Marginal zone B cell production of pathogenic natural antibodies is CR2 dependent but C3 independent

Abstract

Intestinal ischemia/reperfusion (IR)-induced damage requires complement receptor 2 (CR2) for generation of the appropriate natural antibody repertoire. Pathogenic antibodies recognize neo-antigens on the ischemic tissue, activate complement and induce intestinal damage. As C3 cleavage products act as ligands for CR2, we hypothesized that CR2^hi marginal zone B cells (MZB) require C3 for generation of the pathogenic antibodies. To explore the ability of splenic CR2⁺ B cells to generate the damaging antibody repertoire, we adoptively transfered either MZB or follicular B cells (FOB) from C57Bl/6 or Cr2^−/− mice into Rag-1^−/− mice. Adoptive transfer of wildtype CR2^hi MZB but not CR2^lo FOB induced significant damage, C3 deposition and inflammation in response to IR. In contrast, similarly treated Rag-1^−/− mice reconstituted with either Cr2^−/− MZB or FOB lacked significant intestinal damage and displayed limited complement activation. To determine if C3 cleavage products are critical in CR2-dependent antibody production, we evaluated the ability of the natural antibody repertoire of C3^−/− mice to induce damage in response to IR. Infusion of C3^−/−serum into Cr2^−/− mice restored IR-induced tissue damage. Furthermore, Rag-1^−/− mice sustained significant damage after infusion of antibodies from C3^−/− but not Cr2^−/− mice. Finally, adoptive transfer of MZB from C3^−/− mice into Rag-1^−/− mice resulted in significant tissue damage and inflammation. Together these data indicate that CR2 expression on MZB is sufficient to induce the appropriate antibodies required for IR-induced tissue damage and that C3 is not critical for generation of the pathogenic antibodies.

Introduction

Tissue damage that occurs in response to an ischemic event is signficantly magnified by the return of blood flow (reperfusion). The inflammatory response mediates ischemia/reperfusion (IR)-induced tissue damage and results in amplified pathology in many clinical conditions including myocardial infarction, stroke and intestinal ischemia (reviewed in (1–2)). Although the precise mechanism of the magnified damage during reperfusion is unknown, cellular alterations which occur during ischemia appear to be recognized by the innate immune response. Because the local response frequently progresses to systemic inflammation and multiple organ failure, treatment of the IR-induced inflammatory response and subsequent tissue damage is the subject of intense investigation.

Tissue damage resulting from IR is proposed to be mediated in part by complement activation after natural antibody recognition of neo-antigens present on the surface of ischemic tissue (2–3). Primarily found as IgM or IgG3 isotypes, natural antibodies appear to be produced by B-1 or MZB in the absence of immunization (4–9). Natural antibodies may provide protection against bacterial (10) and viral pathogens (4, 11). However, many natural antibodies also recognize self antigens and promote tissue damage in response to IR (12). Self-reactive antibodies target several proteins expressed on ischemic tissue including non-muscle myosin heavy chains subtype A and C (13), β2-glycoprotein I (14), U1-ribonucleoprotien (15) and annexin IV (16). Antibody recognition of the neo-antigens leads to complement activation. However, the initiating complement pathway remains unresolved. Although there is strong evidence supporting the lectin-binding pathway (13, 17–18), the presence of C1q deposition on ischemic tissue cannot completely rule out the contribution of the classical pathway (13). Regardless of the specific pathway, the process is antibody dependent with little known about the mechanism of autoreactive natural antibody selection.

In the mouse, complement receptor 2 (CR2) is an alternatively spliced, type I membrane glycoprotein expressed on mature B cells, follicular dendritic cells (FDC) and epithelial cells providing a linkage between innate and adaptive immunity (19). Interestingly, B-1-like MZB express higher levels of CR2 than the B-2 FOB (20). In conjunction with CD19 and CD81, CR2 comprises part of the B cell receptor complex which enhances B cell signaling and activation (21). As a co-receptor, multiple ligands bind CR2 including the C3 cleavage products, iC3b, C3dg and C3d (22–23), interferon-α (24), Epstein-Barr viral coat protein GP350/220 (23, 25–26) and CD23 (27). How the binding of these ligands specifically aids the ability of CR2 to promote an immune response is not known. However, evidence suggests that by binding complement fragments CR2 facilitates the presentation of antigens associated with complement-tagged structures to the B cell receptor (28).

