Costimulation Blockade Alters Germinal Center Responses and Prevents Antibody-Mediated Rejection



De novo donor-specific antibody (DSA) after organ transplantation promotes antibody-mediated rejection (AMR) and causes late graft loss. Previously, we demonstrated that depletion using anti-CD3 immunotoxin combined with tacrolimus and alefacept (AMR regimen) reliably induced early DSA production with AMR in a nonhuman primate kidney transplant model. Five animals were assigned as positive AMR controls, four received additional belatacept and four received additional anti-CD40 mAb (2C10R4). Notably, production of early de novo DSA was completely attenuated with additional belatacept or 2C10R4 treatment. In accordance with this, while positive controls experienced a decrease in peripheral IgM+ B cells, bela- and 2C10R4-added groups maintained a predominant population of IgM+ B cells, potentially indicating decreased isotype switching. Central memory T cells (CD4+CD28+CD95+) as well as PD-1hiCD4+ T cells were decreased in both bela-added and 2C10R4-added groups. In analyzing germinal center (GC) reactions in situ, lymph nodes further revealed a reduction of B cell clonal expansion, GC-follicular helper T (Tfh) cells, and IL-21 production inside GCs with additional belatacept or 2C10R4 treatment. Here we provide evidence that belatacept and 2C10R4 selectively suppresses the humoral response via regulating Tfh cells and prevents AMR in this nonhuman primate model.


antibody-mediated rejection


antigen-presenting cell


B cell activating factor


costimulation blockade


cytotoxic T-lymphocyte antigen 4


donor-specific antibody


germinal center


inducible T cell costimulator




keyhole limpet hemocyanin


lymphocyte function associated antigen 3


mean survival time


peripheral blood mononuclear cell


programmed cell death 1


postoperation day


SLAM-associated protein


signaling lymphocyte activation molecule


central memory T cells


follicular helper T


In the last decade, anti-HLA donor-specific antibodies (DSAs) arising after kidney transplantation, or de novo DSA, have been increasingly recognized as a significant cause of late graft loss [1, 2]. Specifically, approximately 15–30% of renal transplant recipients develop de novo DSA [3, 4], and despite numerous treatment strategies augmenting conventional immunosuppression or targeting the inhibition or removal of B cells, plasma cells, antibodies and/or complement, no satisfactory therapy has been shown to reliably reverse the effects of DSA once established [5].

While reversal of DSA, or desensitization, has proven to be difficult, several approaches have been successful at preventing antibody formation and antibody-mediated rejection (AMR) after transplantation [6, 7]. Among these approaches, the use of costimulation blockade (CoB) to inhibit T cell–dependent antibody production has been well documented. Larsen et al [8] demonstrated enhanced inhibition of anti-sheep red blood cell antibodies with LEA29Y (belatacept; Bristol Myers Squibb) compared to its parent cytotoxic T-lymphocyte antigen 4 (CTLA-4) Ig. Lowe et al [9] observed that blockade of the CD40/40L pathway using anti-CD40 2C10R4 completely blocked antigen-specific antibody production in macaques immunized with keyhole limpet hemocyanin (KLH) antigen. Combined blockade of both CD28:B7 and CD40:40L pathways suppressed DSA formation in kidney-transplanted macaques [10]. Belatacept in clinical kidney transplant trials also has been associated with remarkably little DSA [11]. The mechanisms of DSA inhibition by CoB, however, have not been fully elucidated.

The above studies highlight the requirement of T cell help for humoral responses after transplantation [12, 13], as CD4+ T cell help is necessary for producing long-lasting IgG isotype-switched alloantibodies [14, 15]. In secondary lymphoid organs, T cells primed by antigen-presenting cells (APCs) differentiate into T cell subsets, including the recently defined follicular helper T (Tfh) cells. Tfh cells subsequently migrate into germinal centers (GCs) and interface with GC B cells through a number of surface signaling molecules: CD40-40L, CD28-CD80/86, SAP (SLAM-associated protein)-signaling lymphocyte activation molecule (SLAM; or CD150) family receptors, inducible T cell costimulator (ICOS)-ICOSL, OX40-OX40L, CXCR5-CXCL13, IL-4 and IL-21 receptors, and more [16], resulting in GC B cell differentiation into memory B cells and plasma cells. For both T cell priming by APCs and GC interactions between Tfh cells and B cells, costimulation via the CD28 and CD40 pathways plays a critical role and has immediate therapeutic potential [17, 18]. We suspect that blockade of these costimulation pathways will prevent effective GC reactions and the consequent production of class-switched DSA.

