Periplasmic chaperones apparently deliver many if not all proteins to the outer membrane, suggesting they are assembled like pili 


P. A. DiGiuseppe and T. J. Silhavy


P. A. DiGiuseppe is a graduate student and T. J. Silhavy is the Warner-Lambert Parke-Davis Professor in the Department of Molecular Biology, Princeton University, Princeton, N.J. 

Author profile--T. J. Silhavy

Download a PDF of this article


The outer membrane (OM) serves a protective role, allowing gram-negative bacteria such as Escherichia coli to survive in a variety of harsh environments. Assembling complex proteins within the gram-negative cell envelope, which includes all noncytoplasmic structures and compartments, presents a considerable challenge. The outer membrane of such bacteria is an asymmetric lipid bilayer containing three major classes of proteins--namely, lipoproteins, [beta]-barrel proteins, and multicomponent surface structures such as pili. The surface-exposed leaflet of this bilayer consists of lipopolysaccharide (LPS), the characteristic glycolipid of gram-negative bacteria, while the inner leaflet is composed of simpler phospholipids.

             Outer membrane biogenesis is a fascinating problem. The asymmetry of this bilayer must be maintained during growth, and all protein components must be folded and inserted into this membrane from a cellular environment that lacks small energy-releasing molecules such as adenosine triphosphate (ATP) that are used to drive enzyme-catalyzed reactions. Direct genetic approaches to identifying the cellular components involved in these processes have not proved all that successful.

             Accordingly, we and others are developing alternative strategies for studying OM biogenesis, and these approaches are uncovering signal transduction pathways, such as Cpx and [sigma]E. These strategies involve selecting suppressors that overcome the toxic effects of mistargeted, misfolded proteins. In analogy to the cytoplasmic heat shock response, these signal transduction pathways control expression of genes whose products function in envelope protein folding and targeting. Although much remains to be learned about these pathways, our findings already suggest that all outer membrane proteins follow an assembly paradigm established for pili, with periplasmic chaperones delivering newly synthesized outer membrane proteins to specific assembly sites.


General Configuration of the Gram-Negative Cell Envelope 


Lipoproteins are generally not true integral membrane proteins, and have lipid covalently attached at the amino-terminal cysteine that anchors these molecules in the inner leaflet of the outer membrane. The [beta]-barrel proteins are composed of [beta] sheets that are wrapped into cylinders. Because of this structure, many of these molecules function as porins that passively transport small molecules across the outer membrane permeability barrier. Some, like OmpF, act as molecular sieves, while others, like LamB, exhibit substrate specificity, in this case for maltose poly mers called maltodextrins. Since all [beta]-barrel proteins are surface exposed, they may be used as phagereceptors, such as OmpF for K20 and LamB for [lamda].

             Pili serve to attach bacteria to surfaces, and each of the many different types has specific affinity for certain surfaces. We are concerned here with pili that are assembled in the outer membrane. An example is the P pilus of uropathogenic Escherichia coli.

             The outer and inner membranes of gram-negative bacteria delimit a zone of controlled permeability known as the periplasm. This cellular compartment contains proteins such as the periplasmic-binding proteins that are involved in the active transport of various small molecules, degradative enzymes, and proteins that function in biogenesis. 


Genetic Analysis of Protein Translocation to the Outer Membrane 


A protein destined for the outer membrane or the periplasmic space is made in the cytoplasm in a precursor form with a signal sequence at its N terminus. By using protein fusions and genetic techniques, researchers identified the signal sequence and components of the secretion machinery.

             For example, LamB is a secreted protein, and its expression is induced by maltose. LacZ is a cytoplasmic enzyme that is required for bacterial cells to use lactose as a carbon and energy source. Fusing the signal sequence of LamB to LacZ results in two interesting phenotypes: the presence of this hybrid protein conferred maltose sensitivity and the uninduced level of Lac activity was abnormally low (“Lac down” phenotype). Suppressors of inducer sensitivity that remain Lac+ alter the LamB signal sequence. Absent a functioning targeting signal, the fusion protein remains in the cytoplasm. Conditional-lethal suppressors with increased Lac activity map to genes that specify components of the secretion machinery. Under noninducing conditions, the fusion protein is directed from the cytoplasm and functions poorly (Lac down). Mutations that compromise the secretion machinery, such as a temperature-sensitive secA mutation at the permissive condition, allow a fraction of the fusion protein to remain in the cytoplasm, where it is active (Lac up). Under nonpermissive conditions, however, the function of the mutant protein is abolished and the cells die, indicating that the secretion machinery is essential.


