YaeT (Omp85) affects the assembly of lipid-dependent and lipid-independent outer membrane proteins of Escherichia coli



The Omp85 family of proteins has been found in all Gram-negative bacteria and even several eukaryotic organisms. The previously uncharacterized Escherichia coli member of this family is YaeT. The results of this study, consistent with previous Omp85 studies, show that the yaeT gene encodes for an essential cellular function. Direct examinations of the outer membrane fraction and protein assembly revealed that cells depleted for YaeT are severely defective in the biogenesis of outer membrane proteins (OMPs). Interestingly, assemblies of the two distinct groups of OMPs that follow either SurA- and lipopolysaccharide-dependent (OmpF/C) or -independent (TolC) folding pathways were affected, suggesting that YaeT may act as a general OMP assembly factor. Depletion of cells for YaeT led to the accumulation of OMPs in the fraction enriched for periplasm, thus indicating that YaeT facilitates the insertion of soluble assembly intermediates from the periplasm to the outer membrane. Our data suggest that YaeT's role in the assembly of OMPs is not mediated through a role in lipid biogenesis, as debated for Omp85 in Neisseria, thus advocating a conserved OMP assembly function of Omp85 homologues.


The cell envelope of Escherichia coli consists of the inner (cytoplasmic) membrane and the outer membrane separated by the aqueous periplasmic space. Unlike the symmetric inner membrane phospholipid bilayer, the outer membrane has an asymmetric bilayer consisting of phospholipids in the inner leaflet and lipopolysaccharide (LPS) in the outer leaflet. Both membranes contain proteins synthesized in the cytoplasm. Outer membrane proteins (OMPs), however, must traverse the inner membrane and periplasm prior to insertion in the outer membrane. While the mechanisms by which proteins travel across the inner membrane have become more clearly understood through the elucidation of translocation machinery, it is not well understood how proteins are targeted and inserted into the outer membrane.

The biogenesis of β-barrel OMPs in E. coli has been extensively studied for many years (for a review, see Danese and Silhavy, 1999). A few OMP assembly pathways are well understood such as lipoprotein transport and integration involving the Lol proteins (Tokuda and Matsuyama, 2004) and specific chaperone/usher pathways for the targeting of pilin proteins (Thanassi et al., 1998). Various factors have been identified that aid β-barrel OMP folding and assembly, including Skp, a periplasmic chaperone (Chen and Henning, 1996; Schäfer et al., 1999), SurA, a peptidyl prolyl cis-trans isomerase (Rouvière and Gross, 1996; Rizzitello et al., 2001), and LPS (Ames et al., 1974; Laird et al., 1994). A complete mechanism involving all these factors has not been revealed. Additionally, not all OMPs require these factors for proper folding, suggesting alternative folding pathways for β-barrel OMPs (Werner et al., 2003). A general mechanism by which integral β-barrel proteins are targeted and insert in the outer membrane has remained elusive.

Another protein shown to play a vital role in β-barrel OMP assembly is the periplasmic protease DegP (Strauch et al., 1989). Misfolding or mislocalization of OMPs leads to their protease-assisted removal (Misra et al., 1991). This function of DegP is essential at temperatures above 39°C presumably due to an overabundance of misassembled proteins (Lipinska et al., 1989). DegP has also been shown to be essential at lower temperatures when mutant OMPs are expressed (Misra et al., 2000; CastilloKeller and Misra, 2003). Genetic and biochemical evidence have shown that DegP can also function as a chaperone (Spiess et al., 1999; Rizzitello et al., 2001).

The major pore-forming protein, OmpF, has been shown to require lipid synthesis (Bolla et al., 1988) and the presence of SurA (Kloser et al., 2001) to assemble correctly. TolC, on the other hand, does not need these factors (Werner et al., 2003). This suggests that OmpF and TolC proteins follow two separate folding pathways for assembly. While Skp, SurA and LPS assist folding and/or trimerization of some OMPs but not others, the targeting and insertion of OMPs into the outer membrane may still follow a general mechanism.

A recently discovered OMP in Neisseria meningitidis (Omp85) has been shown to play a role in the insertion of proteins into the outer membrane (Voulhoux et al., 2003). It has been debated whether this is an indirect effect due to an LPS transport role of Omp85 (Genevrois et al., 2003) or if Omp85 plays a direct role in protein insertion in the outer membrane (Voulhoux and Tommassen, 2004). Also, Omp85 is highly conserved among Gram-negative bacteria and homologues exist in eukaryotic organisms (Gentle et al., 2004; Surana et al., 2004). Studies in eukaryotic organisms have found Omp85 homologues play a role in the insertion of β-barrel proteins into the mitochondrial (Kozjak et al., 2003; Paschen et al., 2003) and chloroplast outer membranes (Tu et al., 2004).