We therefore investigated the effects of B7-specific belatacept and CD40-specific 2C10R4 on DSA production and resultant antibody-mediated injury to renal allografts in a preclinical model of de novo DSA formation. We recently observed that by depleting T cells with an anti-CD3 immunotoxin (IT) (A-dmDT390-scfbDb(C207)) and treating with the calcineurin inhibitor tacrolimus and the CD2-specific fusion protein alefacept (lymphocyte function associated antigen 3 [LFA3]-Ig) during repopulation, treated animals exhibited a mean survival time (MST) of 59 days; however, they developed DSAs by 4 weeks after transplantation and renal allografts demonstrated morphologic changes characteristic of antibody-mediated injury, namely transplant glomerulopathy, peritubular capillaritis, and basement membrane thickening and duplication [19]. Here, by adding CoB agents to the AMR-inducing regimen, we examine the effects on de novo DSA formation and explore the specific effects on Tfh differentiation and function, GC reaction, and the downstream humoral response.


Animal selection and transplantation

Many of the materials and methods used in this study have been previously outlined in Page et al [19]. Male rhesus macaques weighing 4–7 kg were tested specific pathogen free and selected for expression of FN18 epitope, the binding site of anti-CD3 IT. Donor-recipient pairs were selected based on avoidance of MHC class I matches (six alleles tested), and class II maximal mismatch among available animals. Each animal underwent native nephrectomy and recipient transplantation. All medications and procedures were approved by the Emory University Institutional Animal Care and Use Committee, and were conducted in accordance with Yerkes National Primate Research Center and the National Institutes of Health guidelines.

Immunosuppression agents

AMR regimen

All animals received the base regimen of anti-CD3 IT (A-dmDT390-scfbDb(C207), 0.025 mg/kg intravenous [IV] twice daily; Massachusetts General Hospital—Dana Farber-Harvard Cancer Center Recombinant Protein Expression and Purification Core Facility, Boston, MA), tacrolimus (Astellas Pharma US, Inc., Deerfield, IL) and alefacept (LFA3-Ig; Astellas Pharma US, Inc.). As the original dose of IT given from postoperation day (POD) 0 to 3 produced significant viremia and weight loss, we reduced the IT dose to 0.025 mg/kg IV twice daily on POD 0 and 1 for the following monkeys: DP7D, FA5K, FA5N, DR6G, DW11 and RDk12. To alleviate potential symptoms of cytokine storm, methylprednisolone 125 mg was IV injected on days the animals received IT. Tacrolimus was started at 0.05 mg/kg intramuscular (IM) injection twice daily on the day of transplantation, and titrated through day 180 to maintain a trough of 8–12 ng/mL. Alefacept was administered at 0.3 mg/kg once weekly for 8 weeks, starting on POD −3, 0, then 7, 14, etc, based on the dosage used in Weaver et al [20]. Rescue immunosuppression with methylprednisolone (40, 62.5 or 125 mg) IM or IV for 3 days was used when the animal exhibited signs of acute rejection or with reduced tacrolimus administration due to weight loss (Figure 1).

Figure 1.

Immunosuppressive regimens, dosing strategy, clinical interventions and outcomes in antibody-mediated rejection (AMR) control and additional costimulation blockade-treated rhesus monkeys.

Study groups

Five animals were assigned to the AMR positive control arm of IT + tacrolimus + alefacept alone, four of which have been previously reported [19]. Four animals were assigned to receive 20 mg/kg IV of LEA29Y (belatacept; Bristol-Myers Squibb, Princeton, NJ) on posttransplant days −3, 0, 3, 7, 14, 28, 42 and 56, in addition to the base AMR regimen. Four animals were assigned to receive 20 mg/kg IV of the chimeric mouse-rhesus IgG4 anti-CD40, 2C10R4 (Nonhuman Primate Reagent Resource, Boston, MA) on posttransplant days 0, 7, 14, 28, 42 and 56, in addition to the base AMR regimen.

