Search

Clinical isolates of the fungal respiratory and systemic pathogen Histoplasma capsulatum have been placed in several different classes by using genomic restriction fragment length polymorphisms (RFLPs), but in general have not been distinguished further. We report here that a polymerase chain reaction (PCR)-based DNA fingerprinting method that has been termed arbitrary primer or random amplified polymorphic DNA (RAPD) PCR can distinguish among isolates in a single RFLP class. In this method, arbitrarily chosen oligonucleotides are used to prime DNA synthesis from genomic sites that they fortuitously match, or almost match, to generate strain-specific arrays of DNA fragments. Each of 29 isolates of RFLP class 2, the group endemic in the American Midwest, was distinguished by using just three arbitrary primers. In contrast, laboratory-derived S and E colony morphology variants of two strains were not distinguished from their R parents by using 18 such primers. Thus, the clinical isolates of H. capsulatum are quite diverse, but their genomes remain stable during laboratory culture. These outcomes suggest new possibilities for epidemiological analysis and studies of fungal populations in infected hosts.

This review summarizes the biology of Histoplasma capsulatum in relation to a wide variety of corresponding pathologies in histoplasmosis. Features of these disease syndromes can be explained in part by natural variations within the fungal population and adaptations made by individual organisms to specific environments. H. capsulatum grows as mycelia and conidia in the soil; once inhaled, the organism undergoes a dramatic morphological and physiological conversion to a yeast form. The yeasts proliferate within the phagolysosomes of macrophages, using a variety of specific strategies for intracellular survival. Even avirulent strains or variants are able to avoid being killed by macrophages and instead establish inapparent or persistent infections. The ingested avirulent organisms assume enlarged shapes similar in appearance to those seen in histological sections of tissues from patients with histoplasmosis. Respiratory tract epithelial cells also appear to play a role in persistence: within them yeasts undergo phenotypic switching akin to the phase variation observed in other pathogens. This particular change involves the loss or modification of cell wall alpha-(1,3)-glucan, which is also correlated with the spontaneous appearance of avirulent variants. The repertoire of adaptive responses and natural variations within this species probably evolved from the need to adjust to a wide range of dynamic environments. In combination with the immune status of the host, these characteristics of H. capsulatum appear to influence the epidemiology, extent, and persistence of histoplasmosis.

We evaluated P388D1 macrophagelike cells as model host cells for studying the intracellular survival and strain-specific virulence of Histoplasma capsulatum. Previously characterized strains which were virulent for mice destroyed monolayers of these cells within a few days. In contrast, related avirulent "smooth" variants failed to do so even after 20 days, although they persisted within P388D1 cells for at least 7 days. On the basis of this observation, we developed a quantitative radiolabel assay to use as an initial screen for virulence. Another cell type lining the respiratory tract was then examined as a potential host for H. capsulatum. Hamster trachea epithelial (HTE) cells readily internalized a variety of strains lacking alpha-(1,3)-glucan in their cell walls; however, the tracheal cells were only rarely infected by organisms possessing this polysaccharide. We subsequently inoculated HTE cells with alpha-(1,3)-glucan-positive strains and enriched for the few yeasts infecting these cells. The progeny resembled smooth variants in terms of colony morphology, the absence of alpha-(1,3)-glucan in their cell walls, and their inability to kill macrophages. Did the HTE cells select for these variant yeasts from the parent inoculum or instigate a change from the parental phenotype? Following a 3-h uptake period, only 2% of the ingested yeasts lacked alpha-(1,3)-glucan. One day later, nearly half of the intracellular organisms lacked this polymer. This rapid conversion of a large proportion of the inoculum suggests some type of environmentally triggered change, perhaps analogous to phase variation seen in many other pathogens. Infection of epithelial cells or some other nonprofessional phagocyte during natural histoplasmosis might give rise to similar variants, thus establishing a reservoir of organisms capable of causing chronic or latent infections.

