ABSTRACT
Lyme borreliosis caused by Borrelia burgdorferi is a persistent infection capable of withstanding the host's vigorous immune response. Several reports have shown that the spirochete's linear plasmids 25 and 28-1 are essential for its infectivity. In this context, it was proposed that Borrelia burgdorferi organisms control their uptake by macrophages and polymorphonuclear leukocytes (PMNs) through plasmid-encoded proteins and that this mechanism confers resistance to phagocytosis. To investigate this proposal, a precise flow-cytometry-based method with human blood was used to study the impact of the plasmids 25 and 28-1 on B. burgdorferi clearance over 150 min and to investigate whether low-passage organisms are more resistant to phagocytosis than high-passage B. burgdorferi. Exposure of human blood PMNs or blood monocytes to fluorescein isothiocyanate-labeled B. burgdorferi B31 organisms lacking the linear plasmids 25, 28-1, or both revealed that all spirochete populations were internalized at the same rate as the wild-type borrelia parent strain B31. Moreover, no differences in phagocytosis kinetics were detected when low- or high-passage wild-type B. burgdorferi B31 or N40 were cocultured with blood cells. Plasmid loss and probable associated surface protein changes due to serial in vitro propagation of B. burgdorferi do not affect the resistance of these organisms to internalization by phagocytic cells. In particular, we found no evidence for a plasmid-controlled (lp25 and lp28-1) resistance of B. burgdorferi to phagocytosis by leukocytes of the host's innate immune system.
Lyme borreliosis is a multisystemic disorder caused by infection with species from the tick-transmitted spirochete group B. burgdorferi sensu lato. After the infection is established in the skin, spirochetes disseminate to other sites, including joints, heart, and nervous system (50).
During their life cycle in tick vectors and mammalian hosts, the spirochetes are confronted with changing environmental conditions such as shifting temperature, pH, and nutrient levels (8, 37). In this context, adaptation to their host(s) requires the ability to regulate gene expression in response to these environmental changes both on chromosomes and on plasmids (8, 41). The complete genome sequence analysis of B. burgdorferi B31 showed that this spirochete contains a linear chromosome with ca. 910 kb and, additionally, 21 linear and circular plasmids with a total of ca. 610 kb (9). More than 80% of the predicted open reading frames carried on the plasmids are not homologous to known sequences (9). Continuous passaging of B. burgdorferi cultures reduces the number of detectable plasmids and consequently changes the surface protein profile (3, 36, 43). Similarly, decreasing infectivity rates were observed in mice after serial in vitro cultivation of B. burgdorferi (35, 43). It was suggested that the plasmids carry genes essential for the infection and that changes in plasmid profiles are associated with the loss of infectivity (18, 43, 47).
Several independent in vivo studies investigating the relationship between plasmid content and the pathogenesis of the spirochetes showed that the linear plasmids lp25 and lp28-1, 25 and 28 kb in size, respectively, are essential for full virulence (24, 25, 40). B. burgdorferi organisms lacking lp25 are completely noninfectious and quickly eliminated by the host, whereas clones lacking lp28-1 have a phenotype with an intermediate infectivity, and subsequent infection is restricted to the joint tissue only. Recently, it was shown that transformation of noninfectious B. burgdorferi clones naturally lacking lp25 and lp28-1 with the corresponding plasmids reestablished infectivity for mice (19). This confirmed that the absence of lp25 and lp28-1 rendered B. burgdorferi completely noninfectious or less infectious.
During natural infection with B. burgdorferi, the cells primarily involved in first-line host defense against this bacterium are phagocytes. In vitro studies demonstrated that the spirochetes are easily and rapidly ingested and killed by these cells (27, 33). Lipoproteins of B. burgdorferi interact with the phagocytes' Toll-like-receptor 2 (22, 26) and CD14 (44, 45), whereby the activation of the innate immunity is initiated. There is also evidence that the cell receptor CR3 (11, 13) and the mannose receptor (10), which are involved in microbial recognition and phagocytosis, play an important role in B. burgdorferi elimination. Lipoproteins and glycosylated proteins of the organisms are thought to act as ligands and mediate attachment and ingestion of the spirochetes (10). Nevertheless, a small fraction of organisms survive in vivo, disseminate within the infected host, and establish a persistent infection (5, 48).
