Biology of Blood and Marrow Transplantation 10:40-48 (2004) 2004 American Society for Blood and Marrow Transplantation 1083-8791/04/1001-0004$30.00/0 doi:10.1016/j.bbmt.2003.09.013

Allogeneic versus Syngeneic Killer Splenocytes as Effector Cells for the Induction of Graft-versus- Tumor Effect

Shoshana Morecki, Elena Yacovlev, Yael Gelfand, Anna Vilensky, Shimon Slavin

Department of Transplantation & Cancer , Cell Therapy & Transplantation Research Center, Hadassah University Hospital, ,

Correspondence and reprint requests: Shoshana Morecki, PhD, Department of BMT, Hadassah University Hospital, Jerusalem 91120, Israel (e-mail: [email protected]).

Received August 8, 2003; accepted September 22, 2003

ABSTRACT The effect of allogeneic versus syngeneic killer cells derived from normal or severe combined immunodefi- ciency disease (SCID) mice was evaluated for induction of antitumor reaction in a murine model of mammary carcinoma. Tumor cells of H-2d origin were injected intravenously into H-2d/b mice 24 hours after total body irradiation (4 Gy). On the following day, lymphokine-activated killer (LAK) splenocytes, derived from either minor (H-2d) or major (H-2b) histocompatibility complex (MHC)–mismatched parental normal mice or MHC (H-2b)–mismatched SCID mice, were given intravenously. LAK cells of H-2d normal or SCID mice, syngeneic to the tumor, were inoculated in parallel. The results show that LAK cells derived from minor histocompat- ibility complex–mismatched or MHC-mismatched parental normal mice improved the probability of tumor- free survival as compared with LAK cells syngeneic to the tumor cells, but they aggravated the severity of graft-versus-host disease. SCID splenocytes serving as a source of natural killer (NK) cells were expanded and activated in vitro by rIL-2 to obtain a sufficient number of DX5؉ CD3؊ CD8؊ NK cells (SCID-LAK). H-2b SCID-LAK cells did not cause graft-versus-host disease and significantly delayed tumor growth compared with syngeneic H-2d SCID-LAK cells, as indicated by tumor colony assays in vitro and adoptive transfer experi- ments. However, the graft-versus-tumor effect was not long lasting, and treated mice finally died of tumor. Our results show an advantage of allogeneic over syngeneic cell therapy for achieving a graft-versus-tumor effect by rIL-2–activated T cells and NK cells. Periodic repetition of NK treatments may be required to achieve more durable antitumor effects. © 2004 American Society for Blood and Marrow Transplantation

KEY WORDS Allogeneic killer cells ● Syngeneic killer cells ● SCID mice ● Graft-versus-tumor effect

INTRODUCTION disease (GVHD) [12]. Furthermore, donor lympho- cyte infusion given after HSCT was shown to play a Immunotherapy strategies that aim to amplify an key role in eradicating malignant cells and preventing immune response and break tolerance to the tumor relapse [3-6,13-15]. The antitumor effects of alloreac- cells are mostly based on the use of tumor-syngeneic/ tive donor , designated as GVL and graft- autologous effector cells in both experimental models versus-tumor (GVT) effects, were ascribed mainly to and cancer patients [1-11]. During the last decade, an allogeneic T lymphocytes that recognize malignant increasing number of protocols based on an allogeneic cells in a major histocompatibility complex (MHC)– reaction between immunocompetent effector cells and restricted fashion and to MHC-nonrestricted natural tumor target cells were developed. This strategy killer (NK) and lymphokine-activated killer (LAK) emerged from clinical observations after allogeneic cells [16-20]. Unlike T cells, which are activated hematopoietic transplantation (HSCT), through recognition of foreign epitopes, it is sug- where there seemed to be a correlation between graft- gested that NK-cell activation is mediated via MHC versus-leukemia (GVL) effects and graft-versus-host class I missing-self epitopes when the negative signals

40 Allogeneic Killer Cells for Cancer of killing-inhibitory receptors (KIR) on the NK cells lands) in culture medium of 4T1 supplemented with Ϫ cannot be neutralized [21-23]. Allogeneic cell therapy 2-mercaptoethanol (5 ϫ 10 5 mol/L) and sodium (alloCT) has pronounced antitumor effects but is of- pyruvate (1 mmol/L; complete LAK medium). After 4 ten complicated by the development of GVHD, which days in a CO2 incubator at 37°C, cells were washed ϩ is ascribed to immunocompetent donor CD4 and and tested for their ability to exert an antitumor re- ϩ CD8 T cells that target organs such as skin, gut, sponse in vivo. liver, and lung [24-27]. In contrast to T cells, alloge- neic NK cells can mediate antitumor responses with- Propagation of LAK SCID Cells out causing GVHD when administered to immuno- Single-cell suspensions of splenocytes and bone compromised mice or humans [17-20], thus marrow (BM) cells derived from SCID mice were suggesting that NK alloreactivity warrants further prepared in complete LAK medium supplemented study as a potentially safer antitumor modality. with 10 ␮g/mL indomethacin (Sigma, St. Louis, MO). Encouraged by the efficient antitumor effects of Cell suspensions at a concentration of 1 ϫ 106/mL alloCT in previous studies, yet aware of the associated were plated in 6-well tissue culture plates (Corning, complications, we compared the efficacy of allogeneic New York, NY) or tissue culture flasks containing versus syngeneic cell therapy (synCT) by using various 3000 IU/mL rIL-2, according to the total number of sources of lymphokine-activated T and NK cells in an cells obtained. Cell and rIL-2 concentrations were attempt to induce GVT effects in a murine model of adjusted with fresh complete medium to 1 ϫ 106/mL mammary carcinoma. and 3000 IU/mL, respectively, every 3 to 4 days.

