Protein A (SpA) of Staphylococcus aureus is endowed with the capacity to interact with the H chain variable region (VH) of human Abs and to target >40% of B lymphocytes. To investigate whether this property represents a virulence factor and to determine the in vivo consequences of the confrontation of SpA with B lymphocytes, we used transgenic mice expressing fully human Abs. We found that administration of soluble SpA reduces B-1a lymphocytes of the peritoneal cavity and marginal zone B lymphocytes of the spleen, resulting in a markedly deficient type 2 humoral response. Single-cell PCR analysis and sequencing of the Ab VH gene repertoire revealed a significant reduction of VH3+ marginal zone B cells. Since the two B lymphocyte subsets targeted are involved in innate immune functions, our data suggest that crippling of humoral immunity by S. aureus represents an immune evasion mechanism that may aggravate recurrent infections.
Staphylococcus aureus is the most pathogenic of the five species of staphylococci and is implicated in a variety of infections (1). It is an opportunistic pathogen that colonizes the skin and mucous membranes (primarily the anterior nasal vestibule) of ∼20–30% of adults and most children without causing clinical symptoms. However, if the skin is damaged by trauma, inoculation by needles, or direct implantation of medical devices, S. aureus may escape from the local lesion and spread through the blood to other body areas, causing a variety of systemic infections that may involve every system and organ. It can cause lethal infections often associated with abscess formation, such as endocarditis and pneumonia. Because of developing resistance to multiple antibiotics, S. aureus is recognized worldwide as a major health threat (2).
The potential of S. aureus to cause a variety of life-threatening conditions is rooted in its propensity to generate resistance to multiple antibiotics, in its capacity to adapt to different environmental conditions, and in its intrinsic virulence, a result of the combined effect of many factors expressed during infection. Virulence factors for S. aureus include exotoxins such as leukocidin, α- and δ-toxins, microcapsules, and coagulase (2).
In addition to secreting toxins, enzymes, and other bacterial components, S. aureus produces protein A (SpA)3 , a prototype staphylococcal surface protein (2). During infection, SpA has antiphagocytic properties that are based on its ability to bind the Fc portion of Igs. In addition to its Fc IgG binding capacity, which has been used for Ig purification, SpA also interacts with a subset of human Ig VH region, a property responsible for its superantigen (SAg) activity for B cells (3). In contrast to conventional Ags, B cell SAgs interact with conserved framework regions of Igs, outside of the Ab paratope. By virtue of this interaction, they can target a large proportion (>40%) of human B cells.
In humans, Ig VH genes have been categorized into seven families of sequence homologous members. With approximately half of the total pool of functional genes, the VH3 family is the largest family that imparts protective humoral responses against pathogens (4, 5). Because of its high specificity for human Igs encoded by VH3+ genes (6) (>50% of the Ab repertoire), SpA is potentially capable of impacting a significant proportion of the human B cell repertoire. It remains unclear whether this property represents a virulence factor used by S. aureus to subvert the humoral immune response. In addition, there is little insight into the in vivo effects of SpA on B lymphocytes expressing human VH3+ Igs, partly because of the unavailability of an appropriate experimental system. In the present study, we made use of transgenic mice expressing fully human Igs to investigate the in vivo consequences of the confrontation of SpA with B lymphocytes. The data reveal a novel mechanism used by this B cell SAg to cripple humoral immunity.
Materials and Methods
Transgenic mice and recombinant proteins
The transgenic mice, called “XenoMouse,” used in these experiments were provided by Abgenix. In the mice used (XenoMouse G2KL (XMG2KL)), the endogenous H- and κl-chain loci have been functionally inactivated and human DNA inserts encoding Cμ, Cδ, Cγ2, Cκ, and Cλ, and the majority of human Ig variable genes (66 VH, 32 Vκ, and all Vλ) were incorporated (7). Animals were handled in the animal facility of Broussais Hospital in Paris, according to the principles expressed in the Declaration of Helsinki on the use of animals in research. They were housed in a specific pathogen-free barrier facility and used at 14–16 wk of age.
Recombinant protein A (SpA) was obtained from RepliGen. The protein was produced in Escherichia coli and purified using multiple chromatographic steps (>98% purity by HPLC). It was tested by a validated quality control procedure for endotoxin. SpA expresses binding sites for both VH3+ Igs and the Fc region of IgG. Iodine monochloride modification of SpA selectively inactivates Fc binding activity without affecting its VH3 binding activity. Therefore, we treated SpA with iodine monochloride to inactivate its Fc binding site. As a control protein, hen egg lysozyme (HEL; Sigma-Aldrich) was used.
For preparation of lymphocytes from peripheral blood, mice were bled retro-orbitally in heparinized tubes, and heparinized blood (0.2 ml) was pelleted by centrifugation, after which the cell pellet was collected and viable white cells were isolated using MSL cell separation gradient (Eurobio). Cells were pelleted and resuspended in PBS supplemented with 2% FBS and 0.01% NaN3. Spleen, lymph node, bone morrow (BM) and peritoneal cavity (PeC) cells were also collected. For BM preparation, femurs and tibias were removed aseptically, ground in a mortar containing 1 ml of ice-cold sterile RPMI 1640 (BioWhittaker), and filtered through sterile cotton to remove bone fragments. RBC were lysed with Tris-buffered ammonium chloride. Splenic cells were depleted of erythrocytes using Tris-buffered ammonium chloride. PeC cells were removed by aseptic injection of 5 ml of RPMI 1640 into the PeC, followed by withdrawal of the peritoneal exudate. Erythrocytes were lysed with Tris-buffered ammonium chloride. Lymph node lymphocytes were collected from cervical, axillary, inguinal, and portal mesenteric lymph nodes. Single-cell suspensions isolated from the various compartments (spleen, lymph nodes, or BM) were counted using a Malassez cell chamber and brought to a concentration of 5 × 106 cells/ml in RPMI 1640 containing 2% FBS and 0.01% NaN3.
