MicroRNAs (miRNAs) are ∼22 nucleotide-long noncoding RNAs involved in several biological processes including development, differentiation and proliferation. Recent studies suggest that knowledge of miRNA expression patterns in cancer may have substantial value for diagnostic and prognostic determinations as well as for eventual therapeutic intervention. We performed comprehensive analysis of miRNA expression profiles of 27 sarcomas, 5 normal smooth muscle and 2 normal skeletal muscle tissues using microarray technology and/or small RNA cloning approaches. The miRNA expression profiles are distinct among the tumor types as demonstrated by an unsupervised hierarchical clustering, and unique miRNA expression signatures were identified in each tumor class. Remarkably, the miRNA expression patterns suggested that two of the sarcomas had been misdiagnosed and this was confirmed by reevaluation of the tumors using histopathologic and molecular analyses. Using the cloning approach, we also identified 31 novel miRNAs or other small RNA effectors in the sarcomas and normal skeletal muscle tissues examined. Our data show that different histological types of sarcoma have distinct miRNA expression patterns, reflecting the apparent lineage and differentiation status of the tumors. The identification of unique miRNA signatures in each tumor type may indicate their role in tumorigenesis and may aid in diagnosis of soft tissue sarcomas.
Sarcomas are a heterogeneous group of malignant mesenchymal tumors that can occur in a wide range of age groups. A large number of different diagnoses have been described within this tumor group and pathologic diagnosis can be challenging. Currently few markers exist to help distinguish sarcoma subtypes, yet the recent advent of targeted drug therapies—as in the case of gastrointestinal stromal tumor (GIST) and dermatofibrosarcoma protuberans—makes accurate diagnosis imperative (Weiss and Goldbum, 2001).
MicroRNAs (miRNAs) are short, processed, RNA molecules ∼22 nucleotides in length that can control gene function through mRNA degradation, translation inhibition or chromatin-based silencing mechanisms (Doench and Sharp, 2004). In humans, about 500 miRNAs have been discovered so far (miRBase, Release 9.1; http://microRNA.sanger.ac.uk/sequences) (Griffiths-Jones et al., 2006). In 2004, Calin et al. (2004) reported that more than half of the then known miRNAs were located in fragile sites on chromosomes. DNA copy number abnormalities in these fragile sites can have a direct impact on the miRNA expression levels (Calin et al., 2004). miRNAs can influence a wide variety of biological processes, including development, proliferation and differentiation (He and Hannon, 2004). Accumulating evidence suggests that miRNAs may act as either tumor suppressors or oncogenes that control growth and apoptosis (Esquela-Kerscher and Slack, 2006). In addition, a number of recent studies have highlighted the potential of miRNA profiles for diagnosis and prognosis of some epithelial tumors (Lu et al., 2005) and hematological malignancies (Calin et al., 2005). In this study we demonstrate that miRNA expression patterns correlate with known major sarcoma subtypes and can serve as a new tool in defining their biologic differences.
Global miRNA expression profilings were performed on a series of 27 sarcomas, 5 normal smooth muscle and 2 normal skeletal muscle tissues using microarrays and individual molecule sequencing. The 27 sarcomas analysed represent seven different histological types (Table 1).
Distinct miRNA expression profiles in sarcomas
As a first step in the analysis, we asked whether the miRNA expression signatures of the sarcomas were molecularly distinct. The 87 miRNAs that met the filtering criteria were subjected to hierarchical clustering among the 27 sarcomas, 5 normal smooth muscles and 2 normal skeletal muscles in an unsupervised manner. The clustering algorithm grouped both miRNAs and samples into clusters based on overall similarity in miRNA expression pattern without prior knowledge of sample identity. Clustering based on the 87 miRNAs revealed substantive distinctions in overrepresented and underrepresented miRNAs among the tumors (Figure 1). As is evident from the dendrogram at the top of the cluster pattern, the sarcomas and normal muscles samples clustered into five main groups, whereby almost all synovial sarcoma (SS), rhabdomyosarcomas (RMS), leiomyosarcomas (LMS), normal smooth muscles (NSM) and GIST were grouped according to their diagnosis. Two exceptions occurred, with case 4728 (embryonal RMS; ERMS) being loosely related to SS and case 2798 (LMS) loosely related to the RMS branch. Notably, the two normal skeletal muscle tissues fell into the muscle-derived tumors cluster, that is, RMS. Likewise, the normal smooth muscle samples closely clustered with the LMS samples (branch C and D; Figure 1).
