We developed a nonmyeloablative host conditioning regimen in a mouse model of MHC-mismatched bone marrow transplantation that not only reduces radiation toxicity, but also protects against graft-vs-host disease. The regimen of fractionated irradiation directed to the lymphoid tissues and depletive anti-T cell Abs results in a marked change in the residual host T cells, such that NK1.1+ or DX5+asialo-GM1+ T cells become the predominant T cell subset in the lymphoid tissues of C57BL/6 and BALB/c mice, respectively. The latter “natural suppressor” T cells protect hosts from graft-vs-host disease after the infusion of allogeneic bone marrow and peripheral blood cells that ordinarily kill hosts conditioned with sublethal or lethal total body irradiation. Protected hosts become stable mixed chimeras, but fail to show the early expansion and infiltration of donor T cells in the gut, liver, and blood associated with host tissue injury. Cytokine secretion and adoptive transfer studies using wild-type and IL-4−/− mice showed that protection afforded by NK1.1+ and DX5+asialo-GM1+ T cells derived from either donors or hosts conditioned with lymphoid irradiation is dependent on their secretion of high levels of IL-4.
Recent advances in clinical allogeneic bone marrow (BM)3 transplantation have reduced the toxicity of host preparatory regimens such that they are nonmyeloablative and yet permit the development of chimerism (1, 2, 3, 4). Once mixed chimerism is established, the development of complete chimerism, which is important in the treatment of hemologic malignancies, can occur spontaneously or after donor blood lymphocyte infusions (1, 2, 3, 4). A major complication of BM or hemopoietic progenitor cell transplantation even with nonmyeloablative preparatory regimens is the high incidence of severe graft-vs-host disease (GVHD) caused by T lymphocytes in the injected donor inoculum, especially in MHC-mismatched combinations (3, 5, 6). Depletion of donor T lymphocytes ameliorates GVHD, but can result in an increase in failure of engraftment, tumor relapse, and prolonged immunodeficiency (7, 8, 9, 10, 11).
An alternative approach to prevent GVHD is to use a nonmyeloablative host conditioning regimen that protects the host against donor T cell attack. This has been achieved by altering the APC of the host (12) or by using fractionated irradiation of the lymphoid tissues (total lymphoid irradiation, TLI) that induces regulatory “natural suppressor” T cells in the host (13, 14). However, it is not clear whether these host conditioning regimens can protect against severe GVHD induced by peripheral blood T cells, because in the case of human BM or mobilized hemopoietic progenitor transplantation there is a high level of contamination with these blood T cells (15, 16).
Our recent studies showed that allogeneic donor NK1.1+ TCRαβ+ T cells in the BM can protect mice against lethal GVHD induced by donor NK1.1− TCRαβ+ T cells (17). Protection was dependent upon secretion of IL-4 by NK1.1+ TCR αβ+ T cells (17). The latter T cell subset differs from conventional T cells in several ways, including the extremely high levels of secretion of IL-4 and IFN-γ, as well as the recognition of the CD1 Ag-presenting molecule, a class I MHC-like molecule that is nonpolymorphic (18, 19). In addition, depletion of peripheral NK1.1+ TCRαβ+ T cells is followed by rapid reconstitution within 48 h from rapidly dividing progenitors in the BM, whereas depletion of conventional NK1.1− TCRαβ+ T cells involves a slower reconstitution via the thymus (20).
In the current study we developed a nonmyeloablative conditioning regimen that severely depletes peripheral T cells and thymocytes using fractionated lymphoid irradiation with marrow shielding in combination with anti-mouse thymocyte serum (ATS). After the regimen, the splenic NK1.1+ TCRαβ+ T cells of C57BL/6 mice increased from ∼2% of all TCRαβ+ T cells to >90% in association with a marked shift in the cytokine secretion pattern favoring IL-4. A similar observation was made in BALB/c mice using the DX5 marker reported previously to be coexpressed on NK1.1+ TCRαβ+ T cells, but not on conventional T cells (21). The irradiated BALB/c host mice with the predominant DX5+ TCRαβ+ T cells were protected from acute lethal GVHD induced by the injection of combined allogeneic BM and peripheral blood T cell transplants from untreated C57BL/6 donors. Depletion of DX5+TCRαβ+ T cells occurred after treatment of hosts with anti-asialo-GM1 Abs, a regimen previously shown to remove the protection afforded by fractionated lymphoid irradiation (22). Protection was also dependent on BALB/c host cell secretion of IL-4. Adoptive transfer of splenic T cells with the predominant NK1.1+ TCRαβ+ T cell subset from C57BL/6 donor mice conditioned with lymphoid irradiation also protected against transplants from untreated donors.