The homologue of Omp85 in E. coli is the previously uncharacterized 810 amino acid protein YaeT. We deleted the chromosomal copy of yaeT in a cell containing a plasmid copy of yaeT under an arabinose-controlled promoter of pBAD33. Similar to Omp85, YaeT depletion results in the incorrect assembly of OMPs including TolC, OmpF, OmpC and OmpA. As TolC does not require lipid synthesis for assembly, a defect in its assembly by YaeT depletion cannot be due to an indirect effect on an OMP, Imp (Sampson et al., 1989), which facilitates LPS/phospholipids transport to the outer membrane (Bos et al., 2004). Our data suggest that YaeT is an essential protein that plays a general role in the assembly and/or insertion of OMPs, which follow different folding pathways prior to complete assembly.


yaeT is an essential gene

A newly investigated OMP, Omp85, of N. meningitidis is shown to be essential and required for proper OMP assembly in this organism (Voulhoux et al., 2003). We are interested in determining whether the omp85 homologue in E. coli (yaeT), is essential and whether its product plays a general role in OMP biogenesis, affecting the assembly of both lipid-dependent and lipid-independent OMPs. Attempts to delete the chromosomal copy of the omp85 homologue in E. coli, yaeT, following the lambda red recombinase technique described by Datsenko and Wanner (2000) without a plasmid expressing yaeT, were unsuccessful. Therefore to achieve the desired deletion, yaeT was first cloned into a low-copy-number plasmid (pBAD33; Guzman et al. 1995) from which the arabinose-inducible pBAD promoter controlled its expression. The yaeT deletion was created by replacing all but 40 bases at the 5′ end and 90 bases at the 3′ end of the 2,433-base yaeT open reading frame with a kanamycin resistance (Kmr) cassette.

As yaeT could not be deleted from the chromosome without a plasmid copy of the gene present, yaeT appeared to be essential. This was demonstrated experimentally through growth assays shown in Fig. 1. Initially, growth curves were produced by growing cultures in LB with or without 0.005% arabinose. Under these conditions the growth of the yaeT deletion strain did not significantly deviate from that of the yaeT+ strain after 12 h. To ensure minimal expression of cloned yaeT from the pBAD promoter, 0.4% glucose was added to the cultures. As shown in Fig. 1, when the ΔyaeT::Kmr (pBAD33-yaeT+) strain was grown without arabinose but with 0.4% glucose the culture stopped growing after approximately 8 h. In contrast, the culture grown with arabinose or the culture grown without arabinose or glucose continued to grow as normal. These data indicated that YaeT is essential to the cell.

Figure 1.

Growth assays of ΔyaeT::Kmr (pBAD33-yaeT+) strains. Cultures were grown at 37°C in LB with arabinose (circles), without arabinose (squares), or without arabinose but with glucose (triangles). Growth conditions are described in the text and Experimental procedures. At the 5 h time point cultures were diluted 1:50 to maintain the exponential growth phase.

Genes immediately downstream of yaeT encode proteins involved in OMP (Skp) and LPS (LpxD) biogenesis. The results of Western blot analysis, examining Skp in the yaeT deletion and wild-type strains, showed no difference in Skp protein levels (data not shown). This data along with the ability of the plasmid-encoded yaeT gene to complement the lethality phenotype suggest that the ΔyaeT::Kmr deletion exerts no polar effect on these downstream genes. This is further supported by previous work suggesting skp has its own promoter (Hirvas et al., 1991; Dartigalongue et al., 2001).

YaeT is required for OMP assembly

The demonstration of Omp85's role in Neisserial OMP assembly (Voulhoux et al., 2003) prompted the examination of YaeT's involvement in E. coli OMP assembly. OMP levels and assembly were tested from cultures producing YaeT or depleted for YaeT. Overnight grown cultures were subcultured and repeated dilutions were carried out to maintain exponential growth phase (for details, see Experimental procedures). At the points indicated by arrows in Fig. 2A, samples were removed for western analysis. In the ΔyaeT::Kmr (pBAD33-yaeT+) strain, the depletion of YaeT resulted in a decrease of TolC, OmpF/C and OmpA (Fig. 2B). In contrast to these OMPs, no decrease in the level of an inner membrane protein, AcrA, was observed under YaeT depletion conditions (data not shown). These results show a dramatic effect of YaeT depletion exclusively on the OMPs but not on the inner membrane proteins. Fractionating envelopes through sucrose density gradients further substantiated these results. Envelopes from the yaeT+ strain produced outer and inner membrane peaks in fractions with buoyant densities of 1.208 g ml−1 and 1.132 g ml−1 respectively. On the other hand, the buoyant density of the outer membrane from the ΔyaeT::Kmr (pBAD33-yaeT+) strain depleted for YaeT was only 1.184 g ml−1, whereas the inner membrane density was 1.132 g ml−1. These results showed that the depletion of YaeT affected the biophysical properties and protein levels of only the outer membrane and not the inner membrane. The reduction of both TolC and OmpF/C protein levels is the first evidence of an assembly factor affecting proteins that follow distinctly different folding and assembly pathways (Werner et al., 2003).

