In vivo and in vitro studies on interactions between the components of the hemolysin (HlyA) secretion machinery of Escherichia coli

The glycopeptide antibiotic vancomycin blocks cell wall synthesis in Escherichia coli only when it can reach its target site in the periplasm. In vivo, sensitivity to vancomycin is enhanced in the presence of the hemolysin (hly) determinant of E. coli or its translocator portion hlyBD. Two different mutations in hlyD alter the cell's susceptibility to vancomycin: mutations in the tolC-homologous region of hlyD increase vancomycin resistance, whereas mutations at the 3′-terminus of hlyD lead to hypersensitivity to vancomycin and to the accumulation of large periplasmic and cytoplasmic pools of this antibiotic in E. coli. These effects are only observed in the presence of functional HlyB and TolC, the two other components of the hemolysin secretion machinery. A defect in TolC causes hyperresistance to vancomycin, even when present together with a mutant HlyD protein which in the presence of TolC renders E. coli hypersensitive to vancomycin. Lipid bilayer experiments in vitro revealed specific interactions between TolC and vancomycin or HlyD protein. Second-site suppressor mutations in hlyD and hlyB were obtained, which abolish the hypersensitive phenotype caused by the 3′-terminal mutations in hlyD. Our results are compatible with the idea that (a) TolC, together with the TolC-homologous part of HlyD, forms a pore in the outer membrane through which hemolysin is released and vancomycin taken up; and (b) the C-terminal sequence of HlyD interacts with periplasmic loop(s) of HlyB to form a closed channel spanning the periplasm. Received: 7 April 1997 / Accepted: 28 May 1997     

Membrane fusion proteins of type I secretion system and tripartite efflux pumps share a binding motif for TolC in gram-negative bacteria

The Hly translocator complex of Escherichia coli catalyzes type I secretion of the toxin hemolysin A (HlyA). In this complex, HlyB is an inner membrane ABC (ATP Binding Cassette)-type transporter, TolC is an outer membrane channel protein, and HlyD is a periplasmic adaptor anchored in the inner membrane that bridges HlyB to TolC. This tripartite organization is reminiscent of that of drug efflux systems such as AcrA-AcrB-TolC and MacA-MacB-TolC of E. coli. We have previously shown the crucial role of conserved residues located at the hairpin tip region of AcrA and MacA adaptors during assembly of their cognate systems. In this study, we investigated the role of the putative tip region of HlyD using HlyD mutants with single amino acid substitutions at the conserved positions. In vivo and in vitro data show that all mutations abolished HlyD binding to TolC and resulted in the absence of HlyA secretion. Together, our results suggest that, similarly to AcrA and MacA, HlyD interacts with TolC in a tip-to-tip manner. A general model in which these conserved interactions induce opening of TolC during drug efflux and type I secretion is discussed.     

Type 1 protein secretion in bacteria,the ABC-transporter dependent pathway (Review)

The relatively simple type 1 secretion system in Gram-negative bacteria is nevertheless capable of transporting polypeptides of up to 800 kDa across the cell envelope in a few seconds. The translocator is composed of an ABC-transporter, providing energy through ATP hydrolysis (and perhaps the initial channel across the inner membrane), linked to a multimeric Membrane Fusion Protein (MFP) spanning the initial part of the periplasm and forming a continuous channel to the surface with an outer membrane trimeric protein. Proteins targeted to the translocator carry an (uncleaved), poorly conserved secretion signal of approximately 50 residues. In E. coli the HlyA toxin interacts with both the MFP (HlyD) and the ABC protein HlyB, (a half transporter) triggering, via a conformational change in HlyD, recruitment of the third component, TolC, into the transenvelope complex. In vitro, HlyA, through its secretion signal, binds to the nucleotide binding domain (NBD or ABC-ATPase) of HlyB in a reaction reversible by ATP that may mimic initial movement of HlyA into the translocation channel. HlyA is then transported rapidly, apparently in an unfolded form, to the cell surface, where folding and release takes place. Whilst recent structural studies of TolC and MFP-like proteins are providing atomic detail of much of the transport path, structural analysis of the HlyB NBD and other ABC ATPases, have revealed details of the catalytic cycle within an NBD dimer and a glimpse of how the action of HlyB is coupled to the translocation of HlyA.     

