Synergistic Deposition of C4d by Complement-Activating and Non-activating Antibodies in Cardiac Transplants

Authors


* Corresponding authors: William M. Baldwin III, wbaldwin@jhmi.edu, Barbara A. Wasowska, bwasowsk@jhmi.edu. Both senior authors contributed to this paper equally.

Abstract

The role of non-complement-activating alloantibodies in humoral graft rejection is unclear. We hypothesized that the non-complement-activating alloantibodies synergistically activate complement in combination with complement-activating antibodies. B10.A hearts were transplanted into immunoglobulin knock out (Ig-KO) mice reconstituted with monoclonal antibodies to MHC class I antigens. In allografts of unreconstituted Ig-KO recipients, no C4d was detected. Similarly, reconstitution with IgG1 or low dose IgG2b alloantibodies did not induce C4d deposition. However, mice administered with a low dose of IgG2b combined with IgG1 had heavy linear deposits of C4d on vascular endothelium. C4d deposits correlated with decreased graft survival. To replicate this synergy in vitro, mononuclear cells from B10.A mice were incubated with antibodies to MHC class I antigens followed by incubation in normal mouse serum. Flow cytometry revealed that both IgG2a and IgG2b synergized with IgG1 to deposit C4d. This synergy was significantly decreased in mouse serum deficient in mannose binding lectin (MBL) and in serum deficient in C1q. Reconstitution of MBL-A/C knock out (MBL-KO) serum with C1q-knock out (C1q-KO) serum reestablished the synergistic activity. This suggests a novel role for non-complement-activating alloantibodies and MBL in humoral rejection.

Abbreviations: 
Ig-KO

immunoglobulin knock out

MHC

major histocompatibility complex

MBL

mannose binding lectin

C1q-KO

C1q knock out

MBL-KO

MBL-A/C knock out

C1q/MBL-KO

C1q/MBL-A/C triple knock out

MASP-1

MBL-associated serine protease

Introduction

Preformed and newly formed post-transplantation antibodies are a significant threat to allografts. Clinical studies have established that the presence of alloantibodies to human leukocyte antigens (HLA) correlates with the development of humoral rejection of cardiac (1) and renal transplants (2–4). In addition, episodes of acute rejection have been associated with the subsequent development of chronic rejection (3,5), which is the greatest impediment to the long-term survival of cardiac allografts. For these reasons, in recent years there has been a growing interest in understanding the mechanisms by which alloantibodies contribute to graft rejection.

Alloantibodies have been demonstrated to contribute to the rejection process through their ability to activate the complement cascade. Clinically irreversible graft rejection is correlated with the presence of antibodies to HLA in the circulation and deposition of C4d in the graft (6). This has also been demonstrated experimentally in animal models of graft transplantation (7,8). In previous studies, we have shown that complement-activating alloantibodies administered to immunoglobulin knock out (Ig-KO) cardiac allograft recipients significantly accelerated graft rejection compared to counterparts that had been administered non-complement-activating alloantibodies (9,10). Furthermore, we also demonstrated significantly prolonged cardiac (11) and lung (12) allograft survival in rodents deficient in one terminal complement component, C6. These studies demonstrate that the activation of complement contributes significantly to allograft rejection.

However, the effects of alloantibodies, which do not activate complement are less clear. Clinical studies have shown that in some cases cardiac and renal allograft recipients have high titers of non-complement-activating alloantibodies, as measured by flow cytometry or ELISA (13–15). These antibodies do not necessarily correlate with complement activation and graft survival (15), while antibodies detected by complement activation assays are considered predictive of graft rejection (16). Our previous studies have also shown that IgG1, which does not activate complement, can comprise the dominant antibody response to allogeneic blood transfusions in rats (7) and cardiac allografts in mice (10). In a mouse model of heart transplantation, we have shown that passive transfer of high doses of complement-activating antibodies to Ig-KO recipients significantly shortened the survival time of the allografts (9,10). This effect was dose dependent, and low doses of complement-activating alloantibodies did not accelerate graft rejection. Non-complement-activating alloantibodies administered over a wide range of doses also did not accelerate graft rejection. In contrast, cardiac allografts were vigorously rejected within 48 h in mice that had been given low doses of complement-activating alloantibody in combination with a high dose of non-complement-activating alloantibody. We therefore concluded that complement-activating and non-activating alloantibodies can synergize to accelerate graft rejection.