Allograft and immune monitoring

Peripheral blood was obtained weekly by femoral venipuncture for allograft and immune monitoring. Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll method using 5 mL of lymphocyte separation medium (Mediatech, Inc., Manassas, VA) per sample. Washed PBMCs were surface stained with the following antibodies: Alexa-Fluor 700 or PerCP Cy5.5 conjugated anti-CD3 (SP34-2), PerCP-Cy5.5 conjugated anti-CD4 (L200), APC-Cy7 or Pacific Blue conjugated anti-CD8 (RPA-T8) and Alexa Fluor ®700 conjugated IgG (G18–145) (all from BD Pharmingen, San Diego, CA); PE-Cy7 conjugated anti-CD28 (CD28.2), eFluor450 conjugated anti-CD95 (DX2) and APC conjugated anti-human CD279 (J105) (all from eBioscience, San Diego, CA); APC-Cy7 conjugated CD20 (2H7) and Pacific Blue conjugated IgM (MHM-88) (both from Biolegend, San Diego, CA). FITC conjugated goat anti-human IgD (δ chain specific, catalog no. 2030-02) was obtained from Southern Biotech (Birmingham, AL). PE conjugated CXCR5 (710D82.1) from Nonhuman Primate Reagent Resources and PE conjugated anti-human CXCR5 (MU5UBEE; only for Figure 5B) from eBioscience were used for Tfh cell staining. Cells were then processed to detect Ki-67 using intracellular staining with Fix/Perm solution (eBioscience) and PE conjugated anti-Ki-67 (BD Pharmingen). Lymph node tissues were crushed through a 100 μm strainer, washed and stained with the above antibodies. Samples were collected with an LSRII flow cytometer (BD Biosciences, San Jose, CA) and analyzed using FlowJo software 9.2 (Tree Star, Ashland, OR).

Detection of DSA

Alloantibody production was retrospectively assessed by flow cytometric crossmatch of donor PBMCs with serially collected recipient serum samples. Donor PBMCs were coated with ChromPure goat IgG (Jackson ImmunoResearch, West Grove, PA) and incubated with recipient serum. Cells were stained with FITC-labeled anti-monkey IgG (KPL, Inc., Gaithersburg, MD), PE CD20 (BD Pharmingen) and PerCP CD3 (BD Pharmingen). A twofold increase in mean fluorescence intensity from pretransplant values was used to define alloantibody positivity, with all positive shifts being greater than 100 shift in mean fluorescence intensity.

Histology, immunofluorescence staining and quantitative image analysis

Renal allograft tissues were obtained at time of rejection or sacrifice, fixed in 10% neutral buffered formalin and embedded in paraffin. Embedded tissue blocks were sliced into 5-μm thick sections, deparaffinized in xylene and rehydrated through graded ethanol to water for H&E and PAS stains as well as immunohistochemical analysis. For GC and Tfh cell staining, lymph node biopsies were fixed by immersion in neutral-buffered 10% formalin and embedded in paraffin. Immunofluorescence staining was performed for CD4, CD20, CD3, Ki67, IL-21 and programmed cell death 1 (PD-1) to identify the GC in lymph nodes, as previously described [21]. In brief, 4- to 5-µm tissue sections were deparaffinized and rehydrated. Heat-induced epitope recovery of organs were performed with DIVA Decloaker and then blocked with the SNIPER reagent (Biocare, Walnut Creek, CA) for 15 min and in PBS/0.1% Triton X-100/4% donkey serum for 30 min at room temperature. Subsequently, the sections were incubated with goat anti-human PD-1 (R&D Systems, Minneapolis, MN), mouse anti-human Ki67 (clone MM1; Vector, Burlingame, CA), rabbit or mouse anti-human CD20 (Thermo Scientific, Rockford, IL; or clone L26 [Novocastra Laboratories, UK]), mouse anti-human CD4 (clone BC/1F6; Abcam, Cambridge, MA), rabbit polyclonal anti-human IL-21 (AbD Serotec, Raleigh, NC) and rat anti-human CD3 (clone CD3-12; AbD Serotec) antibodies diluted 1:20–1:100 in blocking buffer for 1 h at room temperature. Thereafter, the sections were incubated with secondary antibodies (Alexa Fluor 488/Cy3/Cy5 conjugated appropriate donkey anti-mouse/rabbit/rat/goat antibodies) diluted 1:1000 (Jackson ImmunoResearch) in blocking buffer for 30 min at room temperature. Finally, the sections were stained with Hoechst 33342 (Invitrogen, Carlsbad, CA) diluted in water to 0.0025% for 10 min at room temperature, rinsed again and mounted in warmed glycerol gelatin (Sigma, St. Louis, MO) containing 4 mg/mL n-propyl gallate (Fluka, Switzerland). Between each step, the sections were washed three times with blocking buffer. All images were acquired with an Axio Imager Z1 microscope (Zeiss, Jena, Germany) using various objectives. For quantification, the stained area was calculated using AxioVs40 V4.8.1.0 program (Zeiss). Positive fluorescence signals of Ki67 and IL-21 within follicles were measured using Image J1.43u (NIH, Bethesda, MD). Labeled cells (PD-1+ CD4+ cells) were be manually counted using FluoView software 1.7 (Olympus, Tokyo, Japan). Transplant pathologists reviewed all histology and immunohistochemistry in a blinded fashion.


