The effects of granulocyte-macrophage colony-stimulating factor on tumour-infiltrating lymphocytes from renal cell carcinoma.

It has been shown that granulocyte-macrophage colony-stimulating factor (GM-CSF) can induce specific and non-specific anti-tumour cytotoxicity and also stimulates the proliferation and function of peripheral lymphocytes and thymocytes. GM-CSF and interleukin 2 (IL-2) act synergistically on peripheral lymphocytes for the induction of a highly effective cytotoxic cell population. Thus, the goal of our investigation was to study the effects of GM-CSF upon expansion, proliferation and in vitro killing activity of tumour-infiltrating lymphocytes (TILs) from renal cell carcinoma (RCC). TILs from seven consecutive tumours were cultured with GM-CSF (500 or 1000 nmol ml-1) without IL-2 supplementation, with suboptimal doses of IL-2 (8 and 40 U ml-1) plus GM-CSF (1000 nmol ml-1), and with a dose of IL-2 (400 U ml-1) which sufficed alone to induce TIL development plus GM-CSF (500 or 1000 nmol ml-1). GM-CSF alone or together with suboptimal doses of IL-2 was not able to induce or facilitate TIL development in these cultures. When GM-CSF at both concentrations studied was added to optimal doses of IL-2 the resulting TIL populations proliferated significantly better and faster (+66%), resulting in a higher cell yield (+24%) at the time of maximal expansion of the TIL cultures. The length of the culture periods of TILs was not affected by GM-CSF when compared with the control cultures supplemented with IL-2 alone. In vitro killing activity of TIL populations stimulated with IL-2 and GM-CSF remained unspecific, but lysis of the autologous tumour targets as well as the allogeneic renal tumour targets was significantly enhanced (+138%) as compared with the corresponding control TILs stimulated with IL-2 alone. Lysis of the natural killer (NK)-sensitive control cell line K562 and the NK-resistant Daudi cell line remained unchanged even though FACS analysis of TILs cultured with IL-2 and 1000 nmol of GM-CSF demonstrated a significantly higher proportion of cells expressing the CD56 molecule (+50%). Specific receptors for GM-CSF could not be demonstrated on TILs from RCC. Our data demonstrate that GM-CSF alters the biological behaviour of IL-2-activated TILs from renal cell carcinoma in terms of proliferation, in vitro killing activity and cell-surface molecule expression.(ABSTRACT TRUNCATED AT 400 WORDS)

Granulocyte-macrophage colony-stimulating factor (GM-CSF) is a multilineage glycoprotein cytokine which is synthesised by a variety of cell types, such as T and B lymphocytes (Chan et al., 1986;Herrmann et al., 1988;Pluznik et al., 1989), macrophages (Thorens et al., 1987;Fibbe et al., 1988), fibroblasts (Kaushansky et al., 1988) and endothelial cells (Sieff et al., 1987). Expression of GM-CSF has also been documented in certain solid tumours (Zinzar et al., 1985), and myeloid leukaemia cells are also believed to be a potential pathophysiological source of this cytokine (Oklamura et al., 1988;Fiedler et al., 1990). We have recently shown that tumour-infiltrating lymphocytes (TILs) from renal cell cancer (RCC) are able to release a wide array of various cytokines, including GM-CSF (Steger et al., 1994). GM-CSF acts as a potent growth factor both in vitro and in vivo, stimulating proliferation and maturation of myeloid progenitor cells, giving rise to neutrophilic and eosinophilic granulocytes and monocytes (Begley et al., 1988;Lopez et al., 1986;Metcalf et al., 1986;Silberstein et al., 1986;Kaufman et al., 1989).