             Adding maltose induces the fusion protein, jamming secretion machinery and killing the cell. Specifically, the LacZ component of the hybrid protein folds faster than the protein can be translocated, and thus jams the translocon (Fig. 1). The LamB-LacZX90 mutant protein cannot fold, and copies of this protein are translocated efficiently to the periplasm where they form disulfide-bonded aggregates. Strains producing this misfolded, mistargeted fusion protein also exhibit inducer sensitivity (Fig. 1).

             These studies suggest that proteins with signal sequences are translocated into the periplasm posttranslationally in an unfolded state. Yet, how are these proteins folded and targeted in the periplasm, if ATP is not available outside the cytoplasm?


Envelope Stress Responses 


Fusions provide a way to identify mutations that internalize ordinarily exported proteins within the cytoplasm. However, the sec mutations affect targeting to both the periplasm and the outer membrane. Finding a method to identify mutations specific to only the OM-targeting pathway proved difficult. In the absence of a direct selection for specific OM-targeting factors, we employed an indirect method, based on the heat shock pathway (HSP).

             The HSP helps to maintain proper cytoplasmic protein physiology during times of stress. The HSP consists of chaperones that refold misfolded proteins and proteases that degrade proteins if refolding fails. However, some members of this pathway are also used for protein folding under nonstress conditions. We reasoned that generating stress in the envelope might identify factors involved in protein folding in the periplasm. One such strategy involved using fusions.

             The LamB-LacZX90 fusion protein confers maltose sensitivity by forming protein aggregates in the periplasm. We selected for maltose resistance and thus identified a major class of suppressors as gain-of-function mutations that alter the CpxA (conjugative plasmid expression) sensor kinase. The cpxA gene is in an operon with its cognate response regulator, cpxR. Initially, because the cpxA mutations have pleitropic effects, the role of CpxA was unclear.

             One clue as to the function of CpxA came from the observation that the periplasmic toxicity of LacZ was relieved by degradation by the DegP protease. Indeed, the gain-of-function mutations in cpxA, known as cpxA*, upregulate degP transcription. A second clue came from multicopy suppressor studies designed to elucidate how LacZ causes periplasmic toxicity. Through this screen NlpE, a lipoprotein associated with the outer membrane, was identified. Overproduction of NlpE upregulates degP in a CpxRA-dependent manner, but unlike the gain-of-function cpxA mutations, overproduction of NlpE does not cause pleiotropic phenotypes. This observation suggests that degP is a bona fide member of the Cpx regulon. 

             Meanwhile, Carol Gross and her colleagues at the University of California, San Francisco, were focusing on an alternate sigma factor, [sigma]E. They found that [sigma]E also regulates degP transcription. They also found that overproducing outer membrane proteins, such as OmpC, results in a [sigma]E -dependent increase in degP transcription. Conversely, null mutations in ompR, the response regulator that regulates the transcription of the genes for the OmpF and OmpC porins, decreases [sigma]E activity.

             These observations suggested to us a connection between [sigma]E and the outer membrane [beta]-barrel proteins. Perhaps Cpx and [sigma]E sense and respond to misfolded proteins in the envelope, much like the HSP system responds to misfolded proteins in the cytoplasm. Genes regulated by these systems therefore might encode factors involved in folding and targeting proteins in the envelope following translocation.


