Solution structure of the lipoyl domain of the chimeric dihydrolipoyl dehydrogenase P64K from Neisseria meningitidis.

The antigenic P64K protein from the pathogenic bacterium Neisseria meningitidis is found in the outer membrane of the cell, and consists of two parts: an 81-residue N-terminal region and a 482-residue C-terminal region. The amino-acid sequence of the N-terminal region is homologous with the lipoyl domains of the dihydrolipoyl acyltransferase (E2) components, and that of the C-terminal region with the dihydrolipoyl dehydrogenase (E3) components, of 2-oxo acid dehydrogenase multienzyme complexes. The two parts are separated by a long linker region, similar to the linker regions in the E2 chains of 2-oxo acid dehydrogenase complexes, and it is likely this region is conformationally flexible. A subgene encoding the P64K lipoyl domain was created and over-expressed in Escherichia coli. The product was capable of post-translational modification by the lipoate protein ligase but not aberrant modification by the biotin protein ligase of E. coli. The solution structure of the apo-domain was determined by means of heteronuclear NMR spectroscopy and found to be a flattened beta barrel composed of two four-stranded antiparallel beta sheets. The lysine residue that becomes lipoylated is in an exposed beta turn that, from a [1H]-15N heteronuclear Overhauser effect experiment, appears to enjoy substantial local motion. This structure of a lipoyl domain derived from a dihydrolipoyl dehydrogenase resembles that of lipoyl domains normally found as part of the dihydrolipoyl acyltransferase component of 2-oxo acid dehydrogenase complexes and will assist in furthering the understanding of its function in a multienzyme complex and in the membrane-bound P64K protein itself.

Structure of a malaria parasite antigenic determinant displayed on filamentous bacteriophage determined by NMR spectroscopy: implications for the structure of continuous peptide epitopes of proteins.

The NANP repeating sequence of the circumsporozoite protein of Plasmodium falciparum was displayed on the surface of fd filamentous bacteriophage as a 12-residue insert (NANP)(3) in the N-terminal region of the major coat protein (pVIII). The structure of the epitope determined by multidimensional solution NMR spectroscopy of the modified pVIII protein in lipid micelles was shown to be a twofold repeat of an extended and non-hydrogen-bonded loop based on the sequence NPNA, demonstrating that the repeating sequence is NPNA, not NANP. Further, high resolution solid-state NMR spectra of intact hybrid virions containing the modified pVIII proteins demonstrate that the peptides displayed on the surface of the virion adopt a single, stable conformation; this is consistent with their pronounced immunogenicity as well as their ability to mimic the antigenicity of their native parent proteins.

The protein capsid of filamentous bacteriophage PH75 from Thermus thermophilus.

The PH75 strain of filamentous bacteriophage (Inovirus) grows in the thermophilic bacterium Thermus thermophilus at 70 degrees C. We have characterized the viral DNA and determined the amino acid sequence of the major coat protein, p8. The p8 protein is synthesized without a leader sequence, like that of bacteriophage Pf3 but unlike that of bacteriophage Pf1, both of which grow in the mesophile Pseudomonas aeruginosa. X-ray diffraction patterns from ordered fibres of the PH75 virion are similar to those from bacteriophages Pf1 and Pf3, indicating that the protein capsid of the PH75 virion has the same helix symmetry and subunit shape, even though the primary structures of the major coat proteins are quite different and the virions assemble at very different temperatures. We have used this information to build a molecular model of the PH75 protein capsid based on that of Pf1, and refined the model by simulated annealing, using fibre diffraction data extending to 2.4 A resolution in the meridional direction and to 3.1 A resolution in the equatorial direction. The common design may reflect a fundamental motif of alpha-helix packing, although differences exist in the DNA packaging and in the means of insertion of the major coat protein of these filamentous bacteriophages into the membrane of the host bacterial cell. These may reflect differences in the assembly mechanisms of the virions.

Recognition of the lipoyl domain is the ultimate determinant of substrate channelling in the pyruvate dehydrogenase multienzyme complex.

