Minor Histocompatibility Antigens on Canine Hemopoietic Progenitor Cells¹

Minor histocompatibility Ags (mHA)³ play an important role in allogeneic stem cell transplantation. In donor-recipient combinations identical for major histocompatibility Ags (HLA) mHA are responsible for complications such as graft rejection or graft-vs-host disease (GvHD). However, they are also responsible for the beneficial graft-vs-leukemia reaction (GvL) (1, 2, 3). The GvL effect is used in adoptive immunotherapy with donor lymphocyte transfusion (DLT) for patients with leukemic relapse after allogeneic stem cell transplantation. In chronic myelogenous leukemia, for example, the rate of long-enduring complete remissions after DLT is up to 80% (4, 5, 6). Prior animal experiments in our laboratory have demonstrated that DLT with unmodified T cells was able to convert mixed hematological chimerism into complete donor chimerism in DLA-identical littermates (7).

In acute leukemia the GvL effect of donor lymphocytes is inferior and generally not sufficient for a long-lasting remission. Using T cells immunized against mHA disparities of the host may improve the GvL effect of donor lymphocytes, but it also carries the risk of inducing severe GvHD. GvHD may be avoided if the generation of CTL is achieved with mHA presented exclusively on hemopoietic cells and not present on other tissues of the patient. The first promising results using mHA-specific CTL have been obtained in vitro (8), but to date no sufficient clinical data are available. A large animal model for using mHA-specific CTL in vivo would be a valuable tool regarding the safety of this promising treatment.

In the past many advances in transplantation biology have been derived from preclinical studies in dogs. In our laboratory adoptive immunotherapy with donor lymphocyte transfusion was developed in the canine model (7). Tolerance was induced by transplanting an animal with T cell-depleted bone marrow from its dog leukocyte Ag (DLA)-identical littermate. Sixty days after transplantation donor lymphocytes were transfused, converting the chimerism from mixed to full donor without inducing GvHD (7). Presumably residual hemopoietic stem cells of the host expressing mHA serve as targets for a mHA-directed cytotoxic reaction by donor T cells. Unfortunately, to date few data have been available concerning mHA in the dog (9, 10).

In the present study we characterized canine mHA on a cellular level in DLA-identical littermates. We have defined the conditions for production of APC from lineage-negative bone marrow. We studied the immunization of T cells against mHA in vitro and in vivo. We developed a suspension culture assay (Δ-Assay) to test T cell immunity against hemopoietic progenitor cells (HPC). Finally, we investigated canine families for segregation of mHA and for the identification of DLA restriction elements.

In our studies 1- to 16-year-old purebred beagles of both sexes with an average weight of 13 kg were used. These animals were raised in the kennels at the GSF-National Research Center for Environment and Health (Neuherberg, Germany). All dogs were healthy, regularly dewormed, and vaccinated against distemper, leptospirosis, parvovirus, and canine hepatitis. DLA were determined using alloimmune antisera against DLA-A and DLA-B Ags as defined in the Canine Leukocyte Ag Workshop (11). Cells were incubated with polyclonal DLA antisera and lysed through the addition of complement, and the percentage of dead cells was analyzed in FACS with propidium iodide staining. DLA-identical littermates were preselected by DLA serotyping and mutual nonreactivity in MLR. This was confirmed on the basis of highly polymorphic MHC-associated microsatellite markers (12). Genomic DNA (200 ng) was mixed with 0.5 μM tetra-chlorofluorescein-labeled forward primer, 0.5 μM reverse primer, 10× PCR buffer (200 mM Tris-HCl (pH 8.4) and 500 mM KCl), 1.5 mM MgCl₂, 1 mM of each dNTP, and 2.5 U of Taq polymerase (Invitrogen, San Diego, CA) in a final volume of 50 μl. Samples were amplified by denaturation at 94°C for 5 min, followed by 30 cycles of denaturation at 94°C for 20 s, annealing at 60°C for 20 s, and extension at 74°C for 30 s. A final extension step was performed for 5 min at 74°C. The fragment length of the samples was assessed with an ABI PRISM 377 DNA Sequencer (PE Applied Biosystems, Foster City, CA). Littermates were determined as DLA-identical if they both showed the same fragments in the microsatellite PCR.