We have developed a simple and efficient transformation system for the dimorphic fungus Histoplasma capsulatum. Mutants of H. capsulatum defective in orotidine-5'-monophosphate pyrophosphorylase were transformed to prototrophy by the cloned URA5 gene of the filamentous fungus Podospora anserina. Abortive and mitotically stable transformants were obtained. The stable transformants had integrated copies of the plasmid, some in tandem head-to-tail orientation. Free plasmid identical to the transforming plasmid was present in some of the transformants. We obtained a transformation efficiency of up to 30 transformants/micrograms DNA for plasmid pPAura5-1 (9.2 kb). pPW2001, a smaller plasmid (4.7 kb) derived from pPAura5-1, transformed H. capsulatum more efficiently (up to 155 transformants/micrograms DNA).

The Tn5tac1 transposon contains a strong outward-facing promoter, Ptac, a lacI repressor gene, and a selectable Kanr gene. Transcription from Ptac is repressed by the lacI protein unless an inducer (isopropyl-beta-D-thiogalactopyranoside [IPTG]) is present. Thus, Tn5tac1 generates insertion mutations in Escherichia coli with conditional phenotypes because it is polar on distal gene expression when IPTG is absent and directs transcription of these genes when the inducer is present. To test the usefulness of Tn5tac1 in Bordetella pertussis, a nonenteric gram-negative bacterial pathogen, we chose the bifunctional adenylate cyclase-hemolysin determinant as an easily scored marker to monitor insertional mutagenesis. Tn5tac1 delivered to B. pertussis on conjugal suicide plasmids resulted in Kanr exconjugants at a frequency of 10(-3) per donor cell, and nonhemolytic (Hly-) mutants were found among the Kanr colonies at a frequency of about 1%. Of eight independent Kanr Hly- mutants, two were conditional and exhibited an Hly+ phenotype only in the presence of IPTG. Using a new quantitative assay for adenylate cyclase based on high-pressure liquid chromatography, we found that enzymatic activity in these two strains was specifically induced at least 500-fold in a dose-dependent fashion over the range of 0 to 125 microM IPTG. These data show that Ptac serves as a promoter, lacI is expressed and is functional, and IPTG can induce Ptac transcription in B. pertussis. Adenylate cyclase expression in whole cells, culture supernatants, and cell extracts from these strains depended upon IPTG, suggesting that the insertions do not merely alter secretion of adenylate cyclase-hemolysin. Other virulence determinants under control of the vir locus are expressed normally, implying that these Tn5tac1 insertions specifically regulate adenylate cyclase-hemolysin expression. We conclude that Tn5tac1 insertion mutations permit sensitive, exogenous control over the expression of genes of interest, providing a useful tool for studying virulence and other important traits of diverse bacterial species.

The locations of three cleavages that can occur in mouse 45S pre-rRNA were determined by Northern blot hybridization and S1 nuclease mapping techniques. These experiments indicate that an initial cleavage of 45S pre-rRNA can directly generate the mature 5' terminus of 18S rRNA. Initial cleavage of 45S pre-rRNA can also generate the mature 5' terminus of 5.8S rRNA, but in this case cleavage can occur at two different locations, one at the known 5' terminus of 5.8S rRNA and another 6 or 7 nucleotides upstream. This pattern of cleavage results in the formation of cytoplasmic 5.8S rRNA with heterogeneous 5' termini. Further, our results indicate that one pathway for the formation of the mature 5' terminus of 28S rRNA involves initial cleavages within spacer sequences followed by cleavages which generate the mature 5' terminus of 28S rRNA. Comparison of these different patterns of cleavage for mouse pre-rRNA with that for Escherichia coli pre-rRNA implies that there are fundamental differences in the two processing mechanisms. Further, several possible cleavage signals have been identified by comparing the cleavage sites with the primary and secondary structure of mouse rRNA (see W. E. Goldman, G. Goldberg, L. H. Bowman, D. Steinmetz, and D. Schlessinger, Mol. Cell. Biol. 3:1488-1500, 1983).