Flow Cytometry Analysis MATERIALS AND METHODS Peripheral blood mononuclear cells (5 ϫ 105) Mice were stained directly with phycoerythrin anti-mouse ϩ BALB/c (H-2d) (BALB), C57BL/6 (H-2b) (C57) Pan NK cells (DX5) to detect NK1.1 cells, fluores- d d/b DBA/2 (H-2 ) (BALB/cXC57BL/6) F1 (H-2 )(F1), cein isothiocyanate (FITC) anti-mouse CD3 (17A2), and C.B-17/IcrHsd-scid (H-2d; SCID, severe com- and FITC anti-mouse CD8a (53-6.7). Phycoerythrin bined immunodeficiency disease) (C.B-17) mice, aged rat immunoglobulin M isotype standard (R4-22) and 9 to 12 weeks, were obtained from Harlan Sprague FITC rat immunoglobulin G2b isotype standard Dawley (Indianapolis, IN). C57BL/6J-PrkdcϽscidϾ/ (A95-1) were used as isotype controls. All reagents SzJ (H-2b) (C57 SCID) mice were obtained from were purchased from BD PharMingen (San Diego, Jackson Laboratory (Bar Harbor, ME). All mice were CA). All samples were preincubated for 10 min on ice maintained in the Hadassah University Hospital in with unlabeled rat anti-mouse CD32/16 antibodies accordance with the national laws and regulations for (Southern Biotechnology Associates, Inc., Birming- the protection of animal welfare. ham, AL) to prevent nonspecific staining through the Fc fraction of the labeled antibodies. Staining was Tumor Cells performed as described previously [31]. 4T1 is a tumor cell line established from a cell 4T1 Clonogenic Assay subpopulation isolated from a single spontaneously arising mammary tumor in a BALB/c (H-2d) mouse Lung metastases were measured as already de- [28]. 4T1 cells were maintained in 10% fetal bovine scribed [32]. Briefly, lungs were harvested from 3 mice serum, 2 mmol/L glutamine streptomycin 100 ␮g/ of each experimental group 13 to 14 days after tumor mL, penicillin 100 U/mL, and 1% nonessential amino inoculation. Single-cell suspensions were prepared in acid (culture medium of 4TI) and prepared for intra- serial dilutions in culture medium of 4T1 including 60 venous (IV; 0.25 mL per mouse) and intradermal (0.1 ␮mol/L 2-amino-6 mercaptopurine (Sigma) and mL per mouse) inoculation or were prepared for in plated in 100 ϫ 20 mm tissue culture dishes (Becton vitro assays as described previously [29]. Yac-1 (H-2a) Dickinson Labware, Franklin Lakes, NJ). Tumor col- (Yac) is a cell line sensitive to the cytotoxic activity of onies were fixed and stained with 0.03% methylene NK cells [30] and acts as target cells in chromium 51 blue 10 to 14 days after plating. The number of tumor release assays. Cells were kept at 37°C in a humidified colonies was calculated per mouse lung for each ex-

5% CO2/air incubator. perimental group relative to 100% 4T1 colonies in the control group of untreated mice. Preparation of LAK Normal Splenocytes Measurement of Cytotoxic Activity Spleen cells (1 ϫ 106/mL) were plated in 225-cm2 tissue culture flasks (Costar, Cambridge, MA) con- Cells harvested from normal LAK and SCID- taining 6000 IU/mL recombinant interleukin (rIL)-2 LAK cultures were washed, diluted to 5 ϫ 106/mL, (Proleukin; Chiron BV, Amsterdam, The Nether- and tested in a standard 4-hour chromium 51 release

BB&MT 41 S. Morecki et al.

Table 1. Effects of Syngeneic versus Allogeneic Cell Therapy on Survival of Tumor-Inoculated Mice

Cause of Death No. Disease- Survival,* Free Effector Cells Median (Range) No. Mice Tumor-Related GVHD-Related Survivors

Tumor only 33 (20-137) 31 31 0 0

F1 splenocytes 33 (22-205)‡ 26 25 0 1 BALB splenocytes 28 (25-50) 24 24 0 0 C57 splenocytes 17 (13-20) 15 0 15 0 BALB LAK cells 33 (27-50) 18 18 0 0 DBA LAK cells† 41 (28-180)‡ 18 14 2 2