Bone marrow samples were aspirated from the humerus of dogs under general anesthesia. Bone marrow mononuclear cells were separated over Ficoll (Seromed, Berlin, Germany), washed twice for thrombocyte removal, and resuspended in IMDM supplemented with either 10% inactivated FCS or 5% inactivated pooled dog serum. Peripheral blood was sampled by venipuncture, and mononuclear cells were separated over Ficoll, washed, and kept in RPMI 1640 supplemented with either 10% inactivated FCS or 5% inactivated pooled dog serum.

Lineage-negative (lin⁻) bone marrow cells were obtained by depletion of mononuclear marrow cells of granulocytes and monocytes by Dog 15-7 Ab (rat-anti-dog IgG 2a), mature T cells by Dog 17-4-8 Ab (rat anti-dog IgG 2a), and mature B cells by Dog 22-2 Ab (rat anti-dog IgG 2a) (13). Cells (1 × 10⁸) were incubated with 1000 μl of each Ab culture supernatant for 30 min on ice. Cells were washed twice with PBS supplemented with 0.5% BSA and incubated with magnetic goat anti-rat Dynabeads M-450 (Dynal, Hamburg, Germany) at a cell to bead ratio of 1:4 under permanent shaking for 30 min at 4°C. The positive cells were removed twice using the magnetic particle collector (Dynal). The supernatant containing the lineage-negative bone marrow fraction was aspirated. The cells were cultured in IMDM (Seromed) supplemented with 5% inactivated pooled normal dog serum, 1% l-glutamine (Life Technologies, Eggenstein, Germany), 1% penicillin/streptomycin (Life Technologies), and 1% sodium pyruvate (Seromed) at 1.5 × l0⁶ cells/ml in six-well plates (Corning Costar, Bodenheim, Germany) in a volume of 3 ml or in cell culture flasks (Greiner, Frickenhausen, Germany) in a volume of 10 ml.

For differentiation of lin⁻ bone marrow cells into DCs the following cytokines were tested: human recombinant GM-CSF (hu-rec GM-CSF; Novartis Pharmaceutical, Nurnberg, Germany) at 300 or 800 U/ml, hu-rec IL-4 at 50–800 U/ml, hu-rec IL-1β (both from Genzyme, Russelsheim, Germany) at 10 U/ml, and hu-rec TNF-α (ICC Chemikalien, Ismaning, Germany) at 50–1000 U/ml. Every second day of culture half the medium was exchanged, and cytokines were added according to the initial activity. For the uptake, processing, and presentation of foreign Ag to autologous T cells in vitro-generated canine DC were pulsed with 25 μg/ml keyhole limpet hemocyanin (KLH; Sigma-Aldrich) for 2 days.

The effects of LPS (Sigma-Aldrich, Taufkirchen, Germany), poly I:C (Sigma-Aldrich), and CpG oligonucleotides (Metabion, Martinsried, Germany) on the maturation of DCs was analyzed. Cells were grown as previously described, maturated with 200 ng/ml LPS, 12.5 μg/ml poly I:C, or 1 μg/ml CpG for 48 h before pulsing the cells with KLH.

In the course of investigation we could not find any effect of hu-rec IL-4, hu-rec IL-1β, and hu-rec TNF-α on the generation of DCs. For the generation of mHA-specific CTL we used exclusively DCs generated with GM-CSF (800 U/ml) for 10–12 days.

For morphological characterization, cytospins (Shandon, Frankfurt, Germany) of DCs were stained with Wright’s solution (Merck, Darmstadt, Germany). Immunocytochemistry was performed by incubating cytospins with dog-specific Abs (13), followed by alkaline phosphatase-conjugated streptavidin and Fast Red (Multilink detection kit; Biogenex, Heidelberg, Germany) according to the manufacturer’s specifications. Slides were counterstained with hematoxylin for 40 s.

For FACS of surface markers, DCs were washed and incubated for 30 min with anti-dog Abs specific for CD1a (CA9.AG5) and MHC class II (dog 26-1) (13), followed by FITC- or PE-labeled goat anti-rat Abs as secondary Ab. Monocytes were stained with anti-human CD14 (Dianova, Hamburg, Germany), which was reported to be cross-reactive with the dog-CD14 Ag (14). Analysis was performed on a FACScan (BD Biosciences, Heidelberg, Germany). T cells and CTL were stained for canine CD3 (CA17-2A12), CD4 (CA13-1), and CD8 (dog 10-8) surface Ags according to the procedure described above.

