Polyreactive Antibodies Developing Amidst Humoral Rejection of Human Kidney Grafts Bind Apoptotic Cells and Activate Complement



Antibody mediated rejection (AMR) is associated with a variety of graft-reactive antibodies following kidney transplant. To characterize these antibodies, we immortalized 107 B cell clones from a patient with AMR. In a previous study, we showed that six clones were reacting to multiple self-antigens as well as to HLA and MICA for two of them, thus displaying a pattern of polyreactivity. We show here that all six polyreactive clones also reacted to apoptotic but not viable cells. More generally we observed a nearly perfect overlap between polyreactivity and reactivity to apoptotic cells. Functionally, polyreactive antibodies can activate complement, resulting in the deposition of C3d and C4d at the surface of target cells. Testing the serum of 88 kidney transplant recipients revealed a significantly higher IgG reactivity to apoptotic cells in AMR patients than in patients with stable graft function. Moreover, total IgG purified from AMR patients had increased complement activating properties compared to IgG from non-AMR patients. Overall, our studies show the development of polyreactive antibodies cross-reactive to apoptotic cells during AMR. Further studies are now warranted to determine their contribution to the detection of C4d in graft biopsies as well as their role in the pathophysiology of AMR.


antibody mediated rejection


donor-specific antibodies


human embryonic kidney


monoclonal antibody


nondonor-specific antibodies.


Antibody mediated rejection (AMR) is one of the leading causes of graft loss following kidney transplantation. The pathogenesis of AMR is believed to include both immunologic and nonimmunologic factors (for review: [1]). A central feature of AMR is the presence of circulating graft reactive antibodies that were either present before transplantation (presensitization) or developed de novo posttransplant as a component of the alloresponse. These antibodies have been described as donor-specific antibodies (DSA) when they recognize allelic HLA molecules expressed by the donor, nondonor-specific antibodies (NDSA) when they recognize allelic HLA nonexpressed by the donor, and autoantibodies when they recognize self-antigens [2-9]. We recently reported the development of another kind of antibodies during AMR that display a broad range of reactivity [10]. These polyreactive antibodies can bind several unrelated antigens such as DNA, insulin, LPS as well as multiple HLA molecules for some of them. Using a molecular experimental approach, we found that one such monoclonal polyreactive antibody was produced by a highly expanded memory B cell clone that represented ∼0.2% of all circulating B cells in a patient with AMR. Remarkably, the patient's serum reacted to the same HLA alleles as the ones recognized by this high frequency clone, suggesting that the polyreactive mAb it produced had contributed to the overall serum reactivity.

In mice, polyreactive antibodies are produced by B-1 B cells, a subset of “innate” B cells residing primarily in the peritoneal cavity and nonlymphoid tissues. B-1 B cells are also known to secrete “natural antibodies,” reacting to cells undergoing programmed cell death and involved in their removal. Despite the apparent connection, the identity between polyreactive antibodies and antibodies reactive to apoptotic cells has never been formally demonstrated. Here, we examined the capacity of polyreactive antibodies developing in patients with AMR to react to apoptotic cells, reflecting an ongoing B-1 B cell response.

Another important question relates to the clinical relevance of polyreactive antibodies and their role in AMR alongside more specific antibodies. It is now clear that DSA mediate graft tissue destruction through the activation the complement system, as revealed by the deposition of the molecule C4d on graft endothelium. In our studies we investigated whether polyreactive antibodies can also activate complement and contribute to the pathophysiology of AMR.