The aim of this study was to investigate a potential mechanism of the synergistic effect of complement-activating and non-activating alloantibodies in the acceleration of allograft rejection, namely that non-complement-activating alloantibodies can synergistically activate complement in conjunction with complement-activating antibodies. In the process, we have developed and characterized an anti-mouse C4d antibody to study C4d deposition in a mouse model of antibody-mediated cardiac allograft rejection. We then used this antibody to study whether non-complement-activating antibodies can synergize with complement-activating antibodies in our model of humoral graft rejection, and examined the roles of soluble complement-activating serum proteins, namely mannose binding lectin (MBL) and C1q in the interaction between these antibodies.

Materials and Methods

Development of polyclonal rabbit anti-mouse C4d antibody

To develop a polyclonal anti-mouse C4d antibody, a 14-mer peptide (RPTAPRSPTEPVPQ) corresponding to amino acids 1223–1236 of mouse C4 was synthesized. A cysteine was added to the C-terminal to conjugate to keyhole limpet hemocyanin for immunization. Rabbits were immunized with 200 μg of the peptide emulsified in complete Freund's adjuvant and boosted five times in two-weekly intervals with 100 μg of the peptide in incomplete Freund's adjuvant. The rabbits were bled 7 days after the fifth booster and peptide-specific antibodies were isolated by affinity purification.

Western blotting

C57BL6 mouse serum was incubated for 0, 1 or 4 days in a 37°C water bath in order to cause spontaneous cleavage of serum C4 into C4d as described by Anderson and Stroud (17). A solution containing 1% serum was then boiled and run on a precast 7–15% Tris-HCL gel (Bio-Rad Laboratories, Hercules, CA) under nonreducing conditions for 2 h at 100 V. The gel was transferred to a PVDF membrane (Bio-Rad Laboratories) for 1 h at 80 V, then blocked for 3 h in 1% BSA at room temperature. The membrane was incubated overnight at 4°C in either a 1:10 000 dilution of the polyclonal rabbit anti-mouse C4d antibody or a 1:10 dilution of a commercially available rat antibody to mouse C4 (clone 16D2, Abcam, Inc, Cambridge, MA), then washed five times before incubation with a 1:50 000 dilution of a peroxidase-conjugated donkey anti-rabbit secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA) for 1 h. After five washes, the membrane was developed with an ECL reagent (Amersham Biosciences, Piscataway, NJ) before being exposed to Hyperfilm ECL (Amersham Pharmacia Biotech, Buckinghamshire, UK).

Immunohistology

Full cross sections through the center of cardiac grafts obtained at the time of sacrifice were fixed in acidic 60% methanol, embedded in paraffin, and sectioned at 7 μm. C4d staining was conducted on methanol fixed tissue using our newly developed rabbit anti-mouse C4d antibody and a secondary peroxidase-conjugated donkey anti-rabbit antibody (Jackson ImmunoResearch Laboratories).

Alloantibodies

Alloantibodies to mouse MHC class I and isotype controls were purified from hybridomas obtained from the American Type Culture Collection (ATCC, Manassas, VA). Clone 16.1.2N produces an antibody specific for H-2Kk and Dk antigens of IgG2a subtype. Clone 15.1.5P generates an antibody specific for H-2Kk and Dk of IgG2b subtype. Clone AF3–12-1–3 generates an IgG1 antibody specific for H-2Kk. 16.1.2N and 15.1.5P are both capable of activating complement, while AF3–12-1–3 does not activate complement (10,18). Clone RPC5.4 and MPC-11OUA (ATCC) produced IgG2a and IgG2b isotype control antibodies, respectively. Isotype control of IgG1 subtype (MOPC-31C) was obtained from BD Biosciences (San Jose, CA). All hybridomas were grown in serum free media (Invitrogen, Carlsbad, CA), and purified antibodies were obtained from supernatants by purification on a Protein G column (Amersham Biosciences). Concentrations of the antibodies were determined by ELISA kits specific for their respective subtypes (Bethyl Laboratories, Inc., Montgomery, TX).

