SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Supplementary material
  9. References

Bartonella henselae, the agent of cat-scratch disease and vasculoproliferative disorders in humans, is a fastidious facultative intracellular pathogen, whose interaction with macrophages and endothelial cells (ECs) is crucial in the pathogenesis of these diseases. However, little is known about the subcellular compartment in which B. henselae resides. Two hours after infection of murine macrophages and human ECs, the majority of B. henselae-containing vacuoles (BCVs) lack typical endocytic marker proteins, fail to acidify, and do not fuse with lysosomes, suggesting that B. henselae resides in a non-endocytic compartment. In contrast to human umbilical vein endothelial cells, bacterial death and lysosomal fusion with BCVs is apparent in J774A.1 macrophages at 24 h. This phenomenon of delayed lysosomal fusion requires bacterial viability, and is confined to the BCV itself. Using magnetic selection, we enriched for transposon-mutagenized B. henselae trapped in lysosomes of macrophages 2 h after infection. Genes affected appear to be relevant to the intracellular lifestyle in macrophages and ECs and include some previously implicated in Bartonella pathogenicity. We conclude that B. henselae has a specific capacity to actively avoid the host endocytic pathway after entry of macrophages and ECs, from within a specialized non-endocytic membrane-bound vacuole.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Supplementary material
  9. References

The fastidious, slow-growing and facultative intracellular pathogen Bartonella henselae typically causes cat-scratch disease (CSD), a self-limiting lymphadenopathy related to a cat scratch or bite (Anderson and Neuman, 1997). In immunocompromised individuals (e.g. HIV-infected patients), chronic B. henselae infection may cause tumorous proliferations of endothelial cells (ECs) in the skin and internal organs, referred to as bacillary angiomatosis (BA) and peliosis hepatis (PH) respectively (Slater et al., 1992; Adal et al., 1994). The intracellular niche for sustaining such chronic infections is, however, unknown.

Infection of ECs, monocytes and macrophages by B. henselae in vitro (Musso et al., 2001; Resto-Ruiz et al., 2002) results in inhibition of host cell apoptosis (Kirby and Nekorchuk, 2002; Kempf et al., 2005a), and in the secretion of vasculoproliferative compounds [e.g. vascular endothelial growth factor, adrenomedullin and interleukin-8 (Kempf et al., 2005b)] which may be responsible for the endothelial proliferation characteristic of BA and PH (Kempf et al., 2001; Resto-Ruiz et al., 2002). However, neither the cellular mechanisms underlying chronic Bartonella infections nor the bacterial pathogenicity factors involved in these processes are fully understood.

The ingestion and destruction of microorganisms is an important host defence that occurs in ‘professional’ phagocytes (macrophages and neutrophils) but also, to a lesser extent, in non-professional phagocytes such as endothelial and epithelial cells. However, once within the host cells, some intracellular pathogens control the fate of their membrane-bound compartments, circumventing host defences, and thus sustain chronic infections (reviewed in Scott et al., 2003).

Little is known of the exact compartment in which B. henselae resides, and bacterial genes implicated in intracellular survival in ECs and macrophages have not been identified. The VirB type IV secretion system (TIVSS) of Brucella spp. perturbs endosomal maturation in professional and non-professional phagocytic cells (Comerci et al., 2001; Delrue et al., 2001), with mutants either directly targeted to lysosomes for degradation, or unable to reach the endoplasmic reticulum (ER) to establish an intracellular replication niche. A VirB TIVSS is important in subverting multiple human EC functions related to cytoskeletal alterations and to inflammation, apoptosis and proliferation after invasion by B. henselae (Schmid et al., 2004; Schülein et al., 2005), although effects on intracellular localization are unknown.

We therefore attempted to delineate the intracellular niche of B. henselae within macrophages and ECs. Morphological observations using immunofluorescence and electron microscopy revealed that after uptake by J774A.1 macrophages and human umbilical vein endothelial cells (HUVECs), dead Bartonella are immediately transported along the endosomal pathway to fuse with lysosomes, whereas this is clearly delayed by live wild-type B. henselae. Using an unbiased magnetic screen to enrich for lysosomes, we identified four genes associated with intracellular lifestyle using a transposon library of B. henselae mutants. The range of mutants obtained by this approach suggests a complex process of intracellular persistence and avoidance of phagolysosomal killing by B. henselae.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Supplementary material
  9. References

Invasion of macrophages and ECs by B. henselae

Invasion and intracellular survival of B. henselae WT was quantified in J774A.1 macrophages and HUVECs, by testing the recovery of viable organisms after cell lysis at 1, 2, 3, 4, 5, 6, 8 and 24 h. This was greatest in J774A.1 at 2 h (0.71 ± 0.06% of the initial inoculum), however, B. henselae were poorly recovered from J774A.1 at 24 h (0.04 ± 0.01%). Identical experiments in ECs (not shown) were entirely consistent with published observations, and in contrast to J774A.1, B. henselae were hugely expanded at 24 h in HUVECs as previously shown (Kempf et al., 2000). Based on these data, subsequent results are primarily from experiments performed with an invasion time of 2 h.

Results of transmission electron microscopy (TEM) were also consistent with invasion and intracellular survival data over a 24 h period. Cells were infected with B. henselae for 2 h (Fig. 1A and C) or 24 h (Fig. 1B and D). In both cell types, at 24 h post infection, there was decrease in the number of extracellular bacteria and an increase in number of intracellular bacteria, consistent with other experiments using double immunofluorescence (not shown). Also consistent with our data (above) there was a significant change in morphology of bartonellae 24 h after infection showing clumps of degraded bacteria in J774A.1 (Fig. 1B). By contrast, intracellular Bartonella in HUVECs appeared viable by TEM at the same time point (Fig. 1D).

image

Figure 1. Transmission electron microscopy of J774A.1 and HUVECs infected for 2 h (A and C) or 24 h (B and D) with viable B. henselae WT. Bacteria are retained within membrane-bound compartments (arrows). There is a significant change in morphology of Bartonellae 24 h after infection showing clumps of degraded bacteria in J774A.1 (B) but not in HUVECs (D). Scale bar: 1 µm.

Download figure to PowerPoint

Bartonella henselae inhibits apoptosis and induces VEGF secretion in J774A.1 macrophages

J774A.1 cells were treated with the apoptosis-inducing agent and NF-κB inhibitor PDTC, as described earlier (Wahl et al., 2001; Kempf et al., 2005a), and/or infected with viable B. henselae. PDTC consistently decreased the viability of J774A.1 cells at 8 h to 55% ± 5% compared with untreated controls, and this was restored to 100% in B. henselae-infected cells (Fig. S1A). Moreover, infection with B. henselae resulted in a significantly increased (approximately sevenfold) VEGF secretion in J774A.1 macrophages (Fig. S1B). These data reveal that B. henselae inhibits apoptosis and triggers secretion of VEGF in murine macrophages, as has been shown in human monocytes (Kempf et al., 2005a).

