Functional Evidence of Novel Tumor Suppressor Genes for Cutaneous Malignant Melanoma¹

CMM⁴ is a frequently occurring cancer with a high risk of mortality from metastatic disease (1). The overall incidence of the disease is increasing, probably as a consequence of increased exposure to UV through sunbathing (1). Genetic alterations are known to underlie the development of CMM, and investigation of primary and metastatic lesions has delineated a number of critical chromosomal locations. In general, these segments tend to show LOH and, in a few cases, homozygous deletion (2).

LOH and homozygous deletion at 9p21 represent the most commonly observed abnormalities in primary melanoma (50–60% of cases; Refs. 3 and 4). Deletion mapping, largely in melanoma cell lines, identified the 9p21-encoded cyclin-dependent kinase inhibitor CDKN2^INK4A/p16 as a TSG with a role in both sporadic and familial CMM (5, 6). Intriguingly, the observation that melanoma kindreds show 9p21 linkage but not p16 mutation and the retention of a functional p16 gene in a significant number of primary melanomas suggests the presence of another important melanoma TSG at this chromosomal location (5, 6, 7, 8). In addition, a high frequency (30–35%) of LOH of 10q alleles has been reported in primary melanoma (9, 10). The recently identified TSG PTEN/MMAC1 has been proposed as a target of 10q22–qter losses (11), and Walker et al. (12), using 10q LOH data, mapped a second potential TSG distal to PTEN/MMAC1, raising the possibility of multiple TSGs on 10q with a role in melanoma formation.

The presence of other TSGs important in melanoma development is indicated by the high frequency of LOH on chromosomes 1, 6, and 11 in regional melanoma metastases. Transfer of normal copies of these chromosomes to human CMM cell lines was shown to significantly reduce tumorigenicity and/or metastatic potential (13, 14, 15, 16). Hence, chromosome transfer studies have demonstrated the validity of the technique as a functional complementation approach to the identification of TSGs (13, 14, 15, 16).

To investigate chromosomes 9 and 10 for the presence of novel TSGs, we used MMCT to the human metastatic melanoma cell line UACC-903. Hybrid clones were examined for suppression of tumorigenicity using anchorage-independent growth in soft agar (13) and tumor growth in athymic nu/nu immune-deficient mice. The availability of two variants of chromosome 9 (chromosomes 9a and 9b) harboring microdeletions at 9p21 that ablate IKN4A/B function (Fig. 1) provided an opportunity to obtain evidence for an additional TSG on the short arm of chromosome 9.

Here, we provide the first functional evidence of a novel melanoma TSG or TSGs on chromosome 9p and also identify an additional TSG on chromosome 10. The results provide a basis for accurate subchromosomal localization of these genes as a prelude to molecular cloning.

The highly tumorigenic human malignant melanoma cell line UACC-903 (a kind gift from Dr. J. M. Trent, National Cancer Institute, Bethesda, MD) was routinely cultured at 37°C in RPMI 1640 supplemented with 10% fetal bovine serum, 2 mm l-glutamine, and 50 μg/ml gentamicin (Life Technologies, Inc., Gaithersburg, MD).

The UACC-903-derived monochromosome hybrids were generated using a human monochromosome library maintained in mouse A9 cells as somatic cell hybrids, by methods described in detail elsewhere (17). A9:monochromosome donor hybrids were exposed to 0.075–0.1 μg/ml colcemid (Sigma-Aldrich, Dorset, United Kingdom) for 48 h to induce metaphase arrest and micronucleation. Chromosome transfer was performed as described by Cuthbert et al. (17). After 24 h, 5 × 10⁵ cells were plated into 10-cm Petri dishes and, following a further 48 h incubation, hybrid cells were selected in 400 units/ml hygromycin B (Calbiochem-UK, Nottingham, United Kingdom) in complete medium. After 3–4 weeks, hygromycin B-resistant colonies were isolated.

To ensure identical growth conditions, the parental UACC-903 cell line was transduced with the defective amphotropic retrovirus tgLS-HyTK (18), and several clones were obtained with similar growth characteristics. One clone was selected and used as a control cell line for further studies. Control cells and the monochromosome hybrids were routinely cultured and maintained in the presence of 200 units/ml hygromycin B.

Assessment of anchorage-independent growth was carried out by use of the soft agar cloning method described previously (19). Cells (1 × 10⁴) in 0.3% Noble agar (Difco) were plated onto 0.6% Noble agar base layers in triplicate using six-well dishes. After 21 days incubation colonies that were >50 μm in diameter were counted.