To study the uptake of B. burgdorferi organisms by phagocytic cells, a variety of methods are available for laboratory use. Microscopic and radioisotopic techniques are the two most frequently applied methods to determine the phagocytosis rates of phagocytes for B burgdorferi organisms (30, 31). Furthermore, a flow cytometric technique, which offers several advantage over the other methods, was established by Banfi et al. (1) This rapid and reproducible technique makes it possible to count large numbers of individual cells and at the same time allows the identification of cells engaged in phagocytosis. The most important advantage of this method, however, is that the distinction between extracellular attached and ingested spirochetes is made possible, while it does not require additional purification and separation of the phagocytic cells used in the assay (1, 12).
Using the in vitro borrelia elimination assay based on microscopic evaluation to study the spirochetal eradication by human blood phagocytes, Georgilis et al. stated that highly infective, low-passage B. burgdorferi organisms resist phagocytosis much better than high-passage spirochetes with low infectivity (18). Consequently, these authors suggested that the loss of infectivity in vivo after continuous in vitro passaging of the spirochetes is the result of variations that effect the interaction of spirochetes with phagocytic cells. In this context it was hypothesized that the factor(s) responsible for resistance to phagocytosis might be encoded on plasmids. The goal of our study was therefore to find out whether resistance to elimination of B. burgdorferi by phagocytes is controlled by proteins encoded on the two plasmids lp25 and lp28-1 using the precise flow cytometric technique.
MATERIALS AND METHODS
Bacterial strains and growth conditions.Passage 4 (P-4) of B. burgdorferi sensu stricto strain B31 clone 5A4 (parent strain), 5A5 (deficient in lp25 and lp28-1), 5A8 (deficient in lp28-1), and 5A13 (deficient in lp25) were used in the present study. All strains were kindly provided by J. E. Purser and S. J. Norris (University of Texas-Houston Medical School, Houston, Texas) and have been characterized previously (40). Also, nonclonal populations of B. burgdorferi sensu stricto strains B31, P-10, and P-53 and B. burgdorferi sensu stricto strains N40, P-5, and P-66 were included in the study. The spirochetes were routinely maintained in liquid Barbour-Stoenner-Kelly (BSKII) medium (2) supplemented with 6% rabbit serum and incubated at 33°C.
lp25 and lp28-1 plasmid content of each clone was assessed by PCR. The sequences of the primers used in the present study were identical to those published by Labandeira-Rey and Skare (25).
Preparation of fluorescein-labeled bacteria. B. burgdorferi cells from a 3-ml culture containing 108 cells/ml were labeled with fluorescein as described by Cinco et al. (12). The cultures were centrifuged at 4,500 × g and washed with Hanks’ balanced salt solution (HBSS) containing 0.1% gelatin. The cells were resuspended in 1 ml of filtered HBSS with 0.1% gelatin at pH 7.4 and 0.2 mg/ml of fluorescein isothiocyanate (FITC; Sigma-Aldrich, Taufkirchen, Germany). After a 45-min incubation in the dark and at room temperature, the cells were washed twice and resuspended in HBSS with 0.1% gelatin.
Bacterial fluorescence was confirmed on an Axioskop 2 plus microscope equipped with a 40X-plan-NEUFLUAR objective lens (Zeiss, Jena, Germany). Borrelia cell counts were adjusted to 2 × 107 cells/ml and verified by dark-field microscopy using a Petroff-Hausser counting chamber.
Quenching of extracellular bacterial fluorescence.Quenching of extracellular fluorescence was achieved either with 0.2 to 1.0 mg/ml of trypan blue (Fluka-Chemie AG, Buchs, Germany) or 0.25 mg/ml of crystal violet (Sigma-Aldrich, Taufkirchen). Bone marrow-derived macrophages (BMDMΦ), which are routinely cultured in our laboratory, were used for this test. Bone marrow cells from femurs of 6- to 10-week-old sacrificed mice were flushed out from the medullary cavities with the Dulbecco’s modified Eagle medium (Biochrom, Berlin, Germany). A total of 105 cells/ml were cultivated in the presence of 30% L-conditioned medium containing macrophage colony-stimulating factor derived from the cell line L929 (7). After a 10-day differentiation period at 37°C in a humidified atmosphere with 8% CO2, macrophages were washed twice with Dulbecco’s modified Eagle medium and used for quenching studies. To test for specific macrophage surface markers, these cells were labeled with a FITC-conjugated rat anti-mouse monoclonal F4/80 antibody (Caltag, CA) and evaluated by flow cytometry.
