Lgt activity under the stress conditions employed in this study, perhaps through interactions with Lgt. A second possibility is that YciB and DcrB affect lipid homeostasis and thereby impact Lgt activity. The dcrB-null phenotypes grown under low-salt and low-temperature stress mimic those of the yciB dcrB mutant grown under low-salt stress but higher temperatures (Fig. 5A). Consequently, we hypothesize that inactivation of yciB leads to changes in lipid composition that simulate the physical effects of lower temperatures, and in this background, DcrB becomes essential for lipoprotein maturation. Remarkably, the crystal structure (PDB identifier 6E8A) of the DcrB ortholog in Salmonella shares significant similarity with the mycobacterial periplasmic protein LpqN. LpqN has been suggested to bind lipids, thereby influencing lipid transport and overall cell envelope biology in that bacterium (59, 60) (see Fig. S6 in the supplemental material). That DcrB in E. coli and related bacteria may serve a lipid-binding function, perhaps in concert with IM proteins such as Lgt that are involved in lipid modification pathways, is therefore an attractive possibility.

MATERIALS AND METHODS

Strains, plasmids, primers, and growth conditions.

Strains and plasmids used in this study are listed in Tables 1 and 2 (see also Tables S1 and S2 in the supplemental material). Primers and plasmid construction are described in the supplemental material. Reported E. coli strains are derivatives of MG1655 or BW25113. An micL deletion derivative of MG1655 was generated using the λ Red recombination method (61). Deletion mutants were generated by P1 phage transduction using the Keio mutant strain collection (obtained from the Coli Genetic Stock Center [CGSC]) as donors. For construction of multiple deletion mutants, the chromosomal kanamycin resistance (Kan^r) cassette was excised via FLP-FRT recombination (61). Cells were grown in LB (1% tryptone, 0.5% yeast extract, and 1% NaCl), referred to as LB (1% NaCl), or LB with no NaCl, referred to as LB (0% NaCl). Antibiotics were used at concentrations of 20 μg · ml⁻¹ (chloramphenicol [Cm]), 50 μg · ml⁻¹ (kanamycin [Kan]) or 25 μg · ml⁻¹ when selecting for integrants, 100 μg · ml⁻¹ (ampicillin [Amp]), and 10 μg · ml⁻¹ (tetracycline [Tet]) when selecting for integrants. For induction conditions, l-arabinose at 0.02% or 0.2% final concentrations, and isopropyl-β‐d‐1‐thiogalactopyranoside (IPTG) at various concentrations reported in the figure legends were used.

TABLE 1.

Strains used in this study

Strain	Genotype	Source or referencea
MG1655	K-12 F− λ− IlvG− rfb-50 rph-1	Laboratory collection
TB28	MG1655 lacIZYA::frt	62
PAP9403	BW25113 lgt::Kanr pBAD18s-lgt	35
AM134	MG1655 yciB::frt	29
AM135	MG1655 dcrB::frt	29
AM519	MG1655 yciB::frt dcrB::frt pBAD33-yciB	29
AM562	MG1655 yciB::frt dcrB::frt	29
AM814	AM135 lpp::Kanr
AM866	AM519 skp::frt	29
AM871	MG1655 skp::frt dcrB::Kanr
AM974	AM519 nlpE::Kanr
AM975	AM519 surA::Kanr
AM1137	TB28 yciB::frt	29
AM1138	TB28 dcrB::frt	29
AM1139	TB28 yciB::frt dcrB::frt pBAD33-yciB	29
AM1245	TB28 nadA::Tn10 attλ PcpxP::lacZ	29
AM1249	TB28 nadA::Tn10 attλ PrprA::lacZ	29
AM1251	TB28 nadA::Tn10 attλ PrpoH::lacZ	29
AM1253	AM1137 nadA::Tn10 attλ PcpxP::lacZ	29
AM1255	AM1137 nadA::Tn10 attλ PrprA::lacZ	29
AM1258	AM1138 nadA::Tn10 attλ PcpxP::lacZ	29
AM1260	AM1138 nadA::Tn10 attλ PrprA::lacZ	29
AM1263	AM1139 nadA::Tn10 attλ PcpxP::lacZ	29
AM1265	AM1139 nadA::Tn10 attλ PrprA::lacZ	29
AM1266	AM1139 nadA::Tn10 attλ PrpoH::lacZ	29
AM1436	AM519 pDSW206-lgt
AM1441	AM134 pDSW206-lgt
AM1442	AM135 pDSW206-lgt
AM1646	AM519 cpxR::Kanr	29
AM1650	AM866 micL::Kanr
AM1665	AM1263 nlpE::Kanr
AM1676	AM519 pTrc99A-lgt
AM1680	AM1266 skp::frt
AM1711	AM562 pTrc99a-lgt
AM1714	AM135 lppΔK58 ynhG(ldtE)::Kanr
AM1715	AM562 lppΔK58 ynhG(ldtE)::Kanr
AM1740	AM1263 lpp::Kanr
AM1754	AM1245 pDSW206-lgt
AM1755	AM1249 pDSW206-lgt
AM1756	AM1253 pDSW206-lgt
AM1757	AM1255 pDSW206-lgt
AM1758	AM1258 pDSW206-lgt
AM1759	AM1260 pDSW206-lgt
AM1760	AM1263 pDSW206-lgt
AM1761	AM1265 pDSW206-lgt
AM1765	AM519 micL::frt surA::Kanr
AM1772	AM135 pTrc99a-lgt
AM1774	MG1655 pTrc99a-lgt
AM1779	AM1265 lpp::Kanr
AM1780	MG1655 rcsB::Kanr pTrc99a
AM 1815	AM1779 pDSW206-lgt
AM1816	AM1740 pDSW206-lgt
AM1825	TB28 skp::Kanr nadA::Tn10 attλ PrpoH::lacZ

^aUnless otherwise stated, strains were constructed in this study.