Athymic female CD1 (nu/nu), 4–6 week old mice (Charles River, Ltd., Kent, United Kingdom), were injected s.c. into the right flank with 5 × 10⁶ cells suspended in 0.2 ml of RPMI. Tumors were measured, and the volumes were calculated as described previously (20). Mice bearing tumors of >1.5 cm³ were sacrificed and, in a selection of animals, the lungs, liver, and spleen were recovered to check for the presence of metastases. In addition, tumor cells were harvested for cytogenetic analysis (by FISH) and extraction of genomic DNA for PCR.

As a preliminary screen to confirm the presence or absence of a transferred chromosome in the UACC-903 hybrids, PCR was performed, generating a product derived from the 187-bp HyTK fusion gene.

To confirm the presence and integrity of the transferred chromosome, FISH was performed on metaphase spreads of UACC-903 hybrids, prepared by standard methods (21). Chromosome paints were derived from flow-sorted human chromosomes and biotin-labeled using degenerative oligonucleotide-primed PCR, as described by Griffin et al. (22). Biotinylated DNA was detected with Cy3 avidin (Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire, United Kingdom) at a dilution of 1:500 in 4× SSC, 0.1% Tween 20, and 1% BSA. Slides were counterstained with 4′,6-diamidino-2-phenylindole in Vectashield mounting medium (Vector Laboratories, Burlingame, CA) and visualized using fluorescence microscopy.

The overall numbers of hybrids derived from the chromosome transfer experiments and those exhibiting suppressed/nonsuppressed growth in soft agar are given in Table 1. For the purposes of this study, an arbitrary CFA of ≤1% compared to parental UACC-903 cells was regarded as a suppressed phenotype. Gross structural integrity of transferred chromosomes was confirmed by FISH analysis and, in all cases, a cytogenetically normal chromosome was observed (representative examples are shown in Fig. 2).

The availability of two variants of chromosome 9 (9a and 9b), which were microdeleted for the p16^INK4A and p15/^INK4B loci (see Fig. 1; Ref. 23) permitted an evaluation of the presence of other TSGs for CMM without the influence of these two cyclin-dependent kinase inhibitors encoded on 9p21. One variant (9a) contained exon-specific deletions of p16^INK4A and p15^INK4B, and the second (9b) harbored a much larger deletion (∼30 cM) spanning markers D9S171 (9p21) to IFNA (9p22; Fig. 1). Transfer of these chromosomes caused a reduction of CFA in soft agar compared to UACC-903 cells and control monochromosome hybrids (Fig. 3, B and C). In addition, the majority of hybrids derived by the transfer of chromosome 10 were completely negative for soft agar colony formation (Fig. 3 E), suggesting the presence of one or more melanoma-associated TSGs on chromosome 10.

Chromosome 6 transfer yielded 12 hybrids, 7 of which exhibited suppressed growth in soft agar. Although tumor growth studies were not performed with these particular hybrids, our soft agar data are concordant with previous investigations that demonstrated suppression in soft agar and reduced tumorigenicity in vivo with chromosome 6 (13, 14, 15).

A clonal derivative of UACC-903 cells transduced with the HyTK retroviral fusion gene (as described in “Materials and Methods”) had a mean CFA of 21%, a value largely unaltered by the transfer of chromosome 5 (mean CFA = 19%). In contrast, transfer of chromosome 15 and 9q reduced CFA to 9 and 5%, respectively (Fig. 3 A).

After injection of UACC-903 parent cells into nude mice, by day 15, all mice had developed tumors at all injection sites with a mean volume of 584 ± 120 mm³ (mean ± SE), a result consistent with previously published data on this highly tumorigenic cell line (13). In addition, hybrids generated with chromosomes 9q, 5, and 15 also formed tumors in mice, consistent with our soft agar observation and confirming that chromosomes 5 and 15 do not confer any suppression of tumorigenicity (see Tables 1,2,3). Moreover, the lack of tumor suppression by 9q maps the novel TSG to 9p. By day 19, all positive control groups of mice were sacrificed due to excessive tumor burden.

Tumor suppression by chromosome variant 9a was evident in all mice injected, and no evidence of metastases was observed following microdissection of the liver, lungs, and spleen. All animals were tumor free at day 15, and this status was maintained until day 37 postinjection, at which point small tumors began to emerge in three of the five animals. The tumors continued to grow slowly, and at day 60 the animals were sacrificed. At this point, tumor volume was still relatively small (126.9 ± 70.7 mm³) compared to parental and control hybrid cells (Table 3). FISH analysis of the tumor cells demonstrated, in some cases, a loss of the introduced chromosome but, in other cells, the chromosome was retained.