Previous studies showed that Cr2^−/− mice are resistant to IR-mediated tissue damage and that administering antibodies from wildtype mice restored damage (29). These studies suggested that Cr2^−/− mice do not generate the autoreactive natural antibodies necessary for IR-induced mesenteric tissue damage (29). Moreover, these data suggest that CR2 may influence the selection of the natural antibody repertoire in such a way that results in an autoreactive subpopulation. Since CR2 is required for generation of the pathogenic antibodies, the CR2 ligands may also be required. Previous studies indicated that C3^−/− mice were also resistant to IR-induced tissue damage (30). However, it is not clear if C3 is required only for complement activation or for binding CR2 and initiating production of autoreactive natural antibodies. We hypothesized that CR2^hi MZB require C3 for generation of the pathogenic antibodies. Our results show that similar to the peritoneal B-1 B cells, the CR2^hi MZB produce the natural antibody repertoire necessary to induce tissue damage in response to IR. In addition, adoptive transfer of splenic B cells (either MZB or FOB) or administering serum from CR2 sufficient, C3^−/− mice to the antibody-deficient Rag-1^−/− mice induced normal levels of damage in response to IR. Together these data indicate that although CR2 is critical, the C3 ligands are not required for production of pathogenic, autoreactive antibodies.

Materials and Methods

Mice

Breeding pairs of C57Bl/6 mice, C3^−/− and Rag-1^−/− mice were purchased from Jackson Laboratories and Cr2^−/− mice (31) were obtained from Dr. G.C. Tsokos. All mice were bred and maintained in a 12-hour light-to-dark, temperature-controlled room and allowed food and water ad libitum in the Division of Biology at Kansas State University. Mice were maintained under specific pathogen free conditions (Helicobacter species, mouse hepatitis virus, minute virus of mice, mouse parvovirus, Sendai virus, murine norovirus, Mycoplasma pulmonis, Theiler’s murine encephalomyelitis virus, and endo- and ecto-parasites). All research was approved by the Institutional Animal Care and Use Committee and conducted in compliance with the Animal Welfare Act and other Federal statutes and regulations concerning animals.

Spleen Cell and B-cell Reconstitution

Peritoneal exudate cells (PEC) were collected from euthanized mice by peritoneal lavage using normal saline and red blood cells were lysed prior to injection. Total splenic B-cells were isolated using EasySep B-cell negative selection magnetic beads (StemCell Technologies). The B220⁺ positive B cell population was sorted into FOB (CD23^hi CD21/35^lo or IgM ^loIgD ^hi) and MZB (CD23^lo CD21/35^hi or IgM ^hiIgD ^lo) by differential staining using a Beckman Coulter MoFlo XDP flow cytometer. Most antibodies were obtained from BioLegend. Anti-CD21/35 was purchased from BD Biosciences and anti-IgD was purchased from eBioscience. Rag-1^−/− mice, 8–12 weeks old, were injected i.v. with 1–2 ×10⁶ cells from one of the following sources: C57Bl/6 whole spleen cells, C57Bl/6 PEC, C57Bl/6 FOB, C57Bl/6 MZB, Cr2^−/− PEC, Cr2^−/− whole spleen cells, Cr2^−/− FOB cells, Cr2^−/− MZB cells, C3^−/−FOB or C3^−/− MZB. Prior to use in experiments, the mice rested for 8–9 weeks to allow reconstitution. Preliminary studies indicated the sorted MZB cells expressed the marginal zone marker, CD9, and did not express B1 B cell markers, CD11b and CD5 (Supplementary figure 1A). In addition, wildtype MZB splenic cells sorted with either (CR2 and CD23) or (IgM and IgD) retained MZB phenotype at 2 mo post adoptive transfer indicating a lack of immature B cells (supplementary figure 1B). After IR or Sham treatment and euthanasia, reconstitution of all Rag-1^−/− mice was verified by staining spleen cells for B220 and IgM and compared to C57Bl/6 control mice.