Lack of typical endocytic markers in B. henselae-containing vacuoles (BCVs)

Electron microscopy of J774A.1 and HUVECs, 2 or 24 h after infection, revealed that all internalized bacteria were within membrane-bound compartments (Fig. 1). Confocal laser scanning microscopy (CLSM), in which extracellular bacteria were clearly distinguished from intracellular bacteria, revealed that only a minority of BCVs (J774A.1: 27 ± 7%; HUVEC: 16 ± 6%) had acquired late endosomal/lysosomal lysosome-associated membrane protein 1 (LAMP1) at 2 h (Figs 2C, 3C, 4B and C). The early endosomal marker proteins early endosome antigen 1 (EEA1; Figs 2D and 3D) and transferrin receptor (TfR; Figs 2E and 3E) were rarely (< 1.5%) detectable on BCVs, even early during infection (15–30 min; data not shown), and neither was detected on BCVs 2 h after infection of J774A.1 or HUVECs. Consistent with 24 h survival data and TEM in J774A.1, bartonellae were almost exclusively found in association with LAMP1 in J774A.1 at 24 h (92 ± 7%; Fig. 4B). By contrast to J774A.1, the percentage of BCVs accumulating LAMP1 in HUVECs was unchanged from 2 to 24 h (LAMP1: 21 ± 3%; Fig. 4C). Normal phagosome maturation in J774A.1 was demonstrated by pulse infection with Listeria innocua or zymosan for 1 h; more than 95% of all zymosan- and L. innocua-containing phagosomes were LAMP1-positive (Fig. 2H) and EEA1-negative (not shown) at 2 h. As expected, rapid and transient accumulation of EEA1 around zymosan or L. innocua-containing phagosomes was detected early (data not shown). Importantly, maturation of zymosan-containing phagosomes was unaffected by co-infection with live B. henselae (Fig. 2I and Fig. S2). Thus, the majority of bacteria-containing vacuoles did not contain typical endocytic markers 2 h after infection with live B. henselae, and the complete avoidance of lysosomal fusion by BCVs observed in HUVECs is apparently only delayed in J774A.1 cells.

image

Figure 2. Distribution of various endocytic markers in J774A.1 infected with B. henselae. Representative confocal images of J774A.1 infected for 2 h with live B. henselae gfp+ and stained with (A) TR dextran, (B) LysoTracker Red, (C) LAMP1, (D) EEA1 and (E) TfR. Note frequent colocalization (arrowheads) of the lysosomal glycoprotein LAMP1 (red) with (F) heat-killed B. henselae in J774A.1 infected for 2 h. (G) Colocalization was confirmed by x-z sectioning. In each figure, an overlay of all three channels is represented. Intracellular bacteria are green (FITC) and extracellular bacteria appear blue (FITC + Cy5). Intracellular B. henselae (green) that colocalize with intracellular markers (red) appear yellow. Normal phagosome maturation was confirmed by strong accumulation of LAMP1 (green) around more than 95% of (H) L. innocua (red) phagosomes (arrowheads). (I) Maturation of zymosan (red)-containing phagosomes was unaffected by co-infection with live B. henselae. Co-localization was counted positive quantified by CLSM (see main text for statistics). Scale bar: 8 µm.

Download figure to PowerPoint

image

Figure 3. Distribution of various endocytic markers in HUVECs infected with B. henselae. Representative confocal images of HUVECs infected for 2 h with live B. henselae gfp+ and stained with (A) TR dextran, (B) LysoTracker Red, (C) LAMP1, (D) EEA1 and (E) TfR. Note colocalization (arrowheads) of LAMP1 (red) with (F) heat-killed B. henselae in HUVECs infected for 2 h. (G) Colocalization was confirmed by x–z sectioning. In each figure, an overlay of all three channels is represented. Intracellular bacteria are green (FITC) and extracellular bacteria appear blue (FITC + Cy5). Intracellular B. henselae (green) that colocalize with intracellular markers (red) appear yellow. Colocalization was counted positive quantified by CLSM (see main text for statistics). Scale bar: 8 µm.

Download figure to PowerPoint

image

Figure 4. Viability of B. henselae is crucial in avoiding lysosomal fusion. (A) Viability in original B. henselae infecting inocula was measured using the Live/Dead BacLight kit. Viable bacteria are labelled green, dead bacteria are red. Data from eight independent experiments revealed that viable (v) B. henselae infection inoculums contained a mean of 15.9 ± 3.6% dead bacteria. Heat-killed (hk) B. henselae inocula (control) showed greater than 99.5% dead bacteria. (B) J774A.1 or (C) HUVECs were infected for 2 h or 24 h with viable or heat-killed B. henselae gfp+. The percentage of BCVs positive for TR dextran, LysoTracker Red or LAMP1 was quantified by CLSM (100–300 phagosomes scored for colocalization per experiment). Values given are the mean and standard deviation from three independent experiments. A minority of BCVs were accessible to TR dextran preloaded into lysosomes at 2 h (J774A.1: 26 ± 7%; HUVEC: 17 ± 5%). LysoTracker Red was mostly not sequestered in the B. henselae vacuole at 2 h (J774A.1: 28 ± 12%; HUVEC: 17 ± 2%). A minority of live BCVs were LAMP1-positive at 2 h (J774A.1: 27 ± 7%; HUVEC: 16 ± 6%). By 24 h, the majority of live B. henselae continued to avoid lysosomal and other acidified compartments in HUVECs (LAMP1: 21 ± 3%; TR dextran: 23 ± 4%; LysoTracker Red: 27 ± 6%) but not in J774A.1 (LAMP1: 92 ± 7%; TR dextran: 90 ± 7%; LysoTracker Red: 96 ± 4%). Heat-killing of B. henselae before infection promoted phagolysosome formation illustrated by heavy colocalization with TR dextran (90 ± 6%), LysoTracker Red (85 ± 1%) and LAMP1 (94 ± 6%) at 2 h. P < 0.05 for all experiments. *Significant difference compared with J774A.1 cells (P < 0.05). n.d. not determined.

Download figure to PowerPoint

Failure of BCVs to fuse with lysosomes or other acidified compartments

To further elucidate whether BCVs fused with lysosomes, the lysosomal network of J774A.1 and HUVECs was labelled by fluid-phase endocytosis with the soluble endocytic probe Texas Red (TR) dextran, prior to infection (Swanson, 1989; Fig. S3). TR dextran accumulated in vesicular structures in J774A.1 and HUVECs, which were distinct from BCVs (Figs 2A and 3A). The presence of the fluorescent marker in the minority of inclusions containing B. henselae (J774A.1: 26 ± 7%; HUVEC: 17 ± 5%; Fig. 4B and C) is consistent with the LAMP1 data (above), and confirms failure of normal phagosome-lysosome formation at 2 h. Also consistent, a lysosomotropic marker (LysoTracker Red DND-99), specific for acidified compartments, stained only a minority of BCVs (J774A.1: 28 ± 12%; HUVEC: 17 ± 2%) after infection with live B. henselae (Figs 2B, 3B, 4B and C). These data show that the majority of live B. henselae avoided lysosomal and other acidified compartments in macrophages or ECs 2 h after infection. This is still apparent at 24 h in HUVECs (TR dextran: 23 ± 4%; LysoTracker Red: 27 ± 6%; Fig. 4C) but not in J774A.1 (TR dextran: 90 ± 7%; LysoTracker Red: 96 ± 4%; Fig. 4B), consistent with LAMP1 and viability data above.

Killing of B. henselae before infection promotes normal phagolysosome formation

Morphological observations using immunofluorescence and electron microscopy show that phagosomes containing inert particles or dead microorganisms follow the endocytic route, to fuse with lysosomes (Pitt et al., 1992). Using Live/Dead BacLight and fluorescence microscopy, we determined the percentage of dead bacteria in B. henselae inocula to be 15.9 ± 3.6% (Fig. 4A). This loss of viability in infecting inocula offers a potential explanation for the variant populations observed above. We next showed that heat-killed B. henselae ingested by J774A.1 frequently colocalized with late endosomal/lysosomal LAMP1 (94 ± 6%; Figs 2F and 4B) and lysosomal TR dextran (90 ± 6%; Fig. 4B) at 2 h. Moreover, the acidotropic probe, LysoTracker Red, effectively stained phagolysosomes containing heat-killed B. henselae (85 ± 1%). Additionally, B. henselae killed by paraformaldehyde before infection were unable to avoid being targeted to macrophage lysosomes (97 ± 4% LAMP1-positive; see Fig. S4A). The small number of killed bacteria which entered HUVECs (< 1%) also colocalized with LAMP1 (Fig. 3F and Fig. S4). These data strongly suggest that viable, but not dead, B. henselae control the fate of their membrane-bound compartments.

