In vitro modulation of human and murine melanoma growth by prostanoid analogues.

The inhibitory effect of various prostaglandin analogues on the anchorage independent growth of murine and human melanoma cells was measured. PGA analogues (which were modified at C-16 and C-18) did not demonstrate any major improvement in activity over PGA alone. These included 16.16-dimethyl PGAl, 16,16-dimethyl-PGA2, 16,16-dimethyl-18-oxa-PGA2 and trans- h-2-15- d acetoxy-16,16-dimetiiyl-18-oxa-ll-deoxy-PGEl-methylester. The thromboxane synthetase inhibitor, U51605, demonstrated weak dnti-proliferative activity. PGO2 (with a ketone at C-11 versus C-9 for PGA and PGE) was the most potent prostaglandin tested. Cells from melanoma lines displayed species differences in their sensitivities. PGAl and PGEl dere the most potent inhibitors of the anchorage independent growth of murine melanoma cells. On human melanoma cells PGD2 was the most active prostaglandin, 2-3 times more potent than PGAl; PGEl was a very weak inhibitor.


INTRODUCTION
Prostaglandins, especially the A :?ries (l-3), are potent modulators of animal tumor cells, both in vitro (l-5) and in vivu (6-10). We have recently reported on tE usrijf the bilayG s%?t agar assay to measure the effect of prostaglandins on anchorage-independent growth (4). Tne A and E series were potent inhibitors of murine Imelanoma growth. The anchorage-independent growth of melanoma cells obtained directly from patients and human Imelanoma cell lines was also arrested by PGAl (11). The major problem in extending these observdtions to the clinic is that the concentrations and exposure times needed to produce irreversible inhibition of growth with these prostaglandins as single agent effrctors would be excessively toxic for patients (12). With this in mind, structural analogues of PGAl were investigdted to determine their activity.

Melanoma cells.and soft agar assay
The origin of the human melanoma cell strain C8146c and murine melanoma cell line CCL 53.1 and conditions for growth in soft agar have been previously described (4,111. PG Here added with the cells to the upper layer for these continuous contact experiments. MIRWS was cloned from the irlIRW humdn melanoma developed in this laboratory (13). The MIRWS cell line was injected into nude mice and the tumor removed 5 weeks later and subcultured in Ham's F-10 containing 10% fetal calf serum. Cells from the first four subcultures were tlsed in the experiments presented.

Prostaglandin effects on the anchorage independent growth of murine melanona cells
We tested the effects of several PGA-like analogues in vitro on the growth of murine melanoma cells in soft agar. Table lrankshe effectiveness of these prostaglandins to inhibit anchorage-independent growth relative to PGO2. The ID50 for each PG was extrapolated from the dose response curves obtained over the 0.1 to 10 !~g/ml concentration range and ranked relative to the ID50 for PGD2. PGD2 was assigned a rank of one for each cell line. The higher the ranking the more potent the PG was in inhibiting anchorage-independent growth. All the PGA-like analogues were less potent than PGAl. The more stable analogue, 16,16-dimethyl-PGAl, was only one-fourth as potent as PGAl in inhibiting murine melanoma growth. PGA2 had very little activity, which corresponded to its effect on human melanoma cells (111. The more stable 16,16-dimethyl analogue, U39770, was significantly more active than PGA2. These analogues retained the ability to totally inhibit the anchorage-independent growth of murine melanoma. The 18-oxa derivative of 16,16-dimethyl PGAz.HR466, did not demonstrate any change in activity in comparison to U39770 and was relatively weak in comparison to PGAl. U51605, a thromboxans A2 synthetase antagonist (141, has been shown to inhibit the monolayer 450 SEPTEMBER 1983 VOL. 26 NO. 3 CNE -no effect on colony formation up to maximum concentration of 4 &ml.
growth of B-l&a murfne melanoma cells (5). However, this analogue did not appreciably inhibit the growth of murine melanoma cells in soft agar (Fig. 1). Even at high concentrations (lOl~g/ml) only a 15% reduction in colony number was observed. Another analogue HR601 tihich is in the fairly stable trans-AZ-ll-deoxy-PGE series and is equipotent to HR466 in vasodepressor activity 15) had no effect on the 1 anchorage-independent growth of murine melanoma cells. PGOE which unlike PGA or PGE has an ll-oxa, was about one-half as potent as PGAT (Fig. 1). Prostaglandin effects on the anchorage-independent growth of human me1 anoma cells PGAl inhibited the growth of human melanoma cells in soft agar in a dose dependent manner (11). We measured the effect of several prostanoid analogues on the growth of cells from two human melanoma lines in soft agar (Table 1). C8146c melanoma cells were not ds sensitive to PGA2. Both 16,16-dimethyl PGA were 88% as potent as PGAl in the inhibition of growth as determined by colony formation. The 18-oxa derivative, HR466 was one-half as potent as U39770. HK466 still retained full agonistic properties effecting a ninety-percent reduction in colony formation at 4 ug/ml. As we had observed previously with two other human melanoma cell lines (111, PGEl was a relatively weak inhibitor of anchorage independent growth. The di-azido prostanoid, U51605, was d very weak inhibitor of anchorage independent growth. The maximum inhibition observed was a 25% reduction in colony-formation at 5 ug/ml. HR601 did not effect the growth of C8146c cells to a maximum concentration of 4 ug/ml. PGD2, with the carbonyl at C-11, was a very strong inhibitor of human melanoma anchorage-independent growth. Complete inhibition of colony formation was obtained with a PGD2 concentration of 1 ug/ml versus the 2.5 ug/ml needed with PGAl.