Mouse model of cardiac transplantation

The mouse model of antibody-mediated cardiac allograft rejection used in this study has been previously described (9). In brief, male C57 BL/6J (H-2b), B10.A (H-2a), C57B-Igh-6tm1Cgn (Ig-KO) and B6.129S4-C4tm1Crr (C4-KO) inbred mice were purchased from Jackson Laboratories (Bar Harbor, ME) and used at 8–12 weeks of age. B10.A (H-2a) hearts were transplanted into wild-type (WT) or Ig-KO C57BL/6 (H-2b) recipients. Heterotopic cardiac transplantation was performed under methoxyfluorane (Pittman-Moore, Inc., Mundelein, IL) inhalational anesthesia using the standard technique as previously described (9,19).

Passive transfer of monoclonal antibodies

Ig-KO and C4-KO cardiac allograft recipients were injected intravenously with IgG2b (clone 15.1.5P) at doses of either 100 or 25 μg. IgG1 (clone AF3–12-1–3) was injected at a dose of 100 μg. Control cardiac allograft recipients were injected with the same dose of isotype-matched control monoclonal antibodies (9). Injections were administered i.v. 10 days after transplantation.

Blocking studies for the determination of shared epitopes between monoclonal alloantibodies

Fifty microliters of aliquots containing 2 × 105 B10.A (H-2a) lymph node lymphocytes were preincubated with various dilutions of monoclonal antibodies specific for H-2a of subtypes IgG1, IgG2a or IgG2b (AF3–12-1–3, 16.1.2N, 15.1.5P, respectively). After a 30-min incubation in 4°C, a second antibody to MHC, different from the first, was added at a concentration that was expected to lie within the linear portion of a previously calculated titration curve. The cells were washed twice, then stained with FITC-labeled anti-IgG subtype antibodies specific for either IgG1, IgG2a or IgG2b (BD Biosciences). The cells were fixed in 1% formalin and analyzed on a FACScan flow cytometer (BD Biosciences).

Measurement of C4d deposition on sensitized lymphocytes by flow cytometry

Activation of complement by different IgG subclasses of alloantibodies was measured by flow cytometry. Fifty microliters of aliquots containing 2 × 105 B10.A (H-2a) lymph node lymphocytes were incubated with various dilutions of monoclonal antibodies specific for H-2a of subtypes IgG1, IgG2a or IgG2b (AF3–12-1–3, 16.1.2N, 15.1.5P, respectively). The cells were washed twice, then incubated for 30 min at 37°C in a Gelatin Veronal Buffer 2+ (GVB2+), (Sigma-Aldrich, St. Louis, MO) containing 15% one of the following mouse serum solutions: C57BL6 normal serum, C1q-knockout (C1q-KO) serum, MBL-A/C knock out (MBL-KO) serum, MBL-KO and C1q triple KO (MBL/C1q-KO) serum or a 1:1 mixture of 15% C1q-KO serum with 15% MBL-KO serum. The C1q-KO and MBL-KO mice have been described previously (20,21). The cells were washed twice, then double stained with either the polyclonal rabbit anti-mouse C4d antibody followed by a goat anti-rabbit-FITC (Jackson ImmunoResearch Laboratories) or an anti-mouse C1q-FITC monoclonal antibody (Cedarlane Laboratories, Burlington, NC) and an anti-mouse CD3-PE (BD Biosciences) antibody. The cells were fixed in 1% formalin, then analyzed on a FACScan flow cytometer. Due to the high level of background staining of B lymphocytes for C4d, the gate was limited to CD3+ T lymphocytes.

