INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bartonella henselae is a fastidious gram-negative bacillus that is capable of causing a wide variety of disease syndromes. The most common B. henselae-associated diseases (BAD) are cat-scratch disease, which is most often seen in immunocompetent children, and bacillary angiomatosis (BA), which is common in AIDS patients and other immunosuppressed individuals. BA is a proliferative disorder of the vascular endothelial cells resulting in the development of tumor-like lesions on the skin and internal organs (36). BA is caused by both B. henselae and Bartonella quintana (17, 32, 38, 39). Bartonella-induced vascular proliferation was first attributed to infection with Bartonella bacilliformis (14). This proliferation is manifested as the eruptive phase (verruga peruana) of Carrion's disease, which is endemic to Peru (26). Although the epidemiology of diseases caused by different Bartonella spp. differs, the ability of Bartonella spp. to induce angiogenic lesions in infected patients represents a common mechanism of pathogenesis that is unique to this genus.

B. henselae is a facultative intracellular bacterium which has been shown to attach and invade human endothelial cells (12). Surface pili are thought to play an important role in the initial attachment to host cells, since nonpiliated strains are less invasive (5). In general, primary isolates of B. henselae are thought to be more heavily piliated than isolates that have been extensively cultivated on laboratory media (5). Invasion of human endothelial cells has been shown to occur with large aggregates of bacteria as well as with individual organisms (12). However, some reports indicate that B. henselae may induce endothelial cell proliferation independent of bacterial invasion (8, 21). The factor from B. henselae that is responsible for causing endothelial cell proliferation and ultimately angiogenesis has not yet been identified. In addition, the mechanism by which the B. henselae factor is delivered to endothelial cells to mediate proliferation has not been elucidated.

Recently an operon has been identified in B. henselae that is homologous and colinear to the virB operon of Agrobacterium tumefaciens (22, 29). The B. henselae virB genes share sequence homology to the virB genes of A. tumefaciens; however, in B. henselae the virB5 gene is replaced by the gene encoding the immunodominant 17-kDa antigen found only in Bartonella spp. (22). Despite the presence of high levels of antibody to the 17-kDa antigen in sera from most patients with Bartonella infections, minimal reactivity with this protein was observed on Western blots using B. henselae cultivated on cell-free laboratory medium (2). This suggests that the 17-kDa protein may not be expressed at high levels when grown on routine culture medium. Because of the importance of the virB operon in (i) intracellular survival for other bacteria, (ii) delivery of effector molecules to host cells mediating pathogenesis in other bacteria, and (iii) apparent low-level expression of the 17-kDa antigen in B. henselae bacteria cultivated on laboratory medium, the induction of this operon in the intracellular environment of host cells was examined.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Growth conditions for cell lines and bacterial strains. Human microvascular endothelial cells (HMEC-1) (1) were cultured in MCBD131 cell culture media (Gibco/BRL, Grand Island, N.Y.) supplemented with 10% fetal bovine serum, 10 ng of epidermal growth factor/ml, and 1 µg of hydrocortisone/ml. Escherichia coli strains were grown using Luria-Bertani broth or agar (Difco, Detroit, Mich.) with appropriate antibiotics (ampicillin, 100 µg/ml; kanamycin, 50 µg/ml). B. henselae strains were cultivated on heart infusion agar (Difco) supplemented with 5% rabbit blood or 1% bovine hemoglobin (chocolate agar) at 37°C in 5% CO₂. For some experiments, B. henselae strains were incubated with supplemented MCBD131 at 37°C in 5% CO₂. Antibiotic concentrations for cultivation of B. henselae strains were 200 µg/ml for streptomycin and 50 µg/ml for kanamycin where indicated. Plasmids and bacterial strains are described in detail in Table 1.

