Predominant expression of alternative PAX3 and PAX7 forms in myogenic and neural tumor cell lines.

PAX3 and PAX7 are closely related paired box family members expressed during early neural and myogenic development. Assay of PAX3 and PAX7 mRNA expression in embryonal rhabdomyosarcoma, neuroblastoma, Ewing's sarcoma, and melanoma cell lines revealed tumor-specific expression patterns similar to the corresponding embryonic lineages. Although the mammalian PAX3 and PAX7 genes were reported to contain eight exons, we found that the predominant PAX3 and PAX7 transcripts in these tumor lines contain previously uncharacterized ninth exons. These splicing events alter the COOH-terminal coding regions of the encoded products but do not alter the transcriptional activity as assayed using a reporter gene with a model PAX3/PAX7 binding site. However, the findings of nearly identical COOH-terminal regions within the corresponding genes of the avian and fish genomes suggest conserved functional roles for these regions that require further investigation.

PAX3 and PAX7 exhibit conserved cis-acting transcription repression domains and utilize a common gain of function mechanism in alveolar rhabdomyosarcoma.

The t(2;13) and t(1;13) translocations of alveolar rhabdomyosarcoma (ARMS) result in chimeric PAX3-FKHR or PAX7-FKHR transcription factors, respectively. In each chimera, a PAX DNA-binding domain is fused to the C-terminal FKHR transactivation domain. Previously we demonstrated that PAX3-FKHR is more potent than PAX3 because the FKHR transactivation domain is resistant to repression mediated by the PAX3 N-terminus. Here we test the hypothesis that the cis-acting repression domain is a conserved feature of PAX3 and PAX7 and that PAX7-FKHR gains function similarly. Using PAX-specific DNA-binding sites, we found that PAX7 was virtually inactive, while PAX7-FKHR exhibited activity 600-fold above background and was comparable to PAX3-FKHR. Deletion analysis showed that the transactivation domains of PAX7 and PAX7-FKHR are each more potent than either full-length protein, and resistance to cis-repression is responsible for the PAX7-FKHR gain of function. Further deletion mapping and domain swapping experiments with PAX3 and PAX7 showed that their transactivation domains exhibit subtle dose-dependent differences in potency, likely due to regions of structural divergence; while their repression domains are structurally and functionally conserved. Thus, the data support the hypothesis and demonstrate that PAX3 and PAX7 utilize a common gain of function mechanism in ARMS.

Genetics and the biologic basis of sarcomas.

Identification of genetic alterations has contributed greatly to the understanding of sarcoma biology. Additionally, detection of these abnormalities is providing new tools for the diagnosis of sarcomas. In this paper, three important new genetic findings from the past year are reviewed, including the t(12;15) translocation of congenital fibrosarcoma, mutation of the putative tumor suppressor gene hSNF5/INI1 in malignant rhabdoid tumor, and the association of c-kit mutations with gastrointestinal stromal tumor. Highlighted are important studies concerning mechanisms of chromosomal translocation, functions of sarcoma-specific fusion proteins, genetic abnormalities other than translocations, molecular diagnosis, and molecular profiling of gene expression. Particular emphasis is placed on information obtained with comparative genomic hybridization and microarray techniques, because these powerful technologies will facilitate the rapid acquisition of data that provide insight into the molecular genetic and biologic basis of sarcomas.

Up-regulation of MET but not neural cell adhesion molecule expression by the PAX3-FKHR fusion protein in alveolar rhabdomyosarcoma.

The 2;13 chromosomal translocation in alveolar rhabdomyosarcoma generates the chimeric protein PAX3-FKHR, which is a powerful transcriptional activator. We hypothesize that PAX3-FKHR regulates downstream effector genes involved in rhabdomyosarcoma tumorigenesis. We evaluated alterations in expression of MET and neural cell adhesion molecule that were proposed previously as downstream targets of wild-type PAX3. We used a myogenic tumor cell culture system and rhabdomyosarcoma tumor specimens to assess candidate gene expression in relationship to various PAX3-FKHR expression levels. We demonstrate that the expression of MET, but not neural cell adhesion molecule, correlates significantly with PAX3-FKHR expression. These findings indicate that MET, which encodes a receptor involved in growth and motility signaling, is a downstream target of PAX3-FKHR in alveolar rhabdomyosarcoma.

Mechanism for transcriptional gain of function resulting from chromosomal translocation in alveolar rhabdomyosarcoma.