Phagocytosis rate of blood leukocytes determined by flow cytometric analysis.Quantification of the phagocytosis rates in blood was performed by flow cytometry as described by Cinco et al. (12). For each reaction, 100-μl aliquots of heparinized blood obtained from healthy donors were mixed with 100 μl of FITC-labeled B. burgdorferi suspension at a ratio of 10:1 (bacteria/monocytes) in 10-ml tubes. Reaction tubes were placed on a shaker with 100 rpm at 37°C for 0, 10, 20, 40, 60, 90, 120, and 150 min. At the end of the incubation period, samples were placed on ice to stop ongoing phagocytosis. Erythrocyte lysis and leukocyte fixation were carried out in 3 ml of fluorescence-activated cell sorting (FACS) lysing solution (Becton Dickinson, Heidelberg, Germany). After a 5-min incubation at room temperature, the tubes were centrifuged at 300 × g for 8 min, and the supernatant was discarded. Cells were resuspended in 5 ml of phosphate-buffered saline and were centrifuged again at 300 × g for 8 min. The resulting cell pellet was resuspended in 200 μl of phosphate-buffered saline. Each phagocytosis assay was carried out in triplicate. Shortly before the phagocytic activity of the sample was assessed, the fluorescence signal of noningested bacteria was quenched with 1.0 mg/ml of trypan blue.
Fluorescence signals were detected with a FACSCalibur (Becton Dickinson, Heidelberg, Germany). The instrument's settings were adjusted and optimized on nonstimulated, human blood leukocytes after erythrocytes were removed by lysis to obtain optimal discrimination of the different leukocyte populations present in blood. After gating on the combination of forward (FSC) and sideward (SSC) light scatter, green fluorescence signals of single cells were collected logarithmically through the FL1 channel and displayed in the form of density plots in combination with FSC (17). The CellQuest software (version 3.1f; Becton Dickinson, Heidelberg, Germany) was used for analysis. Five thousand polymorphonuclear leukocytes (PMNs) were counted. With this adjustment, ca. 1,000 monocytes were observed. In this context, the phagocytosis rate represents the fraction of PMNs or monocytic cells with one or more ingested bacteria (green-FITC fluorescent cells) within the total population, whereas the mean FITC fluorescence intensity (MFI) correlates to the number of ingested bacteria per cell.
Statistical analysis.For statistical evaluations, the Mann-Whitney or Student t test (when a normal distribution of the data was verifiable) was used. The critical P value was chosen to be 0.05 to compare the mean fraction of fluorescent spirochetes ingested by PMNs or monocytes. Comparisons were carried out for each available time point during the coculture of phagocytes and B. burgdorferi organisms.
RESULTS
Phagocytosis assay and quenching optimization.Flow cytometry was used to monitor the uptake of FITC-labeled B. burgdorferi cells by PMNs and monocytes in human blood. Figure 1 shows the increase in the number of phagocytes with ingested, fluorescent-labeled spirochetes over time. Phagocytosis rates increase from <1% at 0 min (left panel) to 17% after 10 min (middle panel), and 40% after 60 min (right panel) of incubation at 37°C. In the absence of FITC-labeled B. burgdorferi, no shifting of the fluorescent signal in phagocytic cells was detected over the incubation time of 150 min.
Flow cytometric analysis of blood PMNs incubated with FITC-labeled B. burgdorferi for 0 min (left panel), 10 min (middle panel), and 60 min (right panel) at 37°C. Fluorescent signals were collected in the FL1 channel and are shown on a log scale. Signal shifting from the left to the right rectangle in the middle and right panels indicated an increased uptake of FITC-labeled spirochetes over time.
Optimal quenching conditions that suppress fluorescent signals from extracellularly attached particles or spirochetes were established utilizing murine BMDMΦ. These cells were labeled with anti-F4/80-antibodies carrying FITC and were subsequently quenched either with crystal violet or trypan blue. The histograms in Fig. 2 show fluorescent signals with distinct peaks. When crystal violet was used at a final concentration of 0.25 mg/ml, only a minimal reduction of the extracellular fluorescence signal was observed. Trypan blue at a concentration of 1.0 mg/ml, however, achieved an improved quenching effect. Microscopic analysis (data not shown) demonstrated that 1.0 mg of trypan blue/ml quenched the extracellular fluorescence effectively without affecting the BMDMΦ cells, whereas quenching with crystal violet even at a concentration of 0.25 mg/ml caused changes in the morphology of the BMDMΦ and cell death, in addition to an incomplete quenching.
