Main

Historically, ABO compatibility between donor and recipient has been required for successful organ transplantation. Transplantation of ABO-incompatible donor organs into recipients with preformed antibodies to donor A or B antigens usually results in 'hyperacute' or acute antibody-mediated rejection, initiated by antibody binding to graft antigens resulting in complement activation and thrombosis of graft vasculature. Transplantation of ABO-incompatible kidney and liver grafts has been accomplished with some success, but aggressive immunosuppressive strategies are required in mature individuals, often including splenectomy1,2,3,4,5,6. ABO-incompatible heart transplantation has been performed rarely, usually unintentionally, because of the high risk of lethal antibody-mediated rejection5.

In 1945, Owen defined the inherent susceptibility to immune tolerance induction during immaturity as a consequence of antigen exposure during gestation7. He reported that dizygotic twin calves exposed to allogeneic blood from shared placental circulation did not reject skin grafts from each other. Burnet then linked immune tolerance to developmental events8. Medawar and colleagues showed that tolerance could be induced intentionally ('acquired' tolerance), whereby introduction of antigens to fetal or neonatal mice resulted in permanent and specific abrogation of the development of immune responsiveness to those antigens9. Evidence emerged showing several mechanisms of neonatally induced tolerance, including functional inactivation (anergy) or deletion of immature alloreactive clones, suppression of alloreactivity by immunoregulatory cells and, in B cells, receptor editing10,11,12,13,14,15,16. Although studied extensively at both T-cell and B-cell levels in mouse models, acquired neonatal tolerance has not been previously shown in humans.

Neonatal tolerance was not generally viewed as a clinically relevant phenomenon because human immune maturation at birth is considerably more advanced than in mice, thus presumably beyond susceptibility to tolerance induction17,18. Nonetheless, the human infant manifests several aspects of immunologic immaturity, notably deficiency of humoral responsiveness to stimulation by carbohydrate ('T-independent type 2') antigens19,20,21. This includes development of isohemagglutinins or 'natural' antibodies to nonself A and B blood group antigens, which remain low during the first months of life22,23. Reasoning that humoral rejection of ABO-incompatible heart grafts would not occur in the absence of preformed donor-specific antibody, we undertook a trial of ABO-incompatible heart transplantation in infants, using standard immunosuppression without splenectomy24. There were no cases of hyperacute or acute humoral rejection, nor clinical problems attributable to blood group incompatibility. We have now studied the B-cell immunobiology of ABO-incompatible infant heart transplant recipients and, for the first time, found evidence of neonatally acquired donor-specific B-cell tolerance in humans.

Results

Ontogeny of blood group–specific antibodies

We analyzed sequential serum samples from 13 recipients of ABO-incompatible grafts and 3 recipients of ABO-compatible grafts to detect circulating A-specific and B-specific antibodies before transplantation.

Whereas serum A-specific antibodies developed normally in group O recipients of heart grafts from group B donors (B→O recipients; Table 1 and Fig. 1a) despite nonspecific immunosuppressants, we noted a persistent deficiency in circulating B-specific antibodies (Fig. 1b). Similarly, we observed a deficiency in development of A-specific antibodies in A→O recipients (patients 5–9), whereas B-specific antibodies developed normally (Fig. 1c,d). Serum A-specific and B-specific antibodies remained negligible in B→A (patients 11 and 13) and A→B (patient 12) recipients (Fig. 1e,f), whereas O→O recipients (patients 14–16) developed high titers of both A-specific and B-specific antibodies (Fig. 1g,h), similar to nonimmunosuppressed individuals. This selective deficiency in isohemagglutinin ontogeny suggests that donor-specific B-cell tolerance has occurred after ABO-incompatible transplantation, induced by exposure to donor blood group antigens during the ill-defined period of immunologic immaturity in the human infant.

Table 1 Characteristics of ABO-incompatible and ABO-compatible heart transplant recipients
Full size table
Figure 1: Development of serum A-specific and B-specific antibodies in recipients of ABO-incompatible and ABO-compatible heart transplants showing a persistent and selective deficiency in donor-specific antibody.
figure 1

Isohemagglutinin titer values indicate the inverse of the highest dilution of serum giving a positive result. Transplantation was performed at (a,b) A-specific (a) and B-specific (b) antibody development in B→O and A2B→O recipients (n = 4). c,d) A-specific (c) and B-specific (d) antibody development in A→O recipients (n = 5). (e,f) A-specific (e) and B-specific (f) antibody development in A→B, B→A and A1B→O recipients (n = 4). (g,h) A-specific (g) and B-specific (h) antibody development in O→O recipients (n = 3).