Cell wall composition of isogenic virulent-avirulent strain pairs of Histoplasma capsulatum varied markedly with respect to alpha-(1,3)-glucan content. When yeast cell walls were fractionated by standard techniques, the avirulent strains contained up to 1,000-fold less alpha-(1,3)-glucan than did their virulent parents. No alpha-(1,3)-glucan could be detected on the surface of the avirulent strain yeast cells if we used a mouse monoclonal antibody that recognized this polymer. A similar relationship between virulence and alpha-(1,3)-glucan has been described for Paracoccidioides brasiliensis. alpha-(1,3)-Glucan is also found in several other pathogenic fungi and may thus be an important common virulence determinant.

Bordetella pertussis is known to release a factor which promotes the loss of ciliated respiratory epithelium and copurifies with a soluble peptidoglycan (PG) fragment termed tracheal cytotoxin (TCT). The objective of this study was to determine whether pertussis organisms turn over and release PG derivatives in addition to TCT. B. pertussis Tohama (phase III) was grown in liquid Stainer-Scholte medium containing [3H]diaminopimelic acid (DAP) to label PG specifically, washed to remove free label, and suspended in fresh medium without [3H]DAP. Molecular sieve chromatography of supernatants obtained from such cultures revealed a single included peak of 3H, the elution volume of which corresponded roughly to a disaccharide peptide monomer standard (ca. 10(3) daltons). This material (i) contained [3H]DAP in acid-hydrolyzable linkage, (ii) comigrated with 1,6-anhydro-N-acetylmuramic acid-containing disaccharide peptides on paper chromatography, (iii) was resistant to degradation by mild alkali, and (iv) was indistinguishable from authentic TCT by high-voltage paper electrophoresis and two reversed-phase high-performance liquid chromatography systems. Together, the data suggest that B. pertussis releases a markedly homogeneous set of PG fragments, consisting principally of TCT, and that TCT is possibly a nonreducing, anhydromuramic acid-containing fragment or a cyclic PG derivative.

Cultured hamster trachea epithelial cells were selected as an in vitro model system to study Bordetella pertussis in the respiratory tract. DNA synthesis by serum-stimulated tracheal cells, in contrast to other cell types tested, was inhibited by the supernatant from log-phase B. pertussis broth cultures. A sensitive microassay with these tracheal cells permitted the development of a chromatographic purification scheme based on aggregation of the biological activity under salt-free conditions. The active fraction from this first stage of purification caused a dose-dependent inhibition of DNA synthesis without a similar effect on RNA or protein synthesis. Organ cultures of hamster tracheal rings, when exposed to this partially purified fraction, developed epithelial cytopathology comparable to that seen during B. pertussis infection. Ciliary activity showed and eventually ceased as ciliated cells were extruded from the ring, leaving an intact but mostly nonciliated epithelium. Further purification of this biological activity was achieved with preparative-scale high-voltage paper electrophoresis. Based on ninhydrin staining and the radioactive profile of material purified from radiolabeled B. pertussis cultures, four fractions were eluted from the paper by descending chromatography. Only component B caused a dose-dependent inhibition of cultured tracheal cell DNA synthesis and epithelial cytopathology in tracheal rings. Combination experiments also demonstrated enhanced inhibition by component B in the presence of component G (oxidized glutathione), a copurifying molecule from the growth medium. Amino acid analysis (five residues), glycine (two residues), cysteine (two residues), and diaminopimelic acid (one residue), as well as muramic acid and glucosamine.

