点击显示 收起
【摘要】
Myxoid/round cell liposarcoma (MLS/RCLS) is the most common subtype of liposarcoma. Most MLS/RCLS carry a t(12;16) translocation, resulting in a FUS-DDIT3 fusion gene. We investigated the role of the FUS-DDIT3 fusion in the development of MLS/RCLS in FUS-DDIT3- and DDIT3-transfected human HT1080 sarcoma cells. Cells expressing FUS-DDIT3 and DDIT3 grew as liposarcomas in severe combined immunodeficient mice and exhibited a capillary network morphology that was similar to networks of MLS/RCLS. Microarray-based comparison of HT1080, the transfected cells, and an MLS/RCLS-derived cell line showed that the FUS-DDIT3- and DDIT3-transfected variants shifted toward an MLS/RCLS-like expression pattern. DDIT3-transfected cells responded in vitro to adipogenic factors by accumulation of fat and transformation to a lipoblast-like morphology. In conclusion, because the fusion oncogene FUS-DDIT3 and the normal DDIT3 induce a liposarcoma phenotype when expressed in a primitive sarcoma cell line, MLS/RCLS may develop from cell types other than preadipocytes. This may explain the preferential occurrence of MLS/RCLS in nonadipose tissues. In addition, development of lipoblasts and the typical MLS/RCLS capillary network could be an effect of the DDIT3 transcription factor partner of the fusion oncogene.
--------------------------------------------------------------------------------
Myxoid/round cell liposarcoma (MLS/RCLS) is the most common subtype of liposarcoma, accounting for about 40% of all cases. The tumor cells are characterized by the chromosomal aberration t(12;16)(q13;p11), which in most cases is the only cytogenetic abnormality.1 This translocation results in a FUS-DDIT3 (also called TLS-CHOP) fusion occurring in more than 90% of cases.2,3 A small percentage of cases carry a variant translocation t(12;22) and an EWSR1-DDIT3 fusion gene.4 None of these fusion oncogenes have been found in tumor types other than MLS/RCLS.
The fusion gene FUS-DDIT3 encodes a protein consisting of the N-terminal half of the FUS protein juxtaposed to the DNA-binding basic leucine zipper transcription factor DDIT3 (also known as CHOP or GADD153).5,6 The FUS-DDIT3 protein can be found in the nuclei of MLS/RCLS cells and maintains the capacity of DDIT3 to form dimers with other leucine zipper-possessing proteins.5 These and other observations suggest that FUS-DDIT3 may act as an abnormal transcription factor.7
FUS (also known as TLS), EWSR1, and TAF15 (the latter also known as TAF2N, TAFII68, or RBP56) are closely related genes that encode RNA-binding proteins. All three genes form fusion oncogenes with several alternative transcription factor-encoding genes.8 Each fusion combination is found in a distinct tumor type.
The mechanism behind the strict tumor-type specificity of these and many other fusion genes has long been the subject of debate.9-11 In "instructive" models, the fusion protein controls cell fate decisions, leading to development of a particular type of tumor. Conversely, in "noninstructive" models, the translocations occur only in susceptible committed cell types as a result of a cell type-specific translocation mechanism, or the fusion protein acts as an oncogene only in a susceptible committed cell type. Secondary events are required for tumor development in both models.
In the present study, we have analyzed the effects of the FUS-DDIT3 fusion gene and overexpression of the normal DDIT3 and the 5' part of FUS in the low-differentiated human fibrosarcoma cell line HT1080. Stable transfectant clones were analyzed for tumor formation in severe combined immunodeficient (SCID) mice, and the morphology of the resulting tumors was studied. Microarray analysis was used for expression profiling of the transfected cell lines, and the cells were also tested in vitro for their responsiveness to adipogenic factors.
【关键词】 myxoid/round liposarcoma oncogene fus-ddit liposarcoma phenotype transfected fibrosarcoma
Materials and Methods
Expression Vectors
Polymerase chain reaction (PCR) fragments containing the full-length coding regions of FUS-DDIT3 type II12 , DDIT3, or the sequences encoding the first 180 amino acids of FUS were cloned into a pEGFPN1 vector (Clontech) in-frame with the enhanced green fluorescent protein (EGFP) sequence as described previously.13 The resulting vectors were designated pFUS-DDIT3EGFP, pDDIT3-EGFP, and pFUSaEGFP. All vectors also contained a G418 resistance selection gene.
Cell Culture and Transfection
The fibrosarcoma cell line HT1080 and MLS cell lines 402-912 and MLS 2645-94 were stored in liquid nitrogen and cultivated in RPMI 1640 (Life Technologies, Grand Island, NY) supplemented with 8% fetal calf serum. Cultures were split 1:3 twice every week. The cells were harvested by trypsination, and 3 x 105 cells were seeded into 10-cm2 flasks and transfected the following day using 10 µl of FuGENE 6 transfection reagent (Roche, Indianapolis, IN) and 3 µg of plasmid DNA per flask. The cells were transferred to 75-cm2 culture flasks after 48 hours, and G418 (Sigma, St. Louis, MO) was added to a final concentration of 800 µg/ml. After 3 days, the G418 concentration was reduced to 500 µg/ml. Surviving cells were harvested after 10 days and seeded in 96-well microtiter plates at a density of 30 cells/plate. Clones showing nuclear EGFP fluorescence were expanded in the presence of 200 µg G418/ml, and recombinant protein expression was confirmed by Western blot analysis.