Reductive acetylation of the lipoyl domain (E2plip) of the dihydrolipoyl acetyltransferase component of the pyruvate dehydrogenase multienzyme complex of Escherichia coli is catalysed specifically by its partner pyruvate decarboxylase (E1p), and no productive interaction occurs with the analogous 2-oxoglutarate decarboxylase (E1o) of the 2-oxoglutarate dehydrogenase complex. Residues in the lipoyl-lysine beta-turn region of the unlipoylated E2plip domain (E2plip(apo)) undergo significant changes in both chemical shift and transverse relaxation time (T(2)) in the presence of E1p but not E1o. Residue Gly11, in a prominent surface loop between beta-strands 1 and 2 in the E2plip domain, was also observed to undergo a significant change in chemical shift. Addition of pyruvate to the mixture of E2plip(apo) and E1p caused larger changes in chemical shift and the appearance of multiple cross-peaks for certain residues, suggesting that the domain was experiencing more than one type of interaction. Residues in both beta-strands 4 and 5, together with those in the prominent surface loop and the following beta-strand 2, appeared to be interacting with E1p, as did a small patch of residues centred around Glu31. The values of T(2) across the polypeptide chain backbone were also lower than in the presence of E1p alone, suggesting that E2plip(apo) binds more tightly after the addition of pyruvate. The lipoylated domain (E2plip(holo)) also exhibited significant changes in chemical shift and decreases in the overall T(2) relaxation times in the presence of E1p, the residues principally affected being restricted to the half of the domain that contains the lipoyl-lysine (Lys41) residue. In addition, small chemical shift changes and a general drop in T(2) times in the presence of E1o were observed, indicating that E2plip(holo) can interact, weakly but non-productively, with E1o. It is evident that recognition of the protein domain is the ultimate determinant of whether reductive acetylation of the lipoyl group occurs, and that this is ensured by a mosaic of interactions with the Elp.

Multiple display of peptides and proteins on a macromolecular scaffold derived from a multienzyme complex.

The acyltransferase components (E2) from the family of 2-oxo acid dehydrogenase multienzyme complexes form large protein scaffolds, to which multiple copies of peripheral enzymes bind tightly but non-covalently. Sixty copies of the E2 polypeptide from the pyruvate dehydrogenase multienzyme complex of Bacillus stearothermophilus assemble to form a pentagonal dodecahedral scaffold with icosahedral symmetry. This protein scaffold can be modified to present foreign peptides and proteins on its surface. We show that it is possible to display two epitopes (MAL1 and MAL2) from the circumsporozoite CS proteins of Plasmodium falciparum and Plasmodium berghei, respectively, and a green fluorescent protein (EGFP), on the E2 surface. Immunization with an E2 scaffold displaying the MAL1 epitope elicited MAL1-specific antibodies in rabbits. EGFP (25 kDa) displayed as an N-terminal fusion in each of the 60 copies of the E2 chain folded into its active form, as judged by its fluorescence and detection in localized foci in Escherichia coli cells in vivo. Simultaneous display of a hexahistidine affinity tag, the MAL1 epitope and the green fluorescent protein, all on the same E2 scaffold, could be achieved by reversible denaturation and reassembly of mixtures of appropriately modified E2 chains. This new methodology offers several important advantages over other current display technologies, not least in the size of insert that can be accommodated and the multiplicity of display that can be achieved.

Sites of limited proteolysis in the pyruvate decarboxylase component of the pyruvate dehydrogenase multienzyme complex of Bacillus stearothermophilus and their role in catalysis.

The E1 component (pyruvate decarboxylase) of the pyruvate dehydrogenase complex of Bacillus stearothermophilus is a heterotetramer (alpha2beta2) of E1alpha and E1beta polypeptide chains. The domain structure of the E1alpha and E1beta chains, and the protein-protein interactions involved in assembly, have been studied by means of limited proteolysis. It appears that there may be two conformers of E1alpha in the E1 heterotetramer, one being more susceptible to proteolysis than the other. A highly conserved region in E1alpha, part of a surface loop at the entrance to the active site, is the most susceptible to cleavage in E1 (alpha2beta2). As a result, the oxidative decarboxylation of pyruvate catalysed by E1 in the presence of dichlorophenol indophenol as an artificial electron acceptor is markedly enhanced, but the reductive acetylation of a free lipoyl domain is unchanged. The parameters of the interaction between cleaved E1 and the peripheral subunit-binding domain of the dihydrolipoyl acetyltransferase E2 component are identical to those of the wild-type E1. However, a pyruvate dehydrogenase complex assembled in vitro with cleaved E1p exhibits a markedly lower overall catalytic activity than that assembled with untreated E1. This implies that active site coupling between the E1 and E2 components has been impaired. This has important implications for the way in which a tethered lipoyl domain can interact with E1 in the assembled complex.