Cloning of the virB promoter region. To define potential promoter sequences of the B. henselae virB operon, a 422-bp virB2 DNA probe (derived from the 5'-most end of the virB operon) was constructed by PCR amplification of B. henselae Houston-1 genomic DNA using the primers VB2F and VB2R (29). The probe was digoxigenin-dUTP labeled for use in Southern blotting as described in the Genius System protocol (Boehringer Mannheim, Indianapolis, Ind.). Hybridized probe was detected using the Phototope-Star chemiluminescence detection kit (New England Biolabs, Beverly, Mass.). A 4.0-kb EcoRI band of B. henselae was identified by Southern blotting and targeted for cloning into pUC19. Potential E. coli clones of the 4.0-kb EcoRI fragment were screened by colony blotting using the virB2-digoxigenin probe under the same conditions as described above.

DNA sequencing. To find the nucleotide sequence of the region upstream of the virB operon, internal and M13 universal primers were used to sequence the insert of pICB.4D7 (Table 1). DNA sequencing was performed by ³⁵S-dATP labeling with the dideoxynucleotide termination system using Sequenase T7 DNA polymerase (Amersham Life Science, Cleveland, Ohio). DNA sequences were recorded and analyzed using DNasis software, version 2.5 (Hitachi, San Bruno, Calif.). The putative promoter inserts of the lacZ (pCB182) and green fluorescent protein (GFP) (pANT3) reporter constructs (described below) were determined by the same method using vector-specific primers flanking the inserted promoter region.

Promoter constructs. To determine if the region upstream of the virB operon has promoter activity, a 362-bp fragment of DNA amplified by PCR using primers PVirF and PVirR (Fig. 1) constructed with XbaI sites was first cloned upstream of the lacZ reporter gene in pCB182 (30). Promoter orientations were confirmed by sequencing, and both the forward (pVirBF) and reverse (pVirBR) directions were obtained after transformation of E. coli CB454 (Table 1).

View larger version (30K): [in this window] [in a new window]   FIG. 1.   DNA sequence of the promoter region of the B. henselae virB operon. The vir box sequence of A. tumefaciens (above) and the homologous region in B. henselae are boxed (where R = A or G; W = A or T; D = A, G, or T; H = A, C, or T; and Y = C or T). The primers used to generate the reporter constructs are in bold. The virB2 start codon is located at position 352. The B. henselae nucleotide sequence is numbered as in GenBank.

Transformation of B. henselae 882Str. Transformations of B. henselae were performed as previously described (25). Briefly, 3-day-old B. henselae 882Str grown on streptomycin chocolate agar was harvested and washed three times in ice-cold 10% glycerol and resuspended in a 100-µl total volume at approximately 10⁹ CFU/ml. To 40 µl of the B. henselae 882Str suspension 1.0 µg of plasmid DNA, derived from E. coli JM109 harboring GFP reporter constructs, was added, and the mixture was transferred to precooled 0.1-cm-gap-width cuvettes (Bio-Rad, Hercules Calif.). Cells were electroporated using a Bio-Rad Pulse Controller II for 4.6 ms with field strength equivalents of 12.5 kV/cm and a constant capacitance of 25 µF. Samples were placed in 1 ml of recovery broth (RPMI 1640 with glutamine, 1% HEPES buffer, 1% sodium pyruvate, 5% fetal calf serum [heat inactivated], 5% rabbit blood lysate) and incubated at 37°C with 5% CO₂ for 7 h. Transformants were selected by growth on heart infusion agar-5% rabbit blood with kanamycin. Confirmation of the transformants as B. henselae 882Str was accomplished by extracting total DNA from each of the putative clones followed by PCR amplification using the B. henselae htrA gene primers CAT1 and CAT2 as previously described (3). Amplification of the kanamycin gene of pANT3 was also performed to confirm the presence of the plasmid using the primers TN903KN1 (5'-CCGATGCGCCAGAGTTTCTGAA-3') and TN903KN2 (5'-ACCTATTAATTTCCCCTCGTCAAAA-3') (25).

