Maturation and secretion of the non-typable Haemophilus influenzae HMW1 adhesin: roles of the N-terminal and C-terminal domains

Authors

  • Susan Grass,

    1. Edward Mallinckrodt Department of Pediatrics and Department of Molecular Microbiology, Washington University School of Medicine, and Division of Infectious Diseases, St. Louis Children’s Hospital, St. Louis, MO, USA.
    Search for more papers by this author
  • Joseph W. St. Geme III

    Corresponding author
    1. Edward Mallinckrodt Department of Pediatrics and Department of Molecular Microbiology, Washington University School of Medicine, and Division of Infectious Diseases, St. Louis Children’s Hospital, St. Louis, MO, USA.
    Search for more papers by this author

Abstract

Non-typable Haemophilus influenzae is a common cause of human disease and initiates infection by colonizing the upper respiratory tract. The non-typable H. influenzae HMW1 and HMW2 adhesins mediate attachment to human epithelial cells, an essential step in the process of colonization. HMW1 and HMW2 have an unusual N-terminus and undergo cleavage of a 441-amino-acid N-terminal fragment during the course of their maturation. Following translocation across the outer membrane, they remain loosely associated with the bacterial surface, except for a small amount that is released extracellularly. In the present study, we localized the signal sequence to the first 68 amino acids, which are characterized by a highly charged region from amino acids 1–48, followed by a more typical signal peptide with a predicted leader peptidase cleavage site after the amino acid at position 68. Additional experiments established that the SecA ATPase and the SecE translocase are essential for normal export and demonstrated that maturation involves cleavage first between residues 68 and 69, via leader peptidase, and next between residues 441 and 442. Site-directed mutagenesis revealed that HMW1 processing, secretion and extracellular release are dependent on amino acids in the region between residues 150 and 166 and suggested that this region interacts with the HMW1B outer membrane translocator. Deletion of the C-terminal end of HMW1 resulted in augmented extracellular release and elimination of HMW1-mediated adherence, arguing that the C-terminus may serve to tether the adhesin to the bacterial surface. These observations suggest that the HMW proteins are secreted by a variant form of the general secretory pathway and provide insight into the mechanisms of secretion of a growing family of Gram-negative bacterial exoproteins.

Introduction

Colonization of a host mucosal surface is an essential early step in the pathogenesis of most bacterial diseases (Beachey, 1981). In general, the process of colonization requires specialized microbial factors called adhesins, which promote adherence to host epithelium and thereby prevent removal by bulk forces (Hultgren et al., 1993, 1996). Two primary classes of adhesins exist, pili (or fimbriae), which are polymeric, hair-like fibres, and non-pilus adhesins, which may be monomeric or oligomeric structures and are difficult to visualize by electron microscopy (Hultgren et al., 1993).

In Gram-negative bacteria, adhesins must be secreted across both the inner (cytoplasmic) and outer membranes to gain access to host cell receptors. Most surface or extracellular proteins are secreted via the general secretory pathway (also called the type II pathway), which contains the Sec complex of proteins for export from the cytoplasm and depends on the presence of a cleavable amino-terminal signal (leader) sequence in the secreted protein (Schatz and Beckwith, 1990). With these proteins, translocation across the outer membrane involves a separate secretion apparatus, which typically consists of multiple components (Pugsley, 1993). Other proteins are secreted independently of the Sec apparatus, using either the type I pathway (for example, Escherichia coliα-haemolysin) or the type III system (for example, the Yersinia Yop proteins and the Salmonella and Shigella invasion antigens) (Felmlee et al., 1985; Holland et al., 1990; Hueck, 1998).

Non-typable (non-encapsulated) Haemophilus influenzae is a Gram-negative bacterium that commonly causes human respiratory tract disease, including otitis media, sinusitis, bronchitis and pneumonia (Turk, 1984). In immunocompromised hosts, non-typable H. influenzae is also an important cause of serious systemic diseases, such as septicaemia, endocarditis and meningitis (St. Geme, 1993). Disease due to non-typable H. influenzae begins with colonization of the upper respiratory tract (Murphy et al., 1987), which appears to be influenced by both pilus and non-pilus adhesins (Rao et al., 1999).

The H. influenzae HMW1 and HMW2 proteins are non-pilus adhesins that were first discovered during efforts to identify the major targets of the serum antibody response to H. influenzae infection (Barenkamp and Bodor, 1990; Barenkamp and Leininger, 1992). Subsequently, these proteins were shown to promote attachment to a variety of cultured human epithelial cells (St. Geme et al., 1993; Hultgren et al., 1993). HMW1 is a 160 kDa protein encoded by hmw1A and HMW2 is a 155 kDa polypeptide encoded by hmw2A, with the predicted amino acid sequences sharing 71% identity and 80% similarity (Barenkamp and Leininger, 1992). Both hmw1A and hmw2A are flanked downstream by two additional genes, designated B and C respectively (Barenkamp and St. Geme, 1994). The 1B and 2B genes encode proteins that are 99% identical, whereas the 1C and 2C genes code for polypeptides that are 96% identical.

In order to facilitate interaction with host cells, HMW1 and HMW2 are localized on the surface of the organism in a process that involves cleavage of a 441-amino-acid N-terminal fragment. As a consequence of cleavage, the mature proteins are 125 kDa and 120 kDa in size respectively. In recent work, we initiated studies of the pathway for secretion of HMW1 and HMW2, which appears to be the same whether these proteins are expressed in H. influenzae or E. coli (St. Geme and Grass, 1998). Cell fractionation experiments and immunoelectron microscopy demonstrated that a periplasmic intermediate occurs. Additional analysis revealed that the proteins ultimately are partially released from the surface of the organism. Examination of plasmid subclones and mutant strains established that the HMWB and HMWC proteins are required for normal processing and secretion and indicated that HMW1B/HMW1C and HMW2B/HMW2C are interchangeable. Additional studies of the hmw1 locus demonstrated that HMW1B is located in the outer membrane and serves to translocate HMW1 across the outer membrane. In the absence of HMW1B, HMW1 remains unprocessed and is degraded in the periplasm, at least in part by the DegP protease.