Heteronuclear NMR studies of the specificity of the post-translational modification of biotinyl domains by biotinyl protein ligase.

The lipoyl domains of 2-oxo acid dehydrogenase multienzyme complexes and the biotinyl domains of biotin-dependent enzymes have homologous structures, but the target lysine residue in each domain is correctly selected for posttranslational modification by lipoyl protein ligase and biotinyl protein ligase, respectively. We have applied two-dimensional heteronuclear NMR spectroscopy to investigate the interaction between the apo form of the biotinyl domain of the biotin carboxyl carrier protein of acetyl-CoA carboxylase and the biotinyl protein ligase (BPL) from Escherichia coli. Heteronuclear multiple quantum coherence NMR spectra of the 15N-labelled biotinyl domain were recorded in the presence and absence of the ligase and backbone amide 1H and 15N chemical shifts were evaluated. Small, but significant, changes in chemical shift were found in two regions, including the tight beta-turn that houses the lysine residue targetted for biotinylation, and the beta-strand 2 and the loop that precedes it in the domain. When compared with the three-dimensional structure, sequence alignments of other biotinyl and lipoyl domains, and mutagenesis data, these results give a clear indication of how the biotinyl domain is both recognised by BPL and distinguished from the structurally related lipoyl domain to ensure correct posttranslational modification.

Swinging arms and swinging domains in multifunctional enzymes: catalytic machines for multistep reactions.

Multistep chemical reactions are increasingly seen as important in a growing number of complex biotransformations. Covalently attached prosthetic groups or swinging arms, and their associated protein domains, are essential to the mechanisms of active-site coupling and substrate channeling in a number of the multifunctional enzyme systems responsible. The protein domains, for which the posttranslational machinery in the cell is highly specific, are crucially important, contributing to the processes of molecular recognition that define and protect the substrates and the catalytic intermediates. The domains have novel folds and move by virtue of conformationally flexible linker regions that tether them to other components of their respective multienzyme complexes. Structural and mechanistic imperatives are becoming apparent as the assembly pathways and the coupling of multistep reactions catalyzed by these dauntingly complex molecular machines are unraveled.

Phage display of peptide epitopes from HIV-1 elicits strong cytolytic responses.

Although much effort has been expended on evaluating recombinant proteins and synthetic peptides as immunogens, they have generally proved incapable of inducing an efficient cytotoxic T-cell (CTL) response. Filamentous bacteriophage fd can display multiple copies of foreign peptides in the N-terminal region of its major coat protein pVIII, 2,700 copies of which make up the virus capsid. Here we show that fd virions displaying peptide RT2 (ILKEPVHGV), corresponding to residues 309-317 of the reverse transcriptase (RTase) of HIV-1, are able to prime a CTL response specific for this HIV-1 epitope in human cell lines. Successful priming also requires a T-helper epitope, pep23 (KDSWTVNDIQKLVGK), corresponding to residues 249-263 of HIV-1 RTase. Supplying this by displaying it on either the same or a separate bacteriophage virion led to activation of antigen-specific CD4+ T cells. Likewise, HLA-A2 transgenic mice immunized with bacteriophage virions displaying peptide RT2 were shown to mount an effective, specific anti-HIV-RT2 CTL response. This unexpected ability to elicit a designated cytolytic T-cell response, in addition to a B-cell response, has important implications for access to the class I major histocompatibility complex (MHC) loading compartment and the development of recombinant vaccines.

Restricted motion of the lipoyl-lysine swinging arm in the pyruvate dehydrogenase complex of Escherichia coli.

The three lipoyl (E2plip) domains of the dihydrolipoyl acetyltransferase component of the pyruvate dehydrogenase (PDH) complex of Escherichia coli house the lipoyl-lysine side chain essential for active-site coupling and substrate channelling within the complex. The structure of the unlipoylated form of the innermost domain (E2plip(apo)) was determined by multidimensional NMR spectroscopy and found to resemble closely that of a nonfunctional hybrid domain determined previously [Green et al. (1995) J. Mol. Biol. 248, 328-343]. The domain comprises two four-stranded beta-sheets, with the target lysine residue residing at the tip of a type-I beta-turn in one of the sheets; the N- and C-termini lie close together at the opposite end of the molecule in the other beta-sheet. Measurement of (15)N NMR relaxation parameters and backbone hydrogen/deuterium (H/D) exchange rates reveals that the residues in and surrounding the lipoyl-lysine beta-turn in the E2plip(apo) form of the domain become less flexible after lipoylation of the lysine residue. This implies that the lipoyl-lysine side chain may not sample the full range of conformational space once thought. Moreover, reductive acetylation of the lipoylated domain (E2plip(holo) --> E2plip(redac)) was accompanied by large changes in chemical shift between the two forms, and multiple resonances were observed for several residues. This implies a change in conformation and the existence of multiple conformations of the domain on reductive acetylation, which may be important in stabilizing this catalytic intermediate.

