Int. J. Cancer: 61, 253-260 (1995) 0 1995 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de I'Union Internationale Contre le Cancer

REJECTION OF MOUSE MELANOMA ELICITED BY LOCAL SECRETION

OR NATURAL KILLER CELLS IN TUMOR REJECTION OF INTERLEUKIN-2: IMPLICATING MACROPHAGES WITHOUT T CELLS

Isao HARA, Hai ?(GUYEN, Yoshizumi TAKECHI, Bernd GANSBACHER, Paul B. CHAPMAN and Alan N. HOUGHTON' Mernonal Sloan-Kettering Cancer Center and Cornell University Medical College, 1275 York Avenue, New York, NY 10021. USA.

Tumor cells transduced with cytokine genes provide a model to study host-effector mechanisms involved in tumor rejection. Local IL-2 production within a tumor site mimics a specific helper-T-cell response, bypassing an immunization phase. Growth of mouse B 16F 10 melanomas transduced with interleu- kin-2 (IL-2) in syrtgeneic hosts were significantly delayed. IL-2- producing B 16F I0 cells were super-transduced with inter- feron-y to up-regulate expression of major-histocompatibility- complex (MHC) antigens. Expression of class-I- or class-ll-MHC molecules did not augment tumor rejection of IL-2-secreting tumor cells. Rejection of IL-2-transduced B 16F I0 cells in synge- neic mice was unaffected by depletion of CD8+ T-cell and NK I. I + natural-killer (NK) cell populations. Tumor rejection occurred in SCID mice even after depletion of NKI.I+ cells, confirming that 'r cells and NK cells were not required for tumor rejection. Histologic examination of sites of tumor rejection showed inflammation, characterized by infiltrates of macrophages, occasional neutrophils, and areas of necrosis. When mice were )treated systemically with macrophage-colony- stimulating factor to expand monocyte pools, tumor rejection was significantly ,augmented further. This study shows that in situ IL-2 production can result in tumor rejection mediated by inflammatory events, possibly involving macrophages, and mim- icking a delayed-type hypersensitivity (DTH) response even in the absence of 'K cells and NK cells. Furthermore, tumor rejection can be enhanced by systemic administration of a cytokine to expanid potential inflammatory cell populations. Q 1995 Wiley-Liss. Inc.

There is considerable evidence from transplantable tumor models that T lymphocytes play a central role in the specificity of tumor rejection. However, large gaps remain in understand- ing the actual eRector mechanisms responsible for tumor rejection. In some cases, the host is able to mount a specific immune response: that leads to tumor rejection, while in other circumstances tuimors can grow and kill the host despite the presence of T cellls that recognize the tumor (Koeppen et a/., 1986).

One approach lo this complexity is to isolate various stages of tumor response into discrete steps. An initial first phase is host recognition o f antigens expressed by the tumor. T lympho- cytes can play a central role at this stage. In transplantable tumor models, tumor rejection is abrogated by depletion of T cells in the host, and immunity can be passively transferred from one host to another by T cells. However, these observa- tions do not directly implicate T cells as effectors of tumor rejection. The second phase, tumor rejection with actual destruction of tumor cells, could involve a variety of immune and inflammatory cell types (Tepper et al., 1989; Colombo et al., 1991; Cavallo et al., 1992; reviewed in Colombo and Forni, 1994).

The introduction of cytokine genes into tumor cells has provided an approach to mimic a helper-T-cell response at the tumor site, bypassing a phase of specific T-cell recognition by the host. We have used the mouse B16F10 melanoma system to study rejection of tumor cells transduced to secrete interleu- kin-2. B16F10 melanoma is a poorly immunogenic tumor that elicits no detectable T-cell response in syngeneic hosts. Expres- sion of IL-2 in B.16F10 cells has been shown to lead to tumor rejection (Fearonl et al., 1990; Dranoff et aL, 1993). We found

that tumor rejection of B16F10 induced by local IL-2 secretion does not require host T cells or natural killer cells. Inflamma- tory responses, particularly involving macrophages, were impli- cated in tumor rejection. Effects of local IL-2 production were enhanced by expanding monocyte populations with systemic treatment using macrophage-colony-stimulating factor (M- CSF). These results show that IL-2 production locally at the tumor site can lead to delayed-type hypersensitivity responses that mediate tumor rejection, and that a systemically adminis- tered cytokine can enhance local tumor rejection mediated through these inflammatory responses.