We examined Bordetella avium for virulence factors common to Bordetella pertussis, including pertussis toxin, filamentous hemagglutinin, adenylate cyclase, dermonecrotic toxin, and tracheal cytotoxin. B. avium produced a dermonecrotic toxin and a tracheal cytotoxin. The dermonecrotic toxin of B. avium is a 155,000-molecular-weight, heat-labile protein which was lethal for mice, guinea pigs, young chickens, and turkey poults and produced dermonecrosis when injected intradermally into guinea pigs, chickens, and turkey poults. High-pressure liquid chromatography of B. avium culture supernatant fluid revealed the presence of a tracheal cytotoxin chemically identical to that produced by B. pertussis. B. avium isolates were negative for B. pertussis-like filamentous hemagglutinin and pertussis toxin when assayed with antibody against B. pertussis filamentous hemagglutinin and pertussis toxin. Furthermore, B. avium failed to induce the clustered CHO cell morphology characteristic of pertussis toxin. Adenylate cyclase assays indicated that B. avium does not produce an extracytoplasmic adenylate cyclase, even after passage through embryonated turkey eggs. Since production of virulence proteins by B. pertussis is regulated by growth in media containing nicotinamide or MgSO4 or by growth at reduced temperatures, we determined the effect of these supplements and growth conditions on production of dermonecrotic toxin by B. avium. Production of dermonecrotic toxin in B. avium was not altered by growth in media containing 100 microM FeSO4 or 500 micrograms of nicotinamide per ml or by growth at 25 or 42 degrees C, but production was significantly decreased by growth in media containing 20 mM MgSO4 and slightly reduced by growth in media containing 500 micrograms of nicotinic acid per ml. These studies revealed that B. avium is similar to B. pertussis in that both species produce a dermonecrotic toxin and a tracheal cytotoxin and production of dermonecrotic toxin is regulated by nicotinamide and MgSO4. The presence of dermonecrotic toxin and tracheal cytotoxin in all Bordetella species indicates that these products may be important virulence factors in bordetellosis.

The etiological agent of whooping cough, Bordetella pertussis, destroys the ciliated epithelial cells lining the large airways of infected individuals. This cytopathology can be reproduced in respiratory epithelium by tracheal cytotoxin (TCT), a small peptidoglycan-related molecule purified from the culture supernatant of growing B. pertussis organisms. Using fast atom bombardment mass spectrometry, we analyzed the positive- and negative-ion spectra of the purified, biologically active material and assigned a mass of 921 daltons to TCT. Analysis of fragment ions in these spectra as well as the spectra of the methyl ester and acetylated derivatives of TCT unambiguously defined the primary structure of TCT as N-acetylglucosaminyl-1,6-anhydro-N-acetylmuramylalanyl-gamma- glutamyldiaminopimelylalanine. TCT is therefore identical with the ciliostatic anhydropeptidoglycan monomer released by Neisseria gonorrhoeae and with the neurologically active slow-wave sleep-promoting factor FSu. These and other structurally related glycopeptides containing muramic acid thus form a family of molecules with remarkably diverse biological activities.

Specific destruction of ciliated epithelial cells lining the large airways is the primary respiratory tract cytopathology associated with human Bordetella pertussis infections. We have purified a single low-molecular-weight glycopeptide, tracheal cytotoxin (TCT), that appears to cause this pathology. By using a combination of solid-phase extraction and reversed-phase high-pressure liquid chromatography, about 700 nmol of biologically active peptide can be isolated from 1 liter of B. pertussis culture supernatant (approximately 60% yield). TCT at concentrations of 1 microM destroyed the ciliated cell population when incubated with respiratory epithelium in vitro. This concentration of TCT is similar to the concentrations found in the culture supernatant of growing B. pertussis. Purified TCT also inhibited DNA synthesis of hamster trachea epithelial cells in a quantitative, dose-dependent fashion. Endotoxin was not detected in the purified material, and neither B. pertussis nor Escherichia coli endotoxin could duplicate the biological activities of TCT. Amino acid and amino sugar analyses of purified TCT revealed the presence of glucosamine, muramic acid, alanine, glutamic acid, and diaminopimelic acid in molar ratios of 1:1:2:1:1. This suggests that TCT, the released ciliostatic principle of B. pertussis, is a disaccharide tetrapeptide subunit of peptidoglycan.

