Journal of Heredity Advance Access originally published online on October 26, 2005
Journal of Heredity 2005 96(7):766-773; doi:10.1093/jhered/esi122
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Chromosome Rearrangements in Canine Fibrosarcomas
From the Centre for Veterinary Science, Department of Veterinary Medicine, University of Cambridge, Madingley Road, Cambridge CB3 0ES, UK
Address correspondence to David R. Sargan at the address above, or e-mail: drs20{at}cam.ac.uk.
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
We have previously reported the use of six- and seven-color paint sets in the analysis of canine soft tissue sarcomas. Here we combine this technique with flow sorting of translocation chromosomes, reverse painting, and polymerase chain reaction (PCR) analysis of the gene content of the reverse paint in order to provide a more detailed analysis of cytogenetic abnormalities in canine tumors. We examine two fibrosarcomas, both from female Labrador retrievers, and show abnormalities in chromosomes 11 and 30 in both cases. Evidence of involvement of TGFBR1 is presented for one tumor.
Cancers are one of the major causes of morbidity and mortality in dogs, and canine cancers have also been used extensively as models for human therapy (reviews Vail and McEwan 2000; Withrow et al. 1991). In general, the same tumor types are recognized in man and dogs. For instance, classification of soft tissue sarcomas in dogs largely follows human lines (Hendrick et al. 1998), although soft tissue sarcomas have a much higher incidence in dogs than in humans, with some breeds having an excess of this tumor type (Dobson et al. 2002; Preister and McKay 1980). An understanding of the cytogenetic changes occurring in canine tumors is essential for understanding their causes, aids in classification, validates their potential as models for human medicine, and may have longer term veterinary implications for stratification and prognostication within tumor types.
Cytogenetic studies of canine tumors have, until quite recently, been impeded by the very difficult karyotype of the dog (2n = 78), for which no internationally agreed standard karyotype was achieved using chromosome staining methods alone. It has not been possible to identify the smaller chromosomes (last 17 autosome pairs) with consistency using conventional staining methods, and few if any cytogeneticists are able to interpret chromosomes containing translocations with accuracy using these techniques. The development of chromosome paints for the dog, combined with enhanced 4',6-diamidino-2-phenylindole (DAPI) staining has allowed the consistent identification of the smaller chromosomes (Breen et al. 1999a,b; Langford et al. 1996; Yang et al. 1999). The chromosome paint reagents, as well as sets of chromosome-specific bacterial artificial chromosomes (BACS) (Breen et al. 2001, 2004) have made it possible to define nonstandard canine karyotypes more reliably and far more completely. Comparative genomic hybridization (CGH) has been combined with the use of chromosome-specific reagents (Thomas et al. 2003), but balanced rearrangements remain inaccessible to this type of analysis.
To facilitate cytogenetic analysis of canine tumors, we have recently developed seven-color chromosome painting in the dog, using paint sets including chromosomes derived from other canids to allow unambiguous identification of those pairs of chromosomes that do not separate in the flow karyotype of Canis familiaris (Milne et al. 2004). This provides a robust method to perform analysis of canine tumor karyotypes that can identify all canine chromosomes, including balanced and unbalanced translocation chromosomes, using a minimum number of metaphases in hybridizations. Here we report further cytogenetic and molecular characterization of canine tumor chromosomes in two fibrosarcomas. For one of these, we used reverse painting with paint preparations incorporating the translocation chromosome and polymerase chain reaction (PCR) analysis of these chromosomes in a detailed characterization of translocation breakpoints. We summarize the results from these tumors as an example of these techniques.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Tumors, Cell Culture, and Metaphase Preparation
Tumors were spontaneously occurring soft tissue sarcomas from the pet population, obtained for analysis after surgical excision as part of normal treatment. Two poorly differentiated fibrosarcomas analyzed here were both from adult female Labrador retrievers. Tumor ME came from the hard palate of a 2-year-old dog, while tumor LE came from the dorsal aspect of the right hock of a dog of 6 years 6 months of age. Both tumors consisted of infiltrative cellular masses, in which large spindle cells were arranged in bundles, whorls, and streams. Nuclei are large, ovoid, and hyperchromatic, with one to three large dark-staining nucleoli. There were one to three mitoses per high-power field.