Full size image

Analysis of successful ABO-incompatible adult human kidney transplants showed the phenomenon of graft accommodation, whereby grafts survived without damage despite intragraft deposition of donor-specific antibodies and complement components such as C4d3,4. Thus, low levels of circulating antibodies may reflect deposition within the graft rather than impaired production. To confirm that deficient serum donor-specific antibodies in ABO-incompatible graft recipients was not a result of intragraft deposition, we assessed endomyocardial biopsies for immunoglobulin and complement deposition in addition to routine analysis for acute rejection. In sequential biopsies obtained during follow-up ranging from 1.5 to 5 years after transplant, immunohistochemistry staining showed no deposition of immunoglobulin (IgG or IgM) or complement (C3 or C4d, or both); (Table 1 and Fig. 2a,b; one representative example is shown of a negative biopsy; comparison is made to a case of antibody-mediated rejection resulting from HLA sensitization). We did not observe any other histologic evidence of antibody-mediated graft damage attributable to ABO incompatibility. Thus, lack of circulating and intragraft antibodies confirmed impairment of antibody production against donor A and B antigens. (Note: As previously reported, the incidence of acute cellular rejection in the original cohort was slightly less in ABO-incompatible graft recipients than in ABO-compatible graft recipients24; chronic vascular rejection manifested as accelerated graft vasculopathy developed in one ABO-incompatible recipient (Table 1).

Figure 2: Immunohistochemical staining showing absence of complement deposition and persistence of donor antigens within endomyocardial biopsy specimens from ABO-incompatible graft recipients.
figure 2

(a) Fluorescence staining for complement component C4d in a specimen from an ABO-incompatible graft. (b) Similar staining for C4d in a graft specimen from a known case of antibody-mediated rejection resulting from HLA sensitization (positive control). (c) A-antigen expression in a graft biopsy from a group A donor 2 years after transplantation into a group O recipient. Arrows indicate positive A-antigen staining. (d) Staining for B antigen in same patient as in c. (e) Staining for A antigen in a graft biopsy from a group B donor 4 years after transplantation into a group O recipient. (f) B-antigen expression in same patient as in e. Arrows indicate positive B-antigen staining.

Full size image

Patients 4 and 10 (both AB→O) demonstrated divergent patterns of A-specific antibody development. Although both patients did not produce clinically significant levels of serum B-specific antibodies (Fig. 1b,f), only patient 10 remained deficient in A-specific antibody (Fig. 1a,e). Notably, patient 10 received a group A1B graft, while patient 4's donor was group A2B. The uncommon A2 subtype is characterized by low-density cell-surface A antigen expression compared to the prevalent A1 subtype25, and group A2B cells have even fewer A antigen sites than A2 cells26. In this setting, exposure to A2B graft antigens would seem to induce tolerance only to the B antigens, whereas A1B antigen exposure was tolerogenic for both A and B antigens.

Absence of agglutination-inhibiting factors

The apparent deficiency of donor-specific isohemagglutinins in sera from ABO-incompatible graft recipients (Fig. 1) could result from interference with donor-specific antibodies by serum factors, such as idiotype-specific antibodies. To assess this possibility, we devised a modified hemagglutination assay in which serum from a group O volunteer was serially diluted using patients' sera rather than saline before incubation with group A or B erythrocytes. Titers at which agglutination of group A and B erythrocytes occurred by group O serum diluted with patients' sera were equal to titers obtained with saline dilution of the same group O serum, that is, A-specific >1:256 and B-specific >1:256. Sera from all 13 ABO-incompatible recipients showed no inhibition of erythrocyte agglutination by serially diluted serum from a single group O volunteer. Thus, even at dilutions of 1:256 (group O serum:patient serum) patients' sera were incapable of preventing erythrocyte agglutination by group O serum. This supports our conclusion that donor-specific isohemagglutination deficiency in recipients of ABO-incompatible transplants results from absence of donor-specific antibody.

Donor antigen persistence

In animal models, persistence of donor antigen has been shown to be important for induction of donor-specific tolerance27,28. Using immunoperoxidase staining, we assessed endomyocardial biopsy specimens for A and B antigen expression. We analyzed at least two sequential biopsies from seven recipients of ABO-incompatible grafts and two recipients of ABO-compatible grafts; we showed expression of A or B antigen, or both, of the heart donor in all cases that were not group O (Table 1 and Fig. 2). A antigen expression is observed in a graft biopsy from a group A donor 2 years after transplantation into a group O recipient (Fig. 2c,d); likewise, persistent B antigen expression is apparent in a biopsy from a group B donor 4 years after transplantation into a group O recipient (Fig. 2e,f).