Of the 28 sarcomas used in this study, 5 have previously been analysed for mRNA expression profiling by gene microarray analysis, and another 7 are part of two ongoing studies. As part of those studies the diagnosis was confirmed by a variety of techniques including immunohistochemistry (IHC) and molecular diagnostic tests (Table 1). Remarkably, in the remaining 16 tumors that were not previously analysed for mRNA expression, the miRNA expression patterns suggested that two cases had been misdiagnosed. Case 1433 was initially diagnosed as ERMS based on its histology and reactivity for desmin, smooth muscle actin and CD99. Only a small sample was available for evaluation and showed significant crush artifact. As a result the typical alveolar growth pattern of alveolar RMS (ARMS) was not appreciated in the initial analysis. Upon profiling of the miRNA expression, this case clustered tightly with the other ARMS included in the study. Subsequent demonstration of the fusion transcript PAX3-FKHR by reverse transcription (RT)–PCR and sequencing in this sample confirms that this case is indeed an ARMS (Supplementary Figure I). The second misdiagnosed case 5918 was initially diagnosed as a GIST based on its origin in the wall of the colon and reactivity for CD34. However this case clustered away from the eight other GISTs and instead clustered loosely with pleomorphic RMS (PRMS) (Figure 1). To reevaluate the diagnosis, we analysed KIT and DOG1 expression by immunostaining, and performed mutation analyses for exon 11 of the KIT gene. Both KIT and DOG1 are highly expressed in the vast majority of GIST (West et al., 2004; Espinosa et al., 2007). Case 5918 did not have immunoreactivity for KIT or DOG1 and no mutations were found in KIT exon 11, the most common mutation in GIST (Fletcher and Rubin, 2007). Histologic review of the tumor suggested a new diagnosis of dedifferentiated liposarcoma. This diagnosis was supported at the molecular level by FISH analysis showing amplification for MDM2 and CDK4, which are abnormalities that are seen frequently in well differentiated and dedifferentiated liposarcoma but rarely in GIST (Binh et al., 2005) (Supplementary Figure I).
Tumor type-specific miRNA expression profiles
Using significance analysis of microarrays (SAM) analysis and permutation tests, we identified the miRNAs that correlated with each tumor class. The results for the top miRNAs of each class with <2% FDR (false discovery rate) are detailed in Table 2a–f. A total of 16 and 12 miRNAs showed significant relative overrepresentation in GIST and LMS, respectively compared to rest of the samples. In GISTs, 10 miRNAs showed relative down regulation. However, no miRNAs were found in LMS that showed relative underrepresentation. LMS has smooth muscle origin; uterine LMS originates from myometrium and nonuterine LMS probably has an origin from smooth muscle of big vessels. SAM analysis identified 10 overrepresented miRNAs and 11 underrepresented miRNA in LMS compared to the normal smooth muscle samples. For SS, 9 miRNAs showed relative overrepresentation and 28 miRNAs showed relative underrepresentation compared to the rest of the samples. Given that ARMS and PRMS are believed to be derived from skeletal muscle progenitor cells, their miRNA expression profiles were compared to that of the normal skeletal muscle in an attempt to identify potential etiologically relevant miRNAs. As only a single ERMS sample was studied in this study, it was excluded from the SAM analysis. Furthermore, only miRNAs with adequate measurements in both skeletal muscle samples were included in this series of SAM analysis. Several miRNAs were relatively underrepresented in both PRMS and ARMS compared to normal skeletal muscle: 3 in PRMS and 11 in ARMS. However, three miRNAs had significant higher expression in ARMS than in the normal skeletal muscle samples.
