Characterization of Gene Amplification–Driven SKP2 Overexpression in Myxofibrosarcoma: Potential Implications in Tumor Progression and Therapeutics

Myxofibrosarcoma, a common soft tissue sarcoma affecting the extremities of the elderly, is characterized by increased metastases after relentless local recurrences (1, 2). The cumulative overall survival rate is approximately 75% at 5 years and few series analyzing myxofibrosarcomas provided incongruent values for various clinicopathologic prognosticators (3–5). Among these, we and others previously addressed the importance of clear margins, which predicted better local control and translated into final survival benefits (3, 5). Somatic alterations in cellular DNA underlie most human cancers and myxofibrosarcoma is genetically known to harbor complex karyotypes (6, 7). However, little is understood about the molecular determinants accounting for the clinical aggressiveness of myxofibrosarcoma.

Genome-wide approaches with derived targeted therapies are now prompting increasing efforts to characterize cancer genomes (6, 8). In this series, global chromosomal aberrations were profiled by ultrahigh resolution array comparative genomic hybridization (aCGH) to identify critical genes of clinical relevance in myxofibrosarcoma. Previously, we found that S-phase kinase-associated protein 2 (SKP2), a negative regulator of p27^kip1 cell-cycle inhibitor, was preferentially overexpressed in myxofibrosarcomas with adverse outcomes (9). In this context, special attention was given to 5p in aCGH analysis to assess whether gene amplification drives increased gene expression of SKP2 and other potential oncogenes located in 5p. To validate this hypothesis, we measured the mRNA expression levels of 3 selected genes with DNA copy number gains on 5p, that is, TRIO, SKP2, and AMACR, all of which were upregulated in tumor samples. In a test set, SKP2 amplification was identified in 38% of primary myxofibrosarcomas and strongly correlated with SKP2 labeling index and poorer outcomes.

Apart from the classic function in promoting cell proliferation and tumor growth, stable SKP2 knockdown with short hairpin RNA (shRNA; shSKP2) in 2 SKP2-overexpressing myxofibrosarcoma cell lines confirmed the promigratory and proinvasive oncogenic attributes of SKP2. This prometastatic phenotype conferred by SKP2 was linked to differential expression of motility-associated genes, including ITGB2, ACTN1, IGF1, and ENAH. In vitro, treatment with bortezomib (Velcade, Millennium Pharmaceuticals) repressed SKP2 expression in myxofibrosarcoma cell lines, which, compared with fibroblasts, were more sensitive to this proteasome inhibitor. Noticeably, bortezomib resulted in significant growth inhibition of myxofibrosarcoma xenografts, indicating ubiquitin-proteasome pathway as a potential therapeutic target.

One each myxofibrosarcoma cell line was established and kindly provided by Dr. Bridge (OH931) and Dr. Ogose (NMFH-1), respectively (10, 11). The KEL-FIB fibroblast cell line was purchased from Taiwan Bioresource and Collection Research Center. The cell lines were maintained with the medium, nutrients, and antibiotics as previously reported (12, 13).

The criteria of diagnosis and assessment for parameters of myxofibrosarcoma, including grading and staging, were elaborated previously and all cases were surgically treated with curative intent (3, 9). Tissue procurement was approved by Institutional Review Board (99-2215B). Twelve fresh tumor specimens were previously analyzed by genome-wide aCGH to evaluate copy number alterations (CNA; ref. 13). These materials, along with 4 additional fresh specimens, were used in real-time reverse transcriptase PCR (RT-PCR) assay to detect mRNA expression of candidate oncogenes on 5p and in quantitative DNA-PCR assay to determine SKP2 gene dosage, respectively. To validate the implication of SKP2 amplification, gene dosage was successfully quantified in 82 independent primary myxofibrosarcomas, excluding those used for aCGH profiling or receiving neoadjuvant radiation or chemotherapy. The clinicopathologic characteristics of samples for aCGH were previously described and those of the test set are summarized in Table 1.