Figure 2.

Examination of TolC, OmpF/C and OmpA levels by Western blot analysis. Cultures were either expressing YaeT (yaeT +) or depleted for YaeT (ΔyaeT::Kmr pBAD33-yaeT+).
A. The graph displays growth curves of YaeT+ (diamond) and ΔyaeT::Kmr (pBAD33-yaeT+) (squares) strains. Arrows with numbers show points at which aliquots of cells were removed to prepare whole cell extracts for Western blot analysis.
B. Lanes 1–4 carries samples from cells grown either in the presence of arabinose (lane 1) or without arabinose but with glucose to achieve YaeT depletion (lanes 2–4). Samples were removed at the specified points designated in the graph. Protein samples were analysed in two gels, one of which was probed for TolC (top) and the other gel with OmpF/C and OmpA (bottom) antibodies. Positions of TolC, OmpF/C and OmpA are shown.

To demonstrate whether YaeT functions as an assembly factor for OMPs, pulse-chase experiments were carried out. An assay has been established that demonstrates TolC assembly as the emergence of a 46 kDa proteinase K-resistant and membrane-bound TolC fragment over time (Werner et al., 2003). This assay was performed with wild-type and ΔyaeT::Kmr (pBAD33-yaeT +) strains. Figure 3A shows the results of TolC assembly under YaeT depletion conditions. The wild-type strain displayed normal TolC assembly in which 70% of the total labelled TolC produced a 46 kDa proteinase K-resistant fragment after 30 min of chase (Fig. 3B). In contrast, TolC assembly in the ΔyaeT::Kmr strain was severely impaired, as after 30 min of chase only 25% of the total TolC protein attained the proteinase K-resistant conformation (Fig. 3B), reflecting that the majority of newly synthesized TolC is not assembling and inserting correctly in the outer membrane. The persistence of unassembled (proteinase K-sensitive) TolC under YaeT depletion conditions reflects an unusual stability of TolC assembly intermediates that was also reported for wild type and a mutant TolC protein in a YaeT+ background (Werner et al., 2003; also see Fig. 4C). It is important to note that the radioactive gel carrying samples from the ΔyaeT::Kmr strain were exposed to the phosphorimager screen for almost twice as long as the gel carrying samples from the yaeT + strain. Thus, taking this into account, the total pool of radioactive TolC available for assembly was also significantly smaller in ΔyaeT::Kmr cells than in yaeT+ cells.

Figure 3.

Effect of YaeT depletion on TolC assemblyA. Cultures were grown approximately 4 h without arabinose and in the presence of glucose. Cells were labelled for 30 s and chased for 1, 10 and 30 min. Samples withdrawn at chase periods were lysed and subjected to either proteinase K treatment (+) or no treatment (–) followed by immunoprecipitation using TolC antibodies. Immuno-complexes were analysed on an SDS-PAGE gel (Werner et al., 2003). The appearance of a faster migrating TolC band (TolC*) in proteinase K-treated samples reflects complete TolC assemblyB. The graph shows TolC assembly efficiencies, which were determined by taking the ratios of the full-length TolC bands from proteinase K-untreated and truncated TolC bands from proteinase K-treated samples. Filled and open squares represent TolC assembly efficiencies from yaeT + and ΔyaeT::Kmr strains respectively.

Figure 4.

Effect of DegP on YaeT depletion and examination of osmotically shocked fractionsA. Growth curves of cultures grown at 30°C in the absence of arabinose are shown. After 3 h of incubation without arabinose, cultures were diluted 1:50 in LB without arabinose but supplemented with glucose and grown for additional 3 h. The strains used were YaeT+ DegP+ (squares), YaeT+ DegP (triangles), YaeT DegP+ (circles), or YaeT DegP (diamonds)B. For protein analysis, YaeT depletion conditions were not as severe as in (A). Here, cultures were grown for only 4 h in the absence of arabinose and in the presence of glucose. Cells were collected after growth in arabinose (indicated by an open arrow) or in the absence of arabinose (indicated by the filled arrow).
C and D. Western analyses of whole cell and osmotic-shocked extracts obtained from cells grown under conditions either favouring YaeT expression (C) or its depletion (D). Osmotic-shocked extracts obtained from YaeT+ DegP and YaeT-depleted DegP cells were also treated with proteinase K (PK) prior to western analysis. Positions of TolC, OmpF/C and OmpA are shown.

