Heterogeneity of cytokine production by human malignant melanoma cells Armstrong CA, Tara DC, Hart CE, Kock A, Luger TA, Ansel JC. Heterogeneity of cytokine productiotT by htimati malignant melanoma cells. Exp Dermatol 1992: 1: 37-45. Abstract: Recent investigations indicate that malignant melanoma cells can produce distinct cytokines. While differences in the production of single cytokines have been observed among different melanoma cell lines, the extent of variability in the production of single and multiple cytokines between individual tnelanoma cell lines has not been as thoroughly investigated. A heterogeneity in rnelanoma cell cytokine production could have important implications for the biology of this aggressive neoplasm since certain cytokines may act as autocrine growth factors or be potent modulators of host immune response to the developing tumor. The purpose of this study is to assess the cytokine production profile of two widely available human melanoma cell lines, A375 and G36I. The A375 cell line constitutively expressed the mRNA for IL-la, IL-lp and PDGF-A, with increased expression of these cytokities after inductioti with PMA. GMCSF mRNA was expressed by the A375 melanoma line only after induction with PMA. No IL-6 mRNA was detected in the A375 melanotna cell line. The cell culture supernatants from the A375 cells likewise contained a parallel increase in IL-1 activity as determined in the DIO bioassay and secreted GM-CSF and PDGF-AA as tneasured by ELISA. In contrast, the G361 cell line did not express IL-I, GM-CSF or PDGF-A mRNA (constitutively or after PMA induction) but expressed only lL-6 mRNA and secreted IL-6 activity after PMA induction. These results demonstrate a significant heterogeneity in the productioti of IL-la, IL-ip, lL-6, GMCSF, and PDGF in two distinct tnelanoma cell lines. This study demonstrates that individual tnelatiotna cell lities express atid secrete tnultiple cytokines both constitutively and after stimulation with PMA. The immunoinodulating and mitogenic properties of these melanoma-derived cytokines may have implications in determining the biologic behavior of different malignant melanomas.

Introduction Recent studies suggest that cytokines play an important role in regulating cellular proliferation atid immunological responses in the skin (1-3). In the epidermis, keratinocytes have been shown to produce multiple potent immunomodulating and rnitogenic cytokines including interleukin-1 (IL-1), interleukin-6 (lL-6), interleukin-8 (IL-8), plateletderived growth factor (PDGF), granulocyte macrophage colony-stimulating factor (GM-CSF), basic fibroblast growth factor (bFGF), transforming growth factor alpha (TGFa), transforming * This work was supported in part by a V, A. Merit Review Grant, The Medical Foundation of Oregon, The Oregon Division of the American Cancer Society, and the Dermatology Foundation,

Cheryl A. Armstrong''^, David C. Tara^ Charles E. Hart\ Andreas Kock"*, Thomas A. Luger'' and John C. Ansel'-^ 'Dermatology Servioe, Veterans Affairs Medical Center, ^Departmetit of Dermatology, Oregon Health Sciences University, Portland, Oregon, •"Zymogenetics Corp,, Seattle, Washington, U,S,A,, *2nd Department of Dermatology, Vienna, Austria, and 'Westfalischen Wilhelms-Universitat, MUnster, Germany

Key words: malignant melanoma: cytokines; interleukin 1; interleukin 6; platelet-derived growth factor: granulocyte macrophage colonystimulating factor Cheryl A, Armstrotig, M,Q, Dermatology Service, Veterans Affairs Medical Center, 3710 S,W U,S, Veterans Hospital Road, Portland, Oregon 97207, U.S.A, Accepted for publication 1 April 1992

growth factor beta (TGFp), tumor-necrosis factor alpha (TNFa) and imtnunosuppressive factors (1-6). Cytokine production by human melanoma cells, the malignant counterpart of epidermal melanocytes, also has been demonstrated in several recent studies. Individual tnelanoma cell lines in tissue culture have been shown to express TGFa, TGFp, bFGF, PDGF, or an autostimulatory mitogen called melanoma growth stimulatory activity (MGSA) (7-10). Several tnelanotna cell lines have been found to produce significant amounts of IL1 and IL-8 (11-14). Single tnelanoma cell lines have been reported to secrete GM-CSF and IL-6 (15-17). U is postulated that these cytokines may contribute to the development of the transformed phenotype by acting as autocrine growth factors or by modulating the host immune response to the 37

Armstrong et al. developing malignant melanoma (reviewed in 18). There is evidenee that individual melanoma eell lines ean express multiple growth faetor genes; however, the heterogeneity in gene expression between individual melanoma eell lines has not been as well eharaeterized (19-22). Additionally, previous studies have not correlated melanoma eytokine seeretion with the expression of multiple eytokine genes (22). In this investigation, we studied two widely available human melanoma eell lines, A375 and G361, for the simultaneous expression and seeretion of multiple potent eytokines: IL-la, IL-ip, IL6, PDGF, and GM-CSF. The distinct differenee in the eytokine produetion profiles of these two malignant melanotna eell lines may have important biologieal eonsequences for the development and progression of these malignant neoplasms.

ineorporated (''H) thymidine (cpm + SEM) in triplieate wells at dilutions ranging from 1:2 to 1:8192. The B9 proliferation assay was used to detertnine seereted IL-6 aetivity as deseribed previously (24). The murine plasmaeytoma eell line B9 was kindly provided by L. Aarden, University of Amsterdam, The Netherlands (25). Briefiy, the melanoma eell supernatants were dialyzed against PBS, filtered through a 0.2 \im filter, then serially diluted in triplieate in 96-well mierotiter plates. 5x lO"* eells cultured in RPMI 1640 supplemented with 5 x 10~^ M2-ME, 5% fetal ealf serum, penicillin, and streptomycin were added per well for a final volume of 200 |il. Cells were ineubated for 68-72 h then pulsed with 1 |.iCi (^H) thymidine/well for 6 h. Proliferation was measured as a funetion of ineorporated •'H thymidine (cpm + SEM) in triplicate wells at dilutions ranging from 1:2 to 1:8192.