Experimental variables were analyzed by Prism statistical analysis program (GraphPad Software 6.0, San Diego, CA) or SPSS 19 (IBM, Chicago, IL) using the log-rank test for differences in graft survival and nonparametric Mann–Whitney or Kolmogorov–Smirnov tests for others. When box and whiskers plots are used, the box size represents the limits of the data for the second and third quartiles, with median shown as a bar. The whiskers define the minimum and maximum of the data represented. Error bars represent the mean ± SD in all bar graphs. Values of p < 0.05 were considered to be statistically significant.


The effect of additional CoB on de novo AMR model

All rhesus macaques received the AMR regimen of CD3 IT, alefacept and tacrolimus maintenance, as outlined in the Methods section. Five animals were assigned as positive AMR controls, four of which were previously reported [19]. Four animals received additional belatacept and four received additional 2C10R4. While the five AMR control animals were sacrificed for graft rejection (MST = 59.4 days), seven of eight animals receiving additional CoB were sacrificed due to poor clinical condition that reached the end-point IACUC criteria (weight loss greater than 25%) or per recommendation of attending Yerkes veterinarian (weight loss less than 25% with poor prognosis) but not due to allograft rejection (MST = 54.8 days). Sacrifice time points were similar in all three groups (n = 0.632), allowing for comparable analysis of organs (i.e. blood, lymph nodes, spleens, allografts) procured at time of sacrifice. One animal (RDk12) from the additional 2C10R4-treated group was sacrificed at day 179 for bladder outlet obstruction caused by bladder stone. Dosing strategy, clinical intervention and reasons for euthanasia are summarized in Figure 1. Serum creatinine level at the time of sacrifice was significantly lower in additional CoB-treated groups compared to AMR controls (0.95 ± 0.27 vs. 5.74 ± 2.17; p < 0.001). The data suggest that additional CoB alleviates early graft rejection in this T cell depletion-induced nonhuman primate AMR model.

CoB suppresses early de novo DSA production shown in an AMR-inducing regimen

The addition of CoB showed dramatic improvement in preventing allograft rejection based on longitudinal evaluation of serum creatinine levels (Figure 2A). As described previously, the AMR regimen-treated animals showed varying degrees of transplant glomerulopathy, arteriolar endothelial injury and C4d deposition [19]. However, explanted grafts revealed a decrease in cellular rejection and antibody-mediated transplant glomerulopathy in belatacept- and 2C10R4-treated animals (Figure 2B). While AMR controls demonstrated glomerular basement membrane thickening and striation on electron microscopy, a hallmark of antibody-mediated injury, both CoB treatments prevented this process (Figure 2C). Notably, production of early de novo DSA was completely attenuated for both CoB-treated groups at 4 and 6 weeks posttransplantation compared to AMR positive controls (Figure 2D). These data suggest that in this depletion-based AMR model, additional CoB suppresses de novo DSA production and antibody-mediated injury.

Figure 2.