Is DegP responsible for the degradation of unassembled OMPs in YaeT-depleted cells?

As shown in Figs 2 and 3, YaeT depletion affects OMP biogenesis leading to reduction in OmpF/C, OmpA and TolC levels. We asked whether this reduction is the result of the DegP-mediated degradation of unassembled OMPs that may briefly accumulate when cells are depleted for YaeT. To test this, a degP::Tn10 allele was introduced by P1 transduction into the ΔyaeT::Kmr strain holding a yaeT+ plasmid. Growth experiments were carried out at 30°C due to the temperature sensitivity of degP strains. The absence of DegP in a yaeT+ background had no ill effect on bacterial growth, but under YaeT depletion conditions the absence of DegP accelerated the cessation of bacterial growth compared with when DegP was normally expressed (Fig. 4A). These results indicated that DegP's proteolytic or chaperone activity is important under YaeT depletion conditions.

We examined OMPs in a DegP minus background depleted for YaeT. Due to a concern that dying DegP-minus YaeT-depleted cells (Fig. 4A) may contribute to fractionation artifacts, we harvested cells after only 4 h of YaeT depletion when cells showed little growth inhibition (Fig. 4B). Western analysis was performed on whole cell extracts and osmotically shocked fractions rich in periplasmic proteins. Figure 4C shows western analysis results from cultures grown with arabinose, thereby maintaining YaeT expression; Fig. 4D displays the results of YaeT depletion after the removal of arabinose. As expected, the depletion of YaeT in a DegP+ background reduced the OmpF/C, OmpA and TolC levels in whole cell extracts (Fig. 4D). This reduction in OMP levels was less prominent when DegP was absent, suggesting that DegP is partly responsible for degrading misfolded OMPs that accumulate when YaeT is depleted.

Analysis of fractions enriched for periplasmic proteins provided some interesting clues as to the nature of the OMP assembly defect in ΔyaeT::Kmr and ΔyaeT::KmrdegP::Tn10 strains. We noted that even under a relatively mild isolation condition, OmpF/C and to some degree OmpA were still detectable in the osmotically shocked fraction of YaeT+ strains (Fig 4C and D). These soluble OMPs were proteinase K resistant (data not shown; see Fig. 4D), indicating that they represent properly assembled population. However, despite this apparent contamination of major OMPs, no TolC could be detected from the periplasm-enriched fraction when YaeT was normally expressed. It is conceivable that a relative abundance of OmpF/C made it easier to detect them in the soluble fraction than those proteins that are less abundant such as TolC.

The examination of soluble fractions from YaeT-depleted cells produced strikingly different results than those from YaeT+ cells (Fig. 4D). First of all, when DegP+ cells were depleted for YaeT, TolC could now be detected in the periplasm-enriched fraction despite the fact that there was an overall reduction in its level when whole cell extracts were analysed. In contrast, OmpF/C and OmpA, which were readily detectable in the soluble fraction of YaeT+ cells, could no longer be detected from the soluble fraction of DegP+ YaeT-depleted cells. When DegP was absent, the levels of soluble TolC and OmpA increased significantly over OmpF/C, which were barely detectable. The elevated populations of soluble TolC and OmpA present in the absence of DegP and under YaeT depletion conditions probably represented DegP protease-sensitive unassembled species. To test this, soluble fractions from YaeT+ DegP and YaeT-depleted DegP cells were treated with proteinase K. Soluble OmpF/C and OmpA from YaeT+ DegP cells were proteinase K resistant. On the other hand, soluble TolC and OmpA from YaeT-depleted DegP cells were proteolysed by proteinase K, suggesting that the degraded population consisted of improperly assembled molecules. These results showed that OMP assembly was compromised when YaeT was depleted from the cell. As a result, a fraction of improperly assembled OMPs accumulated in the periplasm-enriched fraction and became susceptible to DegP-mediated proteolysis. It is important to note that the transient accumulation of a proteinase K-sensitive soluble assembly intermediate is a normal feature of wild-type TolC assembly (Werner et al., 2003). The persistence of a proteinase K-sensitive soluble species was previously observed for an assembly defective TolC protein (Werner et al., 2003).

Is the effect of YaeT depletion on OMP assembly due to a defect in lipid biogenesis?