The t(2;13) translocation of alveolar rhabdomyosarcoma results in tumor-specific expression of a chimeric transcription factor containing the N-terminal DNA-binding domain of PAX3 and the C-terminal transactivation domain of FKHR. Here we have tested the hypothesis that PAX3-FKHR gains function relative to PAX3 as a consequence of switching PAX3 and FKHR transactivation domains, which were previously shown to have similar potency but distinct structural motifs. In transient cotransfection assays with human expression constructs, we have demonstrated the increased ability of PAX3-FKHR to activate transcription of a reporter gene located downstream of multimerized e5, PRS-9, or CD19 DNA-binding sites in three cell lines. For example, PAX3-FKHR was 100-fold more potent than PAX3 as an activator binding to e5 sites in NIH 3T3 cells. To compare transactivation potency independent of PAX3-specific DNA binding, we tested GAL4 fusions of full-length PAX3 and PAX3-FKHR or their respective C-terminal transactivation domains on a reporter with GAL4 DNA-binding sites. In this context, full-length PAX3-FKHR was also much more potent than PAX3. Additionally, the activity of each full-length protein was decreased relative to its C-terminal domain, demonstrating that N-terminal sequences are inhibitory. By deletion analysis, we mapped a bipartite cis-acting inhibitory domain to the same subregions within the DNA-binding domains of both PAX3 and PAX3-FKHR. We have shown, however, that the structurally distinct transactivation domains of PAX3 and PAX3-FKHR differ 10- to 100-fold in their susceptibility to inhibition, thus elucidating a mechanism by which PAX3 gains enhanced function during oncogenesis.

Structural characterization of the FKHR gene and its rearrangement in alveolar rhabdomyosarcoma.

The FKHR gene, which contains a forkhead DNA-binding motif, is fused to either PAX3 or PAX7 by the t(2;13) or t(1;13) translocation in alveolar rhabdomyosarcoma,respectively. These tumors express chimeric transcripts encoding the N-terminal portion of either PAX protein fused to the C-terminal portion of FKHR. To understand the structural basis and functional consequences of these translocations, we characterized the wild-type FKHR gene and its rearrangement in alveolar rhabdomyosarcomas. By isolating and analyzing phage, cosmid and YAC clones, we determined that FKHR consists of three exons spanning 140 kb and that several highly similar loci are present in other genomic regions. Exon 1 encodes the N-terminus of the forkhead domain and is embedded within demethylated CpG island. RNA analyses reveal FKHR transcripts initiate from a TATA-less promoter within this island. Exon 2 encodes the C-terminus of the forkhead domain and a transcription activation domain, whereas exon 3 encodes a large 3' untranslated region. The intron 1-exon 2 boundary precisely matches the FHKR fusion point in the chimeric transcripts found in alveolar rhabdomyosarcomas. Using pulsed-field and fluorescence in situ hybridization analyses, we demonstrate that the 130kb FKHR intron 1 is rearranged in t(2;13)-containing alveolar rhabdomyosarcomas. Our findings indicate that FKHR intron 1 provides a large target for DNA rearrangemnt. Rearrangement of this intron with PAX3 produces two important functional consequences: in-frame fusion of N-terminal PAX3 sequences to the FKHR transcriptional activation domain and disruption of the FKHR DNA binding domain.

Wild type PAX3 protein and the PAX3-FKHR fusion protein of alveolar rhabdomyosarcoma contain potent, structurally distinct transcriptional activation domains.

Alveolar rhabdomyosarcoma (ARMS) is characterized cytogenetically by a t(2;13)(q35;q14) chromosomal translocation involving two transcription factor genes: PAX3 and FKHR. ARMS cells express a PAX3-FKHR fusion protein containing the complete N-terminal, DNA-binding domain of PAX3 and the C-terminus of FKHR. Recently we demonstrated that PAX3-FKHR is a more potent transcriptional activator than PAX3 despite impaired binding to canonical PAX3 binding sites. Therefore, we propose that the gene fusion results in switching of PAX3 and FKHR transactivation domains with distinct structure, potency or function. To compare the PAX3 and putative PAX3-FKHR transactivation domains, we fused C-terminal test fragments to the heterologous GAL4 DNA-binding domain and tested activation of a reporter gene co-transfected into four cell types. GAL4-PAX3 and GAL4-PAX3-FKHR were found to be potent activators exhibiting different concentration-dependent transactivation profiles and distinct structural motifs. Deletion mapping demonstrated essential acidic and/or serine/threonine-rich domains in the extreme 3' ends of their respective coding regions and positive modifying elements in adjacent 5' sequences. These data demonstrate that PAX3 and PAX3-FKHR contain structurally distinct transcriptional activation domains and suggest that a consequent difference in function is important for oncogenesis.

Production of multiple cytokines by cultured human melanomas.

We screened a panel of 8 primary and 21 metastatic melanoma cell lines for constitutive secretion of cytokines. Melanomas expressed bioactivity for TGF-beta (8/25 lines) and IFN (7/12), but not IL-2. Immunoassays detected IL-1 alpha (4/25), IL-1 beta (12/25), IL-6 (13/29), IL-8 (29/29), TGF-beta 2 (5/12) and GM-CSF (11/29), but not IL-3, IL-4, TNF-alpha, or IFN-gamma. There was no preferential association of cytokine production with cells cultured from primary versus metastatic disease, and only IL-8 was produced by all lines tested. These data demonstrate that cultured melanomas produce a variety of cytokines which are potentially capable of influencing tumor growth in vivo.