Identification of known and novel miRNAs by cloning
To validate the miRNA profiles determined by microarray and to search for novel candidate miRNAs or other small RNAs in human sarcomas and normal skeletal muscle tissues, we cloned and sequenced small RNA libraries from 10 sarcomas and 2 normal skeletal muscle tissues. A total of 8134 small RNA clones were sequenced, of which 6350 (78%) could be annotated as known miRNAs (Supplementary Table I). A total of 37 clones (0.5%) could be aligned to the human genome sequences but did not correspond to any annotated RNA species. Eighteen of these small RNA clones were designated as novel candidate miRNAs, based on a predicted hairpin precursor with the following properties (Ambros et al., 2003) (Figure 2): (1) complete containment of the cDNA sequence within one arm of a hairpin, (2) at least 16 nucleotides of the cDNA sequence involved in base-pairing and (3) identification as the lowest free energy structure by mfold. Another five clones fulfilled criteria 1 and 3 but had somewhat fewer duplex base pairs in the hairpin region (Supplementary Table II). The remaining aligned but nonannotated clones either showed poor secondary fold-back hairpin structure predicted by mfold or could not fulfill the criteria listed above (Supplementary Table III).
From 18 relevant clones, 14 novel candidate miRNAs were identified (Figure 2). Of these, candidate-3 appears to be close homolog to the known mouse and rat miR-543. In addition, the sequence of two miRNAs is similar to two previously identified human miRNAs: candidate-1 is similar to miR-28, and candidate-12 is very similar to miR-374. Notably, candidate-12 is located ∼150 bp upstream of miR-421, and candidate-14 is located in the miRNA cluster within the human imprinted 14q32 locus. All candidate miRNAs are conserved in mammals (including chimpanzee, mouse, rat and dog) but not in invertebrates. Interestingly, candidate-13 is mapped to a bidirectional transcript pair, that is, sense strand of intron 22 of coagulation factor VIII (F8; NM_000132) and antisense of coding strand of coagulation factor VIII-associated protein (F8A1; NM_012151).
A comprehensive list of miRNA clones is detailed in Supplementary Table IV. Of 177 known miRNAs expressed in this sample group, some (for example, the let-7 family) are highly represented in all samples analysed. Other miRNAs were found to have substantial expression variation among the tumor types. One striking example, miR-143 was found to be highly abundant in GISTs (172/634 (27%) in case 335 and 156/973 (16%) in case 2000) and LMSs (122/606 (14%) in case 516 and 154/1129 (14%) in case 4612), while only 0–2 clones were found in the other sarcoma types and normal muscle samples that were analysed. Two miRNAs seem to be restricted to specific tumor type: miR-200c was only found in SS and miR-140 and its antisense strand miR-140* were only found in GISTs. In addition, known muscle-specific miRNAs (miR-1, miR-133 and miR-206) were only identified in muscle-derived tumors and normal skeletal muscles. Notably, both miR-1 and miR-133a were more abundant in normal skeletal muscles than the tumors.
Comparison of miRNA expression between microarray and cloning approaches
To determine the consistency of miRNA expression profiles obtained by microarray and cloning methods, we selected all samples that were analysed by both approaches and generated heat maps of the miRNA expression profiles based on the mean ratio or mean cloning frequency of each tumor types. As illustrated in Figure 3, the miRNA expression profiles are strikingly consistent between the two methods.
Differential expression of miR-143 in sarcomas
As miR-143 expression varied significantly among the tumor types by both microarray and cloning, we further analysed its expression by northern analysis. Consistent with the microarray and cloning results, miR-143 was highly expressed in the majority of LMS and all GIST, with little or no miR-143 signal detected in RMS and SS, as well as the case of dedifferentiated liposarcoma (DDLPS) (Figure 4). Notably, expression of miR-143 was barely detected in two LMS. One of these (case 2798) was loosely associated with the skeletal muscle/RMS cluster by clustering (Figure 1) and the other case (2384) was only studied by northern blot and not included in the microarray or cloning analyses.
Using a microarray approach, we characterized miRNA expression profiles in a series of 27 sarcomas from 7 different histological types. We identified four major groups based on the miRNA expression patterns by clustering. Three groups consisted predominantly of the same tumor type, that is, SS, GIST and LMS. It is known that the oncogenesis and cellular origin of these sarcoma types are different. SS is characterized by a t(X;18) translocation that leads to the formation of a fusion oncoprotein SYT–SSX, which is believed to underlie its transformed phenotype (Pretto et al., 2006). Most GISTs are driven by an activating mutation in KIT or PDGFRA (Fletcher and Rubin, 2007), while no unifying molecular event has been uncovered for LMS. Phenotypically, there is evidence to indicate that GIST is derived from interstitial cells of Cajal in the myenteric plexus, while LMS are thought to arise from smooth muscle cells in the uterus or from extra-uterine sites. In contrast, SS belongs to a class of sarcomas for which the cell of origin is unknown; biphasic mesenchymal and epithelial differentiation is common in these tumors (Weiss and Goldbum, 2001). It is therefore plausible that the distinct miRNA expression patterns observed among these sarcoma types are likely reflective of the differences in cell lineage, differentiation and/or oncogenic pathways.