MATERIAL AND METHODS Retroviral vector design and transduction of tumor cells

The retroviral vector constructs used in this study have been described, including thevectors DCA (Hantzopoulos et a/., 1989), DC/TKIL2 (Gansbacher et al., 1990b), and N2ICMVIFN-y (Gansbacher et a/., 1990a). Briefly, the DCAvector was used as a control, in which the human adenosine-deaminase minigene was cloned into the 3' LTR of the N2 vector (Armentano et al., 1987). In the DC/TKIL2 vector, a 0.5-kb DNA fragment encoding human IL-2 cDNA was ligated to the 0.85-kb DNA fragment of the HSV thymidine-kinase promoter and inserted into the polylinker site in the 3' LTR of a modified N2 vector (Hantzopoulos et al., 1989) (4). The prefix DC is short for double copy, where the inserted gene in the 3' LTR duplicates by copying into the 5' LTR when the virus integrates into the genome. To generate the N2ICMVIFN-y vector construct, the 0.8-kb cytomegalovirus promoter was ligated to a 0.6-kb DNA fragment encoding the mouse IFN-y gene and cloned down- stream of the neo gene coding sequence in the N2 vector (Armentano et al., 1987) (7). These retroviral constructs were converted to corresponding virus using established proce- dures; vector DNA was transfected into a helper-free, eco- tropic (DCITKIL2) or amphotropic (DCA and N2/ CMVIFN-y) packaging cell line (Markowitz et a/., 1988a,b). Colonies were isolated following selection with G418, ex- panded to cell lines, and cell-free supernatant was tested for the presence of virus. Cell lines secreting high titer of virus (10h-105/Neo CFU/ml) were used to infect mouse B16F10 melanoma cells. Clones of transduced B16F10 cells were isolated after selection in G418, expanded to cell lines, and secretion of IL-2 and IFN-y into the supernatant was mea- sured. The IL-2 transduced clones secreted physiological levels of IL-2.

IL-2 and interferon-y assays For assay of IL-2 levels, parental or IL-2-secreting B16F10

cells, 3 x lo5, were plated in 6-well plates with 1.5 ml of media, and supernatants were collected after 24 hr. IL-2 activity was measured using IL-2-dependent human primary lymphoblasts

'To whom correspondence and reprint requests should be ad- dressed. Fax: (212) 717-3036.

Received: October 26,1994 and in revised form December 13,1994.

254 H A W ETAL.

1 i -a- FlOlDCA

-0- F1011L2 Lo

0 0 10 20 30 40 50

1 $0

Days After Tumor Injection

FIGURE 1 -Tumor rowth in C57BLi6 mice challenged S.C. with B16F10 melanoma and transduced clones of B16F10. 0, B16F10 parental melanoma. 8, FlO/DCA cells are B16F10 melanoma cells transduced with a control vector. 0, FlO/IL2Lo cells are B16F10 cells transduced with IL-2 vector and secrete relatively low levels of IL-2 (13 U/m1/24 hr). ., FlO/IL2Hi cells are B16F10 cells transduced with IL2 vector and secrete higher levels of IL-2 (27 U/m1/24 hr). Mice received 50,000 cells in the S.C. tissue of the flankon day 0. Tumor size was measured as the product of the greatest diameter X perpendicular diameter. Error bars represent standard deviation of tumor size with 8 mice per group.

TABLE I - REJECTION OF B16F10 MELANOMA TRANSDUCED T O SECRETE INTERLEUKIN-2: COMPILATION OF EXPERIMENTS EVALUATING FOR APPEARANCE O F TUMORS

Numher of tumor-hearing mice (Per cent tumor-bearing mice)5

Day after tumor challenge6

10 15 20 25 30 35 40 50 60

Number of Numher Host TX’ experiment53 of mice4 Tumor’

B16F10 None

FlODC None

FlOIL2Hi None

FlOIL2Hi -CD4

F1OIL2Hi -CD8

FlOIL2Hi -NK

FlOIL2Hi -NKiCDS

6

4

11

3

3

3

2

70

20

20

21

38 51 53

- (1;) (10) (33) (48) (62) (81) (86) (100)

- 16 4 10 12 14 16 -

(0) (0) (0) (0) (6) (11) (17) (22) (33) ‘Tumors used for challenge were parental B16F10 melanoma; FlODC, a B16F10 sub-clone transduced with control retroviral vector

DCA; and FlOILZHi, a B16F10 sub-clone transduced with the retroviral vector DC/TKIL2, as described in “Material and Methods.” In each experiment, 50,000 cells were injected into the flank of mice, which were monitored for appearance of palpable tumors.-*Host tx. Host C57B1 mice were untreated controls (none), depleted of CD4+, CD8+ or NK cells, or pretreated with macrophage-colony- stimulating factor (M-CSF) as described in “Material and Meth~ds.”-~Total number of separate e~periments.-~Total number of mice used in all experiments.-jNumber of tumor-bearing mice (per cent of total mice) in all experiments at each time point. Mice bearing palpable tumors were scored as positive.-‘jDay after tumor challenge. Mice were challenged with 50,000 tumor cells injected S.C. in the flank on day O.-’Designates all mice (100%) bearing tumors.

in a proliferation assay, as described by Zier (1982). For assessment of IFN-y production, supernatant was obtained as above, and IFN-y activity was determined by enzyme-linked immunoassay (ELISA) (Amgen, Thousand Oaks, CA).