Flow cytometry. All samples were run on a FACScan flow cytometer (Becton Dickinson) and analyzed with Cell Quest software (Becton Dickinson) to examine both immunostained bacteria and bacteria expressing GFP. For GFP production, B. henselae 882Str reporter constructs were cultured with HMEC-1 or in MCDB131 alone for 24 h at 37°C. Bacteria cultivated with HMEC-1 were treated with gentamicin (200 µg/ml) for 2 h to kill extracellular bacteria, and cells were harvested using 2 mM EDTA. Titration experiments followed by plating showed that this treatment was sufficient to kill all cell-free B. henselae. Infected HMEC-1 cells were lysed in 0.1% saponin, cell debris was collected by low-speed centrifugation, and the bacteria in the supernatant were subjected to flow cytometry. Suspensions of both cocultivated and cell-free bacteria were first differentiated using forward and side scatter so that gating and analysis would not include any remaining HMEC-1 cells or debris. The gated, viable bacteria were observed on a histogram plot for the presence of GFP (FL1) versus the bacterial cell counts. The detection settings for the FL1 channel were set with respect to the control samples. In the final analysis, histogram plots for the samples were overlaid for GFP expression.

Western blotting. B. henselae Houston-1 was incubated in cell culture media alone or with HMEC-1 for 24 or 48 h at 37°C with 5% CO₂. HMEC-1 cocultures were treated with gentamicin, as described for flow cytometry, and infected cells were harvested in 2 mM EDTA and lysed in 0.1% saponin. Cell debris was collected by low-speed centrifugation. Bacteria remaining in the supernatant from the HMEC-1 lysate and cell-free bacteria were lysed in 1× sample buffer (Novex, San Diego, Calif.). Samples were used for Western blotting by standard methods (40). Rabbit polyclonal serum to the 17-kDa antigen (diluted 1:200) was used and has previously been shown to recognize the 17-kDa antigen of B. henselae, Bartonella clarridgeiae, B. quintana, and Bartonella elizabethae (37). Immunoreactive bands were detected using goat anti-rabbit antibody conjugated with horseradish peroxidase (KPL, Gaithersburg, Md.) diluted 1:5,000 and reacted with the 3,3',5,5'-tetramethylbenzidine (TMB) substrate (KPL).

Labeling of B. henselae Houston-1 with FITC. B. henselae Houston-1 was cultured with HMEC-1 or in cell culture media alone for 4 days at 37°C. Cultures to be analyzed by flow cytometry were gentamicin treated, harvested, and lysed as described for the GFP samples. Both cell-associated and cell-free bacteria were incubated with either normal rabbit sera (NRS) or rabbit anti-17-kDa-antigen serum diluted 1:50 in PBST-BSA (136 mM NaCl, 2.7 mM KCl, 1.8 mM KH₂PO₄, 10.1 mM Na₂HPO₄, 0.5% bovine serum albumin [BSA], 0.1% Tween 20) for 1 h at 37°C. Samples were washed four times in PBST (without BSA). Bacteria were collected after each wash by centrifugation at 3,000 × g. Samples were incubated with goat anti-rabbit fluorescein isothiocyanate (FITC) conjugate (KPL) diluted 1:10 in PBST-BSA for 1 h at 37°C. Samples were washed four times in PBST with centrifugation and analyzed by flow cytometry.

Digital imaging of B. henselae 882Str GFP reporter constructs and FITC-labeled B. henselae Houston-1. B. henselae bacteria expressing GFP or bacteria labeled with specific rabbit antiserum were examined by fluorescence microscopy and analyzed by digital imaging. B. henselae 882Str GFP reporter constructs or Houston-1 bacteria were incubated with HMEC-1 or in media alone on eight-well cell culture slides (Nunc, Inc., Naperville, Ill.). Extracellular bacteria were killed with gentamicin as for flow cytometry. Samples were fixed with 2% paraformaldehyde and either reacted with rabbit sera (described above) or analyzed directly for GFP. DNA was stained with antifade-4,6 diaminido-2 phenylindole (DAPI) (Vector, Burlingame, Calif.). Images of B. henselae 882Str GFP reporter constructs and B. henselae Houston-1 bacteria labeled with rabbit immune sera and FITC were captured using a Leitz Orthoplan 2 microscope with a charge-coupled capture device and the Smart Capture program (Vysis, Downer's Grove, Ill.) as previously described (25). GFP induction or FITC labeling was measured by pixel densities and reported as a ratio of GFP/DNA or FITC/DNA.

