Blood Journal
Leading the way in experimental and clinical research in hematology

Differential CD52 expression by distinct myeloid dendritic cell subsets: implications for alemtuzumab activity at the level of antigen presentation in allogeneic graft-host interactions in transplantation

  1. Gudrun Ratzinger,
  2. John L. Reagan,
  3. Glenn Heller,
  4. Klaus J. Busam, and
  5. James W. Young
  1. 1 From the Laboratory of Cellular Immunobiology, Department of Medicine; Department of Epidemiology and Biostatistics; Surgical Pathology Service, Department of Pathology; Allogeneic Transplantation and Clinical Immunology Services, Division of Hematologic Oncology, Department of Medicine; Memorial Sloan-Kettering Cancer Center, New York, NY; and Weill Medical College of Cornell University, New York, NY.
This article has an Erratum 105(8):3018

Abstract

Alemtuzumab (anti-CD52; Campath 1-H) depletes both host and donor T cells when used in preparative regimens for allogeneic transplantation. This promotes engraftment even after nonmyeloablative conditioning and limits graft-versus-host disease (GVHD) even after unrelated or major histocompatibility complex (MHC) disparate allografts. We asked whether anti-CD52 differentially targets antigen-presenting cells (APCs), in addition to depleting T cells. Monocyte-derived dendritic cells (moDCs) expressed abundant CD52 as expected. Langerhans cells (LCs) and dermal-interstitial DCs (DDC-IDCs), however, never expressed CD52. Immunostaining of skin and gut confirmed the absence of CD52 on these resident DC populations under both steady-state and inflammatory conditions. Although anti-CD52 functions primarily by antibody-dependent cellular cytotoxicity (ADCC) in vivo, assessment of its activity in vitro included complement-dependent lysis of CD52+ cells. Anti-CD52 did not impair DC–T–cell adhesion, diminish DC-stimulated T-cell proliferation, or alter moDC development in vitro. We propose that anti-CD52 abrogates GVHD not only by T-cell depletion, but also by removing moDCs and their precursors. This would mitigate moDC phagocytosis and presentation of host-derived antigens to donor T cells in the inflammatory peritransplantation environment, thereby limiting GVHD. The sparing of LCs and DDC-IDCs by anti-CD52, as well as the recovery of donor-derived moDCs in a less inflammatory environment later after transplantation, may allow all these DCs to exert formative roles in graft-versus-tumor (GVT) reactions and immune reconstitution. Whether these results support a separation of deleterious from beneficial graft-host interactions at the level of antigen presentation, rather than solely at the level of T cells, will require further evaluation.

Introduction

CD52 is a small, 12-amino acid, phosphatidylinositol (GPI)–anchored membrane glycoprotein expressed by lymphocytes, especially T cells, as well as monocytes, macrophages, monocyte-derived dendritic cells (moDCs), and the epithelial cells of the distal epididymis and vas deferens.1-3 All of the functions of CD52 may not yet be known, but it constitutes at least a target for complement-mediated cell lysis and antibody-mediated cellular cytotoxicity.4 5 Complement activation, however, is neither necessary nor sufficient for monoclonal antibody (MAb) depletion of CD52+ cells in vivo.6 7Antibody-dependent cellular cytotoxicity (ADCC) is likely the more important mechanism for depleting CD52+ cells sensitized in vivo by approximately 1000-fold less antibody than required for complement-dependent lysis.8 These amounts of antibody approximate the levels detected in patients receiving alemtuzumab in vivo.9

Alemtuzumab is a recombinant DNA-derived, humanized MAb directed against CD52 that is very efficient in mediating lymphocyte depletion both in vitro and in vivo.10 Treatment with humanized anti-CD52 in vivo mitigates graft-versus-host disease (GVHD) and promotes engraftment, even in adult recipients of unmodified allografts from unrelated or mismatched donors.11 12 This is especially noteworthy because other regimens that achieve comparable degrees of T-cell depletion or immune suppression have not always proven similarly successful in these settings. This led us to ask whether anti-CD52 targets different types of dendritic cells (DCs) in addition to depleting T cells.

We focused on DCs because these are the most potent antigen-presenting cells (APCs) and the most critical to initiation of cellular immune responses.13-15 A growing body of data has also emerged regarding the development of DCs and their hematopoietic relationship to other leukocytes.16-25 From these studies one can distinguish at least 3 different types of myeloid DCs, for which investigators increasingly find different specialized functions. These myeloid DCs comprise at least CD34+ hematopoietic progenitor cell (HPC)–derived Langerhans cells (LCs) and dermal-interstitial dendritic cells (DDC-IDCs), as well as CD14+ blood moDCs.16-25 Phenotypically, all 3 mature myeloid DC types are class II major histocompatibility complex (MHC)bright, CD86++, CD14, CD11c+, and importantly, CD83+.26LCs are CD11b, whereas the other 2 are CD11b+.17 27

CD14+ monocytes express abundant CD52, as do moDCs identified by a MAb CMRF-56.2 We evaluated expression and function of the CD52 antigen on all 3 myeloid DC types, however. We focused especially on the CD52 expression pattern of CD34+HPC-derived, cytokine-generated LCs and DDC-IDCs, as well as their resident counterparts in skin and gut mucosa. We compared these with the well-characterized moDCs generated in vitro, which have all the properties ascribed to and expected of DCs, but which have proven difficult to identify specifically in vivo.

