INTRODUCTION

Melanoma is a significant health concern among the Caucasian population and is typically characterized by aggressive biologic behavior, with eventual metastasis and death. Tumorigenesis is thought to be associated with a multistep, temporal accumulation of genetic alterations (1), and in an attempt to understand its biology, the molecular pathogenesis of these lesions is being continuously investigated. Of the chromosomal abnormalities encountered, allelic imbalances of chromosome 9p21, which harbors the MTS1 gene are those frequently described in both sporadic and familial melanomas (2, 3, 4, 5, 6, 7). Mutations involving the long arm of chromosome 10 have also been reported in sporadic primary and metastatic melanomas at frequencies ranging from 31 to 58% (4, 5, 8, 9, 10, 11) and have been associated with a poor prognosis (4, 8).

The Max interacting protein (MXI1) is a tumor suppressor gene mapped to chromosome 10q24–25. Deletion of genetic material from this site is commonly seen in neural crest–associated tumors such as glioblastomas (12). Rao et al. (11) demonstrated MXI1 loss of heterozygosity (LOH) in 46% of the primary and metastatic desmoplastic melanomas that they studied, postulating that the embryologic neuroectodermal origin of desmoplastic melanomas may account for this similarity with glioblastomas. Extrapolating on this further, it would not be surprising if conventional melanomas should show similar allelic imbalances in the MXI1 gene. Herbst et al. (13), however, in their study of 65 sporadic primary and metastatic melanomas, found only 7 to 13% LOH for this gene. To our knowledge, these are the only two published studies in the English literature on MXI1 in melanomas.

We thus sought to further investigate the role of the MXI1 gene in the pathogenesis and progression of conventional melanomas by using a microdissection-based genotyping technique to look for LOH using two microsatellite markers for this gene. We compared these findings to those seen in the MTS1 gene on 9p21.

MATERIALS AND METHODS

Seven primary, 2 recurrent, and 20 metastatic melanomas were identified from the files of the Presbyterian and Montefiore University Hospitals of the University of Pittsburgh Medical Center. Paraffin-embedded blocks were obtained for each of the 29 cases, and clinical information was extracted from the Medical Archives Retrieval System.

Serial 4-μm-thick unstained histologic sections were microdissected under stereoscopic observation, as described elsewhere (14). For each tumor sample, representative uninvolved dermis of approximately equal size was also obtained as a normal control. The microdissected samples were treated with 50 μL of proteinase K (10 ng/mL) for 2 hours and boiled for 5 minutes to remove enzyme activity.

Each sample was PCR amplified in individual reactions using oligonucleotide primers targeted at microsatellite repeats of the MXI1 and MTS1 genes. For MXI1, we used an intragenic microsatellite (3NT) in the 3′ nontranslating region of the gene (12), and the flanking repeat unit D10s1173 (GenBank accession no. L30341). MTS1 was studied using two flanking microsatellites D9s251 and D9s254 (GenBank accession nos. L18726 and L18050, respectively). The primers for MXI1.3NT were 5′-GGTTACTCCAGTGCCAGTGT-3′ (sense) and 5′-TTAAATACAGGTCCTCTGACCC-3′ (antisense); for D10s1173, they were 5′-CATGCCAAGACTGAAACTCC-3′ (sense) and 5′-AAACCCCAATGCCATAATGG-3′ (antisense); for D9s251, they were 5′-CCTGTGTTGAAATTTTGACTG-3′ (sense) and 5′-ATTTCAGACTTCCTTGTGTTC-3′ (antisense); and for D9s254, they were 5′-TCCTGGGTAATAACTGCCG-3′ (sense) and 5′-CACTCACACACACGCTCAG-3′ (antisense). After initial denaturation at 95° for 10 minutes, 35 cycles of PCR were carried out with denaturation at 94° C for 1 minute, annealing at 58° C for 1 minute, and polymerization at 74° C for 1 minute, with a final 10-minute extension at 74° C. The amplified products were labeled with [P33]-deoxyadenine triphosphate and electrophoresed in 6% polyacrylamide. Loss of heterozygosity was determined by comparing the intensity of the polymorphic bands from normal and tumor tissue. To control for allelic dropout caused by insufficient sample size, the bands from normal tissue were required to be of similar intensity to justify analysis of the tumor samples.

RESULTS

Twenty patients were identified with 7 primary, 2 recurrent and 20 metastatic melanomas (Table 1). Overall, both genes showed similar degrees of allelic loss, 54% for MXI1 and 67% for MTS1. For the MXI1 gene, comparing primary and recurrent/metastatic lesions, 6 and 22, respectively were informative for one or both markers. LOH was seen in 2 (33%) and 13 (59%) of these (Fig. 1). Similarly analyzing MTS1, 5 and 19 of primary and recurrent/metastatic tumors were informative, with 4 (80%) and 12 (63%) showing LOH, respectively.

Table 1 Loss of Heterozygosity in Melanomas for MXI1 and MTS1 Using Two Microsatellite Markers for Each Gene
Full size table
FIGURE 1
figure 1

MXI1.3NT genotyping of melanomas. Lanes 1 to 4 represent Patient 1, who is informative for the microsatellite marker, with the specimen from nonlesional dermis showing two bands of similar intensity (Lane 1). Both the primary (Lane 2) and recurrent (Lane 3) melanomas of the left leg show no LOH. The metastasis to the groin lymph node (Lane 4) shows LOH with increased intensity of the upper band and a corresponding decrease of the lower (arrow). Lanes 5 to 6 represent Patient 4, who is noninformative for the gene locus, with only one band present in the normal sample. Lanes 7 to 9 depict Patient 11 with a primary melanoma of the right cheek (not shown), with informative normal tissue (Lane 7) and metastases to the parotid gland (Lane 8) and neck lymph node (Lane 9). Mutations involved different allelic bands, confirming that each metastasis is derived from genetically distinct clonal populations.