Materials and Methods

Patient characteristics and biological samples

All specimen collection and use in our study was approved by the internal review board of the Massachusetts General Hospital (MGH) or the University Hospital of Parma, Italy. The patient from whom B cell clones were generated has previously been described in detail [10]. Briefly, the patient is a 43-year-old male, who underwent explantation of his second kidney transplant consecutive to rejection. His original disease was focal segmental glomerulosclerosis (FSGS). Low level anti-Class I and Class II antibodies were detected in the patient's serum at the time of transplant nephrectomy. Examination of the explanted graft specimen showed evidence of both chronic and acute rejection with a humoral component. Serial plasma samples used in subsequent experiments were collected from kidney transplant recipients as part of their standard clinical care. All 38 patients with AMR (MGH, N = 20; Parma, N = 18) had a graft biopsy documenting humoral rejection at the time of sample collection. A summary of patient characteristics is provided in Table 1. Plasma samples collected from five healthy donors were also used in the study. Individual causes of end-stage renal failure is reported for all patients in supplementary Table 1. The detection of anti-HLA antibodies in all patients is reported in Supplementary Table S2.

Table 1. Summary of patient group characteristics
 MGHUniversity Hospital of Parma
  • 1p-Values indicate statistical significance between stable and AMR groups.
  • 2The age difference between the control and the stable groups is statistically significant (p = 0.035).
N52020 53018 
Age median (range)38 (33–40)45 (28–82)36 (16–66)0.0132 (28–52)247 (23–75)248 (19–69)0.31
Female, n (%)3 (60%)5 (25%)5 (25%)11 (20%)10 (33%)8 (44%)0.54
DSA, n (%) 0 (0%)20 (100%)  0 (0%)17 (94%) 
Class IN/A0 (0%)12 (60%)<0.001N/A0 (0%)10 (56%)<0.001
Class II 0 (0%)17 (85%)  0 (0%)9 (50%) 
Donation, n (%)
DeceasedN/A13 (65%)8 (40%) N/A29 (97%)15 (83%) 
Living related 5 (25%)5 (25%)0.16 1 (3%)2 (11%)0.23
Living unrelated 2 (10%)7 (35%)  0 (0%)1 (6%) 
Median sampling time (years post-tx)N/A2.65.90.002N/A4.78.10.022
Induction therapy, n (%) 10 (50%)10 (50%)1 26 (87%)13 (72%)0.26
ATGN/A10 (100%)7 (70%)0.21N/A6 (20%)4 (31%)0.7
Other 0 (0%)3 (30%)  20 (67%)9 (69%) 
Immunosuppression regimen, n (%)
Four drug regimen 0 (0%)2 (10%)  0 (0%)0 (0%) 
Three drug regimenN/A9 (45%)11 (55%)0.22N/A19 (63%)12 (67%)1
Two drug regimen 11 (55%)7 (35%)  11 (37%)6 (33%) 
Tacrolimus and/or sirolimus 19 (95%)14 (70%)0.09 24 (80%)10 (56%)0.1
Cy.A and/or azathioprine 1 (5%)10 (50%)0.003 7 (23%)9 (50%)0.11

Isolation and immortalization of B cell clones

The procedure for isolation and immortalization of B cell clones has already been described [10]. In brief, all clones were generated by limiting dilution using Epstein–Barr virus as an immortalization agent. Clonality was confirmed by molecular analysis of Ig heavy chain transcripts as previously described [10].

Western blotting

Human embryonic kidney cells (HEK293) were lysed using RIPA buffer (Boston BioProducts, Worcester, MA) supplemented with protease inhibitor. Cell lysates were homogenized using QIAshredder columns (Qiagen, Valencia, CA). Soluble fractions were separated by SDS–PAGE and transferred to nitrocellulose membrane (Invitrogen, Carlsbad, CA). Membranes were saturated 1 h RT in TBST supplemented with 5% nonfat dry milk and immunoblotted overnight at 4°C with six monoclonal polyreactive antibodies (clones 3E7, 4C9, 4E8, 4F11, 4G4 and 4G10) or one nonpolyreactive control clone supernatant (clone 3D4) at 5 µg/mL. Membranes were then washed, probed with HRP-conjugated goat anti-human IgM polyclonal antibodies (Zymed Laboratories, San Francisco, CA), and revealed with enhanced chemiluminescence (ECL; Amersham Biosciences, Uppsala, Sweden). The concentration of all monoclonal antibodies (mAb), previously assessed by ELISA, was further confirmed by Western blotting. Antibodies were separated by electrophoresis, transferred to a nitrocellulose membrane, and revealed with an anti-IgM antibody (Invitrogen).