Characterization of tenascin secreted by human melanoma cells.

Tenascin is a large glycoprotein of the extracellular matrix. It shows a site-restricted expression during embryogenesis and can be found in adult tissues during wound healing and tumorigenesis. Because of the potential involvement of tenascin in adhesion and invasion during metastasis, the study of the interactions of tumor cells with tenascin is of considerable interest. Using five anti-melanoma monoclonal antibodies to four different epitopes of human tenascin, we found that most melanoma cells secrete tenascin in vitro constitutively. Transforming growth factor beta 1 in the medium increased secretion in tenascin-producing cells. Tenascin was present in sera of melanoma patients, with significantly elevated levels in patients with advanced melanomas as compared to patients with low tumor burden or to normal donors. Normal and malignant melanocytes did not attach to tenascin as substrate within 1 to 2 h and tenascin could also inhibit fibronectin-dependent adhesion. These results indicate that tenascin may play a critical role in cell-substrate interactions of melanoma cells.

Human melanoma cells transcribe interleukin 1 genes identical to those of monocytes.

Two molecular species of the pleotropic cytokine interleukin 1 (IL-1) are produced as products of two distinct genes transcribed by cells of the monocyte-macrophage lineage. We have shown previously that a significant proportion of human melanoma cell lines express IL-1 biological activity, but it has not been demonstrated that this activity is the same as authentic monocyte IL-1 alpha and -beta. Here we report the cloning and sequencing of IL-1 complementary DNAs from a metastatic melanoma cell line and demonstrate that they encode bona fide IL-1 alpha and IL-1 beta. In addition, IL-1 complementary DNAs encoding a different amino acid at position 145 were revealed.

Production of interleukin 1 activity by cultured human melanoma cells.

A panel of melanoma cell lines derived from 7 primary and 20 metastatic lesions was tested for the production of interleukin 1 (IL-1) in standard mouse thymocyte costimulation assays. Constitutively produced IL-1 activity was found in the conditioned media of 4 of 7 primary and 5 of 20 metastatic melanoma cell lines tested. Four of 9 cell lines secreting IL-1 were also shown to contain cell-associated activity in their lysates. Melanoma-conditioned media were, however, unable to support the growth of CTLL, an interleukin 2-dependent cell line. The secreted IL-1 activity was significantly inhibited by antibodies to recombinant IL-1 alpha (3 of 3 lines), but not antibody to recombinant IL-1 beta. When conditioned medium from one cell line was fractionated on a Superose 12 column by fast protein liquid chromatography, a major peak of activity eluted at Mr 22,500-27,500. The presence of 2.2-kilobase mRNA hybridizing a probe for IL-1 alpha and 1.6-kilobase mRNA hybridizing a probe for IL-1 beta was detected by Northern blot in 3 of 4 secreting cell lines but not in a nonsecreting line. Taken together, these results suggest that cultured melanoma cells produce the cytokine IL-1 alpha, although the relationship between melanoma IL-1 and monocyte IL-1 is unclear. The production of IL-1 by melanoma cells is of interest because of its potential roles in the biology of melanoma through direct effects on tumor growth or through indirect effects on adjacent stromal and endothelial cells and infiltrating lymphoid cells.

Characteristics of cultured human melanocytes isolated from different stages of tumor progression.

Normal melanocytes and melanocytes of normal nevi, primary melanoma in the radial (RGP) and vertical (VGP) growth phases, and metastatic melanoma exhibited and maintained phenotypic differences when grown in tissue culture or in experimental animals. Only metastatic and VGP primary melanoma cells were tumorigenic in athymic nude mice and had nonrandom chromosomal abnormalities involving chromosomes 1, 6, and 7. The colony-forming efficiency in soft agar was also highest in these two cell types. A cell line of RGP primary melanoma had characteristics of both benign and malignant cells: nevus-like morphology; nontumorigenicity in nude mice; but karyotypic abnormality of chromosome 6. It also had a ganglioside pattern similar to that of normal melanocytes but not melanomas, i.e., a high GM3 ganglioside content compared to the amounts of GM2, GD2, and GD3 gangliosides. Binding of monoclonal antibodies secreted by hybridomas generated by immunization of mice with VGP primary and metastatic melanoma was highest with cells and supernatants of cultures from advanced melanoma and least with nevus cells. There was no binding to normal melanocytes except with the monoclonal antibodies specific for nerve growth factor receptor or 9-O-acetyl-GD3 ganglioside. On the other hand, monoclonal anti-nevus antibodies bound to melanocytes, nevus cells, and RGP primary melanoma cells but not to VGP primary or metastatic melanoma cells. Cultured human melanocytic cells appear to be a unique model for the study of tumor progression.