F1 LAK cells 38 (30-138)‡ 10 9 0 1 C57 LAK cells 18 (14-20) 10 0 10 0

ϫ ϫ 4 (BALB/c C57BL/6) F1 mice were inoculated with 5 10 4T1 cells 24 hours after receiving 4 Gy of total body irradiation. One day later, 20 to 25 ϫ 106 effector cells derived from normal mice were given to them intravenously, and rIL-2 was administered intraperitoneally for 3 consecutive days at 5 ϫ 104 IU/d, starting from the day of cell therapy. *Survival in days. †Mice treated with DBA rIL-2-activated splenocytes (LAK) received 1.2 ϫ 105 IU/d for 4 consecutive days. ‡Mice were killed without disease. assay, as previously described [33], against 5000 NK- 2-tailed unpaired Wilcoxon test. The Kaplan-Meier resistant (4T1) and NK-sensitive (Yac) target cells. method was used to calculate the probability of tu- mor-free survival as a function of the time after inoc- Irradiation ulation of tumor challenge [34]. Statistical significance Total body irradiation (TBI) was delivered by a between pairs of Kaplan-Meier curves was evaluated 6-MeV linear accelerator (Varian, Palo Alto, CA) at by the Wilcoxon test of Breslow [35]. dose rate of 1.9 Gy/min. Irradiation of single-cell suspensions was performed by using a Gamma Cell RESULTS 220 Excel Research irradiator (MDS Nordion, Ka- LAK Cells of MiHC-Mismatched or nata, Ontario, Canada). MHC-Mismatched Splenocytes Cell Therapy Antitumor activity across minor histocompatibil- ity complex (MiHC) or MHC antigen barriers was F 1 mice were conditioned with 4 Gy of TBI and tested in irradiated F mice inoculated with 4T1 tu- 24 hours later were inoculated IV with 5 ϫ 104 1 4T1 mor cells. TBI was given to allow temporal or perma- tumor cells. On the following day, naive splenocytes nent engraftment of the allogeneic therapeutic cells. and IL-2–activated splenocytes from normal or SCID MiHC-mismatched LAK cells derived from DBA/2 mice (20-30 ϫ 106 - ) were inoculated IV. In most ex mice resulted in a delay in tumor-related death, com- periments, rIL-2 was given intraperitoneally at 5 ϫ pared with syngeneic BALB-LAK cells (the median 104 IU/d in vivo for 3 to 4 days, starting from the day survival was 41 and 33 days, respectively; Table 1). of cell therapy. GVHD symptoms such as hunched The Kaplan-Meier probability of tumor-free survival posture, ruffled fur and cachexia, and survival were was significantly higher after treatment with DBA- checked daily. To distinguish GVHD from tumor LAK cells than with BALB-LAK cells (P ϭ .0014). morbidity, each mouse was tested after death for the Treatment with MHC-mismatched parental C57 presence of tumor metastases in the lungs and liver. LAK cells caused severe GVHD in F1 hosts, which Adoptive Transfer Experiments and Measurement resulted in the death of 10 of 10 mice that died free of of Tumor Growth tumors earlier than mice treated with syngeneic F1 LAK cells (median survival was 18 versus 38 days, Lungs were removed from F1 mice, and cell sus- respectively; Table 1). However, the probability of pensions were inoculated into naive BALB hosts for tumor-free survival was significantly higher after in- detection of residual tumor cells by an in vivo assay in oculation with C57 LAK cells as compared with syn- secondary recipients, as previously described [29]. ϭ geneic F1 LAK cells (P .04) (data not shown). These Growth of intradermally inoculated tumors was mea- results show that cell therapy with MiHC- or MHC- sured once a week [29]. mismatched activated splenocytes improves the prob- ability of tumor-free survival; however, because T Statistical Analysis cells contained in allogeneic LAK cells aggravate Statistical significance of tumor growth assessed GVHD, an alternative cell source needs to be applied by clonogenic assay was evaluated by the standard for safer induction of GVT effects.

42 Allogeneic Killer Cells for Cancer

Table 2. Effects of ␥ Irradiation on Allogeneic Cell Therapy

Killing Activity In Irradiation of Effector Cause of Death Vitro† Normal C57BL/6 Cells In Vitro (7.5 Survival*, Effector Cells Gy) Median (Range) No. Mice Tumor-Related GVHD-Related 4T1 Cells Yac Cells

None 33 (17-49) 37 37 0 ND ND Splenocytes ؊ 13 (9-20) 17 0 17 ND ND Splenocytes ؉ 34 (22-105) 14 14 0 ND ND LAK cells ؊ 26 (14-40) 17 0 17 48 ؎ 0.6 66 ؎ 6 LAK cells ؉ 39 (30-97) 16 16 0 50 ؎ 165؎ 4

Irradiated or nonirradiated naive splenocytes and rIL-2-activated splenocytes (LAK) (30 ϫ 106) were given intravenously 24 hours after ϫ 4 ϫ inoculation of 5 10 4T1 tumor cells into (BALB/c C57BL/6) F1 mice that had received 4 Gy of total body irradiation 24 hours previously. ND indicates not done. *Survival in days. †Killing activity against 4T1 (NK-resistant) or Yac (NK-sensitive) target cells was determined by Chromium 51 release assay. Results represent mean Ϯ SE % cytotoxicity of 3 separate experiments in an effector-target ratio of 100:1.

