Role of conserved residues in structure and stability: tryptophans of human serum retinol-binding protein, a model for the lipocalin superfamily.

Serum retinol binding protein (RBP) is a member of the lipocalin family, proteins with up-and-down beta-barrel folds, low levels of sequence identity, and diverse functions. Although tryptophan 24 of RBP is highly conserved among lipocalins, it does not play a direct role in activity. To determine if Trp24 and other conserved residues have roles in stability and/or folding, we investigated the effects of conservative substitutions for the four tryptophans and some adjacent residues on the structure, stability, and spectroscopic properties of apo-RBP. Crystal structures of recombinant human apo-RBP and of a mutant with substitutions for tryptophans 67 and 91 at 1.7 A and 2.0 A resolution, respectively, as well as stability measurements, indicate that these relatively exposed tryptophans have little influence on structure or stability. Although Trp105 is largely buried in the wall of the beta-barrel, it can be replaced with minor effects on stability to thermal and chemical unfolding. In contrast, substitutions of three different amino acids for Trp24 or replacement of Arg139, a conserved residue that interacts with Trp24, lead to similar large losses in stability and lower yields of native protein generated by in vitro folding. The results and the coordinated nature of natural substitutions at these sites support the idea that conserved residues in functionally divergent homologs have roles in stabilizing the native relative to misfolded structures. They also establish conditions for studies of the kinetics of folding and unfolding by identifying spectroscopic signals for monitoring the formation of different substructures.

Toward rational design of ribonuclease inhibitors: high-resolution crystal structure of a ribonuclease A complex with a potent 3',5'-pyrophosphate-linked dinucleotide inhibitor.

The crystal structure of ribonuclease A (RNase A) in complex with pdUppA-3'-p [5'-phospho-2'-deoxyuridine-3'-pyrophosphate (P'-->5') adenosine 3'-phosphate] has been determined at 1.7 A resolution. This dinucleotide is the most potent low molecular weight inhibitor of RNase A reported to date (K(i) = 27 nM) and is also effective against two major nonpancreatic RNases: eosinophil-derived neurotoxin and RNase-4; in all cases, tight binding in large part derives from the unusual 3',5'-pyrophosphate internucleotide linkage [Russo, N., and Shapiro, R. (1999) J. Biol. Chem. 274, 14902-14908]. The design of pdUppA-3'-p was based on the crystal structure of RNase A complexed with 5'-diphosphoadenosine 3'-phosphate (ppA-3'-p) [Leonidas, D. D., Shapiro, R., Irons, L. I., Russo, N., and Acharya, K. R. (1997) Biochemistry 36, 5578-5588]. The adenosine of pdUppA-3'-p adopts an atypical syn conformation not observed for standard adenosine nucleotides bound to RNase A. This conformation, which allows extensive interactions with Asn 67, Gln 69, Asn 71, and His 119, is associated with the placement of the 5'-beta-phosphate of the adenylate, rather than alpha-phosphate, at the site where substrate phosphodiester bond cleavage occurs. The contacts of the deoxyuridine 5'-phosphate portion of pdUppA-3'-p appear to be responsible for the 9-fold increased affinity of this compound as compared to ppA-3'-p: the uracil base binds to Thr 45 in the same manner as previous pyrimidine inhibitors, and the terminal 5'-phosphate is positioned to form medium-range Coulombic interactions with Lys 66. The full potential benefit of these added interactions is not realized because of compensatory losses of hydrogen bonds of Lys 7 and Gln 11 with the terminal 3'-phosphate and the adenylate 5'-alpha-phosphate, which were not predicted by modeling. The results reported here have important implications for the design of improved inhibitors of RNase A and for the development of therapeutic agents to control the activities of RNase homologues such as eosinophil-derived neurotoxin and angiogenin that have roles in human pathologies.

Crystal structures of ribonuclease A complexes with 5'-diphosphoadenosine 3'-phosphate and 5'-diphosphoadenosine 2'-phosphate at 1.7 A resolution.