Similarly, opsonization of viable B. henselae with anti-B. henselae antibodies before infection also led to a significant increase in the number of BCVs which colocalized with LAMP1 (J774A.1: 67 ± 6% vs. 27 ± 7% with untreated bacteria; HUVEC: 65 ± 7% vs. 16 ± 6%), and also with lysosomal TR dextran (J774A.1: 71 ± 11% vs. 26 ± 7%; HUVEC: 59 ± 6% vs. 17 ± 5%) or LysoTracker Red (J774A.1: 66 ± 9% vs. 28 ± 12%; HUVEC: 68 ± 2% vs. 17 ± 2%) at 2 h.

Viable B. henselae are not found in J774A.1 lysosomal fractions

We then attempted to retrieve viable B. henselae from iron-dextran-loaded J774A.1 lysosomes 2 h after infection. The power and utility of a recently published method (Pethe et al., 2004) was confirmed in the J774A.1 model by showing that iron-dextran particles fed to macrophages colocalized with lysosomal TR dextran and LAMP1 on microscopy (Fig. S5A and B), and that nearly all zymosan particles fed to macrophages were found in TR dextran-labelled lysosomal fractions eluted from the MiniMACS column after magnetic selection (Fig. S5C). Following infection of J774A.1 with B. henselae, fluorescence microscopy (Fig. S6) and cultivation on Columbia blood agar (CBA; not shown) revealed B. henselae to be almost solely within the unbound flow-through from the magnetized MiniMACS column, entirely consistent with fluid-phase and antibody labelling data, above. By contrast, the bound fraction was greatly enriched for TR dextran- and iron-dextran-labelled lysosomes, and almost all heat-killed B. henselae were eluted in this fraction. Results are summarized in Table S1.

Isolation of B. henselae mutants from phagolysosomes and functional characterization

To identify bacterial genes associated with the B. henselae intracellular lifestyle, we screened part of a transposon library of B. henselae (Riess et al., 2003). Iron-dextran preloaded J774A.1 were infected with a pool of 100 transposon-mutagenized B. henselae and mutant-containing phagolysosomes were magnetically isolated 2 h after infection. Viable bacteria retrieved from the bound (lysosome-enriched) fraction were subjected to 5 successive rounds of selection for further enrichment. After the final isolation step, 24 individual colonies were randomly picked, and tested by Southern blot analysis to identify families of related or identical mutants, of which the largest had 16 members (Table 1). Transposon insertion sites were identified by chromosomal sequencing. The most commonly identified region (PAK100) displayed 92% DNA identity to a putative virulence-associated protein (VapA5) identified in the B. henselae Houston-1 genome. Other homologies to B. henselae proteins included a putative haemin-binding protein (HbpD; 95% DNA identity), a d-serine/d-alanine/glycine transport protein (CycA; 97% DNA identity) and an unknown protein (98% DNA identity). Each mutant was isolated from independently created pools in two subsequent independent rounds of screening, to confirm the trafficking findings from the initial screening results. All four mutants (PAK96, PAK100, PAK120 and PAK124) were separately confirmed to be taken up normally by host cells (Fig. 5) but to traffic to acidified lysosomes after infection of J774A.1 and HUVECs (Fig. 6). As expected for bacteria which traffic to lysosomes, these strains all displayed significantly reduced viability in both macrophages and ECs (Fig. 5) although, importantly, none of the mutants showed reduced growth on complete solid media (CBA; data not shown) compared with wild type, and VEGF induction in J774A.1 macrophages following an 8 h infection was normal (Fig. 7), demonstrating that these mutants were not grossly affected in their viability or ability to induce an angiogenic host cell response.

Table 1. Bartonella henselae transposon mutants enriched from phagolysosomes.
B. henselae Marseille mutantDNA sequences flankingtransposon insertion pointsGenBankAccession no.LocusaInterpreted function(% identity)bFrequencyof isolationc
  • a

    . The ORF numbers (locus) refer to the Bartonella henselae str. Houston-1 genomic database (NC_005956).

  • b

    .blast was used to perform sequence similarity searches (http://www.ncbi.nlm.nih.gov/blast/).

  • c

    . Number of identical mutants identified by Southern blot after five successive enrichments, that were confirmed by chromosomal sequencing. 24 mutants were analysed (sequencing of 1 mutant not functional).

  • d

    . Number of identical mutants that were isolated in an independently created pool in a second independent screen.

PAK965′-CAAAGATATTTTGAAGCTGCGTAT-transposon-GCTGCGTATAAGCAACACCTCCTG-3′AY451966 (5′-end) AY451967 (3′-end)BH04810haemin-binding protein D (92%)2(2d)
PAK1005′-TACACATCTCAACCCTGAAGCTTG-transposon-GTTATAATCTGTTCGGTCGTACCA-3′AY451964 (5′-end) AY451965 (3′-end)BH14270virulence-associated protein (95%)16(13d)
PAK1205′-GCTATTATTGGTTGTGCCTTATAC-transposon-GCCTTATACTTTGTAATGCGTGCT-3′AY954249 (5′-end) AY954250 (3′-end)BH10880d-serine/d-alanine/glycine transport protein (97%)1(2d)
PAK1245′-ATGATTACCTAACCCCTCATAAAC-transposon-CTCATAAACAGGAGAGCCAAGGTT-3′AY451968 (5′-end) AY451969 (3′-end)BH13090unknown hypothetical protein (98%)4(3d)
image

Figure 5. Bacterial internalization and intracellular survival in macrophages or ECs. Wild-type B. henseale (WT) or one of four mutants (PAK96, PAK100, PAK120, PAK124) or respective complemented mutants (PAK254, PAK255, PAK256, PAK257), was deposited onto J774A.1 (A) or HUVECs (B) and incubated at 37°C for the periods of time indicated. Bacteria (% of inoculum) were calculated from gentamicin protection assays. Data are the averages of triplicate samples from two identical experiments, and error bars represent the standard deviations. Statistically significant differences compared with the wild type (*), and compared with the respective parent mutant strain (**) are indicated (P < 0.05).

Download figure to PowerPoint

image

Figure 6. LAMP1 colocalization in phagosomes containing B. henselae transposon mutants. (A) J774A.1 macrophages or (B) HUVECs were infected for 2 h with zymosan (zym), live or heat killed (hk) wild-type B. henseale (WT) or one of four mutants (PAK96, PAK100, PAK120 and PAK124) or respective complemented mutants (PAK254, PAK255, PAK256 and PAK257), and then stained for LAMP1. The samples were visualized by CLSM. Data refer to the percentage of internalized bacteria that showed co-staining with LAMP1, based on observations of at least 100 bacteria per coverslip. Data are the averages of triplicate samples from three identical experiments. Statistically significant differences compared with the wild type (*), and compared with the respective parent mutant strain (**) are indicated (P < 0.05). Control mutant strains containing the empty vector (no insert) only (PAK250, PAK251, PAK252 and PAK253; Table 2) were as expected and did not deviate significantly from the original mutants (data not shown). n.d. not determined.

Download figure to PowerPoint

image

Figure 7. Induction of VEGF secretion in macrophages upon infection with B. henselae. J774A.1 were infected with B. henselae WT or the B. henselae mutants PAK96, PAK100, PAK120 or PAK124 respectively (MOI 100). Supernatants were taken 8 h after infection and analysed by ELISA. PMA (25 ng ml−1) was used as a positive control. *Significant difference compared with control (P < 0.05).

Download figure to PowerPoint

Complementation of B. henselae mutants in trans

Genetic complementation of the mutants PAK96, PAK100, PAK120 and PAK124 was performed with plasmid constructs carrying the respective wild-type gene including the putative promoter region (Table 2). The ability of the functional gene to complement each mutant in trans was assessed by intracellular trafficking studies in J774A.1 and HUVECs. Confocal microscopy revealed that genetic complementation restored the ability to arrest phagosome maturation at least partially in all but PAK257 (Fig. 6). Consistent data were obtained in invasion and intracellular survival assays; complementation of the mutants with the exception of PAK257 restored viability of Bartonellae post infection in macrophages and ECs (Fig. 5).