Cell suspensions were isolated before two- or three-color immunofluorescence analysis. Leukocytes (0.5 × 106) were stained at 4°C using predetermined optimal concentrations of different combinations of fluorochrome-labeled Abs for 30 min. Following washes, when necessary, biotin-labeled Abs were revealed by streptavidin-CyChrome (BD Pharmingen). Cell phenotype was determined using the following Abs: anti-mouse CD5-PE (53-7.3), anti-mouse CD45R/B220-FITC or anti-mouse CD45R/B220-CyChrome (RA3-6B2), anti-mouse CD11b-PE (M1/70), anti-mouse CD4-PE (H129.19), anti-mouse CD8-FITC (53-6.7), anti-mouse CD43-FITC (S7), anti-mouse CD21/CD35-FITC (7G6), anti-mouse CD23-PE (B3B4), anti-human IgM-FITC or anti-human IgM-PE (G20-127), and anti-human IgD-FITC (IA6-2). All those mAbs and their corresponding isotype controls were purchased from BD Pharmingen. Anti-human CD19-FITC mAbs (J14.119) was provided by Immunotech and anti-mouse CD19-PE mAbs (6D5) by Clinisciences. VH3+ cells were identified using biotin-conjugated SpA and FITC-labeled anti-CD45R/B220 Ab. After washing and fixation in PBS/1% formaldehyde, the cells were analyzed using a single laser FACScan flow cytometer (BD Biosciences). A minimum of 10,000 events were collected per sample, and data were analyzed with CellQuest (version 3.1) (BD Biosciences). Low angle and orthogonal light scatters were used to exclude dead cells and debris. Cells with the forward and side light scatter properties of lymphocytes were analyzed with fluorescence intensity shown on a four-decade log scale. Positive and negative populations of cells were determined using unreactive isotype-matched mAbs (BD Pharmingen) as controls for background staining. Background levels of staining were delineated using gates positioned to include 98% of the control cells. Detection of apoptotic cells was performed using an Annexin VFITC kit (Immunotech).
Cell sorting and lysate preparation
Splenocytes from control, HEL-, and recombinant SpA-treated mice were bulk sorted using a DakoCytomation MoFlo (DakoCytomation) yielding populations with 97.8–99% purity. Cells were counterstained with 7-aminoactinomycin D (Calbiochem) to exclude cells that had lost membrane integrity from the populations of interest identified as B220+CD23−lowCD21+ marginal zone (MZ) and B220+CD23+CD21+ follicular (FO) cells. Sorted B220+ cells were diluted in buffer (10 mM NaCl, 5 mM Tris-HCl (pH 8) at 25°C, and 0.1% Triton X-100) containing 0.4 mg/ml proteinase K (Sigma-Aldrich), and aliquots containing approximately one cell were dispensed into each well of a PCR plate (USA Scientific), covered with 10 μl of mineral oil and capped. Cell lysates were prepared by incubating the plates in a PTC-100 programmable thermal controller or PTC-225 DNA Engine Tetrad (MJ Research) for 1 h at 55°C followed by 95°C for 10 min to inactivate the enzyme.
Primer extension preamplification and PCR amplification
Genomic DNA from cell lysates was amplified using random 15-mer (Qiagen) and 60 cycles of PCR as described previously (8). After initial denaturation for 10 min at 95°C, each cycle of amplification involved denaturation for 1 min at 94°C, annealing for 2 min at 37°C, programmed ramping of 10°C/s to 55°C, and extension for 4 min at 55°C. Specific H-chain genes were amplified by nested PCR approach in which 5 μl of the preamplification product was used as template for the external amplification. Multiplex VH3+VH4 PCR was performed using the following primers: VH3 external, 5′-CCATGGAGTTTGGGCTGAG-3′; VH4 external, 5′-GAAACACCTGTGGTTCTTC-3′; JH external, 5′-ACCTGAGGAGACGGTGACC-3′; VH3 nested, 5′-GTCCAGTGTSAGGTGCAGC-3′; VH4 nested, 5′-GGTGCAGCTGCAGGAGTCG-3′; JH1245 nested, 5′-GTGACCRTKGTCCCTTGGCCC-3′; and JH36 nested, 5′-TGACCAGGGTKCCMYGGCCC-3′. Touchdown PCR was used for external and nested PCR. For the external PCR, one round of denaturation at 95°C for 5 min was followed by denaturation for 1 min at 94°C, annealing for 1 min at 60°C, and extension for 1.5 min at 72°C. Consecutive cycles of programmed ramping at the rate of 2°C/cycle brought the final annealing temperature to 50°C for VH. Denaturation for 1 min at 95°C, annealing for 1 min at 50°C, and extension for 1.5 min at 72°C was repeated 29 times. The nested PCR was similar except that the annealing temperature ramping program ran from 65°C to 55°C for VH.
