Implications of Haemophilus influenzae Biogroup aegyptius Hemagglutinins in the Pathogenesis of Brazilian Purpuric Fever

Abstract

Brazilian purpuric fever (BPF) is an acute disease caused by Haemophilus influenzae biogroup aegyptius; it is characterized by fever, purpura, and hypotensive shock and is usually fatal. The factors responsible for bacterial virulence and pathogenesis are poorly known. Hemagglutinins have been frequently associated with bacterial virulence, and, in the present study, hemagglutinating activity was detected in extracellular products from H. influenzae biogroup aegyptius strains isolated from patients with BPF. A 60-kilodalton (kDa) molecule absorbable by human O-type erythrocytes was identified by an immunoblot assay; a corresponding fraction was chromatographically purified, and its pathogenic effect was evaluated. Rabbits injected intravenously with either the whole bacterial extracellular product or the 60-kDa fraction showed reactions similar to those seen in patients with BPF: purpura, congestion, and fibrin thrombi in the inner organs. We suggest that this hemagglutinating factor is one of the major pathogenic components of BPF.

Brazilian purpuric fever (BPF) is an acute infectious disease that affects young children and is characterized by fever and the rapid onset of purpura and hypotensive shock, followed by death ensuing 7–10 days after an episode of purulent conjunctivitis [1]. All confirmed BPF cases have occurred in either Brazil [2] or Australia [3]; the bacterial agents isolated are associated with 3 clones of Haemophilus influenzae biogroup aegyptius. Previously, this bacterium was associated exclusively with conjunctivitis, producing seasonal and epidemic infection in hot climates [4, 5]. Major outbreaks of BPF occurred from 1984 [2] through 1990 [6], and sporadic cases also have been reported. At present, BPF is a disease requiring mandatory reporting in Brazil, because BPF agents may potentially lead to new outbreaks [7]. In fatal cases of BPF, hemorrhage occurred in the skin, lungs, and adrenal glands, and histopathologic examination revealed hemorrhage, intravascular microthrombi, and necrosis in various organs, such as the upper dermis, renal glomeruli, lungs, and hepatic sinusoids [1, 7].

Little is known of the determinants of bacterial virulence or the pathogenesis of infection involving H. influenzae biogroup aegyptius. Potential virulence factors associated with H. influenzae biogroup aegyptius, such as pilus proteins, membrane-associated proteins, extracellular proteins, lipooligosaccharides, capsular polysaccharides, and IgA1 protease, have been investigated in the infant rat model and/or in tissue culture, but these factors have not been consistently associated with the pathogenesis of BPF [8–13]. Hemagglutinins have frequently been associated with bacterial virulence, and it has been reported that 83% of H. influenzae biogroup aegyptius strains isolated from patients with BPF agglutinate human erythrocytes [2].

Studies of H. influenzae biogroup aegyptius strains from the BPF clone have mainly examined their bacterial structure, and little information is available with respect to their extracellular products [8]. To our knowledge, there have been no studies relating bacterial extracellular products to pathogenesis. In the present work, we investigate the role of hemagglutinins from the extracellular products of H. influenzae biogroup aegyptius in the pathogenesis of BPF.

Materials and Methods

Bacterial strains and extracellular products. The strains studied were F3283, F1946, F3039, F3031, F3033, and F3042 of H. influenzae biogroup aegyptius from confirmed cases of BPF [2]. In addition, the following non-BPF control strains were used: F1951 and F3119 (from the collection of the Instituto Adolfo Lutz, São Paulo); H. influenzae biogroup aegyptius strain KC1018 (isolated from a patient with conjunctivitis, from the collection of the Special Bacterial Pathogen Laboratory, Division of Bacterial Disease, Centers for Disease Control and Prevention, Atlanta); and H. influenzae type b NCTC7279 strain (National Collection of Type Cultures, London) isolated from a patient with meningitis. These bacterial strains were grown as described elsewhere [14] and centrifuged at 6800 g and 4°C for 15 min. The supernatant was recentrifuged at 27,000 g for 15 min and then at 150,000 g for 3 h, to remove membrane blebs [15] and other insoluble debris. The final supernatant, here termed “extracellular products,” was concentrated 20-fold by vacuum dialysis (cellulose tubing; Sigma-Aldrich) against 0.01 mol/L PBS (pH 7.20 containing 0.02% (wt/vol) sodium azide (PBS-A) and was dialyzed against PBS-A to remove molecules with molecular mass of <12–14 kDa [16].