In considering the mechanism of secretion of HMW1 and HMW2, it is noteworthy that these proteins have an unusual N-terminus with 15 charged amino acids scattered through the first 48 residues. It is also notable that the cleavage event responsible for removal of the N-terminal 441-amino-acid fragment occurs in the periplasm, apparently after interaction with the periplasmic face of HMW1B and prior to translocation across the outer membrane (St. Geme and Grass, 1998). Although a periplasmic intermediate is detected, it remains unclear whether HMW1 and HMW2 secretion involves the Sec system. Similarly, it is unknown whether other proteolytic events precede cleavage between amino acids 441 and 442.

In the present study, we extended our earlier analysis of the mechanism of secretion of HMW1, investigating the roles of the HMW1 N-terminal and C-terminal domains. Our results indicate that the HMW1 N-terminus contains the signal for export from the cytoplasm and appears to interact with the Sec system. Initial proteolytic cleavage occurs between amino acids 68 and 69, after a typical prokaryotic signal sequence that begins around residue 48. Following HMW1 delivery to the periplasm, the region between residues 150 and 166 is essential for translocation across the outer membrane, presumably via interaction with HMW1B. The HMW1 C-terminus appears to anchor the protein to the surface of the organism, facilitating HMW1-mediated adherence.

Results

Cleavage of the HMW1 N-terminal 441-amino-acid fragment is not required for surface localization or adherence

In initial experiments, we examined whether cleavage of the HMW1 441-amino-acid N-terminal fragment might be a prerequisite for HMW1 secretion and extracellular release or for HMW1 folding and adhesive activity. To address this question, amino acids 441 and 443 were both mutated from aspartic acid to glycine, generating the construct pHMW1(A-DPD/GPG), which contains wild-type hmw1B and hmw1C. As shown in Fig. 1, when expressed in E. coli DH5α, the mutated protein failed to undergo the processing event responsible for conversion of the 160 kDa precursor protein to the mature 125 kDa protein. Nevertheless, the 160 kDa protein was localized on the surface of the organism and released extracellularly and remained capable of promoting efficient adherence to cultured epithelial cells (Fig. 1 and Table 1).

Figure 1.

Mutation of the HMW1 cleavage site. Cell sonicates were prepared, periplasmic proteins were extracted and culture supernatants were precipitated from DH5α/pT7–7, DH5α/pHMW1–15 (containing hmw1ABC), and DH5α/pHMW1(A-DPD/GPG) (a derivative of pHMW1–15 with mutation of the cleavage site in HMW1), and Western blot was performed with guinea pig antiserum Gp1031-f, directed against HMW1. In addition, whole cells were fixed with paraformaldehyde and a dot immunoblot was performed, again with antiserum Gp1031-f. Cell sonicates are shown in A, periplasmic proteins are shown in B, culture supernatants are shown in C and dot immunoblots are shown in D. In all panels, samples were loaded as follows: lane 1, DH5α/pT7–7; lane 2, DH5α/pHMW1–15; lane 3, DH5α/pHMW1(A-DPD/GPG). Arrows indicate the mature form of HMW1.

Table 1. Adherence to Chang epithelial cells.

Strain
Adherence
(% inoculum)*
  1. *Adherence was determined in a 30 min assay as described previously ( St. Geme et al., 1993). Values represent means ± standard errors of the means of three measurements from a representative experiment.

DH5α/pT7–70.3 ± 0.1
DH5α/pHMW1–1563.3 ± 4.1
DH5α/pHMW1(A-DPD/GPG)71.8 ± 17.2
DH5α/pHMW(A-Δ2–441)0.4 ± 0.2
DH5α/pHMW1(A-Δ2–68)0.5 ± 0.2
DH5α/pHMW(A-Δ72–441)0.4 ± 0.1
DH5α/pHMW1(A-IAIGI)63.5 ± 4.2
DH5α/pHMW1(A-ITIG)53.8 ± 3.5
DH5α/pHMW1(A-IAIGI/ITIG)0.2 ± 0.0
DH5α/pHMW1A + pHMW1BC55.4 ± 8.8
DH5α/pHMW1A-Δ3′(414) + pHMW1BC0.2 ± 0.0
DH5α/pHMW1A-Δ3′(132) + pHMW1BC0.2 ± 0.0

The HMW1 N-terminal 441-amino-acid fragment contains the signal for export

To address whether the HMW1 N-terminal 441-amino-acid fragment might contain the signal for export across the cytoplasmic membrane, we examined the effect of deletion of amino acids 2–441. The plasmid pHMW(A-Δ2–441) is a derivative of pHMW1–15 that lacks coding sequence for amino acids 2–441 of HMW1 and contains intact hmw1B and hmw1C genes. Following transformation of this plasmid into E. coli DH5α, cell fractions were isolated and examined by Western analysis with antiserum Gp1031-f, directed against HMW1. As shown in Fig. 2 and Table 1, deletion of the HMW1 N-terminus resulted in production of a truncated protein Å 125 kDa in size that was present in whole-cell sonicates but was absent from the periplasm and the culture supernatant and was unable to promote in vitro adherence, thus suggesting a block in export across the inner membrane.

Figure 2.

Deletion of HMW1 N-terminal domains. Cell sonicates were prepared, periplasmic proteins were extracted and culture supernatants were precipitated from DH5α/pT7–7, DH5α/pHMW1–15 (containing hmw1ABC), DH5α/pHMW(A-Δ2–441), DH5α/pHMW(A-Δ2–68) and DH5α/pHMW(A-Δ72–441). Western immunoblot was performed with guinea pig antiserum Gp1031-f, directed against HMW1. Cell sonicates are shown in A, periplasmic proteins are shown in B and culture supernatants are shown in C. Samples were loaded as follows: lane 1, DH5α/pT7–7; lane 2, DH5α/pHMW1–15; lane 3, DH5α/pHMW(A-Δ2–441); lane 4, DH5α/pHMW(A-Δ2–68); lane 5, DH5α/pHMW(A-Δ72–441). Arrows indicate the mature form of HMW1.

To extend these findings, we fused the N-terminal 441 amino acids of HMW1 to PhoA, introduced the resulting construct into E. coli CC118 and then measured alkaline phosphatase activity. As shown in Table 2, the HMW1441–PhoA fusion (pPhoA441) was associated with high-level alkaline phosphatase activity, similar to that seen with the control plasmid pCH39, which contains the β-lactamase signal sequence fused to PhoA.