One microgram of genomic DNA each from OH931 and NMFH-1 myxofibrosarcoma cell lines was extracted and checked for DNA integrity for genomic analysis with 385K oligonucleotide-based microarray (NimbleGen). The methods were essentially as reported for 12 fresh tumor samples except that normal lymphocytes of gender-matched donors were used as reference DNAs (13). DNA labeling with Cy3 and Cy5 fluorescent dyes, hybridization, normalization of oligonucleotide arrays, and window averaging of contained probes were done as previously described (13). The raw data were log₂-transformed and output into Nexus software (BioDiscovery). To finely delineate the breakpoints in array probes, gains and losses in significant regions of CNAs were defined as log₂ ratios of ≥ +0.20 or ≤ −0.20, respectively. To unravel causal genes showing copy number–driven deregulated expression, common regions of alteration were filtered for consecutive makers where the proportion of analyzed tumor and cell line samples was ≥20%. Minimal common regions were local maximums within the common regions of size greater than 1 Mb.

As described in Supplementary Method S1, approximately 1,500 myxofibrosarcoma cells were isolated from each fresh sample by laser capture microdissection (LCM) to quantify the mRNA fold expression of TRIO, SKP2, and AMACR. To determine SKP2 gene dosage, 3,000 cells of interest were microdissected from each paraffin block.

From cell lines and LCM-isolated tumor cells of myxofibrosarcoma, total RNAs were extracted, quantified, and reverse-transcribed. Using predesigned TaqMan assay reagents (Applied Biosystems), we measured mRNA abundance of TRIO (Hs00179276_m1), SKP2 (Hs00180634_m1), and AMACR (Hs001091294_m1) with ABI StepOnePlus System (Supplementary Method S2). The fold expression of target genes relative to normal adjacent tissues, dermal fibroblasts, or vehicle controls was calculated by comparative C_t method after normalization to POLR2A (Hs01108291_m1) as the internal control.

Given that no CNA was detected by aCGH at the corresponding DNA segment, PIK3R1 at 5q13.1 was chosen as the reference gene to determine SKP2 gene dosage, taking into account the potential bias of chromosome 5 polysomy. Real-time DNA-PCR was carried out on LightCycler 2.0 (Roche) using primers and probes with sequences provided in Supplementary Method S3. Because the comparative C_t method might result in skewed values for low-copy templates from LCM-isolated, formalin-fixed samples, we used the following equation to calculate the SKP2 gene dosage:

, where E corresponded to individual efficiency of SKP2 or PIK3R1 amplifying reaction derived from standard curves of serially diluted plasmids cloned with normal SKP2 or PIK3R1 DNA sequence (14, 15). T and N corresponded to tumor and normal calibrator tissues, respectively. To define SKP2 amplification, we adopted a cutoff point of average SKP2/PIK3R1 ratio ≥2 in at least 2 independent assays (Supplementary Method S3).

The immunohistochemical and scoring methods for evaluating SKP2 protein expression on tissue microarrays have been described (9). Another recipient block was similarly constructed to enroll 27 new cases, which, along with those published, encompassed all 82 cases in the test set with determined SKP2 gene dosage.

The lentiviral vectors were obtained from Taiwan National RNAi Core Facility, including pLKO.1-shLacZ (TRCN0000072223: 5′-TGTTCGCATTAT CCGAACCAT-3′), pLKO.1-shSKP2#3 (TRCN0000007532: 5′-CCATTGTCAATACTCTCGCAA-3′), and pLKO.1-shSKP2#4 (TRCN0000007533: 5′-AGGCCAACTATTGGCAACAAA-3′). Viruses were produced by transfecting HEK293 cells with the pLKO.1-shLacZ, pLKO.1-shSKP2#3, or pLKO.1-shSKP2#4 using Lipofectamine 2000. For viral infection, 3 × 10⁶ OH931 or NMFH-1 cells were incubated with 8 mL lentivirus in the presence of polybrene, followed by puromycin selection for stable clones of lentivirus-transduced cells.