In vitro-generated DCs were harvested, washed, irradiated with 20 Gy, and seeded in triplicate into 96-well, flat-bottom plates (Greiner, Frickenhausen, Germany) in serially diluted concentrations. Monocytes were enriched by plastic adherence. For this, 1 × 10⁶/ml PBMC were seeded into plastic dishes and incubated at 38°C for 2 h. Nonadherent cells were removed, and adherent monocytes were harvested. Allogeneic or autologous T cells were separated over nylon wool (15) and added at a final concentration of 2 × 10⁵/well. As positive control for T cell proliferation, triplicate cultures of T cells alone were incubated with 4 μg/ml PHA-M (Sigma-Aldrich, Taufkirchen, Germany) and 40 U/ml hu-rec IL-2 (Genzyme); as negative controls, triplicate cultures with T cells or DCs alone were included. On day 5 cells were pulsed with 1 μCi/well [³H]thymidine (Amersham Pharmacia Biotech, Braunschweig, Germany) for 18 h and harvested onto UniFilter Plate-96 (Canberra-Packard, Dreieich, Germany). [³H]thymidine incorporation was measured on a Top-Counter (Packard, Downers Grove, IL).

For the generation of mHA-specific CTL, only DCs generated by incubation with 800 U/ml GM-CSF were used. PBMC (1 × 10⁷) from the donor were incubated with 2 × 10⁶ irradiated DCs of the recipient at 38°C in a humidified atmosphere of 5% CO₂ in air. After 8 and 12 days additional 2 × 10⁶ irradiated DCs were added to the culture. After 16 days CTL were harvested and tested for their ability to suppress the growth of hemopoietic progenitor cells of the host.

One dog was immunized with DCs from a DLA-identical littermate. The first dose of 1 × 10⁸ DCs was divided into four portions and applied s.c. to the four limbs, followed by two additional injections with 5 × 10⁷ cells each on days 14 and 21, respectively. Twenty-eight days after the first injection T lymphocytes from the blood were harvested and studied for the HPC growth suppression in the Δ-Assay.

CTL produced in vitro were tested for the suppression of CFU in soft agar culture. The lin⁻ marrow cells (target cells; 2 × 10⁵) were mixed with various numbers of irradiated CTL (effector cells), obtaining an E:T cell ratio of up to 5:1. Effector and target cells were resuspended in 1.8 ml of McCoy’s medium supplemented with sodium bicarbonate, sodium pyruvate, Eagle’s MEM vitamins, Eagle’s MEM amino acids, MEM nonessential amino acids, l-glutamine, l-serine, l-asparagine, and penicillin/streptomycin (all from Life Technologies). Recombinant growth factors were added for colony stimulation: 20 ng/ml canine stem cell factor (Amgen, Thousand Oaks, CA), 800 U/ml hu rec GM-CSF (Novartis Pharmaceutical), 1 U/ml hu rec erythropoietin (Roche, Mannheim, Germany), and 3% pooled allogeneic MLR supernatant. The cytotoxic reaction was performed in suspension for 4 h at 38°C. Cells were mixed with 200 μl of melted 3% Bacto agar (Difco, Detroit, MI) and plated into six-well culture plates. Colonies were counted after 14 days of culture at 38°C in a humidified atmosphere. The CFU suppression was calculated by comparison of colony numbers obtained at different E:T cell ratios with colonies grown in the absence of specific CTL.