Table 2.  Bacterial strains and plasmids used in this study.
Strain or plasmidGenotype or descriptionReference or source
Strains
B. henselae
 Wild type (WT)B. henselae Marseille, patient isolate, early passage, expresses BadADrancourt et al. (1996)
 gfp+Marseille WT carrying pCD354 expressing GFPmut2, early passage, expresses BadA; KanrThis study
 PAK96Marseille transposon mutant, transposon integrated within 1.210 kb gene region; KanrThis study
 PAK100Marseille transposon mutant, transposon integrated within 1.138 kb gene region; KanrThis study
 PAK120Marseille transposon mutant, transposon integrated within 1.693 kb gene region; KanrThis study
 PAK124Marseille transposon mutant, transposon integrated within 1.205 kb gene region; KanrThis study
 PAK250PAK96 carrying pBBR1MCS; CmrThis study
 PAK251PAK100 carrying pBBR1MCS; CmrThis study
 PAK252PAK120 carrying pBBR1MCS; CmrThis study
 PAK253PAK124 carrying pBBR1MCS; CmrThis study
 PAK254PAK96 complemented with pPK04; Kanr CmrThis study
 PAK255PAK100 complemented with pPK05; Kanr CmrThis study
 PAK256PAK120 complemented with pPK06; Kanr CmrThis study
 PAK257PAK124 complemented with pPK07; Kanr CmrThis study
 ATCC 49882B. henselae‘Houston-1’ isolated from human blood; published genome sequence (NC_005956)Regnery et al. (1992); Alsmark et al. (2004)
E. coli
 TOP10Host strain used for cloning. F-mcrA Δ(mrr-hsdRMS-mcrBC) Φ80lacZΔM15ΔlacX74 Invitrogen
 recA1 araD139 Δ(ara-leu)7697 galU galK rpsL (StrR) endA1 nupG 
 DH5αHost strain used for cloning. F-φ80lacZΔM15 Δ(lacZYA-argF)U169 recA1 endA1 hsdR17(rk, mk+) phoA supE44 thi-1 gyrA96 relA1 tonAInvitrogen
 MOSBlue cellsHost strain used for cloning. endA1 hsdR17(rk12– mk12+)supE44 thi −1 recA1 Amersham
 gyrA96 relA1 lac[F’proA+B+ lacIqZΔM15:Tn10(TcR)] 
Listeria innocua
 ATCC 33090L. innocua Serovar 6a (DSM 20649)DSMZ, Braunschweig, Germany
Plasmids
 pCD354Encodes GFPmut2; KanrDehio et al. (1998)
 pBBR1MCSMedium copy, broad-host-range cloning vector; CmrKovach et al. (1994)
 pCR-Blunt II-TOPOE. coli cloning vector; KanrInvitrogen
 pPK01pCR-Blunt II-TOPO with a 1.210 kb A96-F/R-PCR fragment from Marseille; KanrThis study
 pPK02pCR-Blunt II-TOPO with a 1.138 kb B100-F/R-PCR fragment from Marseille; KanrThis study
 pPK03pCR-Blunt II-TOPO with a 1.205 kb D124-F/R-PCR fragment from Marseille; KanrThis study
 pPK04pBBR1MCS containing 1.210 kb BamHI-XhoI fragment of pPK01; CmrThis study
 pPK05pBBR1MCS containing 1.138 kb BamHI-XhoI fragment of pPK02; CmrThis study
 pPK06pBBR1MCS containing 1.693 kb C120-F/R-PCR fragment from Marseille inserted at EcoRV site; CmrThis study
 pPK07pBBR1MCS containing 1.205 kb BamHI-XhoI fragment of pPK03; CmrThis study

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Supplementary material
  9. References

Phagosomal maturation is fundamental to the control of intracellular pathogens. However, many have different ways of finding a successful intracellular niche. The diverse nature of these niches indicates that pathogens control the biogenesis of their membrane-bound compartment. The term ‘vacuole’ has already been widely used to refer to the membrane-bound compartments housing several pathogens (Leishmania, Toxoplasma and Salmonella; reviewed in Meresse et al., 1999), and we use this term to describe the intracellular compartment inhabited by B. henselae. In this study, we present evidence that the normal processing of internalized bacteria in macrophages and ECs is perturbed in BCVs. Other intracellular compartments that share some features with BCVs include those occupied by Afipia felis, Legionella pneumophila, Toxoplasma gondii, or Chlamydia trachomatis. By contrast, compartments occupied by pathogenic mycobacteriae, Brucella, or Ehrlichia continuously interact with the endosomal system, although they do not fuse with pre-existing lysosomes (Lührmann et al., 2001).

It is important to note that B. henselae has two major biotypes (Houston-1 and Marseille) which are genetically distinct (Iredell et al., 2003), and may differ significantly in their virulence characteristics (Chang et al., 2002; Yamamoto et al., 2002). We chose to study B. henselae Marseille (Drancourt et al., 1996) because we recently showed that this strain, in contrast to the widely available Houston-1 type strain (ATCC 49882), expresses the important bacterial pathogenicity factor Bartonella adhesin A (BadA; Riess et al., 2004). Although other investigators have used the Houston-1 strain as the primary model for B. henselae studies (Dehio et al., 1997; Musso et al., 2001; Schmid et al., 2004), we favour the use of B. henselae Marseille, with normal BadA expression, for functional studies such as this one.

Our electron microscopic investigations clearly showed that unlike other closely related organisms such as Rickettsia spp. (Clifton et al., 1998), B. henselae resides in membrane-bound intracellular compartments of macrophages and ECs, rather than in the cytoplasm. Remarkably, we observed that only few bacteria were present in those vacuoles from macrophages, whereas in ECs a higher number of bacteria were detectable per vacuole (Fig. 1). These data are consistent with the rapid replication of B. henselae reported after HUVEC infection (Kempf et al., 2000) and the observation of a greater proportion of single-bacterium BCVs in murine macrophages (Musso et al., 2001).

We studied the interaction of B. henselae Marseille with the endocytic pathway and assessed the maturation of the BCV by monitoring acquisition of commonly used endocytic marker proteins, such as LAMP1 (late endosomes and lysosomes), EEA1 (early endosomes) and TfR (recycling endosomes). In both J774A.1 and HUVECs following a 2 h infection, the majority (∼70–80%) of BCVs lacked the expression of these typical endocytic marker proteins, and failed to acquire lysosomal contents (preloaded TR dextran), and to acidify (LysoTracker Red DND-99; Figs 2–4). These data suggest that BCVs do not mature into phagolysosomes or into another type of acidified vacuole, nor are they arrested early endosomal compartments in J774A.1 macrophages or HUVECs, as seen in A. felis (Lührmann et al., 2001). As expected, control experiments with L. innocua and zymosan illustrated normal and undisturbed phagolysosome formation (Lührmann et al., 2001) following a 2 h infection (Fig. 2). Significantly, the normal endosomal trafficking of zymosan in B. henselae co-infected macrophages was demonstrated (Fig. S2), and similar findings have been obtained with A. felis (A.H., unpubl. data). BCVs at 2 h were not identified by markers for the ER (calnexin), for the Golgi apparatus (TEX-1 or trans-Golgi network 46), or for non-clathrin membrane invaginations (caveolae; caveolin-1) in J774A.1 cells or HUVECs (Fig. S7).

We show here that viable B. henselae delay fusion of BCVs with lysosomes. However, consistent with published observations of B. henselae Houston-1 in J774A.1 (Musso et al., 2001), our data indicate killing of B. henselae (Marseille) in macrophages within the first 24 h after infection. By 24 h, >  85% of BCVs in macrophages contained LAMP1 and TR dextran, as well as LysoTracker Red indicating lysosomal fusion and acidification, and bacteria appeared non-viable on TEM. Interestingly, Legionella-containing vacuoles acquire endocytic markers after ∼18 h but, in contrast to B. henselae, L. pneumophila then exploits the endosomal pathway to replicate within macrophages (Sturgill-Koszycki and Swanson, 2000).