In Vitro Adipogenesis
pFUS-DDIT3EGFP-transfected, pDDIT3EGFP-transfected, and wild-type HT1080 cells were seeded in flasks and cultured in RPMI 1640 until they reached 100% confluence, after which the medium was changed to adipogenesis induction medium (PT3004 containing human recombinant insulin, dexamethasone, indomethacin, and 3-isobutyl-L-methylxanthine ; Cambrex, East Rutherford, NJ) or maintenance medium (MM; RPMI 1640 containing 8% fetal calf serum). The cells were treated with adipogenesis induction medium for 3 days, followed by 1 to 3 days in MM. This was repeated three times. Control cultures were fed with only MM following the same schedule. After completed cycles, the cells were cultured for 7 more days in MM with replacement of the medium every 2 to 3 days. The cells were inspected using a microscope, and accumulation of fat was assessed by staining the cells with Oil Red O after fixation with 4% buffered formalin.
Western Blot Analysis
Western blot of total proteins from cultured cells was performed after electrophoresis using precast 3 to 8% NuPage Tris-acetate gels (Invitrogen, Carlsbad, CA), followed by transfer in a Novex minigel electrophoresis apparatus (Novex, San Diego, CA) to Immobilon membranes (Millipore, Billerica, MA) according to the manufacturers?? recommendations. The procedures for antibody binding and detection were as described by the suppliers of the antibodies. The DDIT3 protein was detected using a polyclonal rabbit antiserum (GADD 153, R20, sc793; Santa Cruz Biotechnology, Santa Cruz, CA), and the FUS protein was detected with a polyclonal rabbit antiserum against the N terminus of FUS.
Genome-Wide Gene Expression Analysis
Total RNA was extracted from cultured cells using the Trizol method (Invitrogen). The RNA was further purified using a Qiagen RNAeasy kit and stored at C140??C. Five micrograms of RNA was used for cDNA synthesis with Cy3- and Cy5-labeled nucleotides. cDNA labeling and hybridization to microarray glasses were performed according to the Pronto! Plus Direct Labeling v1.2 labeling and hybridization kit (InVitro, Stockholm, Sweden).
Microarrays were produced at the Swegene DNA Microarray Resource Center, Department of Oncology, Lund University (Lund, Sweden). Human array-ready oligonucleotide libraries (version 2.1, catalog no. 810516; and version 2.1.1 upgrade, catalog no. 810518), comprising approximately 27,000 unique probes representing 23,707 genes, were obtained from Operon Biotechnologies (Cologne, Germany). Probes were dissolved in Universal Spotting Solution (Corning, Acton, MA) and printed in duplicate on aminosilane-coated glass slides (UltraGAPS, catalog no. C40017; Corning) using a MicroGrid2 robot (BioRobotics, Cambridgeshire, UK) equipped with MicroSpot 10K pins (BioRobotics). After printing, arrays were left in a desiccator to dry for 48 hours, rehydrated for 1 second over steaming water, snap-dried on a hot plate (98??C), and UV-cross-linked (800 mJ/cm2).
Image and Data Analysis
Hybridized microarrays were scanned using an Agilent G2565AA microarray scanner (Agilent Technologies, Palo Alto, CA). Fluorescence intensities were extracted using Gene Pix Pro 4.0 software (Axon Instruments Inc., Foster City, CA), and uploaded into Bio Array Software Environment (http://base.thep.lu.se) for further analysis. The LOWESS algorithm14,15 was used for normalization.
Minimum median intensity was set to 1 to avoid data loss when the ratio between the samples and the reference cell line (wild-type HT1080) was being calculated. Genes with differences of at least threefold in signal intensity between the HT1080 reference cell line and the transfected HT1080 variants were scored as up- or down-regulated. Genes with a normalized intensity below 25 in the transfected cell lines were rejected from the lists of up-regulated genes. Similarly, genes with a normalized intensity lower than 25 in the HT1080 reference were rejected from the lists of down-regulated genes. RNA from the MLS cell line 402-91 was also tested, with the HT1080 cell line as a reference.
Statistical Analysis
To estimate the statistical significance of the overlap between differentially expressed genes in 402-91 cells and differentially expressed genes in the pFUS-DDIT3EGFP-and pDDIT3EGFP-transfected lines, two sets of genes corresponding to the up- and down-regulated genes in 402-91 cells were selected at random from the total of 23,707 genes. Similarly, two sets of genes corresponding to up- and down-regulated genes in one of the transfected lines were randomly selected. The number of genes that were identical in the randomly selected groups was then observed. This experiment was repeated 107 times for each pair of up- or down-regulated gene sets.