Table 2. Alkaline phosphatase activity associated with N-terminal fragments of HMW1.

Strain
Alkaline phosphatase
activity*
  1. *Alkaline phosphatase activity was measured as described previously ( Brickman and Beckwith, 1975). Values represent the mean of two measurements from a representative experiment.

CC118/pT7-PhoA713
CC118/pPhoA4412956
CC118/pPhoA3833320
CC118/pPhoA2853120
CC118/pPhoA1464434
CC118/pPhoA685565
CC118/pCH393571

Considered together, these observations suggest that the signal for export of HMW1 resides within the N-terminal 441-amino-acid fragment.

Amino acids 1–68 constitute the HMW1 signal sequence

To further define the signal sequence of HMW1, we constructed a series of PhoA fusion proteins containing progressively smaller fragments of the HMW1 N-terminus. As shown in Table 2, fusions containing the N-terminal 383 amino acids, 285 amino acids, 146 amino acids and 68 amino acids were all associated with alkaline phosphatase activity, suggesting that the initial 68 amino acids are sufficient to direct export out of the cytoplasm. Of note, while the immediate N-terminus of HMW1 bears little resemblance to a signal sequence, residues 48–68 contain features of a typical prokaryotic signal sequence, including a predicted leader peptidase cleavage site following amino acid 68.

To obtain evidence that the initial 68 amino acids function as a typical signal sequence, we examined the possibility that the 68-amino-acid N-terminal fragment is cleaved during HMW1 maturation. Following preparation of a large-volume culture of strain DH5α/pHMW1(A-DPD/GPG) (which expresses HMW1 that no longer undergoes processing after amino acid 441), unprocessed HMW1 was purified from the bacterial surface as described by Barenkamp (1996). N-terminal amino acid sequencing of the secreted protein revealed the sequence SGLQGM, which corresponds to amino acids 69–74 of HMW1, indicating cleavage between residues 68 and 69. With this information in mind, it is noteworthy that a band slightly smaller that the 160 kDa full-length HMW1 protein is present in cell sonicates of DH5α expressing wild-type HMW1 (see Fig. 2A, lane 2).

Consistent with our findings that the N-terminal 68 amino acids direct export of PhoA and are cleaved during the course of secretion, deletion of amino acids 2–68 of HMW1 resulted in a block in export from the cytoplasm (Fig. 2) and loss of the ability to promote adherence (Table 1).

Secretion of HMW1 occurs via a Sec-dependent mechanism

Given that the N-terminus of HMW1 contains features of a typical prokaryotic signal sequence and is cleaved at a predicted leader peptidase cleavage site, we wondered whether export involves the Sec machinery. To begin to address this question, we examined the effect of azide on HMW1 secretion. Low concentrations of azide (2 mM) specifically inhibit the SecA ATPase and block Sec-dependent secretion, resulting in accumulation of preproteins in the cytoplasm (Schatz and Beckwith, 1990; Senior, 1990). In performing this experiment, DH5α/pHMW1–15 was incubated in minimal media, then metabolically labelled with [35S]-trans label. Following radiolabelling for 45 min, HMW1 was immunoprecipitated, resolved by SDS–PAGE and detected by autoradiography. As shown in Fig. 3A, in the absence of azide, two forms of HMW1 were detected, including the proprotein lacking the signal sequence and the 125 kDa mature form. Western analysis confirmed the identity of these proteins (data not shown). Incubation in the presence of azide blocked conversion of the preprotein to either the proprotein or the mature form, suggesting inhibition of export from the cytoplasm to the periplasm, where cleavage of the signal sequence and the remainder of the N-terminal 441-amino-acid fragment occurs.

Figure 3.

Role of the Sec system in HMW1 secretion. Bacteria were metabolically labelled with [35S]-trans label, and HMW1 was immunoprecipitated with antiserum Gp1031-f, then resolved by SDS–PAGE and detected by autoradiography.

A. Lane 1, DH5α/pHMW1–15 without azide; lane 2, DH5α/pHMW1–15 with azide; lane 3, DH5α/pHMW1ΔB without azide; lane 4, DH5α/pHMW1ΔB with azide.

B. Lane 1, MC4100/pHMW1–15; lane 2, PR520 (secE cold sensitive)/pHMW1–15. In both A and B, the arrowhead indicates the HMW1 preprotein, the arrow indicates the HMW1 proprotein (lacking the 68 amino acid signal sequence) and the asterisk indicates the mature form of HMW1 (lacking the N-terminal 441 amino acids).

In interpreting this result, it is noteworthy that HMW1B has a typical prokaryotic signal sequence and is probably secreted by a Sec-dependent mechanism. To confirm that the association between azide treatment and accumulation of full-length HMW1 was independent of an effect on proper localization of HMW1B, we repeated the above experiment with DH5α/pHMW1ΔB (harbouring the hmw1 locus with an in-frame deletion in hmw1B;St. Geme and Grass, 1998). As shown in Fig. 3A, in the absence of azide, the HMW1 proprotein and negligible quantities of the HMW1 mature form were detected, analogous to observations with DH5α/pHMW1–15, except for the relative quantities of the mature form. Furthermore, treatment with azide again resulted in accumulation of full-length HMW1.

To obtain additional evidence for involvement of the Sec system, we introduced pHMW1–15 into strain MC4100 and a MC4100 derivative containing a cold-sensitive mutation in secE (PR520; secEcsE501;Riggs et al., 1988). The resulting transformants were inoculated into minimal medium and incubated at 37°C to an absorbency at 600 nm of 0.3, then shifted to 23°C for another 5 h, radiolabelled with [35S]-trans label for 30 min and immunoprecipitated. As shown in Fig. 3B, in strain MC4100/pHMW1–15, both the preprotein and the mature protein were present in roughly equal quantities. In contrast, in PR520/pHMW1–15, there was a relative abundance of the preprotein, again suggesting inhibition of export from the cytoplasm.