It has been previously shown that intracellular B. henselae (Houston-1 and ATCC 49793) are not associated with LAMP1 in HUVECs (Dehio et al., 1997). Consistent with this, and in contrast to our findings in J774A.1 macrophages (where lysosomal fusion is only delayed), we show that viable B. henselae continues to prevent the normal maturation of its vacuole in HUVECs at 24 h. It has been shown that expression of the surface adhesin BadA prevents phagocytosis of B. henselae by J774A.1 macrophages (Riess et al., 2004). However, B. henselae induction of VEGF, IL-1β and IL-8 may participate in a paracrine angiogenic loop whereby macrophages play the predominant role as the effector cell and ECs are the final target cell, resulting in their proliferation (Resto-Ruiz et al., 2002). Delayed destruction by macrophages, once ingested, might thus confer a biological advantage upon Bartonellae.

The unusual intracellular trafficking of B. henselae requires bacterial viability. Heat-killed B. henselae did mature into LAMP1-positive phagolysosomes after a 2 h infection, and consistent data were obtained using lysosomal TR dextran and LysoTracker (Figs 2–4). Heat-killing has the same consequences for other bacteria defective in phagosome maturation, such as A. felis (Lührmann et al., 2001), Brucella abortus (Arenas et al., 2000), L. pneumophila (Swanson and Isberg, 1996) and Tropheryma whipplei (Ghigo et al., 2002). Avoidance of lysosomal fusion by live B. henselae at 2 h is not absolute, and a lysosomal fate for B. henselae is slightly more common in J774A.1 than in HUVECs (Fig. 4B and C), presumably reflecting phagocytosis of the percentage of non-viable Bartonellae present in the infecting inocula (Fig. 4A) by the macrophages. Not only did BCVs containing heat-killed bacteria mature into phagolysosomes, but B. henselae killed by paraformaldehyde before infection were also unable to delay fusion with lysosomes, confirming that the unusual intracellular trafficking of B. henselae requires bacterial viability. Opsonization of live B. henselae with immune antibodies before infection also led to a significant increase in the number of BCVs which colocalized with late endosomal/lysosomal LAMP1 by 2 h, suggesting that uptake via immunoglobulin receptors increases delivery to lysosomes, consistent with data from pathogens such as Legionella, Chlamydia, Toxoplasma, or Mycobacterium (reviewed in Haas, 1998).

It is logical to propose that active modulation of the host cell, e.g. by the VirB TIVSS of B. henselae (Schmid et al., 2004; Schülein et al., 2005) is more likely to be involved in controlling phagosome maturation, than preformed bacterial surface proteins such as BadA (Riess et al., 2004). It should be noted that delayed phagolysosomal fusion in J774A.1 macrophages was achieved not only with viable B. henselae Marseille but also with viable B. henselae Houston-1, with similar results (Fig. S8). B. henselae Houston-1 lacking BadA expression revealed a small yet insignificant increase in the percentage of LAMP1-positive phagosomes at 2 h when compared with Marseille. The evident survival advantage of lysosomal avoidance in these haem-dependent bacteria appears to be further enhanced by inhibition of apoptosis and triggering of host cell VEGF secretion, as shown here in J774A.1 macrophages (Fig. S1), consistent with data in monocytes (Resto-Ruiz et al., 2002; Kempf et al., 2005a) and ECs (Kirby and Nekorchuk, 2002; Schmid et al., 2004).

While a non-fimbrial adhesin (BadA) and a VirB TIVSS have shown to be important virulence factors for B. henselae (Riess et al., 2004; Schmid et al., 2004; Schülein et al., 2005), little is known about genes involved in the localization and intracellular survival of B. henselae. We therefore applied a new magnetic screening assay (Pethe et al., 2004) to search part of a B. henselae Marseille transposon mutant library for bacteria unable to delay lysosomal fusion, in order to find additional genes required for host cell colonization. After multiple cycles of screening and enrichment, we retrieved viable B. henselae mutants from the lysosomal fraction of iron-dextran preloaded macrophages. Of the 100 transposon mutants in the original pool, we found 4 mutants which consistently trafficked to lysosomes in J774A.1 cells and HUVECs, in line with findings from screening studies in Brucella (Kohler et al., 2002; Kim et al., 2003). More extensive screening might be reasonably expected to yield a greater diversity of mutants, perhaps including those virulence factors which have already been identified (Riess et al., 2004; Schülein et al., 2005). Nevertheless, the PAK100 mutant was very strongly represented (16/24 times; ∼70%) in the final group of recovered mutants, and on two independent occasions, and it is therefore likely that this is an important and specific finding.

The intracellular fate of the mutants in macrophages and ECs was investigated, and each mutant, although fully competent in host cell entry (Fig. 5), failed to evade the endocytic pathway and was targeted to acidified lysosomes following a 2 h infection (Fig. 6). Similarly, no difference in bacterial adherence and internalization was observed between wild type and virB mutant strains of Brucella abortus that were unable to evade the endoytic pathway (Comerci et al., 2001). The wild type's ability to regulate maturation of the vacuole in macrophages and ECs results in a direct survival advantage (Fig. 5). Importantly, none of the mutants showed a growth defect on complete solid media (CBA; not shown) and each induced VEGF secretion in macrophages at wild-type level (Fig. 7). Taken together, these findings suggest that the products of the identified genes influence intracellular trafficking and hence intracellular survival.

PAK96 had a disruption in an open-reading frame with close identity to the hbpD gene of B. henselae Houston-1 (Alsmark et al., 2004). Growth of B. henselae is strongly haem dependent as B. henselae is unable to synthesize haem itself (Sander et al., 2000; Minnick et al., 2003; Zimmermann et al., 2003). Bartonella quintana and B. henselae have the highest reported haemin requirement for bacterial growth in vitro (Myers et al., 1969), and no genes for haemin biosynthesis were identified in either genome (Alsmark et al., 2004). From B. quintana it is known that hbpD is the most abundantly transcribed gene from the hbp family (Minnick et al., 2003), and therefore it is not surprising that a hbpD mutant is attenuated in intracellular survival and pathogenicity in macrophages and ECs. It is unlikely that the observed effect is resulting from a polar effect on genes downstream of hbpD, as complementation in trans restored unusual intracellular trafficking and significantly improved survival in both cell types. We therefore couple the role of hpbD at least in part to the diversion of vacuole trafficking. Nothing is known about the role of the d-serine/d-alanine/glycine transport protein in the pathogenicity of B. henselae. Intracellular survival in macrophages and ECs was severely affected by transposon insertion into the cycA gene, and the mutant (PAK120) was efficiently complemented in trans (Figs 5 and 6). Therefore, it can only be speculated that ‘housekeeping’ genes such as the amino acid transporter for the availability of the neutral amino acids (alanine, serine, glycine) are essential in order to survive in stress conditions in the intracellular environment and are likely to be crucial for viability or pathogenicity of B. henselae in host cells. It may well be that sublethal defects in haemin-binding and amino acid transporter systems are masked in enriched artificial growth media, but impair intracellular survival after entry into the host cell.

Thus, this system should trap mutants unable to compensate for nutritional and other stresses of the intracellular milieu, but also trap those which disrupt bacterial functions specifically affecting phagosome maturation. The exact role of the putative virulence-associated protein in B. henselae pathogenicity and host cell infection remains unclear. Intracellular survival of B. henselae in macrophages and ECs was severely reduced by a transposon insertion into the vapA5 gene, however, the mutant (PAK100) was efficiently complemented in trans (Figs 5 and 6). B. henselae VapA5 showed homology to a putative virulence-associated protein of the agriculturally important plant pathogen Pseudomonas syringae (Buell et al., 2003), the biological function of which is unknown. More data are available for the vap genes of Rhodococcus equi known to be crucial for intracellular survival and replication (Jain et al., 2003) but the B. henselae Vap sequence does not appear to be homologous. The isolated mutant PAK124 showed transposon insertion in a gene encoding an unknown hypothetical protein of B. henselae. Although the mutant was significantly affected in intracellular trafficking properties and intracellular survival in macrophages and ECs, we could not restore these properties by complementation of this gene (including its putative promoter region) in trans. In this case, downstream effects or dysregulation of the in trans complemented gene may provide the explanations. Future work will aim to elucidate the role of these genes in B. henselae infections in greater detail, and the association with virulence in those for which it remains unclear (e.g. vapA5).