Quantitative Reverse Transcriptase-PCR Analysis
Total RNA was isolated, and cDNA was generated as described above but without labeled nucleotides. The primer sequences were as follows (F, forward; R, reverse): CTGFF, 5'-CCG TAC TCC CAA AAT CTC CA-3'; CTGFR, 5'-GTA ATG GCA GGC ACA GGT CT-3'; DNAJA1F, 5'-ACT GGA GCC AGG CGA TAT TA-3'; DNAJA1R, 5'-GAA GCC ACA CAG TGC TTC AA-3'; EFEMP1F, 5'-TGC CAT CAG ACA TCT TCC AG-3'; EFEMP1R, 5'-TGT GCG GAA GGT CCC TAT AC-3'; HSPA1F, 5'-CCG AGA AGG ACG AGT TTG AG-3'; HSPA1R, 5'-GCA GCA AAG TCC TTG AGT CC-3'; MMP3F, 5'-CAG GCT TTC CCA AGC AAA TA-3'; MMP3R, 5'-GTG CCC ATA TTG TGC CTT CT-3'; NPAS2F, 5'-GCC AGA CCG TGT TTC AAA AT-3'; NPAS2R, 5'-AGT AGC GAG TCC TGC TGC TC-3'; PBEF1F, 5'-GGA GCA TCT GCT CAC TTG GT-3'; PBEF1R, 5'-TCG CTG ACC ACA GAT ACA GG-3'; PCNAF, 5'-AAA ATT GCG GAT ATG GGA CA-3'; PCNAR, 5'-GCT GGC ATC TTA GAA GCA GTT-3'; PTX3F, 5'-CCA ATG CGT TCC AAG AAG AT-3'; PTX3R, 5'-TCC ACC CAC CAC AAA CAC TA-3'; SERPINB2F, 5'-CTC AAC AAG TGG ACC AGC AA-3'; SERPINB2R, 5'-CCA TGT CCA GTT CTC CCT GT-3'; SKP2F, 5'-CAT TTC AGC CCT TTT CGT GT-3'; SKP2R, GGG CAA ATT CAG AGA ATC CA-3'; VEGFF, 5'-CCC ACT GAG GAG TCC AAC AT-3'; VEGFR, 5'-AAA TGC TTT CTC CGC TCT GA-3'; GAPDHF, 5'-GTG AAG GTC GGA GTC AAC G-3'; GAPDHR, 5'-GGT GAA GAC GCC AGT GGA CTC-3'; ONP54F, 5'-AGC CCA CTT CTT ACC ACA AG-3'; and ONP56R, 5'-CCA AAT AGG TGC ATG AGT AG-3'.
Real-time PCR was measured in an ABI 7700 sequence detector (Applied Biosystems, Foster City, CA) with SYBR Green as fluorophore. Formation of expected PCR product was confirmed by agarose gel electrophoresis and melting curve analysis. Gene expression data were normalized against ONP and GAPD by geometric averaging.16
Tumor Growth in SCID Mice
Female FOX CHASE SCID mice were used as recipients of transfected HT1080 clones and nontransfected HT1080 cells. Cultured cells were harvested by trypsinization and suspended in phosphate-buffered saline (PBS) before injection. The mice were 5 to 6 weeks old at the time of inoculation. In two separate experiments, 10 and 2 million cells, respectively, were injected subcutaneously into the flank of SCID mice. In the first experiment, three animals were injected with the HT1080 wild-type cell line, and three series of five animals were injected with FUSaEGFP-, FUSD-DITT3EGFP-, and DDIT3EGFP-transfected cells, respectively. The second experiment comprised two animals injected with HT1080 wild-type cells and two groups of five animals injected with FUSD-DITT3EGFP- or DDIT3EGFP-transfected cells. The animals were examined regularly for evidence of tumor growth at the site of injection. Animals with palpable tumors were sacrificed, and the tumors were excised and measured. Pieces were frozen or fixed in formalin and embedded in paraffin for histological examination. Tissue sections (5 µm thick) were deparaffinized and stained following routine protocols.
Fluorescence Microscopy
The cultures were washed twice with PBS and fixed in 4% paraformaldehyde in PBS. After two more washes in PBS, the slides were mounted in an antifade mount containing the DNA binding dye 4,6-diamidino-2-phenylindole dihydrochloride and examined in a fluorescence microscope. Fresh, living cells and tissue fragments were examined directly in culture flasks using an inverted fluorescence microscope.
Immunohistochemistry
Series of 5-µm tissue sections were cut from each biopsy, deparaffinized, rehydrated, and stained with the following monoclonal antibodies: rat anti-mouse CD34 (ABcam ab8158; Abcam, Cambridge, MA) and mouse anti-human CD34 (DAKO, Dakopatts A/S, Glostrup, Denmark). Bound antibodies were visualized using the LSAB second antibody streptavidine biotin peroxidase system (DAKO). Stained sections were examined with a light microscope.
Results
G418 selection for stable transfectant clones revealed a clear difference between cells transfected with FUS-DDIT3, DDIT3, and FUSa. All three constructs gave similar rates of transiently transfected cells, as judged by counting of cells expressing the GFP-tagged proteins by fluorescence microscopy 24 hours after transfection (Table 1) . However, only a low percentage of the G418-resistant pFUS-DDIT3EGFP-transfected cells expressed the FUS-DDIT3EGFP protein after 10 days of selection, whereas more than 90% of the G418-resistant pDDIT3EGFP and pFUSaEGFP-stably transfected cells expressed DDIT3EGFP and FUSaEGFP, respectively. This observation suggests that only a small proportion of the cells expressing the FUS-DDIT3EGFP protein remained viable. We also noticed that the pFUS-DDIT3EGFP-transfected clones expressed considerably less of the GFP-tagged protein than the pDDIT3EGFP- and pFUSaEGFP-transfected cells (Figure 1) . The amount of FUS-DDIT3EGFP protein was also lower compared with the levels of native FUS-DDIT3 protein seen in MLS/RCLS cell lines.