We have formulated two defined media which yield excellent plating efficiency of the yeast phase of Histoplasma capsulatum. Neither requires the addition of conditioned medium or purified siderophores, although the simpler of the two media includes hemin as a source of solubilized iron. Agarose, rather than conventional agar, is the solidifying agent for both media. Plating efficiency usually exceeds 90% with both North American and Central American isolates of H. capsulatum.

Procedures for the selective isolation and cultivation in monolayer of respiratory cells have been developed. This technique requires repeated protease treatment and gradient centrifugation of hamster tracheal or lung tissues and permits the establishment of proliferating cultures of epithelial cells with biologic specialization. Mucus synthesis was monitored in cultured tracheal cells by incorporation of 3H-labeled N-acetyl-D-galactosamine and 14C-serine into glycoprotein as determined by trichloroacetic acid precipitation of growth medium followed by acrylamide gel electrophoresis. For comparative purposes tracheal explants and several established cell lines were also examined. Synthesis and secretion of the glycoprotein macromolecule by tracheal cell monolayers appeared to be regulated by vitamin A since its addition to the culture medium significantly increased both the number of cell-associated granules and glycoprotein secretion. Lung-originated cell cultures were grown to confluence and radio-labeled with 3H-choline in serum-free medium for 24 hr to examine surfactant synthesis. Cell monolayers and growth medium were then extracted by the Folch method, and total radioactive phosphatidylcholine as well as disaturated phosphatidylcholine were determined by thin-layer chromatography and alumina gel fractionation of osmium tetroxide-reactive phospholipid, respectively. Data indicate that these cultures have a marked ability to synthesize and secrete surfactant when compared to other established cell lines. In addition, naturally transformed cells that arose during passage and senescence of the primary cultures were analyzed for their biosynthetic capabilities.

A reliable cell isolation technique was developed to allow the cultivation of cells from the hamster respiratory tract. Repeated thermolysin treatments and gradient centrifugation yielded a cellculture completely free from contamination by fibroblasts. Viable cells could be isolated from as little tissue as a single hamster trachea, but in vitro proliferation occurred only if the hamster was less than 4 months of age. The cultured cells could be repeatedly passaged and subcultured for weeks by employing normal tissue culture techniques. Morphologically, the monolayers appeared to be a homogeneous population of epithelial cells, and successful cloning of freshly isolated single cells resulted in apparently identical cultures. The epiethelial origin of these cells was also suggested by continued growth in minimum essential medium with D-valine substituted for L-valine. The relative ease with which this cell type can be isolated, cultured, and manipulated in vitro should encourage its application as a model of the respiratory epithelium.

The yeast form of the dimorphic fungus Histoplasma capsulatum survives within macrophages after phagocytosis. To do so, it must avoid, inhibit, or resist a variety of toxic oxygen metabolites. Using ferricytochrome c reduction to assay superoxide release, we examined the response of mouse macrophages to the yeast form of various H. capsulatum strains. Doses of zymosan as low as 20 particles per macrophage elicited superoxide, whereas H. capsulatum failed to induce superoxide even at 160 yeast cells per macrophage. This phenomenon was observed with two virulent strains of H. capsulatum (G217B and G186A) and with an avirulent variant of G186A. Over a 15- to 150-min observation period, zymosan stimulated increasing reduction of ferricytochrome c, but H. capsulatum did not. When added concurrently with zymosan, H. capsulatum had no effect on superoxide production. Therefore, H. capsulatum was unable either to inactivate the oxygen radical or inhibit host cell superoxide response to other competent stimuli. Enzymatically generated superoxide reduced ferricytochrome c even in the presence of H. capsulatum, again implying that the organism does not readily inactivate superoxide. This experiment also demonstrated that the yeast did not interfere with the assay used. Thus, rather than inhibiting superoxide generation or inactivating the anion, H. capsulatum yeast cells appear to avoid the toxic effects of superoxide by failing to trigger its release.