Materials and Methods
Male C57BL/6 CD45.2(H-2b) and BALB/c (H-2d) CD45.2 wild-type mice, 8–10 wk old, were purchased from the Department of Comparative Medicine, Stanford University (Stanford, CA). Male C57BL/6 IL-4−/− mice (C57BL/6J-IL4tm1Cgn) and male BALB/c IL-4−/− (BALB/cIL4tm2Nnt) were purchased from The Jackson Laboratory (Bar Harbor, ME). Male C57BL/6 CD45.1 mice were obtained from the colony of Dr. I. L. Weissman (Stanford University).
TLI was delivered to the abdomen, lymph nodes, thymus, and spleen with shielding of the skull, lungs, pelvis, and tail as described previously (13, 14). Irradiation was started on day −24 before transplantation, and 17 doses of 240 cGy each were administered. The last dose of TLI or a single dose of total body irradiation (TBI) was administered to BALB/c mice 24 h before cell infusions (13, 14). The irradiation was performed with a Philips x-ray unit (200 kV, 10 mA; Philips Electronic Instruments, Rahway, NJ) at a rate of 84 cGy/min with a 0.5-mm Cu filter.
Rabbit ATS and rabbit anti-asialo-GM1 antiserum
Rabbit ATS was purchased from Accurate Laboratories (New York, NY). Mice were injected i.p. with 0.05 ml ATS in 0.5 ml saline on days −12, −10, and −8 before BM transplantation, corresponding to days 12, 14, and 16 after starting TLI. Rabbit anti-asialo-GM1 antiserum was purchased from Wako Chemicals (Richmond, VA). In some experiments 20 μl antiserum was injected i.p. immediately after TLI and then again 24 h later.
PBMC were isolated on density gradients (Lymphocyte-M; Cedarlane Laboratories, Hornby, Ontario, Canada) and washed twice in ice-cold RPMI 1640 (Life Technologies, Grand Island, NY). Femoral and tibial bones taken from donor C57BL/6 mice were rinsed, and the residual muscle on the bones was carefully removed. BM cells were prepared by flushing the bones with RPMI 1640, and the cell suspension was filtered through nylon mesh to remove aggregates and washed once before transfusion. Mouse T cell-depleted BM cells and purified CD4+ and CD8+TCRαβ+ T cells from the spleen were sorted on a FACStar flow cytometer as described previously (17). Sorted spleen T cells obtained after TLI were first enriched on immunomagnetic bead columns (Miltenyi Biotech, Friedrich, Germany) using biotinylated anti-Thy1.2 mAb (5a-8; Caltag Laboratories, Burlingame, CA) and streptavidin-conjugated microbeads. Enriched Thy1.2+ cells were stained with allophycocyanin-conjugated anti-TCRαβ (H57-597; BD PharMingen, San Diego, CA) before sorting on a Vantage flow cytometer (BD Biosciences, Mountain View, CA) as described previously (17).
Blood samples for chimerism analyses were hemolyzed with ammonium chloride potassium carbonate to remove red cells. The white cell pellets were washed twice with 0.05% sodium azide staining buffer and incubated on ice for 15 min with saturating concentrations of mAb mixtures as described previously (17). Biotinylated anti-Gr-1 (RB6-8C5) and anti-Mac-1 (M1/70.15), and allophycocyanin-conjugated anti-B220 (RA3-6B2; Caltag Laboratories) as well as FITC-conjugated anti-H-2Kb (AF6-88.5), FITC-conjugated anti-CD45.1 (A20), and PE-conjugated anti-Thy1.2 (53-2.1) mAbs (BD PharMingen) were used for mouse chimerism analyses. After incubation, cells were washed twice and followed by streptavidin-Texas Red (Caltag) staining on ice. Background staining for donor-type cells in normal BALB/c mice was ≤0.5%. For NK1.1+ or DX5+ T cell staining from C57BL/6 and BALB/c mice, splenocytes were incubated with a mAb mixture with PE-conjugated anti-NK1.1 (PK136), PE-conjugated anti-DX5 (DX5), and allophycocyanin-conjugated anti-TCRαβ(H57-597) from BD PharMingen. All mouse cells were incubated with CD16/32(2.4G2) (BD PharMingen) to block the FcRγII/II receptors, and propidium iodide was added to exclude dead cells. Chimerism analysis used a lymphoid gate set by forward and orthogonal light scatter (17).
Details of in vitro stimulation of sorted cell populations with PMA and ionomycin, and analysis of IL-4 and IFN-γ in the 48-h supernatants by ELISAs were described previously (17).
Histopathology of liver, skin, and intestines
Tissues were fixed in formalin and embedded in paraffin blocks, and sections were stained with hematoxylin and eosin.