Quenching of extracellular fluorescent signals. Unstained BMDMΦ cells (M1) and cells that carry FITC-labeled, anti-F4/80 antibodies (M2) are shown. Extracellular signals on identical cells stained with the FITC-labeled antibodies are quenched with 0.2 mg/ml (M3) or 1.0 mg/ml (M4) of trypan blue or 0.25 mg/ml (M5) of crystal violet. The best quenching results were achieved with 1.0 mg/ml of trypan blue.
Comparison of low- and high-passage B. burgdorferi organisms using the phagocytosis assay.Using the flow cytometer's FSC and SSC, PMNs and monocytes were differentiated (gates R1 and R2 in Fig. 3, respectively). The phagocytosis rates of human blood cells exposed to fluorescence-labeled B. burgdorferi from low and high passages were evaluated after bacteria and cells were cocultured at 37°C for 10, 20, 40, 60, 90, 120, and 150 min.
Counter plot of the FSC (cell size) versus the SSC light scatter (cell granularity) of human blood leukocytes. The plot shows PMN (R1, upper gate) and monocyte (R2, lower gate) populations, which were further analyzed in subsequent phagocytosis assays.
During the 150-min incubation period, 46.3% ± 2.7% PMNs (R1) had internalized low-passage (P-10) FITC-labeled B. burgdorferi B31, whereas 40.0% ± 1.1% PMNs had taken up high-passage borrelia (P-53). The comparable phagocytosis rates for monocytes (R2) were 45.3% ± 1.7% for low-passage (P-10) and 43.3% ± 2.0% for high-passage spirochetes (P-53). No statistically significant differences were observed for internalization rates of PMNs and monocytes either for low or high passages of B. burgdorferi (Fig. 4a and b). The MFIs of PMNs or monocytes incubated either with low or high passages of B. burgdorferi, a measure for the number of organisms ingested by single phagocytic cells, did not differ at statistically significant rates (Table 1). To confirm these results, identical experiments with another pathogenic strain, B. burgdorferi N40, were performed. As in previous experiments, phagocytosis rates of PMNs and monocytes for low- or high-passage organisms did not differ (data not shown).
Phagocytosis kinetics for low- (P-10) and high-passage (P-53) FITC-labeled B. burgdorferi B31 by PMNs (a) and monocytes (b). Internalization rates of the phagocytes did not differ significantly due to potential culture adaptation of B. burgdorferi populations. The data represent means ± the standard deviations.
Differences of individual MFIs in relation to the overall MFI of the phagocytes after a 150-min incubation at 37°C with the low or high passages of B. burgdorferi B31
Impact of the B. burgdorferi lp25 and lp28-1 plasmids on the rates of phagocytosis by blood leukocytes.Before phagocytosis assays were performed, the presence or absence of the plasmids lp25 and lp28-1 was verified in B. burgdorferi B31 clones 5A4, 5A5, 5A8, and 5A13 by using PCR (data not shown).
B. burgdorferi B31 clones 5A4 (wild-type), 5A5 (lp25−/lp28-1−), 5A8 (lp28-1−), or 5A13 (lp25−) were rapidly taken up by blood phagocytes within the first hour of exposure. After 60 min, internalization rates reached 40% ± 2.0%, 37.6% ± 1.8%, 35.0% ± 0.5%, and 35.6% ± 0.8%, respectively, for PMNs and 26.6% ± 1.2%, 27.3% ± 1.2%, 28.3% ± 2.6%, and 32.3% ± 2.6% for monocytes. During the following 90 min, phagocytosis rates for all four B. burgdorferi B31 clones increased to 43% ± 0.3%, 40.3% ± 2.3%, 42.3% ± 2.8%, and 41.0% ± 0.5% for PMNs and 32.0% ± 2.6%, 28.6% ± 0.8%, 29.6% ± 3.5%, and 32.3% ± 4.0% for monocytes, respectively (Fig. 5).
Phagocytosis kinetics for FITC-labeled B. burgdorferi B31 clones 5A4, 5A5, 5A8, and 5A13 by PMNs (a) and monocytes (b). Internalization rates of the phagocytes did not differ significantly due to plasmid contents in the B. burgdorferi populations. The data represent means ± the standard deviation.