Cell lysates containing 25 μg protein were separated by 4%–12% gradient NuPAGE gel (Invitrogen), transferred onto polyvinylidene difluoride (PVDF) membranes (Amersham), and probed with antibodies against glyceraldehyde-3-phosphate dehydrogenase (GAPDH; 1:3,000; Chemicon) and proteins of interest. The latter included SKP2 (1:200; Zymed), p27^kip1 (1:500; Novocastra), and, from Epitomics, caspase-3 (1:1,000), cleaved caspase-3 (1:500), ENAH (1:1,000; hMena), β2-integrin (1:250), α-actinin-1 (1:1,000), and insulin-like growth factor 1 (IGF1; 1:1,000). After incubation with the secondary antibody, proteins were visualized by the Chemiluminescence System (Amersham).

DNA synthesis was assessed using an ELISA-based and colorimetric BrdU assay (Roche). OH931 and NMFH-1 cells were plated in 96-well plates at densities of 2,000 and 3,000 cells per well, respectively, for evaluation of DNA synthesis at 24, 48, and 72 hours. After incubation with bromodeoxyuridine (BrdUrd) for 3 hours, the labeling medium was decanted for subsequent fixation and final incubation with anti-BrdU-POD solution. The absorbance of samples was measured using an ELISA reader (Promega) at 450 nm with the reference absorbance at 690 nm.

To assess anchorage-independent growth, 2 × 10⁴ tumor cells were mixed in 1 mL of 0.3% agar with complete medium and seeded onto a 6-well plate containing 1.5 mL of 0.6% agar in complete medium. OH931 and NMFH-1 cells were further cultured for 15 or 10 days, respectively. After staining with 0.005% crystal violet in PBS for 1 hour, colonies >0.1 mm in diameter were counted under the microscope in 10 random fields.

An artificial “wound” was created using a 200 μL pipette tip on confluent cell monolayers in 6-well plates. Photographs were taken at 0, 4, 8, 12, and 24 hours. Quantitative analysis of the wound closure was calculated by measuring the percentage of healed area relative to the initial wound at day 0.

Cell invasion was determined using 24-well Collagen-Based Cell Invasion Assay (Millipore). Briefly, each insert was added with 0.3 mL serum-free medium to rehydrate collagens, replaced with 0.3 mL of serum-free suspension of 10⁶ cells in the upper chamber, and incubated for 24 hours to let cells migrate toward the lower chamber containing 10% FBS. After removing noninvading cells in the upper chamber, cells invading through the inserts were stained with provided dye, dissolved in extraction buffer, and transferred to 96-well plates for colorimetric reading at 560 nm.

To explore potential mediators of SKP2 implicating its prometastatic function, we compared expression profiles between stable shSKP2- and shLacZ-transfected OH931 and NMFH-1 cells using RT-PCR Array (PAHS-128, SABioscience). The PCR reactions for 84 motility-associated genes were run on ABI StepOnePlus System (Applied Biosystems). Expression was normalized to housekeeping genes and presented as fold expression relative to the corresponding shLacZ controls. In both OH931 and NMFH-1 cells, those genes differentially expressed with P < 0.0001 between SKP2-silecning and control conditions and >1.75-fold upregulation or downregulation were selected for validation by immunoblotting.

NMFH-1, OH931, and KEL-FIB cells were seeded in 96-well plates at a density of 5 × 10³ cells per well the day before treatment with vehicle control (0.9% saline) or bortezomib at indicated concentrations (5–100 nmol/L) for 72 hours.

Treated and untreated cells were determined for viability using the XTT Assay Kit (Roche) for 0 to 72 hours. Briefly, the investigated cells were plated on gelatinized 96-well plates at a density of 5 × 10³ cells per well and analyzed on an ELISA microplate reader at 492 nm.

The evaluation of apoptosis upon bortezomib treatment was conducted by measuring caspase-3 and caspase-7 activation with the Caspase-Glo 3/7 Assay Kit (Promega) following the manufacturer's instructions.

