CD28 Blockade Ex Vivo Induces Alloantigen-Specific Immune Tolerance but Preserves T-Cell Pathogen Reactivity

Abstract

Donor T-cells contribute to reconstitution of protective immunity after allogeneic hematopoietic stem cell transplantation (HSCT) but must acquire specific tolerance against recipient alloantigens to avoid life-threatening graft-versus-host disease (GvHD). Systemic immunosuppressive drugs may abrogate severe GvHD, but this also impedes memory responses to invading pathogens. Here, we tested whether ex vivo blockade of CD28 co-stimulation can enable selective T-cell tolerization to alloantigens by facilitating CD80/86-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) signaling. Treatment of human allogeneic dendritic cell/T-cell co-cultures with a human CD28 blocking antibody fragment (α-huCD28) significantly abrogated subsequent allospecific immune responses, seen by decreased T-cell proliferation and of type 1 cytokine (IFN-γ and IL-2) expression. Allo-tolerization persisted after discontinuation of CD28 blockade and secondary alloantigen stimulation, as confirmed by enhanced CTLA-4 and PD-1 immune checkpoint signaling. However, T-cells retained reactivity to pathogens, supported by clonotyping of neo-primed and cross-reactive T-cells specific for Candida albicans or third-party antigens using deep sequencing analysis. In an MHC-mismatched murine model, we tolerized C57BL/6 T-cells by ex vivo exposure to a murine single chain Fv specific for CD28 (α-muCD28). Infusion of these cells, after α-muCD28 washout, into bone marrow-transplanted BALB/c mice caused allo-tolerance and did not induce GvHD-associated hepatic pathology. We conclude that selective CD28 blockade ex vivo can allow the generation of stably allo-tolerized T-cells that in turn do not induce graft-versus-host reactions while maintaining pathogen reactivity. Hence, CD28 co-stimulation blockade of donor T-cells may be a useful therapeutic approach to support the immune system after HSCT.

Keywords: CD28 blockade; alloantigen specific; ex vivo; graft-versus-host disease; preserved pathogen reactivity; tolerance.

Figure 1

Schema of in vitro allo-tolerization and retained pathogen reactivity by α-huCD28-mediated blockade of human T-cells. Alloantigen binding to the respective T-cell receptor (TCR) concurrently with CD28 blockade by α-huCD28 potentially tolerizes human T-cells, while CD80/86 co-stimulatory molecules remain accessible to negative regulators such as cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) (top). Human T-cells are co-cultured with MHC-mismatched human dendritic cells (DCs) presenting alloantigen (primary mixed leukocyte reaction), in the presence of the CD28 blocker α-huCD28. After 7 days of culture, T-cells are washed, rested for 2 days in the absence of α-huCD28, and re-stimulated with (A) the same alloantigen (fresh allogeneic DCs), (B) Candida albicans (autologous DCs), or (C) third-party alloantigen (third-party DCs).

Figure 2

CD28 blockade mediates persistent reduction of alloresponses. Proliferation of human T-cells co-cultured with allogeneic dendritic cells (DCs) in vitro in a primary or secondary allogeneic mixed leukocyte reaction (MLR). (A) Representative histogram and dotplot analyses of the percentage of carboxyfluorescein succinimidyl ester (CFSE)/CPD dilution and the mean fluorescence intensity of CD3⁺ T-cells in the primary (upper panel) and secondary MLR (lower panel), with (+) or without (−) α-huCD28. Quadrants were adjusted according to unproliferated (CFSE^pos/CPD^pos) cells. (B) Proliferation (CFSE^neg/CPD^neg) of human CD3⁺ T-cells and FACS-sorted naive (CD45RA⁺/RO⁻) or memory (CD45RA⁻/RO⁺) T-cells analyzed by flow cytometry. Proliferation is depicted as number of proliferated cells without (○) or with (▲) α-huCD28 after a 7-day primary MLR (left panel) and a 3-day secondary MLR (right panel). Lines represent results of different biological replicates; points represent the mean of triplicate determinations. Different recipient–donor pairs were used as biological replicates. ***p < 0.001, **p < 0.005, *p < 0.05.