Patient 16, a group O individual, required urgent retransplantation (unrelated to ABO status) 3 d after primary ABO-incompatible transplantation (group B donor), receiving an ABO-compatible second graft (group O donor). This patient developed A-specific and B-specific antibodies in a normal pattern (Fig. 1g,h), suggesting that only transient exposure to donor antigens was insufficient to induce donor-specific antibody deficiency.

Assessment of B-cell function

To ascertain cellular mechanisms of donor-specific antibody deficiency, we performed B-cell functional and phenotypic analyses using whole blood samples from a subset of 12 ABO-incompatible and 3 ABO-compatible graft recipients. Anergic lymphocytes have been defined as antigen–specific, nonresponsive cells that can recover responsiveness after in vitro culture with appropriate stimulation29,30. The inability of mouse neonatal B lymphocytes to respond to antigens can be overcome with cytokine stimulation31,32, and antibody production by human neonatal lymphocytes in response to CD3-specific stimulation can be elicited with interleukin (IL)-2 and IL-4 (ref. 33).

To determine whether tolerance in infant recipients of ABO-incompatible grafts resulted from functionally inactive donor-specific B cells, we cultured patient-derived peripheral blood mononuclear cells (PBMC) in vitro with nonspecific cytokine or specific donor antigen stimulation, and measured specific antibody levels in culture supernatants by ELISA. Results were consistent amongst all patients tested (Table 1 and Fig. 3). A-specific IgM production by PBMC from two A→O recipients cultured without stimulation was low, comparable to PBMC from a group A control, whereas cultured PBMC from an O→O recipient exhibited modest A-specific antibody production (Fig. 3a). When cultured with IL-2 and IL-4, A-specific IgM production was markedly increased by PBMC from the O→O recipient, but remained minimal in cultures from the A→O recipients and the group A control (Fig. 3b). We obtained similar results when we cultured cells with group A erythrocytes (Fig. 3c). To evaluate the functional capacity of patients' PBMC under in vitro culture conditions, we measured antibody production to the 'third-party' carbohydrate epitope Gal-α1-3β1-4GlcNAc-R (α-gal), a major xenoantigen against which humans produce 'natural' antibodies34. Production of α-gal-specific IgM was not substantially different amongst cytokine-stimulated culture groups (Fig. 3d). Stimulated PBMC from B→O recipients showed a similar pattern of deficient B-specific IgM antibody (Table 1). IgG isotype antibodies against donor A and B antigens were absent for all patients tested (data not shown). These data suggest that donor-specific antibody deficiency in ABO-incompatible recipients does not result from functional inactivation of antigen–specific B cells.

Figure 3: ELISA analyses of antibody in PBMC culture supernatants showing absence of A-specific antibody production by PBMC from two A→O recipients.
figure 3

(a) A-specific IgM isotype antibodies produced by PBMC cultured without stimulation. (b) A-specific IgM isotype antibodies produced by PBMC stimulated with IL-2 and IL-4 cytokines. (c) A-specific IgM isotype antibodies produced by PBMC stimulated with group A erythrocytes. (d) Antibodies against 'third-party' carbohydrate epitope α-gal produced by IL-2- and IL-4-stimulated PBMC.

Full size image

Visualization of antibody-producing cells

To detect B cells capable of producing donor-specific antibodies, we cultured patients' PBMC with IL-2 and IL-4 stimulation. We visualized A-specific and B-specific IgM-producing cells with ELISPOT assays (Table 1 and Fig. 4). We clearly observed spot formation showing B-specific but not A-specific antibody-producing cells in PBMC from an A→O recipients (Table 1 and Fig. 4a,b) whereas we visualized A-specific but not B-specific antibody-producing cells in PBMC from B→O and A2B→O recipients (Table 1 and Fig. 4c,d). Both A-specific and B-specific antibody-producing cells were observed in PBMC from O→O recipients (Table 1 and Fig. 4e,f), although neither was visualized in PBMC from A→B, B→A and A1B→O recipients (Table 1). ELISPOT assays for IgG isotype antibody-producing cells were negative for all patients tested (data not shown).