Nucleotide sequence accession number. The DNA sequence of the promoter region of the B. henselae virB operon was appended to the existing B. henselae virB sequence deposited in GenBank under accession number U23447.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of the putative virB promoter sequence. Eight clones were identified that hybridized with the 422-bp virB2 probe. All eight clones contained an insert of approximately 4.0 kb that hybridized to the virB2 probe by Southern blotting (data not shown). Nucleotide sequencing upstream of the virB2 gene was performed, and analysis of the region revealed a 13-bp sequence located 259 nucleotides upstream of the start codon of virB2 that has homology to the vir box of A. tumefaciens (Fig. 1). The vir box sequence of A. tumefaciens has been shown to act as the 35 region and is necessary for transcription of the virB operon (9, 15). The B. henselae sequence aligns and matches 12 of the 14 residues (Fig. 1) found in the A. tumefaciens vir box consensus sequence, RWTDCAATTGHAAY (where R = A or G; W = A or T; D = A, G, or T; H = A, C, or T; and Y = C or T) (9). This putative promoter region was further analyzed in both E. coli and B. henselae using reporter gene constructs.

Induction of the GFP reporter gene by the B. henselae virB promoter region. To determine if the region upstream of the virB operon has promoter activity, the 362-bp region defined by PvirF and PvirR (Fig. 1) was cloned upstream of the lacZ reporter gene of pCB182 in both the forward and reverse orientations. The new constructs (pVirBF or pVirBR) were used to transform E. coli CB454, and -galactosidase assays were performed (43). The B. henselae htrA promoter and the E. coli lacZ promoter were used as positive controls, and pCB182 without an insert served as the negative control. Neither pVirBF or pVirBR directed transcription initiation of the -galactosidase reporter gene to significant levels in E. coli CB454 compared to htrA or lacZ promoters (data not shown). This was not surprising, because the vir promoters of A. tumefaciens require the presence of an activator protein (VirG) for induction of the promoter (20). Transcription initiation from the A. tumefaciens virB promoter in E. coli has been reported to be inefficient in the absence of this transcription factor (20). Since similar transcription factors may be required for the B. henselae virB expression, the promoter assays were performed directly with B. henselae.

View larger version (39K): [in this window] [in a new window]   FIG. 2.   Flow cytometry of GFP reporter constructs B. henselae pANT3/882Str (A), B. henselae pVBGFPR/882Str (B), and B. henselae pVBGFPF/882Str after coculture with HMEC-1 for 24 h (C). Samples were analyzed by flow cytometry as described in Materials and Methods. The geometric mean of the fluorescence intensity for each of the samples is designated by the x value.

Digital analysis of GFP induction by B. henselae promoter constructs. All B. henselae reporter constructs were incubated with HMEC-1 cells or in media alone on eight-well culture slides and analyzed by digital imaging as described in Materials and Methods. Coculturing of B. henselae pVBGFPF/882Str with HMEC-1 cells resulted in intracellular aggregates of bacteria producing greater amounts of GFP (Fig. 3F) than when they were incubated in media alone (Fig. 3E). B. henselae pANT3/882Str (Fig. 3A and B) and B. henselae pVBGFPR/882Str (Fig. 3C and D) produced little or no GFP in either the presence or absence of HMEC-1 cells. Intensely fluorescing pockets containing large numbers of B. henselae bacteria can be seen within HMEC-1 cells (Fig. 3F). These large aggregates of intracellular bacteria surrounded by a host cell membrane, termed invasomes, have been well documented (11, 12). To normalize reporter gene fluorescence to the amount of bacterial DNA, gfpmut3 expression is reported as a ratio of GFP (green) to DNA, visualized by DAPI staining (blue), as previously described (25). Intracellular B. henselae pVBGFPF/882Str from HMEC-1 cells produced almost ninefold more GFP than when incubated in media alone (Fig. 4C) and almost ninefold more than B. henselae pVBGFPR/882Str (Fig. 4B) and B. henselae pANT3/882Str (Fig. 4A) cultured in either media alone or with HMEC-1.