Excised tumor samples were trimmed and cultured as previously (Milne et al. 2004). Cells were usually first harvested at or before 10 days in culture to keep secondary cytogenetic rearrangements in culture to a minimum. Metaphase preparations for fluorescence in situ hybridization (FISH) were performed exactly as before (Milne et al. 2004).
Chromosome Sorting
Chromosome preparations were derived from five sources: an established canine kidney cell line (MDCK), short-term canine blood cultures, an established Japanese raccoon dog (Nyctereutes procyonoides viverrinus) cell line, a red fox (Vulpes vulpes) primary fibroblast tissue culture (Graphodatsky et al. 2000; Yang et al. 1999), and primary tumor cells from tumor LE, prepared for chromosome sorting as in the other cell lines. Chromosome sorting used a fluorescence-activated cell sorter (FACS) (Star chromosome sorter; Becton Dickinson, Franklin Lakes, NJ). Approximately 300 copies of the desired chromosomes were sorted. Chromosome-specific paint probes were generated by degenerate oligonucleotide primed PCR (DOP-PCR) of flow-sorted chromosomes using a variation of the method developed by Telenius et al. (1992). Primary DOP amplifications usually used primer 6MW (Telenius et al. 1992; Yang et al. 1999) and were followed by secondary DOP amplification incorporating FITC-dUTP, Cy3-dUTP (direct labeling), or biotin-dUTP.
Seven-Color FISH and Two- and Three-Color FISH
Seven color FISH was performed as previously, using paints from the red fox (V. vulpes; VVU) and the Japanese raccoon dog (N. procyonoides viverrinus) to resolve those chromosome pairs that are not individually separated in the canine flow karyotype (Milne et al. 2004). Six probe mixes were prepared such that each labeled either six or seven canine chromosomes in different colors, and all canine chromosomes (except Y) could be individually identified by a single probe mix. Probe hybridization and visualization were also performed as before. Ten to 15 metaphases were captured for each probe mix. Two- and three-color FISH used conventional canine paints as previously (Yang 1999) to verify the conclusions drawn using the seven-color paint sets.
PCR Genotyping: STS Analysis of Translocation Chromosomes and Microsatellite Analysis of Tumor DNA
Polymerase chain reaction genotyping was carried out on genomic DNA and DOP-amplified products of flow-sorted chromosomes using primers specific for CFA11 and CFA27 (Table 1) and the method of Sargan et al. 2000. Initially most unpublished primer sequences were derived from regions of sequence identity in human and murine genes. Canine genomic sequences were used once these became available on public databases. PCR by standard methods used 50 ng genomic or DOP-amplified sorted chromosome DNA as template. (DOP-amplified templates contain about 85% unique sequence, hence some primer pairs fail to amplify on chromosomal templates.) Exons of TGFBR1 were amplified by PCR from tumor and blood DNA preparations and sequenced using a Beckman CEQ 8000 genetic analysis system (Beckman Coulter, Fullerton, CA).
|
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Karyotype Analysis Shows Abnormalities in CFA11 and CFA30 for Both Tumors
Forward chromosome painting was used to characterize two poorly differentiated fibrosarcomas from adult female Labrador retrievers. Primary cell cultures from each tumor were maintained for the minimum time to obtain sufficient metaphases for cytogenetic study (usually a total of 10 days of tissue culture). Metaphases were analyzed by chromosome painting using a seven-color paint system (Milne et al. 2004). Tumor ME was found to have a deletion of CFA11q and trisomy of CFA30. The derived CFA11 is 40% of the size of normal CFA11 and contains no part of any other chromosome (Figure 1A–C). DAPI banded images are not sufficiently detailed to determine if the truncation resulted from interstitial deletion or breakage and loss of the telomeric region. Trisomy CFA30 was seen in 14 of 15 metaphases. All three copies of CFA30 appeared identical in size and DAPI banding pattern. ME tumor karyotype was established as 2n = 79; der11 (del 11q); +30.
|
Tumor LE contained four translocation chromosomes involving chromosomes CFA4, 11, 27, and 30. These were further examined in two- and three-color painting experiments with appropriate normal chromosome paints (Figure 1D–G). LE tumor karyotype was derived as 2n = 78; t(4;11;30), t(11;27); t(27;11); t(30;4); del 11q.