The animal protocol was approved by Animal Use Committee (98121505). A total of 4 × 10⁶ NMFH-1 cells were harvested, resuspended in RPMI-1640, and 1:1 (v/v) mixed with Matrigel (Collaborative Research), yielding a total volume of 0.2 mL per injection. The mixture was subcutaneously inoculated into right flanks of ten 8-week-old male severe combined immunodeficient (SCID) mice (Biolasco) and allowed for growth until 10 days postinjection (day 0). The mice were randomized into 2 groups with 5 each receiving intraperitoneal injection of 1 mg/kg of bortezomib or 0.9% saline as the control. The treatment was continued for 4 weeks and the mice sacrificed 31 days after implantation of NMFH-1 cells. The tumor volume was calculated using the formula: V = π/6 × length (mm) × width (mm)².

The associations of SKP2 gene dosage with clinicopathologic factors and immunoexpression were evaluated using the χ², Fisher exact, or Wilcoxon rank-sum test as appropriate. The median follow-up duration was 30.6 months (range, 2–229 months) for 82 independently validated cases. The endpoint analyzed was metastasis-free survival and disease-specific survival. The durations were calculated from the date of operation until the occurrence of event or last follow-up appointment. Univariate survival analyses were conducted using Kaplan–Meier plots and compared by the log-rank test. A multivariate model analysis was conducted using Cox proportional hazards regression, including parameters with univariate P < 0.05. The Student t test was used to analyze quantitative RT-PCR, functional, and pharmacologic assays for cell line samples. For all analyses, 2-sided tests of significance were used with P < 0.05 considered significant.

Chromosomal imbalances of a varying degree were detected in all samples subjected to aCGH profiling and analysis with Nexus software identified more recurrent regions of deletions than gains with characteristically high genomic complexity (Fig. 1A). Across the whole genome, recurrent gains were detected in 198 chromosomal regions spanning 3,190 genes, whereas 217 chromosomal regions were nonrandomly lost, involving 6,523 genes (Supplementary Table S1). Recently, copy number gain/amplification of 5p was found to be frequent in myxofibrosarcoma in a large-scale genomic study on various major sarcoma types (6), whereas its candidate oncogenes have not been characterized. Along with 1q, 6p, 7p, 7q, 8p, 18q, 19p, and 19q in this series, the differentially altered regions significantly involved 5p with preferential gains in fresh samples featuring adverse outcomes (P < 0.05), suggesting its potential role in tumor aggressiveness (Supplementary Table S2). We finely mapped 3 major amplicons to 5p15.2-p15.1, 5p15.1-p13.2, and 5p13.2-p11, harboring 59 named genes in total (Fig. 1B). Among these, TRIO at 5p15.2, AMACR at 5p13.3, and SKP2 at 5p13.2 were selected for quantifying fold expression of mRNA in LCM-isolated tumor cells from fresh samples. Real-time RT-PCR revealed that the expression levels of these 3 genes were all significantly upregulated in tumor cells compared with adjacent soft tissues (Fig. 1C), thereby reinforcing their roles as driver oncogenes in myxofibrosarcoma. Furthermore, SKP2 mRNA expression was associated with its genomic DNA dosage (Supplementary Fig. S1) and significantly higher in cases of higher American Joint Committee on Cancer (AJCC) stages (Fig. 1D).

Next, we examined the clinical significance of SKP2 gene dosage in an independent test set of 82 primary myxofibrosarcomas. Present in 31 of 82 (38%; Fig. 2A), SKP2 amplification strongly correlated with SKP2 immunohistochemical overexpression (Fig. 2B and Table 1) and preferentially appeared in cases of axial sites, deep locations, higher grades, and more advanced stages (Table 1). However, there were still 5 of 35 SKP2-overexpressing tumors (14.3%) that were not amplified at the SKP2 gene locus. These findings indicated that alternative mechanism(s) other than amplification may operate to drive SKP2 overexpression in at least a minor subset of myxofibrosarcomas (Table 1). Similar to protein overexpression, SKP2 amplification was highly predictive of worse outcomes for both endpoints (both P < 0.0001; Fig. 2C; Supplementary Table S3). In multivariate comparison, SKP2 amplification remained prognostically independent (Supplementary Table S3), along with axial location for disease-specific survival (P = 0.0004) and with higher grades for metastasis-free survival (P = 0.0011).