ϩ The Effect of Gamma Irradiation on AlloCT titumor effect by an enriched DX5 cell subset ϩ Naive or LAK C57 cells were irradiated in vitro without the CD3 cells responsible for GVHD, we performed a phenotypic analysis of spleen and BM (7.5 Gy) before injection into F1 mice that had been inoculated with 4T1 tumor cells 24 hours earlier. As cells derived from 2 strains of SCID mice lacking the seen in Table 2, both naive and LAK allogeneic ef- T-cell compartment of the immune system. BM cells fector cells given without in vitro irradiation induced of C.B-17 and C57 SCID mice contain a small pro- ϩ a severe allogeneic reaction against the F hosts that portion of DX5 cells, whereas splenocytes derived 1 ϩ finally led to GVHD-related death of 17 of 17 mice in from both strains contain 47% and 52% DX5 cells, each experimental group after a median of 13 and 26 respectively (Table 4). To obtain sufficient number of days, respectively. However, in vitro irradiation of cells for in vivo therapy experiments, cells were prop- effector cells before inoculation into the mice totally agated in vitro in the presence of IL-2 for 10 days, abrogated the GVHD-related death of all hosts. Irra- resulting in 10-fold and 5-fold increments of lympho- diation of naive or LAK cells significantly prolonged cytes derived from spleen cells and BM cells, respec- overall survival as compared with nonirradiated effec- tively (data not shown). Both displayed 60% to 78% ϭ ϭ ϩ ϩ ϩ tor cells (P .0000 and P .0004 for naive and LAK DX5 cells and no CD3 or CD8 T cells or NK T ϩ cells, respectively). Also, irradiated LAK cells signifi- cells. An exception was the low percentage of DX5 cantly prolonged the median of survival over that of cells derived from the BM of C.B-17 mice (35% after ϭ untreated controls (P .002). Killing activity directed 10 days of culture). These results show that SCID against tumor cells and NK-sensitive Yac cells, mea- spleen or BM-derived cells can be propagated in vitro sured in vitro, was not affected by gamma irradiation, ϩ and can provide sufficient numbers of DX5 cells, but all the recipients inoculated with irradiated effec- without contamination of T cells, for application in a tor cells later died of tumor metastases (14/14 and therapeutic experimental model (Table 4). 16/16) after a median survival of 34 and 39 days after treatment with naive or LAK splenocytes, respectively (Table 2). These results show that although irradia- tion of allogeneic effector cells used for immunother- apy prevented GVHD and did not abrogate antitumor Table 3. Phenotypic Analysis of Naive and LAK Splenocytes Derived from Normal C57BL/6 Mice effects in vitro, alloCT induced by the administration of a single dose of irradiated donor lymphocytes was Cell-Surface Marker (% Positive Cells) not sufficient to provide a long-lasting GVT effect in Cell Source CD3 NK1.1 NK1.1 CD3 vivo. Naive splenocytes 32 ؎ 3.3* 3.3 ؎ 0.9 3.6 ؎ 3.1 LAK cells† 45 ؎ 10 20 ؎ 110؎ 3 Phenotypic Characterization of Effector Cells Applied to Cell Therapy Phenotypic analysis was performed by fluorescence-activated cell- sorter analysis. Phenotypic analysis of naive splenocytes revealed Ϯ ϩ ϩ ϩ ϩ *Results represent mean SE values of at least 3 separate samples. 32% CD3 ,3%DX5 , and 4% CD3 DX5 cells. Single-positive cells do not include double-positive DX5ϩ After rIL-2 activation in vitro, all 3 cell subsets in- CD3ϩ cells. creased to levels of up to 45%, 20%, and 10%, re- †LAK cells were prepared by activation of splenocytes in vitro with spectively (Table 3). In an attempt to achieve an an- rIL-2 for 4 days.

BB&MT 43 S. Morecki et al.

Table 4. Fluorescence-Activated Cell-Sorter (FACS) Analysis of Spleen and Bone Marrow (BM) Cells Derived from H-2d (C.B-17) and H-2b (C57) SCID Mice

Before Culture* (% Positive Cells)† After Culture‡ (% Positive Cells) Cell Source NK1.1 CD3 CD8 NK1.1 CD3 NK1.1 CD8 NK1.1 CD3 CD8 NK1.1 CD3 NK1.1 CD8

BM C.B-17 3 ؎ 1.0 2 ؎ 1 0.2 ؎ 0.2 0.4 ؎ 0.4 0 35 ؎ 6.5 0.4 ؎ 0.2 0.5 ؎ 0.3 0.5 ؎ 0.4 0 Spleen C.B-17 47 ؎ 16 6.2 ؎ 2.4 1.2 ؎ 0.4 0.9 ؎ 0.5 0 65 ؎ 1 1.1 ؎ 0.9 1.4 ؎ 1.3 3.6 ؎ 2.3 0 BM C57BL SCID 5.5 ؎ 1.0 1.3 ؎ 1.3 0 1.4 ؎ 1.4 ND 60 ؎ 9 0.1 ؎ 0.1 0 0.1 ؎ 0.1 0 ؎ 0 Spleen C57BL SCID 52 ؎ 13 0.7 ؎ 0.6 0.85 ؎ 0.1 2.8 ؎ 2.0 0 78 ؎ 50؎ 00؎ 0 0.9 ؎ 0.8 0 ؎ 0

ND indicates not done. *Naive spleen and bone marrow cells were derived from C.B-17 or C57BL/6 SCID mice. †Results represent the mean Ϯ SE of at least 3 separate samples of naive and cultured cells. ‡FACS analysis was performed on days 8 to 10 on cultures containing rIL-2 (3000 IU/mL).