Results

Characterization of polyclonal rabbit antibody to mouse C4d

The polyclonal rabbit antibody to mouse C4d was tested for its specificity to mouse C4d by Western blotting, tissue staining and flow cytometry. Western blotting of mouse serum aged for different intervals revealed both an approximately 200 kDa band corresponding to the molecular weight of mouse C4/C4b, and a 41 kDa band corresponding to the molecular weight of C4d (Figure 1A). In contrast, a Western blot of aged mouse serum with a commercially available rat antibody to mouse C4 (clone 16D2) yielded only a single band at 200 kDa. This demonstrates that our polyclonal antibody (left panel) binds to both C4b and C4d while the commercially available anti-C4 antibody reacts primarily with C4/C4b (right panel). C4b is the first split product of C4 that binds to cell membranes and factor I cleaves C4b to the smaller fragment C4d. The cleavage of C4b to produce C4d is a regulation step, and, unless factor I is blocked, all C4b is rapidly converted to C4d.

Figure 1.

Characterization of rabbit polyclonal antibody to mouse C4d. (A) Western blot of the rabbit polyclonal anti-mouse C4d antibody (left panel) and a commercially available rat monoclonal antibody (right panel) to mouse C4 (clone 16D2). Serum from a C57BL6 mouse was aged for 0, 1 and 4 days at 37°C to allow cleavage of C4 to C4d. A 1% solution of aged serum was run on a 4–15% Tris-HCl gradient gel under nonreducing conditions. Blots reacted with the monoclonal antibody to mouse C4 yielded a single strong 200 kDa band corresponding to the molecular weight of murine C4/C4b, whereas blots reacted with our polyclonal antibody to mouse C4d yielded both a 200 kDa band as well as a 41 kDa band corresponding to the molecular weight of C4d. (B) Flow cytometric measurement of C4d deposition on sensitized CD3+ cells using a rabbit polyclonal antibody to mouse C4d. B10.A lymph node lymphocytes were incubated with non-complement-activating IgG1 or complement-activating IgG2a mAbs specific for H-2Kk, then incubated in mouse serum to activate complement through the classical pathway. C4d deposition was measured by flow cytometry using our polyclonal antibody to mouse C4d. A strong signal for C4d was seen on the cells sensitized with IgG2a followed by serum complement (thin solid line). No C4d deposition was seen on cells sensitized with IgG1 (bold solid line) or no antibodies followed by serum complement (interrupted line), as well as cells sensitized with IgG2a followed by heat-inactivated serum (dotted line).

In addition, an in vitro flow cytometry assay was conducted to assess the ability of the antibody to recognize cell-bound mouse C4d. Mononuclear cells from B10.A mice were sensitized with complement-activating IgG2a and non-complement-activating IgG1 monoclonal antibodies. The cells were then incubated in 15% mouse serum. C4d was quantified by flow cytometry using our polyclonal antibody. A strong fluorescent signal was detected on cells sensitized with complement-activating IgG2a monoclonal antibody. In contrast, cells sensitized with non-complement-activating IgG1 monoclonal antibody showed no staining for C4d. Similarly, the negative controls, which included sensitized cells incubated in the presence of 15% heat-inactivated serum and non-sensitized cells incubated in 15% normal mouse serum did not show staining for C4d (Figure 1B). These results indicated that our newly produced antibody recognized C4d component of complement bound to the cell membrane.

To demonstrate the ability of this antibody to stain C4d in tissue we performed immunohistochemical staining on cardiac allograft tissue. Allografts from Ig-KO mice that were injected with high doses of complement-activating IgG2b antibody to MHC class I showed significant staining for C4d on vascular endothelium of arteries and capillaries (Figure 2A). In contrast, no significant staining for C4d was found in hearts of Ig-KO mice that did not receive any treatment (data not shown) or were injected with isotype control antibodies (Figure 3). Similarly, no C4d deposition was observed in B10.A hearts transplanted into C4 knockout recipients that were treated with high doses of IgG2b (Figure 2B). No staining was found in allografts from Ig-KO mice injected with a high dose of IgG2b stained with control rabbit serum (data not shown).

Figure 2.

Immunohistochemical staining of mouse cardiac allografts for C4d after passive transfer of antibodies 10 days after transplantation. (A) Allografts from mice injected with a high dose of IgG2b (100 μg) showed significant deposition of C4d on the vascular endothelium. (B) Allografts from C4 knockout mice injected with high doses of IgG2b antibodies to MHC I showed no staining. (C) Allografts from Ig-KO mice injected with high doses of IgG1 antibodies to MHC (100 μg) alone showed no staining for C4d. (D) Allografts from Ig-KO mice given a low dose of IgG2b (25 μg) alone were negative for C4d. (E) Mice injected with a low dose of IgG2b combined with a high dose of IgG1 (100 μg) had significant deposits of C4d. (F) Similarly, allografts from WT recipients were strongly positive for C4d.