Our findings suggest additional mechanisms that may underlie the efficacy of anti-CD52, which go beyond T-cell depletion. These results may also have important implications for future studies to determine whether deleterious graft-host interactions such as GVHD can be distinguished at the level of antigen presentation from beneficial processes like graft-versus-tumor (GVT) activity and immune reconstitution.

Materials and methods

Media and reagents

RPMI 1640, 10 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid), 1% penicillin/streptomycin (Media Lab Core Facility, Memorial Sloan Kettering Cancer Center [MSKCC]), 50 μM 2-mercaptoethanol (2-ME; Gibco BRL Life Technologies, Carlsbad, CA), 1%l-glutamine (Gibco BRL), and 1% heat-inactivated, autologous human plasma or serum were used for culture of moDCs with cytokines as outlined below. X-VIVO 15 (Biowhittaker, Walkersville, Maryland), without either serum or plasma but supplemented with cytokines as specified, was used to culture CD34+ HPCs for the generation of LCs or DDC-IDCs. The DDC-IDC cultures, however, included an initial 5 days in Iscove modified Dulbecco medium (IMDM), 1% penicillin/streptomycin (Media Lab, MSKCC), 1% l-glutamine (Gibco BRL), and 50 μM 2-ME (Gibco BRL), supplemented with 20% autologous plasma before transfer to X-VIVO 15. Allogeneic mixed leukocyte reactions (MLRs) were cultured in complete RPMI medium, supplemented with 10% heat-inactivated single-donor human serum but no exogenous cytokines.

Fresh human plasma (50% vol/vol) that was not heat inactivated, or commercial human complement (no. S1764, Sigma, St Louis, MO) was used for MAb/complement-dependent lysis. The commercial product, used strictly according to manufacturer's instructions, proved more stable and reproducible.

Recombinant human cytokines used for in vitro generation of DCs were granulocyte-macrophage colony-stimulating factor (GM-CSF; Immunex, Seattle, WA); interleukin 4 (IL-4), tumor necrosis factor α (TNF-α), transforming growth factor β (TGF-β), c-kitligand or stem cell factor, FLT-3 ligand, IL-1β, IL-6 (all from R & D Systems, Minneapolis, MN); and prostaglandin E2(PGE2; Calbiochem, San Diego, CA). Humanized anti-CD52 was obtained as the pharmaceutical alemtuzumab (Berlex Laboratories, Richmond, CA) for functional studies in vitro. Humanized anti-CD20 was obtained as the pharmaceutical rituximab (Genentech, South San Francisco, CA) for functional studies in vitro as a control humanized MAb that would be nonreactive with T cells and myeloid DCs. Doses are specified for the respective cultures in “Cell purification and generation of DCs.”

Cell purification and generation of DCs

All cells were obtained from healthy individuals who were already serving as donors for allogeneic hematopoietic stem cell transplantation (HSCT), either by harvesting bone marrow or pheresing granulocyte colony-stimulating factor (G-CSF)–stimulated peripheral blood stem cells (PBSCs). Donors signed informed consents for research sample collection protocols reviewed by the Memorial Sloan-Kettering Cancer Center Institutional Review Board.

The moDC precursors were obtained from tissue culture plastic-adherent peripheral blood mononuclear cells (PBMCs) after standard separation over Ficoll-Paque PLUS (no. 17-1440-03, Amersham Pharmacia Biotech, Uppsala, Sweden). Tissue culture plastic-adherent mononuclear cells were cultured in GM-CSF (1000 IU/mL) and IL-4 (500 IU/mL) as published.28 Medium and cytokines were replenished every 2 days. At approximately day 6, large forward scatter (FSC), HLA-DR+ cells expressed intracellular CD83, confirming commitment to DC differentiation, but very little cell surface CD83 (not shown). An inflammatory cytokine cocktail was added to these immature moDCs for terminal maturation and activation. This mixture comprised IL-1β (2 ng/mL), IL-6 (1000 IU/mL), TNF-α (10 ng/mL), and PGE2 (5 μM).29 30 By day 8 of culture, large FSC, HLA-DRbright cells were CD14 and more than 90% surface CD83+ and CD86+.

CD34+ HPCs were obtained by positive immunomagnetic selection from bone marrow or G-CSF–elicited PBMCs separated over Ficoll-Paque PLUS (Amersham Pharmacia Biotech) according to manufacturer's instructions (CD34+ isolation kit and LS separation columns, Miltenyi Biotec, Bergisch Gladbach, Germany). LCs and DDC-IDCs were separately generated from CD34+ HPCs in the media described and as previously published17 18 20 22 23. Specific cytokine supplements included GM-CSF (1000 IU/mL), TNF-α (5 ng/mL), c-kitligand (20 ng/mL), and FLT-3 ligand (50 μg/mL), with removal of c-kit ligand and FLT-3 ligand from day 5 to 6 onward. CD34+ HPC cultures were replenished with cytokines and media on day 3 and thereafter approximately every other day.