Ischemia/Reperfusion

Ischemia/reperfusion was performed on ketamine/xylazine anesthetized mice. Following a midline laparotomy, the mice were allowed to stabilize for 30 minutes while maintaining their body temperature using a water-circulating heat pad. Buprenorphine was administered locally for pain and peritoneal desiccation prevented by placing warm, saline moistened gauze over the abdominal cavity. The superior mesenteric artery was then identified, isolated, and a small vascular clamp applied. Ischemia was noted by intestinal blanching. Sham treated animals underwent the same procedure as the ischemic mice without occlusion of the superior mesenteric artery. After 30 min of ischemia the clamp was removed and the blood flow restored for 2 hours. Some experiments reconstituted Rag-1^−/− mice with 200 μl of serum or 100μg of Protein L purified antibody from C3^−/− or Cr2^−/− mice by i.v. injection 15 minutes prior to the resumption of blood flow. As described previously, C3^−/− mice were reconstituted by i.p. injection of 200 μl freshly collected Rag-1^−/− serum 20 min prior to laparotomy (32). The mice were then euthanized and serum, spleen and 2 cm sections of the small intestine were collected and used for histological and other analyses.

Antibody Isotypes

Serum was collected by cardiac puncture following IR or Sham treatment and euthanasia. Aliquots were isotyped and analyzed as per manufacturer’s protocol using a mouse Isotyping Milliplex kit on a Luminex SD200 (Millipore).

Histology and Injury Scoring

A 2 cm intestinal section was immediately fixed in 10% buffered formalin and embedded in paraffin for H&E staining of 8 μm transverse sections. Mucosal injury was graded on a six-tiered scale adapted from Chiu et al. (33) as described previously (29). Briefly, the average injury score (0–6) of 75–150 villi per mid-jejunum tissue section was determined in a blinded manner. Normal villi were assigned a score of zero; villi with tip distortion were assigned a score of 1; a score of 2 was assigned when Guggenheims’ spaces (lifting of the epithelium from the lamina propria to form a space) were present; villi with patchy disruption of the epithelial cells were assigned a score of 3; a score of 4 was assigned to villi with exposed but intact lamina propria with epithelial sloughing; a score of 5 was assigned when the lamina propria was exuding; finally, villi that displayed hemorrhage or were denuded were assigned a score of 6.

C3 Deposition

After Sham or IR, a 2 cm mid-jejunal section was frozen in O.C.T. freezing medium and stored at −80°C until used. Intestinal cryosections (8 μm) were fixed in cold acetone and non-specific binding was blocked using 10% donkey serum in PBS. Tissues were stained for C3 deposition using a rat-anti-mouse C3 antibody (Hycult Biotechnologies) followed by a Texas-red conjugated secondary antibody (Jackson Immunoresearch). Serial sections stained with isotype control antibodies were used as background. A blinded observer examined the slides by fluorescent microscopy using a Nikon 80i fluorescent microscope and acquired images using a CoolSnapCf camera (Photometrics) and MetaVue Imaging software (Molecular Devices).

PGE₂ Determination

Immediately after collection, a 2 cm intestinal section was minced and washed in 4°C freshly oxygenated Tyrode’s buffer (Sigma-Aldrich). The tissue was then incubated at 37°C for 20 minutes and the supernatants collected. Prostaglandin E₂ (PGE₂) concentrations were determined using enzyme immunoassay kits (Cayman Chemical) and standardized to the total tissue protein content as determined by BCA assay (Pierce).

Statistical Analysis

Data are presented as mean ± SEM and were compared by unpaired T test or one-way ANOVA with post hoc analysis using Newman-Keuls test (GraphPad/Instat Software). The difference between groups was considered significant when p <0.05.

Results

CR2⁺ splenic cells restore IR-induced tissue damage to Rag-1^−/− mice

Previous studies indicated that peritoneal B-1 B cells produce naturally occurring antibodies required for IR-induced tissue damage (34). Because CR2^hi splenic MZB are similar to B-1 B cells (35), we hypothesized that MZB may also restore damage to IR-resistant Rag-1^−/− mice. To test this hypothesis, we adoptively transferred either PEC or splenic cells from C57Bl/6 mice into Rag-1^−/− mice, 8–9 wks prior to IR and then determined IR-induced intestinal damage. Similar to previous studies, adoptive transfer of wildtype PEC resulted in significant damage (Fig. 1). As hypothesized, splenic cells also restored IR-induced intestinal injury in Rag-1^−/− mice (Fig. 1). In contrast, IR induced little to no damaging effects in Rag-1^−/− mice after adoptive transfer of either PEC or spleen cells from Cr2^−/− mice (Fig 1). Thus, CR2⁺ splenic B cells are sufficient for IR-induced damage.