Type 1 protein secretion in bacteria, the ABC-transporter dependent pathway (review)

The relatively simple type 1 secretion system in gram-negative bacteria is nevertheless capable of transporting polypeptides of up to 800 kDa across the cell envelope in a few seconds. The translocator is composed of an ABC-transporter, providing energy through ATP hydrolysis (and perhaps the initial channel across the inner membrane), linked to a multimeric Membrane Fusion Protein (MFP) spanning the initial part of the periplasm and forming a continuous channel to the surface with an outer membrane trimeric protein. Proteins targeted to the translocator carry an (uncleaved), poorly conserved secretion signal of approximately 50 residues. In E. coli the HlyA toxin interacts with both the MFP (HlyD) and the ABC protein HlyB, (a half transporter) triggering, via a conformational change in HlyD, recruitment of the third component, TolC, into the transenvelope complex. In vitro, HlyA, through its secretion signal, binds to the nucleotide binding domain (NBD or ABC-ATPase) of HlyB in a reaction reversible by ATP that may mimic initial movement of HlyA into the translocation channel. HlyA is then transported rapidly, apparently in an unfolded form, to the cell surface, where folding and release takes place. Whilst recent structural studies of TolC and MFP-like proteins are providing atomic detail of much of the transport path, structural analysis of the HlyB NBD and other ABC ATPases, have revealed details of the catalytic cycle within an NBD dimer and a glimpse of how the action of HlyB is coupled to the translocation of HlyA.     

Change in the cellular localization of alkaline phosphatase by alteration of its carboxy-terminal sequence

Summary Alkaline phosphatase (AP) is secreted into the medium when the carboxy-terminal 25 amino acids are replaced by the 60 amino acid carboxy-terminal signal peptide (HlyA_s) ofEscherichia coli haemolysin (HlyA). Secretion of the AP-HlyA_s fusion protein is dependent on HlyB and HlyD but independent of SecA and SecY. The efficiency of secretion by HlyB/HlyD is decreased when AP carries its own N-terminal signal peptide. Translocation of this fusion protein into the periplasm is not observed even in the absence of HlyB/HlyD. The failure of the Sec export machinery to transport the latter protein into the periplasm seems to be due in part to the loss of the carboxy-terminal sequence of AP since even AP derivatives which do not carry the HlyA signal peptide but lack the 25 C-terminal amino acids of AP are localized in the membrane but not translocated into the periplasm.     

In vivo and in vitro studies on interactions between the components of the hemolysin (HlyA) secretion machinery of Escherichia coli

The glycopeptide antibiotic vancomycin blocks cell wall synthesis in Escherichia coli only when it can reach its target site in the periplasm. In vivo, sensitivity to vancomycin is enhanced in the presence of the hemolysin (hly) determinant of E. coli or its translocator portion hlyBD. Two different mutations in hlyD alter the cell's susceptibility to vancomycin: mutations in the tolC-homologous region of hlyD increase vancomycin resistance, whereas mutations at the 3′-terminus of hlyD lead to hypersensitivity to vancomycin and to the accumulation of large periplasmic and cytoplasmic pools of this antibiotic in E. coli. These effects are only observed in the presence of functional HlyB and TolC, the two other components of the hemolysin secretion machinery. A defect in TolC causes hyperresistance to vancomycin, even when present together with a mutant HlyD protein which in the presence of TolC renders E. coli hypersensitive to vancomycin. Lipid bilayer experiments in vitro revealed specific interactions between TolC and vancomycin or HlyD protein. Second-site suppressor mutations in hlyD and hlyB were obtained, which abolish the hypersensitive phenotype caused by the 3′-terminal mutations in hlyD. Our results are compatible with the idea that (a) TolC, together with the TolC-homologous part of HlyD, forms a pore in the outer membrane through which hemolysin is released and vancomycin taken up; and (b) the C-terminal sequence of HlyD interacts with periplasmic loop(s) of HlyB to form a closed channel spanning the periplasm.     

Substrate-triggered recruitment of the TolC channel-tunnel during type I export of hemolysin by Escherichia coli.