Material and methods Melanoma cell lines

EEISA for GM-CSE and PDGE

The malignant melanoma eell lines A375 and G361 were obtained from the American Tissue Type Cell Repository (Roekville, Maryland) and maintained in tissue culture in DMEM medium (Gibco, Grand Island, New York) supplemented with 5'%) ealf serum. To investigate eytokine produetion, eells were plated in 100 mm petri dishes, grown to near confluence, and incubated with serum-free media for 12-15 hours prior to stimulation with 20 ng/ ml phorbol 12-myristate 13-aeetate (PMA) (Sigma Chemical Company, St. Louis, Missouri) for 3 or 48 h. Supernatants were removed and stored at — 20 C for use in speeifie bioassays or ELISA and cells were harvested and cellular mRNA was prepared for Northern blot analysis as deseribed below.

Northern blot analysis

Bioassays for IL-I and IL-6 Seereted IL-1 activity was measured in the DIO proliferation assay as previously deseribed with the eloned helper T-eell line, D10.G4.1, generously provided by J. Kaye, Yale University, New Haven, Connecticut (23). Melanoma cell supernatants were dialyzed against phosphate-buffered saline (PBS), filtered through a 0.2 ^m filter, then serially diluted in triplicate in 96-weU mierotiter plates in the presenee of 1.4 |ig/ml eoncanavalin A (Con A; Sigma). D10.G4.1 eells in RPMI 1640 supplemented with 5 X 10"^ M 2-ME, 5% fetal calf serum, penicillin, streptomyein, and glutatnine were added at a density of 2x IO'' eells per well for a final volume of 200 |il. Cells were ineubated for 48-52 h then pulsed for 8 h with 1 ^Ci (^H) thymidine/ well. Proliferation was measured as a funetion of 38

Secreted GM-CSF in melanoma eell supernatants was measured by using a commereially available ELISA kit (Genzyme Corp., Boston, Massaehusetts). Seereted PDGF isoforms (PDGF-AA, AB, and BB) and their eoneentrations were determined by ELISA using dimer-speeifie tnonoelonal antibodies (26),

After the A375 and G361 malignant melanoma cell lines were eultured and indueed with PMA for 3 h as deseribed above, poly A^ mRNA was prepared by the oligo (dT)-eellulose affinity ehrotnatography method (27). Briefly, cells were lysed with lysis buffer (0.2 M NaCl, 0.2 M Tris HCI pH 7.5, 1,5 mM MgClz, 2% SDS, 200 |ig/ml proteinase K), scraped off petri dishes with rubber polieemen, sheared, and ineubated for 30-90 minutes at 37"C. Cell lysates were adjusted to 0.5 M NaCl and ineubated with 0.05 g oligo (dT)-eellulose (Boehringer Mannheim, West Germany) overnight at room temperature. Samples were washed with binding buffer (0.5 M NaCl, 0.01 M Tris HCI pH 7.5) and layered over silieonized glass wool eolumns. Poly A^ mRNA was eluted with 0.01 M Tris HCI pH 7.5, ethanol-preeipitated, and quantitated. Northern blot analysis was performed. RNA was eleetrophoresed in 1% agarose gels containing 0.22 M formaldehyde, transferred to nitrocellulose (Sehleicher and Schuell, Keen, New Hampshire) and prehybridized overnight. Cytokine mRNA was deteeted by using the cDNA probes generously provided by the following sources: IL-la from P. Lomedieo, Hoffman-LaRoehe (Nutley, New Jersey), IL-ip from C. Dinarello, Tufts University/New England

Cytokine heterogeneity in human melanoma eells

Medical Center (Boston, Massachusetts), lL-6 from T. Hirano, Osaka University (Osaka, Japan), GM-CSF from S. Clark, Genetics Institute (Cambridge, Massachusetts), PDGF-A from C, Betscholtz. University of Uppsala (Uppsala, Sweden), v-sis from K, Robbins, NIH, NCI (Bethesda, Maryland), and cyclophilin (IB 15) from J, Douglass, Oregon Health Sciences University (Portland, Oregon). The cDNA cytokine probes were '-Pradiolabeled by the random priming method (28), hybridized overnight, and washed under stringent washing conditions. Blots were exposed to x-ray film at - 7 0 C for 24-72 h.

1B15

-

+

A375

Results

In this study we assessed the cytokine profile of two widely used, distinct malignant melanoma cell lines obtained from the ATCC. The two human melanoma cell lines, A375 and G361, were initially examined for expression of IL-1 mRNA. The A375 melanoma line expressed both IL-la mRNA (2.0 Kb) and 1L-1(3 mRNA (1,6 Kb) as indicated in Fig. la. However, the constitutive expression of IL-la mRNA was 20-fold greater than IL-ip mRNA in these cells. With the addition of the protein kinase C activator, PMA (20 ng/ml x 3 h), to A375 cell cultures there was little change in the

-

+

C

G361

Figure lb. Equivalent amounts of RNA in lanes 1 through 5 were verified by hybridizing eaeh blot with a '-P-Iabeled cDNA probe Ibr cyelophilin (IB15).