Figure 1. CR2⁺ spleen cells or PEC restore IR-induced tissue damage in Rag-1^−/− mice.

Peritoneal exudate cells (PEC) or spleen cells from C57Bl/6 (B6) or Cr2^−/− mice (1–2 × 10⁶)were adoptively transferred to Rag-1^−/− mice 8–9 weeks prior to Sham or IR treatment. Additional controls included Sham or IR treatment of C57Bl/6, Cr2^−/− and Rag-1^−/− mice. Approximately 75–150 villi per H&E stained intestinal section were scored for injury as described in the Materials and Methods. Injury scores for the sham-treated mice for all experimental conditions were pooled. Each bar represents the group average ± SEM with 4–8 mice per group. Using ANOVA, * denotes significant difference (p<0.05) from Pooled Sham treatment, φ denotes significant difference from C57Bl/6 IR, and ψ indicates significant difference from Rag-1^−/− IR.

Adoptive transfer of marginal zone B cells restores IR induced damage and complement activation to Rag-1^−/− mice

To clarify which splenic cell population required CR2 expression and restored damage, we isolated and adoptively transferred wildtype splenic FOB and MZB into Rag-1^−/− mice. Wildtype B220⁺ B cells were isolated and sorted using differential staining and FACS based on CD23 and CR2with MZB expressing CR2^hi, CD23^lo and FOB expressing CR2^lo, CD23^hi. Reconstitution of each animal was confirmed by flow cytometry (data not shown) and further verified by serum IgG1, IgG2b, IgG3 and IgM concentrations (Table I). Both MZB and FOB cells produced IgG1and IgG2b as indicated in Table I. Adoptive transfer of wildtype MZB resulted in significantly more IgM than transfer of wildtype FOB. Importantly, after adoptive transfer of wildtype MZB, IR induced significant damage in Rag-1^−/− mice (Fig 2A, C, D, F). Although transfer of wildtype FOB slightly elevated the IR-induced injury score, the injury score was not significantly different from Rag-1^−/− without adoptively transferred cells and was significantly different from Rag-1^−/− with MZB. As MZB are CR2^hi and CR2 expression is required for production of the pathogenic natural antibodies, MZB (IgM^hi, IgD^lo) and FOB (IgM^lo, IgD^hi) from Cr2^−/− mice were adoptively transferred as a control. Despite equivalent spleen cell numbers, adoptive transfer of FOB or MZB from Cr2^−/− mice resulted in significantly less IgG3 and IgM production compared to adoptive transfer of respective wildtype cells (Table I). In addition, adoptive transfer of Cr2^−/− MZB to Rag-1^−/− mice not only failed to restore tissue damage to wildtype levels in response to IR but also did not significantly enhance injury compared to untreated Rag-1^−/− after IR (Fig 2A, B, C, G). Similar to wildtype FOB, IR induced no significant damage after adoptive transfer of FOB from Cr2^−/− mice (Fig. 2A–E). Together these data suggest that the presence of CR2⁺ MZB is sufficient to induce tissue damage in response to IR.

Table I.

Serum antibody concentration after adoptive transfer, μg/ml

Treatmenta	IgG1	IgG2b	IgG3	IgM
B6 MZB	44.79±21.88b	66.08±24.27	16.95±6.93	118.10±19.89
B6 FOB	91.47±31.78	122.2±26.37	7.81±2.73	78.75±13.69
Cr2−/− MZB	26.08±5.92	51.75±7.28	6.34±1.90	98.07±7.14
Cr2−/− FOB	11.51± 6.05	14.83±5.42	5.39±1.57	56.16±10.40
C3−/− MZB	896.1±103.9	>500±0	24.40±8.10	766.4±119.8
C3−/− FOB	>1000±0	445.3±54.75	24.38±3.45	446.1±93.93

^aRag-1^−/− mice were adoptively transferred with MZB or FOB from either C57Bl/6 (B6) Cr2^−/− or C3^−/− mice 8–9 wk prior to antibody isotype determination by multiplex analysis.

^bData are expressed as mean ± SEM of 5–9 animals per group.