A defining event in type I export of hemolysin by Escherichia coli is the substrate-triggered recruitment of the TolC channel-tunnel by an inner membrane complex. This complex comprises a traffic ATPase (HlyB) and the 478 residue adaptor protein (HlyD), which contacts TolC during recruitment. HlyD has a large periplasmic domain (amino acid residues 81-478) linked by a single transmembrane helix to a small N-terminal cytosolic domain (1-59). Export was disabled by deletion of the ca 60 amino acid residue cytosolic domain of HlyD, even though the truncated HlyD (HlyDDelta45) was, like the wild-type, able to trimerise in the cytosolic membrane, and interact with the traffic ATPase. The mutant HlyB/HlyDDelta45 inner membrane complex engaged the hemolysin substrate, but this substrate-engaged complex failed to trigger recruitment of TolC. Further analyses showed that HlyDDelta45 was specifically unable to bind the substrate. The result suggests that substrate engagement by the traffic ATPase alone is insufficient to trigger TolC recruitment, and that substrate binding to the HlyD cytosolic domain is essential. Analysis of three further N-terminal deletion variants, HlyDDelta26, HlyDDelta26-45 and HlyDDelta34-38, indicated that an extreme N-terminal amphipathic helix and a cytosolic cluster of charged residues are central to the cytosolic domain function. The cytosolic amphipathic helix was not essential for substrate engagement or TolC recruitment, but export was impaired without it. In contrast, when the charged amino acid residues were deleted, the substrate was still engaged by HlyD but engagement was unproductive, i.e. TolC recruitment was not triggered. Our results are compatible with the HlyD cytosolic domain mediating transduction of the substrate binding signal directly, presumably to the HlyD periplasmic domain, to trigger recruitment of TolC and assemble the type I export complex.     

Analysis of the haemolysin secretion system by PhoA-HlyA fusion proteins

Summary We studied the efficiency of the pHly152-derived haemolysin transport system using PhoA-HlyA fusion proteins and different constructs which provide HlyB/HlyD in trans. The optimal C-terminal HlyA signal consists of the last 60 amino acids. Longer stretches of HlyA do not improve the transport efficiency of PhoA-HlyA fusion proteins. The introduction of deletions and/or replacements in the 60 amino acid HlyA signal domain revealed at least three functional regions with different degrees of specificity. Amino acids 1–21 (numbered from the N-terminal part of the 60 amino acid HlyA signal), termed region I, could be replaced by a Pro-containing peptide. The other two regions II and III (amino acids 22–40 and 41–60, respectively) seem to interact directly with the HlyB/HlyD translocator since a PhoA fusion protein which contains either of the two regions was still secreted in a HlyB/HlyD-dependent mode, albeit at low efficiency. An efficient trans-complementing HlyB/HlyD system was only obtained from the pHLy152-encoded hly determinant when the regulatory hlyR element was provided in cis. Secretion of the PhoA-HlyA fusion protein did not interfere with the secretion of HlyA even when the fusion protein was induced to a high level. This suggests that the capacity of the HlyB/HlyD translocation system is high and not normally saturated by its natural HlyA substrate.Dedicated to Prof., Dr. F. Lingens on the occasion of his 65th birthday     

The mechanism of secretion of hemolysin and other polypeptides from Gram-negative bacteria

In the secretion of polypeptides from Gram-negative bacteria, the outer membrane constitutes a specific barrier which has to be circumvented. In the majority of systems, secretion is two-step process, with initial export to the periplasm involving an N-terminal signal sequence. Transport across the outer membrane then involves a variable number of ancillary polypeptides including both periplasmic and outer membrane. While such ancillary proteins are probably specific for each secreted protein, the mechanism of movement across the outer membrane is unknown. In contrast to these systems, secretion of theE. coli hemolysin (HlyA) has several distinctive features. These include a novel targeting signal located within the last 50 or so C-terminal amino acids, the absence of any periplasmic intermediates in transfer, and a specific membrane-bound translocator, HlyB, with important mammalian homologues such as P-glycoprotein (Mdr) and the cystic fibrosis protein. In this review we discuss the nature of the HlyA targeting signal, the structure and function of HlyB, and the probability that HlyA is secreted directly to the medium through a trans-envelope complex composed of HlyB and HlyD.     

Substrate-induced assembly of a contiguous channel for protein export from E.coli: reversible bridging of an inner-membrane translocase to an outer membrane exit pore.

The toxin HlyA is exported from Escherichia coli, without a periplasmic intermediate, by a type I system comprising an energized inner-membrane (IM) translocase of two proteins, HlyD and the traffic ATPase HlyB, and the outer-membrane (OM) porin-like TolC. These and the toxin substrate were expressed separately to reconstitute export and, via affinity tags on the IM proteins, cross-linked in vivo complexes were isolated before and after substrate engagement. HlyD and HlyB assembled a stable IM complex in the absence of TolC and substrate. Both engaged HlyA, inducing the IM complex to contact TolC, concomitant with conformational change in all three exporter components. The IM-OM bridge was formed primarily by HlyD, which assembled to stable IM trimers, corresponding to the OM trimers of TolC. The bridge was transient, components reverting to IM and OM states after translocation. Mutant HlyB that bound, but did not hydrolyse ATP, supported IM complex assembly, substrate recruitment and bridging, but HlyA stalled in the channel. A similar picture was evident when the HlyD C-terminus was masked. Export thus occurs via a contiguous channel which is formed, without traffic ATPase ATP hydrolysis, by substrate-induced, reversible bridging of the IM translocase to the OM export pore.     