To extend these results and assess involvement of leader peptidase more directly, we introduced pHMW1–15 into strain W3110 and an isogenic derivative containing a temperature-sensitive mutation in lep (strain IT41; Inada et al., 1989). The resulting transformants were incubated at 30°C, then shifted to 42°C for 5 min, then radiolabelled with [35S]-trans label for 2 min and immunoprecipitated. In strain W3110/pHMW1–15, the HMW1 preprotein, the HMW1 proprotein lacking the signal sequence and the HMW1 mature form were all detectable. In contrast, in strain IT41/pHMW1–15, the HMW1 preprotein was present but the proprotein and the mature form were virtually absent (data not shown).

Together, these results indicate that HMW1 is exported from the cytoplasm by a Sec-dependent mechanism and suggest that HMW1 interacts directly with the Sec machinery and with leader peptidase.

Deletion of the HMW1 region between amino acids 69–441 results in premature periplasmic degradation.

To begin to define the role of amino acids 69–441, we examined the effect of deletion of this segment. In an effort to avoid interference with leader peptidase activity and release of HMW1 from the inner membrane, our strategy involved deletion of amino acids 72–441, thereby preserving the cleavage site between amino acids 68 and 69. As shown in Fig. 2, examination of whole-cell sonicates of DH5α expressing the mutant protein revealed a 125 kDa band corresponding to mature HMW1 and a faint, slightly larger band corresponding to mature HMW1 plus the 68-amino-acid signal sequence. Examination of periplasmic extracts and secreted proteins revealed very small amounts of mature HMW1 (Fig. 2) and adherence assays demonstrated no HMW1-mediated adherence (Table 1). Consistent with these observations, Western analysis of membrane fractions revealed significant amounts of mature HMW1 in the inner membrane (data not shown).

These results suggest that mutant HMW1 lacking most of the pro-piece between amino acids 69 and 441 is exported from the cytoplasm, then cleaved between amino acids 68 and 69, released from the inner membrane and degraded in the periplasm.

The NTNG motif in the N-terminal extremity of HMW1 influences surface localization and adhesive activity

The N-terminal region of HMW1 shares sequence similarity with three related haemolysins called ShlA (expressed by Serratia marcescens), HpmA (expressed by Proteus mirabilis), and Hhd (expressed by H. ducreyi). In addition, there is sequence similarity with Bordetella pertussis filamentous haemagglutinin (FHA) and with the H. influenzae HxuA protein, which binds haem-haemopexin. In all six proteins, the region of homology contains the sequence NPNG(I/M). In ShlA and FHA, site-directed mutagenesis suggests that the first asparagine in this sequence is critical for productive interaction with the cognate transporter protein present in the outer membrane (Schonherr et al., 1993; Jacob-Dubuisson et al., 1997). In previous work, we demonstrated that the same appears to be true with HMW1. In particular, we found that mutation of NPNGI (residues 150–154) to IAIGI resulted in loss of processing and extracellular release of HMW1 (St. Geme and Grass, 1998). Beyond the NPNG(I/M) motif, ShlA, HpmA, HhdA, FHA and HxuA also contain an upstream region defined by the sequence N(S/P)(N/H)L. In ShlA, replacement of the first asparagine in this motif (N109) abolishes secretion and activation (Schonherr et al., 1993). Mutation of the corresponding asparagine in FHA (N137) has a similar effect (Jacob-Dubuisson et al., 1997). Interestingly, HMW1 lacks an N(S/P)(N/H)L motif but contains the sequence NTNG at residues 163–166, just downstream of the NPNGI. To address the possibility that this region plays a role in HMW1 secretion, we converted NTNG to ITIG and examined the effect of this mutation by itself and in conjunction with mutation of the NPNGI. As shown in Fig. 4 and Table 1, mutation of the NTNG alone had no effect on processing or secretion or on HMW1-mediated in vitro adherence. Mutation of both the NTNG and the NPNGI eliminated processing, surface localization and extracellular release of HMW1 and abolished HMW1-mediated adherence. Of note, the phenotype associated with mutation of both the NTNG and NPNGI was identical to that associated with an in-frame deletion in hmw1B (Fig. 4, lane 6), suggesting that the NTNG and NPNGI sequences or the region between amino acids 150 and 166 may be a point of contact with HMW1B.

Figure 4.

Mutation of the HMW1 NPNGI and NTNG motifs. Cell sonicates were prepared, periplasmic proteins were extracted and culture supernatants were precipitated from DH5α/pT7–7, DH5α/pHMW1–15, DH5α/pHMW1(A-IAIGI) (a derivative of pHMW1–14 with mutation of the NPNGI motif in HMW1), DH5α/pHMW1(A-ITIG) (a derivative of pHMW1–15 with mutation of the NTNG motif in HMW1), DH5α/pHMW1(A-IAIGI/ITIG) (a derivative of pHMW1–15 with mutation of both the NPNGI and the NTNG motifs in HMW1) and DH5α/pHMW1ΔB (a derivative of pHMW1–15 with an in-frame deletion in hmw1B), and Western immunoblot was performed with guinea pig antiserum Gp1031-f, directed against HMW1. In addition, whole cells were fixed with paraformaldehyde and a dot immunoblot was performed, again with guinea pig antiserum Gp1031-f. Cell sonicates are shown in A, periplasmic proteins are shown in B, culture supernatants are shown in C and dot immunoblots are shown in D. In all panels, samples were loaded as follows: lane 1, DH5α/pT7–7; lane 2, DH5α/pHMW1–15; lane 3, DH5α/pHMW1(A-IAIGI); lane 4, DH5α/pHMW1(A-ITIG); lane 5, DH5α/pHMW1(A-IAIGI/ITIG); lane 6, DH5α/pHMW1ΔB. Arrows indicate the mature form of HMW1.