Figure 2. Adoptive transfer of MZB or PEC from C57Bl/6 mice restored IR-induced tissue damage in Rag-1^−/− mice.

A. Follicular (FOB) or marginal zone (MZB) B cells or peritoneal exudate cells (PEC) (1.5–2 × 10⁶) from either C57Bl/6 or Cr2^−/− mice were adoptively transferred to Rag-1^−/− mice 8–9 week prior to Sham or IR treatment. C57Bl/6, Cr2^−/− and Rag-1^−/− mice were subjected to Sham or IR treatment as positive and negative controls. H&E stained intestinal sections were scored for injury as described in Materials and Methods. B-G. Representative H&E stained, formalin fixed intestinal sections from each treatment group are shown. C57Bl/6 IR (B), Rag-1^−/− IR (C),. Rag-1^−/− + C57Bl/6 FOB (D), Rag-1^−/− + Cr2^−/− FOB (E), Rag-1^−/− + C57Bl/6 MZB (F), Rag-1^−/− + Cr2^−/− MZB (G). Each bar represents the group average ± SEM with 7–9 mice per group. Using ANOVA, * denotes significant difference (p<0.05) from Pooled Sham treatment, φ denotes significant difference from C57Bl/6 IR, and ψ indicates significant difference from Rag- 1 ^−/− IR. Injury scores for the sham-treated mice for all experimental conditions were pooled. Magnification for all photomicrographs is X100 and the bar=0.8μm.

Previous studies indicated that IR-induced tissue damage requires complement activation (36–38). After adoptive transfer, we assessed complement activation by C3 deposition on intestinal tissue sections after Sham or IR treatment. Immunohistochemistry of intestinal tissues from wildtype C57Bl/6 mice after IR (Fig. 3B) but not Sham (Fig. 3A) treatment showed significant C3 deposition on the basolateral side of the epithelium along the majority of the villi. As expected, no C3 deposits were detected on Rag-1^−/− and Cr2^−/− mice (Fig. 3E, F) after IR. Tissue derived from IR-treated Rag-1^−/− mice reconstituted with C57Bl/6 MZB cells exhibited patterns of C3 deposition similar to C57Bl/6 mice (Fig. 3C). Adoptive transfer of control Cr2^−/−MZB into Rag-1^−/− resulted in limited intestinal C3 deposition which was visible only at the tips of the villi (Fig. 3G). In contrast, C3 staining of the tissue from Rag-1^−/− mice reconstituted with FOB from either C57Bl/6 or Cr2^−/− mice showed minimal C3 deposition (Fig. 3D, H). C3 deposition suggested decreased complement activation and correlated with the lower injury scores shown in Figure 2. Thus, the antibody repertoire of Rag-1^−/− mice after adoptive transfer with wildtype but not Cr2^−/− MZB activated complement and caused tissue damage similar to wildtype mice in response to IR. Together these data suggest that CR2⁺ MZB cells are sufficient to generate the autoreactive antibody repertoire required for complement-mediated, IR-induced tissue damage.

Figure 3. C3 deposition was restored in Rag-1^−/− mice after adoptive transfer of C57Bl/6 MZB.

Intestinal sections from Rag-1^−/− mice without transfer (E) or after adoptive transfer of Marginal zone B cells (MZB) or follicular B cells (FOB) from C57Bl/6 (B6) or Cr2^−/− mice (C, D, G, H) were stained for C3 deposition after IR treatment. Similarly stained tissue sections from C57Bl/6 Sham (A) or IR (B) or Cr2^−/− IR (F) treated mice are shown as controls. Photomicrographs are representative of 6 photos of each intestinal section with 3–5 animals per treatment group.

Representative microphotograph magnification is X200.

Antibodies from C3^−/− or C4^−/− mice restore IR-induced damage to Rag-1^−/− mice