Tumor cell lines and animal studies B16F10 is a mouse melanoma cell line of C57BL/6 origin

kindly provided by Dr. I. Fidler (Houston, TX) (11). Cell lines were cultured in Eagle’s MEM supplemented with 5% fetal bovine serum. C57BL/6 mice, BALBic w i n u and BALB/c SCID mice (6- to 8-week-old females) were obtained from the Jackson Laboratory (Bar Harbor, ME). Tumor cells were

washed twice in PBS, and 50,000 cells were injected S.C. into the flank of each mouse. Tumor growth was measured in millimeters using a caliper. The Iongest surface length (a) and its perpendicular width (b) were measured, and tumor size reported as a x b. Tumors were checked at least 3 times per week and were allowed to grow to 400 mmz (animals were killed when tumors reached this size). For histologic evalua- tion, tissue at the site of tumor-cell inoculation was fixed in formalin solution, blocked in paraffin, sectioned every 4 microns and stained with hematoxylin and eosin. Statistical analysis of tumor growth was performed using the Bonferroni 2-sided t-test, a conservative analysis to allow for multiple comparisons.

INFLAMMATION AND TUMOR REJECTION 255

Cell-surface markers and magnetic head selection The following monoclonal antibodies (MAbs), obtained

from the ATCC (Rockville, MD), were used for typing class-I and class-IT-MHC ex ression: MAb 28-14-8s (ATCC number

against H-2Khand AF6-120.1.2 (ATCCnumberHB163) against ]-Ah. For flow cytometry, single-cell suspensions were incu- bated with MAbs at 4°C for 30 min. After washing, cells were incubated with fluorescein-isothiocyanate-conjugated rabbit anti-mouse IgG antibodies (Dakopatts, Glostrup, Denmark) at 4°C for 30 min. The stained cells were washed and analyzed using an EPICS-PROFILE 11 flow cytometer (Coulter, Hi- aleah, FI,). For magnetic-bead selection, single-cell suspen- sions of 1 t o 2 x 106 cells were incubated with MAb at 4°C for 30 min. After washing, the cells were re-suspended in 2% fetal bovine serum/PBS and incubated with M-450 Dynabeads coupled t o goat anti-mouse IgG (Dynal, Great Neck, NY) at 4°C for 30 min. Positive cells were selected using a Magnetic Particle Concentrator (MPCl, Dynal) and cultured in Eagle's MEM with 5% fetal bovine serum.

Antibody treatments in vivo Depletion of T cells in vivo was accomplished by i.p.

administration of rat MAb GK1.5 (anti-CD4; IgGZb) and MAb 2.43 (anti-CD8; IgGzb) (both hybridomas were acquired from the ATCC). These MAbs were used as ascites fluids (titer > 1: 10,000 by staining of mouse thymocytes by flow cytometry). MAb preparations (0.2 ml) were injected i.p. at day -3 and every 7 days thereafter. Throughout experiments, these treat- ments depleted respective T-cell sub-population > 97% as determined by indirect immunofluorescence staining and cyto- fluorometric analysis of lymph nodes with MAbs GK1.5 (CD4) or 2.43 (CD8) respectively (data not shown). Natural-killer (NK)-cell depletion was performed using MAb PK136 (anti- NK-1.1) (from ATCC) in C57BL/6 mice. Antibody (0.2 ml) was injected at day -3 and every 7 days thereafter. Because MAb PK136 does not react with NK cells of BALB/c origin, anti-asialo GMI anti-serum (Wako, Richmond, VA) was used in nu/nu or SCID mouse studies. Anti-asialo GMI anti-serum, 50 p1 of anti-serum diluted to 200 pl in PBS, was injected i.p. at day - I and repeatcd every 5 days. Depletion of NK cells was assessed by 4-hr "Cr-release assays, with 5,000 YAC cells as

HB-27) against H-2D R /Ld, AF6-88.5.3 (ATCC number HB158)

Cant r o l H-2K H-2D I - A 7 - 1 77 I1 I 1

7 - 1 7 1 7 - r

FIGURE 2 - Expression of class-I- and class-11-MHC molecules by B16F10 melanoma in vitro and in vivo and FlO/MHCI and HOIMHCI + I1 transduced cells. Cells were stained by immuno- fluorescence with MAb against class-I (H-2K or H-2D) or class-I1 (I-A)-MHC molecules or control MAb and evaluated by flow cytometry as described in "Material and Methods". The X axis represents fluoresccnce intensity and the Y axis represents cell number.

targets and spleen cells as effector cells at effector-to-target ratios of 100:1,50:1 and 25:l; depletion was shown to abrogate completely detectable NK activity. The rat anti-mouse CR3 (CDl lb) MAb 5C6 (IgGZb, Serotec, Kidlington, UK) was injected i.p. at day -1, at a dose of 0.5 mg in 0.5 ml PBS, and subsequently 0.1 mg in 0.1 ml PBS every other day. Recombi- nant human M-CSF (a kind gift from the Genetic Institute, Cambridge, MA) was injected i.p. at day -5 at a dose of 50 pg in 0.4 ml PBS, repeated every 5 days. This dose was shown to expand peripheral-blood mononuclear-cell populations, specifi- cally monocytes, 4- to > 8-fold after 4 to 7 days of treatment.