Deletion of the HMW1 C-terminus eliminates surface association

To begin to define the function of the HMW1 C-terminus, we examined the effect of deletion of this region. Exploiting the ClaI site at the 3′ end of the hmw1A gene, we generated the plasmid pHMW1A-Δ3′(414), which contains hmw1A lacking the 3′ end of the gene, resulting in deletion of the C-terminal 414 amino acids. This plasmid was then coexpressed with pHMW1BC (encoding hmw1B and hmw1C) and cells were fractionated. Examination of these fractions by Western immunoblot revealed a negligible amount of truncated HMW1 in the periplasm but abundant HMW1 in the culture supernatant (Fig. 5). To extend this observation, we performed a whole-cell immunoblot, which demonstrated virtually no surface-associated HMW1 in DH5α/pHMW1A-Δ3′(414) + pHMW1BC compared with DH5α/pHMW1–15 (not shown). Consistent with this finding, DH5α/pHMW1A-Δ3′(414) + pHMW1BC was non-adherent in assays with Chang epithelial cells (Table 1). As shown in Fig. 5 and Table 1, deletion of the C-terminal 132 amino acids of HMW1 reproduced the effect observed with deletion of the C-terminal 414 amino acids.

Figure 5.

Deletion of the HMW1 C-terminus. Cell sonicates were prepared, periplasmic proteins were extracted and culture supernatants were precipitated from DH5α/pT7–7, DH5α/pHMW1–15, DH5α/pHMW1 A-Δ3′(414) + pHMW1BC, and DH5α/pHMW1A-Δ3′(132) + pHMW1BC, and Western blot was performed with guinea pig antiserum Gp1031-f, directed against HMW1. Cell sonicates are shown in A, periplasmic proteins are shown in B, and culture supernatants are shown in C. In all panels, samples were loaded as follows: lane 1, DH5α/pT7–7; lane 2, DH5α/pHMW1–15; lane 3, DH5α/pHMW1 A-Δ3′(414) + pHMW1BC; lane 4, DH5α/pHMW1A-Δ3′(132) + pHMW1BC. Arrows indicate the mature form of HMW1 from wildtype constructs and dots indicate truncated forms of HMW1.

Together these experiments suggest that the HMW1 C-terminus may be important in tethering the HMW1 adhesin to the surface of the organism.

Discussion

In this study, we examined the mechanism of maturation and secretion of the non-typable H. influenzae HMW1 adhesin. As summarized in Fig. 6, our results indicate that HMW1 is synthesized as a preproprotein that undergoes an initial cleavage event between amino acids 68 and 69, via leader peptidase, and a second cleavage event between residues 441 and 442. Subsequently, the protein is translocated across the outer membrane in a process that depends on amino acids 150–166 and the HMW1B outer membrane translocator. Most of the secreted protein remains tethered to the surface of the organism, dependent on the C-terminal region.

Figure 6.

Model of secretion of HMW1. As suggested from our observations, HMW1 is synthesized as a preproprotein. The 68 amino acid N-terminal fragment (indicated by a solid black bar) directs the protein to the Sec translocase, allowing for export from the cytoplasm. Following cleavage between amino acids 68 and 69, the protein is released into the periplasm. The segment from amino acids 69–441 (indicated by a hatched bar) serves as an intramolecular chaperone, preventing periplasmic degradation and mediating interaction with the HMW1B outer membrane translocator (shown as a cylinder with the designation ‘B’). With HMW1 on the periplasmic face of the outer membrane, cleavage between amino acids 441 and 442 occurs and the protein is then transported to the surface of the cell via HMW1B. We hypothesize that the C-terminus of HMW1 facilitates reversible tethering to the bacterial surface, with eventual extracellular release of at least small quantities of protein.

Incubation of organisms in the presence of a low concentration of azide was associated with a block in HMW1 export from the cytoplasm, suggesting involvement of the SecA ATPase. Analysis of HMW1 secretion in a cold-sensitive SecE mutant also revealed a block in export from the cytoplasm, providing further evidence for interaction with the Sec apparatus. Consistent with these observations, determination of N-terminal amino acid sequence of surface-associated HMW1 with a mutated 441–442 cleavage site demonstrated that the initial processing event occurs at a predicted leader peptidase cleavage site (after residue 68). Together these results support the conclusion that HMW1 is secreted by a Sec-dependent pathway.

Given our evidence that HMW1 secretion involves the Sec system, it is interesting that the HMW1 signal sequence is atypical, with a long N-terminal extension characterized by multiple charged residues. A similar extension is present in a number of other high-molecular-weight, extracellular Gram-negative bacterial proteins, including B. pertussis FHA, the H. influenzae Hia and Hsf adhesins (Barenkamp and St. Geme, 1996; St. Geme et al., 1996), the diarrhoeagenic E. coli AIDA-I adhesin (Suhr et al., 1996) and the Shigella flexneri SepA protease (Benjelloun-Touimi et al., 1995), among others (Henderson et al., 1998). Of these proteins, all but FHA appear to belong to the autotransporter family, and all are presumed to be secreted by a Sec-dependent mechanism (Henderson et al., 1998). Experiments with FHA suggest that the N-terminal extension is not essential for export (Lambert-Buisine et al., 1998). Nevertheless, one hypothesis that remains tenable is that this domain functions as an intramolecular cytoplasmic chaperone, preventing premature folding and facilitating export across the cytoplasmic membrane. Alternatively, this domain may interact with a cytoplasmic chaperone. In order to establish either of these possibilities, examination of the kinetics of export may be necessary.

In earlier work, we found that site-directed mutation of the HMW1 NPNGI motif at amino acids 150–154 resulted in loss of processing and extracellular release (St. Geme and Grass, 1998). In the present study, we extended this analysis and investigated the potential role of the NTNG motif at amino acids 163–166. Interestingly, mutation of this region alone had no effect. However, when this region was altered in the context of a mutated NPNGI motif, we were unable to detect HMW1 beyond the cytoplasm, analogous to the phenotype associated with elimination of the HMW1B protein. The suggestion from these results is that the NPNGI and NTNG motifs interact with HMW1B and allow for translocation across the outer membrane; in the absence of productive interaction, HMW1 presumably accumulates in the periplasm and is then degraded. Interestingly, comparison of DH5α expressing HMW1 with mutation of the NPNGI motif and DH5α expressing HMW1 with mutation of both the NPNGI and the NTNG motifs revealed distinguishable phenotypes. In particular, alteration of the NPNGI motif alone resulted in a block in processing between amino acids 441 and 442 and elimination of extracellular release, while concurrent disruption of the NPNGI and the NTNG motifs also abolished surface localization. Together these observations suggest that the interaction between HMW1 and HMW1B consists of multiple discrete steps, including processing, translocation across the outer membrane and extracellular release.