In contrast, those mice injected with hybrids containing the larger microdeletion within 9p21–9p22 (∼30 cM; chromosome 9b), tumor growth was noticeably faster and, by day 15 postinoculation, all mice within this group yielded tumors (Table 3). However, overall tumor volume was significantly less than for any of the control groups (mean volumes at day 15 = 95.91 and 96.24 mm³, respectively; P = 0.006 compared to control groups), and tumor growth continued at a slow rate until day 26 postinjection. At this point excessive tumor volume required that the animals be sacrificed.

The tumor growth study clearly defines an additional TSG or TSGs on 9p that is distinct from the p15 and p16 genes (INK4locus). Furthermore, the observation of a different pattern of tumor suppression, depending upon the size of deletion within 9p21, suggests that two TSG loci may be present. One of these genes appears closely linked to the INK4 locus because complete suppression was observed with chromosome variant 9a. The second of these TSGs appears to locate to a segment on 9p outside the 30 cM (approximate) region between markers D9S171 and IFNA (Fig. 2). Speculation that other TSGs exist close to the INK4 locus is consistent with a number of LOH studies of melanoma in which the minimally deleted region excludes INK4 (7, 8, 9). In addition, we have determined that the parental UACC-903 cells are deleted for p16 (at exons 1 and 2).⁵ Therefore we conclude that the observed tumor suppression mediated by the transfer of chromosomes 9a and 9b indicates the presence of a novel TSG locus rather than a modification of gene expression of the host INK4 locus.

Tumor suppression was conferred by the transfer of chromosome 10 into UACC-903 cells (Table 3). Two chromosome 10 hybrids were injected into separate groups of five and six mice. By day 37 postinjection, no tumors had formed in the group of six mice injected with clone 1 (see Table 3), and only after 40 days postinoculation was tumor growth observed in two mice within that group. Suppression of tumor growth was also observed with the second chromosome 10 hybrid (clone 2; see Table 3). In this case, one of the five animals developed a tumor at day 15 postinoculation and, by day 37, all mice were sacrificed due to excessive tumor volume. Eventual tumor growth in this group of mice was associated with a loss of the introduced chromosome 10 (Fig. 2).

Overall, reduction of both soft agar growth and tumor growth in nude mice were indicative of a major TSG or TSGs on chromosome 10. The location of this gene remains to be determined. LOH studies of melanoma and certain other human cancers indicate that nonrandom losses involving 10q22–qter are a common event. On the basis of these analyses, a novel TSG, designated PTEN/MMAC1, has been mapped to 10q23 and has recently been shown to be mutated in breast, prostate, and kidney cancers and also in gliomas and melanomas (24, 25 26). Loss of function of this TSG has recently been demonstrated in a number of melanoma cell lines including UACC-903 (11). In this particular cell line, a T→G transversion creates a premature stop codon at the third position of codon 76.

With regard to the loss of PTEN/MMAC1 function in the hybrids generated using the UACC-903 cell line, our preliminary analysis of DNA from two suppressed and two “segregant” (nonsuppressing) chromosome 10 hybrids has detected a normal codon 76 (TAT), suggesting that at least part of the gene is retained. Currently, we are performing additional studies to determine the functional status of PTEN in a number of segregant hybrids and conducting additional experiments to confirm the presence of a second TSG distal to PTEN, as proposed by Walker et al. (12).

In a proportion of all chromosome hybrids of UACC-903, there appeared to be an escape from the tumor-suppressive action conferred by the transferred chromosome and, in these cases, CFA in soft agar was >1% (Table 1 and Fig. 3, B–E). These findings are consistent with previous chromosome transfer investigations (e.g., Ref. 23). During MMCT discrete intrachromosomal microdeletions of varying sizes are introduced. If these deletions involve a region of the chromosome containing the putative TSG, then suppression is removed and the segregant hybrid will fail to exhibit a suppressed phenotype in soft agar or suppressed tumor growth in nude mice. Thus, segregant hybrids provide a resource with which to map the location of these novel TSGs as a prelude to molecular cloning.

In conclusion, we have transferred chromosomes 9 and 10 into malignant melanoma cells and have been able to provide strong functional evidence for the presence of additional TSG or TSGs on 9p that function independently of the INK4 locus. Functional evidence for TSG activity associated with chromosome 10 has also been provided. It should be noted that introduction of either of these chromosomes into UACC-903 had no effect on overall cell viability or population doubling times when grown as monolayers, compared to parental (nonhybrid) cells (data not shown). The availability of segregant hybrids provides an opportunity to map and clone these genes. Our results, together with those of others, point to the existence of a number of potential TSGs that appear to be central to melanoma development and its malignant progression. Consequently, a complicated pathway of CMM initiation and progression is emerging that involves the loss of one or several TSGs. The isolation of all melanoma suppressor genes will eventually lead to an understanding of how mutations in the various TSGs fit into the stepwise model of malignant progression and provide useful markers for molecular epidemiological studies.