Figure 3

CD28 blockade inhibits Th1 cytokines in secondary allo-mixed leukocyte reactions (MLRs). Th1-specific cytokine expression of FACS-sorted naive and memory T-cells in primary or secondary allo-MLRs. (A,C) Representative dotplots show IL-2 and IFN-γ expression of naive (A) or memory (C) T-cells treated with (+) or without (−) α-huCD28 in a primary and, after discontinuation of CD28 blockade, in a secondary MLR after 24 h. (B,D) Percentage of IFN-γ, IL-2, IFN-γ/IL-2 expressing, and IFN-γ/IL-2 negative naive (B) or memory (D) T-cells, with (▲) or without (○) prior α-huCD28 exposure, in a secondary MLR after 24 h. Pairs of points are the means of triplicate determinations of individual different recipient–donor biological samples. ***p < 0.001, **p < 0.005, *p < 0.05.

Figure 4

CD28 blockade induces checkpoint signaling. Expression of programmed death 1 (PD-1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) on sorted naive T-cells 24 h after a secondary allogeneic mixed leukocyte reaction (MLR) without (−) or with (+) prior α-huCD28. Cells were gated according to: CD3⁺/live cells/CD4⁺/CTLA-4⁺⁺ or CD3⁺/live cells/CD4⁺/PD-1⁺CTLA-4⁺⁺. CTLA-4⁺⁺ and PD-1⁺CTLA-4⁺⁺ α-huCD28 pre-treated cells were analyzed for cellular proliferation by carboxyfluorescein succinimidyl ester (CFSE)/CPD dilution. (A) Representative dotplots of PD-1 and CTLA-4 expression on naive CD4⁺ T-cells in a secondary MLR (day 1). Numbers represent percent within the CD3⁺ cell population. Cumulative results of CTLA-4 and PD-1 expressions of naive T-cells with (▲) or without (○) prior α-huCD28 after a 1 day secondary allo-MLR. Points represent the mean of replicates from different recipient–donor pairs. (B) T-cells with prior CD28 blockade, highly expressing CTLA-4 or PD-1/CTLA-4, were analyzed for proliferation (CFSE^neg/CPD^neg, proliferated). Representative dotplots of CFSE/CPD staining are shown; bar graphs show the cumulative results of percent proliferated T-cells. (C) CD3⁺ T-cell lysates from a primary MLR with (+) or without (−) α-huCD28 for 0–96 h were examined by Western blot to detect phospho-protein phosphatase 2A (pPP2A), total protein phosphatase 2A (PP2A), Akt phosphorylated on threonine 308 (pAkt Thr) or serine 473 residues (pAkt Ser), total Akt, and vinculin as a control. For densitometric analysis (ratio of phosphorylated/total protein) the ImageJ program was used. *p < 0.05.

Figure 5

CD28-tolerized T-cells retain pathogen reactivity. Proliferation of human CD3⁺ T-cells after re-stimulation with Candida albicans or foreign third-party alloantigen. (A–C) Data points for primary mixed leukocyte reactions (MLRs) are derived from Figure 2B (+ α-huCD28) for comparison. The number of proliferated cells in primary MLRs ranged from 1 × 10² to 9 × 10³. T-cell proliferation, shown as number of proliferated cells, was analyzed by flow cytometry after a 7-day re-stimulation with (A) autologous dendritic cells with C. albicans, (B) without C. albicans as a control for possible antigen-independent proliferation, or (C) with foreign third-party alloantigen. Each point is the mean of triplicate determinations of different recipient–donor pairs used as biological replicates. **p < 0.005; ns = not significant, p > 0.05.