Exploring whether Cpx and [sigma]E Respond to Misfolded Proteins 




To test whether Cpx and [sigma]E sense misfolded proteins, we collaborated with Scott Hultgren at Washington University in St. Louis, Mo., who studies P pili. These OM structures are composed of a variety of subunits, including the major pilin subunit, PapA, which polymerizes to form a helical rod structure that is capped by the PapE containing tip fibrillum and the PapG adhesin. Pilin subunits are made as precursors in the cytoplasm and are translocated to the periplasm by the Sec machinery. Once in the periplasm, the subunits interact with the PapD chaperone, which prevents subunit misfolding and escorts them to the PapC usher in the outer membrane, where the pilus is assembled from its base (Fig. 2). In the absence of PapD, pilin subunits misfold and aggregate in the periplasm, much like periplasmic LacZ. Indeed, expression of the PapG subunit in the absence of the PapD chaperone induces degP through both the Cpx and [sigma]E pathways. Both these pathways are involved in responding to misfolded envelope proteins by upregulating periplasmic protein folding and trafficking factors. At present there appear to be three sets of target genes--namely, those controlled solely by Cpx, those controlled by the [sigma]E pathway, and those coregulated by both pathways (Fig. 3). The Cpx pathway may control upwards of 100 genes, while the [sigma]E pathway may control approximately 50-60 target genes, and the two pathways coregulate several genes, including degP and the skp chaperone.


The CpxRA Signal Transduction System 


CpxA is similar to other inner-membrane sensor kinases, with two membrane-spanning domains, a periplasmic domain, and a conserved cytoplasmic domain that contains a conserved histidine residue where autophosphorylation occurs. Using purified CpxR and cpxA* alleles that map to the cytoplasmic domain of CpxA in in vitro phosphorylation assays, we determined that CpxA has autokinase activity, CpxR kinase activity, and CpxR phosphatase activity. In vivo, phosphorylation of the response regulator CpxR is accompanied by increased transcription of cpxRA, revealing that the Cpx pathway is positively autoregulated.

 The periplasmic domain of CpxA is required for sensing all Cpx-associated envelope stresses, and CpxA apparently senses a single periplasmic signal, most likely misfolded envelope proteins. Regulon member cpxP was identified in a screen of a collection of random lacZ transcriptional fusions. cpxP encodes a small periplasmic protein that is divergently transcribed from the cpxRA operon. Removing CpxP induces the Cpx pathway approximately twofold, while overproducing CpxP represses the regulon, suggesting that CpxP functions as a negative regulator of the Cpx system.

 CpxP may shut down the Cpx pathway under nonstress conditions by interacting with the sensing domain of CpxA. If correct, then spheroplast formation should induce the Cpx pathway simply by releasing CpxP from the periplasm. If CpxP is truly a negative regulator of the Cpx pathway, tethering CpxP to the inner membrane and thereby preventing its release should block induction of this system by spheroplast formation.

 To test this possibility, we constructed a functional MBP-CpxP fusion protein and then introduced a mutation in the signal sequence that prevents its cleavage by signal peptidase, thus producing a version of CpxP that is tethered to the inner membrane. This tethered MBP-CpxP effectively prevents induction upon spheroplasting. In contrast, untethered MBP-CpxP is lost when spheroplasts form, allowing induction. These experiments demonstrate that CpxP negatively regulates the Cpx pathway.




 According to our current model (Fig. 4), in the uninduced state CpxP is bound to the periplasmic domain of CpxA, decreasing the ratio of CpxA kinase to phosphatase activity under nonstressed conditions and keeping CpxR in an unphosphorylated state. When envelope stress induces CpxP, it is titrated from CpxA, perhaps by misfolded proteins, increasing the ratio of CpxA kinase to phosphatase activity and increasing phosphorylated CpxR (CpxR~P). CpxR~P upregulates transcription of target genes, including cpxRA, leading to signal amplification and expression of genes for periplasmic protein folding and trafficking factors to clear the stress, and also cpxP, which eventually leads to feedback inhibition and shutdown of the Cpx pathway. 


The [sigma]E Stress-Responsive Pathway 


The [sigma]E pathway is a stress-response system that is induced not only by misfolded PapG, but also by other factors, including ethanol and increased temperatures. Like the Cpx pathway, [sigma]E is both positively autoregulated and feedback inhibited. This system is also subject to regulatory factors. For example, RseA is an inner-membrane protein that acts as an anti-sigma factor by complexing with and sequestering [sigma]E from core RNA polymerase. RseB is a periplasmic protein that negatively regulates the [sigma]E pathway by stabilizing the RseA-[sigma]E interaction. 