Two amendments were made to these cytokine mixtures. For the specific generation of LCs, TGF-β1 (10 ng/mL) was provided throughout the entire culture period.22 23 For the specific generation of DDC-IDCs, IL-4 (500 IU/mL) was added to suppress macrophage differentiation31 when the cells were recultured at day 5 to 6 in X-VIVO 15 supplemented with GM-CSF and TNF-α but without c-kit ligand and FLT-3 ligand. By day 12, large FSC, HLA-DR+ cells expressed intracellular CD83, confirming commitment to DC differentiation, but little surface CD83. Terminal maturation of these immature LCs or DDC/IDCs was accomplished from day 12 to 14 of culture, by providing the same inflammatory cytokine cocktail used for moDCs. The resultant large FSC, HLA-DRbright cells were CD14, more than 60% to 70% CD83+, and approximately 90% CD86++. LCs were CD11b and DDC-IDCs were CD11b+. The remaining progeny in the total cell population were immature granulocytic cells, especially eosinophils, because intermediate steps to enrich for DCs and deplete other myeloid CD34+ HPC progeny were not undertaken.

T cells were obtained from the PBMC fraction that was nonadherent to tissue culture plastic. Nonadherence and elution from nylon wool columns (Polysciences, Warrington, PA) further purified these cells in excess of 95%.

Phenotypic analyses by flow cytometry

Direct fluorescein isothiocyanate (FITC)–conjugated and phycoerythrin (PE)–conjugated mouse antihuman MAbs included anti-CD3–FITC, anti-CD3–PE, anti-CD14–PE, anti-CD16–PE, anti-CD20–PE, anti-CD11b–PE, anti-CD11c–PE (Pharmingen, Franklin Lakes, NJ); anti-CD4–PE, anti-CD8–PE, anti-CD14–PE, anti-CD14–FITC, anti-CD34–FITC, anti-CD11c–allophycocyanin (Becton Dickinson, Franklin Lakes, NJ); and anti-CD34–PE, anti-CD83–PE (Immunotech, Marseille, France). Isotype controls included IgG1-FITC, IgG1-PE, and IgG2a-FITC (Dako, Carpinteria, CA); and rat IgG2b-FITC (Serotec, Oxford, United Kingdom). Rat antihuman CD52-FITC and its respective rat IgG2b-FITC control (Serotec) were used for phenotypic but not functional assessments of CD52. Annexin-V (Early Apoptosis Detection Kit, Kamiya Biomedical, Seattle, WA) and propidium iodide staining, respectively, distinguished apoptotic and necrotic cells. Cytofluorographic evaluation used a FACScan (Becton Dickinson, Immunocytometry Systems, San Jose, CA), gating for live events. For analysis of specific epitope expression by DCs, candidate cells were first gated for large FSC, HLA-DRbright cells, after which 10 000 events were collected for analysis.

MAb and complement-dependent cell lysis

We used MAb/complement-dependent lysis in vitro as a surrogate measurement of anti-CD52 function in vivo. Assessment by ADCC in vitro, which is the more likely mechanism through which anti-CD52 exerts its effects in vivo,6 proved not to be logistically feasible.

Cells were cultured in complete RPMI supplemented with 50% fresh human plasma without heat inactivation or with commercial complement equivalent to 50% plasma. As negative controls for complement, heat-inactivated (56°C for 30 minutes) normal human plasma or serum was used. Humanized anti-CD52 (alemtuzumab) and nonreactive control humanized anti-CD20 (rituximab) were added at the concentrations indicated for each experiment. Cells were opsonized with MAb for 30 to 40 minutes on ice, thoroughly washed, and exposed to complement or plasma (50% vol/vol) for 1 hour. When certain DCs proved resistant to MAb/complement-dependent lysis, owing to their lack of CD52 expression, complement exposure of MAb-opsonized cells was extended overnight to exclude the possibility that low-level CD52 expression might mediate some degree of cell targeting and lysis by alemtuzumab. After washing, the remaining viable cells were counted directly by trypan blue exclusion on a hemacytometer.

Immunohistochemistry

Immunohistochemical studies were performed on formalin-fixed and paraffin-embedded tissues. The antibodies used included anti–S-100 protein (1:50 000; Biogenix, San Ramon, CA) and anti-CD52 (1:40; Serotec). The tissue sections were exposed to the antibodies in citrate buffer solution at pH 6.0. Detection of the primary antibody was performed with a biotinylated secondary antibody (1:100; Vector, Burlingame, CA) for 30 minutes followed by an avidin-biotin complex system (Vector), using diaminobenzidine tetrahydrochloride (DAB; Biogenix) as chromogen. The slides were counterstained with Mayer hematoxylin (Sigma).