It has been argued that Omp85 in Neisseria influences OMP assembly by affecting LPS assembly or insertion in the outer membrane (Genevrois et al., 2003). However, this assertion was rebutted in a subsequent report (Voulhoux and Tommassen, 2004). If YaeT plays a role in lipid biogenesis in E. coli then an inhibitor, cerulenin, which blocks lipid synthesis should inhibit the assembly of all OMPs that require YaeT. Indeed OmpF assembly is severely inhibited by cerulenin (Bolla et al., 1988) and in deep-rough strains producing truncated LPS molecules (Laird et al., 1994). Unlike OmpF, TolC assembly is unaffected in deep-rough LPS mutants (Werner et al., 2003), suggesting that a defect in LPS biogenesis may not be the reason for the TolC assembly defect observed under YaeT depletion conditions. To further examine a possible requirement of de novo lipid synthesis in TolC biogenesis, we tested the effect of cerulenin on TolC assembly. OmpF assembly was severely inhibited when lipid synthesis was blocked by the addition of cerulenin (Fig. 5A). The absence of OmpF molecules is likely due to the rapid degradation of the unassembled population, as has been previously reported (Bolla et al., 1988). Unlike the dramatic influence on OmpF, virtually no effect of cerulenin on total TolC protein levels (Fig. 5B) or TolC assembly was apparent (Fig. 5C). Therefore, the TolC and OmpF assembly defects observed under YaeT depletion conditions cannot be attributed to a common defect involving lipid synthesis. Examination of LPS from YaeT+ and YaeT-depleted cells by SDS-PAGE followed by silver staining further supported this conclusion, because no differences in LPS levels were apparent in YaeT+ and YaeT-depleted cells (data not shown). Also, there was no difference in the localization of LPS when inner and outer membrane peak fractions from the sucrose density gradients were analysed from YaeT+ and YaeT-depleted cells (data not shown). Taken together, these results strongly suggest that OMP biogenesis defects observed under YaeT depletion conditions are not due to an effect on LPS biogenesis or localization, but possibly due to the disruption of OMP biogenesis machinery.

Figure 5.

Effect of cerulenin on OmpF and TolC biogenesis. Cells were grown and radioactively labelled as described in the text and Experimental procedures. (A) OmpF was directly immunoprecipitated from cell extracts and analysed by SDS-PAGE. For TolC analysis, labelled cell extracts were either treated with proteinase K (C) or untreated (B) prior to immunoprecipitation and SDS-PAGE analyses. The appearance of a proteinase K-resistant TolC fragment (TolC*) demonstrates proper assembly of TolC. Positions of OmpF, TolC and TolC* are shown.


Although the process of OMP biogenesis has been studied for several decades, the mechanisms involved in OMP assembly and insertion into the outer membrane remain poorly understood. The recent discoveries of two OMPs from Neisseria and E. coli, which are shown to play roles in OMP assembly (Omp85; Voulhoux et al. 2003) and LPS transport (Imp; Bos et al. 2004), have provided a much-needed boost in narrowing the gap of our understanding of outer membrane biogenesis in Gram-negative bacteria. Here we show that YaeT, a homologue of Neisserial Omp85, is an essential E. coli protein that is required for the biogenesis of two distinct groups of OMPs that follow lipid-dependent (OmpF/C) and lipid-independent (TolC) folding pathways. Because YaeT itself is believed to be an OMP, it is likely to be involved in assisting other OMPs to insert and assemble in the outer membrane.

Disruption of the chromosomal yaeT gene was achieved only when a copy was simultaneously expressed from a plasmid replicon, showing that the observed lethal phenotype was due to YaeT depletion and not a polar effect on downstream genes that are also involved in OMP (Skp) and LPS (LpxD) biogenesis. This was further confirmed by examining the level of Skp (expressed from a gene located immediately downstream of yaeT), which was found to be unaffected in strains depleted for YaeT (data not shown). During the review of this manuscript, an article published by Wu et al. (2005) also showed that YaeT is essential for cell viability and in its depletion, the assembly of a SurA- and LPS-dependent OMP, LamB, is severely inhibited. These complementary results reiterate the role of YaeT as a general OMP assembly factor.