Table 1. Growth of Transfected HT1080 Cells
Figure 1. Western blot analysis of transfected HT1080 cells and MLS 2645-94 from which the FUS-DDIT3 fusion cDNA was cloned. Sample origin is indicated for each lane. A mix of anti-DDIT3 and anti-FUS antibodies were used to detect all variant proteins. 1, the position of the recombinant FUS-DDIT3EGFP protein; 2, the normal FUS protein present in all cells; 3, the recombinant N-terminal part of FUS (FUSa) linked to EGFP; 4, the native FUS-DDIT3 protein of the MLS cells from which the fusion transcript was cloned; and 5, the recombinant DDIT3EGFP protein. Note that the EGFP tag adds around 30 kd in size to the recombinant proteins. Size markers in kilodaltons.
The sizes of the GFP-tagged proteins were checked by Western blot analysis, and clones showing stable expression of proteins of the correct size were selected for further analysis (Figure 1) .
Tumor Growth in SCID Mice
In the first experiment, 107 cells were inoculated into each mouse. All mice, except one, rapidly developed large tumors, and in the following experiment, the amount of cells was reduced to 2 x 106 per animal. Judging from tumor sizes, the original HT1080 cells grew faster in SCID mice than the three transfected cell lines (Table 2) .
Table 2. Tumor Growth and Morphology in SCID Mouse
Microscopic examination of sections from the tumors revealed that the original HT1080 cells and pFUSaEGFP-transfected cells grew as low-differentiated sarcomas, mainly with poorly defined or sinusoid blood vessels and necrotic areas. In contrast, most of the mice inoculated with the pFUS-DDIT3EGFP- or pDDIT3EGFP-expressing cells developed liposarcomas containing atypical lipoblasts of different sizes (Figure 2A) . Increased proportions of extracellular matrix and small myxoid pools were observed in tumors from the two pFUS-DDIT3EGFP-transfected cell lines, but large amounts of the myxoid substance seen in many naturally occurring MLS tumors were not found. Supplemental Table S1 containing detailed information can be found at http://ajp.amjpathol.org.
Figure 2. A: Microphotographs of sections from SCID mouse tumors of HT1080 cells (left column) and the pFUS-DDIT3EGFP- and pDDIT3EGFP-transfected variants (middle and right columns, respectively). Note irregularly sized lipoblasts in the transfected tumors. Original magnifications, x400 and x600. B: Anti-D34 staining of capillaries in SCID mouse tumors from HT1080 and the two transfected variants. Note thin branching capillary vessels in the transfected tumors. Microphotographs captured at x200 magnification. C: In vitro adipogenesis of HT1080 cells and transfected variants. The upper line of photomicrographs shows the indicated cell lines after adipogenic treatment. Control cultured cells are shown in the lower series of photomicrographs. Accumulated lipids are stained red (Oil Red O staining). Original magnification, x740.
Effects on Vascularization
In humans, most MLS/RCLSs are characterized by a typical plexiform capillary network.17 We found that the tumors grown from cells carrying pFUS-DDIT3EGFP and pDDIT3EGFP contained capillary networks with morphology similar to that found in naturally occurring MLS, whereas the original HT1080 and pFUSaEGFP-transfected tumors mainly contained sinusoid vessels (Figure 2B ; Table 2 ). The effects on capillary morphology prompted us to stain for CD34+ endothelial cells in the tumors. Immunohistochemical staining of tumor sections with antibodies to mouse endothelial cells (anti-CD34), revealed a rich abundance of mouse CD34+ vessels in all tumors (Figure 2B) . No expression of human CD34 was detected, indicating that the capillary networks originated from mouse cells. We also noted that the original HT1080 and pFUSaEGFP transfectants contained considerably more vessels than the pFUS-DDIT3EGFP- and pDDIT3EGFP-transfected tumors.
In Vitro Adipogenesis Induction Experiments
The change in tumor phenotype observed in the SCID mice may reflect an increased sensitivity to in vivo factors that induce adipogenic development in the pDDIT3EGFP and pFUS-DDIT3EGFP cells. To test this hypothesis, we investigated the effects of an in vitro adipogenesis protocol on the HT1080 cells and the transfected sublines. Treatment with adipogenic factors induced a dramatic morphological change of pDDIT3EGFP-transfected cells. The cells developed large vacuoles, and some cells resembled the "signet ring cell" type of adipocytes (Figure 2C) . Oil Red O staining showed increased fat accumulation, but the contents of the main vacuole in each cell showed only partial staining. The pFUS-DDIT3EGFP transfectants showed a slight increase in fat content compared with the original HT1080 line, but this was not increased in adipogenic medium, and no morphological response or formation of large vacuoles was seen. The original HT1080 cells grew faster than the transfected variants. Adipogenic factors induced minor morphological changes, with the formation of small vacuoles in a few cells and a slightly increased accumulation of fat (Figure 2C) .