Allogenic mixed leukocyte reactions

DCs were cocultured with 105 purified allogeneic T cells (alloMLRs) in triplicate round-bottomed microwells at either a constant ratio of 30:1 for T/DCs and variable concentrations of MAbs, or at variable T/DC ratios of 30:1 to 1000:1 in the presence of a constant MAb concentration. Medium for allogeneic MLRs consisted of complete RPMI supplemented with 10% single-donor serum or plasma as described in “Media and reagents,” but with no exogenous cytokines. DCs were extensively washed to remove cytokines before adding to T cells.

Because only moDCs expressed CD52 in appreciable amounts (see “Results”), only moDCs were evaluated in alloMLRs after treatment with anti-CD52. The alloMLRs were cultured in the continuous presence of alemtuzumab, which targeted both moDC stimulators and T-cell responders. Alternatively, the moDCs were pretreated with humanized anti-CD52 (alemtuzumab) or control humanized anti-CD20 (rituximab) and complement, thoroughly washed, and then added to allogeneic stimulators in doses based on the DC yield in the control condition.

Proliferating T cells incorporated [methyl-3H]-thymidine (3HTdR; 1 μCi/well [0.037 MBq]; PerkinElmer Life Sciences, Boston, MA) during the last 12 hours of a 4- to 5-day culture. The amount of3HTdR incorporated was measured in a β scintillation counter (Betaplate, Wallac, Perkin Elmer Life Sciences, Wellesley, MA).

Results

CD52 is differentially expressed on DC subsets and their precursors

CD52 expression was determined by flow cytometry using a FITC-conjugated rat antihuman CD52 compared with a rat IgG2b-FITC control (Figure 1; Table1). Purified T cells were included as positive controls given their known expression of CD52. Fresh PBMCs were also stained and gated for candidate circulating, immature DCs, as CD11c+, lineage-negative cells (CD3, CD14, CD16, CD20), from which we determined that all expressed CD52. Monocytes also expressed abundant amounts of CD52, as expected, and more than 95% of the differentiated immature and mature moDCs expressed CD52 as well. The mean fluorescent intensity (MFI) decreased somewhat with maturation, indicating a decrease in CD52 density on the cell surface.

Fig. 1.

Flow cytometry analysis of CD52 expression by different myeloid DC populations and their precursors.

Flow cytometric analyses were performed to determine CD52 surface expression by immature (surface CD83, intracellular CD83+) and mature (CD83+) myeloid DCs and their precursors. T cells provided a positive control. Monocytes as well as their immature and mature moDC derivatives expressed abundant CD52, although the MFI decreased somewhat with maturation. Circulating, immature DCs, defined as CD11c+, lineage-negative (CD3, CD14, CD16, CD20) cells gated from total PBMCs, also expressed CD52 abundantly with an MFI comparable to that of T cells. A substantial proportion of CD34+ HPCs expressed CD52, although at a relatively low MFI compared with the positive T-cell control. CD11b LCs and CD11b+ DDC-IDCs displayed no detectable expression of CD52 above background, independent of the maturation state. The bold line histograms represent the reactivity of anti-CD52 MAb with the selected population, and broken line histograms depict the isotype control. One experiment representative of 3 is shown.

View this table:
Table 1.

Phenotypic expression of CD52

Approximately 80% of the starting CD34+ HPCs expressed CD52, although at a lower surface density than PBMCs and moDCs based on the MFI. Most importantly, however, neither immature nor mature LCs or DDC-IDCs derived from these CD34+ HPCs in vitro ever expressed CD52 at any subsequent stage of differentiation or maturation. This was confirmed by serial phenotyping from approximately day 3 until the end of the 2-week culture.

CD52 has no proliferative or adhesive function in DC/T-lymphocyte interactions

To evaluate whether CD52 alone influences DC function, allogeneic T cells and DCs were cocultured in the presence of humanized anti-CD52 (alemtuzumab) compared with the nonreactive humanized anti-CD20 (rituximab) control. To avoid the introduction of complement, 10% (vol/vol) heat-inactivated plasma or serum was used. The more potent immunostimulatory, mature CD83+ moDCs were combined in a fixed DC/T-cell ratio with allogeneic T cells (1 DC/30 T cells), and the dose of MAb added to the cultures was varied from 1 μg/mL to 1 mg/mL final. Humanized anti-CD52 did not inhibit the formation of DC/T-cell clusters, from which reactive T-cell blasts emerge, as assessed by direct inspection using inverse phase microscopy (not shown). T-cell proliferation also remained comparable to control conditions at all MAb doses (Figure 2). Hence, we conclude that CD52 plays no role in adhesion or proliferation in allogeneic DC/T-cell interactions.

Fig. 2.

Humanized anti-CD52, in the absence of complement-mediated lysis, does not inhibit the immunostimulatory properties of DCs.