The almost-identical MFIs of the phagocytes incubated with 5A4 (wild-type), 5A5 (lp25−/lp28-1−), 5A8 (lp28-1−), or 5A13 (lp25−) B. burgdorferi B31 (Table 2) combined with the results of the phagocytosis rates demonstrated that all clones were engulfed and taken up by blood phagocytes at identical rates, and no statistically significant differences were observed.
Differences of individual MFIs in relation to the overall MFI of the phagocytes after a 150-min incubation at 37°C with the 5A4 (wild-type), 5A5 (lp25−/lp28-1−), 5A8 (lp28-1−), or 5A13 (lp25−) clones of B. burgdorferi B31
DISCUSSION
Phagocytosis of B. burgdorferi plays a crucial role in the initial control of the pathogen burden, prior to the development of an adaptive immune response. Despite the efficient elimination of spirochetes by the host phagocytic cells (32, 34), spirochetes may persist in vivo and contribute to the later manifestation of Lyme disease (4). The aim of the present study was to find out whether resistance to elimination of B. burgdorferi by the phagocytes of the innate immunity is under the control of plasmids lp25 and lp28-1. In an early study by Georgilis et al. (18), high-passage B. burgdorferi that was not infective in mice was taken up by PMNs and monocytes more efficiently than low-passage organisms. Therefore, it was proposed that plasmid loss due to serial culturing might account for the different phagocytosis rates of B. burgdorferi organisms of low and high passages.
To define the relationship between resistance to elimination of B. burgdorferi by the phagocytes and plasmid content of the spirochetes, we measured the phagocytosis with a sensitive flow cytometric method. This assay allows the analyses of large numbers of cells, while natural phagocytosis conditions are retained as far as possible in whole human blood. The most important advantage of the flow cytometric assay used in the present study compared to other methods is the fact that quenching of extracellular fluorescence prior to the measurements of the phagocytosis rates is possible. A clear distinction between adherent, extracellular spirochetes and ingested, intracellular spirochetes is possible (1, 38, 49). Crystal violet or trypan blue are primarily used as the quenching reagents (12, 21, 38). Considerable differences in quenching efficiencies using dyes from different manufactures were observed (49). Consequently, it was necessary to optimize the experimental conditions in order to obtain the highest quenching efficiency while minimal side effects on cells were generated. Our experiments showed that trypan blue is superior to crystal violet, which has been used by other groups (1, 12, 38), and that crystal violet could not be used in our cytometric assay because of its lysosomotropic effect and its ability to quench even intracellular fluorescence (21, 42, 49).
First, we examined the phagocytosis capacity of PMNs and macrophages in the presence of low and high passages of two different B. burgdorferi strains, B31 and N40. The identical phagocytosis rates and internalization abilities obtained in our study are in clear contrast to those observed by Georgilis et al. (18). They found that high-passage organisms were eliminated at a rate approximately 10 times higher than that measured for low-passage organisms. It is noteworthy that the definition for elimination rates differs in the two studies. We determined the elimination rate by assessing the fraction of phagocytic cells that had internalized FITC-labeled bacteria. Simultaneously, the MFI, which correlates with the number of ingested bacteria per cell, was recorded, and it showed that equal numbers of spirochetes were taken up by the phagocytes. Georgilis et al., however, defined elimination efficiency by a statistically less precise, indirect method, counting microscopically viable spirochetes that had not been internalized or did not adhere to the phagocytic cells. Therefore, it is likely that the discrepant results of the two studies are due to the experimental conditions. Furthermore, we assessed phagocytosis rates with whole blood, whereas culture-derived phagocytes were used in the study by Georgilis et al. (18). In vitro culture of cells, however, has an impact on cells mediated by adhesion to surfaces and by maturation of the blood monocytes to macrophages. These changes are apparent in the form of an upregulation of different maturation-associated antigens and changes in the cell's secretory repertoire (6, 23, 28). It is likely that these changes might influence the interaction of phagocytes with B. burgdorferi.