Detection of hemagglutinating activity. Hemagglutinating activity [17] of whole bacteria (10 cfu/mL) and purified (0.7 mg/mL) and nonpurified (6 mg/mL) bacterial extracellular products was assayed in a hemagglutination test with use of fresh, human type O, Rh-negative erythrocytes. Twenty-five microliters of a 1% erythrocyte solution in PBS was added either to 25 µL of a serial dilution of whole bacteria or to purified or nonpurified extracellular products in U-bottomed microtiter plate wells (Microtitre; Limber Chemical). The plates were held at room temperature for 60 min, and the hemagglutination titers were evaluated. A hemagglutination assay was also done with extracellular products after absorption with use of human type O, Rh-negative erythrocytes. Hemabsorption [18] was done with a 200-µL aliquot of extracellular products from H. influenzae biogroup aegyptius strain F3283 (6 mg/mL in PBS) and with a further purified hemagglutinating fraction (6 mg/mL), with addition of an equal volume of erythrocyte suspension and incubation overnight at 4°C. The suspension was then centrifuged at 12,000 g for 5 min. The hemagglutination assay was repeated 5 times with different extracellular products. Neutralization of the hemagglutinating activity was assayed with use of the monosaccharides d-mannose, d-glucose, d-galactose, and d-mannitol (5 mg/mL in saline) as follows: 25 µL of a serial 2-fold dilution of the sugars was pipetted into plastic microplates, and 25 µL of extracellular products (1-:8) was added to each well and incubated for 15 min at 25°C. Twenty-five microliters of human type O, Rh-negative erythrocytes was added, plates were incubated at room temperature for 30 min, and the hemagglutinating activity was examined.

Separation of the hemagglutinating fraction. The hemagglutinating fraction from H. influenzae biogroup aegyptius strain F3283 was analyzed by size exclusion chromatography. In brief, 5 mL of the bacterial extracellular products was applied to a 2.5×90-cm column of Sephadex G200 Superfine (Amersham Pharmacia Biotech). The column was calibrated with use of the molecular mass markers aldolase (158 kDa), bovine serum albumin (67 kDa), ovoalbumin (43 kDa), and chymotrypsinogen A (25 kDa) (Amersham Pharmacia Biotech). The effluent was collected in 1-mL samples, and protein content was measured at 280 nm [19]. Fractions corresponding to the protein peaks were pooled, lyophilized, and stored at 4°C until use.

Identification of the hemagglutinating fraction. The hemagglutinating fraction was identified by analyzing the extracellular products of the bacterial strains before and after absorption with human type O, Rh-negative erythrocytes, by SDS-PAGE (Sigma-Aldrich) followed by an immunoblot assay [20]. Proteins in sample buffer (2% SDS, 2% 2-mercaptoethanol, and 20% glycerol in 1.25 mol/L Tris hydrochloride [pH 6.8]) were boiled for 3 min, submitted to 10% SDS-PAGE, and stained with Coomassie blue [21]. The immunoblot assay [20] was done with an antiserum against the extracellular products of H. influenzae biogroup aegyptius strain F3283, as described below.

Production of antibody to extracellular products of H. influenzae biogroup aegyptius for immunoblot assay. New Zealand White rabbits weighing ∼3 kg were immunized intradermally with 1 mL (1 mg/mL) of nonpurified extracellular products of H. influenzae biogroup aegyptius strain F3283 in PBS containing Freund's complete adjuvant. The rabbits received a booster every 3 weeks with the same immunogen prepared in Freund's incomplete adjuvant. The increase in antibody titer was checked 7–10 days after each booster by use of an immunodot assay [22] and the same batch of nonpurified extracellular products. After the final test, antisera were also tested by immunoelectrophoresis [23] with nonpurified extracellular products from the studied strains; their respective soluble cell lysates were used as controls. The antisera were titered, and a 1:320 dilution was considered to be optimal for the immunoblot assay [20].

Inoculation of animals. Three groups of 3 male New Zealand rabbits, each weighing ∼3 kg, were used. Rabbits were maintained in individual cages and were allowed free access to food and water. All procedures in this study complied with the ethical guidelines proposed by the institution for experiments with animals, and all experiments reported here were conducted according to the principles set forth in guidelines of the Institute of Laboratory Animals Resources of the National Research Council. Each group received a different solution intravenously. Group 1 received 1 mL of extracellular products containing 6 mg of protein/mL in saline; group 2 received the same amount of purified hemagglutinating fraction; group 3 was injected with 1 mL of sterile, pyrogen-free saline. Twenty-four hours after the intravenous inoculations, the animals were anesthetized by an intravenous injection of barbitone-5,5-diethyl acid (Sigma-Aldrich) and killed. Animals were autopsied, and macroscopic alterations of the body surface and inner organs were recorded. This experiment was repeated twice.

