Identification of a Novel Gene, MASL1 within an Amplicon at 8p23.1 Detected in Malignant Fibrous Histiocytomas by Comparative Genomic Hybridization¹

MFH³ is the most common soft-tissue sarcoma among adults (1). In a recent WHO classification, MFH was defined as “a pleomorphic spindle-cell sarcoma, usually occurring in adults and displaying no distinct line of differentiation” (2). Histologically, MFH defines a heterogeneous group of sarcomas composed of a mixture of fibroblastic, histiocytic, and bizarre cells, often accompanied by variable amounts of inflammatory cells, collagen, and a myxoid substance in the stroma. Whether the tumor originates from histiocytes, fibroblasts, or undifferentiated mesenchymal cells is not known. Of the four histological subtypes of MFH (2), the storiform-pleomorphic and myxoid subtypes are more common, whereas giant cell and inflammatory subtypes are seen less frequently.

Karyotypic abnormalities in MFH are usually complex, with multiple numerical and structural rearrangements. No chromosomal aberrations specific for MFH have been identified thus far. However, telomeric associations, unidentified ring chromosomes, and di-centric chromosomes are frequently seen in this type of tumor, as are cytogenetic hallmarks of gene amplifications, i.e., homogeneously staining regions and double minute chromosomes (3, 4). Several known proto-oncogenes with potential pathogenic importance in the development of tumors in soft tissue have been examined for participation in MFH: (a) the sarcoma-amplified sequence gene (SAS); (b) the human homologue of the murine double minute 2 gene (MDM2); (c) cyclin-dependent kinase 4 (CDK4); and (d) the gene encoding C/EBP homologous protein (CHOP). All of those mentioned have been mapped to chromosome 12q13–15, and all except CHOP have been amplified in more than one-third of the MFHs examined (5). Alterations affecting tumor suppressor genes TP53 (17q13) or RB1 (13q14) have also been reported in approximately one-third of MFH tumors (6, 7).

In the work reported here, we studied 19 MFH tumors by CGH to explore genomic alterations that might affect the development and/or progression of this type of tumor. Together with losses and gains involving several regions of different chromosomes, we found distinct high-level amplifications at six loci: (a) 4q12–21; (b) 8p21–pter; (c) 8q24.1–qter; (d) 9q12–13; (e) 12p11.2–pter; and (f) 15q11.2–15. Therefore, each of these regions may contain one or more potential oncogenes affecting the tumorigenesis of MFH. As the first step toward the identification of such genes, we focused on the 8p amplicon to characterize its molecular structure in detail. From the common region of amplification at 8p23.1, we isolated a novel gene designated MASL1 that contained a highly conserved leucine-rich tandem repeat domain that is characteristic of protein-binding molecules.

All 19 cases were newly diagnosed, untreated patients from whom primary tumor tissue was collected at the time of surgical resection. DNA from frozen samples was extracted according to standard protocols. Clinical findings are summarized in Table 1. Pathological diagnoses of all 19 samples were performed independently by two experts in the pathology of soft tissue tumors.

For CGH (8), DNA samples isolated from tumor specimens were labeled directly with Spectrum green-dUTP (Vysis), and normal DNA was labeled with Texas red-dUTP (DuPont New England Nuclear, Boston, MA) using nick translation. Labeled tumor and normal DNAs (180 ng of each), together with 15 μg of normal Cot-1 DNA (Life Technologies, Inc., Gaithersburg, MD) in a 10-μl hybridization solution (50% formamide, 10% dextran sulfate, 2× SSC), were denatured at 70°C for 5 min and applied to normal (lymphocyte) metaphase spreads. Hybridization conditions and signal detection are described elsewhere (9). Three-color fluorescent images (4′,6-diamidino-2-phenylindole, Spectrum green, and Texas red fluorescence) were collected from each metaphase spread using an epifluorescence microscope (Nikon, Tokyo, Japan) and a cooled charge-coupled device camera (Photometrics, Tucson, AZ). Relative changes in the copy number of DNA sequences were analyzed using a digital image analysis system (Quips-XL software; Vysis). Chromosomal regions were considered to be high-level amplifications when the fluorescence ratio exceeded 1.5. Heterochromatic regions near the centromeres and the entire Y chromosome were excluded from the analysis.