HPC suspension culture in the presence of cytokines allowed measurement of growth inhibition in a Δ-Assay. The technique was adapted from the Δ culture (16) and the CML progenitor cell proliferation assay (17) using lineage-negative bone marrow as target cells. The target cells were plated in triplicate in a 96-well plate at a serial 2-fold dilution, resulting in cell numbers ranging from 16,000 to 125 cells/well in IMDM supplemented with 20% FCS, 1% l-glutamine, 1% penicillin/streptomycin, and 1% sodium pyruvate (all from Life Technologies). The growth of hemopoietic progenitor cells was stimulated by the addition of 100 ng/ml stem cell factor (Amgen), 800 U/ml hu rec GM-CSF (Novartis Pharmaceutical), 2 U/ml hu rec erythropoietin (Roche), 100 ng/ml hu rec IL-3 (Novartis Pharmaceutical), and 3% pooled allogeneic MLR supernatant. CTL were harvested, washed, and irradiated with 2000 cGy. At the start of the Δ culture irradiated CTL were added to the wells in a fixed cell number of 10,000 cells/well, resulting in E:T cell ratios between 0.6:1 and 80:1. After 4 days HPC were pulsed with 1 μCi/well [³H]thymidine (Amersham Pharmacia Biotech) for 18 h and harvested onto UniFilterPlate-96 (Canberra-Packard, Dreieich, Germany). The [³H]thymidine incorporation was measured on a Top-Counter (Packard). The counts per minute were plotted against the starting number of target cells and compared with a control without addition of CTL. Statistical analysis of the results was performed with t test.

After immunomagnetic depletion of differentiated bone marrow cells we recovered 17 ± 10% of the initial population depending on contamination of the aspirate with peripheral blood (data from 12 experiments). The lin⁻ bone marrow cells were of small size and low granularity (compare with Fig. 3,A). The culture conditions for the production of canine DCs from lin⁻ bone marrow cells were studied using different media (RPMI 1640 and IMDM), sera (FCS and pooled canine serum (CS)) and cytokines (GM-CSF, TNF-α, IL-1β, and IL-4) at various concentrations and culture times. DCs were analyzed for the expression of surface markers CD1a and MHC class II and were tested for their stimulatory potency in an allogeneic MLR. As shown in Fig. 1,A, IMDM supplemented with 5% CS or 10% FCS and 800 U/ml GM-CSF supported optimal growth of allostimulatory DCs. Lower concentrations of GM-CSF resulted in a decreased allostimulatory capacity (Fig. 1,B). Addition of IL-4, TNF-α, or IL-1β did not improve the allostimulatory activity of DCs (data not shown) or the total number of DCs. Comparing culture for 8, 12, and 19 days, we determined the optimal culture time for the generation of DCs. After 8 or 12 days DCs were highly stimulatory in allogeneic MLR. After 19 days of culture the allostimulatory activity of DCs was greatly decreased (Fig. 1,C). Although DCs generated in medium supplemented with 10% FCS showed the same stimulatory activity as DCs grown in medium with CS (compare with Fig. 1 A), we decided to use CS to avoid the presentation of FCS as foreign Ag.

Beginning with day 3 the cells grew in small aggregates that were mainly nonadherent or only loosely adherent to the plastic surface of the culture dish. A certain percentage of adherent cells appeared with macrophage- and fibroblast-like morphology. Further on, veiled processes appeared at the edge of growing cellular aggregates, and floating single cells acquired DC-like morphology. Subsequently, under the influence of GM-CSF, granulocytes and macrophages also grew out. After 8 days nonadherent cells, which represented 164 ± 102% of the seeded cells, were harvested. All firmly attached cells were left on the plastic surface. Interestingly, after 19 days of culture the number decreased to 100% of seeded cells, mainly by the disappearance of differentiated cells such as granulocytes. Of cells harvested on day 8, 61 ± 17% acquired a dendritic-like phenotype according to morphological and surface staining criteria.

The effects of LPS, polyriboinosinic polyribocytidylic acid (poly(I:C)) and CpG oligonucleotides on the maturation of canine DCs was analyzed. Lacking the necessary Abs to identify mature DCs in the dog, the maturation state was assessed using the presentation of foreign Ag (KLH) to autologous T cells in an MLR. LPS is capable of maturation of DCs in vitro (18). Maturation of DCs for 48 h with LPS resulted in decreased uptake and presentation of KLH to autologous T cells. The [³H]thymidine incorporation of autologous T cells after stimulation with LPS-treated DCs was only 39.2 ± 21.7% compared with that after stimulation with untreated DCs. Synthetic oligonucleotides expressing CpG motifs were reported to boost maturation of DCs in mice and humans (19, 20). In our animal model the incubation of DCs with CpG oligonucleotides had no effect on the uptake and presentation of KLH to autologous T cells. Finally, maturation of DCs was tried using poly(I:C), which was reported to induce maturation in human DCs (21). We observed decreased uptake and presentation of KLH by poly(I:C)-treated DCs, resulting in decreased stimulation of autologous T cells. The [³H]thymidine incorporation after stimulation with poly(I:C)-treated DCs was only 50.4 ± 32.9% compared with that after stimulation with untreated DCs.