Isolation of Abnormal Chromosomes and Reverse Painting: Tumor LE
Both tumors were flow karyotyped in order to sort abnormal chromosomes. For tumor ME, no aberrant peak was seen in the size range expected for der(11), although one may be obscured by one of the peaks representing the normal chromosomes (data not shown). This has not been pursued further. Tumor LE generated a flow karyotype containing a background suggestive of large numbers of dead cells and fragmented chromosomes (Figure 2A). A cluster of abnormal peaks was observed in the position usually occupied by chromosomes 8 and 11, suggesting that translocation chromosomes were present. Narrow gates were set to collect four subsets of chromosomes within the region. After DOP amplification, only one of these (shown as peak "B" in Figure 2A) gave a paint that hybridized strongly and specifically onto aberrant chromosomes in the tumor (Figure 2B–E) as well as to chromosomes in normal metaphases (data not shown).
|
Painting onto a tumor metaphase showed hybridization to several chromosomes including the whole of at least three translocation chromosomes t(4;11;30), t(30;4), and t(11;27) (Figure 2B–E). There was a small band of hybridization to t(27;11), suggesting that part of the sequences from chromosome 11 present in the other two translocation chromosomes were also present in this chromosome. The DAPI banded karyotype, together with painting onto the more easily recognized fox chromosomes, revealed the presence of the complete chromosomes CFA4, 8, 12, and 30 as well as parts of CFA27 and 11 (Figure 2C). It is likely that CFA4 and 30 sequences come from the translocation chromosomes, as intact CFA4 and 30 are found well away from gate B in the flow karyotype and should not contribute to the signal seen in Figure 2(B,E,F). This suggests that for these chromosomes the karyotype is balanced. Only part of der11 and of the CFA11 segment in t(27;11) hybridized to B (Figure 2C,D), allowing us to determine some of the breakpoints within chromosomes 11 and 27 by examining the gene content of paint B.
Gene Content of the Reverse Paint: Tumor LE
To further characterize breakpoints in tumor LE translocation chromosomes in paint B, individual markers at intervals along the translocation chromosomes were amplified by PCR. The results are summarized in Table 1. Controls show that only 85–90% of STS in this size range will amplify from DOP-amplified paints (and direct amplification of the sorted chromosome targets is less sensitive). The sorted translocation chromosomes support amplification of markers proximal to the transforming growth factor (TGF)-ß receptor one gene (TGFBR1) together with exons 2, 7, and 8 of this gene, but not the other tested exons (3–6 and 9) or markers distal to this gene. The exons of canine TGFBR1 gene as represented in contig 55694 (GenBank accession AAEX01055695) are colinear with TGFBR1 in the human genome. The results suggest a complex rearrangement in the translocation chromosomes in which TGFBR1 is broken and rejoined between exons 8 and 9, but sequences between exons 2 and 7 are also missing from peak B, and have either been lost or moved to another chromosome during the translocation.
The reverse paint includes two fragments of CFA11 in translocation chromosomes, and potentially three breakpoints. Our data do not address which other chromosome (4, 27, or 30) abuts TGFBR1, or which of the two CFA11-containing translocation chromosomes in peak B is under analysis here.