high constitutive expression of IL-la but a 15-fold increase in IL-ip mRNA expression as determined by scanning densitometric analysis (Fig. la, lane 2), In contrast, the G361 cell line did not express I L-l a or IL-ip mRNA either constitutively or after PMA induction (Fig. la, lanes 3 & 4), In these experiments the I L-l-producing human hepatoma cell line SK-HEP-I served as a positive control (Fig. la, lane 5), Equivalent atnounts of RNA in each lane were confirmed by hybridizing the blots used in these experiments with a "P-labeled cDNA probe for the housekeeping gene cyclophilin (1B15) as demonstrated in Fig, lb. The secretion of IL-1 activity by these cell lines was then examined in the DIO bioassay (Fig, 2), A375 cells were found

2.0kb1.6kb-

+ A375

G361

Figure la. The expression of IL-I a and IL-ip niRNA in human melanoma cells. The expression of I L - l a and IL-ip mRNA was examined in the A375 and G36I human melanoma cell lines. RNA was extracted and I L - l a and 1L-I(i expression was determined by Northern blot analysis using '-P-labeled cDNA probes specific for human I L - l a and I L-l (3. Each lane contains 10 ng Poly A ' RNA from either unstimulated cells ( —) or cells stimulated with 20 ng/ml PMA (-f) for 3 hours. A human liver adenocarcinoma cell line (SK-Hep-1) served as a positive control (C).

A375 (-)

I B 1:4

^ ^ 1:8

A37S (•)

^ ^ 1:16

G361 (-)

^M 1:32

HH 1:64

G361 (')

[ZI] 1:128

Figure 2. IL-I bioaclivily detected in supernatants of cultured human melanoma cells. The secreted IL-1 activity of A375 and G36I human melanoma cell lines was measured. Supernatants were eoUected in cells cultured either in serum-free DMEM ( —) or DMEM 4- 20 ng/ml PMA (-i-) for 48 hours. Serial dilutions, of these supernatants were assayed for lL-1 bioactivity in the DIO cell line proliferation bioassay. These dala are representative of triplicate experiments, and are expre.s.sed a.s cpm x 10'.

Armstrong et al. to secrete a significant amount of IL-1 activity cotistitutively which was increased with the addition of PMA (20 ng/ml x48 h). In addition, despite not detecting IL-1 mRNA expression in the G361 cells, low levels of IL-1 activity were detected in the supernatants of PMA-induced G361 cells (Fig. 2). These results could reflect the increased sensitivity of the DIO bioassay for the detection of IL-1 compared to Northern analysis or could be due to G361 supernatant IL-6 acting as a co-mitogen in the DIO assay as previously reported (29). Thus there is a clear quantitative difference in the production of IL-1 by these two different melanoma cell lines. The expression of IL-6 mRNA iti the A375 and G361 melanoma cell lines was then examined (Fig. 3), The A375 cell line did not express IL-6 mRNA constitutively or after PMA induction (20 ng/ml x 3 h). Likewise, the G36] cell line showed no constitutive expression of lL-6 mRNA (Fig. 3, lane 3). However, after PMA induction there was significant IL-6 mRNA detected (1.3 Kb) (Fig. 3, lane 4). Secreted melanoma 11-6 activity was measured in the B9 proliferation assay (Fig, 4), A375 cells

1.3 kb

A375

A37S (-)

A375 (•)

G361

Figure 3. The expression of IL-6 mRNA in human melanoma cells. The expression of IL-6 mRNA in the A375 and G361 melanoma ceil lines was examined. RNA was extracted and IL6 expression was delermined by Northern blot analysis using a ^^P-labeled cDNA probe specific for human IL-6. Each lane contains 10 |ig Poly A+ RNA tVom either unslimulated cells ( —) or cells stimulated with 20 ng/ml PMA ( + ) for 3 hours. A human liver adenocarcinoma cell line (SK-Hep-1) served as a positive control (C).

40

produced little IL-6 activity ( + PMA); however, a significant amount of IL-6 activity was detected in supernatants of uninduced G361 cells which was augmented after PMA induction (20 ng/ml x 48 h). Thus, although no IL-6 mRNA was constitutively apparent in G361 cells, significant secreted IL-6 activity was measured in unstirnulated G361 cells in the B9 bioassay, again reflecting the greater sensitivity of this cytokine bioassay when compared to Northern analysis. In addition, a prozone effect was apparent when IL-6 was measured in the supernatants of PMA-induced G361 cells, with an initial increase in IL-6 activity when supernatants were diluted from 1:2 to 1:16, This may be due to the high concentration of IL-6 in the induced supernatants and/or the presence of an IL6 inhibitor apparent at higher concentrations. Thus G361 but not A375 melanoma cells produce high levels of IL-6 mRNA and IL-6-secreted activity. The production of GM-CSF in the A375 and G361 melanoma cell lines was then studied (Fig, 5). No GM-CSF mRNA was detected in uninduced A375 cells, but considerable expression was apparent after PMA induction (20 ng/ml x 3 h) (Fig, 5, lanes 1 & 2). In contrast, the G361 cell line expressed no GM-CSF mRNA either constitutively or after PMA induction (Fig. 5, lanes 3 & 4). Likewise, non-induced A375 cells secreted no detectable GM-CSF whereas 1.6 pg/ml GM-CSF was detected by ELISA in PMA-induced A375 cells (data not shown). The absence of GM-CSF secretion by the G361 cell line with or without induction by PMA parallels the lack of GM-CSF mRNA expression on Northern blot analysis as above. Thus, A375 but not G361 melanoma cells produce GM-CSF.

C H 1:128

G3ei (-)

Q361 (•)

1:16

1:32

1:256

1:512

Figure 4. IL-6 bioactivity detected in supernatants of cultured human melanoma eells. Supernatants were collected from A375 and G361 melanoma cell lines eultured either in serum-free DMEM ( - ) or DMEM-t- 20 ng/ml PMA (-I-) for 48 hours. Serial dilutions of these supernaUmts were assayed tor IL-6 bioactivity in the B9 murine plasmacyloma proliferation bioassay. These data are representative of triplicate experiments, and are expressed as epm x 10'.