Figure 6

Expansion of allo- versus pathogen-reactive T-cell clones after α-huCD28-mediated blockade. Clonal frequency of CD3⁺ T-cell cultures from a primary recipient–donor pair mixed leukocyte reaction (MLR) and re-stimulation with either 1st party alloantigen, Candida albicans, or third-party alloantigen, as analyzed by deep TCRβ sequencing. (A) Expansion of T-cell clones in a primary MLR without (gray bars) or with α-huCD28 (black bars) and a secondary MLR in the absence of α-huCD28. Top 200 clones are ranked from their lowest to highest frequency in cultures without α-huCD28. The top 10 clones are represented individually in all graphs. (B) Clonal expansion after allo-tolerization and re-stimulation with C. albicans or third-party alloantigen. The frequency of antigen-independent expanded T-cell clones (T-cells stimulated with autologous DCs without C. albicans) was subtracted from the Candida-specific clonal expansion. The top 200 clones were ranked from high to low abundance in C. albicans (red bars) or third-party alloantigen (green bars) re-stimulation cultures and compared to allo-tolerized primary MLRs (black bars). For an in depth investigation, the top 10 most expanded clones are represented individually in all graphs. Clone numbers 2 and 4–10 (left graph) and 2–7 and 9–10 (right graph) are present in primary MLRs, but in low numbers ranging from 0.004 to 0.039%. (C) Unique clones expanded upon stimulation with C. albicans (red bars) or third-party alloantigen (green bars) but not present in secondary MLRs with 1st party alloantigen. Clones are ranked from high-to-low abundance in the top 200 clonal repertoire (numbers at the bottom of the figure).

Figure 7

Ex vivo co-stimulation blocked T-cells do not induce graft-versus-host disease (GvHD) in an MHC-mismatched murine model. Experimental schema (A): T-cells isolated from C56BL/6 spleens were co-cultured with BALB/c dendritic cells (DCs) for 5 days with or without α-muCD28. Lethally irradiated BALB/c mice were transplanted with C56BL/6 bone marrow (BM) and FACS-sorted viable T-cells from mixed leukocyte reaction (MLR) cultures. Mice were monitored for signs of GvHD for 44 days. (B) T-cell proliferation: depicted by the number of proliferated C56BL/6 T-cells upon stimulation with BALB/c DCs together with (+) or without (−) α-muCD28 after a 5-day MLR (left graph), before infusion; cytolytic activity: cytotoxicity of alloreactive CD8⁺ and CD4⁺ T-cells (right graph) against MC38 target cells after 21 h is shown. (C) GvHD score of BALB/c mice transplanted with C57BL/6 BM alone (●, n = 6) or together with non-tolerant (freshly isolated) T-cells (■, n = 7), or T-cells without (○, n = 15) or with prior α-muCD28 treatment ex vivo (▲, n = 12). (D) H&E staining of BALB/c livers, necrotic lesions are indicated by black borders, 10× magnification. Necrosis was assessed by quantifying the affected areas using the ImageJ program (− α-muCD28 n = 11, + α-muCD28 n = 11, freshly isolated Tc n = 5, and BM only n = 6). Outliers were removed by applying Grubbs’ test (alpha = 0.1). The sum of the areas of all measurable lesions in a high power field is depicted. Bars represent 100 µm. (E) Dilated blood vessels are indicated by black arrows, 4× magnification. Vasodilation was quantified as the mean of all vessel diameters in a high power field (− α-muCD28 n = 11, + α-muCD28 n = 10, freshly isolated Tc n = 4, BM only n = 4). Outliers were removed by applying Grubbs’ test (alpha = 0.1). Bars represent 500 µm. Image acquisition in (D,E) was done with a Nikon Eclipse 80i microscope with a build on camera Nikon DS-Fi1. Magnification of the object lenses was 500 μm = 4×, 100 μm = 10×. NIS-Elements F4.30.00 software was used for image acquisition. Data shown comprise two separate experiments, scored in a blinded fashion independently by two observers. **p < 0.005, *p < 0.05.