RESULTS 11.-2 secretion by B16FIO melanoma inhihiis tumorgrowth in vivo

The mouse melanoma cell line B16F10 was transduced with the retroviral vector DC/TKIL2 or control vector DCA. A series of transduced clones were selected, and representative clones secreting physiological levels of IL-2 were chosen for further experiments. Clone, FlO/IL2Hi, secreted relatively higher levels of IL-2 (27.0 U/ml per loh cells per 24 hr) and clone FlO/IL2Lo secreted somewhat lower levels of IL-2 (13.3 Uirnl). Neither parental B16F10 cells nor B16F10 cells trans- duced with the control vector DCA (FIODC cells) produced IL-2. There was no difference in cell proliferation, measured

o io 2 0 - 30 40 i a

500

400

300. T I

I 1 60

1 1 1 , 0

0 10 20 30 40 50 W

TI T Y

0 10 20 3O 40 50 €4

Days After Tumor Injecton

FIGURE 3 -Tumor growth in C57BL/6 mice of B16F10 parental melanoma (0 in all 3 panels) or clones of B16F10 transduced IL-2 FlO/II,2Hi, in top panel) or with I F N y to express class-l- FlOIMHCI, in middle panel) or class-I- and class-II(FlO/

HMCI + 11, H in lower panel)-MHC antigens. The results of all 3 panels are those of one experiment. Tumor size was measured as described in Fi ure 1 and In "Material and Methods". Error bars represent stanfard deviation of tumor size with 7 mice per group.

256 H A M ETAL.

as doubling time during logarithmic growth phase in vitro, of FlO/IL2Hi, FlO/IL2Lo, FlOiDCA or parental B16F10 cells (data not shown).

To examine the effect of IL-2 secretion on tumor growth, 50,000 cells of FlO/IL2Hi and FlO/IL2Lo, parental B16F10, or control transduced FlODC tumors were injected S.C. into the flanks of syngeneic CS7BL/6 mice (Fig. 1 shows a representa- tive experiment, and Table I presents a summary of results of all experiments). In each experiment, control tumors (includ- ing B16F10 and FlODC tumors) appeared in most mice after 10 to 15 days and grew quickly (Fig. 1, Table I). In mice injected with FlO/IL2Lo cells, tumors appeared at approxi- mately the same time as control tumors, between 10 and 14 days, but the rate of subsequent tumor growth was consider- ably slower than that of parental and control tumors (Fig. 1). In mice injected with F1O/IL2Hi cells, tumor onset was delayed and tumors were significantly smaller than B16F10 control tumors ( p < 0.003). Approximately one quarter of mice challenged with FlO/IL2Hi cells did not have tumors at 40 to SO days (Table I). However, even after 50 days, tumors continued to develop (Table I). Rechallenge with untrans- duced parental B16F10 melanoma after rejection of F10/ IL2Hi tumors showed that rejection did not induce substantial protection against parental tumors (data not shown), consis- tent with earlier observations using B16F10 tumors secreting much higher levels of IL-2 (Dranoff et al., 1993). In mice previously challenged with FlO/IL2Hi cells, parental tumors grew as rapidly or with only a slight delay.

Rejection of B16FIO tumors secreting IL-2 The B16F10 cell line for these studies did not express

class-I-MHC molecule Kb or the class-11-MHC molecule I-Ah, and expressed low levels of the Dh molecule (Fig. 2). Low amounts of exogenous mouse interferon-? ( 2 0.1 U/ml) could

induce class-I-MHC expression, and higher levels ( 2 1 U/ml) could induce class-11-MHC antigens (data not shown). During growth in vivo, B16F10 cells up-regulated class-I-MHC mol- ecules (Fig. 2), detected at 2 weeks of tumor growth. Lack of expression of MHC molecules at early stages of growth in vivo could contribute to the tumor’s escape from surveillance by the immune system, since MHC molecules expressed by target- antigen-presenting cells are necessary for recognition by T lymphocytes.

FlO/IL2Hi cells were engineered to express class-I- and class-11-MHC molecules constitutively by transduction with a retroviral vector expressing mouse IFN-? (Watanabe et aL, 1989) (N2ICMVIFN-y vector) (Gansbacher et al., 1 9 9 0 ~ ) (6). Magnetic beads coated with MAb against MHC antigens were used to select double-transduced cells expressing class-I-MHC antigens Kh and Db, and class-11-MHC antigen LAb. Following several rounds of magnetic-bead selection, clonal populations were obtained by limiting dilution that expressed cell-surface MHC antigens. Clones were selected that constitutively ex- pressed class-I-MHC molecules (FlO/MHCI is a representa- tive clone that expressed both Kh and Db), and a clone selected that expressed both class-I- and class-11-MHC antigens (clone FlO/MHCI + 11) (Fig. 3). These clones continued to secrete the same amount of IL-2 (25-30 U/ml) as the parental cell line (FlO/IL2Hi).