Protein-protein interaction revealed by NMR T(2) relaxation experiments: the lipoyl domain and E1 component of the pyruvate dehydrogenase multienzyme complex of Bacillus stearothermophilus.

T(2) relaxation experiments in combination with chemical shift and site-directed mutagenesis data were used to identify sites involved in weak but specific protein-protein interactions in the pyruvate dehydrogenase multienzyme complex of Bacillus stearothermophilus. The pyruvate decarboxylase component, a heterotetramer E1(alpha(2)beta(2)), is responsible for the first committed and irreversible catalytic step. The accompanying reductive acetylation of the lipoyl group attached to the dihydrolipoyl acetyltransferase (E2) component involves weak, transient but specific interactions between E1 and the lipoyl domain of the E2 polypeptide chain. The interactions between the free lipoyl domain (9 kDa) and free E1alpha (41 kDa), E1beta (35 kDa) and intact E1alpha(2)beta(2) (152 kDa) components, all the products of genes or sub-genes over-expressed in Escherichia coli, were investigated using heteronuclear 2D NMR spectroscopy. The experiments were conducted with uniformly (15)N-labeled lipoyl domain and unlabeled E1 components. Major contact points on the lipoyl domain were identified from changes in the backbone (15)N spin-spin relaxation time in the presence and absence of E1(alpha(2)beta(2)) or its individual E1alpha or E1beta components. Although the E1alpha subunit houses the sequence motif associated with the essential cofactor, thiamin diphosphate, recognition of the lipoyl domain was distributed over sites in both E1alpha and E1beta. A single point mutation (N40A) on the lipoyl domain significantly reduces its ability to be reductively acetylated by the cognate E1. None the less, the N40A mutant domain appears to interact with E1 similarly to the wild-type domain. This suggests that the lipoyl group of the N40A lipoyl domain is not being presented to E1 in the correct orientation, owing perhaps to slight perturbations in the lipoyl domain structure, especially in the lipoyl-lysine beta-turn region, as indicated by chemical shift data. Interaction with E1 and subsequent reductive acetylation are not necessarily coupled.

Structural determinants of post-translational modification and catalytic specificity for the lipoyl domains of the pyruvate dehydrogenase multienzyme complex of Escherichia coli.

The lipoyl domains of the dihydrolipoyl acyltransferase (E2p, E2o) components of the pyruvate and 2-oxoglutarate dehydrogenase multienzyme complexes are specifically recognised by their cognate 2-oxo acid decarboxylase (E1p, E1o). A prominent surface loop links the first and second beta-strands in all lipoyl domains, close in space to the lipoyl-lysine beta-turn. This loop was subjected to various modifications by directed mutagenesis of a sub-gene encoding a lipoyl domain of Escherichia coli E2p. Deletion of the loop (four residues) rendered the domain incapable of reductive acetylation by E. coli E1p in the presence of pyruvate, but insertion of a new loop (six residues) corresponding to that in the E2o lipoyl domain partly restored this ability, albeit with a much lower rate. However, the modified domain remained unable to undergo reductive succinylation by E1o in the presence of 2-oxoglutarate. Additional exchange of the two residues on the C-terminal side of the loop (V14A, E15T) had no effect. Insertion of a different four-residue loop also restored a limited ability to undergo reductive acetylation, but still significantly less than that of the wild-type domain. Exchanging the residue on the N-terminal side of the lipoyl-lysine beta-turn in the E2p and E2o domains (G39T), both singly and in conjunction with the loop exchange, had no effect on the ability of the E2p domain to be reductively acetylated but did confer a slight increase in susceptibility to reductive succinylation. All mutant E2p domains, apart from that with the loop deletion (LD), were readily lipoylated in vitro by E. coli lipoate protein ligase A; the E2p LD mutant could be lipoylated only at a significantly lower rate. Likewise, this domain exhibited 1D and 2D NMR spectra characteristic of a partially folded protein, whereas the spectra of mutants with modified loops were similar to those of the wild-type domain. The surface loop is evidently important to the structural integrity of the domain and may help to stabilize the thioester bond linking the acyl group to the reduced lipoyl-lysine swinging arm as part of the catalytic mechanism. Recognition of the lipoyl domain by its partner E1 appears to be a complex process and not attributable to any single determinant on the domain.