View larger version (53K): [in this window] [in a new window]   FIG. 3.   Fluorescence micrographs of B. henselae GFP reporter constructs. Panels A, C, E, and G show B. henselae GFP reporter constructs cultured in media alone. Panels B, D, F, and H show B. henselae GFP reporter constructs cocultured with HMEC-1. Reporter constructs shown are B. henselae pANT3/882Str (A and B), B. henselae pVBGFPR/882Str (C and D), B. henselae pVBGFPF/882Str (E and F), and B. henselae pVBDEL/882Str (G and H). Host cell and bacterial DNA is stained with DAPI. Overall magnification in each panel is ×400.

View larger version (17K): [in this window] [in a new window]   FIG. 4.   Digital analysis of fluorescence microscopy showing induction of GFP reported as a ratio of GFP (green fluorescence) to DNA (blue fluorescence). Promoter constructs were cultured in media alone and with HMEC-1 as indicated for 24 h before images were captured by digital fluorescence imaging. B. henselae clones were incubated in cell culture media alone () or cocultured with HMEC-1 (). B. henselae GFP reporter constructs analyzed were B. henselae 882Str/pANT3 (A), B. henselae pVBGFPR/882Str (B), B. henselae pVBGFPF/882Str (C), and B. henselae pVBDEL/882Str (D).

Upregulation of the VirB protein of B. henselae by HMEC-1. Although the promoter region of the virB operon can be activated when B. henselae is cultured with HMEC-1, the ability of endothelial cells to increase expression of VirB proteins was also investigated. The 17-kDa antigen was selected to be the marker of VirB protein expression because it is highly immunogenic in humans and the gene encoding this protein is in the center of the operon. To determine if there was any increase of expression of the 17-kDa antigen in intracellular B. henselae bacteria cultured with endothelial cells, a Western blot analysis was performed of cell lysates of B. henselae Houston-1 incubated in media alone (Fig. 5, lanes A and C) or with HMEC-1 (Fig. 5, lanes B and D) and reacted with rabbit anti-17-kDa antigen. At both 24 (Fig. 5, lanes A and B) and 48 (Fig. 5, lanes C and D) h, cocultured Houston-1 expressed two to three broad bands which reacted with the rabbit anti-17-kDa antigen serum that were absent in Houston-1 cultured in medium alone. The presence of at least two bands at this molecular mass is most likely due to the processed and unprocessed forms of the 17-kDa antigen that have been previously described (2).

View larger version (105K): [in this window] [in a new window]   FIG. 5.   Western blot of B. henselae Houston-1 cultured in media alone (lanes A and C) or with HMEC-1 (lanes B and D). Samples were cultured for 24 h (lanes A and B) or 48 h (lanes C and D). Molecular weight markers are listed at the left (given in thousands), and the various forms of the 17-kDa antigen are indicated by the arrows.

View larger version (12K): [in this window] [in a new window]   FIG. 6.   Digital analysis of fluorescence microscopy using rabbit antiserum and FITC labeling of B. henselae Houston-1 cultured in media alone () or with HMEC-1 () for 5 days. Cultures were reacted with NRS (A) or with rabbit serum specific for the 17-kDa protein (B). Samples were reacted with antiserum, stained with FITC, and examined by fluorescence microscopy. Digital images were captured and reported as a ratio of FITC (green fluorescence) to DNA (blue fluorescence).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Type IV secretion systems are elaborate mechanisms that some gram-negative bacteria utilize to export products outside the bacterial cell to target host cells. The type IV system is a piliated multiprotein channel that spans both inner and outer bacterial membranes and has been identified in a number of animal, plant, and human pathogens. This secretion system is able to transport diverse types of macromolecules including both DNA and protein. The type IV system of Brucella suis is coded by the virB operon and is required for survival and replication within the host macrophage (13, 31). The ptl operon is required for the exportation of the pertussis toxin of Bordetella pertussis (41), and the virB operon of A. tumefaciens, the most characterized of the type IV systems, is responsible for the delivery of the tumor-inducing DNA (T-DNA) to plant cells (for a review, see reference 7). The virB operon of B. henselae encodes 10 genes, of which 8 have significant homology to the components of the type IV secretion system of A. tumefaciens (22, 29).