To gain functional insights, we first characterized NMFH-1 and OH931 myxofibrosarcoma cell lines for SKP2 gene status and its endogenous expression. As validated by real-time DNA-PCR (data not shown), SKP2 gene was amplified in NMFH-1 cells but remained unaltered in OH931 cells, concordant with the aCGH findings. As compared with KEL-FIB fibroblasts with barely detectable expression, both myxofibrosarcoma cell lines apparently showed endogenous SKP2 overexpression with comparable levels of transcript and protein (Supplementary Fig. S2) and the fold expression was higher in the SKP2-amplififed NMFH-1 cells than in the OH931 cells. The finding of SKP2 expression in OH931 cells similarly implied alternative amplification-independent mechanism(s) in a minority of human samples described above. RNA interference was then used to assess the biologic consequences of SKP2 overexpression, and remarkable SKP2 knockdown was achieved in selected stable clones of both myxofibrosarcoma cell lines at both mRNA and protein levels (Fig. 3A).

We compared the rates of myxofibrosarcoma cells incorporating BrdUrd, which were significantly reduced in SKP2-knockdown OH931 and NMFH-1 cells, corroborating the proliferation-promoting function of SKP2 (Supplementary Fig. S3). In soft agar assay, there were significantly smaller and fewer cell colonies identified in SKP2-silenced OH931 and NMFH-1 cells than in corresponding controls, substantiating the role of SKP2 in enhancing anchorage-independent growth (Fig. 3B). Although the growth-promoting function of SKP2 via induction of p27^kip1 proteolysis has been characterized in various cancer types (16), relatively little is understood about whether and how SKP2 overexpression promotes cancer cell motility and distant metastasis. Therefore, wound-healing assay was conducted and it showed significantly slower wound closure in stable SKP2-knockdown myxofibrosarcoma cells (Supplementary Fig. S4), indicating the promigratory function of SKP2. To elucidate the capability of SKP2 in mediating cell invasion, we seeded OH931 and NMFH-1 cells stably transfected with shSKP2 or shLacZ in the modified Boyden chambers. Twenty-two hours after seeding, significantly less invading tumor cells were counted for shSKP2-transfected cell lines, suggesting the proinvasive property of SKP2 in myxofibrosarcoma (Fig. 3C). RT-PCR expression array further identified 5 motility-associated genes significantly and consistently downregulated by SKP2 silencing in both myxofibrosarcoma cell lines, whereas there were no upregulated genes fitting the selection criteria (Supplementary Table S4). Furthermore, immunoblotting confirmed concomitant downregulation of β2-integrin, α-actinin-1, insulin-like growth factor 1, and hMENA proteins that are encoded by ITGB2, ACTN1, IGF1, and ENAH genes, respectively (Fig. 3D).

We further investigated whether and how bortezomib modulated SKP2 expression and affected myxofibrosarcoma cell growth and survival. Bortezomib treatment decreased cell viability in a dose- and time-dependent manner, with substantial inhibition (>50%) after 48 hours at concentrations as low as 10 and 20 nmol/L in NMFH-1 and OH931 cell lines, respectively (Fig. 4A). However, this inhibitory effect was not seen in dermal fibroblasts even at the dose of 100 nmol/L after 72 hours (Fig. 4A). Using Western blot analyses, we further found that SKP2 protein expression was downregulated by bortezomib at 20 nmol/L in both OH931 and NMFH-1 cells from 24 hours onward, accompanied by a concomitant increase in p27^kip1 expression (Fig. 4B). Because bortezomib functions as a proteasome inhibitor, we clarified whether its effect on SKP2 suppression was regulated through protein degradation. By using cycloheximide to shut down nascent protein synthesis, bortezomib did not promote degradation of endogenous SKP2 protein until 24 hours in both cell lines (Fig. 4C, top). Accordingly, the abundance of SKP2 transcript before and after addition of bortezomib was quantified, showing that mRNA expression was downregulated by more than 50% at 8 and 12 hours after treatment in OH931 and NMFH-1 cells, respectively (Fig. 4C, bottom).