Cytokine heterogeneity in human melanoma cells

1.2 kb

A375

+ C361

Figure 5. The expression of GM-CSF mRNA in hun-ian melanoma eeils. The expr-ession of GM-CSF' mRNA in the A375 and G361 human melanoma cell lines was examined. RNA was extt-acted and GM-CSF expression was deter-mined by Northern blot analysis using a "P-labeled cDNA probe specific for human GM-CSF. Each lane contains 10 jrg Poly A+ RNA from either unstimulated cells ( - ) or eells stimulated with 20 ng/ml PMA ( + ) for 3 hours. A human liver adenoear-cinoma eell line (SKHep-1) served as a positive contr-ol (C).

PDGF expression in the two tnelanotna cell lities was then determined. PDGF-A tnRNA was constitutively expressed by the A375 eell line as shown

2.8

kb-

2.3

kb-

1.9

kb-

in Fig. 6, This eell line expressed the three eharaeteristie PDGF-A tnRNA speeies (2,8, 2.3, and 1.9 Kb) as teported in other eell types (Fig, 6, lane 1), After induetion with PMA (20 ng/tnl x 3 h), there was approximately a 10-foId increase in PDGF-A tnRNA expression (Fig, 6, lane 2). In eonttast, there was no expression of PDGF-A by the G361 eell line either eonstitutively or following PMA induetion (Fig. 6, lanes 3 & 4), Neither eell line exptessed PDGF-B tnRNA. PDGF-AA was deteeted in A375 cell eulture supernatants by ELISA using dimer-speeifie antibodies, which increased 2fold after PMA induction (20 ng/tnl x 48 h), again paralleling the Northern blot studies (Table 1), PDGF-AA was also deteeted in the supernatants of G361 eells (±PMA) but at 10-fold lower levels than for the A375 eells. No deteetable PDGF-AB or PDGF-BB was deteeted in the supernatants of either tnelanotna eell line. Therefore, A375 but not G361 tnelanotna eells produce sitjnifieant amounts of PDGF-AA, Thus, there is a clear differenee in the profile of eytokines produeed by the human tnelanoma eell lines A375 and G36L The A375 eell line expressed and seereted signifieant amounts of IL-la, IL-1 (3, GM-CSF and PDGF-A and the G361 line expressed and seereted signifieant amounts of IL-6 (Table 2), The possible biologieal signifieanee of these findings will be diseussed. Discussion Although individual tnalignant tnelanoma eell lines have been reported to produee a nutnber of cytokines, the produetion of tnultiple eytokines by individual tnelanotna eell lines and eell line-speeifie dilTerenees in eytokine produetion have not been appreeiated. Beeause of the potent eell growth properties and infiammatory aetivities of these eytokines, differenees in tnelanoma eytokine profiles eould have important biologieal and prognostic itnplications for the growth and developtnent of this poorly understood neoplasm. The eharTable 1. Secreted PDGF isolorms" Melanoma Cell tines

A 375

G361

Figure 6. The expression of PDGF-A mRNA in human melanoma cells. The expression of PDGF-A mRNA in the A375 and G361 hur-r-ran melanoma cell lines was examined. RNA was extt-acted and PDGF-A expressioti was detertnined by Northern blot analysis using a "P-labeled eDNA probe specilic I'or human PDGF-A. Each lane contains 10 pg Poly A' RNA from either utistitnulated cells ( —) or cells stimulated with 20 ng/ml PMA (-I-) for 3 hours. A human liver adcnocar-cinoma cell line (SKHep-I) set-ved as a positive contt-ol (C).

A375 (-) A375 (+) G361 (-) G361 (-I-)

AA 0.26 0.56 0.03 0.09

AB

BB

ND"

ND

ND

ND

ND

ND

ND

ND

' The three dimeric forms of PDGF (AA, AB. BB) were analyzed in 48-hour conditioned culture rTredia from either unstimulated human melanoma cells ( - ) Ot cells stimulated with 20 ng/ml PMA (+). The levels of fhe three PDGF dimers were determined by ELISA and are expressed as ng/ml. The level of detection for both PDGF-AB and PDGF-BB was 0.01 ng/ml. ' ND - nof detecfable.

41

Armstrong et al. Table 2, Human melanoma cell cytokine production Summary of A375 and G361 human melanoma cytokine mRNA expression IL-1 a A375 A375 G361 G361

IL-1 (3

IL-6

((+ ((+

GM-CSF -

PDGF -H

^• + + Summary of A375 and G361 human melanoma cytokine secretion

IL-r A375 A375 G361 G361

(-) (+ ((-^

-1-

-6"

GM-CSF'

PDGF-AA'

f f

++

+• +

+ •

' Secreted IL-1 activity determined by D10 bioassay of 48-hour conditioned media, " Secreted IL-6 activity determined by B9 bioassay of 48-hour conditioned media, ' Secreted protein determined by ELISA using 48-hour conditioned media.