YaeT depletion drastically altered OMP levels without influencing the inner membrane proteins. As a result, the buoyant density of the outer membrane was lowered to 1.184 g ml−1 (compared with 1.208 g ml−1 of the yaeT+ strain), but the inner membrane buoyant density remained identical to that of the yaeT+ strain. An examination of LPS under YaeT depletion conditions showed that unlike OMPs, the level and localization of LPS were unaffected, suggesting that the observed reduction in the outer membrane buoyant density was the result of an effect of YaeT depletion on OMPs but not on LPS. In contrast to this conclusion, it was reported that the depletion of Omp85 in N. meningitidis affected LPS export to the outer membrane (Genevrois et al., 2003). Furthermore, these authors reported that Omp85 depletion did not affect the localization of the integral OMPs, PorA and Opa, thus contradicting the report by Tommassen's group (Voulhoux et al., 2003) who argued for a direct role of Omp85 in OMP assembly by demonstrating an interaction between Omp85 and non-native forms of PorA. Additionally, several lines of evidence support the notion that YaeT/Omp85 directly affects OMP assembly. First, we show here and have published previously (Werner et al., 2003) that TolC biogenesis is independent of de novo lipid synthesis, because the addition of a potent lipid synthesis inhibitor (cerulenin, this study) or the presence of a truncated LPS core (Werner et al., 2003) failed to compromise TolC assembly, while abolishing the assembly of an LPS-dependent OMP, OmpF (Bolla et al., 1988; Laird et al., 1994). Therefore, the observed effect of YaeT depletion on TolC assembly cannot be explained by an effect on LPS biogenesis. Second, it has been shown that LPS is not essential in N. meningitidis (Steeghs et al., 1998) and OMPs assemble normally in LPS deficient strains (Steeghs et al., 2001). Thus, the OMP assembly defects observed in an Omp85-depleted strain cannot be due to a defect in LPS biogenesis. Third, homologues of Omp85 are present in bacteria such as Treponema pallidum and Borrelia burgdorferi that lack LPS altogether, thus disfavouring the possibility of a primary role for Omp85 in LPS biogenesis.

The above contradictions can be reconciled by recent reports that have identified another essential OMP, Imp, implicated in LPS/lipid transport to the cell surface (Braun and Silhavy, 2002; Bos et al., 2004). Accordingly, depletion of YaeT/Omp85 in cells would produce a general defect in OMP biogenesis, including that of Imp. A defect in Imp biogenesis in turn would reduce LPS/lipid transport to the outer membrane and thus the assembly of LPS/lipid-dependent OMPs such as OmpF but not that of TolC. The fact that TolC assembly is also defective in the YaeT depleted strain, combined with the observations that the LPS-deficient strain of N. meningitidis shows normal OMP assembly while the Omp85-deficient strain shows an OMP assembly defect, argues strongly in favour of the notion that YaeT/Omp85 are directly involved in OMP assembly. Thus, any effect of YaeT depletion on LPS biogenesis is an indirect consequence of defective Imp biogenesis. Our data suggest this effect is small, if any, because there was no reduction in LPS levels or its localization upon YaeT depletion, whereas OmpF/C, OmpA and TolC levels were significantly reduced.

Even if we settle the argument of a direct role for YaeT/Omp85 in OMP assembly, the question remains as to how these proteins achieve their functions. As has been reported for Omp85 depletion (Voulhoux et al., 2003), we have shown here that YaeT depletion leads to a transient accumulation of OMP assembly intermediates that are protease sensitive and are localized in the periplasm-enriched soluble fraction. As YaeT/Omp85 themselves are OMPs, it is conceivable that they assist in the insertion of soluble assembly intermediates to the outer membrane. Our observation that YaeT/Omp85 depletion affects the biogenesis of two distinct groups of proteins that follow dissimilar folding pathways would suggest that YaeT/Omp85 acts as a general assembly factor.

Experimental procedures

Bacterial strains, growth media and chemicals

All bacterial strains used in this study were derived from MC4100 [FaraD139Δ(argF-lac)U139 rpsL150 relA1 fibB5301 ptsF25 deoC1 thi-1 rbsR] (Casadaban, 1976). MC4100 Δara714 was constructed in two steps. First, an auxotrophic marker, leu::Tn10, linked to the Δara714 allele from LMG194 (Guzman et al., 1995), was transduced into MC4100 and Tcr transductants were tested for arabinose sensitivity (the presence of araD139 allele in MC4100 confers arabinose sensitivity, whereas Δara714, which deletes araBAD genes, does not). In the second step, P1 lysate prepared on MC4100 (leu+) was used to obtain Leu+ transductants from MC4100 leu::Tn10Δara714. Leu+ transductants that simultaneously displayed arabinose insensitivity were purified and used in this study as an MC4100 Δara strain (RAM1292). The E. coli strains and plasmids used in this study are listed in Table 1. Minimal medium (M63) and Luria–Bertani (LB) broth were prepared as previously described (Silhavy et al., 1984). When cultures were grown on minimal medium, 0.2% glucose was used as the sole carbon source. A mixture of [35S]-methionine [35S]-cysteine was purchased from Perkin Elmer Life Sciences. ECF substrate was purchased from Amersham Pharmacia Biotechnologies. Cerulenin was purchased from Sigma-Aldrich. Formalin-killed Staphylococcus aureus cells (Pansorbin) were purchased from Calbiochem. All other chemicals were of analytical grade.