Antitumor Reactivity as Measured by Tumor treated with allogeneic C57 SCID-LAK spleen and Clonogenic Assay In Vitro C57 SCID-LAK BM cells, respectively, died of Antitumor activity was evaluated by tumor clono- GVHD, whereas 15 of 15 and 10 of 10 mice treated with naive splenocytes or LAK cells derived from genic assay in vitro in lung cells derived from F1 mice inoculated with 4T1 tumor cells and treated with LAK normal C57, respectively, died of GVHD (Table 1). cells 13 days before lung harvest. The antitumor ac- These results (Table 5) show that allogeneic SCID– tivity of allogeneic LAK cells derived from MiHC- derived LAK cells constitute a safe source of alloge- mismatched DBA/2 mice or MHC-mismatched nor- neic effector cells for GVT induction without causing mal and SCID C57 mice was compared with that of GVHD. syngeneic F or normal BALB and SCID C.B-17– 1 Effect of AlloCT versus synCT on Antitumor derived LAK cells. Cells isolated from the lungs of Activity In Vivo by Adoptive Transfer tumor-bearing untreated control mice contained 435 Experiments to 1015 tumor colonies in vitro, as determined in 5 separate experiments. The percentage of tumor colo- F1 mice inoculated with 4T1 tumor cells were nies in cells isolated from the lungs of treated mice treated 24 hours later with naive or LAK cells derived relative to 100% tumor colonies in tumor-bearing from either normal allogeneic C57 and SCID mice or control mice is presented in Figure 1. Allogeneic LAK syngeneic F1 and C.B-17 SCID mice. Fourteen days cells derived from MiHC-mismatched or MHC-mis- matched normal and SCID mice were more efficient in preventing tumor growth (4%, 4%, and 1% tumor colonies, respectively) than syngeneic LAK cells de- rived from BALB, F1, and C.B-17 SCID mice (14%, 21%, and 37% tumor colonies, respectively). Com- parison of the outcome of alloCT with that of synCT revealed a nonsignificant advantage of using MiHC- mismatched LAK (P ϭ .08) and a significant advan- tage of using normal MHC-mismatched LAK cells (P ϭ .02) and SCID MHC-mismatched LAK cells (P ϭ .03).

Effect of Allogeneic versus Syngeneic SCID-Derived LAK Cells on the Survival of Tumor-Bearing Mice Figure 1. Tumor colony assay in vitro. The LAK splenocytes F mice inoculated with 4T1 tumor cells were derived from either normal (DBA/2, BALB/c, and C57BL/6) or 1 ϫ treated 24 hours later with LAK cells derived from SCID (C57 and C.B-17) mice were used as effector cells (20-25 6 ϫ spleen or BM cells of C57 SCID and C.B-17 SCID 10 ) to treat (BALB/c C57BL/6) F1 hosts inoculated with 4T1 tumor cells. The number of tumor colonies in the lungs was deter- mice. There was no apparent difference in the median mined on day 14 after tumor inoculation and calculated per mouse survival of experimental groups treated with the vari- lung for each experimental group relative to 100% 4T1 colonies in ous SCID effector cells (Table 5). However, com- the control group of untreated mice. There was a nonsignificant pared with alloCT with effector cells derived from advantage for the comparison of DBA with BALB treatment (P ϭ ϭ normal mice (Table 1), C57 SCID–derived LAK cells .08) but a significant advantage for C57 over F1 treatment (P .02) did not cause GVHD; 0 of 18 and only 2 of 11 mice and for C57 SCID over C.B-17 SCID (P ϭ .03).

44 Allogeneic Killer Cells for Cancer

Table 5. The Effect of Syngeneic Versus Allogeneic SCID Splenocytes and BM Cells on the Survival of Mice Inoculated with 4T1 Tumor Cells

Cause of Death Survival,* No. No. Disease-Free Effector Cells Median (Range) Mice Tumor-Related GVHD-Related Survivors

Tumor only 33 (20-137) 31 31 0 0 C.B-17 SCID-LAK spleen cells 33 (26-64) 22 22 0 0 C.B-17 SCID-LAK BM cells 37 (23-201) 18 18 0 0 C57 SCID-LAK spleen cells 39 (31-84) 18 18 0 0 C57 SCID-LAK BM cells 41 (25-52) 11 9 2 0

ϫ ϫ 4 (BALB/c C57BL/6) F1 mice were inoculated with 5 10 4T1 cells 24 hours after receiving 4 Gy of total body irradiation. One day later, 20 to 25 ϫ 105 effector cells derived from SCID mice were given to them intravenously. All mice received 5 ϫ 104 IU of rIL-2 per day intraperitoneally for 3 consecutive days, starting from the day of cell therapy. *Survival in days. after tumor inoculation, lung cells were adoptively effector cells that were allogeneic to the 4T1 tumor transferred intradermally into naive BALB mice. Se- cells were equally efficient in the induction of antitu- rial weekly measurements of tumor size are presented mor effect (P Ͼ .05 for the comparison of naive C57 in Figure 2, showing that lung cells derived from F1 splenocytes versus C57 LAK cells and for normal C57 mice treated with naive allogeneic C57 and LAK cells LAK versus SCID C57 LAK cells). or C57 SCID-LAK cells did not give rise to any tumor colony growth in the adoptive recipients, whereas syn- DISCUSSION geneic F1 or C.B-17 SCID-LAK cells resulted in ex- ponential tumor growth similar to that observed in LAK cells reacting against MiHC- or MHC-mis- untreated tumor-bearing control mice. Follow-up of matched murine mammary carcinoma cells were survival showed that most secondary hosts inoculated shown to exert GVT effects, thus improving the prob- with lung cells derived from F1 mice treated with ability of tumor-free survival, as compared with no or naive allogeneic C57 and LAK cells or C57 SCID- poor GVT effects induced by syngeneic effector cells. Ͼ LAK cells remained tumor-free for 200 days (Figure However, in most cases, such GVT effects were asso- 3). Kaplan-Meier probability of tumor-free survival ciated with an increased incidence of GVHD, as pre- emphasized the advantage of alloCT over synCT in viously reported with naive splenocytes [29,36]. Be- ϭ exerting antitumor effects (P .0003 for C57 LAK cause allogeneic naive lymphocytes and LAK cells ϭ cells versus F1 LAK cells and P .0001 for C57 contain both T and NK cells, we directed our efforts SCID-LAK cells versus C.B-17 SCID-LAK cells). All toward investigating the GVT effects induced by ir-