Mature CD83+ moDCs were combined with purified allogeneic T cells at a fixed ratio of 1 DC/30 T cells in triplicate round-bottomed microwells. Humanized anti-CD52 (alemtuzumab) was added to the alloMLR cultures in graded doses from 1 μg/mL to 1 mg/mL final concentration. Rituximab, a nonreactive humanized anti-CD20 MAb, was used as a negative control. Complete RPMI medium was supplemented with 10% heat-inactivated (complement-depleted) single-donor normal human serum. T-cell proliferation (3HTdR incorporation) in the anti-CD52 condition was divided by that in the anti-CD20 condition to obtain a percent proliferation relative to normal, using anti-CD20 as a negative control MAb. Proliferation in the control anti-CD20 condition varied between 80 000 and 200 000 among 6 different allogeneic pairings. The figure summarizes the averaged triplicate means from 6 independent experiments; error bars represent the SD for these 6 experiments.

Alemtuzumab (anti-CD52) does not impair development and maturation of moDCs

MAbs have several mechanisms by which they may exert their cellular effects. In addition to complement-mediated cytotoxicity and ADCC, these mechanisms may also include the induction of apoptosis and the inhibition of metabolically active proteins such as cytokines and growth factors.10 32 To evaluate any role of anti-CD52 in this respect, humanized anti-CD52 (alemtuzumab) was added in excess at 1 mg/mL during the generation of moDCs from CD14+ monocytes in 1% heat-inactivated autologous plasma with cytokines. Cell counts as well as flow cytometric analysis for apoptosis and maturation did not reveal any differences between anti-CD52–supplemented cultures compared with controls (Table 2). Anti-CD52 therefore causes neither apoptosis nor inhibition of DC development and maturation from CD52+ CD14+class II MHC+ monocyte precursors.

View this table:
Table 2.

Assessment of anti-CD52 on the development of moDCs in vitro

Humanized anti-CD52 induces complement-mediated lysis in proportion to the level of CD52 expression

We used MAb/complement-dependent lysis in vitro as a surrogate measurement of anti-CD52 function in vivo, which is likely more dependent on ADCC.6 Cells were treated with anti-CD52/complement or control anti-CD20/complement, and percent lysis was calculated based on the recovery of viable cells detected by trypan blue exclusion on a direct hemacytometer count. Humanized anti-CD52 lysed CD52+ cells in the presence of complement and in proportion to the level of surface CD52 expression, with the exception of CD34+ HPCs (Figure 3).

Fig. 3.

Anti-CD52/complement-dependent lysis is proportional to the surface density of the CD52 epitope.

DC subsets and their progenitors were compared for surface expression of CD52 and sensitivity to complement-dependent lysis by humanized anti-CD52. Filled bars indicate the MFI of CD52 expression, plotted against the left y-axis (note log10 scale). Empty bars, plotted against the right y-axis, represent the percentage of cells lysed after opsonization with an excess of humanized anti-CD52 (1 mg/mL alemtuzumab), thorough washing, and exposure either to 50% plasma without heat inactivation or the equivalent of commercial complement. Complement exposure was normally 1 hour at 37°C, but cellular resistance to lysis was also confirmed even after overnight incubation. Expression of CD52 and sensitivity to anti-CD52/complement-mediated lysis in vitro are strongly correlated, with the exception of CD34+ HPCs (see also MFI depicted in Figure 1). This graph summarizes 3 independent experiments, and error bars represent the SD of the triplicate means.

As expected, T cells were lysed very efficiently, almost up to 100%. CD14+ monocytes, immature moDCs, and mature moDCs also proved sensitive to complement-mediated lysis by humanized anti-CD52. LCs and DDC-IDCs, however, proved resistant regardless of their maturation state, in keeping with their lack of CD52 expression. The only CD52+ cells that were resistant to MAb/complement-dependent lysis were CD34+ HPCs, which had a lower surface density of CD52 based on MFI.

Although moDCs expressed abundant CD52, independent of the maturation state, the MFI decreased somewhat with maturation. Mature moDCs proved less sensitive to anti-CD52/complement-mediated lysis than immature moDCs and with much greater variation at the highest dose of complement. Reduced concentrations of complement to 12.5% resulted in diminished lysis and increased survival of approximately 40% of mature moDCs (Figure 4). In contrast, MAb/complement-dependent lysis of immature moDCs and T cells remained complete at all concentrations of complement (Figure 4). MAb was used at 1 mg/mL in the experiments depicted to confirm that resistant cells were insensitive even to anti-CD52 used in excess.

Fig. 4.

Mature moDCs are less sensitive than monocyte precursors and immature moDCs to anti-CD52/complement lysis, in proportion to their lower CD52 expression.

Cells were cultured in triplicate overnight in the presence of humanized anti-CD52 (alemtuzumab) and complement. The source of complement was either fresh human plasma or commercial complement in an equivalent concentration, ranging from 50% to 12.5% plasma in the culture medium. Percent lysis was calculated based on the remaining viable cells that excluded trypan blue on a direct hemacytometer count. Three experiments are summarized, and error bars represent the SD of the averaged triplicate means. MAb was used at 1 mg/mL in the experiments depicted, to confirm that resistant cells were insensitive even to anti-CD52 used in excess. Pairwise comparison of 3 groups performed using the Wilcoxon rank sum statistic yielded a Pbetween .2 and .33 for mature moDCs compared with immature moDCs or T cells in 3 concentrations of human plasma.