Our data define for the first time the vacuolic intracellular compartment of B. henselae in detail, and show that viable B. henselae improve their survival by avoiding the normal endocytic pathway in macrophages and ECs. We also show that genes from several functional families are involved in survival. Further work in analysing this bacterial compartment and a genome-wide screening for pathogenicity factors of B. henselae (e.g. via signature-tagged mutagenesis), will help to understand how B. henselae survives within host cells.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Supplementary material
  9. References

Bacterial strains and culture conditions

Bacterial strains are summarized in Table 2. All experiments using B. henselae were with the Marseille strain (wild type) unless otherwise specified. B. henselae was grown on CBA in a humidified atmosphere at 37°C and 5% CO2. The GFPmut2-encoding plasmid pCD354 [kindly provided by C. Dehio, Basel, Switzerland (Dehio et al., 1998)] was introduced into B. henselae by electroporation (Riess et al., 2003). For production of bacterial stock suspensions, bacteria were harvested from agar plates after four days and frozen in Luria–Bertani (LB)-20% glycerol at −80°C. When necessary, kanamycin (30 µg ml−1) or chloramphenicol (1 µg ml−1) was added to the culture media. E. coli strains DH5α and Top 10 (Invitrogen, Carlsbad, CA) were grown at 37°C in LB media (Difco Laboratories, Detroit, MI) with the following concentrations of antibiotics when appropriate: kanamycin, 50 µg ml−1; chloramphenicol, 50 µg ml−1; ampicillin, 100 µg ml−1. Listeria innocua was cultured in brain–heart infusion (Difco) at 37°C.

Generation of the transposon mutant library

Generation of B. henselae Marseille transposon mutants was performed by electroporation of the EZ::TN <KAN-2> transposon (EZ::TN <KAN-2> Tnp Transposome Kit, Epicentre) as previously described (Riess et al., 2003).

Killing and opsonization of bacteria

Killed B. henselae were obtained before infection by heating bacteria at 60°C for 40 min or by treating with 1% paraformaldehyde (PFA) in PBS for 2 h at 4°C. No bacterial growth was observed during 14 days of culture (data not shown). For some experiments, viable B. henselae were opsonized before infection with mouse ascites (raised against B. henselae Marseille and heated to 60°C for 30 min to abolish the lytic complement activity of serum) diluted 1:100 in PBS-0.1% BSA, for 30 min at 4°C, and washed three times with PBS.

Antibodies and probes

The following primary antibodies were used: goat polyclonal anti-murine early endosomal antigen 1 (EEA1; SantaCruz Biotechnology, Santa Cruz, CA); goat polyclonal anti-calnexin (clone C-20; SantaCruz Biotechnology); mouse monoclonal anti-human EEA1 (BD Transduction Labs, San Diego, CA, USA); mouse monoclonal anti-transferrin receptor (TfR, clone H68.4; Zymed Labortaories, San Francisco, CA); mouse monoclonal anti-caveolin-1 (clone 2297; BD Transduction Labs); mouse monoclonal anti-human lysosome-associated membrane protein 1 (LAMP1; clone H4A3); rat monoclonal anti-murine LAMP1 (clone 1D4B; Developmental Study Hybridoma Bank, University of Iowa, Iowa City, IA); sheep polyclonal anti-human trans-Golgi network (TGN)-46; rabbit polyclonal anti-TEX1 (both property of A.H.); and rabbit polyclonal anti-B. henselae Marseille antibodies (Kempf et al., 2000). Secondary antibodies used were: Cy3-, Cy5-, Texas Red-, and fluorescein isothiocyanate (FITC)-conjugated goat anti-rat, anti-mouse and anti-rabbit IgG as indicated, and Cy3-conjugated rabbit anti-goat and anti-sheep IgG (all from Dianova, Hamburg, Germany). Texas Red dextran, LysoTracker DND-99 and N-hydroxysuccinimide (NHS)-Texas Red were purchased from Molecular Probes (Eugene, OR), zymosan A from Sigma (Deisenhofen, Germany).

Cell culture and infection procedures

The mouse macrophage cell line J774A.1 (ATCC TIB-67; kind gift of K. Ruckdeschel, Munich, Germany) was cultured in RPMI 1640 medium (Biochrom, Berlin, Germany), supplemented with 10% heat-inactivated fetal calf serum (Sigma), 2 mM l-glutamine (Gibco, Karlsruhe, Germany), 1 mM sodium pyruvate (Biochrom), 0.05 mM β-mercaptoethanol (Sigma) and nonessential amino acids (Biochrom). Human umbilical vein endothelial cell culture (passages 2–6) was performed as previously described (Kempf et al., 2000). J774A.1 and HUVEC cultures were maintained in a humidified atmosphere at 37°C, 5% CO2. Unless otherwise indicated, J774A.1 or HUVECs were seeded into 24-well plates (Nunc, Roskilde, Denmark) containing 12 mm glass coverslips (1 × 105 cells well−1) the day before the experiment. Bacterial stocks were washed in PBS and diluted in antibiotic-free cell culture media to obtain the desired multiplicity of infection (MOI). Bacteria were sedimented onto cultured cells by centrifugation for 5 min at 200 g. The actual MOI for each experiment was confirmed by plating serial dilutions of the infection inoculum and calculating the number of colony-forming units (CFU). Internalization of bacteria in macrophages and ECs was quantified by gentamicin survival assays as previously described (Kempf et al., 2000; Riess et al., 2004).

Measurement of cell viability and VEGF induction in J774A.1 macrophages

A CellTiter-96 AQueous non-radioactive cell viability assay (Promega, Mannheim, Germany) was used to assess cell viability as previously described (Kempf et al., 2005a). VEGF induction upon B. henselae infection was determined without antibiotics and FCS to minimize non-specific VEGF secretion. Phorbol-12-myristat-13-acetate (PMA, 25 ng ml−1; Sigma) was used as a positive control. Supernatants were centrifuged to remove insoluble particles and VEGF concentration measured using a murine VEGF165 ELISA kit (Quantikine; R&D Systems, Wiesbaden, Germany).

Intracellular trafficking studies

Cell culture infection.  J774A.1 or HUVECs were infected at a MOI of 20 (Bartonella), 10 (Listeria), or 5 (zymosan) for 2 h (unless otherwise indicated). Zymosan particles and L. innocua were prelabelled with 25 µg NHS-Texas Red (Molecular Probes) per millilitre and 0.1 M sodium carbonate (pH 8.3) for 45 min at 4°C, then washed and used to study normal endocytic trafficking and lysosomal acidification. This labelling had no detectable effect on uptake efficiency or compartmentalization and did not affect bacterial viability (data not shown). After infection, cells were washed three times with cell culture medium, and fresh media added for a 2 h chase incubation at 37°C, 5% CO2. Cells were fixed in 3.75% PFA in PBS (pH 7.4), washed with PBS, and free aldehyde groups were quenched with 50 mM NH4Cl in PBS.

Texas Red dextran labelling of lysosomes.  Fluid-phase pinocytosis of 25 µg of Texas Red (TR) dextran per millilitre of cell culture medium was allowed for 16 h at 37°C to label the endocytic pathway. To allow the TR dextran to accumulate specifically in the lysosomes, cells were chased in label-free medium for 3 h before infection with bacteria.