Figure 4: Visualization of antibody-producing cells by ELISPOT assay showing absence of A-specific and B-specific antibody-producing cells in cultured PBMC isolated from A→O (a,b), B→O (c,d) and O→O (e,f) heart transplant recipients.
figure 4

Spots indicate specific antibody-producing cells.

Full size image

Detection of antigen-specific B lymphocytes

These studies suggest that B-cell elimination, not anergy, is the mechanism of abrogation of B-cell responsiveness to donor A and B antigens in ABO-incompatible graft recipients. To directly ascertain the presence of donor-specific B cells, we sought B lymphocytes bearing A antigen–specific B-cell receptors (BCR) by immunostaining patients' PBMC with biotin-conjugated synthetic group A antigen and FITC-conjugated human CD19-specific monoclonal antibody (Fig. 5). FACS analysis of PBMC from an O→O recipient showed 1.8% of the total B-cell population with BCR specific for A antigen (Fig. 5a), whereas in PBMC from an A→O recipient, this B-cell population was absent (Fig. 5b), similar to the FACS profile of PBMC from a group A negative control (Fig. 5c).

Figure 5: Detection by flow cytometry of B lymphocytes bearing BCR specific to A antigen in PBMC isolated from infant heart transplant recipients.
figure 5

Each graph contains 104 gated CD19+ cells. Percentages indicate the proportion of the CD19+ population bearing A antigen–specific BCR. Panels show analyses of PBMC from (a) an O→O recipient (positive control patient) (b) an A→O recipient and (c) a group A negative control. FITC, fluorescein isothiocyanate; PE, phyloerythrin.

Full size image

Discussion

These studies of immunologic development, performed up to 8 years after ABO-incompatible infant heart transplantation, show a persistent and selective deficiency of circulating antibodies to donor A and B antigens without antibody-mediated damage. Potential explanations for antibody deficiency include: intragraft immunoglobulin deposition ('accommodation'), suppressed antibody production resulting from nonspecific systemic immunosuppression and deficient antibody production resulting from specific B-cell tolerance. Absence of immunoglobulin and complement deposition in biopsy specimens suggests that accommodation has not occurred; normal development of A-specific and B-specific antibodies in O→O infant transplant recipients indicates that immunosuppressants do not suppress isohemagglutinin development. Thus, the preponderance of evidence suggests that these children have acquired donor-specific B-cell tolerance.

Several B-cell tolerance mechanisms have been identified, including anergy (functional inactivation)29,30, deletion35,36,37 and receptor editing38,39,40. In vitro analysis of PBMC from ABO-incompatible graft recipients showed that these cells were simultaneously incapable of producing antibodies against donor-type A and B antigens, yet capable of producing antibodies against an irrelevant third-party polysaccharide antigen, and that B lymphocytes bearing BCR specific for donor-type A and B antigens were absent. These data suggest that anergy does not have a role in this setting and that antigen-specific B cells have been eliminated by deletion or receptor editing.

The site at which the tolerizing interaction occurs between immature recipient B cells and donor A and B antigens is difficult to delineate in humans, thus definitive conclusions cannot be made from these experiments. Tolerance may be induced within the graft; alternatively, or in combination, tolerizing interactions may occur in nongraft sites such as spleen or other lymphoid organs, or in bone marrow or peripheral blood, through migration of graft-derived cells or dispersal of donor antigens through perioperative administration of donor-type noncellular blood products (platelets and plasma). Several findings suggest that transient exposure to donor antigens, such as administration of noncellular blood products, is insufficient to induce tolerance. First, patient 16 (blood group O) received a group B graft for a 3-d period only and was administered group B blood products intraoperatively. Yet after receiving a group O second graft, patient 16 developed B-specific antibody production to 'normal' titers for age (Fig. 1h). Thus, transient exposure to B antigen of the first donor, whether in the graft itself or by group B blood products, was insufficient to induce tolerance. Second, patient 4 (A2B→O) received group A1B platelets and plasma products containing both A and B antigens. Yet, after transplantation with a group A2B graft, patient 4 was not rendered tolerant to A antigen contained in group A1B blood products, developing A-specific antibodies with normal kinetics (Fig. 1a).

The spleen is thought to be important for A-specific and B-specific antibody production, inasmuch as splenectomy has been reported to diminish the production of these antibodies in ABO-incompatible adult kidney transplant recipients1. But it has also been shown that splenectomy in children does not prevent development of polysaccharide-specific antibodies41, showing that production of these antibodies is not limited to the spleen. In our one case of a congenitally asplenic person (patient 6, A→O), B-specific antibody developed normally (Fig. 1d), consistent with the theory that isohemagglutinin production is not limited to the spleen, whereas A-specific antibody production did not develop at all (Fig. 1c), suggesting that interactions leading to tolerance also are not confined to the spleen.