The fourth expression cluster consisted predominantly of normal skeletal muscle tissues and muscle-related sarcomas (ARMS and PRMS) showing histologic, ultrastructural and/or immunophenotypic evidence of skeletal muscle differentiation. Within this cluster, ARMS, PRMS and normal skeletal muscles are separated into their individual subclusters, indicating that each of them possesses a miRNA expression signature that is measurably different from the others. Correspondingly, ARMS and PRMS differ in their oncogenic origin. ARMS is characterized by a PAX3–FKHR or a PAX7–FKHR fusion, while no recurrent molecular event has been found for PRMS or ERMS (Weiss and Goldbum, 2001).
Based on the miRNA expression signatures, one case thought to be ERMS (case 1433) grouped closely with the ARMS, and a previously diagnosed GIST (case 5918) grouped closely with the PRMS, bringing into question the assigned diagnoses. Using a combination of molecular, histopathologic and immunohistochemical analyses, we confirmed that the original diagnoses assigned to these cases were incorrect and that these tumors represented ARMS and DDLPS, respectively. These findings demonstrate that miRNA expression signatures are predominantly unique for each tumor type, suggesting the potential for tumor classification.
Using SAM analysis, we selected miRNAs that are associated with each tumor class. In GIST, miR-221 and miR-222 were expressed at lower level compared to LMS in the series. These miRNAs have been shown to target the 3′-UTR region of KIT in experimental systems (Felli et al., 2005). Given that activating mutation in KIT is present in ∼80% of GIST (Fletcher and Rubin, 2007), lower expression of miR-221 and miR-222 may allow for increased translation of KIT and further enhance its oncogenic influence on the cell.
SAM comparison of LMS and normal smooth muscle samples identified significant overrepresentation of miR-1, miR-133A and miR-133B in LMS. These miRNAs play a major role in myogenesis and myoblast proliferation (Chen et al., 2006). Both LMS and NSM showed underrepresentation of miR-206, an miRNA that is highly expressed in skeletal muscles and implicated in myogenic differentiation (Kim et al., 2006; Politz et al., 2006). Subclassification of LMS has been a challenge. Interestingly, miR-143 was expressed at low levels in 2 of 6 LMS (as seen in case 2798 by miRNA profiling and northern blot analysis and in case 2384 by northern blot analysis alone). Case 2798 also separated from those LMS with high miR-143 expression by clustering, suggesting that LMS is molecularly heterogeneous.
In SS, miR-143 was expressed at very low levels relative to GIST and LMS, as demonstrated by microarray, cloning and northern analyses. Similarly, miR-143 is also reduced in expression in several cancer types, including colorectal neoplasia (Michael et al., 2003) and cervical cancer (Lui et al., 2007). ERK5 (also known as MAPK7) is the only experimentally verified target for miR-143, which is known to promote cell growth and proliferation in response to tyrosine kinase signaling (Esau et al., 2004), however, its role in sarcomagenesis remains unclear. SSX1, a common 3′-fusion partner gene resulting from a t(X;18) in SS (Weiss and Goldbum, 2001), is predicted to be a target for miR-143 by miRBase (http://microrna.sanger.ac.uk/targets/v4/) and Target Scan 3.1 (http://www.targetscan.org/). Since miRNAs target the 3′-UTR region of mRNA transcripts, it is tempting to speculate that underrepresentation of miR-143 in SS tumor cells enables the production of the SYT-SSX1 oncoprotein.