Self-assembly and catalytic activity of the pyruvate dehydrogenase multienzyme complex from Bacillus stearothermophilus.

The pyruvate dehydrogenase multienzyme complex from Bacillus stearothermophilus was reconstituted in vitro from recombinant proteins derived from genes over-expressed in Escherichia coli. Titrations of the icosahedral (60-mer) dihydrolipoyl acetyltransferase (E2) core component with the pyruvate decarboxylase (E1, alpha2beta2) and dihydrolipoyl dehydrogenase (E3, alpha2) peripheral components indicated a variable composition defined predominantly by tight and mutually exclusive binding of E1 and E3 with the peripheral subunit-binding domain of each E2 chain. However, both analysis of the polypeptide chain ratios in complexes generated from various mixtures of E1 and E3, and displacement of E1 or E3 from E1-E2 or E3-E2 subcomplexes by E3 or E1, respectively, showed that the multienzyme complex does not behave as a simple competitive binding system. This implies the existence of secondary interactions between the E1 and E3 subunits and E2 that only become apparent on assembly. Exact geometrical distribution of E1 and E3 is unlikely and the results are best explained by preferential arrangements of E1 and E3 on the surface of the E2 core, superimposed on their mutually exclusive binding to the peripheral subunit-binding domain of the E2 chain. Correlation of the subunit composition with the overall catalytic activity of the enzyme complex confirmed the lack of any requirement for precise stoichiometry or strict geometric arrangement of the three catalytic sites and emphasized the crucial importance of the flexibility associated with the lipoyl domains and intramolecular acetyl group transfer in the mechanism of active-site coupling.

Structure and selectivity in post-translational modification: attaching the biotinyl-lysine and lipoyl-lysine swinging arms in multifunctional enzymes.

The post-translational attachment of biotin and lipoic acid to specific lysine residues displayed in protruding beta-turns in homologous biotinyl and lipoyl domains of their parent enzymes is catalysed by two different ligases. We have expressed in Escherichia coli a sub-gene encoding the biotinyl domain of E.coli acetyl-CoA carboxylase, and by a series of mutations converted the protein from the target for biotinylation to one for lipoylation, in vivo and in vitro. The biotinylating enzyme, biotinyl protein ligase (BPL), and the lipoylating enzyme, LplA, exhibited major differences in the recognition process. LplA accepted the highly conserved MKM motif that houses the target lysine residue in the biotinyl domain beta-turn, but was responsive to structural cues in the flanking beta-strands. BPL was much less sensitive to changes in these beta-strands, but could not biotinylate a lysine residue placed in the DKA motif characteristic of the lipoyl domain beta-turn. The presence of a further protruding thumb between the beta2 and beta3 strands in the wild-type biotinyl domain, which has no counterpart in the lipoyl domain, is sufficient to prevent aberrant lipoylation in E.coli. The structural basis of this discrimination contrasts with other forms of post-translational modification, where the sequence motif surrounding the target residue can be the principal determinant.

Solution structures of apo and holo biotinyl domains from acetyl coenzyme A carboxylase of Escherichia coli determined by triple-resonance nuclear magnetic resonance spectroscopy.

A subgene encoding the 87 C-terminal amino acids of the biotinyl carboxy carrier protein (BCCP) from the acetyl CoA carboxylase of Escherichia coli was overexpressed and the apoprotein biotinylated in vitro. The structures of both the apo and holo forms of the biotinyl domain were determined by means of multidimensional NMR spectroscopy. That of the holo domain was well-defined, except for the 10 N-terminal residues, which form part of the flexible linker between the biotinyl and subunit-binding domains of BCCP. In agreement with X-ray crystallographic studies [Athappilly, F. K., and Hendrickson, W. A. (1995) Structure 3, 1407-1419], the structure comprises a flattened beta-barrel composed of two four-stranded beta-sheets with a 2-fold axis of quasi-symmetry and the biotinyl-lysine residue displayed in an exposed beta-turn on the side of the protein opposite from the N- and C-terminal residues. The biotin group is immobilized on the protein surface, with the ureido ring held down by interactions with a protruding polypeptide "thumb" formed by residues 94-101. However, at the site of carboxylation, no evidence could be found in solution for the predicted hydrogen bond between the main chain O of Thr94 and the ureido HN1'. The structure of the apo domain is essentially identical, although the packing of side chains is more favorable in the holo domain, and this may be reflected in differences in the dynamics of the two forms. The thumb region appears to be lacking in almost all other biotinyl domain sequences, and it may be that the immobilization of the biotinyl-lysine residue in the biotinyl domain of BCCP is an unusual requirement, needed for the catalytic reaction of acetyl CoA carboxylase.