Apart from the cell-cycle arresting effect caused by p27^kip1 reexpression through downregulated SKP2 (17, 18), we sought to elucidate whether cell apoptosis might account for the decreased cell viability in bortezomib-treated myxofibrosarcoma cells. Bortezomib induced activation of caspase-3 with increased cleaved form appearing 24 and 48 hours posttreatment in OH931 and NMFH-1 cells, respectively, whereas the degree was more apparent in the former (Fig. 4B). This apoptosis-promoting effect was also cross-validated by the quantitative chemiluminescent assay showing increased caspase-3/7 activity in both myxofibrosarcoma cell lines at the range of therapeutic doses (≤20 nmol/L; Supplementary Fig. S5).

The potentiality of bortezomib treatment in vivo was implied by the cell model showing that myxofibrosarcoma cells were relatively more sensitive to bortezomib than dermal fibroblasts at the therapeutic doses. Therefore, the efficacy of bortezomib in tumor growth inhibition was examined in murine xenografts of NMFH-1 cells. As compared with the vehicle-treated mice, bortezomib treatment resulted in significant regression of measured tumor volumes from day 29 posttreatment onward and apparent shrinkage of tumor size at the end of day 31 (Fig. 5A and B). There was neither weight loss nor impairment of hepatic and renal function related to bortezomib treatment (Supplementary Fig. S6). On postmortem histologic examinations (Fig. 5C, first row), xenografts in vehicle-treated mice displayed hyperchromatic spindle to pleomorphic sarcoma cells with frequent mitotic activity in an abundant myxoid matrix, similar to the morphologic features of human myxofibrosarcomas. However, bortezomib-treated xenografts were histologically characterized by reduced viable cells with few, if any, mitoses (P < 0.001) and apparent regressing fibrotic change that overrode the myxoid matrix. Immunohistochemically (Fig. 5C, second to fourth rows), there were fewer SKP2-labeling tumoral nuclei (P < 0.001) but increased p27^kip1-postive (P = 0.011) and terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL)-positive (P < 0.001) apoptotic cells in bortezomib-treated group, and vice versa in the vehicle control group.

High-resolution aCGH allows for rapid discovery of novel cancer-associated genes by sharply delineating the breakpoints of CNAs (8). Similar to a variety of carcinomas harboring amplified 5p (19, 20), this chromosomal aberration has been recently shown in different types of sarcomas, including myxofibrosarcomas (6). The current series provided high-resolution global genomic profiling of myxofibrosarcomas and distinguished gains and losses in 198 and 217 chromosomal regions, respectively. Specifically on 5p, there were recurrent DNA copy increases at the loci spanning TRIO, AMACR, and SKP2 genes where concordantly upregulated mRNA expression levels were confirmed in a panel of myxofibrosarcoma tissues samples and cell lines. The latter finding indicated these 3 candidates as coactivated driver oncogenes rather than “bystanders” unrelated to tumorigenesis. Furthermore, multiple levels of evidence with independent validation corroborated SKP2 as a crucial amplified oncogene of myxofibrosarcomas, contributing to cellular selective advantage and disease progression by conferring aggressive phenotypes. Notably, amplification-driven SKP2 overexpression has therapeutic relevance given the SKP2-suppressing effect of bortezomib.