At this point, the mechanism by which HMW1B promotes processing and secretion of HMW1 remains unclear. Of note, HMW1B shows significant sequence similarity with several outer membrane ushers involved in the biogenesis of Gram-negative bacterial pili (Clouthier et al., 1993; Watson et al., 1994; St. Geme and Grass, 1998). Based on recent work with the PapC usher required for assembly of E. coli P pili, ushers are believed to form multimeric outer membrane pores that allow for ordered assembly of pilus subunits into growing pili (Thanassi et al., 1998). Analysis of the secondary structure of HMW1B predicts formation of a β-barrel with a central hydrophilic core (J. W. St. Geme and S. Grass, unpublished data), suggesting the possibility that this protein forms a pore for surface localization of the HMW1 adhesin. Characterization of the function of HMW1B will probably provide insights into the related outer membrane proteins called FhaC (expressed by B. pertussis) (Willems et al., 1994), ShlB (expressed by S. marcescens) (Poole et al., 1988), HpmB (expressed by P. mirabilis) (Uphoff and Welch, 1990), HhdB (expressed by H. ducreyi) (Palmer and Munson, 1995) and HxuB (expressed by H. influenzae) (Cope et al., 1995), all of which are involved in secretion of cognate proteins that share amino-terminal sequence similarity with HMW1.

In considering the function of the region of HMW1 between amino acids 69 and 441, it is noteworthy that deletion of this region resulted in reduced HMW1 in the periplasm and culture supernatant, similar to findings when the N-terminal signal sequence was removed. One possibility is that this region is required for export from the cytoplasm. However, close examination of Western blots revealed evidence of cleavage of the 68-amino-acid signal sequence, suggesting export from the cytoplasm and then degradation in the periplasm. With this information in mind, a more attractive hypothesis is that the pro-piece functions as an intramolecular chaperone, facilitating periplasmic folding of the mature protein and thus protecting against premature degradation by periplasmic proteases. Interestingly, when the cleavage site between amino acids 441 and 442 was mutated, blocking processing and maturation, the protein was still translocated to the bacterial surface and capable of promoting adherence in vitro. Given that cleavage is not essential for either secretion or adhesive activity, the purpose of this event remains unclear.

Deletion of as few as 132 amino acids from the HMW1 C-terminus resulted in very efficient secretion and extracellular release, suggesting that the C-terminal end of the adhesin serves to impede transit through the periplasm and dissociation from the surface of the organism (influencing the kinetics of secretion and extracellular release). It is possible that the C-terminus engages HMW1B, thus tethering the mature protein to the outer membrane. Given that low concentrations of salt and EDTA are sufficient to purify mature HMW1 from the surface of whole bacteria, it is likely that tethering involves non-covalent interactions, such as electrostatic interactions, hydrogen bonds or van der Waals forces. If future analysis demonstrates that the C-terminal truncated protein retains full adhesive activity, the deletion construct may be ideally suited for purifying HMW1 for the purpose of vaccine preparation.

In considering our results, it important to recognize that they were generated by expressing the hmw1 gene cluster in E. coli, raising the possibility that the details of HMW1 processing and secretion may differ in the native H. influenzae background. However, all indicators point to the conclusion that processing and secretion occur identically in E. coli and H. influenzae. In particular, in both backgrounds, secretion is a two-step process with a stable periplasmic intermediate, maturation involves periplasmic cleavage of an N-terminal 441-amino-acid fragment and the mature protein is localized on the bacterial surface and then released in small amounts into the culture supernatant ( St. Geme and Grass, 1998). Furthermore, regarding involvement of the Sec system, the sec genes are known to be present in the H. influenzae genome (Fleischmann et al., 1995).

In summary, our results provide strong evidence that the HMW1 adhesion is secreted by a Sec-dependent, type II pathway with a terminal branch composed simply of an outer membrane translocator, perhaps accompanied by an intramolecular chaperone. The unusual N-terminus suggests evolutionary relatedness to the autotransporter family of proteins. Continued investigation of HMW1 secretion should advance our understanding of pathways for protein secretion in Gram-negative bacteria and facilitate efforts to develop a vaccine effective against non-typable H. influenzae.

Experimental procedures

Strains and plasmids

Bacterial strains and plasmids are listed in Table 3. E. coli DH5α is a non-adherent laboratory strain that has been described previously (Sambrook et al., 1989). E. coli CC118 contains a deletion of the phoA gene and lacks alkaline phosphatase activity (Manoil and Beckwith, 1985). This strain was a generous gift of V. Miller (Washington University). E. coli strains MC4100 and PR520 have been described previously (Riggs et al., 1988) and were provided by J. Beckwith (Harvard Medical School). E. coli strain W3110 and IT41 have been described previously as well (Inada et al., 1989) and were provided by R. Taylor (Dartmouth Medical School).