Reactivity to apoptotic, necrotic and pyroptotic cells

Human jurkat T cell leukemia cells were first cultured overnight with 200 ng/mL of anti-FAS antibody (clone CH11, Millipore, Billerica, MA) to induce apoptosis. A mixture (1:1) of viable and apoptotic jurkat cells were incubated for 30 min at 37°C with either B cell clones' supernatants or the patient's plasma diluted at 1:5. Antibody binding was revealed using a FITC-conjugated anti-IgA/G/M secondary antibody (Invitrogen) for the clone supernatants or a FITC conjugated anti-IgG secondary antibody (Invitrogen) for the patients' plasma. Cells were analyzed using an Accuri C6 flow cytometer (BD Biosciences, San Jose, CA).

In experiments assessing the binding of polyreactive antibodies to various subpopulations of viable and apoptotic cells, jurkat cells where treated with anti-FAS as described above whereas HEK 293 cells were exposed to UV light (240 × 10−3 J) to induce apoptosis using a UV stratalinker 2400 (Stratagene, Santa Clara, CA). Both cell lines where stained 12 h after induction of apoptosis. After staining with polyreactive mAb or a non polyreactive control mAb, as described above, cells were stained with APC-conjugated Annexin-V and propidium iodide kit (eBiosciences, San Diego, CA) according to the manufacturer's instruction. Cells were then analyzed using a FACSVerse or an Accuri C6 flow cytometer (BD Biosciences). The same cells were also analyzed using an Olympus FV1000 confocal microscope.

The monocytic cell line THP-1 was used to assess the reactivity of polyreactive antibodies to pyroptotic cells. THP-1 cells were treated for 1 hour with 10 ng/mL LPS (Sigma–Aldrich, St. Louis, MO) and subsequently incubated overnight with 200 ng/mL of recombinant flagellin from Salmonella typhimurium (Invivogen, San Diego, CA). Pyroptotic and nonpyroptotic (nontreated) cells were then stained with the monoclonal polyreactive antibodies as described above.

ELISA assays

ELISA assays for the detection of antibodies to double stranded DNA (dsDNA), whole protein extract from human embryonic kidney cell line (HEK-293) and insulin were performed as previously described [7]. Antibody binding was revealed with an HRP-conjugated goat anti-human IgG/M/A (Invitrogen) and developed using 3,3′,5,5′-tetramethylbenzidine (TMB; Sigma). Optical density was read at 450 nm.

ELISA to phosphatidylserine (PS) and lysophosphatidylcholine (LPC) were adapted from Pierangeli and Harris [11]. Ninety-six flat bottom culture plates (BD Biosciences) were coated with either PS (Avanti Polar Lipids, Inc., Alabaster, AL) or LPC (Sigma) at 50 µg/mL in ethanol and incubated uncovered overnight at 4°C to allow ethanol evaporation. After three washes, plates were blocked in PBS supplemented with 10% fetal calf serum for 1 h RT and used as described above.

Serum IgG purification

Plasma IgG were purified from patients specimens using the Melon Gel IgG Purification Kit (Thermo Scientific, Rockford, IL) according to the manufacturer's instructions.

Complement opsonization and C4d binding assay

Apoptotic jurkat cells (0.5 × 106 cells) were incubated for 20 min at 37°C with monoclonal polyreactive antibody at 1 µg/mL or purified serum IgG diluted 1:2. Human serum from a healthy donor diluted 1:5 in HBSS was then added as a source of complement and incubated for 15 min at 37°C. After two washes in PBS, cells were incubated for 30 min at 4°C with an anti-C4d antibody (Quidel, San Diego, CA), washed twice again in PBS and incubated for 30 min at 4°C with a FITC-conjugated anti-mouse IgG secondary antibody (BD Biosciences). After two final washes at 4°C, C4d binding was measured on a FACSVerse flow cytometer (BD Biosciences).