Humanized anti-CD52/complement alters moDC stimulation of allogeneic T-cell proliferation in a dose-dependent manner

We evaluated the functional consequences of anti-CD52 targeting of CD52+ cells in the context of a fully allogeneic MLR. LCs and DDC/IDCs were not evaluated in these experiments, owing to their lack of CD52 expression and their resistance to anti-CD52/complement–dependent lysis.

Cells were exposed continuously throughout the alloMLR culture period to humanized anti-CD52 (alemtuzumab)/complement versus control humanized anti-CD20 (rituximab)/complement. Mature moDCs were used in these experiments as the optimal stimulators, even though they were more resistant to MAb/complement lysis (Figure 3) than their less immunostimulatory, day 5 to 6 immature moDC precursors. The dominant effect, as shown in Figure 5A, may therefore have been directed at the T-cell responders. By holding constant the amount of complement, there was a wide range of inferred lytic activity, especially at MAb doses below 30 μg/mL, which correspond to the levels recoverable from patients treated in vivo.9

Fig. 5.

Humanized anti-CD52, in the presence of complement, alters moDC stimulation of allogeneic T-cell proliferation in a dose-dependent manner.

(A) Mature CD83+ moDCs were combined with allogeneic T cells at a fixed ratio of 1 DC/30 T cells in alloMLRs. Humanized anti-CD52 (alemtuzumab) was added in graded doses from 1 μg/mL to 1 mg/mL. Complete RPMI was supplemented with complement-replete human plasma (not heat inactivated) or an equivalent amount of commercial complement. The proliferation of responder T cells in the continuous presence of humanized anti-CD52/complement, divided by the proliferation in the presence of control humanized anti-CD20/complement, yielded a stimulation index at each dose of MAb; 100% control proliferation ranged from 50 000 to 200 000 cpm3HTdR incorporation in 3 different allogeneic combinations. This inhibitory effect is due to MAb/complement-dependent lysis of both stimulator and responder cells; although the mature moDCs are relatively less sensitive, so the predominant effect is likely T-cell–based. The graph summarizes 3 independent experiments, and the error bars represent the SD of the averaged triplicate means from each experiment. Comparison of these 2 groups based on Wilcoxon rank sum test showed P < .02 for doses between 1 μg/mL and 1000 μg/mL. (B) Immature moDCs were opsonized with an excess of humanized anti-CD52 (alemtuzumab, 1 mg/mL final) or nonreactive control humanized anti-CD20 (rituximab, 1 mg/mL final) for 30 to 40 minutes on ice, thoroughly washed, and then exposed to complement or plasma (50% vol/vol) for 1 hour or overnight at 37°C. Both conditions started with the same number of cells and were handled identically. Yields and cell concentrations for reculture and terminal maturation of the moDCs, after MAb/complement treatment, but before addition as stimulators to an alloMLR, were based solely on the control anti-CD20 condition. Already matured DCs proved too variable in their sensitivity to MAb/complement lysis; so immature moDCs were treated, then matured and used for these experiments. Incorporation of3HTdR was assessed during the last 12 hours of a 4- to 5-day culture. The percent inhibition at each T/DC dose was calculated as follows: 100 − (mean cpm of T cells stimulated by anti-CD52–treated moDCs/mean cpm of T cells stimulated by anti-CD20 treated moDCs) × 100. Shown are the results of 3 independent experiments and the SD of the mean percent inhibition at each T/DC dose.

Alternatively, day 5 to 6 immature moDCs were opsonized with anti-CD52, washed, and exposed to complement before subsequent cytokine-induced maturation and addition to allogeneic T cells in the MLR. We used immature moDCs that were matured after MAb/complement exposure, because already matured moDCs do not exhibit consistent susceptibility to humanized anti-CD52 (alemtuzumab)/complement. Pretreatment of the moDCs also spared T cells from the effects of the MAb and instead specifically targeted the moDCs. As shown in Figure 5B, the percent inhibition relative to control humanized anti-CD20 (rituximab) was most pronounced at the lower stimulator doses, likely due to the strength of an allogeneic stimulus when higher moDC doses were used. In any case these lower stimulator doses would approximate T/DC ratios expected in vivo, even under inflammatory conditions.

In lieu of exposing MAb-treated cells to complement for the experiments depicted in Figure 5B, we opsonized immature moDCs with humanized anti-CD52 (alemtuzumab) or control humanized anti-CD20 (rituximab), then cultured these cells with nonadherent autologous PBMCs dosed to provide a ratio of natural killer (NK) cell (CD3CD56+) to moDC of 30:1 to 50:1. Use of the resultant cells (1500 rad 137Cs) as allogeneic MLR stimulators yielded very inconsistent results. Further attempts to incorporate such assessments of ADCC with regard to the function of alemtuzumab were therefore abandoned.