Liver and gut lymphocyte preparations
After hosts were exsanguinated, livers were flushed by injection of heparinized PBS in the right ventricle until the liver became pale. Livers were pressed through a nylon mesh to prepare single-cell suspensions in 2 μM EDTA in PBS. The cell suspension was centrifuged on Ficoll-Hypaque, and the interface layer was collected for lymphocyte staining. Gut, including duodenum through rectum, was collected, cut longitudinally, and rinsed thoroughly with cold RPMI 1640. The rinsed gut was put into EDTA in PBS, minced into <5-mm segments, and suspended by vortex 20–30 s, alternating with incubation at room temperature for ∼2–4 min. After three repeats, cells in the supernatant were filtered through a nylon mesh before separation on a Ficoll-Hypaque gradient for collection of mononuclear cells.
Predominance of NK1.1+ TCRαβ+ and DX5+ TCRαβ+ T cells after TLI
The spleen cells of C57BL/6 and BALB/c mice were examined for the percentage of NK1.1+ or DX5+ TCRαβ+ T cells, respectively, as well as the percentage of all TCRαβ+ T cells before and after fractionated irradiation of the lymphoid tissues (TLI). Fig. 1⇓A shows flow cytometric analyses of C57BL/6 spleen cells stained for the TCRαβ marker vs forward light scatter before and after 2, 8, or 17 irradiation treatments of 240 cGy each, targeted to the thymus, spleen, and lymph nodes while shielding the head, lungs, and hind limbs. In each case the last irradiation dose was given 1 day before lymphoid tissues were harvested. Whereas the TCRαβ+ T cells were 33.2% of live nucleated cells (enclosed in box) before irradiation, the percentage decreased to 25.6, 13.4, and 9.4% after 2, 8, and 17 treatments. Although the mean percentage of TCRαβ+ T cells in the spleen decreased ∼3-fold (30–9%) after 17 treatments, the mean absolute number decreased ∼150-fold (32–0.2 × 106; Table I⇓). Similar marked decreases in the percentages and absolute numbers of TCRαβ+ T cells in the spleen of BALB/c mice after 2, 8, or 17 treatments of irradiation were observed also (Fig. 1⇓B and Table I⇓).
Because a previous report indicated that NK1.1+ TCRαβ+ T cells in the spleen are rapidly reconstituted from BM sources after peripheral depletion (20), the percentage of NK1.1+ TCRαβ+ T cells among gated TCRαβ+ T cells was determined before and after lymphoid irradiation (with marrow shielding) of C57BL/6 mice (Fig. 1⇑A). Whereas NK1.1+ TCRαβ+ T cells accounted for only 1.3% of all TCRαβ+ T cells before irradiation, the percentage increased dramatically after 2, 8, and 17 irradiation treatments, such that 65.9% of T cells expressed the NK1.1 marker after 17 treatments (Fig. 1⇑A). As shown in Table I⇑, the increased mean percentage of NK1.1+ T cells among all T cells was associated with a 4-fold decrease in the mean absolute number of NK1.1+ T cells in the spleen. The mean percentage of NK1.1+ T cells among all T cells was calculated using the percentages shown in the boxes in the right column of Fig. 1⇑A for each animal. However, there was a 30-fold greater reduction of NK1.1− T cells compared with NK1.1+ T cells that resulted in a marked change in the balance of these T cell subsets in the spleen.
The lack of the NK1.1 marker in BALB/c mice (23) did not permit a similar analysis of T cell subsets before and after irradiation. However, we analyzed another marker, DX5, that has previously been reported to be expressed on most NK1.1+ T cells in the spleen (21). Fig. 1⇑A shows that a similar dramatic increase in the percentage of DX5+ TCRαβ+ T cells in the spleen of BALB/c mice occurred after 17 irradiation treatments (1.2–64.4%). Table I⇑ shows that the increased mean percentage of DX5+ T cells was associated with a 4-fold decrease in the absolute number of these cells in the spleen. A similar increase in the percentage of NK1.1+ and DX5+ T cells among all TCRαβ+ T cells was observed in the BM and peripheral blood (data not shown).
To determine whether further peripheral T cell depletion beyond that achieved with irradiation would further increase the percentage of NK1.1+ T cells in C57BL/6 mice, a group of mice was given three doses of a T cell-depleting rabbit ATS on days 12, 14, and 16, with the last of 17 doses of irradiation given on day 24 and the first on day 0. Mice given both irradiation and ATS had 92.4% NK1.1+ T cells among all T cells immediately after the completion of the combined regimen (Fig. 1⇑A). Similarly, BALB/c mice given both irradiation and ATS had 91.8% DX5+ T cells among all T cells (Fig. 1⇑B). The mean increased percentages of NK1.1+ and DX5+ T cells among all T cells after irradiation and ATS was due to the more complete depletion of NK1.1− and DX5− T cells (350- to 1000-fold reduction) compared with that of NK1.1+ and DX5+ T cells (7- to 15-fold reduction; Table I⇑).