Table 1. Bacterial strains and plasmids used in this study.
StrainsCharacteristicsSource or references
 RAM1134BW25113/pKD46 Datsenko and Wanner (2000)
 RAM1291RAM1134/pBAD33-yaeT+This study
 RAM1292MC4100 Δara714This study
 RAM1293MC4100 Δara714/pBAD33-yaeT+This study
 RAM1294RAM1293 ΔyaeT::KmrThis study
 RAM1295RAM1293/pTrc-tolC+This study
 RAM1296RAM1294/pTrc-tolC+This study
 RAM1297RAM1293 degP::Tn10This study
 RAM1298RAM1294 degP::Tn10This study
 RAM1299MC4100 ΔtolC::Cmr/pTrc-tolC+This study
 RAM1300RAM1292 ompF::lacZ+/pTrc-ompF+This study
 pTrc99AApr; expression vectorPharmacia
 pTrc-tolC+Apr Werner et al. (2003)
 pTrc-ompF+Apr Traurig and Misra (1999)
 pBAD18Kmr; expression vector Guzman et al. (1995)
 pBAD18-yaeT+KmrThis study
 pBAD33Cmr; expression vector Guzman et al. (1995)
 pBAD33-yaeT+CmrThis study
 pKD46Apr; Lambda-red recombinase Datsenko and Wanner (2000)
 pKD4Kmr Datsenko and Wanner (2000)

Plasmid construction

To clone yaeT into pBAD33 primers were designed for polymerase chain reaction (PCR) amplification. The forward primer (5′-GGAAGAATTCATAATAACGATGGCG-3′) creates an EcoRI site 9 bases upstream of the first ATG start codon while the reverse primer (5′-CCTTTGTCTAGAACACTTACC AGG-3′) creates a unique XbaI site downstream from the stop codon. Due to the presence of two EcoRI cut-sites in pBAD33, yaeT was first cloned into pBAD18. yaeT DNA was amplified from the chromosome by PCR, digested with EcoRI and XbaI and ligated into the appropriately digested pBAD18 vector. The pBAD18-yaeT + clone was digested with BamHI and XbaI and ligated into the appropriately digested pBAD33 vector. The pTrc clone of tolC was constructed as described by Werner et al. (2003). The pTrc-ompF clone was constructed as described previously (Traurig and Misra, 1999).

yaeT deletion

The deletion of yaeT was performed via the lambda-red recombination technique described by Datsenko and Wanner (2000). The pBAD33 clone of yaeT was transformed into RAM1134 containing the lambda-red recombinase plasmid (pKD46). Primers were designed to delete approximately 2100 base pairs from the internal portion of the yaeT open reading frame. The correct deletion would leave 40 bases at the 5′ end of the yaeT open reading frame and 90 bases at the 3′ end. PCR amplification was performed using the forward primer (5′-GCGATGAAAAAGTTGCTCATA GCGTCGCTGCTGTTTATGAAGGCTGGAGCTGCTTCG-3′), the reverse primer (5′-CTTTTTGAACGGCTGGGCGTAGGA GAACACCAACGGCATATGAATATCCTCCTTAG-3′) and the pKD4 plasmid that contains the Kmr marker. Purified PCR product was electroporated into electrocompetent cells (RAM1291) at 2.5 kV using the Bio-Rad E. coli Pulser. Cultures were incubated in a 30°C water bath overnight with gentle shaking and plated on LBA containing ampicillin (50 µg ml−1), chloramphenicol (10 µg ml−1), kanamycin (12.5 µg ml−1) and l-arabinose (0.01%). PCR amplification was performed on purified colonies to verify if the deletion was successful. The forward yaeT cloning primer and a reverse skp primer were used to determine if the Kmr marker was inserted in the chromosomal yaeT gene. The production of a correct size PCR-amplified DNA fragment confirmed that the majority of the yaeT gene, as intended, was deleted from the chromosome.

YaeT depletion

Overnight cultures were grown at 37°C in LB containing chloramphenicol (10 µg ml−1) and arabinose (0.01%) to allow for basal levels of YaeT from the plasmid during growth. Next day, cultures were subcultured (1:100 dilution) in LB containing arabinose and chloramphenicol. For experiments performed in Fig. 2, cultures after the initial 1:100 dilution were grown to an OD600 of 0.5 at which time 1:17 dilutions were performed in LB containing chloramphenicol, glucose (0.4%) and no arabinose. Growth and dilutions were performed in this media until the ΔyaeT::Kmr (pBAD33-yaeT+) strain could no longer reach an OD600 of 0.5. When the ΔyaeT::Kmr (pBAD33-yaeT+) strain stopped growing, cells were concentrated to attain a final OD600 of 0.5 and used as the last sample.