Previous studies indicated that IR-induced complement-mediated damage was attenuated in C3^−/− mice presumably due to a lack of complement activation; however the specific role of C3 in pathogenic antibody production was not determined (34, 38). Because CR2 is required for IR-induced damage and C3 cleavage products are CR2 ligands, we examined the role of the C3 in generation of the auto-reactive antibodies. To determine if the CR2⁺ B cells found in C3^−/− mice produce the damaging antibody repertoire, serum or antibodies purified from C3^−/− mice were injected into Cr2^−/− or Rag-1^−/− mice prior to IR or Sham treatment. As expected, untreated Rag-1^−/−, C3^−/− and Cr2^−/− mice did not sustain IR-induced tissue damage (Fig. 4A, C). After infusion of C3^−/− serum or purified antibodies into either Rag-1^−/− (Fig. 4A, E–F) or Cr2^−/− (data not shown) mice, significant intestinal damage was observed in response to IR. As CR2 also binds C4b, it was possible that C4 was required for CR2 induction of antibody. However, similar to C3^−/−serum, infusion of serum from C4^−/− mice into Rag-1^−/− induced significant injury (2.03 ± 0.3 injury score). PGE₂ expression was analyzed as evidence of an inflammatory response. Correlating with intestinal damage, infusion of Rag-1^−/− mice with antibodies or serum from C3^−/− mice significantly increased intestinal PGE₂ expression in response to IR (Fig. 4B). In contrast, injection of Cr2^−/− serum or antibodies into Rag-1^−/− mice resulted in minimal IR-induced tissue damage and background levels of intestinal PGE₂ (Fig. 4A, B, D and data not shown). To verify complement activation within these mice, we examined C3 deposition by immunohistochemistry after IR (Fig. 5). Similar to intestinal injury, C3 was deposited in wildtype but not Rag-1^−/− mice (Fig. 5A, B). However, significant C3 was deposited on intestines from Rag-1^−/− mice which received serum from C3^−/− but not Cr2^−/− mice (Fig. 5C, D). Together these data indicate that C3^−/−mice produce the appropriate antibody repertoire to induce complement activation in response to IR.

Figure 4. Immunoglobulin from C3^−/− mice induces injury and PGE₂ production in Rag-1^−/− mice after IR.

A. Mucosal injury was determined from H&E stained intestinal sections from Rag-1^−/− or Cr2^−/− mice which received either protein L purified Ig or serum from C3^−/− or Cr2^−/−mice by i.v. injection 15 min prior to reperfusion. B. PGE₂ was analyzed from supernatants of intestinal tissues as described in the Materials and Methods. Each bar represents the mean ± SEM with 4–6 animals per treatment group. C–F. Representative H&E stained, formalin fixed intestinal sections are shown from each treatment group. C3^−/− IR (C), Rag-1^−/− IR+Cr2^−/− Sera (D), Rag-1^−/−IR+C3^−/− Sera (E), Rag-1^−/− IR+C3^−/− Ig (F). Using ANOVA, * denotes significant difference (p<0.05) from Pooled Sham treatment, φ denotes significant difference from C57Bl/6 IR and ψ indicates significant difference from Rag-1^−/− IR. Magnification for all photomicrographs is X100 and the bar=0.8μm.

Figure 5. Immunoglobulin from C3^−/− but not Cr2^−/− mice induces complement activation in Rag-1^−/− mice.

Intestinal sections from C57Bl/6 (B6) (A) or Rag-1^−/− mice without (B) or with infusion of C3^−/− (C) or Cr2^−/− serum (D) were subjected to IR and then stained for C3 deposition. Photomicrographs are representative of 6 photos of each intestinal section from 3–5 animals per treatment group. Representative microphotograph magnification is X200.

Antibody deficient, C3 sufficient serum restores IR-induced damage to C3^−/− mice

To confirm that antibodies from CR2⁺ B cells are sufficient for IR-induced damage in the absence of C3, C3^−/− mice were infused with fresh Rag-1^−/− serum prior to IR or Sham treatment. After infusion of freshly-obtained, C3-containing serum into C3^−/− mice, the injury score of Sham treated mice was unchanged compared to Sham treated C3^−/− mice. In contrast, infusion of C3⁺, Rag-1^−/− serum into C3^−/− mice prior to IR induced significant injury, although injury was attenuated when compared to wildtype mice (Fig. 6A, C). The attenuated damage is likely due to dilution of C3. In addition, serum reconstitution of C3^−/− mice significantly increased PGE₂ production compared to either Sham or IR treatment of C3^−/− mice (Fig. 6B). To verify that the C3^−/− MZB produce sufficient antibodies for IR-induced intestinal damage, C3^−/− MZB and FOB were adoptively transferred to Rag-1^−/− mice 2 months prior to IR. As expected, MZB from C3^−/− mice restored IR-induced intestinal damage, PGE₂ production and complement activation in Rag-1^−/− mice (Fig. 6 and data not shown). Surprisingly, C3^−/− FOB induced significant intestinal injury and PGE₂ production at a level similar to C3^−/− MZB cells (Fig. 6). In addition, adoptive transfer of either C3^−/− cell type induced significant antibody production (Table I). Together these data indicate that C3 is required for complement activation but is not required as a CR2 ligand in the generation of pathogenic antibodies.