High-resolution (1.7 A) crystal structures have been determined for bovine pancreatic ribonuclease A (RNase A) complexed with 5'-diphosphoadenosine 3'-phosphate (ppA-3'-p) and 5'-diphosphoadenosine 2'-phosphate (ppA-2'-p), as well as for a native structure refined to 2.0 A. These nucleotide phosphates are the two most potent inhibitors of RNase A reported so far, with Ki values of 240 and 520 nM, respectively. The binding modes and conformations of ppA-3'-p and ppA-2'-p were found to differ markedly from those anticipated on the basis of earlier structures of RNase A complexes. The key difference is that the 5'-beta-phosphate rather than the 5'-alpha-phosphate of each inhibitor occupies the P1 phosphate binding site. As a consequence, the ribose moieties of the two nucleotides are shifted by approximately 2 A compared to the positions of their counterparts in earlier complexes, and the adenine rings are rotated into unusual syn conformations. Thus, the six-membered and five-membered rings of both adenines are reversed with respect to the others but nonetheless engage in extensive interactions with the residues that form the B2 purine binding site of RNase A. Despite the close structural similarity of the two inhibitors, the puckers of their furanose rings are different: C2'-endo and C3'-endo, respectively. Moreover, their 5'-alpha-phosphates and 3'(2')-monophosphates interact with largely different sets of RNase residues. The results of this crystallographic study emphasize the difficulties inherent in qualitative modeling of protein-inhibitor interactions and the compelling reasons for high-resolution structural studies in which quantitative design of improved inhibitors was enabled. The structures presented here provide a promising starting point for the rational design of tight-binding RNase inhibitors, which may be used as therapeutic agents in restraining the ribonucleolytic activities of RNase homologues such as angiogenin, eosinophil-derived neurotoxin, and eosinophil cationic protein.

A single amino acid mutation enhances the thermal stability of Escherichia coli malate dehydrogenase.

The stability of wild-type Escherichia coli malate dehydrogenase was compared with a mutant form of the enzyme with the amino acid residue at position 102 changed from arginine to glutamine. The mutation occurs on the underside of a mobile loop which closes over the active-site cleft on formation of the enzyme/cofactor/substrate ternary complex. The mutant enzyme is kinetically compromised while the wild-type enzyme is highly specific for oxaloacetate. The mutant enzyme was shown to be more resistant to irreversible thermal denaturation by thermal inactivation experiments and high-sensitivity differential scanning calorimetry than the wild-type enzyme. In contrast, resistance of both enzymes to reversible unfolding in guanidinium chloride was similar. Circular dichroic spectropolarimetry shows the secondary structures of the enzymes are similar but there is a demonstrable difference in tertiary structure. From the position of the mutation, it is conjectured that the substitution on a mobile surface loop results in partial closure of the loop and greater resistance to thermal inactivation of the mutant enzyme. However, molecular modelling combined with circular dichroic spectropolarimetry indicate that the mutation may have a more widespread effect on the structure than simply partial closure of the mobile surface loop as the environment of distant tyrosine residues is altered. Resistance of the wild-type enzyme to thermal inactivation can be increased by cofactor addition, which may have the effect of partial closure of the mobile surface loop, but has little effect on the mutant enzyme.

Immune responses in humans and animals to meningococcal transferrin-binding proteins: implications for vaccine design.

The results reported here show that the two meningococcal transferrin-binding proteins (TBP1 and TBP2) generate different immune responses in different host species and that there is variation in response dependent on the method of antigen preparation and possibly the route of administration. Mice immunized with either whole cells of Neisseria meningitidis SD (B:15:P1.16) or the isolated TBP1-TBP2 complex from the same strain produced antisera which, when tested against a representative panel of meningococcal isolates by Western blotting (immunoblotting), recognized some but not all heterologous TBP2 molecules. In contrast, rabbit antisera raised to the same preparations were cross-reactive with almost all the TBP2 molecules. The immune response to TBP1 was also host species dependent. Western blot analysis with denatured TBP1 failed to detect antibodies in antisera raised in mice to whole cells or in a rabbit to the TBP1-TBP2 complex but detected broadly cross-reactive antibodies in mouse anti-TBP1-TBP2 complex sera and strain-specific antibodies in rabbit anti-whole-cell serum. Human convalescent-phase sera obtained from five patients infected with meningococci of different serogroups and serotypes contained fully cross-reactive antibodies to TBP2 but no anti-TBP1 antibodies, when examined on Western blots. However, on dot immunoblots, the same patients' sera, as well as the mouse anti-whole cell and the rabbit anti-TBP1-TBP2 complex sera, reacted with purified biologically active TBP1 of strain SD. This indicates that native TBP1, a protein which loses its biological and some of its immunological activities when denatured, is immunogenic and that humans generate cross-reactive antibodies to native epitopes. These observations have important implications for assessing the vaccine potential of TBPs and other meningococcal antigens. Conclusions regarding the usefulness of TBPs as candidate components of meningococcal serogroup B vaccines based on results from certain animal species such as mice, or on methods such as Western blotting, may have little bearing on the situation in humans and may lead to some potentially useful antigens being disregarded.