Analysis of the membrane organization of an Escherichia coli protein translocator, HlyB, a member of a large family of prokaryote and eukaryote surface transport proteins

Haemolysin B (HlyB) is essential for secretion of the 107 x 10(3) Mr haemolysin A protein from Escherichia coli and is a member of a family of highly conserved, apparently ATP-dependent surface proteins in many organisms. We have shown in this study that both HlyB and HlyD fractionate primarily with the cytoplasmic membrane of E. coli and are accessible to proteases after removal of the outer membrane. We have measured experimentally the topological organization of HlyB within the membrane by construction of fusions to beta-lactamase as a reporter. The predicted folding of HlyB, with a minimum of six transmembrane segments, does not always coincide with regions of highest average hydrophobicity. This suggests that HlyB may have a novel organization within the bilayer. From our data and comparative sequence analysis, we have been able to predict very similar topological models for the other members of the HlyB family.     

Identification of individual amino acids required for secretion within the haemolysin (HlyA) C-terminal targeting region

The release of haemolysin from Escherichia coli involves direct secretion across both the inner and outer membranes. Secretion of HlyA is dependent upon a specific membrane export complex composed of HlyB, -D and possibly TolC. HlyA is targeted to the medium via the membrane translocation complex, by a novel C-terminal secretion signal. Previous studies involving deletion and fusion analyses have given contradictory results for the minimal length (20-60 residues) of this HlyA signal region and little is known of the nature of the specific residues and structural features required for function. In this study we have analysed, quantitatively, the effect upon secretion of many point mutations introduced into the HlyA C-terminus. The results indicate the presence of a minimal secretion signal domain whose proximal boundary extends to at least residue -46 and which contains at least four individual residues essential for maximal secretion levels. We propose that such residues act co-operatively, forming multiple contact points with the translocator proteins, with the 'best fit' promoting maximal levels of secretion.     

Topological and functional studies on HlyB of Escherichia coli

Summary The topology of HlyB, a protein located in the inner membrane of Escherichia coli and involved in the secretion of -haemolysin (HlyA), was determined by the generation of HlyB-PhoA and HlyB-LacZ fusion proteins. The data obtained by this biochemical method together with computer predictions suggest that HlyB is inserted in the cytoplasmic membrane by six stable hydrophobic, -helical transmembrane segments. These segments extend from amino acid positions 158 to 432 of HlyB. The cytoplasmic loops between these transmembrane segments are relatively large and carry an excess of positively charged amino acids, while the periplasmic loops are rather small. In addition to these six transmembrane segments, two additional regions in the 78 N-terminal amino acids of HlyB appear to be also inserted in the cytoplasmic membrane. However, the association of these two segments with the cytoplasmic membrane seems to be less tight, since active PhoA and LacZ fusions were obtained by insertion into the same positions of these segments. A LacZ-HlyAs fusion protein carrying, at the C-terminus of LacZ, the 60-amino acid signal sequence of HlyA was not secreted in the presence of HlyB/HlyD. However, transport of this fusion protein into the cytoplasmic membrane appeared to be initiated, as suggested by the tight association of this protein with the inner membrane. A similar close association of LacZ-HlyA_s with the inner membrane was also observed in the presence of HlyB alone but not in its absence. These data suggest that HlyB recognizes the HlyA signal sequence and initiates the transport of HlyA into the membrane.     