Full size image

More than one tumor was sampled in eight patients. Of these, the pattern of LOH in both genes was identical in five. One of the discordant cases (Patient 1) was that of a primary melanoma of the left leg that was negative for LOH in MXI1 and recurred 6 months later with a similar result. The metastasis from 1 year later, however, showed acquisition of LOH in both loci of the MXI1 gene (Fig. 1). MTS1 was not informative. Patient 10 had two distinct metastases sampled within 1 week of each other. The adrenal tumor showed LOH in all four gene loci studied, whereas the lung was positive in only one locus for each gene (MXI1.3NT and D9s251) and was negative in the other two. The primary tumor, a trunk melanoma from 19 years prior, was not available for analysis. The results from Patient 11 were of particular interest, with the two tumors studied representing neck lymph node and parotid gland metastases that were removed 6 and 8 years, respectively, after a right cheek primary. Although on initial inspection the pattern of gene loss appeared to be the same for both lesions, examination of the allelic bands lost for the MXI1.3NT microsatellite revealed that different ones were deleted in the separate metastases and thus represented genetically distinct clones (Fig. 1).

DISCUSSION

Genetic alterations involving the long arm of chromosome 10 occur fairly commonly in primary and metastatic melanomas, with frequencies ranging from 31 to 58% (4, 5, 8, 9, 10, 11). The gene most often implicated is PTEN/MMAC1 (9, 10), a tumor suppressor gene located on 10q23.3. Germline mutations of this gene have been associated with Cowden disease (multiple hamartoma syndrome), a rare autosomal-dominant familial cancer syndrome (15). The MXI1 gene is also located on the long arm of chromosome 10 and has been mapped to chromosome 10q24–25. This latter gene acts as a tumor suppressor by negatively regulating the myc oncoprotein. Allelic imbalances have been demonstrated in glioblastomas (12); however, to our knowledge, there have been only two prior studies in the English literature to determine its possible involvement in melanoma tumorigenesis. Herbst et al. (13) concluded that the low frequency of LOH (7 to 13%) at intragenic markers for MXI1 in conventional melanomas most likely represented random genetic rather than pathogenic events. Conversely, Rao et al. (11) found MXI1 LOH to be relatively frequent in desmoplastic melanomas, seen in 5 of 11 informative cases.

Our study, unlike that of Herbst et al. (13), showed frequent LOH at intragenic and flanking loci of the MXI1 gene, with aberrations seen in 54% (15 of 28) of the informative cases of primary, recurrent, and metastatic melanomas. This is similar to the 45% LOH that was seen in the desmoplastic melanomas studied by Rao et al. (11), although all of our cases were of the conventional type. When the primary melanomas were considered separately, only two of six informative cases (33%) demonstrated LOH. Conversely, 13 of 22 (59%) informative metastatic/recurrent lesions showed genetic loss. Although the number of primary lesions studied is small, this seems to suggest that genetic imbalance involving the MXI1 gene is a late event in tumorigenesis, associated with metastatic potential and spread. Further anecdotal support for this hypothesis is provided by Case 1, in which no LOH was detected in the primary or recurrent melanoma but there was acquisition of LOH in the groin metastasis at both the intragenic and flanking loci of the MXI1 gene (Fig. 1).

The findings seen in Patients 10 and 11 are remarkable for their demonstration of genetic heterogeneity in separate metastases from the same primary. In Patient 10, the adrenal and lung metastases were sampled within 1 week of each other, yet the adrenal tumor showed genetic loss in microsatellites for MXI1 and MTS1 that the lung lesion did not. The synchronous nature of the metastases begs against genetic progression as is seen in Patient 1 but suggests that two distinct clones developed at the same time in the primary tumor and that both acquired metastatic potential. This is further and more convincingly supported by the findings in Patient 11, in which distinct allelic bands were lost in the MXI1 gene in the lymph node and parotid gland metastases (Fig. 1). These findings suggest that the molecular evolution of melanoma is not the result of a uniform, linear progression but involves multiple genetic alterations occurring in clonal subpopulations, each attaining its own metastatic potential. These initial findings support those of others (6, 7), but further larger studies need to be performed to investigate the genetic diversity present in melanomas, as well as to correlate these with biologic behavior.

The MTS1 (CDKN2A) gene on chromosome 9p21 has been frequently demonstrated to show allelic loss in both familial and sporadic cutaneous melanomas. This is a tumor suppressor gene that encodes the cell cycle inhibitor p16INK4a. In this study, LOH for the MTS1 gene was seen in 16 of 24 informative melanomas (66.67%), similar in frequency to previous reports of 47 to 72% (2, 3, 4, 5, 6, 7, 13). Of the five informative primary tumors, four (80%) showed LOH, also in concordance with other investigators who suggest that mutations involving the short arm of chromosome 9 are early events in the evolution of melanoma (5). The MTS1 analysis was used as a positive control gene in our study and validates our methodology and findings with respect to the much less frequently studied MXI1 gene. The finding of similar degrees of allelic loss in both suggests that MXI1 may also play a role in the development of melanoma, as is believed of MTS1.

In summary, the frequency and pattern of allelic imbalances seen in MXI1 suggest that it is a putative tumor suppressor gene on chromosome 10q involved in conventional melanoma progression. Genetic heterogeneity seen in different metastases from the same primary are indicative of the development of clonally diverse populations, each acquiring its own metastatic potential.