Cells expressing class-I- and class-11-MHC antigens were used to further investigate a role for T-cell responses in tumor rejection mediated by in situ IL-2 secretion. No difference in tumor growth was observed for tumors constitutively express- ing class-I- and class-11-MHC antigens (FlO/MHCI and F10/ MHCI + I1 clones) as compared with FlO/IL2Hi tumors (Fig. 3). These results suggest that constitutive expression of class-1- or class-11-MHC antigen by tumor cells did not further enhance tumor rejection in vivo, suggesting that cell types

-4 B16FlO I t FlO/iL2 Hi

500i T I -i- CDB Depletion I 400

300

200

100

0 0 10 20 30 40 50 60

t F10/112Hi f t 0 -

e 2 400

a 300

200

100

0 0 10 20 30 40 50 60

Days After Tumor Injection

~

-8- 816F10

t F1011LZHi

T -A- CD4 Depleion

4 B16F10

-m- F101iLZHi I

NKcCDB

400

300

200

100

0 0 10 20 30 40 50 60

FIGURE 4 - Tumor growth in C57BL/6 mice of B16F10 parental melanoma (0 in all 4 panels) and B16F10 melanoma cells transduced to secrete IL-2 (H in all 4 panels). Mice were depleted of potential effector-cell populations using MAbs against CD4, CD8 and NKl.l antigens, as described in “Material and Methods”. Upper left panel, C57BL/6 mice depleted of CD8+ cells and challenged with FlO/IL2Hi cells. Upper right panel, C57BL/6 mice depleted of CD4+ cells and challenged with FlOIIL2Hi cells. Lower left panel, C57BL/6 mice depleted of NK1.1+ cells and challenged with FlOIIL2Hi cells. Lower right panel, C57BL/6 mice depleted of both CD8+ and NK1.1+ cells and challenged with FlO/ILZHi cells. The results from all 4 panels are from one experiment. Tumor measurements were performed as described in the legend to Figure 1 and in Methods. Error bars are standard deviation of tumor size with 6-8 mice per group.

INFLAMMATION AND TUMOR REJECTION 257

other than T cells could be involved in tumor rejection. However, we cannot rule out that these extremely low levels of IFN--y affected other cell or host functions that counteracted the effect on increased MHC expression.

Depletion of potentfial effector cells: implicating a non- T-cell population in tumor rejection

To investigate what type of effector cells participated in rejection of B16F10 tumors with in situ IL-2 secretion, tumor- bearing hosts weire depleted in vivo of selected effector-cell populations. Depletion of CD8+ and NK1.1+ cells did not significantly influence tumor rejection (Fig. 4, Table I). When NK1.1+ and CD8’ cells were depleted together, there was a trend for 1L-2-secreting tumors to grow faster in depleted than in control mice or in mice depleted of either CD8+ or NK1.1+ populations alone, but this difference was not significant (Fig. 4, Table I). These results suggest that CD8+ and NK+ cells can enhance rejection, but do not play a major role in rejection of B16F10 cells secreting IL-2. Surprisingly, significant augmenta- tion of tumor rejection was observed in mice depleted of CD4+ cells ( p < 0.005 after day 30), and most mice remained tumor-free after ti0 days (Fig. 4, Table 1). This suggests that CD4+ cells either mediate suppression of rejection or directly accelerate growth of F1OlL2Hi tumors.

Rejection of FlOII,L2Hi tumors in T-cell-deficient mice To confirm that T cells are not necessary for rejection of

1L-2-transduced E116F10 cells, FlO/lL-2 Hi cells were injected into SCID and nulnu mice. Tumor rejection was observed in SCID mice (Fig. 5, upper panel) as well as in nulnu mice (data not shown). Moreover, tumor growth in SCID mice was still effectively retarded even after depletion of asialo-GM1+ cells, which includes natural-killer-cell populations (Fig. 5 , lower panel). These results, taken together with the previous experi- ments, confirm that neither T lymphocytes nor natural killer cells are necessary for tumor rejection in this model.

Histological evidence of inflammatory response at sites of tumor rejection

Potential effector cells of rejection were examined in biop- sies taken after tuimor challenge in syngeneic C57BL/6 mice. Biopsies at sites of control-tumor challenge (B16F10 and FlOIDCA) showeld pigmented melanoma cells with essentially no mononuclear or inflammatory cell response (Fig. 6a,b), either within or at the periphery of the tumor. In contrast, sites challenged with FIOIIL2Hi tumor cells showed inflammatory responses, with infiltration of macrophages and occasional neutrophils (Fig. 6 c f ) at days 3 and 10, with associated necrosis by day 10 (Fig. 6e,f ). Infrequent residual melanoma cells were observed at the periphery of these inflammatory sites (Fig. 6c,d), suggesting that these peripheral cells escaped the inflammatory ‘response and eventually led to outgrowth of tumor. Despite heavy infiltration of macrophages at tumor rejection sites, pre-treatment of mice with a MAb against CR3 (CDllb) , an adhesion receptor on macrophages involved in migration, did not inhibit tumor rejection, suggesting that tissue macrophages rather than migrating blood monocytes might be implicated in the inflammatory response (Fig. 7).