Figure 6. Pathogenic antibody production is C3 independent.

Wildtype C57Bl/6 (B6), C3^−/−, or Rag-1^−/− mice were subjected to IR with or without restoration of C3-containing Rag-1^−/− serum into C3^−/− mice or C3^−/− marginal zone cells (MZB) or C3^−/− follicular zone cells (FOB) adoptively transferred to Rag-1^−/− mice. Mucosal injury (A) and PGE₂ (B) were analyzed as described in the Materials and Methods. Representative H&E stained, formalin fixed intestinal sections are shown from each treatment group. C3^−/− IR + Rag-1^−/− Sera (C), Rag-1^−/− IR+C3^−/− MZB (D), Rag-1^−/−IR+C3^−/− FOB (E). Each bar represents the mean ± SEM of 4–8 mice per treatment group. Injury scores for the sham-treated mice for all experimental conditions were pooled. Using ANOVA, *denotes significant difference (p<0.05) from Pooled Sham treatment, φ denotes significant difference from C57Bl/6 IR and ψ indicates significant difference from Rag-1^−/− IR. Magnification for all photomicrographs is X100 and the bar=0.8μm.

Discussion

Previous studies indicated that IR-induced, complement mediated tissue damage requires CR2 for generation of the pathogenic antibodies (12, 29, 38–39). In addition, IR-induced injury required C3 but the exact role of C3 in the production of pathogenic antibodies was unclear (30). As a central component in the complement cascade, C3 is required for complement activation during complement dependent tissue injury. However, C3 cleavage products, iC3b, C3d and C3dg, also bind CR2 resulting in enhanced antibody production. In this study, we demonstrate that although required for complement activation, C3 is not required for generation of the pathogenic antibodies essential for IR-induced tissue damage. Furthermore we established that CR2⁺ splenic MZB are sufficient for IR-induced tissue damage. Thus, the MZB production of pathogenic antibodies required for IR-induced tissue damage is CR2 dependent and C3 independent.

The role of complement, and therefore C3, in IR-mediated tissue damage is well established (reviewed in (40)). However, the requirement of C3 cleavage fragments for production of the appropriate natural antibody repertoire to initiate complement activation was not known. We demonstrated that injection of purified antibodies or serum from C3^−/− mice into Rag-1^−/− or Cr2^−/− mice restored IR-mediated mesenteric damage and inflammation. In addition, adoptive transfer of C3^−/− but not Cr2^−/− MZB to Rag-1^−/− mice resulted in IR-induced damage. Therefore, the CR2 dependent generation of the pathogenic antibody repertoire does not require C3 cleavage fragments. The generation of CR2-dependent pathogenic antibodies in the absence of C3 has been demonstrated previously as the CR2-mediated antibody response to Streptococcus pneumoniae is C3 independent but CD19 dependent (41). Although the specific ligand(s) of CR2 has not been elucidated, IFN-α is a strong candidate as a recent study showed that IFN-α enhances MZB production of pathogenic autoantibodies (42). Finally, the current studies do not rule out enhanced antibody production due to C3 fragments as previously shown to occur in the absence of CR2 (43).

Since C3 degradation products are the primary ligand for CR2 during antibody production, we expected that C3 and CR2 deficient mice would produce similar antibody repertoires. Surprisingly, transfer of C3^−/− sera or Ig purified from C3^−/− mice resulted in significant IR-induced tissue damage. In addition, adoptive transfer of C3^−/− FOB significantly increased antibody production. Thus, in the absence of C3, both cell types produce pathogenic antibodies, which induce intestinal damage in response to IR, as well as IR-induced inflammation. As C3 degradation products are an important ligand for CR2 and CR2 deficient B cells are not tolerized to soluble hen egg lysozyme (44), it is possible that C3 is critical to induction of tolerance of the CR2^lo FOB. This hypothesis is supported by recent data indicating that C3 deficiency up-regulates CR3 expression on splenic B cells (45) and CR3 and iC3b are critical for TGF-β and IL-10 production which lead to tolerance (46). In contrast, excessive CR2 decreases auto-antibody production (47). Thus, in the absence of the CR2 ligand (C3^−/− mice), both FOB and MZB produce increased autoantibodies. It is also possible that C3^−/− mice have an increased number of B cell precursors or B1 B cells which contaminated the sorted cells. However, the adoptively transferred, wildtype and C3^−/− B cells appeared similar when stained for B cell markers after IR (data not shown).