In further phenotypic studies the expression of CD4 and CD8 markers on NK1.1+ T cells in the C57BL/6 spleen before and after irradiation was determined. Representative two-color flow cytometric analyses of NK1.1 vs CD4 or CD8 markers on gated NK1.1+ TCRαβ+ T cells before and after eight or 17 doses of irradiation are shown in Fig. 2⇓A. The CD4+NK1.1+ T cells remained the major subset and increased from 62 to 76% of all NK1.1+ T cells during the course of irradiation (Fig. 2⇓A). In contrast, the CD8+NK1.1+ T cells decreased from 14 to 4% during the same period (Fig. 2⇓A). The percentage of CD4−CD8−NK1.1+ T cells was determined by staining with anti-NK1.1 mAb vs combined anti-CD4 and CD8 mAbs. These double-negative NK1.1+ T cells decreased slightly from 24 to ∼20% after irradiation (Fig. 2⇓A).
A previous study showed that asialo-GM1+ cells mediated protection against GVHD in BALB/c hosts given TLI, because injection of anti-asialo-GM1 Abs into hosts before allogeneic BM transplantation markedly increased the incidence of severe lethal GVHD (22). Accordingly, we determined whether an injection of anti-asialo-GM1 Abs given immediately after 12 doses of TLI and then repeated 24 h later depleted DX5+TCRαβ+ T cells in the spleen of BALB/c mice. Fig. 2⇑B shows that after TLI the percentage of TCRαβ+ T cells was 3% (enclosed in box of left upper panel), and that ∼38% of the T cells were DX5+ (right upper panel). After two injections of anti-asialo-GM1 Abs, the TCRαβ+ T cells in the spleen of TLI-treated mice were ∼1% (left lower panel) and the discrete population of DX5+ cells among the gated TCRαβ+ T cells was no longer observed (right lower panel). The percentage of DX5+ cells among TCRαβ+ T cells was reduced ∼10-fold in the latter mice. Thus, anti-asialo-GM1 Abs effectively depleted DX5+TCRαβ+ T cells compared with DX5−TCRαβ+ T cells.
Changes in T cell secretion of cytokines after TLI
Previous studies showed that just after the completion of TLI, spleen T cells stimulated in vitro with anti-CD3 mAb secreted markedly increased levels of IL-4 and reduced levels of IFN-γ compared with T cells from unirradiated mice (24). In addition, alloreactive T cell clones obtained within 3 wk after TLI have been reported to be shifted toward a Th2 cytokine pattern (25). Because NK1.1+ T cells secrete very high levels of IL-4 (18, 19), we isolated the NK1.1− and NK1.1+ T cell from the C57BL/6 spleen after 17 treatments of TLI and compared their capacities to secrete IL-4 and IFN-γ after stimulation in vitro with PMA and calcium ionophore. The yield of sorted cells immediately after TLI was too low for analysis; therefore, mice were rested for 7 days before harvesting and sorting the spleen cells. At that time point, there is a marked recovery in the absolute number of NK1.1− TCRαβ+ T cells. Fig. 3⇓A shows the two-color profiles of the NK1.1+ and NK1.1− TCRαβ+ T cells at that time point, and boxes R1 and R2 enclose the two populations of sorted cells, respectively. After a 48-h stimulation, the mean concentration of IL-4 in the supernatants from the sorted NK1.1+ TCRαβ+ T cells was ∼2800 pg/ml, whereas the mean concentration in the supernatants from the NK1.1−TCRαβ+cells was ∼100 pg/ml (Fig. 3⇓B). Although the secretion of IL-4 was markedly increased in NK1.1+ compared with NK1.1− T cells, the secretion of IFN-γ was slightly decreased (mean, ∼1600–1250 pg/ml; Fig. 3⇓B).
To determine the effect of changes in the composition of NK1.1+ and DX5+ T cells on the pattern of cytokine secretion of all TCRαβ+ T cells after irradiation, sorted TCRαβ+ T cells were obtained from the spleen of C57BL/6 and BALB/c mice before and after eight doses of irradiation. At the latter point, the mean percentage of NK1.1+ T cells among all T cells was 27% (Table I⇑). Yields of sorted T cells immediately after eight doses of irradiation were sufficient for analysis without resting the mice. Fig. 3⇑C shows that the concentration of IL-4 from the sorted T cells from C57BL/6 mice before irradiation was ∼75 pg/ml, whereas after irradiation, the level increased ∼10-fold to 700 pg/ml. The concentration of IFN-γ using cells harvested before irradiation was ∼1200 pg/ml, and increased ∼2-fold to ∼2200 pg/ml after irradiation (Fig. 3⇑C).