Figure 3.

Extent of C4d staining in arteries and capillaries of cardiac allograft tissue. A score of 0 indicates no C4d deposition, and a score of 4 indicates strong, linear deposition of C4d. Each individual symbol represents one allograft. Arteries and capillaries were scored separately. Cardiac allografts from Ig-KO mice that had been administered a combination of IgG1 with IgG2b had high C4d deposition scores which were comparable to grafts from WT recipients. Cardiac allografts from recipients treated with a low dose of IgG2b alone, or a high dose of IgG1 alone showed very little or no C4d deposition. Similarly, all isotype controls for each group had no C4d deposits.

Synergistic deposition of C4d by complement-activating and non-activating alloantibodies in vivo

Immunohistochemical stains of cardiac allografts were conducted to assess whether complement-activating and non-activating antibodies synergize to activate complement in vivo. In passive transfer experiments, we have found that complement-activating and non-activating alloantibodies synergize to accelerate graft rejection (9). Ig-KO mice that were given a low dose (25 μg) of a complement-activating IgG2b or high dose (100 μg) of a non-complement-activating IgG1 monoclonal antibody 10 days after transplantation had over 90% graft survival 2 days after injection of alloantibodies. However, irreversible and uniform rejection of cardiac allografts within 2 days after injection of alloantibodies was observed in all Ig-KO mice that were treated with a low dose of IgG2b in combination with a high dose of IgG1 alloantibodies.

To test whether antibody-mediated cardiac allograft rejection in vivo correlates with the extent of C4d deposition in grafts, we stained cardiac allograft tissues for C4d. Ig-KO mice injected with high doses (100 μg) of IgG1 antibody to MHC class I alone (Figure 2C) showed no staining for C4d. Similarly, cardiac allografts from Ig-KO mice given a low dose (25 μg) of IgG2b antibody to MHC class I alone were also negative for C4d (Figure 2D). In contrast, cardiac allografts from Ig-KO mice treated with a high dose of IgG2b (100 μg) showed significant deposition of C4d on the vascular endothelium in arteries and capillaries (Figure 2A). Likewise, mice injected with a low dose of IgG2b (25 μg) combined with a high dose of IgG1 had significant deposition of C4d on the vascular endothelium of the graft (Figure 2E). The pattern and extent of C4d staining in both groups were similar to the staining in WT allograft recipients (Figure 2F) that demonstrated high levels of IgG1 as well as IgG2a and IgG3 in the circulation (10).

The degree of C4d deposition on grafts from all experimental groups were assessed on a scale of 0–4 by an experienced pathologist, with a score of ‘0’ indicating no to minimal C4d staining, and ‘4’ indicating a strong, linear staining for C4d on the vascular endothelium (Figure 3). Overall, C4d deposition in grafts from mice given a low dose of IgG2b combined with a high dose of IgG1 (n = 5) were comparable to that seen in WT graft recipients (n = 9). Little or no staining for C4d was found in any other group of Ig-KO recipients, including mice receiving isotype control antibodies (n = 12), low dose of IgG2b alone (n = 5) or high dose of IgG1 alone (n = 5).

Synergistic deposition of C4d by complement-activating and non-activating alloantibodiesin vitro

In order to validate our findings that complement and non-complement-activating antibodies synergize to deposit C4d in vivo, we performed an additional in vitro assay to demonstrate the same synergistic activation of complement, in the absence of other potentially confounding variables. We modified a flow cytometry assay to measure C4d bound to cells in the presence of alloantibodies. To this end, various dilutions of complement-activating and non-activating alloantibodies, monoclonal antibodies specific for MHC class I, were used to sensitize B10.A lymph node lymphocytes. The cells were washed, incubated in 15% normal mouse serum solution and stained for C4d.