Innate-like, peritoneal B-1 B cells produce natural antibodies which restore IR-induced injury (48). Both B-1 B cells and MZB produce IgM and IgG3 isotypes, recognize microbial and self antigens and participate in T-independent antigen responses (reviewed in (35)). A recent study suggested that both MZB and B-1 B cells use a similar mechanism for production of the pathogenic natural antibodies (7). The use of either phosphoinositide 3-kinase p110δ deficient mice or a specific inhibitor of p110δ inhibited production of cardiac myosin antibodies by both MZB and B-1 B cells (7). In the current study, adoptive transfer of either PEC B-1 B cells or MZB to Rag-1^−/− mice restored IR-induced intestinal damage and inflammation as identified by increased complement activation and PGE₂ production. Thus both innate-like B cell populations produce the pathogenic antibodies sufficient for IR-induced tissue damage possibly via a distinct molecular mechanism from B-2 B cells.

Previous results indicated that Cr2^−/− mice have normal total Ig concentrations despite expressing significantly less IgM and IgG to annexin IV, a primary IR-induced antigen (16, 49). In the current study, adoptive transfer of Cr2^−/− cells also resulted in dramatically lower serum IgM and IgG3 titers compared to adoptive transfer of wildtype cells. As the total B cell number recovered from the spleens after reconstitution was not different, the decreased antibody titer reflects the role of CR2 in antibody production. Similar to these results, recent evidence indicated that compared to wild-type mice, Cr2^−/− mice expressed less IgM and IgG3 in response to T-independent type 2 antigens (41). The decreased production of antibody was also complement independent (41). Although the exact function of CR2 in production of these pathogenic antibodies remains unknown, antibody generation is C3 independent.

Although previous studies indicate that CR2 is required for intestinal IR-induced tissue damage, the specific cell types that require CR2 is not clear. MZB express high CR2 levels, but other host cells, including the follicular dendritic cells (FDC) (50) and T cells (51) also express CR2. Adoptive transfer of MZB from Cr2^−/− mice to CR2 sufficient Rag-1^−/− mice did not restore IR-induced damage. Despite attenuated tissue damage, immunohistochemistry detected limited C3 deposition at the villi tips in tissue from Rag-1^−/− mice reconstituted with Cr2^−/− MZB cells. In contrast, no C3 deposition was detectable on intestinal tissue of IR-treated Cr2^−/− mice. It is possible that CR2 expression on the host cells induces production of small quantities of pathogenic antibodies by the Cr2^−/− MZB. This hypothesis is supported by previous studies of CR2 and germinal center formation. Although optimal germinal center formation requires CR2 on both B cells and FDC, CR2 expression on either B cells or FDC is sufficient for germinal center formation (50, 52–53). Another possibility is that CR2⁺ B cells contribute to the IR-mediated intestinal damage in an antibody-independent manner. B cell depletion of wildtype mice prior to IR resulted in attenuated intestinal damage suggesting an additional role for CR2⁺ B cells (54). Therefore, although other cell types may also require CR2, MZB require CR2 expression to provide the natural antibody repertoire necessary for IR-induced mesenteric tissue damage.

In summary, we demonstrate that similar to B-1 B cells, the CR2⁺ MZB also produce antibodies which induce pathogenicity in response to IR. Importantly, although CR2 expression is required, the generation of these pathogenic antibodies is C3 independent. Further investigation is required to elucidate the exact role of CR2 in the generation of the natural antibody repertoire and the specific CR2 ligand(s) inducing these pathogenic antibodies.

Abbreviations used

IR

ischemia/reperfusion

MZB

marginal zone B cells

FOB

follicular B cells

PEC

peritoneal exudate cells

FDC

follicular dendritic cells