GM-CSF is involved in the host defence mechanisms and is a potent factor in activating macrophages for tumour cell killing. Activated macrophages can be non-specifically cytotoxic for tumour cells in an MHC-independent fashion (Fidler and Schroit, 1988). They also can specifically recognise tumour cells in vivo, thus playing an important role in host surveillance against autochthonous transformed neoplastic cells (Fidler, 1985). Vaccination with irradiated tumour cells engineered to secrete murine GM-CSF has been shown to induce specific anti-tumour immunity (Dranoff et al., 1993). TILs from RCC are able to secrete GM-CSF upon stimulation (Steger et al., 1994), and this ability to secrete GM-CSF upon autologous tumour stimulation was recently shown to correlate positively with the clinical response after TIL immunotherapy in melanoma patients (Schwartzentruber et al., 1994). In addition, there are reports that GM-CSF can stimulate proliferation or function of T-cell lines. It has been demonstrated that GM-CSF can serve as a growth factor for the IL-2/IL4 dependent T-cell line HT-2, acting through a pathway which is distinct from that of IL-2 or IL4 (Kupper et al., 1987;Woods et al., 1987). Herbelin et al. (1989,1990) reported that GM-CSF and IL-1 act synergistically to stimulate thymocyte proliferation via an IL-2-independent pathway. Santoli et al. (1988) demonstrated that GM-CSF can support the growth of cells within the lymphoid lineage and exert potent amplifying effects on IL-2-induced T-cell growth in vitro. Moreover, in a recent evaluation of GM-CSF by Masucci et al. (1990), it was demonstrated that GM-CSF and IL-2 act synergistically on peripheral lymphocytes with the induction of a highly effective cytotoxic cell population.
IL-2-based immunotherapy and adoptive immunotherapy with in vitro-activated lymphocyte-activated killer (LAK) cells and TILs are increasingly applied in the therapy of human solid tumours Topalian et al., 1988). RCC is one of the more extensively investigated human cancers in which these novel forms of anti-cancer therapy have shown activity ( Weiss et al., 1992). In recent publications our group has demonstrated that several cytokines, such as IL-4 (Tso et al., 1992), IL-6 (Lee et al., 1991) and IL-7 (Ditonno et al., 1992), are able to modulate TILs derived from RCC under certain culture conditions. This fact suggests that a variety of cytokines, and perhaps growth factors as well, are involved in the activation of tumour-derived and specific immunocompetent cells, rather than 1L-2 alone.
Based on the mentioned reports demonstrating some activity of GM-CSF on T cells, we designed experiments to investigate the influence of GM-CSF on TILs from RCC. T'he results of these experiments show that GM-CSF, when added to optimal concentrations of IL-2, has emkarkable modulatory effects on the growth, expansion, proliferation and the in vitro cytolytic activity of RCC TILs. Possible clinical implications will be discussed.

Mateil and mhos
Lymphocyte cultures TILs were cultured from the primary tumour of seven patients with RCC. Tumours were obtained from the operating room, minced into small pieces and enzymatically digested overnight in RPMI-1640 culture medium (Celigro, Mediatech, Washington, DC, USA) containing 0.01% hyaluronidase type V, 0.002% DNAse type I, 0.1% collanase type IV (Sigma, St Louis, MO, USA), 2 mM Lglutamine (Gibco, Grand Island, NY, USA) and 50 Lg ml-' gentamicin. The resuting sing-cell suspensions were then passed over single-step Ficoll-Hypaque density gradients (LSM, Organon Tenika, Durham, NC, USA). The mixture con ng both TILs and tumour cells rieved from the gradient interfaces was washed, counted and either cryoprsrved for use as targets in cytotoxicity assays or cultured in six-well tissue culture plates (Costar, Cambridge, MA, USA, or Falcon, Becton Dickinson Labware, Lincoln Park, NJ, USA) at a density of 0.5 x 106 cellsml-' in medium consisting of RPMI-1640 plus 10% heat-inactivated human AB serum (Irvine Scientific, Santa Ana, CA, USA), 50IUmlI penicillin, 50tg ml ' streptomycin (JHR Bioscences, Lenexa, KS, USA) and 2mM L-glutmjn.