Figure 2. Tumor size measured in adoptive transfer recipients. ϫ 6 ϫ Lung cells (1.5 10 ) from primary (BALB/c C57BL/6) F1 mice Figure 3. Survival of secondary BALB/c recipients inoculated in- were transferred intradermally into naive secondary BALB/c hosts tradermally with 1.5 ϫ 106 lung cells derived from primary (BALB/ on day 14 after tumor inoculation. Primary F1 recipients received 4 c x C57BL/6) F1 mice that were inoculated with tumor cells 14 days ϫ 4 Gy of total body irradiation 24 hours before inoculation of 5 10 previously. Primary F1 recipients received 4 Gy of total irradiation 4T1 tumor cells and 20 to 25 ϫ 106 splenocytes derived from 24 hours before tumor inoculation and 20–25 ϫ 106 splenocytes various sources 1 day after tumor inoculation. Results represent 1 of derived from various sources 1 day after tumor inoculation. Exper- 2 experiments. Experimental groups consisted of 5 to 9 mice in each imental groups consisted of 10 to 12 mice pooled from 2 experi- experiment. ments.

BB&MT 45 S. Morecki et al. radiated LAK cells, in which radiosensitive T cells sis-inducing ligand expression on activated NK cells, cannot exert GVHD, or we used a source of NK cells which suggests that through cytokine activation, anti- alone in an attempt to prevent T cell–mediated tumor effects might be enhanced via NK activation in GVHD. Indeed, irradiated LAK did not cause GVHD vivo by intrinsic factors or by exogenous rIL-2 or but were unable to mediate a long-lasting GVT effect. other cytokine administration [39]. NK activity is The possible antitumor activity of NK cells in the rather complex and regulated via the inhibitory and setting of HSCT independently of GVHD has been activating receptors [23,39,40]. Blockading their in- suggested in several recent reports [17-20]. In our hibitory receptor (KIR) molecules has been shown to study, cell sources enriched with NK cells syngeneic enhance antitumor activity directed against murine or allogeneic to the tumor cells were evaluated as a leukemia [41]. However, because NK killing activity is possible source of safer cell therapy after HSCT. After mediated through the “missing self” of a KIR-like rIL-2 activation and expansion in vitro, LAK cells ligand [22], allogeneic, intentionally mismatched derived from C.B-17 and C57 SCID mice had no (rather than matched) NK cells should be more effec- ϩ ϩ ϩ CD3 or CD8 cells and consisted mainly of DX5 tive in triggering the eradication of residual tumor cells, which exerted cytotoxic activity in vitro against cells. As shown in our 4T1 model and as also indicated 4T1 tumor cells and NK-sensitive Yac cells, similarly in a recently reported clinical trial in patients with to LAK cells derived from normal splenocytes (data acute myelogenous leukemia treated with haploiden- not shown). tically mismatched stem cell allografts in an elegant As reported previously, SCID-derived cells given work by Ruggieri et al. [19,20], allogeneic NK allo- to mice with and without MCA-38 tumor cells after reactivity seems to have a therapeutic advantage in allogeneic HSCT did not induce GVHD in the re- prevention of relapse when mismatching of NK allo- cipients [17,18]. Unlike previous reports in mice in- reactivity exists in the direction of graft versus host. oculated with MCA-38 tumor, in our 4T1 tumor Similarly, when NK alloreactivity is in the direction of model, syngeneic and allogeneic SCID-LAK cells host versus graft, the incidence of rejection is de- were compared and were administered after nonmy- creased, suggesting that alloreactive NK cells may also eloablative TBI without HSCT. As evidenced by the be effective in eliminating residual hematopoietic cells tumor colony assay and adoptive transfer experiments, of host origin as conditioning for HSCT [19,20]. NK cells allogeneic to the tumor exerted efficient Because allogeneic NK cells did not cause GVHD in GVT effects in tumor-bearing mice, in contrast to our study or in other trials [17-20], activated alloge- NK cells syngeneic to the tumor. Unfortunately, neic intentionally mismatched NK cells could be used probably because of a lack of memory NK cells, such as tumor-selective killer cells against hematopoietic GVT effects were short acting and thus not sufficient tumor cells and other tumor cells susceptible to NK to exert a long-lasting in vivo effect. Because alloge- cell–mediated lysis. neic NK cells, as well as irradiated LAK cells, includ- LAK cells derived from normal BALB mice did ing radiosensitive T cells, did not cause GVHD, the not cause severe GVHD symptoms in F1 hosts, and possibility of exploiting their GVT capability over none of theses hosts died from GVHD. It is interest- longer periods of time through the use of multiple ing to note that although BALB and C57 mice were doses given at short intervals should perhaps be con- equally matched with the mismatched F1 hosts, the sidered. intensity of GVHD induced by C57BL/6-derived Taken together, our results support the use of cells dramatically exceeded that induced by BALB- allogeneic cell–mediated immunotherapy based on the derived cells. The intensity of GVHD and its lethal use of T cell–depleted NK cells for the induction of consequences are determined by the degree of MiHC safe antitumor activity. NK cells have been shown to and MHC mismatching, as well as by other genetic mediate hybrid resistance, in which parental donor factors, such as the cytokinetic potential of T cells, and hematopoietic cells, normal and malignant alike, may additional factors that cannot be controlled intention- be rejected by the F1 host even though T cells are ally. In our experimental models as in clinical practice, genetically tolerant to the allograft [37]. Our working the same degree of MiHC or MHC mismatching may hypothesis predicts that similarly to the way that NK result in uneventful transplantation in some individu- cells can break tolerance to self in the course of hybrid als and lethal GVHD in others. The fact that tumor- resistance, such cells might also serve as effector cells syngeneic SCID-derived LAK cells did not exert an for breaking tolerance to tumor cells. Furthermore, antitumor effect further emphasizes the major role of rIL-2–activated NK cells express on their surface the mismatched NK alloreactivity in exerting an antitu- tumor necrosis factor–related apoptosis-inducing li- mor response. gand that mediates killing of many transformed cells In summary, these experiments, in accord with our but spares most normal somatic cells [38,39]. Various previous investigations and supported by indirect data cytokines known for their antitumor effects are also from clinical studies, suggest a clear advantage of able to induce tumor necrosis factor–related apopto- alloCT over synCT in exerting GVT effects and em-