In myxofibrosarcomas, SKP2 amplification was seen in 38% of an independent set of primary tumors, generally in line with the detection rate of aCGH profiling and amplification frequencies of canonical oncogenes reported (21). Moreover, the increase in SKP2 copy number was reflected at the protein level with a strong association between gene amplification and SKP2 overexpression. However, it should be added that alterative pathways might also operate because SKP2-overexpressing OH931 cell line and 14% of such myxofibrosarcomas lacked gene amplification. SKP2 depends on SCF–SKP2 E3 ligase activity to exert its oncogenic function toward the promotion of cell proliferation and tumorigenesis via proteasome-mediated degradation of its substrates, such as p27^Kip1 (16, 22, 23). Given this classical attribute, it is conceivable to find positive correlations between SKP2 amplification and adverse clinicopathologic factors, such as increasing tumor size and mitotic rate, in high-grade myxofibrosarcomas. Our in vitro assay also confirmed that SKP2 knockdown led to decreased BrdUrd uptake in both myxofibrosarcoma cell lines, signifying attenuated DNA synthesis. Furthermore, anchorage-independent growth was impaired in SKP2-knockdown myxofibrosarcoma cells, as shown by significantly fewer and smaller colonies in soft agar assays. Importantly, SKP2 amplification independently portended worse outcomes in primary myxofibrosarcomas, along with higher grades for metastasis-free survival and with axial locations for disease-specific survival.

Distant metastasis is the leading cause of mortality in soft tissue sarcomas, a major obstacle in treatment (24, 25). Apart from inducing cell-cycle entry, emerging evidence implies that SKP2 overexpression may play a role in promoting metastatic propensity in common carcinomas, such as prostatic and esophageal cancers (26, 27). As SKP2 amplification was an adverse harbinger of metastasis in primary myxofibrosarcomas, this prompted us to further examine whether SKP2 silencing can lead to attenuated cell migration and/or invasion. In this series, SKP2-knockdown myxofibrosarcoma cells not only delayed wound healing but also significantly decreased invading cells in Transwell assays, strengthening the role of SKP2 in tumor metastasis. Recently, upregulated matrix metalloproteinase (MMP)-2 and MMP-9 were shown to mediate invasive behavior conferred by SKP2 (28). Subsequently, SKP2 was found to increase RhoA transcription by coordinating Myc and Miz1 in an E3 ligase–independent manner, thereby promoting cancer metastasis (26). Nevertheless, neither RhoA nor MMP-2 or MMP-9 was significantly downregulated in both SKP2-knockdown myxofibrosarcoma cell lines in our RT-PCR expression arrays, suggesting cell type–dependent differences in the prometastatic downstream mediators of SKP2.

At both the mRNA and protein levels, we showed consistent and significant downregulation of a focused panel of motility-associated genes after SKP2 knockdown. Of these, ITGB2 encoding β2-integrin was downregulated, whereas the genes encoding β1- and β3-integrins, albeit better characterized in cancer metastasis (29, 30), were not differentially expressed by RNA interference. To drive membrane protrusions of migratory cells, integrins are known to provide physical links between actin cytoskeleton and extracellular matrix and transduce growth factor–mediated signaling (29, 30). Specifically for β2-integrin, it is known to interact with α-actinin to induce actin polymerization and form a heterodimeric αLβ2 integrin receptor overexpressing in breast cancers (31, 32). In addition, the expression and phosphorylation of α-actinin-1 are required for pressure-induced colon cancer cell adhesion by facilitating Src recruitment to β1-integrin to enhance activation of focal adhesion kinase (33). Regarding the proinvasive and prometastatic functions, IGF1 can stimulate cell motility by relocating integrins to the leading edge of migrating cells and upregulate important proteases needed for extracellular matrix degradation (34). It is therefore interesting to find downregulated IGF1 expression in SKP2-silenced myxofibrosarcoma cells. Actually, the ligand binding of IGF1 to IGF1 receptor has been shown to activate signaling cascades in several soft tissue sarcomas, thereby contributing to their tumorigenesis (35, 36). Notably, SKP2 knockdown also negatively regulated expression of ENAH gene encoding the hMENA protein, a member of enabled (Ena)/vasodilator-stimulated phosphoprotein (VASP) family. Overexpression of hMENA, especially its invasion isoform, can drive EGF-dependent cell invasion and promote breast cancer invasion and metastasis by cooperating with HER2 signaling (37, 38). Certainly, sophisticated studies are needed to decipher the exact roles and regulatory mechanisms of these targets in SKP2 signaling pathway involving myxofibrosarcoma metastasis.