Table 3. Bacterial strains.
Strain/plasmidRelevant genotype/descriptionReference or source
Strain (E. coli)
DH5αφ80 dLacZΔM15ΔlacUdeoRrecAendA1Life Technologies
CC118ΔlacΔphoA Manoil and Beckwith (1985)
MC4100FaraDΔ(lac)U169relArpsL Riggs et al. (1988)
PR520MC4100secEcsE501argE::Tn10 Riggs et al. (1988)
W3110FmcrAmcrBIN(rrnD-rrnE)1 l Sambrook et al. (1989)
IT41W3110lep9 (am) TetR Inada et al. (1989)
Plasmid
pHMW1–15pT7–7 with hmw1 gene cluster St. Geme et al. (1993)
pHMW1–14
Same as pHMW1–15 with an additional 1.5
kb of upstream sequence
Barenkamp and Leininger (1992)
pT7–7Cloning vector, AmpR Tabor and Richardson (1985)
pMMB67Cloning vector, AmpR Furste et al. (1986)
pACYC184Cloning vector, TetRCamRChang and Cohen (1978)
pCH39pKT279 containing β-lactamase signal Hoffman and Wright (1985)
sequence fused to PhoA 
pT7-HMW1ApT7–7 containing hmw1 A gene as aThis study
XbaI–HgiAI fragment 
pHMW1ApACYC containing hmw1 A gene as a St. Geme and Grass (1998)
SpeI–SalI fragment 
pHMW1BCpMMB67 with hmw1B and hmw1C St. Geme and Grass (1998)
pPHO7Derivative of pTZ18R containing phoA Gutierrez and Devedjian (1989)
flanked on both sides by multiple cloning sites 
pT7-PhoA
Derivative of pT7–7 containing BamHI–PstI
fragment with phoA from pPHO7
This study
pPhoA68
Derivative of pT7-PhoA with first 68 amino
acids of HMW1 fused in-frame to PhoA
This study
pPhoA146
Derivative of pT7-PhoA with first 146
amino acids of HMW1 fused in-frame to PhoA
This study
pPhoA285
Derivative of pT7-PhoA with first 285
amino acids of HMW1 fused in-frame to PhoA
This study
pPhoA383
Derivative of pT7-PhoA with first 383
amino acids of HMW1 fused in-frame to PhoA
This study
pPhoA441
Derivative of pT7-PhoA with first 441
amino acids of HMW1 fused in-frame to PhoA
This study
pHMW(A-Δ2–441)
Derivative of pHMW1–15 with deletion of
sequence encoding amino acids 2–441 in HMW1
This study
pHMW1(A-Δ2–68)
Derivative of pHMW1–15 with deletion of
sequence encoding amino acids 2–68 in HMW1
This study
pHMW(A-Δ72–441)
Derivative of pHMW1–15 with deletion of
sequence encoding amino acids 72–441 in HMW1
This study
pHMW1(A-IAIGI)Derivative of pHMW1–14 with mutation of St. Geme and Grass (1998)
NPNGI at positions 150–154 in HMW1 
pHMW1(A-ITIG)
Derivative of pHMW1–15 with mutation of
NTNG at positions 163–166 in HMW1
This study
pHMW1(A-IAIGI/ITIG)
Derivative of pHMW1–15 with mutation of
NPNGI at positions 150–154 and NTNG at positions
163–166 and in HMW1
This study
pHMW1ΔBDerivative of pHMW1–15 with in-frame St. Geme and Grass (1998)
deletion in hmw1B 
pHMW1(A-DPD/GPG)Derivative of pHMW1–15 with mutation ofThis study
DPD at positions 441–443 in HMW1,
abolishing the cleavage site
 
pHMW1A-Δ3′(414)
Derivative of pACYC184 containing
hmw1 A with deletion of the 3′ end
beginning at the ClaI site
This study
pHMW1 A-Δ3′(132)
Derivative of pACYC184 containing
hmw1 A with deletion of the 3′ end
beginning at the NruI site
This study

Culture conditions

E. coli strains were grown on Luria–Bertani (LB) agar or in LB broth and were stored at −80°C in LB broth with 50% glycerol. Antibiotic concentrations used to select for plasmids included 100 µg ml−1 ampicillin, 50 µg ml−1 kanamycin and 30 µg ml−1 chloramphenicol.

Plasmid construction

To generate PhoA fusion constructs, the phoA gene from pPHO7 was excised as a BamHI–PstI fragment and inserted into pT7-7 digested with BamHI and PstI, creating pT7–PhoA. Fragments containing the hmw1A promoter and coding sequence for increasing segments of the HMW1 N-terminus were amplified from pHMW1–15 by polymerase chain reaction (PCR) and inserted into the BamHI site in pT7–phoA. Restriction mapping was performed to identify recombinant plasmids with inserts in the proper orientation and in-frame with the phoA gene.

The plasmids pHMW1(A-Δ2–68) and pHMW(A-Δ72–441) contain the hmw1 gene cluster with deletions of the N-terminus of HMW1, and pHMW1(A-DPD/GPG), pHMW1(A-ITIG), and pHMW1(A-IAIGI/ITIG) contain the hmw1 gene cluster with point mutations in the N-terminus of HMW1. pHMW1(A-Δ2–68), pHMW1(A-ITIG), and pHMW1(A-IAIGI/ITIG) were constructed using recombinant PCR to generate a 1.6 kb XbaI–BamHI fragment harbouring the relevant sequence. The resulting fragment was digested with XbaI and BamHI, then ligated in place of the wild-type SpeI–BamHI fragment in pHMW1–14. pHMW(A-Δ72–441) was constructed using recombinant PCR to generate a 0.6 kb XbaI–BamHI fragment with the relevant sequence. This fragment was digested with XbaI and BamHI and ligated in place of the XbaI–BamHI fragment in pHMW1–15. pHMW1(A-DPD/GPG) was constructed using recombinant PCR to generate a 2.5 kb MfeI fragment with the intended mutations, which was inserted in place of the wild-type MfeI fragment in pT7-HMW1A. The resulting plasmid was digested with BglII and BamHI to generate a 5 kb fragment, which was inserted in place of the wild-type BglII–BamHI fragment in pHMW1–14. The plasmid pHMW1A-Δ3′(414) contains the hmw1 A gene with ≈ 1.2 kb deleted from the 3′ end. It was constructed by digesting pT7-HMW1A with ClaI to remove the hmw1A 3′ end, then religating. The resulting plasmid was digested with BglII and ClaI, generating a 4 kb fragment, which was purified and ligated into BamHI–ClaI-digested pACYC184. pHMW1A-Δ3′(132) contains the hmw1A gene with ≈ 0.4 kb removed from the 3′ end. It was constructed by digesting pHMW1A with NruI and HindIII, then blunt-ending and religating.

With all mutant constructs, the presence of the desired mutations and the integrity of surrounding DNA were confirmed by nucleotide sequencing.

Recombinant DNA methods

DNA ligations, restriction endonuclease digestions and gel electrophoresis were performed according to standard techniques (Sambrook et al., 1989). Plasmids were introduced into E. coli by electroporation (Dower et al., 1988) or the rubidium chloride method of transformation (Kushner, 1978).

Cell fractionation and protein analysis

Whole cell sonicates were prepared by resuspending bacterial pellets in 10 mM HEPES, pH 7.4 and sonicating to clarity. Periplasmic proteins were extracted using the osmotic shock method as described by Slonim et al. (1992). Proteins released from the surface of the bacterium were recovered by precipitating culture supernatants with 10% trichloracetic acid (Barenkamp and St. Geme, 1994). To collect culture supernatants, bacteria were pelleted by ultracentrifugation at 100 000 g for 60 min. The purity of fractions obtained using these techniques was established in earlier work (St. Geme and Grass, 1998).