 The current model for [sigma]E signal transduction suggests that, under nonstress conditions, RseB, RseA, and [sigma]E form a complex within the inner membrane, keeping [sigma]E from interacting with core RNA polymerase and thus shutting down this pathway. RseB binds aggregated periplasmic proteins and is thought to function as a sensor for the [sigma]E pathway. RseA is sequentially cleaved by the inner membrane-anchored DegS protease and then the inner membrane-embedded YaeL protease, releasing [sigma]E and activating the pathway.

 The Cpx and the [sigma]E pathways share a number of features, including positive autoregulation and feedback inhibition. Removing CpxP or RseB does not fully derepress the Cpx or [sigma]E pathways, but induces them approximately twofold. In the absence of CpxP or RseB, both pathways show wild-type induction in response to envelope stress. Like RseB, it is possible that CpxP binds misfolded envelope proteins and functions as a chaperone.


Why Are There Two Envelope Stress Response Pathways? 


Given that the Cpx and the [sigma]E pathways overlap at both the levels of input (activating signals) and output (target gene induction), and that the two pathways share several regulatory features, why are there two envelope stress pathways? While both pathways sense and respond to misfolded envelope proteins, each pathway has distinct primary functions. The Cpx pathway apparently monitors the assembly of cell appendages such as pili, while the [sigma]E pathway seems to monitor the assembly of [beta]-barrel proteins in the outer membrane. 

 Pyelonephritic strains of E.coli have a large number of long P pili extending from their cell surfaces that help cells adhere to human kidney cells. Electron micrographs reveal that cpxRA null mutant derivatives of these strains have shorter pili. Cpx-regulon member DsbA is critically important for the proper folding of the PapD chaperone, which is required for proper pilus assembly. Meanwhile, DegP degrades misfolded subunits in the periplasm. These observations strongly suggest that Cpx functions in pilus biogenesis.

 Recall that [sigma]E regulates degP transcription in response to changes in the levels of outer membrane [beta]-barrel proteins. The C-terminal amino acids of several [beta]-barrel proteins are conserved, and DegS binds peptides ending with this conserved C-terminal sequence. This binding activates DegS, which cleaves RseA and triggers a cascade of cleavage events that activate the [sigma]E pathway. These findings are also consistent with the idea that the [gj]E pathway functions primarily to monitor [beta]-barrel proteins.


The Cpx Pathway and Adhesion 


Knowing that the Cpx pathway plays a role in P pili assembly, it seemed possible that the Cpx pathway plays a role in surface sensing. Using Cpx-regulated transcriptional lacZ fusions and small hydrophobic glass beads, we showed that Cpx is induced when cells attach to hydrophobic surfaces. Moreover, transcriptional activation requires not only an intact CpxRA pathway, but also the outer membrane lipoprotein NlpE.

 Glass beads provide a convenient artificial surface. Under these conditions, pili are not required for adhesion, and the biological significance of these studies is not yet clear. Since NlpE is required for cells to adhere to hydrophobic surfaces, it follows that NlpE is an upstream component of the Cpx signal transduction pathway. One possibility is that adhesion damages the outer membrane, and NlpE functions as a general outer membrane damage sensor for the Cpx pathway. An equally likely possibility is that adhesion sends a signal to the Cpx pathway that is distinct from that of envelope stress, and that the role of NlpE is specific to adhesion.

 The Cpx pathway is induced by a variety of disparate stresses. For instance, alkaline pH, high levels of the lipid II biosynthetic intermediate of the enterobacterial common antigen (ECA), and high levels of phosphatidylethanolamine also can induce Cpx. Each of these envelope stresses signal through the periplasmic domain of CpxA and do not require NlpE for signal transduction; thus, the role of NlpE is specific to surface sensing.


The [Sigma]E Pathway: [Beta]-Barrel Targeting to the Outer Membrane 


P pili are assembled using a well-characterized chaperone-usher system. Following translocation to the periplasm by the Sec machinery, pilin subunits interact with the PapD chaperone in the periplasm. PapD escorts those subunits to the PapC usher in the outer membrane where the pilus is assembled (Fig. 2). We suggest that other outer membrane proteins also follow this assembly paradigm.