To narrow the region at 8p21–23 that was amplified at a high level in one tumor (tumor 1) in the CGH study, we performed revFISH (10) on elongated prophase chromosomes, using whole-genome DNA from this tumor as the probe. Elongated prophase chromosomes were prepared from cultured lymphocytes according to previously published methods (11).

Southern hybridization was carried out according to standard protocols (12) to identify genomic fragments within the amplified region at 8p21–23. The following seven cosmids on 8p21–23.3 that contain ESTs or STSs were used as probes in this experiment: (a) 0757C03 (GEN-131E06); (b) 0600H06 (GEN-003G09); (c) cos.024H10 (GEN-024H10); (d) cos.505G01 (GEN-505G01); (e) 0155D08 (GEN-001B12); (f) cos.D8S1819 (D8S1819); and (g) cos.D8S550 (D8S550; Table 2). The ESTs and the data concerning their chromosomal locations were obtained from the GENOTK human cDNA database (http://genotk.genome.ad.jp). EcoRI-digested tumor DNAs and human placental DNAs (5-μg samples) were electrophoresed in 0.8% agarose gels and transferred to nylon membranes (BIODYNE, Pall, NY). Each probe was labeled with [α-³²P]dCTP using random primers and hybridized to the prehybridized filter. We analyzed signals with a BAS1000 image analyzer (Fuji, Tokyo, Japan), calculated the degree of amplification, and then exposed the filters for autoradiography at −80°C for 1 day.

We screened a human fetal brain cDNA library (10 × 10⁶ plaques; Stratagene) using GEN-024H10 as the probe and subsequently determined the DNA sequences of positive clones using a 377 ABI autosequencer (Perkin-Elmer Corp.) and a protocol that has been described previously (13).

Total RNAs were obtained from frozen samples of MFH tumors using Trizol (Life Technologies, Inc.) according to the manufacturer’s instructions. cDNA generated from 0.2 μg of total RNA with Superscript II (Life Technologies, Inc.) was used as the template for each PCR. All reactions involved an initial denaturation at 94°C for 2 min followed by 25 cycles at 94°C for 30 s, 58°C for 30 s, and 72°C for 30 s on a GeneAmp PCR system 9600 (Perkin-Elmer Corp.). The integrity of each RNA template was controlled through the amplification of β-actin, which revealed similar signals in all samples. Primer sequences were as follows: (a) MASL1, 5′-GCATTTCCAGCCACCTCAG-3′ and 5′-ACAGAAAGCCCTCCACA TAC-3′; and (b) β-actin, 5′-CCAGAGATGGCCACGGCTGCT-3′ and 5′-TCCTTCTGCATCCTGTCGGCT-3′. RT-PCR products were electrophoresed through a 3.0% GTG agarose gel, Southern blotted onto nylon membranes, and hybridized with a [γ-³²P]dATP-labeled internal oligonucleotide of GEN-024H10, 5′-ATCCTCAATGTCTTCTTCCAG-3′.

Northern blots containing 2μg of polyadenylated RNA in each lane from 1 of 16 different normal human tissues(Clonetech) were hybridized with a random-primed ³²P-labeled probe corresponding to nucleotides 2361–3858 of MASL1, according to the manufacturer’s instructions. The blots were washed with 0.1× SSC/0.1% SDS at 50°C, and signals were analyzed with a BAS1000 image analyzer (Fuji).

A total of 16 of 19 tumors examined (84%) exhibited genomic imbalances. A schematic summary of the copy number aberrations detected by CGH in each case is shown in Fig. 1. Underrepresentation of the smallest region of overlap was seen on 3p25–pter (two cases; 11%), 10p12–pter (two cases; 11%), 13q21 (two cases; 11%), and 18q22–qter (two cases; 11%). Overrepresentation of the smallest region of overlap was seen on 4q12–13 (seven cases; 37%), 5p15.3–pter (five cases; 26%), 8q24.1–qter (four cases; 21%), 6p12–21 (three cases; 16%), and 12p12–pter (three cases; 16%). High-level amplification of DNA was identified at six different loci (4q12–21, 8p21–pter, 8q24.1–qter, 9q12–13, 12p11.2–pter, and 15q11.2–15) in one tumor each. The high-level amplification on the short arm of chromosome 8 was present in tumor 1 (Fig. 2 A).