The issue of whether or not phagocytized Histoplasma capsulatum yeasts evade phagosome-lysosome fusion (P-LF) has been debated by several investigators. To resolve this problem, yet avoid drawbacks associated with the conventional assays of P-LF (electron microscopy and the acridine orange assay), we used fluorescein isothiocyanate-labeled dextran (FITC-dextran) to monitor P-LF in the macrophage-like cell line P388D1.D2. Controls indicated that FITC-dextran could be used to distinguish between evasion of P-LF by Toxoplasma gondii and phagolysosome formation following ingestion of Saccharomyces cerevisiae. Phagosomes containing H. capsulatum clearly fused with FITC-dextran-labeled lysosomes at a rate comparable to that observed for S. cerevisiae. This was true for several strains of H. capsulatum including two avirulent strains derived in this laboratory. Varying the dose of H. capsulatum did not alter the percentage of phagolysosomes formed. Our results indicate that H. capsulatum is one of a small number of organisms which is able to survive in phagolysosomes.

Two regions of mouse rDNA were sequenced. One contained the last 323 nucleotides of the external transcribed spacer and the first 595 nucleotides of 18S rRNA; the other spanned the entire internal transcribed spacer and included the 3' end of 18S rRNA, 5.8S rRNA, and the 5' end of 28S rRNA. The mature rRNA sequences are very highly conserved from yeast to mouse (unit evolutionary period, the time required for a 1% divergence of sequence, was 30 X 10(6) to 100 X 10(6) years). In 18S rRNA, at least some of the evolutionary expansion and increase in G + C content is due to a progressive accretion of discrete G + C-rich insertions. Spacer sequence comparisons between mouse and rat rRNA reveal much more extensive and frequent insertions and substitutions of G + C-rich segments. As a result, spacers conserve overall G + C richness but not sequence (UEP, 0.3 X 10(6) years) or specific base-paired stems. Although no stems analogous to those bracketing 16S and 23S rRNA in Escherichia coli pre-rRNA are evident, certain features of the spacer regions flanking eucaryotic mature rRNAs are conserved and could be involved in rRNA processing or ribosome formation. These conserved regions include some short homologous sequence patterns and closely spaced direct repeats.

The combined use of non-aggregating Histoplasma capsulatum strains and a defined medium which allows quantitative plating of the yeast phase has allowed us to select 5-fluoroorotic acid (5-FOA)-resistant mutants of this dimorphic fungus. Approximately two-thirds of the 5-FOA-resistant strains were auxotrophic for uracil; all were deficient in orotidine-5'-monophosphate pyrophosphorylase (OMPpase) activity. One class of OMPpase mutant (alpha), which retained a low level of OMPpase activity, was auxotrophic in the yeast phase (37 degrees C) but grew slowly in the mycelial phase (25 degrees C) without exogenous uracil. This phenotype was not due to a temperature-sensitive OMPpase activity. Both wild-type and alpha mutants had a higher OMPpase activity in the mycelial phase than the yeast phase; this increased activity may be sufficient to allow mycelial growth of alpha mutants.

A selection procedure was developed which allowed us to isolate spontaneous isogenic avirulent clones from virulent strains of Histoplasma capsulatum. The avirulent yeasts had a unique phenotype: they did not aggregate like the parental strains but grew as dispersed budded and unbudded single cells in liquid medium. On solid medium, the avirulent variant strains grew as smooth-textured colonies, whereas the virulent parental strains grew as rough convoluted colonies. Virulence testing in mice demonstrated that the smooth variants gave 50% lethal dose values similar to those of the avirulent Downs strain. Growth curves for the paired rough and smooth strains were similar. Furthermore, they had the same protein profiles when crude cell fractions were separated on one-dimensional polyacrylamide gels or when whole-cell extracts were separated by two-dimensional gel electrophoresis. Electrophoresis of culture supernatants, however, revealed a difference in a released low-molecular-weight peptide that may be related to virulence. In addition to their usefulness in comparative virulence studies, these avirulent strains should prove valuable for H. capsulatum genetic experiments because of the unique ability of these yeasts to grow without clumping.