 Lipoproteins are made as precursors in the cytoplasm, translocated through the Sec machinery, and modified at the periplasmic face of the inner membrane. Following modification, mature lipoproteins destined for the outer membrane interact with the LolCDE ABC transporter, and in an ATP-dependent manner are released from the inner membrane to form a complex with the carrier periplasmic protein LolA. The LolA-lipoprotein complex then crosses the periplasm and interacts with LolB, which serves as an assembly site for integrating lipoproteins into the outer membrane.

 [Beta]-barrel proteins are likely escorted through the periplasm to outer membrane assembly sites as well. Several periplasmic chaperones have been implicated in [beta]-barrel folding and targeting. For example, biochemical evidence supports a role for SurA in LamB assembly. However, mutants that lack a particular chaperone typically show only modest defects in [beta]-barrel assembly and, often, no defect is detectable. One explanation is that there are functionally redundant chaperones in the periplasm.

 Evidence for functional redundancy for periplasmic chaperones comes from the genetic demonstration of synthetic lethality. For example, the Skp and DegP chaperones share a redundant function with SurA. Mutants lacking any one of these three proteins grow well under ordinary laboratory conditions. Double mutants lacking both Skp and SurA, or DegP and SurA, are nonviable. 

 Interesting candidates for outer membrane assembly sites have recently been revealed--for example Imp (OstA). The gene for Imp is located in an operon with the SurA chaperone. Imp contains N-terminal and C-terminal cysteine-containing structural motifs that are shared with outer membrane usher proteins such as FimD and PapC. Like these proteins, Imp forms high-molecular-weight complexes in the outer membrane. Moreover, Imp is essential. Depletion studies show that, in the absence of Imp, newly synthesized membrane proteins and lipids are mislocalized. These results are consistent with an important role for Imp in envelope biogenesis. 




Braun, M., and T. J. Silhavy. 2002. Imp/OstA is required for cell envelope biogenesis in Escherichia coli. Mol. Microbiol. 45:1289-1302. 


Cosma, C. L., P. N. Danese, J. H. Carlson, T. J. Silhavy, and W. B. Snyder. 1995. Mutational activation of the Cpx signal transduction pathway of Escherichia coli suppresses the toxicity conferred by certain envelope-associated stresses. Mol. Microbiol. 18:491-505. 


Danese, P. N., and T. J. Silhavy. 1997. The sigma(E) and the Cpx signal transduction systems control the synthesis of periplasmic protein-folding enzymes in Escherichia coli. Genes Dev. 11:1183-1193. 


Jones, C. H., P. N. Danese, J. S. Pinkner, T. J. Silhavy, and S. J. Hultgren. 1997. The chaperone-assisted membrane release and folding pathway is sensed by two signal transduction systems. EMBO J. 16:6394-6406. 


Mecsas, J., P. E. Rouviere, J. W. Erickson, T. J. Donohue, and C. A. Gross. 1993. The activity of sigma E, an Escherichia coli heat-inducible sigma-factor, is modulated by expression of outer membrane proteins. Genes Dev. 7:2618-2628. 


Otto, K., and T. J. Silhavy. 2002. Surface sensing and adhesion of Escherichia coli controlled by the Cpx-signaling pathway. Proc. Natl. Acad. Sci. USA 99:2287-2292. 


Raivio, T. L., M. W. Laird, J. C. Joly, and T. J. Silhavy. 2000. Tethering of CpxP to the inner membrane prevents spheroplast induction of the cpx envelope stress response. Mol. Microbiol. 37:1186-1197. 


Snyder, W. B., and T. J. Silhavy. 1995. Beta-galactosidase is inactivated by intermolecular disulfide bonds and is toxic when secreted to the periplasm of Escherichia coli. J. Bacteriol. 177:953-963. 


Walsh, N. P., B. M. Alba, B. Bose, C. A. Gross, and R. T. Sauer. 2003. OMP peptide signals initiate the envelope-stress response by activating DegS protease via relief of inhibition mediated by its PDZ domain. Cell 113:61-71.