Systemic administration of M-CSF enhances tumor rejection of IL-2-transduced melanoma cells

Histologic examination and depletion of host cell popula- tions in vivo suggested that inflammatory, non-T-cell popula- tions, including macrophages, played a role in rejection of FlOIIL2Hi tumors. To further explore a role for macrophages, we used systemic administration of the cytokine M-CSF to expand the monocyte pool in host C57BL/6 mice. Treatment with M-CSF alone did not elicit rejection of control B16F10 tumors (data not shown). However, M-CSF significantly in- creased rejection of FlO/IL2Hi tumors ( p < 0.03), further supporting a role of macrophages in rejection of IL-2-

400 ’OO1

+I- FlOlDCA

I /GI

400 5001 t F101112Hi

t NK Depletion

0 5 10 15 20 25 30 35 40 45

Days After Tumor Injection

FIGURE 5 - Tumor growth in SCID mice of B16F10 melanoma cells (0), B16F10 cells transduced with control vector (FlO/DCA, O), or B16F10 cells transduced with IL-2 (FlO/IL2Hi, m). In the lower panel, SCID mice were depleted of NK cells using antibody against asialo GM1 and challenged with FlO/IL2Hi cells. The results in the top and bottom panels are from one experiment. Tumor measurements were performed as described in Figure 1 and in “Material and Methods”. Error bars represent standard deviation of tumor size with 7 to 8 mice per group.

transduced tumor cells (Fig. 7). Two-thirds of mice treated with M-CSF remained free of tumor for more than 80 days vs. < 15% of F1OIIL2Hi mice at 60 days (Table I; and data not shown).

DISCUSSION

These experiments were initiated to study host effector mechanisms elicited by IL-2 secretion locally at tumor sites in a poorly immunogenic tumor. A primary source of IL-2 produc- tion in the organism is T cells. Therefore, this model would mimic an activated helper-T-cell response, producing an in situ reaction mimicking a limited DTH reaction. B16F10 mela- noma has many features that may reflect the poor immunoge- nicity of metastatic tumors in humans (Fidler, 1973). Because B16F10 expresses little or no MHC antigens initially after injection in vivo (although class-I-MHC antigens are up- regulated in vivo), we wished to explore the role of MHC antigens in tumor rejection in this model. For this purpose, B16F10 cells were selected, using transduction with inter- feron-y to express class-I (Kb and Db) or class-11-MHC antigens (I-Ab).

One of two outcomes was expected. In the first scenario, rejection would depend on, or be augmented by, expression of MHC antigens, implicating T cells in a primary role in tumor rejection. However, tumor rejection induced by in situ IL-2 secretion turned out to be unrelated to level of MHC expres- sion. In a second scenario, we expected that NK-cell response would mediate rejection that was perhaps dependent on lack

25 8 HARA ETAL.

FIGURE 6 - Histological examination of subcutaneous tissues of C57BLi6 mice at sites of challenge with B16F10 melanoma and B16F10 melanoma transduced with IL-2 (FlOIIL2Hi cells). (a) Low-power view ( 1 0 0 ~ ) of B16F10 melanoma showing mass of pigmented epithelioid tumor cells with no mononuclear cell infiltrate; (b) high-power view of A ( 2 0 0 ~ ) showing melanoma cells and no mononuclear-cell response; (c) low-power view ( 1 0 0 ~ ) of site of challenge with FlOIIL2Hi cells at 3 days, showing mononuclear cells surrounded by S.C. connective tissue; (d) high-power view ( 2 0 0 ~ ) of day 3 tumor in panel c of mononuclear-cell infiltrate; predominantly macrophages (thick arrow) and occasional viable B16F10 melanoma cells (thin arrow); (e) low-power view ( 1 0 0 ~ ) of site of challenge with FlOIIL2Hi at 10 days, showing central area of necrosis (N) surrounded by mononuclear cells and occasional neutrophils; (f) high-power view ( 2 0 0 ~ ) of panel e at 10 days, showing necrosis at the top border with occasional viable tumor cells (larger epithelioid cells), mononuclear cells and occasional neutrophils.

of MHC-antigen expression. NK cells express intermediate- affinity IL-2 receptors, and can be expanded and activated by IL-2 (Robertson and Ritz, 1990). More importantly, the negligible class-I-MHC expression by B16F10 would provide a potentially suitable target for NK cells, since class-I-MHC antigens expressed by target cells appear to provide a negative signal to NK cells. Surprisingly, rejection of B16/IL2Hi cells did not require NK cells.

Thus, we had to consider other possibilities. Cavallo et al. (1992) showed that rejection of an IL-2-transduced mouse

mammary-tumor-elicited rejection involved neutrophils, with only a minor effect of CD8+ T cells and no role for CD4+ T cells or NK cells. In the case of B16F10 melanoma, histologic examination showed that inflammatory responses were also elicited by IL-2 secretion, composed mainly of macrophages and occasional neutrophils. This response led to necrosis and extensive but incomplete destruction of tumor cells. Although there are similarities with the IL-2-transduced mammary tumor model of Cavallo et al. (1992), there are also several differences. First, CD8+ cells played no detectable role in

INFLAMMATION AND TUMOR REJECTION

m a 300-

6 b

5

I-‘ 200-

4 - E 100-

259

t FIOllLZHi

t CR3 Depletion

0 1 0

* TI

Days after Tumor Injection

FIGURE 7 -Tumor growth in C57BL/6 mice of B16F10 melanoma cells (0, B16 Ave) and B16F10 melanoma cells transduced with IL-2 (m, FlOIILWi). Mice were treated with MAb against CR3 (A, CR3 depletion) or with M-CSF (0) and challenged with FlOIIL2Hi cells. Tumor measurements were performed as described in Figure 1 and in “Material and Methods”. Error bars represent standard deviation of tumor size with 6 to 8 mice per group.