Of the various toxins and virulence-related factors produced by Bordetella pertussis, only one has been demonstrated to reproduce the specific respiratory epithelial cytopathology characteristic of the pertussis syndrome. That molecule is tracheal cytotoxin (TCT), which is released by B. pertussis during log phase growth. An HPLC-based method has allowed us to purify TCT from culture supernatants, resulting in a preparation with undetectable levels of endotoxin and which is homogeneous by all analytical criteria, including fast atom bombardment-mass spectrometry (FAB-MS). Exposure to purified TCT specifically damages ciliated epithelial cells, causing ciliostasis and extrusion of these cells. Other species of Bordetella, which generate remarkably similar respiratory tract infections and ciliated cell-specific pathology, produce a chemically identical TCT. Compositional analysis and FAB-MS have unambiguously defined the structure of TCT as N-acetylglucosaminyl-1, 6-anhydro-N-acetylmuramylalanyl-gamma-glutamyl-diaminopimelylalanine+ ++. This particular disaccharide-tetrapeptide composition and arrangement reveals that TCT is apparently formed by cleavage of peptidoglycan. Unlike other gram-negative bacteria, however, B. pertussis seems to be very selective in its release of cell wall fragments: greater than 95% of soluble peptidoglycan in culture supernatants is TCT. The structure of TCT places it in the "muramyl peptide" family, a group of structurally related molecules that are responsible for a diverse array of biological activities. Neisseria gonorrhoeae also releases muramyl peptides (one of which is identical to TCT) that can cause ciliated cell-specific damage like that seen during gonococcal infection of fallopian tube mucosa. In addition, TCT is absolutely identical in structure to FSu, a potent sleep-promoting factor isolated from humans.(ABSTRACT TRUNCATED AT 250 WORDS)

Twenty isolates of the dimorphic, pathogenic fungus Histoplasma capsulatum were divided into three classes based on comparisons of restriction enzyme digests of their mitochondrial DNA and rDNA. The majority of isolates, including most North American strains and the African H. capsulatum var. duboisii variants, belong to class 2. Isolates from Central America and South America make up class 3. The attenuated Downs strain is the only member of class 1.

The most consistent pathological feature of pertussis is the selective colonization and subsequent destruction of ciliated cells in the respiratory epithelium. Phase I B. pertussis can reproduce the human respiratory tract cytopathology during in vitro infection of hamster tracheal organ cultures. However, most isolated biologically active components produced by B. pertussis (lymphocytosis-promoting factor, adenylate cyclase, dermonecrotic toxin, etc.) have no apparent cytopathic effect on the respiratory epithelium. We have purified a glycopeptide from the culture supernatant of virulent B. pertussis that mimics completely the ciliated cell pathology characteristic of pertussis (5). Tracheal cytotoxin (TCT) is released during log phase broth culture and consists of 15 amino acid residues as well as two amino sugars. The selective biological activity of TCT has been studied in tracheal organ cultures by light and electron microscopy. A series of pathological changes precedes the eventual extrusion of ciliated cells, while all other epithelial cell types appear ultrastructurally normal. TCT also causes a dose-dependent inhibition of DNA synthesis in cultured hamster trachea epithelial cells, providing a quantitative bioassay to monitor TCT activity during purification steps. Previously, TCT could be completely purified from oxidized glutathione (a major contaminant from the culture medium) only by high-voltage paper electrophoresis, a procedure not well suited for large scale work. We have now substituted a final column chromatography step that separates TCT from oxidized glutathione and other contaminating peptides. This change and other preparative scale adaptations now allow us to purify 150-250 nmol of biologically active TCT from one liter of culture supernatant (3-4 X 10(13) bacteria), a ten-fold increase over our previous batch size with no increase in processing time.(ABSTRACT TRUNCATED AT 250 WORDS)