C4d immunofluorescence procedure

Apoptotic jurkat cells (1 × 105) opsonized by complement molecules as described above were coated on a glass slide by centrifugation in a Shandon Cytospin3 for 20 min at 200 rpm. Slides were then air dried for 20 min, blocked with Avidin D (100 µg/mL, Vector, Burlingame, CA) for 20 min followed by three PBS washes. D-Biotin (10 µg/mL, Sigma) was added for 20 min followed by three PBS washes. B cell clones' supernatants containing either a nonpolyreactive antibody (3D4) or polyreactive antibodies (3E7, 4G10) were added and incubated for 30 min. After three washes with PBS, a mouse monoclonal anti-C4d antibody (1:100, Quidel) or anti-C3d antibody (1:100, Quidel) was added and incubated for 30 min followed by three PBS washes. Biotinylated horse-anti-mouse IgG (H&L, 1:100, Vector) was added and incubated for 30 min followed by three PBS washes. FITC-streptavidin (1:50, Vector) was then added and incubated for 30 min followed by three PBS washes. Lastly, slides were covered slipped with Aquamount and visualized using an Olympus BX60 microscope.

C4d deposition and activated caspase 3 staining on human kidney biopsies

Consecutive frozen sections for each staining were air dried for 30 min and rinsed in PBS. Slides were then stained for C4d as described above, or for activated caspase 3 (BD Biosciences, 1/50) for 1 h and revealed with a goat anti rabbit IgG secondary antibody (1/200, Vector) subsequently detected with Cy3 streptavidin (1/5000, Jackson Immunoresearch, West Grove, PA).

Statistical analysis

A Student's unpaired t-test was used to compare the reactivity of stable and AMR serum samples as well as C4d binding to viable and apoptotic cells (Figures 5 and 6). Comparisons of patient's characteristics between the AMR and stable groups (Table 1) were based on Fisher's exact test. Correlation between polyreactivity and reactivity to apoptotic cells (Figure S2) was determined using a statistical analysis based on a two-tailed nonparametric Spearman's test.


Polyreactive antibodies bind to apoptotic cells

As described elsewhere [10], we generated a series of immortalized B cell clones from an explanted graft specimen obtained from a patient who underwent transplant nephrectomy for acute and chronic antibody mediated rejection. One hundred seven clones were established using this approach. We previously reported that six of them reacted to multiple unrelated antigens [10], hence displaying a profile characteristic of polyreactivity [12]. We confirmed here the polyreactive profile by probing protein lysate prepared from the human embryonic kidney cell line HEK293 with the monoclonal antibodies (mAb) secreted by the six clones. As reported in Figure 1, polyreactive mAb reacted to a wide array of proteins whereas a control nonpolyreactive mAb (3D4) did not.

Figure 1.

Polyreactivity of supernatant from selected B cell clones. A protein lysate prepared from the human kidney cell line HEK 293 was probed by Western blotting with supernatant from six polyreactive clones and one nonpolyreactive clone (3D4). Control lane was probed with the secondary antibody alone. The picture on the bottom shows the signal generated by the probing of the supernatants with an anti-IgM antibody, as a control of the IgM content in each supernatant.

“Natural” preformed antibodies reactive to apoptotic determinants have been described in the blood of healthy human donors (for review: [12-15]). We examined whether polyreactive monoclonal antibodies produced by immortalized B cell clones could also bind to apoptotic cells. A mixture of apoptotic and viable jurkat cells was stained with mAb secreted by polyreactive immortalized B cell clones and assessed by flow cytometry. As shown in Figure 2A, polyreactive mAb reacted to apoptotic whereas a control nonpolyreactive mAb (3D4) did not. Remarkably, viable cells were not stained by the polyreactive mAb, indicating that the reactivity was restricted to apoptotic cells. The ability of polyreactive antibodies to bind cells undergoing microbe-mediated inflammatory cell death (pyroptosis) was also assessed using flow cytometry. Staining patterns were comparable to those observed with apoptotic cells (Figure 2B).