Microarray Expression Analysis
Microarray-based expression analysis of HT1080 cells, pDDIT3EGFP- and pFUS-DDIT3EGFP-transfected HT1080 cells, and MLS-derived cell line 402-91, showed that only small numbers of genes were differentially expressed in the original HT1080 line and the transfected variant cell lines. In pDDIT3EGFP-transfected HT1080 cells, 36 genes were up-regulated and 98 were down-regulated, and in pFUS-DDIT3EGFP transfectants, 51 genes were up-regulated and 82 were down-regulated, three times or more compared with H1080 cells. In MLS line 402-91, 364 genes were up-regulated and 309 genes were down-regulated, three times or more compared with the HT1080 cells. To test whether FUS-DDIT3 and DDIT3 induced or repressed genes in an MLS/RCLS like pattern, we searched our microarray data for genes that were up- or down-regulated both in the transfected cell lines and in the 402-91 cell line. Many of the genes that were found to be differentially regulated in the pFUS-DDIT3EGFP- and pDDIT3EGFP-transfected HT1080 cells were also found to differ in expression between MLS 402-91 and the HT1080 cell line. (Tables 3 and 4 ; Figure 3 ). The probability of obtaining a similar overlap in up- and down-regulated genes by random selection among the 23,707 genes on the microarray is less than 10C7 in each of the comparisons. This shows that when DDIT3 and FUS-DDIT3 are expressed in HT1080 cells, their gene expression changes toward an MLS 402-91-like pattern. Supplemental data from the microarray experiments is available at http://ajp.amjpathol.org.
Table 3. Named Genes Up-Regulated in HT1080 Transfected with FUS-DDIT3 and DDIT3, and in MLS 402-91
Figure 3. Number of genes that were at least three times up-regulated () and down-regulated () in DDIT3- and FUS-DDIT3-expressing HT1080 cells as compared by microarray analysis with control HT1080 cells. Shaded parts of bars indicate number of genes that are identical with genes up- or down-regulated in MLS 402-91 cells. Total number of genes that were represented on the CHIP was 23,707.
Discussion
The MLS/RCLS-specific FUS-DDIT3 belongs to the EWS group of fusion oncogenes. All fusion oncogenes of this group contain the 5' part of one of the closely related genes FUS, EWSR1, and TAF15 as 5' partner and one of many different transcription factor genes as 3' partner.18 Each combination of fusion partners is specifically found in one or, in a few examples, two different tumor types. FUS-DDIT3 and EWSR1-DDIT3 are found only in MLS/RCLS.
FUS-DDIT3 has been previously reported to block terminal differentiation of preadipocytes in vivo and in vitro,19,20 and transgenic mice that expressed FUS-DDIT3 in all tissues developed MLS/RCLS-like tumors that arose in adipose tissue.21 These reports lead to the hypothesis that MLS/RCLS develops from preadipocytes carrying FUS-DDIT3 that are incapable of terminal differentiation.19-24
The observation that the majority of human MLS/RCLS tumors arise in or between the large muscles of the limbs, and rarely in adipose tissues,1 calls this hypothesis into question. In the present experiments with pFUS-DDIT3EGFP- and pDDIT3EGFP-transfected xenografted HT1080 cells, we observed a FUS-DDIT3- and DDIT3-induced switch from a low-differentiated sarcoma to an MLS/RCLS-like morphology. Microarray expression profiling showed that transfection of FUS-DDIT3 or DDIT3 induced a significantly changed expression pattern of HT1080 cells toward an MLS-like profile. Taken together, our morphology and microarray data show that FUS-DDIT3 can drive a primitive mesenchymal tumor cell toward an MLS/RCLS phenotype by promoting or inhibiting expression of a limited number of genes. Thus, FUS-DDIT3 may act as an instructive factor to induce a liposarcoma phenotype. Similar results, indicating instructive activities of EWS group fusion oncogenes, have been published or suggested by other research groups.25-28 A phenotype-determining mechanism was also suggested in 1914 by Peyton Rous29 for sarcomas caused in chickens by "filterable agents," later identified as oncogene-carrying retroviruses. Our results also show that expression of the nontransforming fusion partner gene DDIT3 induces a similar switch in morphology and expression pattern. These findings support the hypothesis that the 3' transcription factor partner genes of the EWS group of fusion oncogenes determine the tumor entity.
In vitro experiments with transfected mouse preadipocytic cells have shown that FUS-DDIT3 blocks adipocyte differentiation, because the cells fail to exit the cell cycle.22 In contrast, DDIT3-expressing cells exit the cell cycle and develop into adipocytes. This is compatible with our finding that HT1080 cells carrying pFUS-DDIT3EGFP failed to show increased fat accumulation on treatment with adipogenic factors, and similar results were obtained with the FUS-DDIT3-carrying MLS cell line 402-91 (data not shown). The control-cultured pFUS-DDIT3EGFP-transfected cells showed a minor increase in fat content compared with the original HT1080 cell line. This may be a sign of an interrupted adipogenesis, but it may also be an effect of the slower growth of the transfected cell line in adipogenic medium allowing time for more fat accumulation.
The negative in vitro adipogenesis results obtained with FUS-DDIT3-transfected 3T3-fibroblast or HT1080 cells or MLS/RCLS cell lines are in conflict with data obtained from primary cultures of MLS/RCLS tumor cells.24 However, unknown in vivo factors may add to the adipogeneic response, and these factors may be lost after prolonged in vitro culture. In vivo factors might also explain the fact that tumors from the transfected cells and human MLS/RCLS tumors contain lipoblasts far more advanced than those obtained after adipogenic treatment of MLS/RCLS cell lines or FUS-DDIT3-transfected preadipocytes.