Proteins were resolved by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE) using 7.5% polyacrylamide gels (Laemmli, 1970). Western blots were performed with guinea pig polyclonal antiserum Gp1031-f, which is directed against HMW1 and was used in a dilution of 1:1000.

Radiolabelling and immunoprecipitation

To determine whether the Sec system is involved in export of HMW1, strain DH5α/pHMW1–15 was inoculated into LB broth and incubated for Å 8 h. Subsequently, the culture was diluted in minimal medium containing all 20 essential amino acids and incubated overnight. The culture was diluted again, this time in minimal medium containing all amino acids except methionine and cysteine, and then grown to an absorbence at 600 nm of Å 0.5. Sodium azide was added to achieve a final concentration of 2 mM and the culture was incubated for another 15 min. Finally, [35S]-trans label was added and labelling was allowed to proceed for 45 min. Labelling was quenched with cold methionine and cysteine, and samples were prepared for immunoprecipitation.

To examine the influence of SecE on export of HMW1, strains MC4100/pHMW1–15 and PR520/pHMW1–15 were inoculated in LB broth and then minimal medium as described above. Ultimately dilutions were performed in minimal medium containing all 20 essential amino acids except methionine and cysteine. Cultures were grown to an absorbence at 600 nm of Å 0.3, then shifted to a temperature of 23°C and incubated for another 5 h. Two millilitre samples were removed and labelled with 100 µCi of [35S]-trans label for 30 min and were then prepared for immunoprecipitation.

To examine the role of leader peptidase in HMW1 secretion, strains W3110/pHWM1–15 and IT41/pHMW1–15 were incubated, as described above, at 30°C. Subsequently they were diluted in minimal medium containing all amino acids except methionine and cysteine and incubated at 30°C to an absorbence at 600 nm of Å 0.6. Subsequently, they were shifted to 42°C and incubated for another 5 min. Next, 2 ml samples were removed and labelled with 100 µCi of [35S]-trans label for 2 min and were then prepared for immunoprecipitation.

For immunoprecipitation, samples were centrifuged and pellets were solubilized in 25 µl of 1% SDS, 10 mM Tris, pH 7.5, 1 mM EDTA. After heating at 95°C for 3 min, mixtures were diluted with 200 µl of 2% Triton X-100, 50 mM Tris, pH 7.5, 0.25 M NaCl and centrifuged in a microfuge for 5 min. Supernatants were transferred to fresh tubes, 3 µl of guinea pig serum Gp1031-f was added to each tube and mixtures were incubated overnight at 4°C. The following day, 50 µl of ImmunoPure immobilized recombinant protein A (Pierce) was added to each sample and mixtures were incubated at room temperature for 2 h. Beads were pelleted by centrifuging in a microfuge at 4000 r.p.m. for 2 min and then washed five times with 2% Triton X-100, 50 mM Tris, pH 7.5, 0.25 M NaCl. Ultimately samples were heated at 95°C for 5 min, and proteins were resolved by SDS–PAGE. Samples were detected by autoradiography.

Whole-cell (dot) immunoblots

For whole-cell immunoblots, bacteria were grown to an optical density at 600 nm of 0.5, then washed twice with PBS, fixed for 20 min at room temperature with 4% paraformaldehyde in PBS, and resuspended in PBS to an optical density at 600 nm of 0.6. Aliquots of 100 µl were applied to nitrocellulose filters using a dot-blot manifold apparatus (Schleicher and Schuell). Samples were incubated for 30 min and then pulled through the filter by vacuum suction. Following blocking for 1 h with 0.5% blocking agent (Boehringer Mannheim) in Tris-buffered saline, surface-exposed HMW1 was detected using guinea pig antiserum Gp1031-f.

Alkaline phosphatase assays

Alkaline phosphatase activity was measured as described by Brickman and Beckwith (1975). Bacteria were incubated in 3 ml of LB broth to an optical density at 600 nm of 0.5, then pelleted and resuspended in 0.3 ml of 1 M Tris, pH 8. A 10 µl volume of each suspension was mixed with 100 µl of 0.4% paranitrophenyl phosphate (Sigma 104) in flat-bottom 96-well plates. Reactions were allowed to proceed at room temperature for 10 min and were stopped by the addition of an equal volume of 1 M K2HPO4. Enzymatic activity was quantitated spectrophotometrically at 420 nm and standardized according to the density of the bacterial suspension as measured at an optical density of 590 nm. With all strains, production of a PhoA fusion protein was confirmed by Western analysis using a rabbit polyclonal antiserum raised against PhoA (3-Prime, 5-Prime, Boulder, CO, USA).

Determination of N-terminal sequence

HMW1 was purified from the surface of DH5α/pHMW1(A-DPD/GPG) as previously described (Barenkamp, 1996), then separated on a 7.5% SDS–polyacrylamide gel and electrotransferred to a polyvinylidene membrane. After staining with Coomassie brilliant blue R-250, the Å 160 kDa protein was excised from the membrane and submitted to Midwest Analytical. Amino-terminal sequence determination was performed by automated Edman degradation using a Perkin-Elmer Biosystems model 477 A sequencing system.

Adherence assays

Adherence assays were performed with Chang epithelial cells [Wong-Kilbourne derivative, clone 1-5c-4 (human conjunctiva); ATCC CCL20.2] as described previously (St. Geme et al., 1993; St. Geme et al., 1996). Percentage adherence was calculated by dividing the number of adherent colony-forming units per monolayer by the number of inoculated colony-forming units.

Acknowledgements

We thank S. Loosmore and Y.-P. Yang for providing guinea pig antiserum Gp1031-f, V. Miller for providing strain CC118, J. Beckwith and C. Lee for providing strains MC4100 and PR520, and R. Taylor for providing strains W3110 and IT41. This work was supported by Public Health Service Grant 1RO1 DC-02873 from the National Institute on Deafness and Other Communication Disorders awarded to J.W.S. and funding from Connaught Laboratories.

Ancillary