Figure 2.

Reactivity to apoptotic and pyroptotic cells. (A) Cell culture supernatants from three selected polyreactive clones as well as a control nonpolyreactive clone 3D4 were used to stain a mixture of viable and apoptotic jurkat cells. Results are reported after gating on viable cells (upper panels) or apoptotic cells (lower panels). Filled gray histograms show the signal generated by staining with the secondary antibody alone. (B) Reactivity to pyroptotic cells. Cell culture supernatants from three selected polyreactive clones as well as from the nonpolyreactive clone 3D4 were used to stain either viable (upper panel) or pyroptotic (lower panel) monocytic THP-1 cells. Filled gray histograms show the signal generated by staining with the secondary antibody alone. (C) Binding of polyreactive antibodies on viable, early apoptotic, late apoptotic or dead cells. A mix of viable and FAS-treated jurkat cells were stained with polyreactive mAb, Annexin V and propidium iodide (PI; upper panel). Binding of each clone on viable cells (green gate), early apoptotic (orange gate), late apoptotic (red gate) and dead cells (black gate) is depicted as histograms. The same experiment was performed on a mix of viable and UV-treated HEK-293 cells (lower panel). (D) HEK293 cells stained for flow cytometry experiment were visualized by confocal microscopy. Signal for Annexin V (red), PI (blue) and polyreactive clone (green) is shown for each polyreactive antibody. All pictures were acquired using the same settings for consistency purposes.

We next used FITC conjugated Annexin V and propidium iodide (PI) to determine at what stage of apoptosis polyreactive mAb reacted to jurtkat cells. Using these markers, we were able to discriminate four populations of jurkat cells: viable (Annexin V negative, PI negative), early apoptotic (Annexin V positive, PI negative), late apoptotic (Annexin V positive, PI positive) and dead cells (Annexin V negative, PI positive). All polyreactive mAb reacted exclusively to late apoptotic jurkat cells (Figure 2C, upper panel). Similar results were found for HEK293 cells even though the Annexin V/PI staining could discriminate only viable and late apoptotic populations for these cells (Figure 2C, lower panel). As revealed by confocal microscopy, polyreactive mAb staining on HEK 293 cells did not co-localize with Annexin V or PI staining (Figure 2D), ruling out the hypothesis that these antibodies were solely reacting to exposed DNA in late apoptotic cells. In an attempt to determine whether previously described antigenic structures exposed at the surface of apoptotic cells would be recognize by polyreactive antibodies, we measured the binding of these antibodies to phosphatidylserine (PS) and lysophosphatidylcholine (LPC) by ELISA. As depicted on Figure S3, only clone 3E7 and 4G10 were reactive to these two antigens.

To examine the association between polyreactivity and capacity to bind apoptotic cells, we assessed the reactivity of all antibodies secreted by the 107 immortalized B cell clones to apoptotic jurkat cells (Figure S1A). The reactivity is reported in Figure 3 together with the cumulative reactivity to dsDNA, insulin and a lysate of HEK293 cell line as a surrogate marker of polyreactivity. As shown in this figure, a strong association was observed between reactivity to apoptotic cells and polyreactivity, supporting the view that polyreactive antibodies and natural antibodies binding to apoptotic cells are identical. This association between polyreactivity and reactivity to apoptotic cells is statistically significant (P = 0.0053, Figure S2).

Figure 3.

Correspondence between polyreactivity and reactivity to apoptotic cells. Cumulative reactivity index (OD ratio between values obtained with the clone of interest and a control clone after subtraction of background values obtained with the secondary antibody) to dsDNA, insulin and lysate of HEK293 cells (right x-axis) for all 107 B cell clones are compared with reactivity index to apoptotic cells for the same clones (left x-axis).