Localization of antigenic domains on the major subunits of Bordetella pertussis serotype 2 and 3 fimbriae.

Antibody-binding domains on the major subunits of Bordetella pertussis serotype 2 (Fim2) and 3 fimbriae (Fim3) have been identified using synthetic peptides which were screened for recognition by anti-protein monoclonal antibodies (mAbs). The presence of non-contiguous fimbrial epitopes was demonstrated by both anti-Fim2 and anti-Fim3 mAbs, several of which recognized at least two peptides that were discontinuous in the amino acid sequence of the corresponding subunits. The specificity of one mAb, 51/24, directed against Fim2, was investigated by replacement-set analysis of a 10-residue peptide, and revealed that antibody binding to the peptide was dependent on the sequence N94PQ96 which is non-conserved in Fim3. Furthermore, proline at residue 95 was found to be essential for mAb 51/24 binding. The specific anti-Fim3 mAb, AG3A, was found to recognize the 10-residue carboxy-terminal peptide from both Fim3 and, unexpectedly, from Fim2. This result suggests that mAb AG3A serospecificity at the protein level is determined by a conformational constraint which prevents mAb AG3A binding to the Fim2 C-terminal domain. Several free peptides containing amino acid residues which comprise part of the Fim2 and Fim3 epitopic domains were prepared as immunogens. One of these peptides was immunogenic in the mouse, indicating the location of a T-helper cell epitope within the peptide sequence, and induced a strong anti-peptide antibody response. The other peptides each required immunization as a conjugate with a carrier protein for anti-peptide antibody stimulation. All four anti-peptide antibody preparations only weakly recognized fimbriae-coated ELISA plates.(ABSTRACT TRUNCATED AT 250 WORDS)

Localisation of a receptor-recognition domain on the S3 subunit of pertussis toxin by peptide mapping.

Overlapping 10-amino-acid peptides, which consecutively span the amino acid sequence of the S3 subunit of pertussis toxin, were synthesised on polyethylene pins and screened for their ability to bind the glycoprotein fetuin. Fetuin binding was localised to a single peptide comprising amino acids 46-55. A free peptide, (E)S3c, of longer sequence (S3 amino acids 44-58) was also found to bind alpha-1-acid glycoprotein, mixed brain gangliosides and fetuin. (E)S3c also recognised asialofetuin but with a lower apparent affinity relative to fetuin. The single tryptophan residue of the peptide yielded a fluorescence-emission maximum of 355 nm. In the presence of either ganglioside or the phospholipid L-alpha-lysolecithin, but not N-acetylneuramin-lactose or lactosylceramide, the emission intensity of (E)S3c was enhanced and the emission maximum blue-shifted to 340 nm by ganglioside, or to 345 nm by L-alpha-lysolecithin. Monosialogangliosides, disialogangliosides, and trisialogangliosides, when fluorescence-titrated, were each found to bind the peptide with a similar dissociation constant of 4.4 +/- 2.8 microM. These findings demonstrate that region 44-58 of the pertussis-toxin S3 subunit is likely to be involved in the recognition of both glycosylated and phospholipid constituents of target-cell membranes.

Some implications of structural collapse during freeze-drying using Erwinia caratovora L-asparaginase as a model.

When the enzyme Erwinia caratovora L-asparaginase was freeze-dried in mixtures of lactose and sodium chloride, biological activity and protein structure were preserved during drying. However, by altering the ratios of the excipients in the formulation it was possible to obtain products which were pharmaceutically acceptable or unacceptable as assessed by the criteria of dried cake appearance, moisture content or ease of reconstitution.

Effects of guanidinium hydrochloride on the structure and immunological properties of Bordetella pertussis fimbriae.