Costimulation blockade prevents allograft rejection and alloantibody production. (A) Serum creatinine level of AMR positive control, belatacept-treated and 2C10R4-treated groups. AMR positive control group (n = 5, red line), receiving CD3-immunotoxin, alefacept and tacrolimus (trough 8–12 ng/mL), had early alloantibody production within 4 weeks from transplantation (the first detection of DSAs from AMR positive controls is indicated by ▾) and were sacrificed for graft failure. Belatacept-treated (n = 4, blue line) and 2C10R4-treated (n = 4, purple line) groups received the AMR regimen plus additional belatacept or additional 2C10R4 for 8 weeks (duration and frequency indicated by arrows along the x-axis); these animals failed to demonstrate DSA production and retained graft function, but were sacrificed for nonrejection end-point criteria. Animal survival was equivalent among the three groups (p = 0.632). (B) Histological examination of necropsy kidney specimens from healthy control, AMR control, belatacept-added and 2C10R4-added group. Representative H&E staining portrays evidence of cellular and AMR in the AMR positive control group, including tubulointerstitial inflammation and transplant glomerulopathy. These features were prevented in animals receiving additional costimulation blockade. Original magnification ×200 (bar scale: 100 μm). (C) An electron microscopy image of glomerular basement membranes from healthy control, AMR control, bela-added and 2C10R4-added group. AMR control animals showed thickening, lamination and duplication of glomerular basement membranes. (D) De novo DSA production was inhibited in animals receiving belatacept and 2C10R4. *p < 0.05 versus AMR control. AMR, antibody-mediated rejection; DSA, donor-specific antibodies.

CoB prevents the maturation of B cells in peripheral blood and memory T cell populations in secondary lymphoid organs

All animals exhibited an initial drop in peripheral blood B cells, as previously observed [19]; AMR controls increased their peripheral counts back to baseline by 1 month after transplantation, but animals treated with CoB suppressed this repopulation (Figure 3A). Compared to AMR controls, CoB-treated animals had fewer memory-phenotype B cells (CD27+CD20+) and a predominance of IgM+ B cells by day 28 (Figure 3B). Consistent with this finding, CoB-treated animals had reduced proliferation of class switched IgG+ B cells in peripheral blood (Figure S1). These data suggest that CoB suppresses B cell differentiation to memory and isotype switching—and thus maintains a more naïve phenotype of remaining circulating B cells.

Figure 3.

Costimulation blockade prevents B cell repopulation and maturation, and maintains naïve CD4 T cell phenotype in secondary lymphoid organs. (A) Peripheral blood B cells decreased after induction immunosuppression, as reported previously [18]. In contrast to AMR controls, CoB-treated animals failed to repopulate B cells. *p < 0.05. (B) Memory type B cells (CD20+CD27+) are decreased and immature phenotype (IgM+CD20+) B cells are predominant in CoB-treated animals. The drop in IgM+ B cells among AMR controls coincides with DSA production and IgG+B cell proliferation (Figure S1). (C) Peripheral blood T cells remain decreased in all groups. The isolated difference at 35 days may represent onset of allograft rejection in AMR controls. *p < 0.05. (D) Central memory CD4 T cells (CD4+CD28+CD95+) were significantly decreased and CD4 naïve T cells (CD4+CD28+CD95) increased in bela- (n = 4, blue) and 2C10R4-added (n = 3, purple) animals in the spleen and inguinal lymph nodes compared to AMR positive controls (n = 5, red) at time of necropsy (One recipient from 2C10R4-added group was censored from this analysis due to its necropsy time point performed at 179 days posttransplant; the subject was sacrificed for nonrejection end-point criteria; Figure S4). NS > 0.05. AMR, antibody-mediated rejection; CoB, costimulation blockade; DSA, donor-specific antibodies.

All animals had sustained peripheral T cell depletion after transplantation (Figure 3C). Consistent with previous report, rapid phenotypic switching from naïve to memory T cells was observed in AMR controls. However, both CoB-treated groups showed decreased percentages of central memory T cells (Tcm; CD4+CD28+CD95+) compared to the AMR controls and maintained percentages of naïve T cells (CD4+CD28+CD95) compared to the healthy controls, particularly in secondary lymphoid organs (Figure 3D), which likely reflects the requirement of costimulation for T cell differentiation after priming by APC. Additionally, Tfh cells have been speculated to differentiate into the Tcm phenotype [22], and this bulk decrease in Tcm could be surmised to have decreased T cell help for B cell activation.