Polyreactive antibodies activate complement, leading to C4d deposition on target cells

Detection of the complement molecule C4d deposited on graft tissue biopsies, especially on peri-tubular capillaries, is a major tool for the diagnosis of AMR (for review: [16]). It is also a sign of activation of the complement in situ presumably responsible for some levels of tissue damage during rejection. We thus assessed the capacity of polyreactive mAb generated in our lab to also activate complement and trigger the deposition of C4d on target cells. As shown in Figure 4A, polyreactive mAb effectively induced deposition of C4d on the surface of apoptotic jurkat cells whereas a nonpolyreactive control mAb did not (black bars). Deposition of C4d on apoptotic cells was closely associated with the capacity of the mAb to bind these cells (Figure 4A, gray bars). As illustrated in Figure 4B, the capacity to activate complement was dependent on the mAb concentration. The deposition of C4d and C3d on the surface of apoptotic cells in the presence of polyreactive antibodies was further confirmed by immunofluorescence assays (Figure 4C). In this experiment, complement activation was abrogated by prior heat inactivation of the serum used as a source of complement.

Figure 4.

Capacity of polyreactive antibody to activate complement. (A) Association between antibody binding and C4d deposition on apoptotic cells. Median fluorescence of antibody binding to apoptotic jurkat cells (left x-axis) is compared with the median fluorescence of C4d binding on jurkat cells after antibody binding (right x-axis). (B) Three polyreactive antibodies (3E7, 4F11 and 4G10) and one nonpolyreactive antibody (3D4) were incubated at various concentrations with apoptotic jurkat cells in the presence of a source of complement. C4d binding was revealed by flow cytometry. (C) C3d and C4d staining of apoptotic jurkat cells after binding of nonpolyreactive antibody 3D4 or polyreactive antibodies 3E7 and 4G10. For each condition, a control using heat-inactivated serum as a source of complement has been added. For consistency purposes, all pictures were acquired using the same settings.

Enhanced IgG reactivity to apoptotic cells in the serum of patients with AMR

We took advantage of the fact that polyreactive antibodies also bind apoptotic cells to detect them in the serum of kidney transplant recipients in relation with AMR. We first measured the reactivity to viable and apoptotic jurkat cells in serum samples collected at MGH from 5 healthy donors, 20 control kidney transplant recipients with stable graft function and 20 patients with AMR. Virtually no IgG binding was observed on viable cells using either serum from healthy donors, stable, or AMR patient serum (Figure 5, left panel). In contrast, the reactivity of serum IgG from AMR patients to apoptotic jurkat cells was significantly higher than that of stable patients (Figure 5, right panel). Remarkably, patients whose end-stage renal failure was caused by autoimmune diseases did not appear to have higher serum reactivity to apoptotic cells (Figure S4). Although serum samples from AMR patients were collected at a later time point posttransplant than from non-AMR patients (Table 1), no correlation was observed between serum reactivity to apoptotic cells and sampling time (Figure S5). Polyreactive IgG also appeared to develop at time of AMR as revealed by lower reactivity to apoptotic cells detected in serum samples collected before AMR (Figure S6).

Figure 5.

Serum reactivity to apoptotic cells. The reactivity of serum samples collected at Massachusetts General Hospital from healthy controls (N = 5), stable patients (N = 20) and AMR patients (N = 20) to viable (left panel) or apoptotic (right panel) jurkat cells was measured by flow cytometry using a FITC-conjugated anti-IgG secondary antibody.