Culture conditions GM-CSF (Sandoz, Basle, Switzrland) at concentrations of 500 nmol ml-' and 1000 nmol ml-' was added to the singlecell suspensions, either alone or together with various concentrations of interluin 2 (8, 40, 400 U ml-'; Hoffmann La Roche, Nutley, NJ, USA; 400 U m1' equals 1000 Cetus U ml-' or 6000 IU ml-'). TIL cultures supplemeted with the same concentrations of IL-2 alone served as controls. All cell cultures were maintained at 37C and 5% carbon dioxide and passaged as needed to maintain a concentration of 1 x 10' to 1.5x 106 TILsml-. Proliferation To determine the proliferation of TIL cultures [3HI1 thymidine uptake assays were performed. AWroximately 5 x 10' TILs per well were cultured for 4 days in 96-well flat-bottom microtitre plates (Costar) in 100p1 of complete medium supplmented with GM-CSF and/or IL-2 at concentrations cited in the text. Triplicate wells were pulsed with 0.5 jCi of [3HJTdR (Dupont, Boston, MA, USA) for 24 h and then harvested for scintillation counting using a PHD

Cytotoxicity assays
The cytotoxic activity of TILs grown in IL-2 and/or GM-CSF was tested in. vitro in a standard 4 h 5'Cr release assay against fresh (cryopreserved) autologous tumour, one allogeneic tumour target (TU 59), K562, a NK-sensitive erythroleukaemia cell line, and Daudi cells, a NK-resistant lymphoma. A total of 5 x 107 target cells in a vohlume of 2 ml were labelled with 200 pCi of 5'Cr (ICN Radiochemica, Irvine, CA, USA) for 1 h at 3TC and washed three times before use. The 5 x 103 targets and the appropriate number of effectors at seveal effector-target (E/T ratios (40:1, 20:1, 10:1 and 5:1) were plated in triplicate in a total of 0.2 ml of medium in 96-well round-bottom microtitre plates. After a 4 h incubation period the plates were centrifuged at 800 r.p.m. for 3 min and 100 p1 of the supernatant was harvested, and counted on a gamma counter. The percentage specific lysis was determined as: Experimental countsspontaneous counts Total countsspontaneous counts x 100 Target cells incubated in medium alone and with 1% sodium dodecyl sulphate were used to determine spontaneous and maximal relase of chromium respectively. Cytotoxicity is expressed as lytic units (LU) per 1 x 10' cells. One lytic unit is defined as the number of effector cells needed to lyse 30% of 5 x 103 target cells.

GM-CSF receptor assays
For the detection of cell-urface receptors specific for GM-CSF a previously published ligand-binding assay with 1'Ilabelled GM-CSF was used (DiPersio et al., 1988).

Statistical anaysis
The significn of differences in number of lytic units in assay and differences in percentages of positive cells of FACS analysis was determined the Wilcoxon signed-rank test A P-value <0.05 was conside to indicate signia and two-tailed P-values were used.

Redls
Growth, expansion andproliferation The results of growth, maximal expansion and days in culture of TIL cultures under the various culture conditions are depicted in Table I. GM-CSF alone and when added to suboptimal concentrations of IL-2 (8, 40 U ml') failed to induce TIL growth in all cultures. Adding 500 nmol ml-' or 1000 nmol ml-' GM-CSF to the cell cultures together with a concentration of IL-2 which was alone sufficient to induce TIL growth (400 U ml-') resulted in the development of TILs. The expansion of TILs cultured in IL-2 + GM-CSF was significantly higher at both concentrations of GM-CSF when compared with TILs grown in IL-2 alone. The culture period of all TIL cultures did not differ regardless of the culture conditions.
The results of the [3HJTdR incorporation assays can be seen in Table II. TILs incubated with IL-2 and GM-CSF at both concentrations studied proliferated significantly better than TILs grown in IL-2 alone.
Cytolytic activity Four hour 5"Cr-release assays were performed in all seven TIL populations at an early stage of culture (days 25-45) and at later stage of culture (days 53-72). Fresh (cryopreserved) autologous and the allogeneic renal target cells (TU 59) were available for all experiments. At an early stage of culture TILs grown with IL-2 and GM-CSF showed an enhanced killing activity against the autologous tumour target as well as against the allogeneic renal tumour target (Table III). The killing behaviour against the NK-cell sensitive K562 target and the NK-resistant Daudi cell target was unchanged when compared with the killing behaviour of corresponding TIL cultures activated with IL-2 alone. Killing of all TIL populations tested was always non-specific, as the allogeneic renal target, K562 and Daudi cells were lysed equally well or better independently of the culture condition. The same pattern in killing was observed at the second evaluation at a later stage of the cultures (data not shown), and killing remained also nonspecific.
Phenotypical analysis Phenotypical analysis of TIL cultures (Table IV) supplemented with IL-2 or GM-CSF and IL-2 revealed similar percentages of cells positive for CD3 and CD3/CD56 respectively. TILs grown with 1000 nmol ml-' GM-CSF and IL-2 showed a significantly higher percentage of CD56+ cells, while expression of CD56 was unchanged in TILs activated with 500 nmol ml-' GM-CSF + IL-2 when compared with TILs activated with IL-2 alone. The percentages of CD8+ and CD4+ cells were similar independent of the culture conditions. Expression of the IL-2 receptor (CD25) was unaffected by GM-CSF.