Initially, we conducted an experiment to develop a dose-dependent titration curve. Based on this curve, we determined the optimal concentration of IgG2a and IgG2b for future experiments. The optimal concentration was considered as the dose of complement-activating alloantibody that would cause a minimal detectable C4d deposition. A concentration of 0.4 μg/mL and 0.015 μg/mL (low dose) was chosen for IgG2a and IgG2b, respectively (Figure 4A).

Figure 4.

C4d deposition on CD3+ cells measured by flow cytometry. (A) Titration curve of C4d deposition as a function of IgG concentration. Increasing amounts of a complement-activating antibodies to MHC class I of either subtype IgG2a or IgG2b, or a non-complement-activating IgG1 were added to aliquots of B10. A mouse lymph node mononuclear cells, which were then incubated in 15% mouse serum. C4d deposition was measured by flow cytometry on CD3+ gated cells. IgG1 caused no detectable C4d deposition at doses from 0 to 50 μg/mL. In contrast, IgG2a caused a dose-dependent increase in C4d deposition at doses from 0.75 to 10 μg/mL and IgG2b caused an even greater dose-dependent increase in C4d deposition at doses from 0.05 to 0.4 μg/mL. (B) and (C) Demonstration of synergistic activation of complement by IgG1 and IgG2a or IgG2b. Increasing amounts of non-complement-activating IgG1 were added in combination with a low dose of complement-activating IgG2a (0.4 μg/mL) (Figure 4B) or IgG2b (0.015 μg/mL) (Figure 4C). C4d deposition increased significantly in a dose-dependent manner with increasing concentrations of IgG1. In contrast, an isotype control IgG1 antibody did not alter the low level of C4d deposition caused by the low dose of IgG2a or IgG2b alone.

We found that low doses of either IgG2a or IgG2b that were insufficient to cause significant C4d deposition on their own, synergized with increasing concentrations of the non-complement-activating IgG1 antibody. IgG1 caused a dose-dependent increase of C4d deposition on lymphocytes sensitized with low doses of IgG2a or IgG2b (Figure 4B,C, respectively). Adding increasing amounts of an IgG1 isotype control to IgG2a or IgG2b had no effect on C4d deposition, indicating that binding of antibodies to the cell is necessary for the synergistic activation of complement.

Analysis of epitopes recognized by complement-activating and non-activating monoclonal alloantibodies

In order to determine whether the three monoclonal antibodies used in our experiments recognized shared epitopes, we conducted blocking studies using a flow cytometry assay. IgG1 antibody was unable to block binding of the IgG2a or IgG2b antibody to the B10.A cells, indicating that IgG1 does not recognize shared epitopes with either IgG2a or IgG2b. In contrast, IgG2a and IgG2b blocked each other very strongly (data not shown). It has been previously shown that the binding of an antibody to MHC molecules could significantly increase the binding affinity of another antibody to the same MHC molecule (22). However, in our experiments the initial cross-linking of MHC molecules by IgG1 antibodies did not result either in significant increase of binding IgG2a or IgG2b that were added after IgG1 or increase of C1q binding (data not shown).

The role of MBL in the synergistic deposition of C4d by complement and non-complement-activating alloantibodies

IgM (23,24) and IgG (25) have been reported to bind MBL. Therefore, we tested whether MBL plays a role in the synergistic activation of complement by the combination of complement-activating and non-activating monoclonal antibodies using sera from C1q-KO, MBL-KO or C1q/MBL-KO mice.

The absence of C1q in serum completely eliminated the ability of high doses of both IgG2a and IgG2b alone to deposit C4d on the cell surface, as well as the ability of IgG1 to synergize with low doses of IgG2a and IgG2b to deposit C4d (Figures 5A,B). The absence of MBL decreased significantly C4d deposition by high doses of IgG2a or IgG2b, and also by combinations of IgG1 with low doses of IgG2a and IgG2b. Reconstitution of the MBL-deficient serum with C1q-deficient serum restored both the ability of a high dose of IgG2a or IgG2b alone, as well as both IgG1/IgG2a and IgG1/IgG2b combinations, to deposit C4d. To confirm the role of MBL in the synergistic deposition of C4d by IgG1, we measured C4d deposition over an extended titration curve of IgG1 in combination with low doses of IgG2a (Figure 5C) and IgG2b (Figure 5D). This experiment demonstrated that the reconstitution of MBL-deficient serum with C1q-deficient serum restored the ability of IgG1 to synergize with IgG2a or IgG2b at each concentration of IgG1 tested. These data indicate that both C1q and MBL participate in complement activation by single subclass of antibodies as well as in combinations of complement-activating and non-activating antibodies.