Serum IgG purified from AMR patients show enhance reactivity to apoptotic cells and increased complement activation capabilities

We next assessed the capacity of serum IgG from patients with AMR to activate complement. To eliminate the effect of IgM, which can also activate complement, we first purified IgG from serum samples. These experiments were carried out using specimens collected at the University Hospital of Parma. As reported in Figure 6A, the reactivity of total IgG purified from patient with AMR (N = 18) to apoptotic cells was significantly higher than the reactivity of IgG purified from healthy donors (N = 5) or kidney transplant recipients with stable condition (N = 30). In presence of complement, these purified IgG can activate the classical pathway of the complement leading to C4d deposition at the surface of targeted apoptotic cells. As depicted in Figure 6B, opsonization with IgG purified from patients with AMR led to significantly higher levels of C4d deposition compared to IgG purified from healthy donors or kidney transplant recipients with stable condition. We did not observe any correlation between complement activation capacity and C4d deposition on apoptotic cells and sampling time (Figure S5).

Figure 6.

Capacity of serum IgG to bind apoptotic cells and activate complement. (A) Purified serum IgG from 48 patients treated at the Parma University Hospital (30 stable and 18 AMR samples) as well as IgG purified from five healthy donors were tested for reactivity to apoptotic jurkat cells by flow cytometry. (B) C4d binding on the surface of apoptotic jurkat cells after incubating the purified IgG mentioned above with normal serum as a source of complement was assessed by flow cytometry.

C4d deposition can co-localize with apoptotic cells in vivo

To strengthen the in vivo relevance of our findings, we carried out immunohistochemical staining for C4d and activated caspase 3 staining on consecutive section of kidney biopsies obtained from five patients with AMR. All specimens displayed a bright C4d staining. Apoptotic cells, as detected by the presence of activated caspase 3, was scarce in two biopsies (data not shown) and abundant in three patients. C4d deposition appeared to co-localize with apoptotic cells in two of the three samples (Figure 7, cases 1 and 2) while the staining pattern was clearly distinct in the third biopsy (case 3).

Figure 7.

C4d and activated caspase 3 staining on kidney biopsies. Consecutive sections of kidney biopsies from three patients with AMR were stained for C4d (green, left panel) or activated caspase 3 (red, right panel) as a marker of apoptosis.


In a previous report we described antibodies detected amidst AMR that are neither allospecific nor autoreactive but rather polyreactive in that they react to a broad range of antigens as different as nucleic acids and HLA molecules [10]. Here we show that these polyreactive antibodies also react to apoptotic cells. The correspondence between polyreactivity and reactivity to apoptotic cells, which can only be revealed at the clonal level, had not been described before. The exact antigenic determinants recognized by polyreactive antibodies at the surface of apoptotic cells are still not well known. Work by Kim et al. [17] identified lysophosphatidylcholine (LPC), a phospholipid constituent of the cell membrane modified during programmed cell death, as the main target of natural IgM on apoptotic cells. In our study, mAb reactive to apoptotic jurkat cells were all polyreactive and therefore did not appear to be specific to a single antigenic structure. The fact that two of six polyreactive antibodies reacted to LPC as well as phosphatidylserine corroborates this point.

Although our in-depth characterization of polyreactive mAb was carried out on a single kidney transplant recipient, we could extend the findings to other patients with AMR. More specifically, polyreactive IgG could be detected at higher levels in AMR patients than in transplant recipients with stable graft function, suggesting that they develop as a component of the immune response associated with this form of rejection. Their cause and origin, however, are still obscure. Autoantibodies reacting to cryptic antigens that become accessible after cell death have previously been described. For this reason, we cannot exclude that part of the serum reactivity to apoptotic cells is due to the presence of monoreactive autoantibodies. Nevertheless, we showed a very strong association between polyreactivity and reactivity to apoptotic cells among 107 generated B cell clones. These findings strongly suggest that the majority of antibodies reacting to apoptotic cells are polyreactive.