PCR products were separated by electrophoresis on 1.2% agarose gels. Products were purified (Edge Biosystems) and sequenced using the ABI Prism Big Dye Terminator Cycle Sequencing kit (Applied Biosystems). Sequences were purified (Edge Biosystems) and analyzed on an automated capillary sequencer (ABI Prism 3100 Genetic Analyzer; Applied Biosystems). Germline sequences were determined by using the JOINSOLVER sequence analysis program (9). A rearrangement was considered productive if the VDJ junction maintained the reading frame into the JH segment. Rearrangements that were out of frame or introduced a stop codon during the rearrangement at a junction were considered nonproductive.
Induction of T cell-dependent and -independent responses
Groups of four mice were pretreated with SpA before immunization. To induce a T cell-dependent response, tetanus toxoid (TT) (Aventis) was preincubated with 2% Alhydrogel (Brenntag Biosector) for 1 h, and mice were injected i.p. (15 μg/mice). For induction of a T cell-independent response, mice received 50 μg of DNP-Ficoll (Biosearch Technologies) in PBS by the i.p. route. Mice were bled before and 7 days after immunization.
An ELISA was used to quantify total IgM and IgG, and IgM and IgG reactive with SpA in mouse plasma. Briefly, microtiter plates (MaxiSorp, Nunc Immunoplate; eBioscience) were coated overnight with goat Abs (1 μg/ml in borate-buffered saline (pH 8.4)) to human γ-chain (I3382; Sigma-Aldrich) or to human μ-chain (I0759; Sigma-Aldrich). After blocking with 1% PBS-BSA, plasma samples diluted in 1% PBS-BSA were incubated for 2 h at 37°C with the sensitized wells. Bound Abs were detected by HRP-labeled anti-human-ΙgG (A0170; Sigma-Aldrich) or -IgM Abs (68392; ICN Biomedicals). For detection of IgM and IgG reactive with SpA, microplates were coated with SpA and the ELISA processed as described above. A human mAb (M3: IgM, VH3, Vκ1), provided by Dr. J. P. Bouvet (Paris, France) (10), and a pool of human sera were used as calibration standards. SpA binding Igs were calculated as follows: (concentration of VH3+ Ig/concentration of total Ig) × 100.
For detection of TT- and DNP-specific Abs, microtiter plates were coated overnight with either TT (2 μg/ml in PBS) or DNP-BSA (10 μg/ml in PBS). After blocking with 1% PBS-BSA, serum samples diluted in 1% PBS-BSA were incubated 2 h at 37°C with the sensitized wells. Bound Abs were detected by HRP-labeled anti-human-ΙgG (A0170; Sigma-Aldrich) or -IgM Abs (68392; ICN Biomedicals).
To assess the significance of the observed differences in the groups of treated mice, all experimental results were tested with Mann-Whitney nonparametric U test and the χ2 test. Results were considered significant at p < 0.05, to be highly significant at p < 0.01, and to lack significance at p > 0.05. The results of tests for significance are provided in the figure legends.
Because SpA from S. aureus expresses binding sites for both VH3+ Igs and the constant region domain of IgG, it was first necessary to abrogate its Fcγ binding site by iodination. Such a modified SpA binds VH3+ Igs but not VH3− mAbs (3, 11). All experiments were performed using this modified VH3+ Ig-specific SpA and XMG2KL transgenic mice (7). The mouse IgH and Ιgκ loci of these rodents were inactivated, and their humoral immune system has been restored by introduction of human germline VH, Vκ, and Vλ loci. These mice secrete human Igs and, when immunized, mount Ag-specific Ab responses with somatic mutation, isotype switching, and affinity maturation (12). Their B lymphocytes express fully human surface Igs and secrete IgM and IgG2 bearing κ or λ human L chains. Half of the VH genes present on the translocus belong to the VH3 gene family, a figure that closely mimics the complexity of the human Ig VH locus. This similarity in genomic composition and content translates into phenotypic homologies in surface Ig expression. As in human normal peripheral B cells, which express ∼50% of SpA+ Igs, approximately half of peripheral B cells from XMG2KL mice are able to bind SpA that has been rendered devoid of its Fc binding (Fig. 1). The binding of iodine monochloride-treated SpA to VH3 Igs has been reported previously by immunochemistry methods (12). Additionally, individual CD19+ B cells were sorted, and their VH genes sequenced from genomic DNA of single cells as described previously (8). VH3 and VH4 family members were analyzed and the ratio of B cells expressing VH3 and VH4 was 3:1, as described previously (8). CD19+ B cells that bound biotinylated iodine monochloride-modified SpA were identified with streptavidin-FITC. Individual SpA binding B cells were sorted, and VH3 and VH4 genes amplified from genomic DNA of individual cells were sequenced and their sequences analyzed. Of 69 SpA binding B cells analyzed, all 69 expressed VH3 gene family members. These included VH3–11 (n = 6), VH3–13 (n = 2), VH3–15 (n = 13), VH3–23 (n = 31), VH3–30 (n = 1), VH3–30-3 (n = 2), VH3–53 (n = 8), VH3–66 (n = 2), VH3–72 (n = 1), and VH3–74 (n = 3). Thus, the XMG2KL transgenic mouse expresses VH3+ B lymphocytes capable of interacting with SpA in a SAg manner and represents a suitable experimental model to investigate the in vivo effects of this exotoxin.