The sorted chromosomes include CFA27 sequences from a single translocation chromosome, t(11;27). Markers proximal to ACVR1B, but also those between WNT1 and EPLIN are missing from the translocation chromosome. Both radiation hybrid (RH) mapping and sequence assembly of CFA27 suggests colinearity with HSA12 (Guyon et al. 2003; Hitte C, personal communication). Markers that are centromeric on CFA27 are telomeric on HSA12p, while those that are telomeric on CFA27 are centromeric on 12p. Hence it is not yet clear whether there is a local reversal of colinearity of CFA27 with respect to human chromosome 12 in this region of the map, consistent with a single breakpoint in LE's CFA27, or whether a complex rearrangement has also taken place on this chromosome.
Sequences of TGFBR1
To decide whether the potential loss of TGFBR1 activity in these tumors affected both copies of the gene, coding exons 2–9 of TGFBR1 were amplified and sequenced from normal canine (greyhound) DNA, DNA from cell lines derived from each tumor (ME and LE), and DNA from a blood sample (LE). At least one copy of all available exons of TGFBR1 was present in DNA isolated from each tumor, and no sequence abnormalities or heterozygous bases were found compared with the published genomic sequences or our own sequencing of each of the normal DNA samples. Margins were removed from tumors prior to culture, but they undoubtedly contain normal stromal cells, so we cannot rule out the possibility that it is this DNA that we have sequenced. However, metaphases in these cell lines showed consistent karyotypic changes, so it is likely that the sequencing result does reflect the tumor genotype.
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Two canine fibrosarcomas from animals of the same breed and sex share aspects of tumor karyotype in that both have abnormalities in CFA30 and CFA11. In one case only two and in the other case five rearrangements can be found by complete chromosome painting. This simplicity of karyotype and similarity of phenotype suggests that the shared features in the karyotype may relate to the etiology of the tumors.
Canine chromosome 30 corresponds approximately to human HSA15q14-15q24. This region is rich in oncogenes and has been identified as a "region of increased tumor expression" by transcriptome analysis (Zhou et al. 2003). Gains of chromosomal material as well as translocations are frequently seen in soft tissue synovial sarcoma, and rearrangements are also seen in fibrosarcoma (Mitelman database). It is possible that the translocations seen in tumor LE led to positional up-regulation of genes on CFA30.
The involvement of CFA27 in tumor etiology for LE is also possible: the breakpoint studied here is close to WNT10b, WNT1, and NR4A1. However, the losses in t(11;27) studied here may well be balanced through the der 27 (t(27;11)) chromosome, which could not be isolated from the flow karyotype, and has not been further characterized. The best characterized changes in tumor LE are in canine chromosome 11, and most particularly the rearrangement of at least one copy of TGFBR1.
TGFBR1 (activin A receptor type II-like kinase, 53 kDa) is a serine-threonine kinase that interacts with TGFBR2 to mediate intracellular signaling in response to TGF-ß (Ebner et al. 1993). TGFBR activity controls the mesoderm-inducing transcription factors SMAD2, SMAD3, and SMAD4 through phosphorylation (Inman et al. 2002). In a variety of situations, TGF-ß signaling through TGFBR1 exerts control on cell proliferation and causes differentiation. Loss of TGFBR1 expression or TGF-ß responsiveness has been noted in tumors of many types, including carcinomas and sarcomas. TGF-ß signaling activates at least one of the CDKN2 family of inhibitors of the cyclin D-dependent kinases Cdk4 and Cdk6, which phosphorylate RB1 and thereby inactivate RB1's growth suppressive functions (Seoane et al. 2001). In fibroblasts, suppression of CDKN2 expression is a major method of escape from senescence associated with telomere shortening (Drayton et al. 2004).
The absence of successive markers suggests that complex rearrangements have occurred at both the translocation breakpoints investigated here, in which sequences amounting to a few thousands (for CFA11) or hundreds of thousands of base pairs (CFA27) have gone missing from the translocation chromosome upstream of the translocation breakpoints. Complex translocation breakpoints are now well recognized: intrachromosomal inversion may occur prior to (or as an early step in) a translocation at or near the inversion site. Our results note only the presence or absence of a gene in the FACS peak rather than gene order. Hence they are compatible with an inversion model in which sequences missing at the breakpoint on one translocation chromosome will be found on the reciprocal translocation, but could also fit a gene deletion model. For the CFA11 breakpoint examined here, all TGFBR1 exons are present in at least one copy in tumor DNA. We are now performing a heterozygosity analysis using polymorphic microsatellites that will allow us to discriminate copy number in the tumor, and thus differentiate between these alternatives, and also map any other deleted sequences.