Secretion of active β-lactamase to the medium mediated by the Escherichia coli haemolysin transport pathway

An in frame gene fusion containing the coding region for mature β-lactamase and the 3′-end of hylA encoding the haemolysin secretion signal, was constructed under the control of a lac promoter. The resulting 53 kDa hybrid protein was specifically secreted to the external medium in the presence of the haemolysin translocator proteins, HlyB and HlyD. The specific activity of the β-lactamase portion of the secreted protein (measured by the hydrolysis of penicillin G), approximately 1 U/μg protein, was close to that of authentic, purified TEM-β-lactamase. This is an important example of a hybrid protein that is enzymatically active, and secreted via the haemolysin pathway. Previous studies have indicated that haemolysin is secreted directly into the medium, bypassing the periplasm, to which β-lactamase is normally targeted. This study indicated, therefore, that normal folding of an active β-lactamase, can occur, at least when fused to the HlyA C-terminus, without the necessity of entering the periplasm. Despite the secretion of approximately 5 μg/ml levels of the active β-lactamase fusion into the medium, there was maximally only a 50% detectable increase in the LD₅₀ for resistance to ampicillin at the individual cell level. This result suggests that, normally, resistance to ampicillin requires a high concentration of the enzyme close to killing targets, i.e. in the periplasm, in order to achieve significant levels of protection.     

A heterologous membrane protein domain fused to the C-terminal ATP-binding domain of HlyB can export Escherichia coli hemolysin.

The hydrophobic-rich NH2-terminal 34 amino acids of a tetracycline resistance determinant (TetC) were fused to the COOH-terminal 240 amino acids of the hemolysin transporter, HlyB, which contains a putative ATP-binding domain. This hybrid protein replaced the NH2-terminal 467-amino-acid portion of HlyB and could still export the Escherichia coli hemolysin (HlyA). Export by the hybrid protein was approximately 10% as efficient as transport by HlyB. Extracellular secretion of HlyA by the TetC-HlyB hybrid required HlyD and TolC. The extracellular and periplasmic levels of beta-galactosidase and beta-lactamase in strains that produced the hybrid were similar to the levels in controls. Thus, HlyA transport was specific and did not appear to be due to leakage of cytoplasmic contents alone. Antibodies raised against the COOH terminus of HlyB reacted with the hybrid protein, as well as HlyB. HlyB was associated with membrane fractions, while the hybrid protein was found mainly in soluble extracts. Cellular fractionation studies were performed to determine whether transport by the hybrid occurred simultaneously across both membranes like wild-type HlyA secretion. However, we found that HlyA was present in the periplasm of strains that expressed the TetC-HlyB hybrid. HlyA remained in the periplasm unless the hlyD and tolC gene products were present in addition to the hybrid.     

Analysis of the haemolysin transport process through the secretion from Escherichia coli of PCM, CAT or β-galactosidase fused to the Hly C-terminal signal domain

Secretion of haemolysin (HlyA) is secA independent, but depends upon two accessory membrane proteins, HlyB and HlyD, encoded by the hly determinant. A fourth (cytoplasmic) protein, HlyC, is required to activate HlyA post-translationally, but has no role in export. Deletion studies have previously shown that the HlyA molecule contains a targeting signal close to the C-terminus which specifically directs its secretion to the medium. This targeting signal has been variously located within the terminal 27, 53, 60 or 113 amino acids. In this paper, we have sought to confirm the presence of a C-terminal targeting signal and to analyse the specificity of the Hly transport system through fusion of C-terminal fragments of HlyA to heterologous polypeptides. A C-terminal fragment (23 kDa) of HlyA, when fused at the C-terminus, efficiently promoted the secretion of the eukaryotic protein prochymosin (PCM) to the medium via HlyB and HlyD. This result is in contrast to previous findings that prochymosin, preceded by the alkaline phosphatase signal sequence, cannot be translocated across the Escherichia coli inner membrane. The HlyA targeting domain was also used to secrete to the medium varying portions of chloramphenicol acetyltransferase (CAT) and 98 per cent of the beta-galactosidase (LacZ) molecule (both E. coli cytoplasmic proteins). In the case of the PCM and CAT fusions the efficiency of secretion was reduced as the proportion of the PCM and CAT molecule increased. This result is consistent with inhibition of secretion through the irreversible folding of the larger passenger protein fragments, or the occlusion of the HlyA targeting signal by upstream sequences. Analysis of the nature of the C-terminal domain promoting secretion of prochymosin, demonstrated that shortening the signal domain from 218 to 113 amino acids significantly reduced the efficiency of secretion. This result may also reflect the importance of maintaining an independently folded signal motif well separated from a passenger domain.     