Denaturation of Bordetella pertussis fimbrial preparations by guanidinium hydrochloride (GdnHCl) has been characterized using static light scattering, c.d., fluorescence and antibody recognition. The susceptibility of Fim2 + 3 (a mixed preparation of two fimbrial types) to GdnHCl was found to be highly dependent on pH; as the pH was increased from pH 7.2 to 10.5, the concentration of GdnHCl required to induce 50% denaturation was decreased. At pH 10.5, Fim2 + 3 was denatured by GdnHCl in a three-step pathway comprising: (1) formation of a pre-denaturational intermediate at less than 1.0 M-GdnHCl; (2) dissociation of the fimbrial polymer into subunits between 2 M- and 3.2 M-GdnHCl; and (3) subunit unfolding between 2.8 M- and 3.6 M-GdnHCl. A similar pathway was also found for the denaturation of the individual fimbrial types, Fim2 and Fim3, except that unfolding of either subunit commenced at a lower GdnHCl concentration (2.2 M) than that found for the mixture of fimbriae, Fim2 + 3. The second step in the denaturation pathway, dissociation into subunits, was partially reversible, but the renaturation and reassociation of fully unfolded subunits to form fimbriae-like structures was not achieved. These findings demonstrate that the GdnHCl denaturation of complex polymeric proteins is unlikely to follow a reversible two-state denaturation pathway, and support the involvement of a chaperone-like protein in the folding and assembly of the fimbriae in vivo. Measurement of the ability of anti-fimbrial monoclonal antibodies to recognize intermediates in the denaturation pathway enabled the identification of two types of epitope which were dependent on different aspects of fimbrial tertiary/quaternary structure.

A spectroscopic and conformational study of pertussis toxin.

The conformation of native pertussis toxin has been investigated by secondary structure prediction and by circular dichroism, fluorescence and second-derivative ultraviolet absorption spectroscopy. The far-ultraviolet circular dichroic spectrum is characteristic of a protein of high beta-sheet and low alpha-helix content. This is also shown by an analysis of the circular dichroic spectrum with the Contin programme which indicates that the toxin possesses 53% beta-sheet, 10% alpha-helix and 37% beta-turn/loop secondary structure. Second-derivative ultraviolet absorption spectroscopy suggests that 34 tyrosine residues are solvent-exposed and quenching of tryptophan fluorescence emission has shown that 4 tryptophan residues are accessible to iodide ions. One of these tryptophans appears to be in close proximity to a positively charged side-chain, since only 3 tryptophans are accessible to caesium ion fluorescence quenching. When excited at 280 nm, the emission spectrum contains a significant contribution from tyrosine fluorescence, which may be a consequence of the high proportion (55%) of surface-exposed tyrosines. No changes in the circular dichroic spectra of the toxin were found in the presence of the substrate NAD. However, NAD did quench both tyrosine and tryptophan fluorescence emission but did not change the shape of the emission spectrum, or the accessibility of the tryptophans to either the ionic fluorescence quenchers or the neutral quencher acrylamide.

Binding of NAD+ to pertussis toxin.

The equilibrium dissociation constant of NAD+ and pertussis toxin was determined by equilibrium dialysis and by the quenching of the protein's intrinsic fluorescence on titration with NAD+. A binding constant, Kd, of 24 +/- 2 microM at 30 degrees C was obtained from equilibrium dialysis, consistent with the previously determined value for the Michaelis constant, Km, of 30 +/- 5 microM for NAD+ (when the toxin is catalysing the ADP-ribosylation of water and of dithiothreitol). The intrinsic fluorescence of pertussis toxin was quenched by up to 60% on titration with NAD+, and after correction for dilution and inner filter effects, a Kd value of 27 microM at 30 degrees C was obtained, agreeing well with that found by equilibrium dialysis. The binding constants were measured at a number of temperatures using both techniques, and from this the enthalpy of binding of NAD+ to toxin was determined to be 30 kJ.mol-1, a typical value for a protein-ligand interaction. There is one binding site for NAD+ per toxin molecule.

The secondary structure of protein G', a robust molecule.

The secondary structure of recombinant streptococcal Protein G' was predicted and compared with spectropolarimetric data. The predicted secondary structure consisted of 37 +/- 4% alpha-helix and 30 +/- 5% beta-sheet, whereas the values obtained from c.d. data were 29 +/- 2% alpha-helix and 41 +/- 3% beta-sheet. An alpha-helix-beta-sheet/turn-alpha-helix motif is conjectured to comprise the Fc-binding unit. The c.d. spectra in the near u.v. and far u.v. show that the Protein G' molecule is stable to heating at 100 degrees C and to extremes of pH (pH 1.5 to 11.0). The protein retained biological activity at these extremes. The molecule uncoils above pH 11.5 in a time-dependent fashion. Unfolding of the molecule in guanidinium chloride was monitored by c.d. and fluorescence emission; 3 M-guanidinium chloride was required to unfold the protein by 50%. The protein was completely unfolded in 5.5 M-guanidinium chloride and fully refolded with restoration of activity after removal of guanidinium chloride.