SAM analysis comparing PRMS to normal skeletal muscle revealed that two of the known muscle-specific miRNAs (miR-1 and miR-133) are relatively underexpressed in PRMS. Several myogenic genes are regulated by these two miRNAs (Chen et al., 2006). For ARMS, miR-335 seems to be of particular interest as it is overrepresented miRNA compared to normal skeletal muscle and notably, miR-335 resides in the intron 2 of MEST (also known as PEG1). MEST has been indicated to play a role in muscle differentiation (Yan et al., 2003), and its mRNA expression is high in ARMS (Baer et al., 2004). MEST is a downstream target of PAX3, the gene involved in the PAX3–FKHR fusion that is typical for alveolar rhabdomyosarcomas (Mayanil et al., 2001). It thus appears that the PAX3–FKHR fusion may influence the transcription of miR-335 that has several predicted targets. These predicted targets for miR-335 (by Target Scan 3.1; http://www.targetscan.org) include CHFR, which is lost in many tumors, and HAND1 and SP1, which function in mesoderm or muscle differentiation (Supplementary Table V).
Novel candidate miRNAs
Using a cloning approach, we discovered a significant number of novel miRNAs and other small RNAs, indicating that some of these small RNAs are unique to a specific tumor type or are expressed at levels, that escaped detection in previous cloning experiments. Although one might expect some novel miRNAs to be expressed specifically in one tumor type, it is conceivable that others would be expressed in a range of tumor types. For example, candidate-1, candidate-7, candidate-10 and candidate-12 were also found in cervical cancer samples (Lui et al., 2007); and candidate-5 was also observed in colon cancer cell line (WOL, unpublished data). Analysis for conservation of fold-back structure revealed that all the 14 novel candidate miRNAs were conserved in mammals including chimpanzee, dog, mouse and rat. Together with their structural characteristics they fulfill the criteria for novel miRNA species as described by Kim (Kim, 2005). The final distribution profile of these miRNAs awaits the testing of many more samples. Novel small RNAs that do not fulfill current miRNA criteria were identified and these could represent: (1) miRNAs that fail to meet the arbitrary miRNA criteria, (2) miRNA-like molecules formed by slightly divergent synthetic mechanisms and (3) other small RNAs such as natural siRNAs that might be formed by completely different mechanisms. Alternatively, these could also represent spurious single-stranded RNA transcripts or common degradation products of longer cellular RNAs.
Our dual approach to miRNA profiling has provided a comprehensive analysis of miRNA expression patterns in different histological types of soft tissue sarcomas. The miRNA expression signatures are clearly distinct among the tumor types studied, implicating their role in tumorigenesis in these tumors and their potential as diagnostic markers or even therapeutic targets.
Materials and methods
Tissue samples were as follows: eight GIST, seven SS, six LMS, one DDLPS, six RMS, five NSM and two normal skeletal muscle samples. The RMS included three different histological subtypes, that is, three ARMS, two PRMS and one ERMS. The fresh frozen tissues were collected from surgical specimens at Stanford University Medical Center and Vancouver General Hospital. All cases were centrally reviewed and the diagnoses were further confirmed by ancillary IHC and molecular studies (Table 1). The study was approved by the local ethical committees of Stanford University and the British Columbia Cancer Agency.
miRNA array printing
miRNA microarrays used in the study were printed at the Stanford functional genomics facility (www.microarray.org). The arrays contained a total of 668 probes spotted in duplicate. The 668 probes represent 328 known human miRNAs, 113 mouse miRNAs, 45 rat miRNAs, 154 predicted human miRNAs and 28 control probes (Ambion, Austin, TX, USA). The complete list of probes is given in Supplementary Table V1.
RNA isolation and miRNA array experiments
Total RNAs were extracted from frozen tumor samples using mirVana miRNA isolation kit (Ambion). Reference RNA (XpressRef Universal Total RNA) was obtained from SuperArray (Frederick, MD, USA). miRNA was further enriched from 25 μg of total RNA using a microcon YM-100 column (Millipore, Billerica, MA, USA), and indirectly labeled with Cy3 or Cy5 amine reactive dyes (Amersham Biosciences, Buckinghamshire, UK) using mirVana miRNA labeling kit (Ambion). Hybridization was at 42 °C for 12–16 h. Arrays were washed and immediately scanned using a GenePix 4000B array scanner (Axon Instruments, Foster City, CA, USA).