B16F10 rejection, and there was only minimal detectable protection against rechallenge with parental tumor even after rejection of class-I-MHC or class-I- and class-11-MHC-positive transduced tumors. Second, the early phase of rejection in the IL-2-transduced tnammary tumor model showed a classical inflammatory response, with infiltration of neutrophils at day 3, evolving to macrophages by day 10. A macrophage infiltrate appeared to dominate rejection of IL-2-transduced B16F10, even at an early phase. It is unclear how IL-2 would elicit macrophage and nleutrophil responses in the absence of T cells or NK cells, but it has been shown that macrophages and neutrophils express IL-2 receptors (Wei et al., 1993; Herrmann et al., 1985), and it i s conceivable that high local concentrations of IL-2 directly activated these inflammatory cells. Because rejection is observed in NK-depleted SCID mice, it is unlikely that there is cross-priming by T cells or NK cells to elicit a DTH-like response. The observation that M-CSF treatment augmented rejecti’on further supports a central role for macro- phages as effector cells of tumor rejection in this model.

A number of qytokines have been transduced into tumor cells in mouse model systems (including IL-2, IL-4, IL-6, IL-7, IL-12, G-CSF, tumor necrosis factor-a and interferon-y) (reviewed by Columbo and Forni, 1994). One finding is that almost any cytokinle that acts on immune or inflammatory cells can induce tumor rejection, for instance mediated by T cells (e.g., IL-2 and IL-6), natural killer cells (eg. , IL-2), eosinophils and macrophages (e.g., IL-4) or granulocytes (e.g., G-CSF). It also appears that 3 single cytokine can elicit different types of rejection, depending on the particular tumor, eg., IL-2 elicit- ing T-cell-, NK-cell- or macrophage-dependent tumor rejec- tion (Colombo and Forni, 1994). For more immunogenic tumors, expression of IL-2 by tumor cells has led to rejection that is dependent on T cells and/or natural killer cells (Fearon et al., 1990; Colombo and Forni, 1994), but this does not rule out a role for inflammatory cells such as macrophages or neutrophils. In the present study, B16F10 melanoma rejection elicited by in sitii IL-2 was not dependent on T cells or NK cells, and appeared to be mediated by macrophages and possibly neutrophiils within the local inflammatory responses, without any detectable participation of T- or NK-effector-cell populations. It is interesting to compare these results with those of Cavallo et al. (1992), where tumor rejection by local IL-2 secretion elicited a neutrophil response that was impli- cated in tumor rejection without major participation by T cells or NK cells.

There is generally a requirement for macrophages in DTH responses. This has been demonstrated in lethally irradiated mouse models, where DTH can be transferred only with lymphocytes and bone-marrow presumably as a source of monocytes (Volkman and Collins, 1971). Blood monocytes infiltrate into DTH sites, mature into tissue macrophages, and become the major infiltrating cells at the DTH site (far outnumbering transferred T cells). It is not clear how IL-2 secreted by B16F10 cells elicits this DTH-type response, but IL-2 could trigger the release of secondary cytokines and/or directly activate monocytes and neutrophils that express IL-2 receptors (Wei et al., 1993; Herrmann et al., 1985). It appears that a DTH reaction elicited by B16F10 cells secreting IL-2 can be markedly augmented by expanding the monocyte pool in the host. These results bring to mind the FBL mouse leukemia model (Greenberg et al., 1987). Rejection of disseminated FBL leukemia is dependent on CD4+ cells (perhaps as a source of local cytokine secretion, e.g., IL-2) and in vitro correlative studies suggest that macrophages are crucial for tumor rejec- tion. Our results in the B16F10 model suggest that IL-2 can be a primary cytokine to produce a DTH reaction capable of rejecting tumors. Rejection of FlO/IL2Hi cells did not lead to protection against challenge with parental B16F10 cells, a result similar to that of Dranoff etal. (1993) using B16F10 cells secreting at least 100-fold more IL-2, despite the presence of substantial numbers of tissue macrophages at sites of tumor rejection for antigen presentation. On the other hand, immuni- zation of syngeneic mice with irradiated B16F10 cells express- ing granulocyte-macrophage-colony-stimulating factor (GM- CSF) led to protection against parental tumor (Dranoff et al., 1993). Because it appears that tissue macrophages are not sufficient for immunization against B16F10 during a DTH response, it is possible that other cell types stimulated by GM-CSF (eg. , dendritic cells) are necessary for immunization against B16F10 melanoma.

ACKNOWLEDGEMENTS

This work was supported by grants R01 CA56821 and PO1 CA593.50 and CA33049 from the National Cancer Institute, Swim Across America, and the Louis and Anne Abrons Foundation.