Detection and characterization of intermediates in the folding of large proteins by the use of genetically inserted tryptophan probes.

L-Lactate dehydrogenase from Bacillus stearothermophilus was rebuilt by using site-directed mutagenesis to produce an enzymically active, tryptophan-less enzyme by replacing all the wild-type tryptophans (80, 150, and 203) by tyrosines. Nine single tryptophan-containing active enzymes were constructed from this enzyme by genetically replacing one of the tyrosines 36, 85, 147, 190, 203, 237, 248, 279, or 285 by tryptophan. The equilibrium and the time-resolved tryptophan fluorescence intensity and anisotropy were used to report unfolding events in guanidine hydrochloride (GHCl) monitored from these nine defined positions. Three structural transitions, half complete at 0.55, 1.7, and 2.8 M GHCl, were identified and defined four folding intermediates, I (native), II (expanded monomer 1), III (expanded monomer 2), and IV (random coil), stable at 0, 1, 2.2, and 4 M GHCl, respectively. Intermediate II is a globular monomer. All the probed alpha-helices and most of the beta-structure was intact. There was an increase in the rate but not the extent of the mobilities of six of the probed tryptophan side chains, indicating loss of tertiary structure. Circular dichroism (CD) showed all the secondary structure to be intact. Intermediate III is monomeric and still globular, but the tryptophan anisotropy indicated an increase mobility at positions 36, 85, 190, 203, 279, and 285. Helix alpha-B is further disrupted but helices alpha-1F, alpha-2G, and alpha 3G were still rigid. CD showed half the secondary structure to be still intact. Intermediate IV is a random coil in which all tryptophans have complete rotational freedom and the helix CD signal is lost.(ABSTRACT TRUNCATED AT 250 WORDS)

Phase II trial of whole-cell pertussis vaccine vs an acellular vaccine containing agglutinogens.

An acellular pertussis vaccine containing agglutinogens 2 and 3, pertussis toxin, and filamentous haemagglutinin was developed by the Centre for Applied Microbiology and Research in the UK. 188 infants were entered into a randomised blind trial and received either the acellular or a whole-cell vaccine, combined with diphtheria and tetanus toxoids, in a 3, 5, and 8-10 month schedule. Local reactions were similar in the two groups but significantly fewer infants had systemic symptoms after the acellular vaccine. Mean log-antibody titres to the agglutinogen and toxin components were higher with the acellular than with the whole-cell vaccine. Persistence of antibodies one year after the third dose was also better in the acellular group.

Recognition of pertussis toxin by antibodies to synthetic peptides.

Eight synthetic peptides, selected from the amino acid sequence of pertussis toxin (PT) subunits S1, S2, S3 and S4, were assessed for their ability to induce protein-recognizing and neutralizing antibodies. Seven of these peptides, prepared as conjugates of either keyhole limpet haemocyanin or tetanus toxoid, induced significant levels of antibody, all of which reacted with SDS-denatured PT on Western blots. Six of the antibodies bound to PT-coated ELISA plates; this binding was inhibited by homologous peptide antigen. However, none of the antibodies, including those directed against the N-terminus of subunit S1, were able to attenuate in vivo or in vitro toxin-dependent activity. Further investigation revealed that only one antibody, specific for the C-terminus of S1 (peptide Slc, 237-255), could recognize the conformation of native PT in solution. The other five antipeptide antibodies which reacted with PT-coated ELISA plates did not recognize PT when captured onto ELISA plates via either a monoclonal antibody or fetuin, unless the conformation of the toxin had been relaxed by reduction with dithiothreitol. Conversely, the native PT-recognizing response of peptide Slc did not bind the conformationally relaxed PT molecule. From this study, it appears likely that a peptide capable of inducing PT-neutralizing antibody must closely resemble the conformation of the cognate sequence in the native protein.

Serotyping Bordetella pertussis strains.

The serotyping scheme for Bordetella pertussis, developed in the 1950s, depends on the presence or absence of various strains of three major agglutinogens, two of which have been shown to be fimbrial proteins, and several minor agglutinogens, the biochemical nature of which is unknown. This article reviews the reasons for the confusion which has recently arisen over the nomenclature of the serotype antigens and proposes a simplified serotyping scheme based on the fimbrial components.