Microarray data analysis
Fluorescence ratios (sample/reference) were calculated using GenePix software. miRNA arrays were normalized and data was uploaded to Stanford Microarray Database (http://genome-www5.stanford.edu/). To limit the measurement errors, only miRNA spots with a ratio of signal over background of at least 2.5 in either Cy3 or Cy5 channels were included. Further, miRNA spots were filtered based on those where expression levels differed by at least fourfold in at least three arrays. Finally miRNA spots with >80% good data were selected. A total of 87 miRNAs passed the filtering criteria and were used for further analysis. Unsupervised hierarchical clustering analysis and SAM were then performed as described previously (Nielsen et al., 2002).
Small RNA isolation and cloning
Small RNA extracted using mirVana miRNA isolation kit (Ambion) was used as starting material for cloning procedure of Lau et al. (2001) with slight modifications. Purified small RNAs were ligated with pre-adenylated 3′-adaptor oligonucleotide, gel purified and subjected to a second ligase reaction with a 5′-adaptor oligonucleotide. The gel-purified, doubly ligated RNA was reverse transcribed using Superscript II (Invitrogen, Carlsbad, CA, USA) and RT primer, followed by PCR amplification using the RT primer and a forward primer. A second PCR was performed using the RT primer and a second forward primer (Supplementary Table VII). The PCR product was purified by phenol/chloroform extractions and then digested with Ban I (NEB, Beverly, MA, USA) for concatemerization using T4 DNA ligase (NEB). Concatamers ranging from 600 to 1000 bp were isolated from a low-melting-point agarose gel, processed with Taq polymerase, and cloned into the pCR4-TOPO vector using the TOPO TA cloning kit (Invitrogen). Colony PCR was performed using the M13 forward and reverse primers, and the PCR products were prepared for sequencing using shrimp alkaline phosphatase and exonuclease I (USB Corporation, Cleveland, OH, USA).
Small RNAs obtained by cloning were compared with functionally annotated sequences using BLAST (blastn, http://www.ncbi.nlm.nih.gov/blast/), BLAT (http://genome.ucsc.edu), miRBase Release 9.1 (http://microrna.sanger.ac.uk/sequences/search.shtml) and simple text searches. For each cloned small RNA, the best alignments to a functionally annotated sequence (not more than one error) were used to assign a functional category to the small RNA. Putative novel small RNAs were analysed using mfold version 3.2 (Zuker, 2003) to identify potential precursor structures.
Total RNA (20 μg) from snap-frozen tissues was fractionated on a denaturing 15% polyacrylamide gel. The gels were then transferred to Hybond-N+ membranes (Amersham Biosciences), fixed by ultraviolet cross-linking at 1200 μJ and baked (80 °C) for 1 h. Membranes were then hybridized overnight at 55 °C in PerfectHyb Plus hybridization buffer (Sigma, St Louis, MO, USA), together with a locked nucleic acid (LNA)-modified oligonucleotide probe complementary to the mature miR-143 (Supplementary Table VII) that was labeled with terminal transferase (NEB) and biotin-16-dUTP (Roche Diagnostics, Indianapolis, IL, USA). Subsequently, the blots were washed at 55 °C for 15 min each in 2 × standard saline citrate/0.1% SDS and 0.2 × SSC/0.1% SDS. After washes, the blots were incubated in blocking solution (1 × Phosphate buffered saline, pH 7.4/0.05% Tween 20/0.1% SDS/0.5% blocking reagent, Roche Diagnostics) for 1 h and then in streptavidin–alkaline phosphatase conjugate (USB Corporation) for 1 h, followed by three washes each in buffer A (1 × PBS, pH 7.4/0.05% Tween 20/0.1% SDS) and buffer B (0.1 M Tris–HCl, pH 9.5, 0.1 M NaCl) at room temperature. The blots were then incubated with chemiluminescent substrate CDP-Star (GE Healthcare, Piscataway, NJ, USA) and exposed to Kodak BioMax XAR film (New Haven, CT, USA).
Sections were cut at 4 μm, deparaffinized in xylene and hydrated in a graded ethanol series. The primary antibodies used were DOG1 (mouse monoclonal, 1/50; clone DOG1.1 unpublished data), CD117 (rabbit polyclonal, 1/200; Dako, Carpinteria, CA, USA), and CD34 (mouse monoclonal, 1/80; clone 581/CD34, BD Biosciences, San Jose, CA, USA). The antigen retrieval and IHC were performed as previously described (West et al., 2004).