According to the morphological criteria described for human DCs (22, 23), canine DCs were large and showed a typical appearance of the nucleus and characteristic veiled cytoplasmatic processes (Fig. 2,A). Immunocytochemical staining of canine DCs showed strong expression of MHC class II (Fig. 2,B) and CD1a (Fig. 2,C) surface Ags, but no staining with an isotype control (Fig. 2 D).

Canine DCs were further analyzed on FACScan after immunofluorescent staining with dog-specific Abs. After 10 days of culture the forward and side scatter characteristics showed a nonlymphoid population of large size and dense granularity comprising 80–85% of all cells (Fig. 3,B). This population stained brightly for MHC class II and CD1a (Fig. 3, C and D, respectively), and it was negative for the monocyte Ag CD14 (Fig. 3 E). Different Abs detecting human costimulatory molecules; for example, B7.l, B7.2, or CD40 were not cross-reactive for canine DCs (data not shown).

DCs are the major stimulatory component in an allogeneic MLR. One of the main criteria in the functional assessment of DCs is their ability to prime and stimulate the growth of allogeneic T cells. Our T cell preparations (nylon wool enrichment) reached a purity of ∼95%, as analyzed by the expression of CD3 surface Ag in immunofluorescent staining (data not shown). The allogeneic stimulation of T lymphocytes with DCs in a DLA-mismatched situation was compared with the stimulation with PBMC (Fig. 4 A). Canine DCs showed up to a 100-fold stronger stimulation of allogeneic T cells than PBMC derived from the same animal.

A characteristic capacity of DCs is their ability to take up, process, and present a self or foreign Ag to autologous T cells (23). To test the ability of DCs to present a foreign Ag, we used KLH. The animals were never immunized with or been in contact with this protein. DCs grown under our conditions presented KLH to autologous T cells, resulting in a high stimulation in autologous MLR (Fig. 4,B). In contrast, autologous DCs not loaded with KLH failed to stimulate autologous T cells. PBMC and enriched monocytes pulsed with KLH were not able to stimulate autologous T cells (Fig. 4 B). The absence of mitogenic activity in the KLH preparation has been demonstrated by incubation of T cells alone with 25 μg/ml KLH, resulting in the absence of T cell proliferation (data not shown). Similar results were also obtained using tetanus toxin instead of KLH as a foreign Ag (data not shown).

The Δ-Assay was established in DLA-mismatched combinations with E:T cell ratios between 0.6:1 and 80:1. The results showed that even at the lowest E:T cell ratio, the suppression of [³H]thymidine uptake by HPC was nearly complete (Fig. 5,A). At an E:T cell ratio of 5:1, HPC growth inhibition was >90% in all analyzed DLA-mismatched combinations (12 of 12; Table I). CTL did not suppress either the growth of autologous HPC (Fig. 5B) or that of HPC from DLA-mismatched dogs not sharing DLA with the stimulator dog.

Immunization of T lymphocytes in vitro against mHA required DCs of DLA-identical littermates and restimulation on days 8 and 12. The mHA-specific CTL suppressed the growth of stimulator-derived HPC in 53% (8 of 15) of DLA-identical littermate combinations analyzed. In 13% (2 of 15) we observed an almost complete inhibition of HPC growth (inhibition >90%) at an E:T cell ratio of 5:1, comparable to that with DLA-mismatched combinations. In 40% (6 of 15) the inhibition of HPC growth was intermediate (inhibition ranging from 31 to 73%) at an E:T cell ratio of 5:1. Inhibition of HPC growth was not detectable (<10% compared with the control) in 47% of combinations (7 of 15; Table II). Generation of mHA-specific CTL was not successful using PBMC instead of DCs as APC (data not shown).