Histopathologic analysis. Tissue specimens were taken from the kidneys, liver, lungs, and spleen of each animal, fixed in 0.01 mol/L phosphate-buffered 10% formalin (pH 7.4), embedded in paraffin, processed for histopathologic analysis by routine methods, and stained with hematoxylin-eosin.

Results

When whole bacteria were used, H. influenzae biogroup aegyptius strains isolated from patients with BPF and from patients without BPF showed hemagglutinating activity. However, when the extracellular products alone were used, only strains from patients with BPF presented hemagglutinating activity (table 1). The hemagglutinating activity of H. influenzae biogroup aegyptius strain F3283 could be neutralized by absorbing its extracellular products with erythrocytes. However, the hemagglutinating activity could not be inhibited by d-mannose, d-glucose, d-galactose, or d-mannitol.

Table 1.

Hemagglutinating activity of control strains and of Haemophilus influenzae biogroup aegyptius strains and their extracellular products from confirmed cases of Brazilian purpuric fever (BPF).

Analysis of the extracellular products by SDS-PAGE showed that H. influenzae biogroup aegyptius strain F3283 presented 38-kDa and 60-kDa molecules that were absent from the extracellular products of the control strains (KC1018 and NCTC7279) (figure 1). The immunoblot assay with rabbit antiserum against extracellular products from BPF strains revealed bands of 38, 43, 55, and 60 kDa. After absorption of the bacterial products with human O type erythrocytes, the 55- and 60-kDa bands were removed (figure 2).

Figure 1.

SDS-PAGE (10%) of extracellular products from Haemophilus influenzae biogroup aegyptius strain F3283 isolated from a patient with Brazilian purpuric fever (lane A); H. influenzae control strain KC1018 from conjunctivitis (lane B); H. influenzae type b control strain NCTC7279 (lane C); and culture medium (lane D). Nos. on the right indicate molecular weight in kilodaltons.

Figure 2.

Immunoblotting of extracellular products from Haemophilus influenzae biogroup aegyptius strain F3283 before (lane a) and after (lane b) absorption with human type O erythrocytes with antibody to extracellular products from Brazilian purpuric fever strain F3283 in rabbit serum, 12.5% SDS-PAGE, and transfer to nitrocellulose for blotting.

The extracellular products from strain F3283 were fractionated by size exclusion chromatography, resulting in 3 protein peaks I, II, and III (figure 3, top). Peak I corresponded to a 60-kDa protein without other protein contamination (figure 3, bottom). Compared with the corresponding band from whole extract analysis, peak I showed a slightly lower molecular mass. However, we believe it is the same molecule and that this slight difference may be due either to the different preparation of the material (only the purified fraction was lyophilized) or to contaminants associated with the molecule in the whole extract that slightly changed the molecular mass. The hemagglutinating activity was detected only in the first peak, corresponding to a 60-kDa protein, for which molecular mass was confirmed by SDS-PAGE (figure 3, bottom). This 60-kDa molecule was removed by absorption with human O type erythrocytes (data not shown).

Figure 3.

Top, Gel filtration chromatography (Sephadex G200) of extracellular products from Haemophilus influenzae biogroup aegyptius strain F3283. Molecular mass standards of 67 and 43 kDa were used to calibrate column. Bottom, Purified hemagglutinating fraction, corresponding to first left peak, submitted to 10% SDS-PAGE: lane A, material from peak 1; lane B, whole extracellular product from Brazilian purpuric fever strain. Nos. on the right indicate molecular weight in kilodaltons.

To study the in vivo pathogenic effect, rabbits were inoculated intravenously with either extracellular products, chromatographically purified hemagglutinating factor (fraction corresponding to the 60-kDa peak), or saline alone (3 animals/group). Those animals injected with bacterial extracellular products or purified hemagglutinating factor were killed 24 h after challenge, corresponding to 18 h after the initial signs (i.e., elevated body temperature). The control animals that received saline alone were killed at same time and presented no symptoms or lesions of the body or inner organs.

The animals from the groups injected with bacterial products presented both macroscopic and microscopic alterations. The most striking finding was ecchymosis due to suffusions on parts of the body surface, which was clearly seen when the skin was removed (figure 4). This manifestation was more widely disseminated and intense in all the animals injected with whole bacterial extracellular product.

Figure 4.

Dorsal surface of rabbits showing hemorrhages after skin removal. Rabbits were injected with purified hemagglutinating fraction of extracellular product from Haemophilus influenzae biogroup aegyptius strain F3283 (A), whole extracellular product from H. influenzae biogroup aegyptius strain F3283 (B), or medium (C).