A single injection of SpA impacts B-1a cells in the PeC and VH3+ B cells in the spleen
Although the majority of B lymphocytes in an individual belong to the recirculating conventional B cell subset, termed B-2, a subset of specialized B lymphocytes, termed B-1, can be distinguished by tissue distribution, cell surface phenotype, BCR signal generation, and capacity for self-renewal (13, 14). B-1 cells express high levels of IgM and low levels of IgD and B220, lack CD23, and when found in the PeC, express the macrophage marker CD11b (Mac-1). Many B-1 cells, termed B-1a, also express CD5, a coreceptor that may contribute to their unique characteristics (15). In postnatal life, the highest proportion of B-1 cells is found in the PeC, where they account for up to >60% of total B cells. To determine the impact of SpA on B cells, transgenic mice received one i.p. injection of the recombinant protein devoid of its Fcγ binding site (1 mg in PBS). Mice were sacrificed the following day, and their B cell populations quantitated by FACS analysis, excluding macrophages. Compared with control untreated mice (35 ± 11.1%), SpA-injected mice had increased percentages (46.2 ± 5.5%) of B-2 cells (IgM+CD11b−) in the PeC. The treatment had no effect on the B-1b cell population (IgM+CD11b+CD5−), which was unchanged in the two transgenic mouse groups (25.9 ± 6.4 and 23.5 ± 11.7%, respectively). This markedly contrasts with the B-1a cell subset (IgM+CD11b+CD5+), which was significantly reduced in treated mice (24.4 ± 1.6%, p < 0.05) compared with the control group (37.5 ± 3.3%). Even injections of lower doses of SpA (500, 250, and 125 μg) induced a specific reduction in the number of B-1a cells.
This single injection of SpA resulted in an important diminution of splenic VH3+ B cells (B220+SpA+ B cells), which were decreased by 24.8% compared with mice not receiving SpA. Other cell populations were not affected. Thus, the percentages of CD4+ T cells were 47.2 ± 1.5 and 42.5 ± 3.8 for untreated and SpA-treated mice, respectively. Those of CD8+ T cells were 25.8 ± 0.4 and 25.02 ± 3.1 for untreated and SpA-treated mice, respectively. As will be shown below, the reduction of total B cells is more related to MZ B cell depletion.
Several mechanisms could account for the observed diminution of B cells in treated mice, including alterations of homing, decreased production and/or developmental blockade of the corresponding progenitors, and accelerated cell death. To gain insight into the cause of B cell disappearance, we determined the percentages of apoptotic and necrotic B cells in the spleen of the treated mice by annexin V and propidium iodide labeling. We found that B cell death was increased by 53% (p < 0.03) in injected mice as compared with control mice. Specifically, the percentages of annexin V+, propidium iodide+ B cells were 10.4 ± 2.4 and 22.1 ± 1.6 in untreated and SpA-treated mice, respectively.
These results demonstrate that a single i.p. SpA injection rapidly impacts B-1a cells present in the body cavities and induces a specific diminution of VH3+ B cells in the spleen. After only 16 h, SpA treatment triggers a significant increase of apoptotic B cells in the spleen.
Repeated administration of SpA does not affect B cells in the BM or the lymph nodes
During bacterial infection, important amounts of exotoxins are released into the circulation, resulting in longer and repeated exposure of SpA to cells of the immune system. To investigate the SAg effects on B cells under conditions that mimic bacterial infection, the transgenic mice received five i.p. injections of soluble SpA in PBS every other day. A control group of mice received HEL under the same experimental conditions. Mice were sacrificed 13 days after the last injection, and their cell populations in different lymphoid organs were analyzed by flow cytometry.
We found that pro-B (B220lowCD43+IgM−) and pre-B (B220lowCD43−IgM−) cell percentages were unchanged in SpA- and HEL-injected mice (pro-B: 19.6 ± 0.7 and 18.3 ± 2.5%; pre-B: 57.3 ± 0.7 and 55.5 ± 1.7%, respectively). The percentages of immature cells (B220+IgM+IgD−) and mature recirculating cells (B220+IgM+IgD+) were also unchanged compared with HEL-treated mice (immature: 16.0 ± 0.8 and 16.3 ± 1.3%; mature: 14.4 ± 1.8 and 13.2 ± 2.1%, respectively). Finally, the percentages of mature B cells and CD4+ and CD8+ T cells were similar in the lymph nodes of the two experimental groups.
SpA treatment induces a rapid and specific decrease in plasma VH3+Igs
To follow the effect of the SpA administration on secreted Igs, mice were bled at different time points during the injection protocol, which was initiated on day 1. Plasma samples were collected before the first injection at day 0 and at day 5, day 11, and day 21. HEL- and SpA-treated mice exhibited unchanged concentrations of total IgM in the plasma until day 11. However, by day 21, total IgM concentrations were significantly increased in the SAg group, compared with the control group (p < 0.002) (Fig. 2,A). Notably, no changes in IgG concentrations were evident in the HEL- and SpA-treated groups during the observation period (Fig. 2 B).