A similar increase in the concentration of IL-4 was observed with sorted TCRαβ+ T cells from BALB/c mice before and after irradiation (Fig. 3⇑C). Whereas the mean level before irradiation was ∼200 pg/ml, after irradiation the mean level rose to ∼800 pg/ml. Although a rise of the IL-4 secretion was observed using C57BL/6 and BALB/c mice, the secretion of IFN-γ was considerably different between the strains. Before irradiation, the mean level of IFN-γ in the BALB/c mice (∼550 pg/ml) was about one-half that of the C57BL/6 mice. After irradiation the level in BALB/c mice did not change appreciably and was ∼4-fold below that of the irradiated C57BL/6 mice. This resulted in a reversal of the ratio of concentrations of IL-4:IFN-γ after irradiation in the two strains; in C57BL/6 the ratio was ∼1:4, whereas in BALB/c mice it was ∼1:0.75. Before irradiation of C57BL/6 and BALB/c mice the ratios were ∼1:16 and 1:3, respectively.
Protection against GVHD after TLI
NK1.1+ TCRαβ+ T cells from the BM of C57BL/6 donor mice inhibit the ability of donor NK1.1− TCRαβ+ T cells to induce severe GVHD in lethally irradiated BALB/c hosts (17). Inhibition of GVHD is dependent on the secretion of IL-4 by NK1.1+ TCRαβ+ T cells (17). We theorized that the predominance of DX5+ TCRαβ+ T cells and the increased secretion of IL-4 by all TCRαβ+ T cells in BALB/c mice conditioned with TLI would protect against GVHD induced by C57BL/6 T cells after allogeneic BM transplantation. Accordingly, BALB/c mice were conditioned with either single-dose lethal (800 cGy) TBI or 17 doses of TLI and given a single i.v. injection of C57BL/6 BM cells with or without C57BL/6 PBMC.
Fig. 4⇓A shows that the BALB/c mice given 800 cGy TBI and an i.v. injection of C57BL/6 BM cells (3 × 106) all survived at least 100 days without clinical signs of GVHD. As shown above, addition of 0.5 × 106 C57BL/6 PBMC to the marrow resulted in the death of all hosts by day 48 (Fig. 4⇓A). Three i.p. injections of ATS on days −12, −10, and −8 failed to protect the irradiated hosts, and all died by day 44 (Fig. 4⇓A). Control hosts given irradiation without cells all died by day 14 (Fig. 4⇓A). BALB/c hosts given sublethal (450 cGy) instead of lethal (800 cGy) TBI uniformly died after the infusion of the combined C57BL/6 marrow and PBMC transplantation with or without the addition of ATS (data not shown).
BALB/c hosts given TLI all survived for at least 100 days regardless of whether they received a subsequent infusion of C57BL/6 marrow cells 1 day after the last dose of TLI (Fig. 4⇑B). However, the addition of 0.5 × 106 C57BL/6 PBMC to the marrow cells resulted in the death of 60% of hosts by 100 days (Fig. 4⇑B). Nevertheless, survival of the latter hosts was significantly improved (p < 0.001, by log-rank test) compared with that of hosts given TBI and the same cell inoculum. Injection of ATS during the second week of TLI (days −12, −10, and −8) protected the hosts further such that the combined infusion of marrow and PBMC failed to kill any hosts during a 120-day observation period (Fig. 4⇑C). Control hosts given the same preparatory regimen without a cell infusion or with only a marrow cell infusion all survived at least 120 days also (Fig. 4⇑C). CD1−/− BALB/c hosts given the same preparatory regimen and C57BL/6 BM and PBMC had increased mortality, and 70% died by day 35 (data not shown).
Histopathological analyses of the tissues of six host wild-type mice that received either the TBI or TLI and ATS preparatory regimen and a combined infusion of BM cells and PBMC were performed when hosts developed clinical GVHD or when protected hosts were sacrificed >100 days after the cell infusions. In all cases microscopic analysis of the liver, small and large intestines, and skin was performed using hematoxylin and eosin staining to look for evidence of GVHD. All hosts given TBI developed typical microscopic changes in GVHD in the skin, including epidermal hyperplasia, a dermal inflammatory infiltrate, subepidermal blistering, and necrotic keratinocytes (Fig. 5⇓A). In the large intestine changes included crypt apoptosis and inflammation, and atrophy of mucin-containing glandular cells (Fig. 5⇓B). These changes were not observed after 100 days in six of six hosts given the TLI and ATS regimen as shown in representative tissue secretions from a host in Fig. 5⇓, E and F. The epidermis was one to two cells thick (Fig. 5⇓E), and plump mucin-containing cells lined the intestinal crypts (Fig. 5⇓F). The microscopic appearance of the skin and intestines of the latter hosts was similar to that of control hosts given the TLI and ATS regimen and injected with BM cells without PBMC (Fig. 5⇓, C and D). Examination of tissue sections from the liver and small intestines showed minimal abnormalities in hosts given either TBI or TLI and ATS and an injection of BM cells and PBMC (data not shown).