Polyreactive antibodies are primarily IgM and are considered a normal self-reactive component of the immune system. Class-switched polyreactive IgG have also been described [18, 19], but are usually associated with pathogenic properties. In systemic lupus erythematosus (SLE) for instance, polyreactive antibodies may become pathogenic solely by class switching from IgM to IgG [20]. Similarly a number of studies showed that polyreactive IgM have a protective, anti-inflammatory, function while class switch polyreactive IgG are pro-inflammatory [21-24]. How polyreactive IgM switch to pathogenic IgG has not been thoroughly clarified but in many circumstances the level of natural antibodies appear to be driven by exposure to apoptotic cells (for review: [23]). Antibody mediated rejection is associated with an increase of apoptotic cells in various regions of the kidney graft [25-28]. This burst of apoptosis could contribute to the class switching of B cell clones reactive to apoptotic cells. If this hypothesis is correct, polyreactive antibodies would develop secondary to a primary event responsible for cell death in the graft. In line with our findings, the study by Cardinal et al. [29] shows that LG3, a bioactive C-terminal fragment of perlecan released during apoptosis, can fuel the production of anti-LG3 antibodies, which in turn work as accelerator of vascular injury.

Our previous experiments demonstrated the unexpectedly high frequency of clone 4G10 in the blood of a patient with ongoing rejection [10]. This frequency implied a considerable expansion in vivo that was consistent with the accumulation of somatic mutations found in the immunoglobulin heavy chain region as well as the expression of the CD27 memory marker by the corresponding immortalized B cell clone. This clone is also reactive to several self-antigens, class I antigens and apoptotic cells. We hypothesize that this clone is not unique but rather, is representative of a discrete subset of memory, somatically mutated polyreactive B cells, producing “natural IgM” under physiological conditions. Upon activation by apoptotic cells in the inflammatory context associated with graft rejection, these polyreactive B cells would expand, undergo class switch recombination and produce IgG. The expansion of natural antibodies and autoantibodies upon stimulation by apoptotic cells has already been described in animal models [30].

C4d deposition is a hallmark of AMR. It is well accepted that deposition of this molecule on vessels in graft biopsies results from DSA binding to HLA on the donor endothelium. We showed here that polyreactive mAb as well as polyreactive IgG purified from the serum of AMR patients have the capacity to activate complement resulting in C4d deposition on the surface of the target cells in vitro. We also observed a co-localization between C4d deposition and activated caspase 3 in three of five biopsy samples and among them in two of three biopsies which has detectable apoptotic cells. Taken together, these findings suggest that serum polyreactive antibodies can also account for the deposition of C4d in the graft tissue alongside DSA in patients with AMR. The exact contribution of polyreactive antibodies to C4d deposition relative to DSA is difficult to estimate. Remarkably, in a recent report, Haidar et al. [31] established a correlation between C4d deposition on the surface of erythrocytes and AMR in kidney transplant recipients. Since erythrocytes are of recipient origin and could therefore not be targeted by DSA, it is tempting to hypothesize that polyreactive antibodies are at play in this phenomenon. Of note, similar C4d deposition on red blood cells has also been described in patients with lupus and is used as a surrogate marker of SLE activity [32-34], bringing further argument for this hypothesis. Additional studies are now warranted to determine whether these complement-binding polyreactive antibodies are also cytotoxic as C4d binding is not always associated with graft damages [16].

Overall, we report here on the capacity of polyreactive antibodies to react to apoptotic cells and activate complement. We also describe the detection of such polyreactive IgG in the blood of kidney transplant recipients with ongoing AMR. These findings support the hypothesis than AMR is accompanied by a polyreactive B cell response in addition to the monospecific B cell response usually recognized as responsible for DSA. Further investigation will advance our knowledge of these neglected elements of B cell immunity following solid organ transplantation and their precise contribution to the pathophysiology of graft rejection.


The authors thank Dr. Bernard Collins for providing the biopsy samples and Patricia Della Pelle for superb technical assistance.

This work was supported by the Fahd and Nadia Alireza's Research Fund, the Roche Organ Transplantation Research Foundation (ROTRF) and the National Institute of Health, National Institute of Diabetes, Digestive and Kidney Diseases Grant DK083352.


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.