In untreated mice, VH3+ IgMs represent ∼60% of total IgM (Fig. 2 C), a figure comparable to its expression in human serum. In vivo SpA exposure led to a rapid loss of VH3+ IgM in the plasma (36 ± 2.7% at day 5) compared with HEL-treated mice (58.6 ± 2.5%, p < 0.01), and this diminution persisted throughout the observation period. Interestingly, despite the significant increase of the total IgM concentration observed at D21 in the SpA-treated group, VH3+ IgMs remained significantly lower as compared with control mice.
As seen in human serum (11), VH3+ IgG concentrations in the plasma of the transgenic humanized mice were lower than VH3+ IgMs approximating 40% of total IgG. Because SpA binding is sensitive to replacement mutations in Ig VH regions, this probably reflects the presence of somatic mutations in the VH genes of the mice. As shown in Fig. 2 D, SpA treatment decreased the VH3+ IgG percentages, and this effect was detectable at day 5 (12.6 ± 2.1%, p < 0.01). These results indicate that SpA treatment induces a rapid decrease of plasma VH3+ Igs. At day 21, the amounts of IgM increased for reasons that are unclear.
Repeated injections of SpA affect B-1a, but not B-1b or B-2, cells in the PeC
As shown above, a single SpA injection induces a marked decrease of the B-1a subpopulation. To determine whether repeated administration of this SAg affects other cell subsets, we analyzed peritoneal cell populations of transgenic mice receiving five injections. As expected, there was a 45% reduction of the B-1a subpopulation (CD11b+CD5+) in the SAg-injected mice, compared with the control-treated group (p < 0.05) (Fig. 3 A). However, peritoneal B-2 cell numbers were not affected by repeated SAg injections, and their percentages were similar in the HEL-group (41.5 ± 7.4%) and in the SpA group (45.8 ± 2.6%). The percentages of B-1b cells (CD11b+CD5−) were also comparable in the two groups, and T cells were not affected by five injections of SpA. These results indicate that SpA impacts the B-1a subpopulation in the peritoneum and even repeated administration of this SAg does not modify B-1b or B-2 cell numbers.
Specific reduction of the MZ subpopulation in the spleen
Postnatal B lymphocyte development in the BM ultimately leads to generation of a heterogeneous peripheral B cell population, and a series of successive developmental stages have been characterized in the mouse spleen depending on their cell surface marker expression. Newly formed B cells (B220+CD23−CD21−) generated in the BM emigrate to the white pulp of the spleen, wherein two types of transitional (T1: B220+CD21−IgM+; and T2: B220+CD21+IgM+) B cell precursors exist (16). Transitional B cells can further differentiate into FO or MZ B cells, which differ in their localization and cell surface markers. Resident MZ cells (B220+CD23 −lowCD21+) are located at the junction of white and red pulps of the spleen, which is intimately associated with the marginal sinus. They secrete preferentially IgM and IgG, a reflection of their role in primary T-independent responses. FO recirculating B cells (B220+CD23+CD21+) produce mainly IgG1 or other Ig isotypes and engage in T cell-dependent responses (17).
Because of these anatomical and functional dichotomies in B cell subpopulations, we sought to characterize the effect of SpA treatment on splenic B lymphocyte development. We found that the number of MZ cells was reduced by 40.3% (p < 0.03) in humanized transgenic mice exposed to SpA injections, compared with control treated mice (Fig. 3,B). By contrast, other B cell populations, including the relatively abundant FO B cells, were not significantly affected (Table I). Interestingly, the repeated administration of SpA resulted in a loss of VH3+ B cells (a 37% reduction compared with HEL-treated mice, p < 0.04) in the spleen that was more severe than that seen in humanized transgenic mice injected with one single dose of SpA (a 24.8% decrease compared with control mice).
|Splenic Populations .||HEL-Treated Mice .||SpA-Treated Mice .|
|B220+SpA+||2.4 ± 0.3||1.5 ± 0.2∗|
|NF/B-1||1.9 ± 0.3||1.9 ± 0.3|
|MZ||1.2 ± 0.1||0.7 ± 0.1∗∗|
|FO||1.9 ± 0.5||2.7 ± 0.3|
|Splenic Populations .||HEL-Treated Mice .||SpA-Treated Mice .|
|B220+SpA+||2.4 ± 0.3||1.5 ± 0.2∗|
|NF/B-1||1.9 ± 0.3||1.9 ± 0.3|
|MZ||1.2 ± 0.1||0.7 ± 0.1∗∗|
|FO||1.9 ± 0.5||2.7 ± 0.3|
Two groups of five mice were treated with five i.p. injections of SpA or HEL every other day. Mice were sacrified at day 21, and splenic cell populations were analyzed. B cell subpopulations were identified by the expression of cell surface markers.
MZ, B220+CD23−/lowCD21+; FO, B220+CD23+CD21+; NF/B-1, B220+CD23−CD21−. Enumeration of B220+SpA+ B cells was performed on total spleen cells and includes all the B cell subpopulations; ∗, p < 0.04, ∗∗, p < 0.03. Values of p were determined by the Mann-Whitney U test. The values shown are ± SE.