For TolC assembly assays (Fig. 3), overnight cultures were grown and subcultured as described above for YaeT depletion experiments in Fig. 2. At an OD600 of 0.3, cultures were diluted 1:50 in LB containing chloramphenicol and glucose but lacking arabinose. This dilution was repeated once and when the cultures reached an OD600 of 0.3 (approximately 4 h after the removal of arabinose) aliquots were removed and pulse-labelled for the TolC assembly assay. ΔyaeT::Kmr (pBAD33-yaeT+) cultures were still able to grow a short time after aliquots were removed which ensured living cultures were being assayed.

Cultures for experiments involving the lack of DegP were grown at 30°C due to temperature sensitive phenotypes of DegP strains. Overnight cultures were grown and subcultured as described above. In Fig. 4A, the subcultures were grown to an OD600 of 0.5 and diluted 1:50 into LB containing chloramphenicol and glucose but no arabinose. A second dilution was performed when cultures reached an OD600 of 0.5 (approximately 3 h). For protein analysis in Fig 4C and D, bacterial cultures were grown to an OD600 of 0.3 (approximately 2 h) in the presence of arabinose. Dilutions (1:17) were performed in LB containing chloramphenicol and glucose but no arabinose. This dilution was repeated once again after the cultures reached an OD600 of 0.3 and cultures were grown approximately 2 h. At this point, culture densities were adjusted to an OD600 of 0.3 and aliquots were used for Western analyses as shown in Fig 4C and D.

For the sucrose gradients, cultures were grown and diluted as described for Fig. 2 except 1:25 dilutions were performed. The buoyant membrane densities described were from samples taken at the last dilution (approximately 4 h without arabinose).

Membrane analysis

Whole cell envelopes were prepared by French press cell lysis as previously described (Misra, 1993). Separation of the whole cell envelopes into inner and outer membrane fractions was achieved through sucrose density gradients (30–55%) as described by Misra et al. (2000). After centrifugation, fractions were collected and densities were calculated by measuring the refractive index of each fraction at 25°C. Samples from each fraction were analysed by SDS-PAGE.

Western blot analysis

Protein samples were prepared by solubilizing cells in 2% SDS and heating at 100°C for 5 min. A gentle osmotic shock method of Ariéet al. (2001) was used to obtain samples enriched in the periplasmic proteins. These samples were analysed on mini SDS-polyacrylamide gels and transferred to polyvinylidene difluoride (PVDF) membranes (Immobilon-Millipore). After transfer, the membranes were incubated for 1.5 h with primary antibodies raised against appropriate antigens. The membranes were washed and goat anti-rabbit alkaline phosphatase conjugated IgG secondary antibodies were added for 1 h. The membranes were exposed to ECF substrate for 5 min and analysed using a phosphorimager.

Radioactive labelling and assembly assays

Cells were transformed with pTrc-tolC to allow better detection of TolC. Cultures were grown in LB broth containing ampicillin (50 µg ml−1), chloramphenicol (10 µg ml−1) and glucose (0.4%) to an OD600 of 0.3. Cells were pelleted by centrifugation at 3000 r.p.m. in GS-6 centrifuge (Beckman) with GH 3.7 rotor and resuspended in M63 salts. The cells were pelleted again and resuspended in minimal medium containing 0.2% glucose as the carbon source. Ten minutes prior to labelling, 0.4 mM isopropyl-β- d-thiogalactopyranoside (IPTG) was added to the culture to induce TolC synthesis. Cells were labelled as described previously (Werner et al., 2003). To examine the effect of cerulenin, RAM1299 and RAM1300 cells were grown in minimal media containing glycerol as the carbon source to an OD600 of 0.3. Either cerulenin (100 µg ml−1) or methanol (solvent) was added to the cultures 30 min prior to labelling and IPTG (0.4 mM) was added 10 min prior to labelling to induce TolC or OmpF.

TolC assembly was assayed in cells with or without proteinase K as described previously (Werner et al., 2003). To analyse OmpF, membrane proteins were extracted from the cells by the gentle detergent lysis procedure (Misra, 1993) and boiled. Labelled extracts were mixed with immunoprecipitation buffer (1% Triton X-100, 50 mM Tris-HCl, pH 7.5) containing appropriate antibodies and incubated rocking at 4°C for a minimum of 4 h. Immunocomplexes were precipitated with Pansorbin, washed several times and analysed by SDS-PAGE. The gels were dried and exposed to phosphor storage screens (Molecular Dynamics and Amersham Pharmacia Biotechnologies). The screens were developed using a phosphorimager.


This work was supported by a grant from the National Institutes of Health (R01-GM048167). We are thankful to Leanne Misra for critically reading the manuscript.