RT–PCR detection of fusion transcript
Total RNA (1 μg) was reverse transcribed with thermoscript II reverse transcriptase and random hexamers. The resulting cDNA was subjected to PCR amplification using specific primers for SYT–SSX and PAX–FKHR, as detailed in Supplementary Table VII.
Fluorescence in situ hybridization
Fluorescence in situ hybridization for copy number detection of MDM2/CDK4 was performed on a paraffin section of case 5918 using standard techniques (Shimada et al., 2006) to confirm the diagnosis of dedifferentiated liposarcoma.
Mutation analyses of KIT and PDGFRA genes
GIST tumors were analysed for mutations in exons 9, 11, 13 and 17 of the KIT gene and exons 12 and 18 of PDGFRA using a combination of denaturing high-performance liquid chromatography and direct sequencing, as previously described (Heinrich et al., 2003).
Ambros V, Bartel B, Bartel DP, Burge CB, Carrington JC, Chen X et al. (2003). A uniform system for microRNA annotation. RNA 9: 277–279.
Baer C, Nees M, Breit S, Selle B, Kulozik AE, Schaefer KL et al. (2004). Profiling and functional annotation of mRNA gene expression in pediatric rhabdomyosarcoma and Ewing's sarcoma. Int J Cancer 110: 687–694.
Binh MB, Sastre-Garau X, Guillou L, de Pinieux G, Terrier P, Lagace R et al. (2005). MDM2 and CDK4 immunostainings are useful adjuncts in diagnosing well-differentiated and dedifferentiated liposarcoma subtypes: a comparative analysis of 559 soft tissue neoplasms with genetic data. Am J Surg Pathol 29: 1340–1347.
Calin GA, Ferracin M, Cimmino A, Di Leva G, Shimizu M, Wojcik SE et al. (2005). A MicroRNA signature associated with prognosis and progression in chronic lymphocytic leukemia. N Engl J Med 353: 1793–1801.
Calin GA, Sevignani C, Dumitru CD, Hyslop T, Noch E, Yendamuri S et al. (2004). Human microRNA genes are frequently located at fragile sites and genomic regions involved in cancers. Proc Natl Acad Sci USA 101: 2999–3004.
Chen JF, Mandel EM, Thomson JM, Wu Q, Callis TE, Hammond SM et al. (2006). The role of microRNA-1 and microRNA-133 in skeletal muscle proliferation and differentiation. Nat Genet 38: 228–233.
Doench JG, Sharp PA . (2004). Specificity of microRNA target selection in translational repression. Genes Dev 18: 504–511.
Esau C, Kang X, Peralta E, Hanson E, Marcusson EG, Ravichandran LV et al. (2004). MicroRNA-143 regulates adipocyte differentiation. J Biol Chem 279: 52361–52365.
Espinosa I, Lee CH, Kim M, Rouse B, Subramanian S, Montgomery K et al. (2007). A novel monoclonal antibody against DOG1 is a sensitive and specific marker for gastrointestinal stromal tumors. Am J Surg Path (in press).
Esquela-Kerscher A, Slack FJ . (2006). Oncomirs—microRNAs with a role in cancer. Nat Rev Cancer 6: 259–269.
Felli N, Fontana L, Pelosi E, Botta R, Bonci D, Facchiano F et al. (2005). MicroRNAs 221 and 222 inhibit normal erythropoiesis and erythroleukemic cell growth via kit receptor down-modulation. Proc Natl Acad Sci USA 102: 18081–18086.
Fletcher JA, Rubin BP . (2007). KIT Mutations in GIST. Curr Opin Genet Dev 17: 3–7.
Griffiths-Jones S, Grocock RJ, van Dongen S, Bateman A, Enright AJ . (2006). miRBase: microRNA sequences, targets and gene nomenclature. Nucleic Acids Res 34: D140–D144.
He L, Hannon GJ . (2004). MicroRNAs: small RNAs with a big role in gene regulation. Nat Rev Genet 5: 522–531.
Heinrich MC, Corless CL, Duensing A, McGreevey L, Chen CJ, Joseph N et al. (2003). PDGFRA activating mutations in gastrointestinal stromal tumors. Science 299: 708–710.