These combined findings suggest, in accordance with animal tolerance models27,28, that persistent exposure to donor antigens is required for tolerance induction, possibly in the graft itself, and that tolerance is dependent on a certain degree of antigen expression. Consistent with these results, experiments with 3-83 (H-2k,b-specific) immunoglobulin transgenic mice showed that specific self-reactive B cells are eliminated by deletion or receptor editing upon repeated exposure to antigen-bearing cells42.

The developmental limits of susceptibility to tolerance induction to donor A and B antigens have not yet been delineated. Although patient 9 (A→O) already had established A-specific antibody production prior to transplant at age 14 months (titer 1:128; Fig. 1c), A-specific antibody was depleted by plasma exchange during the operative procedure and did not reaccumulate to clinically significant levels by 5 years after transplant, showing a persistent antibody deficiency similar to younger A→O recipients. Furthermore, A-specific antibody-producing cells could not be detected by ELISPOT assay (Table 1). Thus, acquisition of 'immune competence,' shown by antibody production before transplant, did not preclude continued susceptibility to apparent tolerance induction. This raises the possibility that the window during which ABO-incompatible transplantation could be performed might be prolonged by repeated intentional exposure to nonself A and B antigens before transplant. This single case notwithstanding, caution must be advised for consideration of ABO-incompatible transplantation in children in whom isohemagglutinin production has already begun to develop.

Both systemic immunosuppression and (neonatal) thymectomy have been shown to interfere with tolerance induction in experimental settings43,44,45,46,47. That tolerance developed in ABO-incompatible infant recipients despite these factors may be related to the fact that polysaccharide antigens such as blood group substances are considered to be 'T-cell independent'48, and most systemic immunosuppressive drugs are directed predominantly at T-cell function, which is largely mature in neonates49.

Whether tolerance to donor A and B antigens in infants is also associated with tolerance to donor HLA has not been addressed in these studies. Although tolerance may develop to donor HLA, one could also speculate that the developmental differences in maturation of human immune responses to carbohydrates versus proteins19,21 may result in susceptibility to tolerance induction to carbohydrate antigens (A and B) concomitant with 'sensitization' to protein antigens (HLA).

Neonatally induced transplantation tolerance has not been shown previously in humans. The setting of ABO-incompatible infant heart transplantation is unique in that it allows an early immune intervention and a clear strategy of assessing immunity versus tolerance to defined donor antigen(s). We have delineated the mechanism of successful ABO-incompatible infant heart transplantation as “actively acquired tolerance of foreign cells”9, and characterized here as donor antigen–specific B-cell elimination. For the first time, we have shown that human infants are sufficiently immature for intentional induction of immunologic tolerance, at least in the B-cell compartment to T-independent A and B antigens. Through understanding the mechanisms underlying tolerance in this setting, it may be possible to expand the potential applications of this strategy to older patient populations, which could result in decreased mortality for patients awaiting transplantation and increased usage of donor organs. Additionally, our findings have significance for xenogeneic transplantation in which similar issues of polysaccharide antigen and antibody production remain major obstacles to success50.

Methods

Patients and biologic samples.

We studied 16 cardiac allograft recipients between February 1996 and October 2002 (Table 1); 13 infants (median age at transplant 10.4 weeks) received grafts from ABO-incompatible donors and 3 infants (median age 20.4 weeks) received ABO-compatible grafts. Of the 16 transplant recipients, 15 recipients are clinically well at current ages ranging from 2.2 to 8.4 years; one recipient died (age 2.5 years) due to accelerated graft vasculopathy (undetectable donor-specific antibody at time of death). We reported heart transplant and immunosuppression details previously24. Near-total thymectomy was performed as a routine part of infant cardiac surgery; splenectomy was not performed. We obtained peripheral serum samples for isohemagglutinin detection at defined time points after transplant. Endomyocardial biopsies for clinical rejection surveillance were obtained at defined time points; we performed immunohistochemistry staining to detect antibody-mediated rejection and donor antigen persistence. For a subset of 15 transplant recipients (Table 1), additional whole blood samples were obtained for B-cell functional analysis. We performed experiments in compliance with institutional guidelines and approved by the Research Ethics Board of The Hospital for Sick Children. Informed consent was obtained from transplant recipients' parents.