On the other hand, differences in spirochete populations with regard to plasmid content might impact on the interaction between B. burgdorferi and phagocytes, since the modified expression of surface molecules might hamper their internalization by phagocytes. In both studies, spirochetes originated from culture and were passaged intensively. Plasmid loss due to continuous culturing and freezing procedures is known to occur with high frequency in B. burgdorferi (3, 36, 43). It was noted that clonal heterogeneity in plasmid content was detected in the primary outgrowth population from a colony (20). The results obtained showed that the utilization of clonal cultures with a well-characterized plasmid profile is essential in studies, especially when genetic elements are suggested to be involved.
In the present study we focused on plasmids lp25 and lp28-1 because of their essential role for the infectivity of B. burgdorferi (15, 25, 40). The absence of lp25 is associated with a complete loss of infectivity, while the lack of lp28-1 resulted in reduced infectivity (24, 25, 39). Purser et al. provided evidence that the BBE22 gene carried on lp25 is the locus required for infectivity of B. burgdorferi (39). Complementation studies and enzymatic analysis demonstrated that the gene product has nitroaminidase activity and is probably required for the biosynthesis of NAD. lp28-1, on the other hand, carries the vlsE gene (16) encoding a surface-localized lipoprotein that undergoes extensive recombination during the course of infection (51) and the variable regions of the central domain form immunogenic epitopes on the exposed surface of the protein (14, 29). To determine whether lp25 and/or lp28-1 influence elimination of B. burgdorferi by PMNs and monocytes, the phagocytosis rates for wild-type B. burgdorferi (5A4) and plasmid-deficient clones lacking lp25 (5A13), lp28-1 (5A8), or lp25 and lp28-1 simultaneously (5A5) were determined by utilizing the flow cytometric method. Regardless of which B. burgdorferi clone was used for the phagocytosis assay, ca. 40% of the PMNs and 30% of the monocytes had internalized spirochetal organisms after 150 min of incubation. The numbers of borrelia cells internalized in these phagocytic cells were almost identical for the four different clones. In vivo studies in normal and immunocompromised mice showed that the lp25 mutant of B. burgdorferi can be cleared quickly within 48 h of inoculation (24), and it was demonstrated that the absence of an adaptive immune response had no effect on lp25-mediated infectivity (19). From these results it was concluded that the inability of these mutants to survive within the mammalian host is related to the absence of lp25 gene products that are either important for cellular metabolism or for the innate immune response (19, 24). Results obtained showing similar phagocytosis kinetics for the lp25-deficient clones and the wild-type strain provided evidence that gene products of lp25 do not contribute B. burgdorferi elimination by phagocytosis and that the differential clearance of clones lacking these plasmids in comparison to the wild type cannot be ascribed to the innate immunity, as suggested by Labandeira-Rey et al. (24). However, these results are in accordance with the proposition of Purser et al. (39). These authors suggested that the protein encoded by the BBE22 gene has a physiological function in the life cycle of B. burgdorferi rather than an effect on the host immune response.
lp28-1, on the other hand, seems to influence the host's immune response more intense. A progressive loss of the vlsE-carrying plasmid is correlated with a low percentage of mice that developed arthritis (46). It was shown that immune evasion is correlated with proteins encoded on lp28-1 and that the clearance of these mutants is due to the mechanisms of the adaptive immunity (24, 39). The effect of the proteins encoded on lp28-1 on the innate immunity is yet unknown. Our results demonstrating identical phagocytosis rates of PMNs and monocytes for B. burgdorferi lacking lp28-1 and the wild type indicated that missing the proteins encoded by this plasmid do not influence the elimination of B. burgdorferi by phagocytic cells belonging to the innate immunity.
Our results show that clonal cultures of B. burgdorferi with identified plasmid components combined with a sensitive flow cytometric method provide a powerful tool to clarify the role of different genetic elements of B. burgdorferi in the innate immunity. We also demonstrated that proteins encoded on lp25 and lp28-1, the essential plasmids required for the pathogenicity of B. burgdorferi in mammalian hosts, are not essential for bacterial elimination by phagocytic cells belonging to the host's first line defense system.
ACKNOWLEDGMENTS
We thank Joye E. Purser and Steven J. Norris for kindly providing the B. burgdorferi clones used in this study. We also thank Uwe Mueller and Nicole Schuetze for help with the software of the FACS station. Furthermore, we thank Brain Summers for critically reading the manuscript and Gottfried Alber for supporting this project and ongoing research.
FOOTNOTES
- Received 22 February 2005.
- Returned for modification 11 April 2005.
- Accepted 19 May 2005.
- Copyright © 2005 American Society for Microbiology