Repeat sequences in the Bordetella pertussis adenylate cyclase toxin can be recognized as alternative carboxy-proximal secretion signals by the Escherichia coliα-haemolysin translocator

The 1706-residue adenylate cyclase toxin (CyaA) of Bordetella pertussis is an RTX protein with extensive carboxy-proximai glycine and aspartate-rich repeats. CyaA does not have a cleavable amino-terminal signal peptide and can be secreted across both bacterial membranes of the Escherichia coli cell envelope by the α-haemolysin (HlyA) translocator (HlyBD/TolC). We performed deletion mapping of secretion signals recognized in CyaA by this heterologous translocator. Truncated proteins with N–terminal and internal deletions were secreted at levels up to 10 times higher than intact CyaA and similar to HlyA. A secretion signal recognized by HlyBD/ToiC was found within the last 74 residues of CyaA. However, secretion of CyaA was reduced but not abolished upon deletion of the last 75 or 217 residues, indicating that at least two additional secretion signals recognized by HlyBD/TolC are within CyaA. One of them was localized to the repeat sequence between residues Asp-1587 to lle-1631. Interestingly, a conserved acidic' motif (Glu/Asp)-(X)₁₁-Asp-(X)_3/5-(Glu/Asp)-(X)₁₄-Asp was found in the C-terminal sequences of HlyA, CyaA and the two secreted CyaA derivatives. We speculate that the presence and spacing of acidic residues may be an important feature of secretion signals recognized by the haemolysin translocator.     

Mutations in HlyD, part of the type 1 translocator for hemolysin secretion, affect the folding of the secreted toxin

HlyD, a member of the membrane fusion protein family, is essential for the secretion of the RTX hemolytic toxin HlyA from Escherichia coli. Random point mutations affecting HlyA secretion were obtained, distributed in most periplasmic regions of the HlyD molecule. Analysis of the secretion phenotypes of different mutants allowed the identification of regions in HlyD involved in different steps of HlyA translocation. Four mutants, V349-I, T85-I, V334-I and L165-Q, were conditionally defective, a phenotype shown to be linked to the presence of inhibitory concentrations of Ca2+ in extracellular medium. Hly mutant T85-I was defective at an early stage in secretion, while mutants V334-I and L165-Q appeared to accumulate HlyA in the cell envelope, indicating a block at an intermediate step. Mutants V349-I, V334-I, and L165-Q were only partially defective in secretion, allowing significant levels of HlyA to be transported, but in the case of V349-I and L165-Q the HlyA molecules secreted showed greatly reduced hemolytic activity. Hemolysin molecules secreted from V349-I and V334-I are defective in normal folding and can be reactivated in vitro to the same levels as HlyA secreted from the wild-type translocator. Both V349-I and V334-I mutations mapped to the C-terminal lipoyl repeat motif, involved in the switching from the helical hairpin to the extended form of HlyD during assembly of the functional transport channel. These results suggest that HlyD is an integral component of the transport pathway, whose integrity is essential for the final folding of secreted HlyA into its active form.     

Energetically distinct early and late stages of HlyB/HlyD-dependent secretion across both Escherichia coli membranes.

The alternative secretion pathway which exports hemolysin across both Escherichia coli membranes into the surrounding medium is directed by an uncleaved C-terminal targeting signal and the membrane translocator proteins HlyD and HlyB. In order to identify stages and intermediates in this unconventional secretion process we have examined the effect of inhibition of the total proton motive force (delta P) and its components during the in vivo HlyB/HlyD-dependent export of a 22.4 kDa secretion competent HlyA C-terminal peptide (Actp). Secretion of Actp was severely inhibited by the proton ionophore carbonylcyanide m-chlorophenylhydrazone (CCCP), which collapses simultaneously membrane potential delta psi and the proton gradient delta pH, and also by valinomycin/K+, a potassium ionophore which disrupts delta psi. The inhibition of secretion by valinomycin/K+ was ameliorated by imposition of a pH gradient, the second component of the delta P, and selective depletion of delta pH by nigericin also blocked secretion. This indicates that, as in the secretion of beta-lactamase to the periplasm, HlyB/D-directed secretion requires delta P itself and not specifically one of its components. However, inhibition of HlyB/D-dependent secretion was only marked when CCCP, valinomycin/K+ or nigericin were present during the early stage of Actp secretion; at a later stage the secretion was not significantly inhibited. HlyB/D-dependent secretion appears therefore to share with conventional secretion across the cytoplasmic membrane an early requirement for delta P, but comprises in addition a late stage which does not require delta P, delta psi or delta pH. The translocation intermediate identified in the delta P-independent late stage of secretion was associated with the membrane fraction. Analysis of the protease accessibility of this intermediate in whole cells and spheroplasts showed that it was not in the periplasm, nor was it exposed on the cell surface or on the periplasmic faces of either the inner or outer membranes. This may reflect its close association with the inner membrane or a membrane translocation complex.