Chimerism in BALB/c hosts
To determine the extent of chimerism in long term surviving BALB/c hosts, PBMC were harvested at 100 days after the allogeneic cell infusion and stained for a donor-type surface marker using anti-H-2Kb mAb vs T cell (Thy-1.2), B cell (B220), and granulocyte (Gr-1) and macrophage (Mac-1) markers. Fig. 6⇓A shows the two-color flow cytometric analysis of PBMC from a host given TLI, ATS, and an infusion of only C57BL/6 BM cells. Mixed chimerism was observed for T cells, B cells, monocytes, and granulocytes, and donor cells from each lineage (right upper quadrants) accounted for 5.1, 36.5, and 8.6% of the PBMC, respectively. A similar pattern of mixed chimerism was observed in BALB/c hosts given TLI, ATS, and a combined injection of C57BL/6 marrow cells and PBMC (Fig. 6⇓B). Hosts conditioned with the myeloblative TBI regimen and given an injection of only C57BL/6 marrow cells were complete chimeras for all lineages tested (Fig. 6⇓C). Table II⇓ summarizes the mean percentages of donor-type cells within the PBMC harvested from the different groups of surviving hosts at 100 days. All hosts given TLI or TLI and ATS developed mixed chimerism after the injection of allogeneic BM cells with or without donor PBMC (Table II⇓). In contrast, all hosts given TBI and a marrow cell infusion were complete chimeras.
Early expansion of donor T cells
To determine the effect of the host conditioning regimens on the early expansion of donor T cells in the lymphoid tissues and on the early invasion of donor T cells into the nonlymphoid tissues, BALB/c hosts given either TLI and ATS or TBI and ATS, and an infusion of C57BL/6 combined BM (3 × 106) and spleen (6 × 106) cells were euthanized 12 days after the cell injection. Spleen cells were used as a source of peripheral T cells instead of PBMC to increase the number of peripheral T cells infused to ∼2 × 106/host. A variety of tissues were harvested from the hosts, and mononuclear cells were enriched from each tissue as described in Materials and Methods. Staining of donor T cells derived from the injected BM or spleen was distinguished using congenic C57BL/6 donor mice expressing either the CD45.1 marker (spleen cell donors) or the CD45.2 marker (BM donors). BALB/c hosts expressed the CD45.2 marker. Fig. 7⇓ shows the two-color flow cytometric analysis of the harvested mononuclear cells stained for TCRαβ vs the CD45.1 marker. Almost all donor cells in hosts conditioned with TLI or TBI and ATS were CD45.1 TCRαβ+ T cells (enclosed in boxes). Because both BM donors and hosts were CD45.2, the latter T cells were derived from the injected donor spleen cells in all host tissues tested, including peripheral blood, spleen, liver, gut, and BM. In all tissues the percentages of donor T cells among the mononuclear cells was markedly increased in hosts given TBI compared with those given TLI (Fig. 7⇓). This was particularly striking in the blood, liver, and gut, where the differences were ∼15-, 14-, and 40-fold, respectively.
Table III⇓ shows the mean percentages and absolute numbers of donor T cells derived from the injected C57BL/6 spleen cells in the mononuclear cells harvested from the different tissues on day 12. The differences in the absolute number of cells in the hosts conditioned with TLI or TBI reflect the differences in donor T cell expansion and invasion into the host tissues. The largest differences in absolute numbers were observed in the blood (∼300-fold), gut (∼200-fold), and liver (∼1500-fold; Table III⇓). Thus, the ability of the TLI and ATS regimen to protect against GVHD compared with TBI is reflected in the markedly reduced early expansion of donor T cells in these hosts, especially in the nonlymphoid tissues, despite the development of stable mixed chimerism. It is of interest that the differences in donor T cell expansion in spleen and BM in TBI and TLI conditioned hosts (∼3-fold) were considerably less than those observed in the nonlymphoid tissues (∼200- to 1500-fold).
IL-4 and protection against GVHD
To determine whether resistance of TLI-conditioned BALB/c hosts to GVHD is dependent on host secretion of IL-4, experiments were performed in which wild-type and IL-4−/− BALB/c hosts were given TLI and BM cells from C57BL/6 IL-4−/− donors. These donors were used to eliminate the contribution of donor IL-4 in vivo, because donor marrow NK1.1+ T cells inhibit GVHD via IL-4 secretion (17). Whereas 90% of the treated wild-type hosts given IL-4−/− marrow cells survived 120 days, IL-4−/− hosts died more rapidly (p < 0.007, by log-rank test), and only 33% survived during this time interval (Fig. 8⇓B). In additional experiments wild-type and IL-4−/− BALB/c hosts were given TLI and marrow cells and PBMC from C57BL/6 IL-4−/− donors. Again, the survival of IL-4−/− hosts was significantly reduced (p < 0.03) compared with that of wild-type hosts (Fig. 8⇓B).