Labelling of acidified compartments with LysoTracker Red DND-99.  Infected J774A.1 or HUVECs were washed, and incubated with 50 nM LysoTracker Red DND-99 in cell culture medium for 1.5 h at 37°C, fixed, labelled sequentially with B. henselae-specific antibodies, and immediately visualized by CLSM.

Analytic and quantitative immunofluorescence.  For all staining protocols, the formaldehyde-fixed cells were washed three times in PBS at the beginning and following each incubation step. For differential staining of intracellular and extracellular B. henselae, fixed cells were sequentially incubated with blocking solution (0.2% BSA in PBS) for 15 min, rabbit anti-B. henselae antibodies for 1 h, Cy5-conjugated goat anti-rabbit IgG antibodies for 1 h, 0.1% Triton X-100 in PBS for 15 min, blocking solution for 15 min, rabbit anti-B. henselae antibodies for 1 h and FITC-conjugated goat anti-rabbit IgG antibodies for 1 h. The latter two antibody incubations were not required if host cell infection was performed with B. henselae gfp+. Next, cells were serially incubated with primary antibodies directed against a specific host intracellular marker for 1 h, and appropriate Cy3- or Texas Red-conjugated secondary antibodies for 1 h. Stained specimens were mounted in Fluoroprep (bioMérieux, Nürtingen, Germany).

Samples stained for immunofluorescence were viewed with a Leica DM IRE 2 confocal laser scanning microscope (Leica, Bensheim, Germany). Three different fluorochromes could be detected simultaneously representing the green (FITC or gfp+), red (Cy3 or Texas Red) and blue (Cy5) channels. Phase contrast was visualized using a fourth channel. Fluorescence images were acquired sequentially to avoid non-specific channel interference. Images were digitally processed with Photoshop 7.0 (Adobe Systems).

In all confocal micrographs, extracellular bacteria are clearly distinguished from intracellular bacteria. Unless otherwise indicated, intracellular bacteria appear in green because of the absence (no colocalization) of a signal in the red channel, or, in the case of a colocalization (fusion) with phagosomal marker staining, in yellow. Colocalization was confirmed by examining individual organelles by x–z sectioning. Extracellular bacteria appear pale blue as a result of the superimposition of signals both in the green channel and the blue channel. Between 100 and 300 individual intracellular bacteria (phagosomes) were scored for colocalization in at least five random fields for each experiment. Results are expressed for the percentage of phagosomes positive for each respective intracellular marker as the mean and standard deviation of three independent determinations.

Control immunofluorescence stainings for specificity of fluorescent probes.  Macrophage and EC lysosomes were selectively prelabelled with TR dextran or LysoTracker Red, as determined by double immunofluorescence experiments with antibodies to late endosomal/lysosomal LAMP1, which virtually completely colocalized with the TR dextran or LysoTracker Red preloaded organelles (Fig. S3). Little or no colocalization was seen with antibodies to late endosomal rab7 or early endosomal EEA1, as expected for a lysosomal compartment (data not shown).

Transmission electron microscopy

Infected cells were analysed by TEM as previously described (Kempf et al., 2000) using a Zeiss EM 902 transmission electron microscope (Zeiss, Oberkochen, Germany) operating at 80 kV.

Live–dead population studies

Viability of B. henselae inocula was measured using the Live/Dead BacLight kit (Molecular Probes) according to the manufacturer's instructions. Briefly, an aliquot of the original inoculum used for infection was incubated with a combination of SYTO 9 (live organisms: green) and propidium iodide (dead organisms: red) for 15 min, and examined by CLSM, with heat-killed B. henselae as a control. At least 1600 bacteria were counted per experiment, in at least eight random fields. The percentage of dead bacteria was expressed as the mean and standard deviation of eight independent experiments.

Magnetic separation of B. henselae-containing phagolysosomes

Preparation of iron-dextran and lysosomal enrichment.  Iron-dextran was prepared as previously described (Rodriguez-Paris et al., 1993; Pethe et al., 2004) and the solution was tested prior to use to confirm that it was not toxic for target J774A.1 cells (data not shown). The method of lysosomal enrichment was carried out as recently published (Pethe et al., 2004) with some modifications, as follows. Six-well plates confluent with J774A.1 (2 × 106 cells well−1) were incubated with 5 ml of iron-dextran in water (∼40 mg ml−1 iron-dextran) mixed 1:1 with 2 × RPMI 1640 medium supplemented with 20% FCS. Cells were incubated for 2 h, then rinsed and chased overnight in label-free medium. One hundred B. henselae Marseille transposon mutants (Riess et al., 2003) were pooled together and used to infect J774A.1 at a MOI of 20 in 5 ml of cell culture medium. Bacteria were dispersed by 5 passages through a 25-gauge needle before centrifugation onto the monolayer. After 2 h, the macrophages were rinsed and chased in fresh media for a further 2 h, then rinsed and scraped into 5 ml of cold homogenization buffer (250 mM sucrose, 0.5 mM EGTA, 0.1% gelatin, 20 mM HEPES, pH 7.0). The cell suspension was centrifuged at 700 g for 10 min, resuspended in 1 ml of cold homogenization buffer, and lysed by multiple passage (∼12–15 times) through a 25-gauge needle. The homogenate was subjected to low-speed centrifugation (300 g for 10 min at 4°C) to remove nuclei and unlysed cells. A MiniMACS column (Miltenyi Biotec, Bergisch-Gladbach, Germany) placed in the magnetic holder was equilibrated with homogenization buffer and the postnuclear supernatant applied to the column. The homogenate (unbound fraction) was collected by gravity. The column was washed with 2 ml of homogenization buffer and then removed from the magnetic holder, and the bound (lysosomal) fraction eluted with 2 ml of elution buffer (0.5% Tween 20 plus 0.05% SDS in water). This was centrifuged (2000 g for 10 min), the supernatant discarded, and the pellet resuspended in 500 µl of LB medium. Four 100 µl aliquots were plated on CBA supplemented with kanamycin (30 µg ml−1) for 12–14 days, and the recovered bacterial mutants harvested for further enrichment. The screen was performed on two independently created pools with five rounds of selection/enrichment each. The recovered bacteria from each pool were thus enriched for mutants defective in delaying phagolysosomal fusion (Pethe et al., 2004), but which retained invasiveness, and viability on complete solid media.

Controls for the magnetic approach of lysosomal enrichment.  Several controls were performed to substantiate the validity of the magnetic screen. J774A.1, preloaded with iron-dextran or with both iron-dextran and TR dextran, were infected for 2 h with FITC-conjugated Zymosan A BioParticles (Molecular Probes) or with live or heat-killed B. henselae gfp+. Bacteria from the bound and unbound fractions were separately cultivated on CBA with kanamycin, and the populations of bacteria or zymosan (green) and lysosomes (red) from each fraction assessed using fluorescence microscopy (see Fig. S5C and S6).

J774A.1, preloaded with iron-dextran and TR dextran but not infected, were fixed (24-well plates, coverslips) with PFA or processed (6-well plates) by magnetic selection. Coverslips were stained with FITC-conjugated LAMP1 and observed by CLSM using phase contrast, FITC and Texas Red channels (see Fig. S5A). The bound (lysosomal) fraction was eluted from the MiniMACS column using PBS and a 200 µl aliquot was centrifuged onto a glass microscope slide for 10 min at 500 r.p.m. (Cytospin 4 Cytocentrifuge; Thermo Shandon, Pittsburgh, PA), air-dried and fixed with PFA. The slide was stained with FITC-conjugated LAMP1 and observed by CLSM (see Fig. S5B).

Live or heat-killed B. henselae gfp+ were incubated with different concentrations (∼20 and ∼40 mg ml−1) of iron-dextran, washed with PBS and then processed through magnetic columns as described above. Bound and unbound fractions were compared using fluorescence microscopy (see Fig. S9).