Several of the genes of the B. henselae virB operon also contain functional domains that are found in the A. tumefaciens virB operon. The Walker A NTPase binding domain of VirB4 and VirB11 and the peptidase cleavage site of VirB2 are found in both organisms (29). Although VirB2 of A. tumefaciens is posttranslationally processed at this peptidase cleavage site (16, 18), no evidence of B. henselae VirB2 processing has been reported. It is interesting that the virB5 gene has been replaced by the gene coding for the immunodominant 17-kDa antigen of B. henselae. In A. tumefaciens both VirB2 and VirB5 are subunits of the conjugative pilus (28). The presence of a signal sequence in the 17-kDa antigen suggests that this protein is localized to the membrane of the bacterium (2). Strong immunoreactivity of patient sera to the 17-kDa antigen further strengthens the possibility that this protein is exposed on the outer surface of the bacterium, possibly in conjunction with pili.

A 14-bp section of the B. henselae virB promoter region shares homology to the 35 promoter sequence (vir box) of the A. tumefaciens vir promoters (Fig. 1). The lack of GFP production by B. henselae pVBDEL/882Str (Fig. 3 and 4) demonstrates that an 87-bp region containing this sequence is required for virB promoter activity in B. henselae. The vir box is highly characterized in A. tumefaciens and is present in each of the promoters of the Ti (tumor inducing) plasmid vir operons (10). The A. tumefaciens vir box acts as a 35 promoter sequence and facilitates the binding of phosphorylated VirG, which is the activator protein of a two-component regulation system (9, 15, 20, 23, 33). The binding of VirG to the vir box is required to allow the alpha subunit of RNA polymerase to bind, form the holoenzyme, and initiate transcription (20). In addition, the vir boxes of B. henselae and A. tumefaciens share an A-T-rich region in the center of the sequence which is common to activator binding regions of other two-component regulatory systems and is believed to be necessary in DNA-protein interactions (15). Further experiments are under way to determine the role of the putative vir box in transcription and intracellular induction of the virB operon of B. henselae.

The possibility of other promoter regions initiating transcription of individual or multiple virB genes has not been eliminated. However, no additional known promoter sequences upstream of the individual B. henselae virB genes have been described (2, 22, 29). The absence of sizable intergenic regions between most B. henselae virB genes, considered together with the colinear orientation of the B. henselae virB genes compared to the A. tumefaciens virB operon, suggests that the B. henselae virB operon is polycistronic and is transcribed from a single promoter.

In addition to the virB operon of A. tumefaciens, the regulation of other type IV secretion system operons has been well studied. Transcription of the ptl operon of Bordetella pertussis is under control of the ptx promoter, which is activated by a two-component regulatory system (27, 35). The ptl genes of Bordetella pertussis are immediately downstream of the ptx operon, which codes for the production of the pertussis toxin proteins (41). The promoter upstream of the ptx operon is regulated by the BvgA and BvgS two-component regulatory system (27). This same promoter and regulation system is also responsible for the transcription of the Bordetella pertussis ptl operon that codes for the type IV secretion system and is necessary to export the pertussis toxin from the cell (4). Both the ptx and ptl genes are cotranscribed as a single polycistronic mRNA from this promoter. The induction of GFP by the virB promoter region of intracellular B. henselae but not in media alone suggests that a similar activator system may also be at work in this organism. The inability of the B. henselae virB promoter sequence to induce a reporter gene in E. coli further strengthens the theory that a Bartonella-specific protein is needed for initiation of transcription. However, we have observed that the virB promoter is activated by other cell types including HeLa (epithelial origin), HGF-1 (fibroblast origin), and differentiated THP-1 (human macrophage), suggesting that induction is not endothelial cell specific (data not shown).