For phenotypic analysis of mHA-specific CTL, selected samples were stained for CD4 and CD8 surface Ags. Effector cells were mostly CD8 positive (84.4 ± 7.1%) compared with CD4 (9.8 ± 9.7%). Effector cells taken from nonreactive littermate combinations showed no clear distinction between CD4 and CD8 cells (CD4, 43.8 ± 19.5%; CD8, 50.4 ± 37.1%).

In some cases mutual stimulation of DLA-identical littermates resulted in HPC growth suppression in only one direction. One-way reactivity of female CTL against male HPC was observed in two DLA-identical littermate combinations of the same family (Table II, dogs 1465 × 1461 and 1466 × 1462). This reactivity suggested the involvement of a Y-associated mHA. We also observed one-way reactivity between cells of sex-identical littermates (Table II, dogs 483 × 486, 486 × 483, and 309–311).

The Δ Assay correlated very well with the inhibition of the growth of CFU by mHA-specific CTL in a soft agar culture (Table III). In four combinations an E:T cell ratio of 5:1 was tested. Suppression of HPC growth correlated with suppression of CFUs for both DLA-identical and DLA-mismatched combinations.

We immunized a dog in vivo with DCs from its DLA-identical littermate. After 4 wk the ability of freshly aspirated PBMC from this animal to suppress the growth of HPC from the DC donor was analyzed in a Δ Assay. We found an intermediate suppression of HPC growth of 62% compared with the control at an E:T cell ratio of 5:1 (Fig. 6). This was in the same intermediate range of HPC growth suppression as that observed after in vitro immunization of the same littermate combination (in vivo, 62% inhibition; in vitro, 44% inhibition).

The segregation of mHA was studied in two canine families. The mHA-specific CTL were produced by stimulation of DLA-identical littermates. These cells were tested against HPC from all littermates and the parents if available. In the families analyzed, mHA-specific T cells recognized two and three of five family members (Table IV, Family A) and three of four family members (Table IV, Family B). DLA restriction elements could be either DLA-A2 alone or DLA-A2 and B4 (Table IV), but further tests in unrelated situations are necessary to clearly identify the DLA restriction elements involved.

Adoptive immunotherapy with DLT after allogeneic stem cell transplantation provides a potent strategy to treat hematological malignancies (5). However, the use of DLT is limited by the occurrence of GvHD and the poor response of acute leukemia (3). A possible improvement in DLT is the sensitization of donor T lymphocytes against Ags presented by leukemia cells. These Ags could be leukemia-specific proteins resulting from chromosomal recombinations and mutations (24, 25), overexpressed normal proteins (26, 27), or mHA exclusively expressed on hemopoietic cells (28, 29). In a murine model adoptive transfer of T lymphocytes primed against a single synthetic mHA peptide resulted in the eradication of leukemic cells without causing GvHD (30). In humans the minor Ag HA1 is regularly expressed predominantly on hemopoietic cells (28). It appears feasible to use ex vivo-generated CTL against HA1 for the treatment of recurrent leukemia. Following adoptive immunotherapy hemopoiesis of the host, including residual leukemic cells, is completely replaced by the graft (8). To date several mHA with restricted tissue distribution have been defined using hemopoiesis-specific CTL from transplanted patients (29). Such CTL directed against mHA on hemopoietic cells have been shown to suppress the growth of human leukemia in NOD/SCID mice (31), but clinical data are still missing.

The dog has often served as a preclinical model for transplantation studies, including adoptive immunotherapy with donor lymphocytes (7). The conversion of a mixed hematological chimerism into a complete donor chimerism is used as a model for leukemia. In the present study we demonstrated cellular immunity to hemopoietic progenitor cells in vitro. Previously we have shown that immunization of the donor against the host enhanced the reaction of donor lymphocytes against hemopoietic cells of the host, resulting in a rapid conversion of mixed to full donor chimerism (7, 32). This conversion was much faster than the gradual conversion observed in dogs receiving DLT from unsensitized donors.