acterization of the ability of individual tnelanoma cell lines to express a wide range of growth factor genes simultaneously in culture has recently been reported (22). These investigators found that melanoma cell lines constitutively expressed multiple growth factor genes including bFGF, PDGF-A, PDGF-B, TGF-pl, TGF-a, MGSA, IL-la, and ILlp. In this study, we have demonstrated heterogenous cytokine production of a number of other cytokines both constitutively and following PMA induction in two human melanoma cell lines. The cell line A375 constitutively expressed IL-la, ILlp, and PDGF-A mRNA. After PMA induction, this melanoma cell line also expressed GM-CSF mRNA and exhibited a significant increase in expression of IL-la, IL-1 p, and PDGF mRNA. Likewise, the conditioned media from the A375 cells contained IL-1 bioactivity and secreted PDGF-AA and GM-CSF as tneasured by ELISA. No IL-6 mRNA was detected in A375 cells and only low levels of IL-6 bioactivity were detected in A375 supernatants. In contrast, the G361 cell line expressed abundant amounts of IL-6 mRNA and secreted IL-6 activity. Low levels of IL-1 activity and PDGF-AA were detected in G36I supernatants. The finding that a single melanoma cell line can simultaneously produce these distinct cytokines suggests that these factors may interact to play a role in the development and growth of this neoplasm in vivo. We have previously reported that several other melanoma lines can produce lL-1 (11). IL-1 activity was demonstrated in supernatants from human melanoma cell lines derived from 1 primary nodu42

lar melanoma and five metastatic melanoma lines. Another study reported constitutive secretion of IL-1 activity in four of seven primary and five of 20 metastatic melanoma cell lines tested (12). The ability of two melanoma cell lines to express lL-1 tnRNA as well as a wide range of other growth factors has recently been demotistrated (22). The biological implications of IL-1 production by malignant melanoma cells are uticlear. Because of the pleotropic effects of this cytokine, melanoma-derived IL-1 could facilitate or inhibit the growth and development of these neoplasms. Support for the ability of IL-1 to facilitate melanoma growth is found in a recent report of an animal model of metastatic melanotna (30). The simultaneous injection of recombinant IL-1 and human melanoma cells in nude mice resulted in a significantly higher number of pulmonary metastases than were seen with tumor cells alone, although the mechanistn of action was not clear (30). IL-1 has beeti reported to be mitogenic for a number of cell types including keratinocytes, fibroblasts, and at least one melanoma cell line (12, 31, 32), although it has not yet been clearly established as an autocrine growth factor for melanoma in vivo. Alternatively, IL-1 secretion by melanotnas could contribute to tumor growth and dissemination by suppressing host defense mechanisms, since the injection of recombinant IL-1 has been reported to induce systemic immune suppression in mice (33). The release of IL-1 by tnelanotna cells could also mobilize host defenses against the IL-1-secretitig tumor. IL-1 is capable of activating a number of lymphoid cells including T cells, B cells, atid monocytes (34). These IL-1-activated immune cells secrete a number of other cytokines such as TNFa, interleukin-2 (IL-2), IL-6, and IL-1 itself, which have direct or indirect anti-tumor effects (34--37). Finally, IL-1 has been reported to be directly cytostatic and/or cytotoxic for melanoma cells in vitro (38, 39). There is evidence, then, that IL-l can augment or suppress melanoma cell proliferation in vitro (12, 38, 39). The quantitative differetices in the production of IL-1 by different melanoma cell lines as demonstrated in this study may also contribute to the heterogeneity of the host response to these pigmented neoplasms. Thus the role of IL-1 production by melanoma cells on the biology and the progression of this tumor is complex, difficult to predict from in vitro studies, and needs to be tested experimentally in animal models. In the present investigation we determined that a malignant melanoma cell line can express lL-6 mRNA and secrete IL-6 activity as initially reported (16). This was confirmed in another study that demonstrated that a melanoma cell line produced IL-6 after induction with recombinant IL-

Cytokine heterogeneity in human melanoma cells

la but showed no eonstitutive IL-6 secretion or mRNA expression (17). IL-6 is a pleotropie cytokine that shares many biological activities with IL1, sueh as the induetion of fever and acute-phase plasma proteins and activation of B and T cells as well as natural killer eells (37, 40, 41). A number of different epithelial and mesenehymal cells produee IL-6 (39, 40). IL-6 is a potent growth faetor for human B-eell hybridomas and its expression has been reported in various human malignant tumors (41-45). There is evidence that IL-6 may promote malignant disease as well as evidenee to support an anti-tumor effect of IL-6 in other systems. In a reeent study, when non-tumorigenie IL6-dependent murine plasmaeytoma eells were transfeeted with IL-6, the eells exhibited autoerine growth in vitro and the IL-6 cDNA transfectants were highly tumorigenie when inoculated subeutaneously in syngeneie miee (46). In the skin, IL-6 has been reported to function as an autoerine growth faetor for keratinocytes and for AIDS-Kaposi sarcoma eells (47, 48), In eontrast, in a murine model that generated lung and liver micrometastatie disease, treatment ofmiee with reeombinant IL-6 delivered intraperitoneally led to regression of established metastases without apparent toxieity (49). Sinee host immunosuppression by irradiation eliminated the tumor regression mediated by administration of rIL-6, it appeared that the IL-6 response was not a direet effect on the tumor but required a radiosensitive eomponent of the host immune system. Additionally, there is evidenee that IL-6 may act as an autocrine antiproliferative faetor for an IL-6 produeing melanoma cell line in vitro (17). Using a syngeneie animal model, we have evidence that BI6-derived murine melanoma cells transfeeted with the murine IL-6 gene form subeutaneous melanomas that are signifieantly smaller and less lethal than those formed by nontransfected melanoma eells, suggesting that IL-6 may have a potent anti-tumor effeet in this model (50). In this study we confirmed that malignant melanoma cell lines can produce PDGF. PDGF is a potent mitogen for eells of tnesenehymal origin and is ehemotactic for monocytes and neutrophils (51, 52), A wide number of eell types produee PDGF including monocytes, macrophages, endothelial cells, keratinoeytes, and some human tumor eell lines (6, 51-53). Some human melanoma cells have previously been shown to produee a PDGF-like growth factor and express mRNA for PDGF (9). Additionally, PDGF mRNA expression has been demonstrated in melanoma cell lines also expressing several other growth factors (22). Our finding that PDGF was expressed and secreted by only o n e of the two melanoma cell lines tested again