In contrast to the insensitivity seen in dermal fibroblasts, we also substantiated that low concentrations (10–20 nmol/L) of bortezomib could effectively suppress cell viability of both SKP2-overexpressing myxofibrosarcoma cell lines independent of SKP2 gene status. This bortezomib-induced decrease in the cell number was achieved by both activated caspase-mediated apoptosis and downregulated SKP2 expression, findings in keeping with those reported in the ovarian and colorectal cancer cell models by Uddin and colleagues (17, 18). As expected, the SKP2-suppressing effect was accompanied by concomitant reexpression of p27^Kip1 though inhibition of proteasomal degradation. Intriguingly, SKP2 downregulation by bortezomib treatment might be ascribed to transcriptional repression of SKP2 gene expression. This inference was based on the finding that there was remarkably decreased mRNA expression, rather than increased protein degradation, in both myxofibrosarcoma cell lines treated. The actions of bortezomib are pleiotropic (17, 18, 39, 40), whereas our data clearly showed that bortezomib treatment in SKP2-overexpressing myxofibrosarcomas could achieve cytotoxicity in vitro. Furthermore, the pharmacologic effect of bortezomib in cell models was reinforced by the in vivo studies of myxofibrosarcoma xenografts, showing significant tumor regression with lowered SKP2 labeling, increased p27^Kip1 expression, and increased apoptosis detected by TUNEL assay. However, it took almost 1 month for the xenografts treated with bortezomib to show significant growth inhibition. This suboptimal effect in vivo suggested that combination therapy is more desirable to treat myxofibrosarcomas than bortezomib alone at concentrations achievable in serum. In this regard, the role of ubiquitin-proteasome system has been recently highlighted in treatment of some solid cancers where synergistic antitumor effect was shown by combination of bortezomib and chemotherapeutic or novel targeted agents (41, 42).

In conclusion, aCGH profiling has been established SKP2 as a crucial molecule in promoting disease progression of myxofibrosarcomas, with independent in vitro and in vivo validation to characterize its prognostic, biologic, and therapeutic implications. Gene amplification represents a predominant mechanism driving mRNA and protein overexpression of SKP2, which, apart from accelerating cell growth, confers aggressive phenotypes by promoting cell migration and invasion, possibly through regulation of motility-associated genes. Furthermore, bortezomib treatment resulted in decreased viability of myxofibrosarcoma cells in vitro and regression of xenograft growth in vivo through suppression of SKP2 and induction of apoptosis, indicating the potentiality of ubiquitin-proteasome pathway as a therapeutic target in myxofibrosarcomas.

The complete aCGH database of myxofibrosarcoma samples and cell lines is available at Gene Expression Omnibus (GEO) under accession number GSE35483 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE35483).

No potential conflicts of interest were disclosed.

The authors thank the Genomic Core Laboratory (CMRPG880251) and Tissue Bank (CMRPG870461) of Chang Gung Memorial Hospital-Kaohsiung Medical Center, Kaohsiung, Taiwan, for critical technical assistance. The authors also thank Drs. Bridge and Ogose for providing the myxofibrosarcoma cell lines.

This work was supported in part by grants from National Science Council, Taiwan (96-2320-B-182A-008-MY3 to H.-Y. Huang; 99-2320-B-384-001-MY2 to C.-F. Li) and Chang Gung Memorial Hospital (CMRPG870753 and XMRPG890061, H.-Y. Huang).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.