46 Allogeneic Killer Cells for Cancer phasize the potential use of intentionally mismatched tigen pulsed donor lymphocytes. Bone Marrow Transplant. 2001; NK cells, possibly in repeated doses, to achieve safe 28:795-798. and effective immunotherapy of minimal residual dis- 11. Dudley ME, Wunderlich JR, Robbins PF, et al. Cancer regres- ease. Clinical studies currently in progress are aiming sion and autoimmunity in patients after clonal repopulation to further exploit the use of allogeneic cell–mediated with antitumor lymphocytes. Science. 2002;298:850-854. 12. Weiden PL, Sullivan KM, Flournoy N, Storb R, Thomas ED. immunotherapy, possibly in conjunction with HSCT, Antileukemic effect of chronic graft- versus-host disease: con- for prevention or minimizing relapse, which continues tribution to improved survival after allogeneic marrow trans- to be the single major barrier to successful HSCT. plantation. N Engl J Med. 1981;304:1529-1533. 13. Kolb HJ, Mittermuller J, Clemm C, et al. Donor leukocyte transfusions for treatment of recurrent chronic myelogenous ACKNOWLEDGMENTS leukemia in marrow transplant patients. Blood. 1990;76:2462- We thank the Danny Cunniff Leukemia Research 2465. 14. Collins RH Jr, Shpilberg O, Drobyski WR, et al. Donor leu- Laboratory, the Gabrielle Rich Leukemia Research kocyte infusions in 140 patients with relapsed malignancy after Foundation, the Cancer Treatment Research Foun- allogeneic bone marrow transplantation. J Clin Oncol. 1997;15: dation, the Novotny Trust, the Szydlowsky Founda- 433-444. tion, the Fig Tree Foundation, and Ronne and 15. Slavin S, Morecki S, Weiss L, et al. Donor infu- Donald Hess for their continuous support of our on- sion: the use of alloreactive and tumor-reactive lymphocytes for going basic and clinical research. immunotherapy of malignant and nonmalignant diseases in conjunction with allogeneic stem cell transplantation. J Hema- tother Stem Cell Res. 2002;11:265-276. 16. Weiss L, Reich S, Slavin S. Allogeneic cell therapy in murine REFERENCES B-cell leukemia (BCL1): 2. The role of non-activated and 1. Rosenberg SA, Lotze MT, Yang JC, et al. Prospective random- rIL-2-activated CD4ϩ and CD8ϩ T cells in immunotherapy ized trial of high-dose interleukin-2 alone or in conjunction for leukemia. Cytokines Cell Mol Ther. 1999;5:153-158. with lymphokine-activated killer cells for the treatment of pa- 17. Murphy WJ, Longo DL. The potential role of NK cells in the tients with advanced cancer. J Natl Cancer Inst. 1993;85:622- separation of graft-versus-tumor effects from graft-versus-host 632. disease after allogeneic bone marrow transplantation. Immunol 2. Fefer A, Benyunes M, Higuchi C, et al. Interleukin-2 ϩ/Ϫ Rev. 1997;157:167-176. lymphocytes as consolidative immunotherapy after autologous 18. Asai O, Longo DL, Tian ZG, et al. Suppression of graft-versus- bone marrow transplantation for hematologic malignancies. host disease and amplification of graft-versus-tumor effects by Acta Haematol. 1993;89(suppl 1):2-7. activated natural killer cells after allogeneic bone marrow trans- 3. Slavin S, Naparstek E, Or R, Weiss L. Prevention of GVHD plantation. J Clin Invest. 1998;101:1835-1842. and induction of graft vs leukemia (GVL) effects: will it ever be 19. Ruggeri L, Capanni M, Martelli MF, Velardi A. Cellular ther- possible? Bone Marrow Transplant. 1988;3:208-209. apy: exploiting NK cell alloreactivity in transplantation. Curr 4. Slavin S, Ackerstein A, Nagler A, Naparstek E, Weiss L. Cell- Opin Hematol. 2001;8:355-359. mediated cytokine-activated immunotherapy (CCI) of malig- 20. Ruggeri L, Capanni M, Urbani E, et al. Effectiveness of donor nant hematological disorders for eradication of minimal resid- alloreactivity in mismatched hematopoietic ual disease (MRD) in conjunction with conventional transplants. Science. 2002;295:2097-2100. or bone marrow transplantation (BMT). Blood. 21. Karre K, Ljunggren HG, Piontek G, Kiessling R. Selective 1990;76(suppl 1):566a, (abstr.). rejection of H-2-deficient lymphoma variants suggests alterna- 5. Slavin S, Naparstek E, Nagler A, Ackerstein A, Kapelushnik Y, tive immune defence strategy. Nature. 1986;319:675-678. Or R. Allogeneic cell therapy for relapsed leukemia following 22. Seaman WE. Natural killer cells and natural killer T cells. bone marrow transplantation with donor peripheral blood lym- Arthritis Rheum. 2000;43:1204-1217. phocytes. ExpHematol. 1995;23:1553-1562. 23. Smyth MJ, Hayakawa Y, Takeda K, Yagita H. New aspects of 6. Slavin S. Immunotherapy of cancer with alloreactive lympho- natural-killer-cell surveillance and therapy of cancer. Nat Rev cytes. Lancet Oncol. 2001;2:491-498. Cancer. 2002;2:850-861. 7. Verneris MR, Baker J, Edinger M, Negrin RS. Studies of ex 24. Ferrara JL, Deeg HJ. Graft-versus-host disease. N Engl J Med. vivo activated and expanded CD8ϩ NK-T cells in humans and 1991;324:667-674. mice. J Clin Immunol. 2002;22:131-136. 25. Crawford SW, Longton G, Storb R. Acute graft-versus-host 8. Rosenberg SA, Packard BS, Aebersold PM, et al. Use of tumor- disease and the risks for idiopathic pneumonia after marrow infiltrating lymphocytes and interleukin-2 in the immunother- transplantation for severe aplastic anemia. Bone Marrow Trans- apy of patients with metastatic melanoma. A preliminary report. plant. 1993;12:225-231. N Engl J Med. 1988;319:1676-1680. 26. Kolb HJ, Schattenberg A, Goldman JM, et al. Graft-versus- 9. Falkenburg JH, Wafelman AR, Joosten P, et al. Complete leukemia effect of donor lymphocyte transfusions in marrow remission of accelerated phase chronic myeloid leukemia by grafted patients. European Group for Blood and Marrow treatment with leukemia-reactive cytotoxic T lymphocytes. Transplantation Working Party Chronic Leukemia. Blood. Blood. 1999;94:1201-1208. 1995;86:2041-2050. 10. Slavin S, Ackerstein A, Morecki S, Gelfand Y, Cividalli G. 27. Ueno NT, Rondon G, Mirza NQ, et al. Allogeneic peripheral- Immunotherapy of relapsed resistant chronic myelogenous leu- blood progenitor-cell transplantation for poor-risk patients kemia post allogeneic bone marrow transplantation with alloan- with metastatic breast cancer. J Clin Oncol. 1998;16:986-993.