Southern blotting analysis and nucleotide sequencing

The extraction of the B. henselae chromosomal DNA and Southern hybridization was performed as previously described (Riess et al., 2003). Transposon insertion sites were identified by using the sequencing primers <KAN-2>FP-1 (3′-end of transposon) and <KAN-2>RP-1 (5′-end of transposon; Riess et al., 2003). Sequence data were compiled and analysed by MacDNASIS V3.7 (Molecular Biology Insights, Cascade, CO) for open reading frame (ORF) identification and analysis. The blast search algorithm (http://www.ncbi.nlm.nih.gov/blast/) was used to search for sequence similarity.

Preparation and manipulation of DNA

Plasmids used or generated in this study are given in Table 2. Oligonucleotide primers used for cloning steps are included in Table 3. DNA manipulations were carried out using standard procedures. Chromosomal DNA from B. henselae WT was prepared as previously described (Riess et al., 2003). PfuTurbo DNA polymerase (Roche, Basel, Switzerland) was used to obtain polymerase chain reaction (PCR) products. Plasmids were propagated in E. coli (DH5α or MOSBlue) and isolated with a QIAprep Miniprep kit (Qiagen, Hilden, Germany). PCR purification and DNA purification from methylene blue stained agarose gels were performed with a QIAquick Purification kit according to the manufacturer's instructions (Qiagen). Reactions with calf intestinal alkaline phosphatase, DNA ligase and restriction enzymes were performed as specified by the supplier (Fermentas, St. Leon-Rot, Germany). The Zero Blunt TOPO PCR Cloning Kit (Invitrogen) and pMOSBlue Blunt-ended Cloning Kit (Amersham, Uppsala, Sweden) was used according to the manufacturer's instructions. Electroporation in E. coli or B. henselae was performed as previously described (Riess et al., 2003).

Table 3.  Primers (5′-3′) used for cloning and complementation of B. henselae transposon mutants.
NameSequenceReference or source
  • a

    . Denaturation 95°C, 5 min; amplification 29 cycles (95°C, 60 s; ×°C, 30 s; 72°C, 90 s); extension 72°C, 7 min. Annealing (x°C): A96, 60°C; B100, 47°C; C120, 56°C; D124, 58°C.

  • b

    . Denaturation 94°C, 5 min; amplification 30 cycles (94°C, 60 s; 55°C, 30 s; 72°C, 30 s); extension 72°C, 7 min.

A96-FaTAAAAAGCTGAACTCTCTTTAACCCThis study
A96-RCCAGAAAATGGCATAATCTATTAAAThis study
B100-FaCACCTTACAGTGTTCTGGACGGThis study
B100-RATGGTTTATGGAGGAGGCCCThis study
C120-FaCTCATATAACACTAAATCTTCTTATCCGCThis study
C120-RTTGGTTACCCTTTTTTGATATGCCThis study
D124-FaTCTCCATTATAGGCTTTAACGCThis study
D124-RTTGCTATCATCAAATTGTGCCThis study
CAT-FbACATGGAAGCCATCACAAACGThis study
CAT-RCGCCTGATGAATGCTCATCCThis study
M13-FGTAAAACGACGGCCAGInvitrogen
M13-RCAGGAAACAGCTATGACInvitrogen

Cloning and verification of complemented B. henselae mutants

Oligonucleotide primers (Table 3) constructed from sequencing reactions of B. henselae transposon mutants PAK96 (primers: A96-F/R), PAK100 (B100-F/R), PAK120 (C120-F/R) and PAK124 (D124-F/R), were used in PCR with chromosomal DNA from B. henselae WT to generate respective 1.210 kb, 1.138 kb, 1.693 kb and 1.205 kb products containing the required ORF and a putative promoter region.

Polymerase chain reaction products were cloned into pCR-BLUNT II-TOPO using the Zero Blunt TOPO PCR Cloning Kit to create pPK01, pPK02 and pPK03 respectively. These plasmids were digested with BamHI and XhoI and each insert ligated into pBBR1MCS. The resultant plasmids (pPK04, pPK05 and pPK07, respectively), were electroporated in the respective B. henselae transposon mutant (PAK96, PAK100 and PAK124) to produce the complemented mutants PAK254, PAK255 and PAK257.

The 1.693 kb insert was cloned into EcoRV-digested and dephosphorylated pBBR1MCS using the pMOSBlue Blunt-Ended Cloning kit. The resultant plasmid pPK06 was electroporated in B. henselae transposon mutant PAK120 to create complemented mutant PAK256.

Single strand sequencing by primer walking on plasmids and PCR products was performed using a BigDye Terminator Cycle Sequencing Ready Reaction Kit (PE Applied Biosystems) and an ABI Prism 3700 automated DNA sequencer to ensure that no mutations had been introduced by PCR. In addition to the universal M13 primer for pBBR1MCS, sequence-specific oligonucleotides were made on an automated DNA synthesiser (Applied Biosystems).

Four of each of the resultant chloramphenicol- and kanamycin-resistant recombinants (PAK254, PAK255, PAK256, PAK257) were randomly selected and screened by colony PCR for the chloramphenicol acetyltransferase (cat) gene (410 bp), insert product and insert plus transposon product using the appropriate primers (Table 3; data not shown). Each of these clones was expanded on CBA supplemented with chloramphenicol and kanamycin and then stored at −80°C.

The vector pBBR1MCS was electroporated in PAK96, PAK100, PAK120 and PAK124 to produce the vector-only equivalent controls for PAK254, PAK255, PAK256 and PAK257 respectively.

Statistical analysis

Unless otherwise indicated, all experiments were performed at least three times. Paired Student's t-test was used for statistical analyses. A value of P < 0.05 was considered statistically significant.

Data deposition

The sequence data have been submitted to the GenBank database. Accession numbers: AY451964, AY451965, AY451966, AY451967, AY451968, AY451969, AY954249 and AY954250.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Supplementary material
  9. References

We thank David G. Russel (Ithaca, New York, USA) for providing the magnetic screen protocol before publication and Christoph Dehio (Basel, Switzerland) for providing pCD354. We also thank D. Neumann, A. Schaefer, E. Januschke and B. Fehrenbacher for excellent technical assistance and especially Ingo B. Autenrieth for continuous support. This work was supported by grants from the Deutsche Forschungsgemeinschaft (DFG) to V.A.J.K and A.H., and the ‘Landesforschungsschwerpunktprogramm’ of the Ministry of Science, Research and Arts Baden-Württemberg and from the University of Tübingen (Fortüne-Programm) to V.A.J.K. P.K. is the recipient of an Australian Postgraduate Award and Westmead Millennium Foundation research scholarship stipend enhancement.

Supplementary material

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Supplementary material
  9. References

The following material is available for this article online

Fig. S1. A. B. henselae inhibits host cell apoptosis.

B. Induction of VEGF secretion upon infection of J774A.1 cells with B. henselae.

Fig. S2. The extent of maturation of zymosan-containing phagosomes was unaffected by co-infection of J774A.1 macrophages with live B. henselae.

Fig. S3. Control immunofluorescence staining for specificity of fluorescent probes.

Fig. S4.B. henselae killed by paraformaldehyde before infection were unable to avoid being targeted to lysosomes.

Fig. S5. Controls for the magnetic screen.

Fig. S6. Isolation of phagolysosomes from J774A.1 macrophages.

Fig. S7. Distribution of markers proteins of caveolae, ER and Golgi in J774A.1 and HUVECs infected with B. henselae.

Fig. S8. J774A.1 were infected with viable B. henselae Houston-1 or B. henselae Marseille for 2 or 24 h, or with zymosan for 1 h, and then stained with LAMP1. The mean percentages of the bacteria- and zymosan-containing phagosomes positive for LAMP1 were quantified by CLSM.

Fig. S9. Iron-dextran does not interact non-specifically with B. henselae.

Table S1. Amount of bacteria and lysosomes after magnetic selection of J774A.1 cells infected with B. henselae.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. Supplementary material
  9. References