Despite the strong effect of immunized cells in vivo on chimerism, their activity could not be easily demonstrated in vitro. The chromium release assay was not sensitive enough to detect immunity against mHA. Ideally the growth of CFUs should be suppressed by mHA-specific CTL. However, the technique of suppression of CFU growth is very laborious and time consuming. Hemopoietic growth factors have become readily available in a recombinant form; they can be used for the stimulation of HPC growth in suspension culture (16). The so-called Δ-Assay allows the measurement of HPC growth by the incorporation of tritiated thymidine. The same method has been used by Smit et al. (17) to detect reactions against leukemic progenitor cells in patients with chronic myelogenous leukemia. In the dog HPC were enriched by the depletion of lineage-positive cells using Abs to T cells, B cells, monocytes, and granulocytes (13). Optimal growth of HPC was obtained by the addition of recombinant cytokines. The addition of MLR supernatant was helpful to calibrate the potentially stimulatory effect of T cells by secreted cytokines. In all DLA-mismatched combinations the suppression of thymidine uptake was observed as well as in 53% of DLA-identical combinations. The suppression of thymidine uptake correlated well with the suppression of CFU growth in selected samples. The Δ Assay for detection of CTL against mHA worked well in the dog model, and it showed homology to similar assays in human patients with chronic myelogenous leukemia (17) and possibly also acute myeloid leukemia (our unpublished observations).

CTL have been produced against viral Ags and tumor Ags by culturing T cells from donors immune to these Ags. From nonimmune donors cytotoxic T cell responses against mHA could be elicited in vitro using DCs (8). CTL responses could be generated by native DCs differing in mHA from the responder as well as autologous DCs pulsed with an immunogenic peptide. In this study we show the generation of CTL against hemopoietic progenitor cells using allogeneic DCs in the dog.

The optimal conditions for the generation of canine DCs and mHA-specific CTL are not known. Human DCs can be generated from CD34⁺ bone marrow progenitor cells or PBMC (23, 33, 34, 35, 36). The differentiation into this lineage is primarily regulated by GM-CSF and is supported by the activity of TNF-α and IL-4, pushing the cells down along the myeloid pathway (23, 34, 37, 38, 39). For human bone marrow-derived DCs, the cytokine requirements and conditions for in vitro cultivation have been worked out (40, 41). The inclusion of stem cell factor (35, 36, 38, 42) and/or Flt-3 ligand (43) in combination with GM-CSF and TNF-α enhances DC outgrowth from human bone marrow, providing a method for ex vivo generation of large numbers of DCs for clinical purposes (44).

Recently, Hagglund et al. (45) described the in vitro generation of canine DCs with allostimulatory activity and DC-like phenotype from CD34⁺ bone marrow progenitor cells using human GM-CSF, human Flt3 ligand, and human TNF-α. In our study we observed no activity of human TNF-α on the outgrowth of DCs, and human Flt3 ligand was only weakly cross-reactive with canine cells. Moreover, we found that canine DCs did not respond to human IL-4. We chose to generate canine DCs from lineage-negative marrow, including not only CD34⁺ stem cells but also more differentiated progenitors of the myeloid pathway, using GM-CSF. Canine DCs were produced by our method, which showed morphology, immune phenotype, and functional characteristics comparable to those of human DCs. DCs were treated with LPS, poly I:C, and CpG oligonucleotides, reagents reported to enhance DC maturation in vitro (18, 19, 20, 21, 46). The strong capability of our DCs for Ag uptake and processing suggested an immature phenotype. Unfortunately, no Abs for costimulatory molecules were available for the dog, so we could only rely on Ag presentation to autologous T cells for DC characterization. DCs incubated with LPS and poly(I:C) showed a reduction in Ag uptake, suggesting a more mature type. On the other hand, incubation of DCs with CpG oligonucleotides had no effect on Ag presentation. We found no influence of matured DCs on the production of mHA-specific CTL, so we generally used DCs not matured with LPS or poly(I:C). Most likely, DC maturation was induced after incubation with allogeneic T cells.

We immunized fresh PBMC with DCs from a DLA-identical littermate. DLA type was determined by serotyping with antisera and MLR as well as genotypically on the basis of highly polymorphic microsatellite markers (12). To exclude minute differences in the serotypes, all animals were the offspring of parents with four defined DLA haplotypes and were taken from the same litter. The nature of mHA was demonstrated by the allogeneic reaction between littermates with identical DLA type and the inheritance of the marker and its restriction by DLA Ags, most likely DLA-A2 in the families studied here (Table IV).