Because the majority of CD57BL/6 spleen T cells at the end of TLI were NK1.1+ T cells, we assayed the ability of sorted TCRαβ+ T cells from TLI-treated wild-type donor C57BL/6 spleen cells to inhibit GVHD induced by sorted TCRαβ+ T cells from untreated wild-type C57BL/6 spleen cells in BALB/c hosts conditioned with TBI. Fig. 8⇑A shows that control BALB/c hosts given TBI and 1.5 × 106 T cell-depleted marrow cells from untreated C57BL/6 donors all survived at least 100 days. T cell-depleted marrow cells were used to eliminate any regulatory effects of donor marrow T cells on GVHD in these hosts. The addition of 0.1 × 106 sorted spleen T cells from untreated wild-type donors to the marrow cells resulted in the death of 90% of hosts by 90 days. In contrast, the addition of 0.1 × 106 sorted spleen T cells from TLI-treated wild-type donors failed to induce GVHD, and all hosts survived 100 days (Fig. 8⇑A). When BALB/c hosts were given T cell-depleted C57BL/6 marrow cells and a mixture of sorted spleen T cells from untreated and treated wild-type donors, the rapidity of death was significantly reduced compared with that in the group given only marrow and sorted spleen T cells from untreated donors (p < 0.5), as judged by the log-rank test. In contrast, addition of spleen T cells from treated IL-4−/− instead of treated wild-type donors resulted in more rapid death of the hosts compared with the group given marrow and untreated spleen T cells (Fig. 8⇑A).
Our previous studies showed that inbred mice and rats, and outbred dogs conditioned with TLI before allogeneic BM transplantation were resistant to acute GVHD compared with hosts conditioned with TBI (13, 14). BALB/c spleen cells after TLI treatment were reported to adoptively transfer resistance to GVHD in TBI-conditioned hosts injected with C57BL/6 splenic T cells (26). Transfer of resistance was removed by depletion of Thy-1+ cells with anti-Thy-1 mAb and complement (26). These regulatory cells were called natural suppressor cells (27), because they suppressed GVHD in a variety of host and donor combinations and did not interfere with chimerism. Furthermore, spleen cells assayed immediately after TLI were found to secrete high levels of IL-4 and low levels of IFN-γ after in vitro activation with anti-CD3 mAb compared with normal spleen cells (24). Thus, a change in the composition of T cells and their cytokine profile after TLI was theorized to contribute to protection against GVHD (24). Protection was also reported to be dependent upon the presence of asialo-GM1+ cells, because depletion of the latter cells by injection of anti-asialo-GM1 Abs into hosts given TLI markedly increased their sensitivity to lethal GVHD after subsequent allogeneic BM transplantation (22). Taken together these studies suggested that T cells expressing asialo-GM1, a marker shared with NK cells, may play an important role in ameliorating GVHD.
In the current study the role of NK1.1+ TCRαβ+ T cells was investigated, because this subset shares markers with NK cells, secretes high levels of IL-4, and has recently been shown to protect against GVHD induced by NK1.1− TCRαβ+ T cells (17). Immunofluorescent staining and multicolor analysis of the spleen cells in C57BL/6 and BALB/c mice showed a progressive increase in the percentage of NK1.1+ TCRαβ+ T cells and DX5+TCRαβ+ T cells, respectively, as the number of treatments of fractionated lymphoid irradiation increased. After 17 treatments, the latter cells rose from ∼1% in unirradiated controls to ∼60–70% among the residual TCRαβ+ T cells. Addition of ATS to the irradiation regimen resulted in a further increase to >90% of all TCRαβ+ T cells. Among the NK1.1+ T cells, the predominant subset was CD4+, and few CD8+ cells were observed.
The marked change in the T cell subset composition occurred without an expansion in the absolute number of NK1.1+ or DX5+ T cells and was instead due to a more profound depletion of NK1.1− and DX5− T cells. This remarkable alteration in subsets is probably explained by the ability of the BM that is shielded during TLI to restore NK1.1+ T cells by virtue of a rapid expansion of progenitor cells in response to peripheral depletion (20). In contrast, replenishment of NK1.1− or DX5− T cells is likely to be delayed from progenitors in the thymus that are damaged during thymic irradiation (28). Injection of anti-asialo-GM1 Abs into BALB/c host given TLI rapidly depleted the DX5+ TCRαβ+ T cells and indicated that these cells are asialo-GM1+. The latter cells are likely to be the asialo-GM1+ cells reported previously to protect against GVHD after TLI (22).
As expected, the sorted NK1.1+ TCRαβ+ T cells in the spleen after TLI secreted considerably higher levels of IL-4 after in vitro activation compared with the sorted NK1.1− TCRαβ+ T cells. The increased proportion of NK1.1+ and DX5+ T cells among all T cells after TLI resulted in an altered cytokine secretion pattern by all sorted TCRαβ+ T cells from C57BL/6 and BALB/c mice, such that the levels of IL-4 secretion and the ratio of IL-4 to IFN-γ secretion were markedly increased. This is likely to explain the high levels of IL-4 secreted in response to anti-CD3 mAb activation of whole spleen cells after TLI as well as the polarization toward Th2 alloreactive T cell clones reported previously (24, 25).