SpA treatment results in depletion of MZ B cells expressing VH3 family genes
To determine whether SpA exposure affects the repertoire of expressed VH genes, B220+ splenocytes from control, HEL-, and SpA-treated mice were sorted into separate populations of MZ or FO B cells, and VH genes were amplified from genomic DNA of individual B cells from each subset. SpA treatment had an impact on the expressed repertoire of MZ B cells but not on that of FO B cells (Fig. 4). Within the FO subpopulation, VH3 family members were expressed by 60.6, 60.2, and 68.5% of B cells from control, HEL-, and SpA-treated mice, respectively. In contrast, the VH repertoire of MZ B cells from SpA-treated mice exhibited a significant decrease (p < 0.02) of VH3 rearrangements (36.5%), as compared with control mice (57.1%) and HEL-treated mice (48.4%). The loss of over 20% of the VH3 B cells in the MZ after SpA exposure was compensated by an increase in B cells expressing VH1 (23.8%) and VH4 genes (34.9%). In control mice, 11.7% of the MZ B cells expressed VH1 and 24.7% expressed VH4 genes. We then calculated the absolute number of VH3 expressing B cells in each subset. In the control mice there were 1.2 × 106 VH3+ FO B cells and 0.7 × 106 VH3+ MZ B cells. In SpA-treated mice there were 1.9 × 106 VH3+ FO B cells and 0.3 × 106 VH3+ MZ B cells. Therefore, the reduction of MZ B cells reflects the decrease of the absolute number of VH3 expressing B cells in this subset, a consequence of the VH3+ B cell targeting capacity of SpA.
Specific impact of SpA on the type 2 T cell-independent response
To determine the effect of SpA treatment on the humoral response, groups of SpA-treated mice were immunized with either a T cell-dependent Ag (TT) or a type 2 T-independent Ag (DNP-Ficoll). As expected, TT immunization induced the production of specific IgM and IgG Abs in both control and SpA-treated immunized mice (Fig. 5,B). By contrast, DNP-Ficoll immunization induced the production of specific IgM Abs in control mice but not in SpA-treated mice (Fig. 5 A). Consistent with the reduction of B-1a and MZ B cells, these results reveal that SpA has a functional impact on the T cell-independent branch of the humoral response.
B cells are continuously generated from hemopoietic progenitors, first, in fetal liver and, then, in adult BM. Following sequential progression through developmental stages expressing cytosolic and cell surface markers, emerging immature B cells emigrate into the peripheral immune system and give rise to heterogeneous peripheral B cell subpopulations with distinct anatomical locations and functional properties. Whereas the mechanisms that shape B cell development in normal conditions are relatively well defined, there is little knowledge of the direct impact of pathogen-derived products on human B lymphocytes. Because S. aureus strains expressing SpA exhibit a higher virulence than others (2) and because this protein binds human Igs, we turned our attention to the consequences of interactions of SpA with B lymphocytes in vivo. In the present study, we used a transgenic mouse model to investigate the impact of SpA Fab-mediated B cell SAg activity on target human Igs in vivo. The results revealed that SpA specifically reduces VH3+ B cells and impacts two B cell subpopulations, B-1a and MZ, known to exert innate-like functions in the immune system. This reduction had functional consequences on the type 2 humoral response but not on the T cell-dependent B cell response. We propose that this deleting effect represents a novel mechanism used by some infectious agents as a survival strategy that allows them to persist in mammalian hosts.
Characterization of the humoral response showed that VH3+ IgMs remained depressed in SpA-treated mice throughout the observation period, as compared with HEL-injected mice. However, injections of SpA had no negative effects on total IgM concentrations in the plasma. At day 21, 12 days after the last dose of SpA, there was an important increase of total IgM concentration that could represent a compensatory effect of the depletion resulting from SpA injections. By contrast, the significantly decreased levels of VH3+ IgG relapsed after day 5 to become moderately decreased at day 21, perhaps reflecting the slow clearance of the last SpA bolus on day 9. Taken together, these data suggest that the specific deletion resulting from SpA injections is stronger on VH3+ IgM cells than on VH3+ IgG cells.
In body cavities, B-1 cells are overrepresented compared with other lymphoid sites. Injecting SpA into the peritoneum cavity, we found that this SAg has no effect on B-2 cells, even after repeated administration. This distinctive unresponsiveness of B-2 cells is reminiscent of functional disparities previously reported for B-1 and B-2 cells. B-1 cells have been shown to display an altered responsiveness to BCR ligation compared with their B-2 cell counterparts. For example, they exhibit a diminished ability for intracellular Ca2+ mobilization, aberrant proliferation and increased apoptosis upon BCR cross-linking (18). Although anti-μ activation drives B-2 cells into the S phase, B-1 cells fail to progress into the cell cycle. Conversely, while B-2 cells require a combination of a phorbol ester and a calcium ionophore to enter the S phase, treatment with only phorbol ester stimulates B-1 cells (reviewed in Ref.18). Unlike splenic B cells, translocation of the transcription factor NF-κB to the nucleus does not occur following BCR cross-linking on peritoneal B-1 cells. This disparity in response to BCR ligation may be related to CD5-mediated negative regulation (19). Interestingly, similar changes in BCR signal transduction have been reported in anergic B cells (20, 21), known to be prone to BCR-mediated cell death. It has been suggested that the microenvironment of the peritoneum, and perhaps other pleural cavities, provides a unique milieu that favors the induction of the hyporesponsiveness of B-1 cells. The lack of SpA effect on B-2 cells in the PeC may therefore reflect their intrinsic resistance to BCR-mediated ligation in this microenvironment.