CoB suppresses GC reconstruction after T cell depletion

Since the GCs are the sites for clonal B cell expansion, affinity maturation and isotype switching [23], we evaluated GC development within lymph nodes. We previously reported that T cells, even with profound depletion, were not completely depleted from the lymph nodes compared to those in blood, spleen and bone marrow [19]. Interestingly, despite incomplete depleting T cells, CD3-IT abolished GC structure in the lymph nodes as well as spleen (Figure 4A). Based on CD20 staining areas, the average B cell follicle size was comparable across all treatment groups as well as in healthy controls (Figure 4B and C). This result suggests that the B cell populations in lymph nodes are largely unaffected by potent immunosuppression, unlike in peripheral blood. GC size and frequency did not vary significantly between healthy and AMR positive controls, given the heterogeneic GC responses in normal healthy rhesus macaques (Figure S2). However, GC frequency and size were dramatically reduced in the belatacept- and 2C10R4-treated groups (Figure 4D and E); in contrast, belatacept treatment alone only minimally reduced GC frequency and size (Figure S3), as previously shown [24]. Since we also observed initial obliteration of GC after T cell depletion, this underscores the effect of combining CoB with T cell depletion, as neither belatacept alone or T cell depletion without CoB (i.e. AMR positive controls) was able to elicit this effect.

Figure 4.

Posttransplant follicle and GC morphology in secondary lymphoid organs. (A) Representative immunofluorescence image of GC staining with CD3 (blue), CD20 (red) and Ki67 (green) in lymph node and spleen sections from untreated (healthy control) and CD3-immunotoxin-treated rhesus monkey (posttransplant day 13). (B) Immunofluorescence analysis for CD20 (red) staining in lymph node sections from healthy control, AMR positive control, additional belatacept, and additional 2C10R4-treated recipients. Original magnification ×200. (C) Image J quantitation of the positive fluorescence signals of CD20 in lymph node sections. Follicle size based on CD20+ B cells was similar across all groups, including healthy controls. (D) Hoechst nuclear staining in paraffin-embedded lymph node sections. Original magnification ×200. GCs in lymph nodes of rhesus macaques were identified by Hoechst nuclear staining. This less intensely stained GC area was confirmed as corresponding to the same area containing proliferating (Ki67+) B cells (Figure S2). (E) ImageJ quantitation of the less intensely stained Hoechst area. GC size was significantly reduced in bela- and 2C10R4-added animals compared to AMR positive controls (*p < 0.05). These data are representative lymph node sections from healthy controls (n = 4), AMR positive controls (n = 5), bela-added (n = 4) and 2C10R4-added (n = 3) rhesus macaque recipients. AMR, antibody-mediated rejection; GC, germinal center.

PD-1+CD4 T cells in lymph nodes

Given the role of CD28 and CD40 signaling during T cell differentiation we hypothesized that inhibition of T cell help, especially Tfh cells, may be involved in the decrease in DSAs seen in treated animals because highly specialized Tfh cells are crucial to GC development [25, 26]. We thus evaluated lymph node cells for Tfh phenotypes from tissues procured at necropsy time points. Tfh cells, which tightly regulate the GC reaction, are characterized by high expression of CXCR5, BCL-6, PD-1 and ICOS [27]. We confirmed by flow cytometry that PD-1hi CD4 T cells were greatly reduced in CoB-treated recipients (Figure 5A). In contrast, CXCR5+CD4 T cells in the lymph node showed no difference among treatment groups (data not shown). It is difficult, however, to extrapolate these populations as representing GC-Tfh, given that PD-1 is expressed by other CD4+ T cells [28], and that the nature of CXCR5+CD4 T cells in rhesus macaques is unclear [27]. However, flow analysis of lymph node cells from healthy animal with newly defined antibodies (which was not available during the present study) revealed high level of homogeneous PD-1+ expression on CXCR5+BCL-6+ CD4 T cells (Figure 5B).

Figure 5.

Decreased PD-1+CD4+ T cells in lymph nodes in additional costimulation blockade-treated animals. (A) Flow cytometry contour plots show PD-1 expression among CD4 T cells in draining inguinal lymph nodes from healthy controls (n = 5) as well as AMR positive controls (n = 5), bela-added (n = 4), 2C10R4-added (n = 3) rhesus macaque recipients. The frequency of the gated population (PD-1hiCD4+) is shown as a bar graph. *p < 0.05, **p < 0.01, ***p < 0.001, versus AMR control. (B) Flow cytometry analysis of CD4 T cells based on the expression of CXCR5 and BCL-6 in the lymph nodes of rhesus macaque. Cells were first gated for CD4 T cells and gated based on the differential expression of CXCR5 and BCL-6. High PD-1 expression was shown in CXCR5+BCL-6+ CD4 T cell subsets. Numbers in dot plots indicate the percentage of cells in each gate. AMR, antibody-mediated rejection; PD-1, programmed cell death 1.