Microscopic alterations were present in all animals fromboth groups and included congestion, focal hemorrhage, and fibrin thrombi (figure 5). These alterations were of similar intensity in both groups and were present in at least 2 of the organs examined (liver, kidneys, lungs, and spleen).

Figure 5.

Histopathology of kidney of a rabbit injected with purified extracellular product from Haemophilus influenzae biogroup aegyptius strain F3283. Kidney section shows focal hemorrhage. Hematoxylin-eosin staining; original magnification, ×40.

Discussion

Bacteria from both BPF and non-BPF strains are known to agglutinate red blood cells [4]. Hemagglutinins have been found in other bacteria, such as Bordetella pertussis [24] and Vibrio cholerae [25], in both whole bacterial preparations and their extracellular products. In the present study, we analyzed the hemagglutinating activity of extracellular products. All extracellular products from H. influenzae biogroup aegyptius strains from BPF cases showed hemagglutinating activity that has not been reported before. No hemagglutinating activity was observed in the control non-BPF strains, although hemagglutinins may have been present in very low concentrations, undetectable by the method used. Cell-associated hemagglutinins may lose activity during storage; however, in our study, both control and BPF strains were maintained under similar conditions for the same period in the culture collection, and thus this possibility is unlikely. The extracellular hemagglutinins from BPF clone F3283 were not inhibited by sugars such as d-mannose, d-glucose, d-galactose, and d-mannitol, which suggests that these are not lectin-like components. Further studies are in progress to test for other simple and complex carbohydrates.

Protein bands of 38 and 60 kDa were present in the extracellular products from strain F3283 that were not present in those from non-BPF clones KC1018 and NCTC7279 when assayed by SDS-PAGE. An extracellular 38-kDa protein has been described from BPF clones [7]; however, when we analyzed the gel filtration eluates, only the first peak showed hemagglutinating activity. A 60-kDa component was identified by SDSPAGE in this peak.

The immunoblot assay with anti-H. influenzae biogroup aegyptius extracellular products revealed 4 antigenic bands of 60, 55, 43, and 38 kDa; the 60- and 55-kDa bands disappeared after hemabsorption. The 55-kDa band was not seen by SDS-PAGE on gel filtration, which may be related to the assay sensitivity.

The pathogenesis of BPF is poorly known. Some studies have focused on the virulence of BPF clones, revealing, in the infant rat, a role for a 145-kDa surface protein linked to the intravascular survival of the bacteria, maintaining bacteremia [10]. Lipooligosaccharides also have been related to virulence of H. influenzae biogroup aegyptius [11]. However, the disease is characterized by widespread hemorrhage and intravascular microthrombi in different tissues, which is suggestive of disseminated intravascular coagulation [1, 8]. Some studies have shown an in vitro cytotoxic effect of BPF clones on endothelial cells that detach and aggregate in the presence of bacteria [12, 13]. Although vascular wall damage may be related to hemorrhage, the consequences are usually more restricted than is observed in BPF cases. In this study, we observed initially a hemagglutinatingeffect of the extracellular products of BPF clones. When either bacterial extracellular products or a chromatographically purified, 60-kDa hemagglutinating factor were injected intravenously in rabbits, we observed hemorrhage on the body surface macroscopically and congestion, focal hemorrhage, and fibrin thrombi microscopically in various organs, as occurs in cases of disseminated intravascular coagulation. The more-intense effect in rabbits of the nonpurified hemagglutinating extracellular product, compared with that of the purified fraction, suggests that the 60-kDa protein is not the only factor responsible for the alteration, although it is certainly the most important because it induced significant alterations alone. Although the 38-kDa extracellular protein [8] revealed in this study does not promote hemagglutination, it may be a cofactor in the development of purpura. Lipooligosaccharides may be involved in pathogenesis, but these were probably absent from our preparation of extracellular products because the dialysis step during processing would have removed all low-molecular-mass molecules of <14 kDa. To our knowledge, this is the only study to show the effect of a BPF clone-derived factor that induces alterations very similar to those seen in human cases. These data suggest that the hemagglutinating extracellular product of the BPF clone, specifically the 60- kDa protein, is the main pathogenic factor linked to the hemorrhagic manifestations of BPF.

Acknowledgments

We thank Venâncio A. F. Alves (Instituto Adolfo Lutz) for assistance in the histopathologic analysis.

Financial support: Conselho Nacional de Pesquisa (grant 30123396-9 to H.G.).