Although BALB/c hosts conditioned with either lethal (800 cGy) or sublethal (450 cGy) TBI with or without addition of ATS succumbed to acute GVHD after an injection of combined C57BL/6 BM and PBMC, hosts conditioned with TLI or TLI and ATS showed marked resistance to GVHD. Even in the case of sublethally (450 cGy) irradiated hosts, infusion of allogeneic peripheral blood cells resulted in the death of all recipients by 14 days (29). Histopathologic analysis of the liver, gut, skin, and pancreas of the hosts given TLI and ATS showed no microscopic evidence of GVHD at 100 days after transplantation. Nevertheless, these hosts were all mixed chimeras, with donor-type cells found in all lineages tested, including T cells, B cells, granulocytes, and macrophages. Hosts conditioned with TBI that subsequently developed severe GVHD after combined marrow and spleen cell transplantation showed a marked early expansion of donor-type T cells derived from the injected spleen cells in lymphoid and nonlymphoid tissues. The most dramatic expansion was observed in the gut and liver, with 35–70% of mononuclear cells extracted from these tissues on day 12 being made up of donor spleen-derived TCRαβ+ T cells. In contrast, minimal early expansion of donor T cells was observed in the same tissues in hosts conditioned with TLI and ATS, and only 2% of mononuclear cells from the gut and liver were donor T cells. It is likely that the inability of the donor T cells to invade and expand especially in the host nonlymphoid tissues is responsible for protection after the TLI and ATS regimen.
Because the increase in the percentage of DX5+asialo-GM1+TCRαβ+ T cells in BALB/c hosts given TLI was associated with increased IL-4 secretion and protection against GVHD, we compared the severity of GVHD in wild-type and IL-4−/− hosts conditioned with TLI. These hosts received BM cells with or without blood mononuclear cells from IL-4−/− C57BL/6 donors to eliminate donor contribution of IL-4. The survival of the wild-type hosts was significantly improved compared with that of the IL-4−/− or CD1−/− hosts, indicating that host CD1-reactive cells that express/secrete IL-4 contribute to protection against GVHD. In additional experiments, sorted splenic T cells from C57BL/6 wild-type donors conditioned with TLI inhibited the ability of untreated wild-type donor splenic T cells to induce GVHD in wild-type BALB/c hosts conditioned with TBI. The inhibitory activity of the T cells from TLI-treated donors was expected based on our previous studies of natural suppressor T cells found after TLI (27).
The current study shows that this inhibitory activity is probably mediated by the secretion of IL-4 by the adoptively transferred T cells, because sorted T cells from IL-4−/− C57BL/6 mice given TLI failed to inhibit GVHD and worsened the survival of BALB/c hosts. The cellular mechanisms by which IL-4 protects against GVHD are not clear. One possibility is that IL-4 regulates the interaction between donor T cells and host APCs such that donor T cell activation is impaired or shifted to a Th2 pattern with reduced GVHD potency (30, 31). In vitro PMA and ionomycin activation of sorted chimeric donor and host cells obtained from TLI-treated hosts 100 days after BM transplantation showed secretion of a Th2 pattern of cytokines (high IL-4 and low IFN-γ; data not shown). It is possible that secretion of IFN-γ by host T cells also plays a role in protection. However, we did not address this issue in the current study.
In conclusion, the T cells with the predominant NK1.1+ and DX5+ asialo-GM1+ subsets found after fractionated lymphoid irradiation can transfer protection to hosts against GVHD induced by donor alloreactive T cells. Cloned NK1.1+ T cell lines have been reported to protect against GVHD (32). In addition, protection is lost after depletion of asialo-GM1+ or IL-4-secreting cells. The experimental findings indicate that the residual host NK1.1+ and DX5+asialo-GM1+ T cells prevent the rapid expansion and invasion of donor T cells in the host nonlymphoid target tissues of GVHD and affect the patterns of host and donor T cell cytokine secretion.
We thank Mary Hansen and Sharon Dickow for preparation of this manuscript, and Aditi Mukhopadhyay for technical assistance.
↵1 This work was supported by National Institutes of Health Grants P01HL57443, R01HL58250, and R01AI37683.
↵2 Address correspondence and reprint requests to Dr. Samuel Strober, Division of Immunology and Rheumatology, Department of Medicine, CCSR Building, Room 2215-C, Stanford Medical Center, MC5166, 300 Pasteur Drive, Stanford, CA 94305-5166. E-mail address:
3 Abbreviations used in this paper: BM, bone marrow; GVHD, graft-vs-host disease; ATS, anti-mouse thymocyte serum; TLI, total lymphoid irradiation; TBI, total body irradiation.
- Received December 20, 2000.
- Accepted June 14, 2001.
- Copyright © 2001 by The American Association of Immunologists