CoB suppresses Tfh cells and B cell clonal expansion in lymph nodes

GC-Tfh cells are structurally confined within the GC, flow analysis of minced lymph node cells is unable to preserve its locational information. To identify GC-Tfh cells in situ, we immunostained the draining lymph nodes to first demarcate the B cell follicles and GCs for structural context. Clonal B cell expansion (Ki67+CD20+) inside GCs was significantly higher in the AMR positive controls, but suppressed in animals treated with additional CoB (Figure 6A), consistent with the decrease in circulating peripheral blood IgG+CD20+ B cells (Figure S1). Once the boundaries of the B cell follicles and GCs were established, GC-Tfh cells were identified according to their expression of PD-1. GC-Tfh cells (PD-1hiCD4+) in lymph nodes were significantly decreased in the belatacept-added group compared to AMR control (Figure 6B). Similarly, IL-21 production in follicles, reflecting GC-Tfh function, was greatly reduced in both belatacept- and 2C10R4-added groups (Figure 6C). Furthermore, IL-21 staining within follicle showed a strong correlation with the DSA level (r = 0.87). IL-21 has been implicated as a key regulator for Tfh cell development. Tfh cells are also known to express high levels of IL-21, which is required for cognate B cell help and GC formation [29]. The downstream effect of decreased IL-21 corresponds to the above finding of decreased clonal expansion of B cells with CoB. A less dramatic difference of GC-Tfh cells was noted in 2C10R4-treated monkeys relative to AMR monkeys, which could be due to PD-1 staining signal from other sources outside of GC, such as activated CD4 cells (non-GC-Tfh). These data suggest that adding CoB affects Tfh cells and their function in T cell–depleted animals that otherwise preserve Tfh function, B cell clonal expansion and DSA production.

Figure 6.

Effects of costimulation blockade on B cell clonal expansion, GC-Tfh and IL-21 production. (A) B cell clonal expansion in GCs was significantly increased in AMR group (*p < 0.05 vs. healthy control). The increased B cell proliferation found in AMR positive controls was significantly suppressed by additional belatacept and 2C10R4 (*p < 0.05 vs. AMR) treatment. Proliferating B cells in GCs are identified by CD20, Ki67 and CD3 staining. (B) Tfh cells were significantly reduced in bela-added (*p < 0.05 vs. AMR) but not in 2C10R4-added (p = 0.1429 vs. AMR) rhesus macaques. CD4, CD20, PD-1 were simultaneously analyzed for Tfh cells in the GC. (C) IL-21 expression was significantly reduced in bela-added and 2C10R4-added animals compared to AMR positive controls (*p < 0.05). These data are representative lymph node sections from healthy controls (n = 4), AMR positive controls (n = 5), bela-added (n = 4) and 2C10R4-added (n = 3) rhesus macaque recipients. Original magnification ×200 (scale bar: 50 μm). AMR, antibody-mediated rejection; GC, germinal center; PD-1, programmed cell death 1; Tfh, T follicular helper.


The inhibition of T cell–dependent antibody production by CoB has been previously reported, as discussed above, but the mechanisms through which humoral responses are inhibited are less clear. We assessed the blockade of CD28 and CD40 pathways in a rigorous, clinically relevant preclinical model that paralleled our observations from human trials. In our clinical trials, T cell depletion with alemtuzumab (CAMPATH-1H) with rapamycin maintenance has been associated with increased DSA production during repopulation [30], and preliminary evidence suggests that this can be prevented by belatacept [31]. In our primate AMR model, we found that rapidly proliferating CD4 memory T cells that were resistant to T cell depletion correlated with early alloantibody production. The current study illustrates that adding CoB to this model decreased CD4 Tcm population in spleen and lymph nodes as well as de novo alloantibody production. This decrease likely represents inhibition of T cell activation via APC-driven CD28 signal that triggers the naïve T cells to become effector T cells in conjunction with T cell receptor signal. Differentiation and survival of Tfh cells are highly dependent on cognate B cell interaction. Costimulatory molecules CD40L, CD28, ICOS, PD-1 and OX40 are involved in antigen-specific contact with B cells expressing MHC class II [16]. CoB might alter this cognate interface resulting in the diminished GC response, B cell expansion/differentiation and downstream DSA production. Furthermore, T cell depletion resulted in early eradication of GC, supported by addition of CoB treatment, in contrast to AMR controls that had fully reconstructed, hyperplastic GC. The early effect of T cell depletion on GC suggests that the depletion process may intimately affect Tfh cells. That GC responses return in the AMR controls also suggests that Tfh cells are specifically inhibited by CoB. The inhibition of GC B cell clonal expansion and Tfh function (with IL-21 expression) required the combination of both CoB and T cell depletion, as belatacept alone did not accomplish this effect.