The DDIT3-transfected HT1080 cells accumulated large amounts of fat, and many cells developed a "signet ring" lipoblast phenotype when treated with adipogenic factors in vitro. The original HT1080 cells showed a much weaker response. We speculate that forced expression of DDIT3 triggered the first steps of an adipocyte differentiation program that resulted in responsiveness to adipogenic factors.
Searches in our microarray results for FUS-DDIT3- and DDIT3-induced or -suppressed genes reported to have some role in adipogenesis identified DSIPI. DSIPI encodes a leucine zipper type of transcription factor that has been shown to inhibit adipogenesis.30 The reduced expression of DSIPI in FUS-DDIT3- and DDIT3-transfected cells may be an important factor behind the morphological shift seen in the SCID mouse tumors.
In previous studies, only FUS-DDIT3 (but not DDIT3) was found to transform normal cells or cause tumors in transgenes, thus excluding the normal DDIT3 as a transforming gene.19,21,22,31 Our observation that pDDIT3EGFP-transfected HT1080 cells grew as myxoid liposarcomas with the capillary network and abnormal lipoblasts shows that constitutive expression of the normal DDIT3 protein may induce an MLS-like morphology when expressed in a mesenchymal tumor. In this context, it is important to note that our DDIT3-carrying vector construct lacked the 5'-untranslated region of the DDIT3 transcript, which is known to play an important role in limiting production of the protein at a translation checkpoint.32 This critical part of the DDIT3 transcript is also deleted in the MLS/RCLS fusions. Both MLS/RCLS cells and our pDDIT3EGFP-transfected HT1080 cells contain more of variant DDIT3 than the levels of DDIT3 found in normal lipoblasts or adipocytes (data not shown). Further analysis of FUS/EWSR1-DDIT3-negative MLS/RCLS-like tumors is required to find out whether de-regulated DDIT3 expression may be associated with a liposarcoma phenotype.
Most MLS/RCLS tumors are characterized by a fine plexiform network of capillaries.17 A similar capillary network emerged in most tumors of pFUS-DDIT3EGFP- and pDDIT3EGFP-transfected cells (Figure 2) . Tumors of original HT1080 or pFUSaEGFP-transfected cells showed a different pattern, dominated by poorly defined sinusoid vessels. Staining of tumor sections with anti-human and anti-mouse CD34 antibodies showed that all endothelial cells were of mouse origin. Thus, tumor-produced factors affect the angiogenesis and vessel morphology. The lower expression of VEGF in the FUS-DDIT3- and DDIT3-transfected cells may partially explain the differences in capillary formation. Another factor that may affect capillary morphology is interleukin-6.33 Interleukin-6 has recently been reported to be produced by MLS/RCLS cells and up-regulated in DDIT3- or FUS-DDIT3-transfected HT1080 cells.34 DOL54, an angiogenesis-inducing gene previously reported to be induced by DDIT3 and FUS-DDIT3,35 was equally expressed in HT1080 and its transfected variants (data not shown).
Our analysis showed that the expression of a limited number of genes was changed in the DDIT3- or FUS-DDIT3-containing HT1080 cells. The small overlap in differentially expressed genes between the two transfected cell lines may indicate that DDIT3 and its oncogenic derivative, FUS-DDIT3, differ with regard to their effects on target genes. This interpretation is supported by several other studies of the EWS group of fusion oncogenes, showing that addition of the N termini of EWSR1 or FUS protein to transcription factors changes their activities as transcriptional regulators. We found, however, that several of the differentially regulated genes, most notably stress-induced genes, were overexpressed both in DDIT3-transfected HT1080 cells and in the MLS 402-91 line. The pFUS-DDIT3EGFP-transfected HT1080 cells differed in this respect. It is possible that expression of DDIT3 or FUS-DDIT3 must exceed a threshold level before induction of stress response genes occurs and that the low level of FUS-DDIT3 in the transfected HT1080 cells was insufficient to trigger the stress response genes and some other DDIT3 target genes.
SERPINB2 was one of the most strongly up-regulated genes common to pFUS-DDIT3EGFP-transfected HT1080 cells and MLS cell lines. This gene was not induced in pDDIT3-transfected HT1080 cells, indicating an FUS-DDIT3-specific induction mechanism. Because FUS-DDIT3, but not DDIT3, is a transforming gene, SERPINB2 may be an important factor in the transformation process. SERPINB2, also known as PAI2, encodes a plasminogen activator inhibitor that may switch between different conformations.36 However, the protein has also been reported as a nuclear protein, binding to the pRB protein product of the retinoblastoma tumor suppressor gene.37 The multiple roles of the SERPINB2 protein in cell cycle control and extracellular matrix regulation make it interesting in the context of FUS-DDIT3-induced transformation, and this finding will be followed up in our continued research.