Agglutinogens and fimbriae of Bordetella pertussis.

Agglutinogen 2 (AGG2) of Bordetella pertussis is a fimbrial antigen and therefore a potential adhesin and acellular vaccine component. AGG2 was found to dissociate only under harsh conditions into the subunits of mol. wt. 22500 seen in SDS-PAGE. Results from studies of agglutinogen 3 (AGG3) are presented which confirm previous findings from this Laboratory that AGG3 is also a fimbrial protein but with a subunit mol. wt. of 22000. The amino acid sequence of AGG2, deduced from the nucleotide sequence of the gene encoding it, was used as a basis for synthesis of three peptides. Coupled to Keyhole Limpet Haemocyanin (KLH), the peptides were immunogenic in mice, inducing antibodies which bound well to homologous peptide in ELISA but poorly to intact fimbriae. Monoclonal and polyclonal serotype-specific antibodies failed to react significantly with the peptides or their KLH-conjugates. These results indicate that the synthetic peptides do not represent the serotype 2 epitope. Mice immunized with purified AGG2 or AGG3 were found to be protected against respiratory infection with B. pertussis. Results presented here indicate that this protection is, to a large extent, serotype-specific and that immunization of mice with AGG2 or AGG3 can lead to a change in serotype of the infecting strain. These results are analogous to findings from epidemiological studies of the protection induced in children by whole cell vaccines. They reaffirm the importance of both AGG2 and AGG3 as components of whole cell and acellular vaccines.

Pertussis vaccine: present status and future prospects.

Use of killed whole-cell Bordetella pertussis vaccines has been a major factor in control of symptomatic whooping cough (pertussis). In the UK, diminished public confidence in the safety of this vaccine led to a reduction in vaccine acceptance which correlated with an increase in the incidence of pertussis. There is a need for acellular pertussis vaccines of low toxicity which, ideally, will prevent colonization and also protect against the disease symptoms. Vaccine design can rely increasingly on knowledge of the roles of individual bacterial components in the pathogenesis of pertussis. Serotype-specific agglutinogens 2 and 3 (fimbriae) and filamentous haemagglutinin are among surface components of B. pertussis which probably mediate adhesion to the respiratory mucosa. Systemic effects of pertussis can largely be attributed to the lymphocytosis promoting factor (pertussis toxin). Vaccines containing detoxified toxin plus one or more purified adhesins are envisaged at present.

The location of surface antigens of Bordetella pertussis by immuno-electron microscopy.

Immuno-electron microscopy using colloidal gold-tagged monoclonal antibodies (McAbs) has been used to detect antigens on the surface of Bordetella pertussis cells. McAbs to serotype-specific agglutinogens 2 and 3 labelled fimbriae in a serotype-specific manner. An attempt was made to determine whether individual fimbriae of serotype 1.2.3 cells bear both antigens 2 and 3. In double labelling experiments, individual cells were found to label with McAbs to antigen 2 (3 nm gold) and to antigen 3 (15 nm gold). Many fimbriae labelled with only one of these reagents, but there were instances where both labels could have been attached to the same fimbria. No fimbrial labelling was obtained with McAb to agglutinogen 1. Gold-tagged McAbs to filamentous haemagglutinin (FHA) labelled neither the fimbriae nor the surface of B. pertussis cells. Unfixed cells on electron microscope grids appeared to shed FHA readily. After fixation, a small proportion of cells bore aggregates of FHA which labelled specifically with anti-FHA McAb-gold. Results with one McAb to lymphocytosis promoting factor indicated that this antigen may be more intimately associated with the surface of B. pertussis than is FHA.

Release and purification of fimbriae from Bordetella pertussis.

A competitive ELISA has been used to monitor the release of fimbriae from 1.2.3 serotype of B. pertussis after treatment by different methods and fimbriae have been purified from homogenates or KSCN extracts by chromatography. Fimbriae purified from different serotypes have been studied by inhibition of bacterial agglutination, immunodiffusion and SDS-polyacrylamide gel electrophoresis. Fimbriae purified from a 1.2 serotype have been labelled in immunoelectron microscopy with a IgM monoclonal antibody to agglutinogen 2 and evidence is presented that fimbriae purified from a 1.3 serotype also carry the agglutinogen 3 specificity. A difference in subunit molecular weight has been found between fimbriae purified from 1.2 and 1.3 serotypes.