Resident populations of LCs and DDC-IDCs do not express CD52, similar to the progeny generated in vitro under cytokine-supported conditions

One of the hazards of drawing conclusions based on cells generated in vitro under the aegis of nonphysiologic doses of recombinant cytokines is that the culture itself may have artifactually skewed the results. We therefore examined CD52 protein expression immunohistochemically on archival human tissue (Figure6). The specimens included 2 samples each of normal skin and intestinal tissue, skin biopsies from patients with drug hypersensitivity reactions, and skin biopsies from patients with GVHD. None of the biopsies showed coexpression of S-100 protein–positive cells with CD52. Thus, resident epidermal or mucosal LCs were immunonegative for CD52. Likewise, coexpression of CD52 and S-100 protein–positive inflammatory cells of the dermis never occurred. Hence, DDC-IDCs were also immunonegative for CD52. These results argue definitively against the possibility that exogenous cytokines in vitro could have modulated CD52 surface expression compared with that found on resident populations of DCs in vivo.

Fig. 6.

Cutaneous or mucosal DCs do not express CD52 in vivo.

(A) Normal skin. Intraepidermal DCs (LCs) are immunopositive for S-100 protein but negative for CD52. Apart from dermal inflammatory CD52+ T-cell aggregates that could be seen as an artifact of the punch biopsy in some sections (not shown), no CD52+dendritic profiles were seen in the dermis. Factor XIIIa staining confirmed the presence of dermal DCs (not shown). (B) Normal colon. CD52+ T lymphocytes surround a lymphoid follicle in a colonic biopsy. Mucosal DCs, which are immunopositive for S-100 protein, are negative for CD52. Germinal center B cells stained in this paraffin-embedded and -fixed tissue were also negative as expected. (C) GVHD affecting the skin. Intraepidermal and dermal DCs are immunopositive for S-100 protein, but negative for CD52. Inflammatory T lymphocytes that are destroying the dermal-epidermal interface as part of the graft-host reaction are CD52+, but there are no CD52+ dendritic profiles. (D) Drug hypersensitivity reaction in the skin. CD52+ lymphocytes are seen in the superficial dermis. S-100 protein–positive DCs are present in the epidermis and superficial dermis but are consistently negative for CD52. Staining was performed as described in “Immunohistochemistry.” Original magnification × 100.

Discussion

The successful application of anti-CD52 in clinical transplantation derives in large part from its depletion of alloreactive T cells. This mechanism supports recent reports that recombinant DNA-derived, humanized anti-CD52 (alemtuzumab), administered in vivo, reduces the incidence and severity of GVHD after allogeneic HSCT while preserving the GVT benefit of the allograft, although follow-up is admittedly still limited.11 12Investigators have also now shown that anti-CD52 MAb eliminates host moDCs, compared with standard chemotherapy and radiation-based preparative regimens, but does not impair recovery of donor-derived moDCs.3

We asked whether, in addition to T lymphocytes, distinct types of DCs differentially expressed CD52 for targeting by alemtuzumab. Using 3 carefully defined populations of myeloid DCs, we found that moDCs expressed CD52 to the exclusion of other myeloid DC populations. The most surprising and novel finding was that LCs and DDC-IDCs were, in fact, always negative for the CD52 epitope, either as resident populations in normal or inflamed skin or gut or as cytokine-generated progeny of CD34+ HPCs in vitro. Although the operative mechanism by which humanized anti-CD52 (alemtuzumab) exerts its effects in vivo is likely by ADCC,6 MAb/complement-dependent lysis provided a useful approximation of anti-CD52 function for our studies in vitro.

Alemtuzumab, or anti-CD52, caused complement-dependent lysis in proportion to the amount of surface CD52 expression and was therefore lytic only for moDCs but not for either immature or mature LCs or DDC-IDCs. CD34+ HPCs also proved resistant, perhaps reflecting their lower surface density of CD52 expression. This resistance, however, is also consistent with the well-documented preservation of transplantable stem cells after anti-CD52 purging in vitro.33 34 It may also account for the paucity of actual cases of aplasia from treatment with alemtuzumab in vivo.

CD52 exerted no adhesive or proliferative functions between DCs and T cells, because anti-CD52 alone in the absence of complement inhibited neither DC/T-cell aggregation nor alloantigen-specific T-cell proliferative responses. CD52 also had no effect on moDC development because binding of CD52 by the MAb did not alter cytokine-driven differentiation in vitro.

The fact that anti-CD52 MAb therapy with alemtuzumab does not result in increased GVHD, given the persistence of LCs and DDC-IDCs, suggests 2 possible explanations. One is that the dominant effect of alemtuzumab is to cause such profound T-cell depletion that survival of LCs and DDC-IDCs is irrelevant. Our data suggest an additional possibility, which is that elimination of moDCs from the inflammatory environment early after transplantation removes a highly phagocytic and potent APC that could otherwise present antigen from dying host cells to surviving or newly generated donor T cells.