MLS-like tumors develop from adipose tissues in FUS-DDIT3 transgenic mice, whereas human MLS/RCLSs most often arise inside large muscles. The differences between naturally occurring human tumors and those developing in transgenic mice suggest that the target cells and/or microenvironmental requirements for transformation by FUS-DDIT3 may differ in the two species. We have reported previously that most pFUS-DDIT3EGFP-transfected human cells die shortly after the fusion protein is expressed,13 and the HT1080 cell line was no exception (Table 1) . The single clone that we succeeded in retrieving was one of few stable FUS-DDIT3EGFP- expressing HT1080 clones observed in the cloning step. Sequencing of the inserted FUS-DDIT3 and Western blot confirmation of protein size showed that the cells expressed the expected EGFP-tagged type II FUS-DDIT3 protein but at a low level. The low level of FUS-DDIT3EGFP was probably of importance for the survival of this clone. The surviving clone may also have originated from a rare variant of the HT1080 cells that was compatible with expression of FUS-DDIT3. A similar selection mechanism may take place in the development of MLS/RCLS in humans. Observations of toxic effects have also been reported for the related EWSR1-FLI1 fusion oncogene found in Ewing sarcoma. Mutations or silencing of CDKN2A/p16 or TP53 were found to increase cell survival of EWSR1-FLI1-transfected cells.38 This suggests that cells with loss of p16 or p53 functions may be preferential target cells for EWSR1-FLI1 transformation. Similarly, unknown mutations or epigenetic changes may increase the compatibility with FUS-DDIT3 expression. In this context, it is interesting to note that the HT1080 cell line carries a normal copy of the TP53 but has deleted the CDKN2A/p16.
We conclude that forced expression of FUS-DDIT3 and the normal DDIT3 induces a switch in HT1080 cells toward an MLS/RCLS-like morphology and gene expression pattern. The results also indicate that FUS-DDIT3 and DDIT3 may act as instructive factors to induce a liposarcoma phenotype and that the transcription factor partner DDIT3 of FUS-DDIT3 is the tumor type determining part of this EWS-group fusion oncogene.
Table 4. Named Genes Down-Regulated in HT1080 Transfected with FUS-DDIT3 and DDIT3, and in MLS 402-91
Table 4A. Continued
【参考文献】
Fletcher CDM, Unni KK, Mertens F: Tumors of Soft Tissue and Bone. 2000 IARC Press Lyon, France
Åman P, Ron D, Mandahl N, Fioretos T, Heim S, Arheden K, Willen H, Rydholm A, Mitelman F: Rearrangement of the transcription factor gene CHOP in myxoid liposarcomas with t(12;16)(q13;p11). Genes Chromosomes Cancer 1992, 5:278-285
Antonescu CR, Elahi A, Humphrey M, Lui MY, Healey JH, Brennan MF, Woodruff JM, Jhanwar SC, Ladanyi M: Specificity of TLS-CHOP rearrangement for classic myxoid/round cell liposarcoma: absence in predominantly myxoid well-differentiated liposarcomas. J Mol Diagn 2000, 2:132-138
Panagopoulos I, Åman P, Mandahl N, Mitelman F: Two distinct FUS breakpoint clusters in myxoid liposarcoma and acute myeloid leukemia with the translocations t(12;16) and t(16;21). Oncogene 1995, 11:1133-1137
Crozat A, Åman P, Mandahl N, Ron D: Fusion of CHOP to a novel RNA-binding protein in human myxoid liposarcoma. Nature 1993, 363:640-644
Rabbitts TH, Forster A, Larson R, Nathan P: Fusion of the dominant negative transcription regulator CHOP with a novel gene FUS by translocation t(12;16) in malignant liposarcoma. Nat Genet 1993, 4:175-180
Sanchez-Garcia I, Rabbitts TH: Transcriptional activation by TAL1 and FUS-CHOP proteins expressed in acute malignancies as a result of chromosomal abnormalities. Proc Natl Acad Sci USA 1994, 91:7869-7873
Åman P: Fusion genes in solid tumors. Semin Cancer Biol 1999, 9:303-318
Barr FG: Translocations, cancer and the puzzle of specificity. Nat Genet 1998, 19:121-124
Åman P: Fusion oncogenes in tumor development. Semin Cancer Biol 2005, 15:236-243
Åman P: Fusion oncogenes. Semin Cancer Biol 2005, 15:159-161
Panagopoulos I, Mandahl N, Ron D, Hoglund M, Nilbert M, Mertens F, Mitelman F, Åman P: Characterization of the CHOP breakpoints and fusion transcripts in myxoid liposarcomas with the 12;16 translocation. Cancer Res 1994, 54:6500-6503
Thelin-Järnum S, Göranssson M, Schweizer-Burguete A, Olofsson A, Åman P: The Myxoid liposarcoma specific TLS-CHOP fusion protein localizes to nuclear structures distinct from PML nuclear bodies. Int J Cancer 2002, 97:446-450
Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J, Speed TP: Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res 2002, 30:15
Cleveland WS, Devlin SJ: Locally weighted regression: an approach to regression analysis by local fitting J Am Stat Assoc 1988, 83:596-610