BB&MT 47 S. Morecki et al.

28. Dexter DL, Kowalski HM, Blazar BA, Fligiel Z, Vogel R, 35. Breslow W. A generalized Kruskal-Wallis test for comparing K Heppner G. Heterogeneity of tumor cells from a single mouse sample subject to unequal patterns of censorship. Biomedika. mammary tumor. Cancer Res. 1978;38:3174-3181. 1970;57:579-594. 29. Morecki S, Yacovlev E, Diab A, Slavin S. Allogeneic cell ther- 36. Morecki S, Yacovlev E, Gelfand Y, Uzi I, Slavin S. Cell therapy apy for a murine mammary carcinoma. Cancer Res. 1988;58: with preimmunized effector cells mismatched for minor histo- 3891-3895. compatible antigens in the treatment of a murine mammary 30. Kiessling R, Klein E, Wigzell H. “Natural” killer cells in the carcinoma. J Immunother. 2001;24:114-121. mouse. I. Cytotoxic cells with specificity for mouse Moloney 37. Yu YY, George T, Dorfman JR, Roland J, Kumar V, Bennett leukemia cells. Specificity and distribution according to geno- M. The role of Ly49A and 5E6(Ly49C) molecules in hybrid type. Eur J Immunol. 1975;5:112-117. resistance mediated by murine natural killer cells against nor- 31. Morecki S, Gelfand Y, Nagler A, et al. Immune reconstitution mal T cell blasts. Immunity. 1996;4:67-76. following allogeneic stem cell transplantation in recipients con- 38. Ashkenazi A, Dixit VM. Death receptors: signaling and mod- ditioned by low intensity vs myeloablative regimen. Bone Mar- ulation. Science. 1998;281:1305-1308. row Transplant. 2001;28:243-249. 39. Smyth MJ, Cretney E, Takeda K, et al. Tumor necrosis factor- 32. Pulaski BA, Ostrand-Rosenberg S. Reduction of established spontaneous mammary carcinoma metastases following immu- related apoptosis-inducing ligand (TRAIL) contributes to in- notherapy with major histocompatibility complex class II and terferon gamma-dependent natural killer cell protection from B7.1 cell-based tumor vaccines. Cancer Res. 1998;58:1486-1493. tumor metastasis. J ExpMed. 2001;193:661-670. 33. Morecki S, Leshem B, Weigensberg M, Bar S, Slavin S. Func- 40. Diefenbach A, Raulet DH. Innate immune recognition by tional clonal deletion versus active suppression in transplanta- stimulatory immunoreceptors. Curr Opin Immunol. 2003;15: tion tolerance induced by total-lymphoid irradiation. Trans- 37-44. plantation. 1985;40:201-210. 41. Koh CY, Blazar BR, George T, et al. Augmentation of antitu- 34. Kaplan E. Non-parametric estimation from incomplete obser- mor effects by NK cell inhibitory receptor blockade in vitro and vation. J Am Stat Assoc. 1958;53:457-481. in vivo. Blood. 2001;97:3132-3137.

48