Figure 5.

Synergistic C4d deposition on CD3+ cells in MBL- or C1q-deficient sera measured by flow cytometry. Combinations of complement-activating antibodies to MHC class I of either subtype IgG2a (A) or IgG2b (B), and a non-complement-activating IgG1 were added to aliquots of B10. A mouse lymph node mononuclear cells, which were then incubated in 15% of one of four sera: normal mouse, C1q-KO, MBL-KO, C1q/MBL-KO or a 1:1 mixture of MBL-KO with C1q-KO. C4d deposition was measured by flow cytometry on CD3+ gated cells. In the MBL-KO serum group, C4d deposition by high dose IgG2a (6 μg/mL) alone as well as high dose IgG2b (7.5 μg/mL) alone was significantly lowered, and synergistic deposition of C4d by a combination of IgG1 (6 μg/mL) with low dose of either IgG2a (0.2 μg/mL) or IgG2b (0.03 μg/mL) was minimal compared to normal mouse serum. No complement activation was seen in any of the C1q-KO or the MBL/C1q-KO serum groups. A 1:1 mixture of the MBL-KO serum with the C1q-KO serum reconstituted the ability of high doses of IgG2a or IgG2b, as well as the combinations of IgG1/IgG2a and IgG1/IgG2b to deposit C4d. Deposition of C4d over an extended titration curve of IgG1 in combination with low doses of IgG2a (0.2 μg/mL) (C) or IgG2b (0.03 μg/mL) (D) was measured using serum from MBL-KO or C1q-KO mice. Addition of increasing amounts of IgG1 to either low doses of IgG2a or IgG2b resulted in minimal change of C4d deposition with serum from MBL-KO or C1q-KO mice. However, IgG1 caused a dramatic IgG1-dose-dependent C4d deposition when MBL-KO serum was reconstituted with C1q-KO serum. Each data point for one experimental condition represents an average from three replicates ±SD.

Discussion

Clinical studies have documented that the deposition of C4d in tissue biopsies correlates with antibody-mediated rejection of cardiac and renal allografts (26–31). C4d has now become a standard marker for antibody-mediated rejection in renal and cardiac transplantation (32,33). In order to make our results more clinically relevant, we developed a polyclonal rabbit antibody specific to a peptide of mouse C4d. This peptide is homologous to the peptide used to prepare the polyclonal antibody to human C4d that is utilized in clinical studies (28,29,33).

In this study, we have shown significant C4d staining in mouse cardiac allograft recipients injected with a high dose of complement-activating alloantibodies. These findings are consistent with our previous study in rat cardiac allografts that C4d staining correlates with the presence of circulating alloantibodies (8). Moreover, the strong, linear and diffuse deposits of C4d along the vascular endothelium in mouse hearts are the same as in rat and human cardiac allografts (28,31).

In addition, we have documented that non-complement-activating antibodies can synergize with complement-activating antibodies to activate complement, as measured by C4d deposition in vivo and in vitro. These findings indicate a mechanism by which some non-complement-activating alloantibodies contribute to humoral allograft rejection.

Titers of non-complement-activating alloantibodies in transplant recipients can be significant in relation to complement-activating alloantibodies (14,15). Moreover, their presence can increase the risk for development of early acute rejections after kidney transplantation (13). Results from animal models have also shown that non-complement-activating alloantibodies can be present in higher titers than complement-activating alloantibodies in response to presensitization by blood transfusion (7), as well as in acute rejection of H-2 incompatible mouse cardiac allografts (9,10). However, whether non-complement-activating alloantibodies contribute to humoral rejection is still unclear.