Stålberg A, Zoric N, Åman P, Kubista M: Quantitative real-time PCR for cancer detection: the lymphoma case. Expert Rev Mol Diagn 2005, 5:221-230
Fletcher CDM, Unni KK, Mertens F: Pathology and Genetics of Tumors of Soft Tissue and Bone. 2002 IARC Press Lyon, France
Kovar H: Context matters: the hen or egg problem in Ewing??s sarcoma. Semin Cancer Biol 2005, 15:189-196
Zinszner H, Albalat R, Ron D: A novel effector domain from the RNA-binding protein TLS or EWS is required for oncogenic transformation by CHOP. Genes Dev 1994, 8:2513-2526
Kuroda M, Ishida T, Takanashi M, Satoh M, Machinami R, Watanabe T: Oncogenic transformation and inhibition of adipocytic conversion of preadipocytes by TLS/FUS-CHOP type II chimeric protein. Am J Pathol 1997, 151:735-744
Perez-Losada J, Pintado B, Gutierrez-Adan A, Flores T, Banares-Gonzalez B, del Campo JC, Martin-Martin JF, Battaner E, Sanchez-Garcia I: The chimeric FUS/TLS-CHOP fusion protein specifically induces liposarcomas in transgenic mice. Oncogene 2000, 19:2413-2422
Barone MV, Crozat A, Tabaee A, Philipson L, Ron D: CHOP (GADD153) and its oncogenic variant, TLS-CHOP, have opposing effects on the induction of G1/S arrest. Genes Dev 1994, 8:453-464
Ron D: TLS-CHOP and the role of RNA-binding proteins in oncogenic transformation. Curr Top Microbiol Immunol 1997, 220:131-142
Tontonoz P, Singer S, Forman B, Sarraf P, Fletcher J, Fletcher C, Brun R, Mueller E, Altiok S, Oppenheim H, Evans R, Spiegelman B: Terminal differentiation of human liposarcoma cells induced by ligands for peroxisome proliferator-activated receptor gamma and the retinoid kappa receptor. Proc Natl Acad Sci USA 1997, 94:237-241
Teitell MA, Thompson AD, Sorensen PH, Shimada H, Triche TJ, Denny CT: EWS/ETS fusion genes induce epithelial and neuroectodermal differentiation in NIH 3T3 fibroblasts Lab Invest 1999, 79:1535-1543
Rorie CJ, Thomas VD, Chen P, Pierce HH, O??Bryan JP, Weissman BE: The Ews/Fli-1 fusion gene switches the differentiation program of neuroblastomas to Ewing sarcoma/peripheral primitive neuroectodermal tumors. Cancer Res 2004, 64:1266-1277
Hu-Lieskovan S, Zhang J, Wu L, Shimada H, Schofield DE, Triche TJ: EWS-FLI1 fusion protein up-regulates critical genes in neural crest development and is responsible for the observed phenotype of Ewing??s family of tumors. Cancer Res 2005, 65:4633-4644
Perez-Mancera PA, Sanchez-Garcia I: Understanding mesenchymal cancer: the liposarcoma-associated FUS-DDIT3 fusion gene as a model. Semin Cancer Biol 2005, 15:206-214
Rous P, Murphy JB: On the causation by filterable agents of three distinct chicken tumors. J Exp Med 1914, 19:52-68
Shi X, Shi W, Li Q, Song B, Wan M, Bai S, Cao X: A glucocorticoid-induced leucine-zipper protein, GILZ, inhibits adipogenesis of mesenchymal cells. EMBO Rep 2003, 4:374-380
Perez-Losada J, Sanchez-Martin M, Rodriguez-Garcia MA, Perez-Mancera PA, Pintado B, Flores T, Battaner E, Sanchez-Garcia I: Liposarcoma initiated by FUS/TLS-CHOP: the FUS/TLS domain plays a critical role in the pathogenesis of liposarcoma. Oncogene 2000, 19:6015-6022
Jousse C, Bruhat A, Carraro V, Urano F, Ferrara M, Ron D, Fafournoux P: Inhibition of CHOP translation by a peptide encoded by an open reading frame localized in the chop 5'UTR. Nucleic Acids Res 2001, 29:4341-4351
Motro B, Itin A, Sachs L, Keshet E: Pattern of interleukin 6 gene expression in vivo suggests a role for this cytokine in angiogenesis. Proc Natl Acad Sci USA 1990, 87:3092-3096
Göransson M, Elias E, Stålberg A, Olofsson A, Andersson C, Åman P: Myxoid liposarcoma FUS-DDIT3 fusion oncogene induces C/EBP beta-mediated interleukin 6 expression. Int J Cancer 2005, 1:1
Kuroda M, Wang X, Sok J, Yin Y, Chung P, Giannotti JW, Jacobs KA, Fitz LJ, Murtha-Riel P, Turner KJ, Ron D: Induction of a secreted protein by the myxoid liposarcoma oncogene. Proc Natl Acad Sci USA 1999, 96:5025-5030
Wilczynska M, Lobov S, Ohlsson PI, Ny T: A redox-sensitive loop regulates plasminogen activator inhibitor type 2 (PAI-2) polymerization. EMBO J 2003, 22:1753-1761
Darnell GA, Antalis TM, Johnstone RW, Stringer BW, Ogbourne SM, Harrich D, Suhrbier A: Inhibition of retinoblastoma protein degradation by interaction with the serpin plasminogen activator inhibitor 2 via a novel consensus motif. Mol Cell Biol 2003, 23:6520-6532
Deneen B, Denny CT: Loss of p16 pathways stabilizes EWS/FLI1 expression and complements EWS/FLI1 mediated transformation. Oncogene 2001, 20:6731-6741
作者单位：From the Department of Oncology,* and the Lundberg Laboratory for Cancer Research, Department of Pathology, the Gothenburg, University, Sahlgrenska University Hospital, Gothenburg; the Department of Genetics and Pathology, Rudbecklaboratoriet, Uppsala; and the Swegene Facility for Bioinformatics, Go