We offer several lines of evidence in support of this reasoning. First of all, despite the absence of randomized trials, comparable levels of T-cell depletion by other transplantation regimens have not regularly achieved success similar to anti-CD52–containing regimens in terms of immune reconstitution and reduced GVHD, especially in older adults receiving unrelated or mismatched allografts. This is true even when G-CSF–elicited, T cell–depleted PBSCs containing disproportionate numbers of Th2-inducing DCs35 have been used as the source of the allograft. Alemtuzumab may therefore mediate either qualitative differences in T-cell depletion or differentially affect antigen presentation in contrast to other T-cell depletion approaches. Secondly, infections by viruses like cytomegalovirus (CMV) seem not to be exerting negative effects on patient survival after alemtuzumab-facilitated transplantations, to the same extent as comparable CMV reactivation rates after other methods of T-cell depletion,11 suggesting more effective immune reconstitution. Although follow-up is still limited, relapse is so far not a greater cause of treatment failure after preparative regimens using anti-CD52.11 12 We would therefore propose that preservation of LCs and DDC-IDCs after anti-CD52 treatment allows these myeloid DCs to play formative roles in the redevelopment of acquired and beneficial immunity.

Several lines of evidence support the concept of moDCs as prime candidates for eliciting GVHD reactions, at least acutely. Monocytes and moDCs are intensely phagocytic,36 especially compared with other myeloid DCs, such as LCs and DDC/IDCs. In the inflammatory environment of an allogeneic transplantation, circulating monocyte precursors and immature moDCs would be ideally suited for uptake of dying host cells. Differentiation into mature moDCs and presentation of these host antigens to reactive clones of T cells circulating through secondary lymphoid organs would follow. Recent data also indicate that lipopolysaccharide (LPS) and CD14 are critical to the induction of experimental acute GVHD,37 38 further implicating a specific role for CD14+ moDCs in the sensitization arm of this process.

In the case of hematopoietic stem cell allografts, moDCs of either host39 or donor origin could phagocytose dying host cellular antigens for presentation to and sensitization of engrafting donor-derived T cells. These would in turn cause GVHD in the periphery, especially in those sites that most abundantly express the same MHC antigens, for example, skin, gut, liver, and lymphoid organs. In the case of other nonmyeloablative preparative regimens that allow persistence of host T cells, donor moDCs could even sensitize host T cells by the direct pathway, leading to host resistance and nonengraftment or rejection.

Similar logic applies to solid organ allografts, where MHC disparities are the rule rather than the exception and where chronic rejection and long-term graft survival remain problematic. Donor moDCs transferred among the so-called passenger leukocytes in an allograft could directly sensitize host T cells. Alternatively, host moDCs could phagocytose and present donor-derived cellular antigens from dying cells in the inflammatory environment of the transplanted allograft. Humanized anti-CD52 therapy should alter or eliminate both processes.

These results have important implications for the activity of alemtuzumab in allogeneic transplantation. Like other studies in this area,40-42 our findings support greater attention to the afferent arm of the immune response, rather than focusing on alloreactive T-cell responders/effectors to the potential exclusion of evaluating antigen presentation. This leads us to hypothesize that the undesirable complications of GVHD or rejection (host-versus-graft) may be distinguishable from the beneficial graft-versus-leukemia or GVT effects exerted by hematopoietic allografts, based on differences in afferent sensitization of an immune response by different types of myeloid DCs.

If true, this would predict that the selective elimination of moDCs by anti-CD52 in the inflammatory environment early after transplantation may promote the long-term tolerance and graft survival that has been so difficult to achieve with only T cell–targeted immune suppression in mismatched or unrelated host-donor pairings. Accordingly, preservation of resident LCs and DDC-IDCs may be as important to generating GVT and reconstituting normal cellular immunity as is depletion of highly phagocytic moDCs to the reduction of GVHD in the early and highly inflammatory posttransplantation environment. This does not exclude a role, however, for moDCs in the generation and maintenance of peripheral tolerance or GVT at later time points when the inflammatory cytokine milieu is substantially diminished. It also does not exclude a more dominant effect of anti-CD52 on T-cell responder populations, regardless of the DC populations that may be left intact or not. These concepts merit further specific testing in preclinical animal models and clinical trials.

Acknowledgments

We thank Eileen Walsh, RN, Frieda Toomasi, RN, the nurses and attending physicians of the Allogeneic Bone Marrow Transplantation Service, and the Allogeneic Transplant and Cytotherapy Laboratory staff, especially Nancy H. Collins, PhD, Sharon Bleau, and Zankar Desai, all for assistance with sample procurement and processing. We additionally acknowledge the technical assistance and expertise of Kristin Iversen in performing the immunohistochemical studies. We also appreciate the assistance of Scott Freeswick, RPh, in providing us with alemtuzumab and rituximab for these studies.

Footnotes

  • Prepublished online as Blood First Edition Paper, October 10, 2002; DOI 10.1182/blood-2002-04-1093.

  • Supported by grants P01 CA 23766 (J.W.Y.), R01 CA 83070 (J.W.Y.), and P01 CA 59350 (J.W.Y.) from the National Cancer Institute, National Institutes of Health; and LLS 6124-99 (J.W.Y.) from the Leukemia and Lymphoma Society of America.

  • Correspondence: James W. Young, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021; e-mail: youngjw{at}mskcc.org.

  • The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 U.S.C. section 1734.

  • Submitted April 10, 2002.
  • Accepted September 6, 2002.

References

View Abstract