When comparing the fate of the B-1a and B-1b cell subsets in mice injected repeatedly with SpA, our experiments show a specific decrease of the percentage of B-1a, but not B-1b, cells in the PeC. This finding conflicts with previous studies of neonatal SpA inoculation showing a persistent “hole” among the B-1a and B-1b cells (22). However, in that study, the T15i transgenic “knock in” mice used have B cells expressing the canonical T15/S107 VH gene rearrangement, and most of their peritoneal VHT15-expressing B cells express diverse endogenous L chains and bear the phenotype of B-1b cells (23). By contrast, the humanized mice we have used do not exhibit a biased pattern of B-1a/B-1b distribution in the PeC. It is also possible that, because B-1b cells can replenish from progenitors in the BM under conditions of suspended feedback regulation (13), SpA neonatal treatment could alter generation of B-1b cells at this critical window of B cell development but seems unable to impact them at the adult stage.
The selective effect of SpA administration on CD5+ B-1a, as opposed to CD5− B-1b, cells we have observed may in fact reflect functional differences among these two peritoneal subsets. First, CD5 is a negative regulator of BCR signal transduction (19) and has a role in maintaining tolerance in anergic B cells (15). Second, the B-1b subset may have different cytokine requirements, i.e., be less T and B cell dependent or more macrophage-like than the B-1a subset (14). Third, B-1a (CD5+) cells are derived from fetal liver and have a self-renewing capacity that explains their persistence in adult mice. Fourth, the fate of a particular B cell depends on the specificity of its BCR and its density and on the microenvironment in which selection events occur (24). For example, transitional splenic B cells (T1 and T2) express distinct surface markers, and their BCR responsiveness are different (25). Most notably, the T2 subset generates proliferative, antiapoptotic, and differentiation signals in response to BCR engagement. In contrast, the T1 subset is relatively unresponsive to BCR stimulation. By the same token, BCR ligation in the PeC may have different impacts on lymphocyte survival of B-1a and B-1b cells.
In parallel with the marked depression of B-1a cells in the peritoneum, we found that SpA specifically decreases MZ B cells expressing VH3 genes in the spleen. It is intriguing to speculate on the lack of alterations on other splenic B cell subpopulations. First, T1 and T2 B cells are short-lived, and newly formed splenic B cells have half-lives of 3–4 days, which is consistent with their short transit time in the splenic compartment. Therefore, it is possible that the rapid differentiation of these B cell subsets in a short time-window has masked SpA effects. Second, with life spans of a few months, FO B cells represent a dominant subset of the mouse spleen (23). Although the numbers of FO B cells were not modified by SpA treatment, the density of CD23 molecules on these cells was increased (Fig. 3 B). CD23 is the low-affinity IgE FcR expressed predominantly on mature virgin B cells and lost after activation, and its expression on splenic B cells in culture is up-regulated by IL-4 and CD40 (26, 27). Because B cell SAgs interacting with VH3+ IgE have been demonstrated to induce IL-4 secretion in human FcεRI+ cells (28), overexpression of CD23 on FO B cells could be a consequence of the SpA activation of the FcεRI+ cells.
We favor the view that the disparity in responses of MZ and FO B cells to SpA reflects a functional dichotomy in these two splenic subpopulations. It has become clear that MZ B cells represent a B cell subset that differs in many respects from the predominating subset of mature recirculating FO B cells. MZ B cells are noncirculating, intermediate-sized lymphocytes located in a distinct area surrounding the B cell follicles and the periarteriolar lymphoid sheath of the spleen. Increasing evidence indicates that MZ B cells are involved in T cell-independent type 2 responses to polysaccharide Ags that form the major constituents of cell walls of encapsulated bacteria such as S. pneumonia, N. meningitis, and H. influenza (29). Their phenotype (less condensed nuclear chromatin, CD21high, CD23low, CD35high) and their higher basal levels of CD80 and CD86 (30) suggest that they are in a somewhat activated state that might result from Ag experience. When stimulated ex vivo, purified MZ B cells respond generally more rapidly to stimulation (LPS, dextran-conjugated anti-IgM or IgD Abs, or CD40 ligation) than FO B cells (30). Even though they do not proliferate after activation, BCR stimulation of MZ cells does elicit early signaling events, including Ca2+ mobilization. Thus, the activated phenotype of MZ B cells in combination with their unique anatomical location in the spleen may reflect their specific immunological functions, e.g., to respond rapidly to bloodstream infections. Although the molecular mechanisms regulating the differentiation into MZ vs FO B cells have not been fully defined, studies of mice transgenic for, or deficient in, several genes have suggested that MZ B cell development could be related to chemotactic migration to the MZ (31, 32, 33). In addition, the specificity and surface density of the BCR are critical in determining lineage commitment to different B cell subsets (24), with weak BCR signaling permissive for MZ B cell development and intermediate signals required for FO B cell development (34). It was recently found that Notch signaling facilitates generation of MZ B cells while it suppresses generation of FO B cells (35). It is therefore conceivable that, in addition to impacting MZ and FO B cells in different microenvironments of the spleen, SpA-BCR interactions are of different strengths, which leads to different effects on cell survival. Naturally, it is also possible that the SpA treatment led to defects in the motility and responsiveness to chemokines critical for the migration of MZ B cell precursors to the appropriate sites. This hypothesis has been used to explain the loss of MZ B cells in mice lacking Pyk-2 tyrosine kinase, Lsc, or DOCK-2 (31, 32, 33), and recent studies have shown a critical role of the integrins LFA-1 and α4β1 in the localization and retention of MZ B cells (36).