To define the smallest region exhibiting amplification at 8p21–pter, we applied revFISH, using tumor 1 DNA as the probe, to elongated prophase chromosomes. FISH signals of the amplicon for this MFH tumor (Fig. 2 B) narrowed the amplicon cytogenetically to band 8p23.1.

We used seven cosmids containing STSs or ESTs on 8p21–pter as probes. These sequences were sublocalized cytogenetically by FISH as follows: (a) GEN-131E06 and GEN-003G09 on 8p21–22; (b) D8S550, GEN-024H10, and GEN-505G01 on 8p23.1; and (c) D8S1819 and GEN-001B12 on 8p23.2–pter (data not shown). The amplification status of each of these loci in 15 MFHs is given in Table 2; the minimal overlapping region amplified was between D8S1819 and D8S550. This region contained two ESTs, GEN-024H10 and GEN-505G01, which were amplified in two of the tumors (tumors 1 and 4; 13%). However, a high copy number amplification had been detected by CGH at 8p21–pter only in tumor 1, suggesting that the physical size of the 8p amplicon in tumor 4 might be less than 2 Mb (8). A rough estimation of the copy number, based on a comparison of the hybridization signals of amplified DNA versus those of normal DNA, revealed about 8- and 4-fold amplifications of GEN-024H10 and GEN-505G01 in tumor 1 and about 6- and 3-fold DNA amplifications of each in tumor 4, respectively (Fig. 3).

We screened a human fetal brain cDNA library using GEN-024H10 as a probe. The alignment of cDNA sequences from 10 positive clones revealed a 6242-bp transcript containing a single open reading frame of 3159 bp. The initiation codon for translation was considered to be at nucleotides 417–419 because the consensus sequence for the initiation of translation (Kozak’s rule) is well conserved. Fig. 4 shows the predicted 1052-amino acid sequence of the gene product.

The PSORT program (14) for the prediction of protein localization sites indicated one hydrophobic region in the middle (residues 486–502), although a hydropathy plot (15) predicted no hydrophobic regions. ATP/GTP-binding site motif A [(AG)-X₄-G-K-(ST)] was found at residues 416–423. Leucine residues repeated as every seventh amino acid can form a structural motif known as the leucine zipper (16). The deduced gene product would contain three leucine zipper regions (residues 68–89, 102–123, and 962–983), each with four leucine residues spaced at 7-amino acid intervals (L-X₆-L-X₆-L-X₆-L). A homology search using the FASTA program (Genome Net) revealed that the predicted protein is similar to the acid-labile subunit protein (ALS), a component of a ternary complex with IGF-I or IGF-II and IGFBP-3 (17). ALS belongs to a family of proteins with a conserved leucine-rich repeated motif (18). When the 11 leucine-rich tandem repeats (23 residues each) of the new protein are aligned (Fig. 4), it is apparent that certain residues are highly conserved, and the following consensus sequence can be deduced: P-X-X-α-X-X-L-X-X-L-X-X-L-X-L-X-X-N-X-α-X-X-α (X, any amino acid; α, aliphatic amino acids L, I, or V). We designated the novel gene MASL1.

Northern analysis was carried out to determine the level of MASL1 expression in various human tissues. A 6.6-kb transcript was expressed ubiquitously (Fig. 5,A). To compare the levels of MASL1 transcription in MFH tumors and normal tissues, quantitative expression studies were performed using RT-PCR. As shown in Fig. 5 B, this gene was overexpressed not only in tumor 1 but also in tumors 3, 4, 5, 6, 7, 9, and 11. As a rough estimate, the expression of MASL1 increased about 10-fold in tumor 1, the only tumor that had revealed high-level amplification at 8p21–23 in CGH analysis.

In our CGH analysis of MFH, 84% of the tumor samples showed gains and/or losses of DNA sequences in at least one chromosomal region. The minimal common regions of the most frequent underrepresentations affected 3p25–pter (11%), 10p12–pter (11%), 13q21 (11%), and 18q22–qter (11%). Among them, only 3p25–pter contains a known candidate gene, VHL. However, because nonrandom losses at 13q21 were also detected in a CGH analysis of MFH reported by Larramendy et al. (19), that region is very likely to contain tumor suppressor genes.