suggests that some but not all melanoma eell lines are eapable of produeing this eytokine. The A375 cell line demonstrated only A chain PDGF mRNA and seereted only PDGF-AA, whieh supports the possibility that this is the predominant PDGF gene expressed in melanoma eells as is the case in other tumor cell lines (54, 55). We have shown that one of the two human melanoma cell lines investigated expressed GM-CSF mRNA. A eorresponding inerease in GM-CSF secretion was deteeted in the melanoma eell culture supernatants. GM-CSF was first deseribed as one of the hematopoetie growth and differentiation factors with multiple biologic aetions including the induction of the maturation of undifferentiated bone marrow hematopoietie eells into mature granuloeytes and macrophages (56, 57). It is produeed by a variety of eells such as maerophages, aetivated T lymphocytes, endothelial eells, bone marrow stem eells, and keratinoeytes (56-58). Most recently, conditioned medium from the A375 cell line has been shown to contain seereted GM-CSF whieh stimulated human monoeytes and U937 eells (a human monoeytoid cell line) to produee TNF (15). While the biologic consequences of GM-CSF produetion by melanoma eells are not yet established, this report suggests that local or systemic effeets of melanoma-derived GM-CSF may lead to the activation of immature or mature granuloeytes and maerophages with antitumor properties and thus limit the growth and development of this tumor. Paradoxieally, other investigators have found that GM-CSF produetion by some tumor eell lines was immunosuppressive and aetually aeeelerated tumor metastasis (59). Additionally, GM-CSF was found to aet as an autoerine growth faetor in aeute myeloblastie leukemia eells where the growth of these eells could be bloeked by the addition of a neutralizing antiserum to GM-CSF (60). It is not known if GM-CSF has any growth-promoting activities on melanomas. In conclusion, reeent studies indieate that melanoeytie eells produce a number of eytokines and express eell surfaee reeeptors for different eytokines (18, 61). In this study, we demonstrate that individual malignant melanoma eell lines vary dramatieally in the types of cytokines they produce. Additionally, this demonstration of multiple eytokine production by a single melanoma cell line continues to expand our understanding of the interactive network of cytokines within tumor-bearing skin. In addition to potential autoerine growth and transforming effeets, these melanoma-derived cytokines have been postulated to have paraerine growth effeets on the stromal eells in the epidermis and dermis. There is evidenee that the paraerine phenotypic transformation in certain cell lines de43

Armstrong et al.

pends on multiple growth factors rather than any one specific growth factor (62). Additionally, the immunomodulating effects of a particular combination of cytokines produced in vivo could both induce and inhibit the production of cytokines by other cells, leading to variations in the host immune response to the melanoma. The heterogeneity of cytokine production among different melanoma cell lines may play a role in determining the biologic behavior of different malignant melanomas. Future studies must focus on the in vivo role of different cytokines on the development and progression of these neoplasms. In our laboratory we are currently testing the biologic importance of specific melanoma-derived cytokines in the growth and development of this neoplasm using a syngeneic animal model with an initial observation that IL-6 has a potent anti-tumor effect in this model (50), These and related studies may lead to new diagnostic and therapeutic approaches to manage this poorly understood and lethal neoplasm.

References 1. Ansel J, Perry P, Brown J, et al. Cytokine modulation of keratinocyte cytokines. J Invest Dermatol 1990: 94: I01S-107S. 2. Kupper T S. Interleukin I and other human keratinocyte cytokines: moleeular and functional characterization. Adv Dermatol 1988: 3: 293-301. 3. Rothe M, Falanga V. Growth factors: their biology and promise in dermatologie diseases and tissue repair. Arch Dermatol 1989: 125:1390-1398. 4. Kupper T S. Mechanisms of cutaneous inflammation: interactions between epidermal cytokines, adhesion moleeules, and leukoeytes. Arch Dermatol 1989: 125: 1406-1412. 5. Kock A, Schwarz T, Kirnbauer R, et al. Human keratinocytes are a souree for tumor necrosis factor alpha: evidence for synthesis and release upon stimulation with endotoxin or ultraviolet light. J Exp Med 1990: 172: 1609-1614. 6. Damm D, Shipley G, Hart C, Ansel J. The expression and modulation of PDGF-A and PDGF-B in normal human keratinoeytes. In: Oppenheim J, Powanda M C, Kluger M J, Dinarello C, eds. Moleeular and Cellular Biology of Cytokines. New York: Wiley-Li.ss, 1990: pp. 111-115. 7. DeLarco J E, Pigott D A, Lazarus J A. Ectopic peplides released by a human melanoma eell line ihat modulate lhe transformed phenotype. Proc Natl Aead Sci USA 1985: 82: 5015-5019. 8. Mcscatelli D, Presta M, Joseph-Silverstin J, Rifkin D B. Both normal and tumor eells produce basic fibroblast growth factor. J Cell Physiol 1986: 129: 273-276. 9. Westermark B, Johnsson A, Paulsson Y, et al. Human melanoma cell lines of primary and melaslalie origin express the genes encoding the chains of platelet-derived growth factor (PDGF) and produce a PDGF-like growth factor. Proc Natl Acad Sci USA 1986: 83: 7197-7200. 10. Richmond A, Lawson D H, Nixon D W, Stevens J S, Chawia R K. Extraction of a melanoma growlh-stimulatory activity from culture medium conditioned by the HsO294 human melanoma cell line. Cancer Res 1983: 43: 2106-2112. 11. Koek A, Schwarz T, Urbanski A, et al. Expression and