In vitro models have demonstrated mechanisms by which antibodies can augment or inhibit tissue injury (9, 10, 34, 35). Antibodies have been shown to activate NK cells, neutrophils and macrophages through their Fc receptors (36–38). Recent findings by Reed et al. suggest that cross-linking of HLA class I molecules by antibodies to HLA may contribute to chronic graft rejection by promoting endothelial transduction of intracellular activating signals and cell proliferation (39,40). Narayanan et al. have shown that the exposure of human aortic endothelial cells to saturating concentrations of MHC class I antibodies and complement can induce caspase 3-dependent apoptotic cell death, while sub-saturating concentrations of the same antibodies conferred resistance to antibody- and complement-mediated lysis (35).

Clinically, prospective transplant recipients are screened for potentially harmful antibodies to HLA. Antibodies detected by flow cytometry alone are considered as ‘permissible’, meaning that a transplant can be performed with a high risk of rejection (41). In contrast, antibodies detected by complement-dependent cytotoxicity (CDC) test are ‘non-permissible’, indicating an unacceptably high risk of graft rejection (42).

Wahrmann et al. have recently developed a technique that allows detection of antibody-mediated activation of complement by measuring deposition of C4d on the surface of microparticles coated with HLA antigens (43). Their data indicate that the presence of complement-activating antibodies to HLA class I antigens is associated with an inferior graft survival. Kushihata et al. (15) have demonstrated that mixtures of complement-activating antibodies can cause synergistic deposition of C3b on cells as measured by flow cytometry. Our results suggest that the detection of C4d on cells or beads by flow cytometry can reflect an interaction of complement-activating and non-activating antibodies to MHC class I.

Our investigation into the mechanism of synergistic complement activation yielded novel findings. The results from experiments utilizing the C1q- and MBL-deficient sera suggest that while C1q is the component absolutely required for synergistic effect of complement activation by a combination of complement-activating and non-activating antibodies, MBL significantly augments this synergistic effect.

Although the exact mechanism by which MBL binds to a certain combination of complement-activating and non-activating antibodies is not known, it is possible that the glycosylation patterns of the antibodies used in our studies may be partly responsible for the observed synergistic effects. Malhotra et al. (25) previously reported that certain circulating IgG molecules in patients with rheumatoid arthritis can activate complement through the MBL pathway. The mechanism of this activation has been attributed to IgG glycoforms known as G0 glycoforms. G0 glycoforms lack galactose in the Fc region of IgG and terminate in N-acetyl glucosamine (GlcNAc), which allows binding of MBL.

McMullen et al. (23) have demonstrated that MBL can bind human IgM using surface plasmon resonance, and that IgM cause lysis of sensitized red blood cells more effectively in MBL-sufficient sera than MBL-deficient sera. It is important to note that not all IgM antibodies have the ability to activate MBL. It might be attributed to different efficiency to activate complement by hexameric versus pentameric forms of IgM (44,45).

To date, studies examining the role of MBL in human allograft rejection have been limited and conflicting. In kidney allograft biopsies, the presence of C4d was associated with diffuse H-ficolin and IgM deposition in the peritubular capillaries (46) or was co-localized with MBL-associated serine protease 1 (MASP-1) (47). Berger et al. found a correlation between higher serum MBL levels and more severe rejections of renal transplants (48). Interestingly, MBL-deficient cardiac allograft recipients have been reported to have increased acute rejection episodes and allograft vasculopathy (49). These results clearly indicate that further studies to examine the role of MBL in allograft rejection are warranted.

In summary, we have provided data supporting the hypothesis that non-complement-activating alloantibodies contribute significantly to the pathology of antibody-mediated cardiac allograft rejection through their ability to synergize with low titers of complement-activating alloantibodies to promote activation of complement. We have demonstrated that both C1q and MBL strongly contribute to this synergy in vitro.

Acknowledgments

We thank Dr. Marina Botto for supplying the original colony of C1q-KO mice used for this study. This work was supported by NIH grants for BAW RO1-HL63948, for WMB RO1-AI42387 and PO1-HL56091, for GLS RO1-HL56068, RO1-HL52886, P50-DE016191 and RO1-DE017821 and ROTRF grant ID#508303540 for BAW.

Ancillary