Kim HK, Lee YS, Sivaprasad U, Malhotra A, Dutta A . (2006). Muscle-specific microRNA miR-206 promotes muscle differentiation. J Cell Biol 174: 677–687.
Kim VN . (2005). MicroRNA biogenesis: coordinated cropping and dicing. Nat Rev Mol Cell Biol 6: 376–385.
Lau NC, Lim LP, Weinstein EG, Bartel DP . (2001). An abundant class of tiny RNAs with probable regulatory roles in Caenorhabditis elegans. Science 294: 858–862.
Lu J, Getz G, Miska EA, Alvarez-Saavedra E, Lamb J, Peck D et al. (2005). MicroRNA expression profiles classify human cancers. Nature 435: 834–838.
Lui WO, Pourmand N, Patterson BK, Fire A . (2007). Patterns of known and novel small RNAs in human cervical cancer. Cancer Res 67: 6031–6043.
Mayanil CS, George D, Freilich L, Miljan EJ, Mania-Farnell B, McLone DG et al. (2001). Microarray analysis detects novel Pax3 downstream target genes. J Biol Chem 276: 49299–49309.
Michael MZ, O'Connor SM, van Holst Pellekaan NG, Young GP, James RJ . (2003). Reduced accumulation of specific microRNAs in colorectal neoplasia. Mol Cancer Res 1: 882–891.
Nielsen TO, West RB, Linn SC, Alter O, Knowling MA, O'Connell JX et al. (2002). Molecular characterisation of soft tissue tumours: a gene expression study. Lancet 359: 1301–1307.
Politz JC, Zhang F, Pederson T . (2006). MicroRNA-206 colocalizes with ribosome-rich regions in both the nucleolus and cytoplasm of rat myogenic cells. Proc Natl Acad Sci USA 103: 18957–18962.
Pretto D, Barco R, Rivera J, Neel N, Gustavson MD, Eid JE . (2006). The synovial sarcoma translocation protein SYT-SSX2 recruits beta-catenin to the nucleus and associates with it in an active complex. Oncogene 25: 3661–3669.
Shimada S, Ishizawa T, Ishizawa K, Matsumura T, Hasegawa T, Hirose T . (2006). The value of MDM2 and CDK4 amplification levels using real-time polymerase chain reaction for the differential diagnosis of liposarcomas and their histologic mimickers. Hum Pathol 37: 1123–1129.
Subramanian S, West RB, Corless CL, Ou W, Rubin BP, Chu KM et al. (2004). Gastrointestinal stromal tumors (GISTs) with KIT and PDGFRA mutations have distinct gene expression profiles. Oncogene 23: 7780–7790.
Weiss SW, Goldbum JR . (2001). Soft tissue tumors. Mosby: St Louis.
West RB, Corless CL, Chen X, Rubin BP, Subramanian S, Montgomery K et al. (2004). The novel marker, DOG1, is expressed ubiquitously in gastrointestinal stromal tumors irrespective of KIT or PDGFRA mutation status. Am J Pathol 165: 107–113.
Yan Z, Choi S, Liu X, Zhang M, Schageman JJ, Lee SY et al. (2003). Highly coordinated gene regulation in mouse skeletal muscle regeneration. J Biol Chem 278: 8826–8836.
Zuker M . (2003). Mfold web server for nucleic acid folding and hybridization prediction. Nucleic Acids Res 31: 3406–3415.
This work was supported by NIH grant CA112270, a grant from LifeRaft and the Department of Pathology, Stanford University. SS was supported by a postdoctoral fellowship from the US National Institutes of Health.
About this article
Cite this article
Subramanian, S., Lui, W., Lee, C. et al. MicroRNA expression signature of human sarcomas. Oncogene 27, 2015–2026 (2008) doi:10.1038/sj.onc.1210836
Inhibition of microRNA-183 expression resists human umbilical vascular endothelial cells injury by upregulating expression of IRS1
Drug Delivery (2019)
Computers in Biology and Medicine (2019)
Scientific Reports (2019)
Journal of Cellular Physiology (2019)
Determination of the Binding Constant between Oligonucleotide-Coupled Magnetic Microspheres and Target DNA
ACS Omega (2019)