44

release of interleukin-1 by ditTerent human melanoma cell lines. J Natl Cancer Inst 1989: 81: 36-42. 12. Bennieelli .1 L, Elias J, Kern J, Guerry D. Produetion of interleukin I activity by cultured human melanoma eells. Cancer Res 1989: 49: 930-935. 13. Forster E, Kirnbauer R, Urbanski A, Kock A, Luger T A. Human melanoma cells produce interleukin 8 whieh functions as an autocrine growth factor. J Invest Dermatol 1991: 96: 608. 14. Zachariae C O C, Theslrup-Pedersen K, Matsushima K. Expression and seeretion of leukocyte chemotactic cytokines by normal human melanocyles and melanoma cells. J Invest Dermatol 1991: 97: 593-599. 15. Sabatini M, Chavez J, Mundy G R, Bonewald L F. Stimulation of tumor necrosis factor release from monoeytic cells by the A375 human melanoma via granulocyte-macrophage eolony-stimulating factor. Caneer Res 1990: 50: 2673-2678. 16. Berger C A, Damm D C, Koek A, Ansel J C. Heterogeneity of cytokine production by human malignant melanoma cells. J Invesl Dermatol 1990: 94: 407. 17. Morinaga Y, Suzuki H, Takatsuki F, et al. Contribution of IL-6 lo the anliproliferalive effect of IL-1 and tumor neerosis factor on lumor cell lines. J Immunol 1989: 143: 3538-3542. 18. Kock A, Schwarz T, Micksche M, Luger T A. Cyiokines and human malignant melanoma immuno- and growthregulatory peptides in melanoma biology. In: Nathanson L. ed. Melanoma Research: Genetics, Growth Factors, Melastases, and Anligens. Boston: Kluwer Academic Publishers, 1991: pp. 41-66. 19. Pichon F, Lagarde A E. Auloregulation of MeWo metastalic melanoma growth: characterization of intracellular (FGF, MGSA) and seereted (PDGF) growth factors. .1 Cell Physiol 1989: 140: 344-358. 20. Lizonova A, Bizik J, Grofova M, Vaheri A. Coexpression of lumor-associaled a^-macroglobulin and growth factors in human melanoma cell lines. J Cell Biochem 1990: 43: 315-325. 21. Chevenix-Trench G, Martin N G, Ellem K A O. Gene expression in melanoma cell lines and cultured melanocyles: correlation between levels of c-src-1, e-myc and p53. Oncogene 1990: 5: 1187-1193. 22. Rodeck U, Melbcr K, Kath R, et al. Constitutive expression of multiple growth factor genes by melanoma cells but not normal melanocytes. J Invest Dermalol 1991: 97: 20-26. 23. Kaye .1, Gillis S, Mizel S B, et al. Growth of a cloned helper T cell line induced by a monoelonal anlibody specific for the antigen receptor: interleukin I is required for the expression of receplors for interleukin 2. J Immunol 1984: 133: 1339-1345. 24. Kirnbauer R, Koek A, Schwarz T, el al. INF-P2, B cell differentiation factor 2, or hybridoma growlh factor (lL-6) is expressed and relea.sed by human epidermal cells and epidermoid carcinoma cell lines. .) Immunol 1989: 142: 1922-1928. 25. Aarden L A, DeGroot E R, Schaap O L, Lansdorp P M. Production of hybridoma growth factor by human monoeytes. Eur J Immunol 1987: 17: 1411. 26. Hart C E, Bailey M, Curtis D A, et al. Purification of PDGF-AB and PDGF-BB from human platelet extracts and identifiealion of all three PDGF dimers in human platelets. Biochemistry 1990: 29: 166-172. 27. Badley J E, Bishop G A, St. John T, Frelinger J A. A simple, rapid melhod for lhe puril'ieation of poly A ' RNA. BioTechniqties 1988:6: 114-116. 28. Feinberg A P, Vogelstein B. A teehnique tor ladiolabeling DNA restriction endonuclease fragments to high specific activity. Anal Bioehem 1983: 132: 6-13.