Coupled with the initiation of transcription by the virB promoter region, the expression of genes encoding potential surface proteins is also of interest. The 17-kDa antigen, encoded by a gene found in the middle of the virB operon of B. henselae, is expressed at low levels when grown on laboratory media. A faint immunoreactive protein at 17 kDa can be noted on Western blots of whole-cell lysates of B. henselae grown on laboratory media when they are reacted with sera from patients diagnosed with BAD. However, the 17-kDa band was considerably stronger in intracellular B. henselae Houston-1 from HMEC-1 (Fig. 5). The strongest evidence for the upregulation of the 17-kDa antigen gene by an external stimulus is the increased FITC labeling of intracellular B. henselae Houston-1 reacted with the 17-kDa-antigen-specific rabbit sera (Fig. 6). This increase of protein expression induced by human endothelial cells helps explain why the 17-kDa antigen is not expressed at high levels in bacteria grown on laboratory media yet antibodies to this protein appear at high titers in humans who have been diagnosed with BAD. This would suggest that the 17-kDa antigen is expressed by the bacterium at high levels during infection, and due to its probable presence on the cell surface it represents a good target for the humoral immune response.

It is not currently known what specific environmental stimulus is required to activate the virB operon. The results of each of the experiments described in this study show that the virB operon of B. henselae is activated in intracellular bacteria. The requirement for host cell invasion for virB activation has not yet been defined. However, experiments conducted using GFP constructs to infect HMEC-1 in the absence of gentamicin treatment resulted in the presence of multiple-cell-associated bacteria that did not exhibit fluorescence (unpublished data). Thus, invasion and not merely attachment appears to be required for induction of the virB operon of B. henselae, and it is likely that the intracellular environment of the host cell contains the specific stimulus which is necessary for induction.

B. henselae is able to adhere to and invade a variety of different cell types including epithelial and endothelial cells (5, 6, 12). However, the interaction of B. henselae with endothelial cells is most interesting and results in proliferation and ultimately angiogenesis. The proliferative activity of B. henselae has been reported as both a membrane-associated (8) and a secreted diffusible (21) factor. The specific B. henselae protein responsible for this proliferation has not yet been identified. Additionally, it is not known how the effector molecule is delivered to the host endothelial cell. If the proliferative factor is exported through a type IV secretion system, this factor may be present in membrane preparations at high enough concentrations to induce cell proliferation, as reported by Conley et al. (8), and could also be secreted from the bacterium, as reported by Maeno et al. (21). The delivery of the pertussis toxin and T-DNA by type IV secretion systems indicates that this system plays an important role in delivery of biologically significant effector molecules in other bacteria. Similarity of gene structure and regulation of the virB operon of B. henselae to those of other bacteria suggests that this operon may play a similar transport role in delivery of biologically important effector molecules to endothelial cells.

The specific function of each of the B. henselae virB genes is still unknown; however, the presence of an operon with significant homology to other type IV secretion systems suggests that some effector molecule is secreted from the bacteria into the environment or host cell. The expression of the virB operon is upregulated in intracellular bacteria and appears to require a Bartonella-specific component. Although the specific activator molecule is not known, the upregulation of the 17-kDa antigen demonstrates that this operon is expressed after stimulation by endothelial cells. It is tempting to couple this secretion system with the delivery of the angiogenic factor, because of the ability of endothelial cells to induce VirB expression and the ability of B. henselae to proliferate endothelial cells. Currently B. henselae virB mutants are being constructed to determine the role this operon plays in attachment, invasion, or proliferation of endothelial cells. Although the exact mechanism of angiogenesis is probably complex and requires the expression of a variety of proteins, the role of virB operon-encoded proteins in this process should be considered.