Blood group determination and isohemagglutinin assay.

We determined blood types using standard blood bank procedures. Serum A-specific and B-specific antibody titers were measured using agglutination tests with erythrocytes of known blood type ('reverse' blood typing). We diluted patient serum with saline in ratios ranging from 1:1 to 1:256 and mixed the serum with group A or B erythrocytes. A modified version of the isohemagglutination assay was used to test for serum factors that might inhibit erythrocyte agglutination by A-specific or B-specific antibodies (Supplementary Methods online).

Immunohistochemical staining.

We stained endomyocardial biopsy sections for deposition of immunoglobulin and complement components and for persistent expression of donor A or B antigens (Supplementary Methods online) and a cardiac immunopathologist examined the sections.

Cell isolation and cultures.

We layered patient peripheral blood onto HISTOPAQUE-1077 solution (Sigma). Following centrifugation, we removed serum and stored it at −30 °C. PBMC at the serum-HISTOPAQUE interface were washed and either used immediately for cell culture or flow cytometry, or resuspended in fetal bovine serum (FBS) containing 10% dimethyl sulfoxide (Sigma) for long-term storage in liquid nitrogen. We cultured 1 × 105 cells per well in RPMI 1640 medium supplemented with 10% FBS, 5 μg/ml bovine insulin (Sigma), 5 μg/ml human transferrin (Sigma), 5 ng/ml sodium selenite, 50 μM 2-mercaptoethanol, 100 U/ml penicillin and 100 μg/ml streptomycin. For erythrocyte-stimulated cultures, 2 × 105 cells per well of washed human erythrocytes were added on the first day of culture. We did not change medium before supernatant collection. In cytokine-stimulated cultures, 200 U/ml human recombinant interleukin(IL)-2 (PeproTech Canada) and 50 U/ml human recombinant IL-4 (PeproTech Canada) were added on the first day of culture.

Enzyme-linked immunosorbent assay (ELISA).

We coated polystyrene plates (Corning) with 5 μg/ml synthetic group A or B trisaccharide conjugated to bovine serum albumin (V-Labs). Nonspecific binding sites were blocked with PBS-Tween containing 2% FBS. We cultured patients' cryopreserved PBMC with or without stimulation. After 10 d, culture supernatants were incubated in wells for 1 h; bound antibodies were detected using horseradish peroxidase–conjugated goat antibodies to human IgM or IgG (Southern Biotechnology, 1:4,000 dilution) followed by color development with 3,3′,5,5′-tetramethylbenzidine (Sigma). We stopped the reaction with sulfuric acid after 30 min and measured absorbance at 450 nm. Preparation of a standard curve allowed antibody concentration to be determined from optical density. Negative controls were PBMC from blood group A (A-specific negative control) and blood group B (B-specific negative control) volunteers; assay results for negative control samples were 0–4 ng/ml; assay results >5 ng/ml were scored as positive.

Enzyme-linked immunospot (ELISPOT) assay.

We coated nitrocellulose membranes of 96-well MultiScreen-HA plates (Millipore) with 5 μg/ml synthetic group A or B trisaccharide conjugated to bovine serum albumin. Cryopreserved PBMC were cultured for 10 d with IL-2 and IL-4. After washing, we cultured cells for 24 h in MultiScreen-HA plates and stained them with biotin-conjugated goat antibodies to human IgM or IgG (Southern Biotechnology, 1:4,000 dilution). Incubation with streptavidin-horseradish peroxidase was followed by color development with AEC solution. Negative controls, in which we did not visualize any spots, were PBMC from blood group A (A-specific negative control) and blood group B (B-specific negative control) volunteers.

Flow cytometry.

We immunostained PBMC for direct detection of B lymphocytes bearing A antigen–specific BCR. PBMC (5 × 105 cells) were incubated with 1.25 μg/ml synthetic group A tetrasaccharide-PAA-biotin conjugate (GlycoTech) and FITC-conjugated human CD19-specific monoclonal antibody (PharMingen, 10 μl/106 cells), followed by strep-tavidin-phycoerythrin. We used group A and O PBMC as negative and positive controls, respectively, with at least two samples for each experimental group. We gated CD19+ cells using FACScan flow cytometer (BD Biosciences) and analyzed data with WinMDI Version 2.8. The detection limit of group A antigen-specific BCR-bearing cells in the negative control population was approximately 0.2% of total CD19+ B cells.

Note: Supplementary information is available on the Nature Medicine website.