Cytokine heterogeneity in hnman melanoma cells 29. Mizutaiii H, May L T, Scligal P B, Kupper T S. Synergislic inleraclion.s ol" lL-1 and IL-6 in T cell acliviition: mitogeii tiul nol antigen receptor-induced proliferation of cloned T helper cells is enhanced by exogenous IL-6. .1 Immunol 1989; 143: 896 901. 30. Giavazzi R, Garofalo A, Bani M R, et al. Interleukin 1indueed augmentation of experimental metastases from a human melanoma in nude mice. Cancer Res 1990: 50: 4771_4775. 31. Ristow H F. A major factor contriliuling to epidermal proliferation in inflammatory skin diseases appears to be interleukin 1 or a related protein. Proc Natl Acad Sci tJSA 1987: 84: 1940-1944. 32. Schmidt .1 A, Mizel S B, Cohen D, Green 1. Interleukin I, a potential regulator of fibroblast proliferation. J Immunol 1982: 128: 2177-2182. 33. Robertson B, Gahring L, Newton R, Daynes R. In vivo administration of interleukin I to normal mice depresses their capacity to elicit contact hypersensitivity responses: prostaglandins are involved in this modirication ol immune function. .1 Invest Dermalol 1987: 88: 380-387. 34. Dinarello C A. lnterleukin-l and its biologically related eytokines. Adv Immunol 1989: 44: 15.3-205. 35. Sugerman B .1, Aggaiwal B B, Hass P E, Figari 1 S, Palladino M A, Shepard H M. Recombinant human tumor necrosis factor-a: effects on proliferation of nonnal and transformed cells in vitro. Science 1985: 230: 943-945. 36. Rosenberg S A, Lotze M T, Mull L M, et al. Observations on the systemic administration of autologous lymphokineaetivated killer cells and recombinant interleukin-2 to patients with metastatic cancer. N lingl .1 Med 1985: 313: 1485-1492. 37. Luger T A. Krutmann .1, Kirnbauer R, et al. lFN-p2/lL6 augments the aetivity of human natural killer cells. .1 Immunol 1989: 143: 1206-1209. 38. Lachman L B. Dinarello C A, Llansa N D. Fidler I J. Natural and reeombinant human interleukin 1-P is cytotoxic for human melanoma cells. .1 Immunol 1986: 136: 3098-3102. 39. Onozaki K, Matsushima K, Aggaiwal B B, Oppenheim .1 J. Human interleukin 1 is a cytocidal factor for several tumor cell lines. .1 Immunol 1985: 135: 3962-3968. 40. Sehgal P B, May L T. Human interferon-p2. .1 tnterferon Res 1987: 7: 521-527. 41. Yasukawa K, Hirano T, Watanabe Y, et al. Structure and expression of human B eell stimulatory factor-2 (BSF-2/ IL-6) gene. EMBO .1 1987: 6: 2939-2945. 42. Wong G H W, Goeddel D V. Tumor necrosis faetor-alpha and -beta inhibit virus replication and synergize with interferons. Nature 1986: 323: 819-822. 43. Van Snick .1, Cayphas S, Vink A, et al. Purification and NH2-terminal amino acid sequence of a T-cell derived lymphokine with giowth factor aetivity for IJ-eell hybridomas. Proc Natl Acad Sci USA 1986: 83: 9679-9683. 4 4 Kawano M, Hirano T, Matsuda T, et al. Autocrine generation and requirement of BSF-2/1L-6 for human multiple myeloma. Nature 1988: 332: 83-85. 4 5 . Oster W, Cicco N A, Klein H, et al. Partieipation ol the cytokines interleukin 6, tumor necrosis factor-alpha, and interleukin 1-beta seereted by aeute myelogenous leukemia blasts in autocrine and paraerine leukemia growth control. J Clin Invest 1989: 84:451-157.

46. Tohyama N, Karasuyama H, Tada T. Growth autonomy and tumorigenicity of interleukin 6-dependent B cells transfeeted with interleukin 6 cDNA. J Exp Med 1990: 171: 389-400. 47. Grossman R M, Krueger J, Yourish D, et al. Interleukin 6 is expressed in high levels in psorialic skin and stimulates proliferation of cultured human keratinocytes. Proe Natl Acad Sci USA 1989: 86: 6367-6371. 48. Miles S A, Rezai A R, Salazar-Gonzalez .1 F, et al. AIDS Kaposi sarcoma-derived cells produce and respond to interleukin 6. Proe Natl Aead Sei USA 1990: 87: 40684072. 49. Mule .1 J, Mclntosh J K, .lablons D M, Rosenberg S A. Antitumor activity of recombinant interleukin 6 in mice. .1 Exp Med 1990: 171: 629-636. 50. Armstrong C A, Koppula S V, Tara D C, Ansel J C. The effeet of melanoma-derived interleukin 6 on melanoma arowth and development in a murine model. .1 Invest Dermatol 1991: 96: .S48. 51. Heldin C t-1, Westerniark B. Growth faetors: mechanism of action and relation to oncogenes. Cell 1984: 37: 9-20. 52. Ross R. Platelet-derived growth faetor. Ann Rev Med 1987: 38: 71-79. 53. WiUiams LT. Stimulation of paraerine and autocrine pathways of cell proliferation by platelel-derived growth factor. Clin Res 1988: 36: 5-10. 54. Betsholtz C, .lohnsson A, Heldin C-H, et al. eDNA sequenee and chromosomal localization of human plateletderived growth factor A-chain and its expression in tumor cell lines. Nature 1986: 320: 695-699. 55. Johnsson A, Betsholtz C, Heldin C-H, Westermark B. Antibodies against platelet-derived growth faetor inhibit acule transformalion by simian sarcoma virus. Nature 1985: 317: 438^40. 56. Metealf D C, Begley C G, .lohnson G R, et al. Biologic properties in vitro of a recombinant human granuloeytemaerophage eolony-stimulalinc faetor. Blood 1986: 67: 37-45. 57. SielT C A, Emerson S G, Donahue R E, et al. Human reeombinant granulocyte-maerophage colony-stimulating factor: a multilineage hematopoietin. Science 1985: 230: 1171-1173. 58. Kupper T S, Lee F, Coleman D, Chodakewitz .1, Flood P, Horowitz M. Keiatinoeyte derived T-cell growth factor (KTGF) is identical to granuloeyte maerophage colony stimulating faetor (GM-CSF). .1 Invest Dermatol 1988: 91: 185-188. 59. Tsuehiya Y, Igarashi M, Suzuki R, Kumagai K. Production of eolony-stimulating faetor by tumor cells and the faetormediated induction of suppressor cells. .1 tmmunol 1988: 141: 699-708. 60. Young D C, Griffin .1 D. Autocrine seeretion of GM-CSF in aeute myeloblastie leukemia. Blood 1986: 68: 1178-1181. 61. Herlyn M, Clark W H. Rodeck U, Maneianti L, Jambrosic .t, Koprowski H. Biology of Disease: biology of tumor progression in human melanocytes. Lab Invest 1987: 56: 461-474. 62. van Zoelen E .1 .1, van Oostwaard T M J, de Laat S W. The role of polypeptide growlh factors in phenotypic transformation of normal rat kidney cells. } Biol Chem 1988: 263: 64-68.

45