The observation that the SAg we have used acts specifically on B-1a and MZ B cells is reminiscent of repertoire analysis, suggesting that some MZ B cells have BCR that are similar to those of B-1 cells (30). Further evidence pointing to possible maturation of B-1 cells to become MZ B cells comes from studies in IL-7-deficient mice (37) and in mice whose Rag-2 genes can be inactivated after birth (23). Both strains lack recirculating cells but have B-1 cells and MZ B cells. Because B-1 cells exhibit the propensity of self-renewing, it is possible that the MZ B cells that developed in these mice maturated from B-1 cells (13). It can then be proposed that the SpA impact on B-1 cells blocks MZ B cell development.
It is important to consider the functional consequences of B-1 and MZ B cell depletion resulting from SpA administration. B-1 cells produce the majority of natural Abs (NAbs) and contribute substantial amounts of IgA (reviewed in Ref.18). The secreted Igs are generally germline encoded and are often reactive with carbohydrates on self-Ags, which might be shared with determinants on bacterial, viral, or tumor determinants. Studies of mice lacking NAbs have demonstrated their essential role in providing early protection against a variety of pathogens (38, 39). Thus, NAbs appear to be an early B cell subset essential for “housekeeping” responsibilities in the organism, targeting senescent RBC, oxidized membrane lipids, the products of apoptosis and intestinal microflora, and they play a role in innate immunity (40, 41).
In addition to B-1 cells, MZ B cells are poised to provide a first line of response to pathogens (30). Splenic MZ B cells appear to respond to the T cell-independent type 2 Ags DNP-Ficoll and NP-Ficoll and form an integrated part of innate immunity. They provide a strategically located buffer of polyreactive B cells in the spleen that can rapidly produce NAbs (mainly IgM) upon blood-borne and life-threatening infections with encapsulated bacteria or other microorganisms. Recent studies suggest that the immune response to Streptococcal polysaccharides requires their localization to MZ B cells with subsequent transfer to FO dendritic cells (42) and that, in a systemic infection, CD11low DC provide critical survival signals to MZ B cells. In local T cell-independent immune responses, PeC macrophages provide similar support to B-1 cells (43). Thus, because of their enrichment in the PeC, B-1 cells are ideally suited to respond to Ags that enter the organism through the gut epithelium, and MZ B cells are located so as to be able to efficiently encounter and respond to blood borne pathogens.
Persistent and recurrent S. aureus infections point to mechanisms that obstruct the development of efficient functions. The bacterium produces a series of factors that may impart various survival strategies, allowing its persistence in mammalian hosts (2). Three of them, the exotoxin SpA (44) and enterotoxins A and D (45, 46), exhibit a SAg activity for human B cells. Our findings reveal that, by virtue of its SAg activity, SpA deletes B-1a and MZ subpopulations in vivo and cripples innate immunity. It is remarkable that protein L of Peptostreptococcus magnus (47) also exhibits a potential to target B lymphocytes with innate immune functions (48). With the recent demonstration based on a transgenic mouse system that SpA can “co-opt” the self-tolerance pathways of BCR-mediated apoptosis and impact B cells important for defense against systemic infection (22, 49, 50), our data suggest that elimination of B cell subpopulations with innate immune functions may represent an important immune-evasion function that may potentiate ongoing and recurrent infections.
Finally, we would like to emphasize that in our studies of SpA (this article) and protein L (48) effects on B cell subsets, we used mouse strains expressing completely different human Ig transloci. The “five-feature” mice (51) used to study protein L do not express human VH3+ Igs. In contrast, while SpA targets VH3+ human Igs, protein L is specific for κ-L chain+ human Igs. Yet, we found that the two proteins used in soluble form result in similar B cell deleting effects in the two different mouse strains. Thus, the overall data suggest that the two bacteria, S. aureus and P. magnus, have evolved to produce proteins with SAg activity for either VH3+ or κ-L chain+ human Igs but have developed a similar strategy to subvert a common branch of B cell-mediated immune defense.
We are indebted to Abgenix (Drs. Larry Green and Shelley Belouski) for the transgenic mice used in these experiments. We thank Dr. Robert Girard for gift of DNP-BSA.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from the Institut National de la Santé et de la Recherche Médicale (Paris, France) and Agence Nationale de Recherches sur le Sida (ANRS) (Paris, France). M.V. was supported by a predoctoral fellowship from ANRS.
Abbreviations used in this paper: SpA, S. aureus, protein A; BM, bone marrow; FO, follicular; HEL, hen egg lysozyme; MZ, marginal zone; NAb, natural Ab; PeC, peritoneal cavity; SAg, superantigen; T1, transitional 1; T2, transitional 2; TT, tetanus toxoid; V, variable gene segment; XMG2KL, XenoMouse G2-KL.