The minimal common regions of the most frequent overrepresentations affected 4q12–13 (37%), 5p15.3–pter (26%), 8q24.1–qter (21%), 6p12–21 (16%), and 12p12–pter (16%). Chromosomes 4q12–13, 8q24.1–qter, 6p12–21, and 12p12–pter contain known proto-oncogenes (PDGFAR, MYC, CCND3, and KRAS, respectively). Although the 5p15.3–pter region is a commonly reported site of high-level gain or amplification in several types of tumors, no proto-oncogenes have been identified there as of yet.

We identified high-level DNA amplifications at six different locations (4q12–21, 8p21–pter, 8q24.1–qter, 9q12–13, 12p11.2–pter, and 15q11.2–15) in one tumor each. revFISH on elongated prophase chromosomes narrowed the distinct high-level amplification on the short arm of chromosome 8 to the neighborhood of 8p23.1. We and another group of investigators have previously detected frequent amplifications at 8p22.1–23.1 in primary gastric cancers (9, 20). Those results prompted us to focus on refinement of the amplicon at 8p23.1 in MFH as the first step toward identifying oncogenes involved in multiple types of neoplasms.

An initial Southern blot analysis with seven cosmids containing STSs or ESTs that had been mapped to 8p21–23.3 revealed that the minimal region covered by overlapping amplifications lay between D8S1819 and D8S550, a fragment that contains the ESTs GEN-024H10 and GEN-505G01. GEN-024H10 was amplified in 2 (tumors 1 and 4) of the 18 MFH tumors (13%).

From a human fetal brain cDNA library, we isolated a 6242-bp full-length cDNA (MASL1) corresponding to GEN-024H10, which contained a single open reading frame of 3159 bp. The primary structure of the deduced MASL1 product showed several features and unexpected sequence homologies that suggested critical functions for this protein. A striking feature was the presence of 11 tandem leucine-rich repeats of 23 amino acids; this motif usually includes proline as the first residue, repeated leucines in several positions, and asparagine as the eighteenth residue. The highly conserved nature of this pattern indicates that leucine-rich repeat regions are important for maintaining the structural integrity and/or the function of proteins that contain them. This motif may participate in binding, because the functions of known proteins containing it relate to protein-protein interactions (21, 22). The repeated motif of MASL1 is very similar to one that occurs in yeast adenylate cyclase (23), an enzyme that is activated by RAS and controls progression over the start point in the G₁ phase of the yeast cell cycle (24). Possession of this conserved motif suggests that MASL1 protein may play an important role in signal transduction and the control of cell growth in humans.

Moreover, the predicted amino acid sequence of MASL1 bears a 35.9% homology to the human acid-labile subunit protein (ALS). In the ternary complex with IGFs and IGFBP, ALS has a central role in regulating the bioavailability of circulating IGF (25). ALS binds to IGFBP-3 only in serum; in this ternary form, the growth factors are apparently unable to cross the capillary barrier (26). However, when ALS interacts with cell-associated glycosaminoglycans, which can act as receptors for other circulating molecules (27), it dissociates readily from the ternary complex so that growth factors in the binary complex can pass the capillary barrier. Passage of the IGFs from circulation into the tissues might depend on this specific dissociation mechanism (28). Therefore, if MASL1 is analogous to ALS and binds to a protein similar to IGFBP-3, it may regulate cell growth.

Our quantitative RT-PCR studies revealed that the expression of MASL1 was significantly enhanced in tumors 1 and 4, both of which had shown distinct amplification of DNA at this locus. The results strongly suggest that DNA amplification is one of the mechanisms that can lead to the up-regulation of MASL1; in fact, overexpression of MASL1 was found in 57% (8 of 14) of the tumors examined by RT-PCR, although only one of the tumors had shown high-level amplification at 8p23.1 by CGH. Together with the predicted amino acid structure of MASL1, the data seem to indicate that the deregulation of this gene is an important pathogenetic factor for MFH.