GM-CSF receptor analysis
Two TIL populations were analysed for their ability to express the GM-CSF receptor. One population was activated with IL-2 400 U ml-' alone and the other with IL-2 400Uml' +GM-CSF 100OnmolmlV'. In neither of the   E d mICSu -TUs GG Sg etaf two TIL populations tested could GM-CSF receptors be detected.
In summary, the addtion of GM-CSF to optimal concentrations of IL-2 resulted in a 25% increase in expansion, in a 66% increase in thymidine incorporation and in a 50% increase in CD56 expression of TILs from RCC. Moreover, there was a 138% increase in killing capacity of the tested IL-2/GM-CSF TILs against the autologous and allogeneic renal tumour targets, while the killing behaviour against the NK-sensitive cell line K562 and the NK-resistant cell lne Daudi remained unchanged or tended to be lower.
GM-CSF exerts a wide array of biological actitities on many cell types. Besides its stimulatory function on the proliferation of immature progenitors, it was soon recognised that GM-CSF could also enhance differentiated functions of mature effector cells (Lopez et al., 1983;Vadas et al., 1983;Weisbart et al., 1985Weisbart et al., , 1986. Although still somewhat controversial, some wel-designed in vitro studies have ckarly demonstrated that the activity of GM-CSF is not resticted to monocytes/macrophage and granulocytes. Also, the proliferation and growth of T cells (Kupper et al., 1987;Woods et al., 1987;Santoli et al., 1988;Herbelin et al., 1989Herbelin et al., , 1990 as well as their in vitro killing behaviour (Masucci et al., 1990) can be modulated by GM-CSF.
The first goal of our study was to investigate the ability of GM-CSF to induce TIL development from single-:ell suspensions derived from RCC. GM-CSF at both concentrations (500 nmol ml-', 1000 nmol ml-') investigated failed to induce TIL development from RCC specimens when used as single activator. Also, when GM-CSF was added to tumour/ lymphocyte suspensions together with suboptimal concentrations of IL-2 (8, 40 U ml-'), no TIL development was observed. Only when the primary cell cultures were supplmented with GM-CSF and concentrations of IL-2 (400 U ml-') which sufficed alone for the activation and expansion of TILs was TIL development observed. Thus, GM-CSF cannot be asumed to be an independent growth factor for TILs derived from RCC, nor does GM-CSF facilitate the development of TILs when suboptimal doses of IL-2 are used.
However, TILs activated with GM-CSF and IL-2 differ signifiantly, in terms of proliferation, expansion and in vitro kilng behaviour, from TILs activated with EL-2 alone. TILs grown with GM-CSF + IL-2 proliferated better than TILs activated with IL-2 alone, but the possible culture period was not affected. This enhancement of proliferation, coupled with similar time periods for which TILs could be maintained in culture, resulted in a snificantly higher and more rapid expansion for TILs grown with either 500 umol ml-' or 1000 umol ml-' GM-CSF and IL-2. These results are in good agreement with the limited data available for peripheral T cells. Santoli et al. (1988) have demonstrated that GM-CSF enhances the short-term responsiveness of peripheral T cells to IL-2, and GM-CSF also potentiates the long-term growth of non-activated human lymphocytes and of lectinand Agactivated T cells in the presence of IL-2. Although in clnical investigations the number of activated cells reinfused to the patients and the effectiveness of treatment demonstrates no correlation thus far, most dinical protocols require the expansion of TIL cultures to at kast 1 x 109 to 1 x 101" cls.
Such cell counts are usually reached within 4-6 weeks of culture . Thus, this higher proliferation rate of TILs resulting in high cell counts when GM-CSF + IL-2 are used for activation would shorten the culture period, allowing an earlier onset of adoptive immunotherapy after surgical removal of the primary tumour.
Unlike TILs derived from melanoma (Itoh et al., 1986;Muul et al., 1987), the killing behaviour of TlLs derived from RCC is non-specific in general, yet certain clones of RCC TILs have been isolated and exert autologous tumourspecific cytotoxicity Schendel et al., 1993).