260 HARA ETAL.

REFERENCES

ARMENTANO, D., Yu, S.-F., KANTOFF, P.W., VON RUDEN, T., ANDER- SON, W.F. and GILBOA, E., Effect of internal viral sequences on the utility of retroviral vectors. J. Wro/., 61, 1647-1650 (1987). CAVALLO, F., GIOVARELLI, M., GULINO, A,, VACCO. A,. STOPPAC- CIARO, A,, MODESTI, A. and FORNI, G., Role of neutrophils and CD4+ T lymphocytes in the primary and memory response to non- immunogenic murine mammary adenocarcinoma made immunogenic by IL-2 gene. J. Immunol., 149,3627-3635 (1992). COLOMBO, M.P., FERRARI, A. STOPPACCIARO, A,. PARENZA. M., RODOLFO, M., MAVILIO, F. and PARMIANI, G., Granulocyte-colony- stimulating-factor gene transfer suppresses tumorigenicity of a murine adenocarcinoma in vivo. J. exp. Med., 173,889-897 (1991). COLOMBO, M.P. and FORNI, G., Cytokine gene transfer in tumor inhibition and tumor therapy: where are we now? Immunology Today,

DRANOFF. G., JAFFEE, E., LAZENBY, A,, GOLUMBEK, P., LEVITSKY, H., BROSE, K., JACKSON, V., HAMADA, H., PARDOLL, D. and MULLIGAN, R., Vaccination with irradiated tumor cells engineered to secrete murine granulocyte-macrophage-colony-stimulating factor stimulates potent, specific, and long-lasting anti-tumor immunity. Proc. nat. Acad. Sci. (Wash.), 90,3539-3543 (1993). FEARON, E.R., PARDOLL, D.M., ITAYA, T., GOLUMBEK, P., LEVITSKY, H.I., SIMONS, J.W., KARASUYAMA, H., VOGELSTEIN, B. and FROST, P., Interleukin-2 production by tumor cells bypasses T-helper function in the generation of an anti-tumor response. Cell, 60,397403 (1990). FIDLER, I.J., Selection of successive tumour lines for metastasis. Nature (Lond.), 242,148-149 (1973). GANSBACHER. B., BANNERJI, R., DANIELS, B., ZIER, K., CRONIN, K. and GILBOA, E., Retroviral vector-mediated gamma-interferon gene trans- fer into tumor cells generates potent and long lasting anti-tumor immunity. Cancer Res., 50,7820-7825 (1990~). GANSBACHER, B., ZIER, K., DANIELS, B., CRONIN, K., BANNERJI, R. and GILBOA, E., Interleukin-2 gene transfer into tumor cells abrogates tumorigenicity and induces protective immunity. J. exp. Med., 172,

GREENBERG, P.D., KERN, D.E., JENSEN, M.C., KLARNET, J.P. and CHEEVER, M.A., Effector mechanisms by which adoptively transferred T cells promote tumor eradication. Prog. clin. biol. Rex, 244, 127-135 (1987).

15,48-51(1994).

1217-1224 (1990b).

HANTZOPOULOS, P.A., SULLENGER, B.A., UNGERS, G. and GILBOA, E., Improved gene expression upon transfer of the adenosine-deaminase minigene outside the transcriptional unit of a retroviral vector. Proc. nat. Acad. Sci. (Wash.), 86,3519-3523 (1989). HERRMA", F., CANNISTRA, S.A., LEVINE, H. and GRIFFIN, J.D., Expression of interleukin-2 receptors and binding of interleukin-2 by gamma interferon induced human leukemic and normal monocytic cel1s.J. exp. Med., 162, 1111-1116 (1985). KOEPPEN, H., ROWLEY, D.A. and SCHREIBER, H., Tumor-specific antigens and immunological resistance to cancer. In: R.M. Steinman and R.J. North (eds.), Mechanisms of host resistance for infectious agents, tumors and al/ografls, pp. 359-386, Rockefeller University Press, New York (1986). MARKOWITZ, D., GOFF, S. and BANK, A,, Construction and use of a safe and efficient amphotropic packaging cell line. Virology, 167, 400406 (1988~). MARKOWITZ, D., GOFF, S. and BANK, A,, A safe packaging line for gene transfer. Separating viral genes on two different plasmids. J. Mrol., 62, 1120-1124 (198%). ROBERTSON, M.J. and RITZ, J., Biology and clinical relevance of human natural killer cells. Blood, 76,2421-2428 (1990). TEPPER, R.I., PATTENGALE, P.K. and LEDER, P., Murine interleukin-4 displays a potent anti-tumor activity in vivo. Cell, 57,503-512 (1989). VOLKMAN, A. and COLLINS, F.M., The restorative effect of peritoneal macrophages on delayed-type hypersensitivity following ionizing radia- tion. Cellular Iinmunol., 2,552-561 (1971). WATANABE, Y., KURIBAYASHI, K., MIYATAKE, S., NISHIHARA, K., NAKAYAMA, E., TANIYAMA, T. and SAKATA, T., Exogenous expression of mouse interferon gamma cDNA in mouse neuroblastoma C1300 cells results in reduced tumorigenicity by augmented anti-tumor immunity. Proc. nat. Acad. Sci. (Wash.), 86,9456-9460 (1989). WEI, S., BLANCHARD, D.K., LIU, J.H.. LEONARD, W.J. and DJEU, J.Y., Activation of tumor-necrosis-factor-alpha production from human neutrophils by interleukin-2 via interleukin-2-receptor beta. J. Immu- nol., 150,1979-1987 (1993). ZIER, K., Functional and antigenic properties of cultured T-cells in the cell-mediated lympholysis assay. Hum. Immunol., 4,147-156 (1982).