The ClIRW5 human imelanoma cell strdin was not as sensitive to
PGAl (Table 1).
It took twice the concentrdtion of PGAl to produce an inhibition comparable to the effect on iBl46c. PGA2 hdd only one-third the potency of PGAl. The dimethyl derivatives, U3Y770 and U42416, were very potent inhibitors of anchordge independent growth. The PGAl analogue was abOUt as effective as PGAl, whereas U39770 was from 1.6-fold more potent than PGAl. However, 16,16-dimethyl-18-oxa-PGA2 was only one-fifth as potent as 16.16-dimethvl-PGA?. As was observed for cells from the C8146c cell line,HR601 was al& inactive. Ml605 had no significant effect on the growth of these cells in soft agar. As was the case with C8146c cells, PGD2 was the most active prostaglandin tested. It was 3-fold more potent than PGAl for the inhioition of IdIRW.5 growth in soft agar.

PGD2 was the [most potent proStdglandin
we have tested on human melanoma cell growth in soft agar; it also was active on murine melanoma cells. The direct anti-proliferative effects on rnurine melanoma may explain some of its previously observed anti-metastdtic properties which have been attributed to the inhibition of platelet aggregation (10). Recently PGD2 was shown to exert a dose-dependent inhibition of the growth of cells from human leukemia (16) and neuroblastoma lines (171. PGD2 has a ketone at C-11 whereas PGA and PGE have the ketone at C-9. PGF which does not hdve d ketone also was inactive (4,11,17). The presence and position of the ketone in the cyclopentane ring is not the only determinant for activity, since PG81 was relatively inactive (11). PGBl may represent a special SEPTEMBER 1983 VOL. 26 NO. 3 453 case because the position of the double bond in the cyclopentane ring fixes the molecular confornation into a unique secondary structure (18). Experience to date suggests that the relationship and type of substitutions at C-9 and C-11 are important for inferring anti-tumor activity to the prostaglandin and suggests that other structural analogs of the cyclopentane ring should be tested.
The 16,16-dinethyl-PGA analogs were as active as PGAl in inhibiting anchorage independent growth of hunan melanoma cells. However, they hadlsharply reduced activity on the nurine nelanona cells. Though the 16,16-dinethyl nodification did not greatly enhance the observed potency of these analogues in vitro, one would predict because of their increased stability thaFUm and U42416 would be nore active in vivo. However, conversion of C-18 to oxygen greatly reduced actixt=d suggested that 18-oxa derivatives need not be tested further.
HR601 was totally inactive on the growth of cells from the three melanomas tested. Since the 18-oxa modification sharply reduced the activity of 16,16-dinethyl-PGA2 (Table 11, one can not ascribe the lack of activity to A-2, 11-deoxy and/or 15 cl-acetoxy modifications. Further testing with sinpler 11-deoxy-PGEl and a-2,11-deoxy-PGEl is needed before one can assign structure-function relationships. It is worthy that HR601 was biologically active in other in vitro system and was very active on antihypertensive dogs (16).
It is difficult to draw absolute conclusions from the results with the thronboxane A2 synthetase inhibitor U51605 (141, and the &-2-11-deoxy-PGEl derivative, HR601, because of the multiple Modifications. For example, with U51605 one does not know if weak anti-tuner activity was due to the 9,11-azo or the 15-deoxy. Honn and Meyer found this analogue inhibited the growth of B16a, a nurine nelanona cell line (51. U51605 had negli cells from the nurine cell line (CCL 53.1 used here. 3 ible effect on the growth of One should note we are neasuring anchorage independent growth in a semi-solid media, whereas the previous investigation neasured mono-layer growth in a liquid oedia. U51605 did denonstrate some inhibitory activity on one of the huraan nelanona cell lines, Table 1. So there nay be heterogenity in the sensitivity of nelanonas to this analogue. An important and disturbing result was the differences in sensitivity between the cells from nurine CCL 53.1 and the hunan nelanona cell lines. On nurine melanoma cells PGAl and PGEl were equal in potency (11) and nore potent than PGD2, which had significantly greater activity than PGA2 (Fig. 1). Human melanoma displayed a completely different sensitivity, with PGD2 being more potent than PGAl which in turn was greater than PGEl and PGA2. These results suggest caution should be exercised in transposing prostaglandin data in mrine system to predict results or activity in human cell lines.