TILs activated with GM-CSF and IL-2 showed a different killing behaviour in vitro when compared with the corresponding TIL cultures activated with IL-2 alone. The addition of GM-CSF to the culture medium resulted in enhanced killng of the autologous tumour target and the allogeneic renal tumour target. In contrast, lysis of the NK-sensitive K562 erythroleukaemia cell line and the NK-reistant Daudi lymphoma cell line ained unaffected. The percentag of CD3+, CD3+/CD56+, CD4+ and CD8+ cells were similar in all TIL populations independent of the culture condition.
The higher percentage of cells positive for the NK marker CD56 in TIL populations cultured with high concentrations of GM-CSF and IL-2 appears not to be responsible for the demonstrated enhanced kIilling, since TIL cultures supplmented with the lower concentration of GM-CSF also showed enhanced klling and a similar percentage of these cells were CD56+ when compared with TILs activated with IL-2. Furthrmore, the killing activity against the NKsenstive K562 cell hne rained uaffected independent of the concentration of GM-CSF used. The differences in the pattern of target lysis between TEL populations activated with IL-2 or GM-CSF + IL-2 were maintained over time in our experiments. Despite this of lytic activity of the autolgous tumour target the killing behaviour of the tested TIL populations was always non-specific as the allogeneic targets were always lysed equally well or better.
However, the fact remains that GM-CSF is able to enhance the lytic activity of RCC TILs against renal targets only.
These data are in part imilar to findin rding the ability of GM-CSF to induce LAK-cell activity in peripheral Iymphocytes. Masucci et al. (1990) demonstrated that p l lymphocytes activated with IL-2 and GM-CSF lysed Daudi cells and the human colorectal caer cell line SW918 signcntly better than IL-2-activated LAK cels In that study, a 10-fold lower dose of IL-2 was required when GM-CSF was added as compared with IL-2 alone to generate a cytotoxic cell population with the same lytic activity. The authors asumed that GM-CSF might render more cells susceptible to IL-2 stimulation, since a higher cell fraction expraessi CD25 when stimulated with IL-2 and GM-CSF.
However, this is not rlted in our results, since the expression of CD25 in RCC TILs was similar whether or not GM-CSF was added to the medium. The mode of action of GM-CSFs activity on lymphocytes in general and TILs in particular remains to be elucidated.
Although the presence of a specific receptor might be assumed and the murine receptor specific for GM-CSF has been demonsta in two cell ines of T-lymphoyte origin (Park et al., 1986), the presence of the GM-CSF receptor on human lymphocytes has not been thoroughly investigated and has not as yet been demonsta (Gasson, 1991). We were not able to demonstrate the presei c of receptors secific for GM-CSF on the surface of mature RCC TILs. Although the expression of low nmmbers of the GM-CSF receptor on a minor subfraction of the TIL populations invesited might have been undetectable with the ligandbindng assay used, the pathway through which GM-CSF modulates T-cell actions, or at least TIL actions, appears to be an indiret rather than a direct one. After tumour processing the singkl l suspensions also contain macrphages and other mononuclear cells. Since GM-CSF has been shown to activate M opges to enhance non-specific and specific immune reonses tgainst tumour cells, it might be speculated that one of these indirect effects could be the activation and stimulaton of cells other than lymphocytes to reles cytokines with T-cell-activating propertes In summary, GM-CSF is not able to induce TIL development from RCC or to facitate TIL activation induced by IL-2. However, TILs from RCC cultured with IL-2 and GM-CSF demonstrate snificantly higher proliferation, resulting in a higher expansion of TIL cultures, and exert a higher killing activity against renal tumour targets in vitro.
These findings provide a rational basis for the use of GM-CSF in the expansion of TILs, and further investigations are warranted. Clinical experience with systemic GM-CSF is more or less lhimted to the us of this cytokie to shorten chemotherapy-induced leucpenia. Unlike sysemic IL-2 administration, GM-CSF application is rarely associated with serious sideeffcts (Morstyn et al., 1988;Horn et al., 1991;Steger et al., 1992). Based on our results and the known properts of GM-CSF in improving host defence in immunooompromised patients by means of enhancd cytokine release and enhancement of cytolytic activity of neutrophils, eonnophils and macrophages (Weisbart, 1989), one might also speculate that GM-CSF could be of therapeutic value when admin lly to patints reciving IL-2-based adoptive immunotherapy with TLs. The mode of action of GM-CSF in enhancing T-cell-mediated cytotoxic effects is not completely understood. Further in vitro and in vivo investigations with TILs and GM-CSF are needed to elucidate these issues. GG Steger is a 1991 Max Kade Foundation grant reipient