In the majority of cases (53%) we observed a suppression of HPC growth with immunized CTL from DLA-identical littermates. In these animals growth suppression of HPC was due to differences in mHA. Phenotypic analysis showed that growth suppression was mainly mediated by CD8⁺ cells, indicating the presence of MHC class I-restricted Ags. This is in conjunction with known mHA in humans and mice. In ∼47% no suppression of HPC growth could be detected. This was due to an absence of mHA-specific CTL, resulting in a mixed population of unspecific cells, consisting of CD4⁺ and CD8⁺ cells in roughly equal proportions. In these littermate combinations there was no sufficient difference in immunogenic mHA. It is known that in the mouse immune responses are only initiated by a small number of highly immunogenic mHA (47, 48, 49, 50, 51). In DLA-identical siblings hemopoietic transplants of bone marrow and PBL produced GvHD in only 50–70% of the dogs (52). In dogs there may be a tightly restricted mHA polymorphism due to breeding selection. Alternatively, only a limited number of mHA loci may be eliciting strong immune reactions detectable by in vitro screening systems. These differences could also explain the varying levels of HPC growth inhibition we observed in our experiments.

Of special interest were some pairs of DLA-identical littermates showing one-way reactivity of mHA-specific CTL. These could be sorted into two categories: female cells reacting against male cells or cells from the same gender. Female cells involved in a one-way reaction most strongly inhibited the growth of male cells. As in other species (53, 54, 55) this hints at the involvement of a Y chromosome-encoded mHA, especially since both cases were observed in the same family. However, further studies at the molecular level are necessary to prove the existence of a Y chromosome-located mHA in the dog. Y chromosome-associated mHA are normally not restricted to cells of hemopoietic origin (56), but the level of expression may differ in various tissues. Adoptive immunotherapy with a Y chromosome-encoded mHA would be feasible in certain circumstances.

One-way reactivity between sex-matched animals could be attributed to either a homo/heterozygous situation or the involvement of an immunodominant allele of an mHA, with the other allele not presented. An example of such a case is found in human HA-1. Only the H allele is functioning as a T cell epitope and is recognized by CTL, whereas the R allele of HA-1 is not presented to T cells (57). A third case of one-way reactivity involving three DLA-identical female animals (Table II, dogs 309, 310, and 311) indicates either the unlikely possibility of a mHA with more than two alleles or the involvement of at least two different minor Ags.

We have shown the production of mHA-specific CTL in vitro. To compare the efficacy of these cells we also immunized one animal in vivo with DCs from its DLA-identical littermate. The HPC growth suppression induced by mHA-specific CTL after in vitro immunization was comparable to the HPC growth suppression with T cells after in vivo immunization of a donor (Fig. 6). Additional experiments will be directed at comparing the capabilities of ex vivo-generated CTL with in vivo-immunized T cells in a transplantation setting.

The exclusive suppression of hemopoietic cells by mHA-specific CTL was not shown in our study. The mHA not restricted to hemopoietic systems would have yielded the same results in our analysis. In following studies we want to identify reactions against nonhemopoietic cells, such as fibroblasts or keratinocytes, in a modified Δ-Assay. This would give a hint of whether the mHA we found are restricted to the hemopoietic system and would lead only to a GvL reaction after adoptive immunotherapy or if the mHA are expressed ubiquitously and would lead to severe GvHD after transfusion of CTL. Another possibility for the identification of tissue distribution would be to test mHA-specific CTL in a skin explant model (58). In any case only transplantation and adoptive immunotherapy will give answers regarding GvHD and conversion of chimerism after transfusion of mHA-specific CTL obtained from immunized donors or generated in vitro using host DCs.

After establishing the tools for the generation of mHA-specific CTL in the dog, we are now prepared to use our animal model as an in vivo model for the treatment of leukemic relapse with mHA-specific CTL. In >50% of the DLA-identical animals we were able to induce a reaction against mHA regardless of the DLA type of the animal. This is a very promising starting point for exploitation of the preclinical canine model of mixed chimerism for treatment of leukemia in humans.

We thank A. Arco-Zinneberg, G. Werner, and M. Hagemann for technical assistance.