TABLE 2.

Plasmids used in this study

Plasmid	Relevant characteristics	Source or reference
pBAD33	p15A Cmr araC Para	63
pDSW206	pTrc99a, promoter attenuated (−10, TATAAT→CATTAT; −35, TTGACA→TTTACA)	64
pTrc99A	pBR/ColE1 Ampr lacI Plac::empty	64
pAM1	pBAD33-yciB	29
pAM7	pTrc99a-lgt	This study
pAM8	pDSW206-lgt	This study

Spot viability assays.

Cultures grown under permissive conditions overnight at 37°C were pelleted, washed, and normalized to an optical density at 600 nm (OD₆₀₀) of 1.0 in LB (1% NaCl or 0% NaCl) prior to spotting on to agar medium. Cell suspensions were serially diluted from 10⁻² to 10⁻⁶, and 4 μl from each dilution was spotted on agar plates with or without salt and grown at 37°C overnight or at 20°C for 60 to 72 h, at which point the plates were imaged (Gel-Doc system; Syngene).

Microscopy.

Cells were grown overnight with 0.02% l-arabinose when depleting yciB or lgt and subcultured 1:150 into LB (1% NaCl). Cells were imaged on 1% agarose pads by phase microscopy using a Nikon Eclipse Ti microscope with a charge-coupled device (CCD) camera as described previously (29). Microscope slides were maintained at 37°C using the TC‐500 temperature controller (20/20 Technology). Image processing was done in NIS‐Elements software (Nikon).

β-Galactosidase activity assays.

Assays were performed essentially as described previously (29). Strains grown overnight under permissive conditions were subcultured 1:150 in LB (1% or 0% NaCl) and grown at 37°C for 2 to 3 h. A 200-μl aliquot of each culture was transferred to a 96‐well flat-bottomed plate and OD₅₉₅ values were read using a plate reader (SpectraMax 190; Molecular Devices). Next, 100-μl aliquots of cultures were transferred to a new 96‐well plate and mixed with 10 μl of PopCulture reagent (EMD Millipore) containing 0.4 unit · μl⁻¹ of rLysozyme (Novagen) and incubated at room temperature for 30 min. After lysis, 30 μl of lysate was transferred to a new 96‐well plate, mixed with 150 μl of Z‐buffer containing o‐nitrophenyl‐β-galactoside (ONPG; 0.8 mg · ml⁻¹) and β-mercaptoethanol (0.2%), and placed in a plate reader maintained at 28°C. The OD₄₁₅ values were read every minute for an hour. Miller units were derived using a modified equation as described previously (29).

Whole-cell lysate preparation and immunoblotting.

For Lpp immunoblotting, cells were prepared as described previously (29). Briefly, strains grown overnight under permissive conditions (0.02% arabinose for PAP9403) were subcultured 1:150 in LB (1% NaCl) and grown at 37°C for 3 h. Approximately 3.2 × 10⁸ cells were collected (10,500 × g; 1.5 min) for each strain. Cell pellets were resuspended in 100 μl Bugbuster (EMD) with benzonase (50 U · ml⁻¹) and incubated at room temperature for 15 min to lyse the cells. Lysates were mixed with 200 μl 2× Tricine-SDS sample buffer (Novex) containing 4% β-mercaptoethanol and boiled in water for 5 min. Samples were resolved on 16% Tricine gels (Novex) with Tricine-SDS sample buffer at 125 V for 2.5 h. Proteins were transferred onto a nitrocellulose membrane (0.2 μm). A primary rabbit anti-Lpp polyclonal antibody was used at 1:500,000. Infrared fluorescence goat anti-rabbit secondary antibody (IRDye 800CW; LI-COR Biosciences) was used at 1:20,000.

Laurdan generalized polarization assay.

Laurdan spectroscopic measurements of bacterial cells were conducted essentially as described previously (49). WT and mutant E. coli cells were grown overnight in LB (1% NaCl) at 37°C and were subcultured to an OD₆₀₀ of 0.03 under the same conditions and grown for 2 h. At this point, cells were collected and washed in prewarmed (37°C) Laurdan buffer (50 mM Na₂HPO₄-NaH₂PO₄ [pH 7.4], 0.1% glucose, and 150 mM NaCl), and resuspended in prewarmed Laurdan buffer with 10 μM Laurdan (Sigma). Samples were incubated in the dark at 37°C with shaking at 250 rpm for 1 h. The dye-labeled cells were washed twice in Laurdan buffer, and 200-μl aliquots of the washed cells were distributed in quadruplicates to a 37°C prewarmed, black, flat and clear-bottomed 96-well microtiter plate (Greiner). As a control, WT cells were treated with the membrane fluidizer benzyl alcohol (30 mM) for 5 min prior to measurement. Generalized polarization (GP) of Laurdan was measured using a SpectraMax M5 fluorimeter maintained at 37°C, with the excitation wavelength set at 350 nm and emission detection at 440 nm and 490 nm. The emission shift was measured using the following formula: GP = (I₄₄₀ − I₄₉₀)/(I₄₄₀ + I₄₉₀).