Neither CoB nor T cell depletion alone was sufficient for producing long-term effects on the GC reaction and antibody response. Our clinical experience with T cell depletion using alemtuzumab showed up-regulated humoral responses, with increased DSA and serum B cell activating factor (BAFF) levels [32]. A significant subset of multiple sclerosis patients treated with alemtuzumab had increased IL-21 levels, which correlated with development of autoimmune thyroiditis [33]. Thus, despite extensive T cell depletion, T cells refractory to depletion may continue to provide B cell help. CD28 and CD40 have been known to be required for GC formation, as animals deficient in one or both molecules cannot induce proliferative B cell expansion, class-switch recombination, and GC formation in response to T cell–dependent antigens [34, 35]. However, while treatment with CTLA-4 Ig (Abatacept) suppressed acquisition of Tfh cell phenotype in lymph nodes of a murine rheumatoid arthritis model, clonal expansion of B cells remained intact [36]. Also, macaques treated with long-term belatacept were found to have only minimal to moderate reductions in GC size, compared to the profound reduction we observed when combining belatacept with T cell depletion [24]. Animals also developed anti-KLH antibodies upon cessation of belatacept, demonstrating the reversible and transient nature of T cell–dependent antibody inhibition by belatacept alone [24]. This implies that targeting Tfh cells by combining T cell depletion with CoB has a more profound and lasting downstream effect on humoral responses, namely the alloantibody response. This is further supported by our long-term surviving monkey, RDk12, who failed to demonstrate DSAs at 179 days, well beyond completing an anti-CD40 treatment course on day 56 (Figure S4). This effect was likely donor-specific and not due to generalized over-immunosuppression, as the animal was able to mount GC reactions comparable to healthy controls at 179 days.

Due to its clinical outcome (poor prognosis/weight loss) and drug availability, the exact regimen is not clinically translatable. While a poor therapeutic agent, the anti-CD3 IT in combination with tacrolimus and alefacept was efficient at producing DSA and AMR and thus served as a robust preclinical nonhuman primate AMR model. Our results show that adding CD28 or CD154 CoB can suppress Tfh cell differentiation, B cell clonal expansion, GC reaction and DSA formation that conventional immunosuppressants such as tacrolimus fail to prevent in our AMR model. Moreover, the novel finding that CoB combined with T cell depletion specifically produces these effects on the B cell has implications for potential clinical use, particularly for the reduction of humoral responses. The data support further testing of CoB and depletional induction therapy in a clinically translatable regimen to reduce posttransplant humoral response. While risks of over-immunosuppression must be considered, these results support the use of belatacept and CD40 blockade in combination with lymphodepletion to prevent AMR.


This work was supported by the NIH 1U01AI074635 awarded to SJK, NIH R01AI078775 to FV and collaboration with the Yerkes National Primate Center (Yerkes base Grant OD P51POD111). We would like gratefully acknowledge: Drs. Andrew Adams and Mandy Ford for critical review of the manuscript; Frank Leopardi and Kelly Hamby for assisting in animal surgeries; the YNPRC staff and the expert assistance of Drs. Elizabeth Strobert and Joe Jenkins for animal care; the Emory Transplant Center Biorepository team for their work in weekly viral monitoring; and Hong Yi and the Robert P. Apkarian, Integrated Electron Microscopy Core of Emory University, for their work in tissue analysis for electron microscopy. Reagents used in this study were provided by the Nonhuman Primate Reagent Resource (HHSN272200900037C).

Authors' Contributions

EJK designed experiments, performed surgical procedures and cared for experimental macaques, conducted in vitro experiments, interpreted data and prepared the manuscript. JK designed experiments, performed surgical procedures, cared for experimental macaques, conducted in vitro experiments, interpreted data and prepared the manuscript. ACG conducted in vitro experiments. JJH performed histology and immunohistochemistry, interpreted data and prepared the manuscript. ABF interpreted data (pathologist). NNI, FV and ADK interpreted data and prepared the manuscript. SJK conceived of experimental design, performed surgical procedures, cared for experimental macaques, interpreted data and prepared manuscript.


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