Recognition of HIV-derived B and T cell epitopes displayed on filamentous phages.

The amino acid sequence of HIV reverse transcriptase (RT) from residue 248 to residue 262 was expressed on the surface of filamentous phage fd, fused to the major coat protein gVIIIp. The chimeric phage was used to assess the ability of anti-RT (248-262) human T cell lines and clones to become activated by the phage-displayed peptide. The RT peptide displayed on phage was recognized by the T-cells and induced production of Abs. However, not all T cells raised against the synthetic RT (248-262) peptide could respond. Lack of recognition did not depend on differences in the ability of different APCs to present the phage, but was apparently determined by the TCR specificity. The results presented here may be relevant to the design of recombinant protein-based subunit vaccines.

Effects of temperature and Y21M mutation on conformational heterogeneity of the major coat protein (pVIII) of filamentous bacteriophage fd.

Solid-state NMR spectroscopy was used to analyze the conformational heterogeneity of the major coat protein (pVIII) of filamentous bacteriophage fd. Both one and two-dimensional solid-state NMR spectra of magnetically aligned samples of fd bacteriophage reveal that an increase in temperature and a single site substitution (Tyr21 to Met, Y21M) reduce the conformational heterogeneity observed throughout wild-type pVIII. The NMR results are consistent with previous studies indicating that conformational flexibility in the hinge-bend segment that links the amphipathic and hydrophobic helices in the membrane-bound form of the protein plays an essential role during phage assembly, which involves a major change in the tertiary, but not secondary, structure of the coat protein.

Principles of quasi-equivalence and Euclidean geometry govern the assembly of cubic and dodecahedral cores of pyruvate dehydrogenase complexes.

The pyruvate dehydrogenase multienzyme complex (Mr of 5-10 million) is assembled around a structural core formed of multiple copies of dihydrolipoyl acetyltransferase (E2p), which exhibits the shape of either a cube or a dodecahedron, depending on the source. The crystal structures of the 60-meric dihydrolipoyl acyltransferase cores of Bacillus stearothermophilus and Enterococcus faecalis pyruvate dehydrogenase complexes were determined and revealed a remarkably hollow dodecahedron with an outer diameter of approximately 237 A, 12 large openings of approximately 52 A diameter across the fivefold axes, and an inner cavity with a diameter of approximately 118 A. Comparison of cubic and dodecahedral E2p assemblies shows that combining the principles of quasi-equivalence formulated by Caspar and Klug [Caspar, D. L. & Klug, A. (1962) Cold Spring Harbor Symp. Quant. Biol. 27, 1-4] with strict Euclidean geometric considerations results in predictions of the major features of the E2p dodecahedron matching the observed features almost exactly.

Interaction of the globular domains of pIII protein of filamentous bacteriophage fd with the F-pilus of Escherichia coli.

Gene 3 protein (pIII), a minor coat protein at one end of the filamentous bacteriophage fd, is involved in initiating the infection by the virus of Escherichia coli cells that display an F-pilus. Infection is thought to start with the adsorption of the D2 domain of pIII to the tip of the pilus, retraction of the pilus, and penetration of the E. coli cell membrane mediated by an interaction between the D1 domain of pIII and the Tol protein complex in the membrane. A subgene encoding the pIII-D1D2 di-domain was created, and the subgene was successfully overexpressed in E. coli cells. Domains D1 and D2 were separated after limited proteolysis of a modified pIII-D1D2 (designated pIII-D1D2.trp) into which two tryptic cleavage sites were introduced at appropriate points. The purified pIII-D1D2 di-domain and pIII-D2 domain were able to bind to the F-pilus, competing with the wild-type pIII and delaying infection by the intact filamentous phage. The pIII-D1 domain was unable to bind to the F-pilus by this criterion. This provides conclusive evidence that the pIII-D2 domain is responsible for the adsorption to the tip of the F-pilus and can achieve this in the absence of domain D1, opening the way to identifying the molecular basis of the interaction of pIII-D2 with the pilus.