Soft tissue sarcomas with translocation breakpoints at human 9q22 have been recorded several times. Many of the 9q22 translocation chromosomes are thought to represent NR4A3::EWS1 fusions connected with (mainly pediatric) chondrosarcoma. Two cases in the Mitelman database have translocations at t(9;12) (q22;q13) (equivalent to the t(11;27) recorded here). One of these is a carcinoma while the other is a soft tissue synovial sarcoma in the trunk wall of an adult female, unusual in that it does not have the translocation chromosome t(X;18) (p11;q11) juxtaposing SYT::SSX (present in 95% of human synovial sarcomas; see Panagopoulos et al. 2001). The latter tumor may correspond more closely to the tumors reported here.
The tumors presented here are part of a series of soft tissue sarcomas that are the first solid tumors in the dog in which unambiguous and complete assignment of all chromosomes, including translocation chromosomes, by chromosome painting has been reported (Milne et al. 2004). Here, a complete cytogenetic description has directly informed partial characterization at the molecular level. In order to elucidate more fully the relationship between karyotypic abnormalities and tumor behavior and prognosis in the dog, and to provide a greater understanding of the comparative value of canine sarcomas to human medicine, a far larger number of tumors will have to be analyzed at this level. The rapid advances in technology in this field now make this a real possibility.
|
Acknowledgments |
|---|
We thank Drs. Christophe Hitte, Francois Galibert, Ewen Kirkness, and Elaine Ostrander for providing unpublished canine RH mapping and sequence analysis information. These studies were supported by the Pet Plan Charitable Trust and Cancer Research UK. BM received a Domestic Research Studentship from Cambridge University. This paper was delivered at the 2nd International Conference on the "Advances in Canine and Feline Genomics: Comparative Genome Anatomy and Genetic Disease," Universiteit Utrecht, Utrecht, The Netherlands, October 14–16, 2004.
|
Footnotes |
|---|
Corresponding Editor: Elaine Ostrander
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
-
Breen M, Bullerdiek J, and Langford CF, 1999a. The DAPI banded karyotype of the domestic dog (Canis familiaris) generated using chromosome-specific paint probes. Chromosome Res 7:401–406.[CrossRef][ISI][Medline]
Breen M, Hitte C, Lorentzen TD, Thomas R, Cadieu E, Sabacan L, Scott A, Evanno G, Parker HG, Kirkness EF, Hudson R, Guyon R, Mahairas GG, Gelfenbeyn B, Fraser CM, Andre C, Galibert F, and Ostrander EA, 2004. An integrated 4249 marker FISH/RH map of the canine genome. BMC Genomics 5:65.[CrossRef][Medline]
Breen M, Jouquand S, Renier C, Mellersh CS, Hitte CH, Holmes NG, Cheron A, Suter N, Vignaux F, Bristow AE, Priat C, McCann E, Andre C, Boundy S, Gitsham P, Thomas R, Bridge WL, Spriggs HF, Ryder EJ, Curson A, Sampson J, Ostrander EA, Binns MM, and Galibert F, 2001. Chromosome-specific single-locus FISH probes allow anchorage of an 1800-marker integrated radiation-hybrid/linkage map of the domestic dog genome to all chromosomes. Genome Res 11:1784–1795.
Breen M, Thomas R, Binns MM, Carter NP, and Langford CF, 1999b. Reciprocal chromosome painting reveals detailed regions of conserved synteny between the karyotypes of the domestic dog (Canis familiaris) and human. Genomics 61:145–155.[CrossRef][ISI][Medline]
Dobson JM, Samuel S, Milstein H, Rogers K, and Wood JL, 2002. Canine neoplasia in the UK: estimates of incidence rates from a population of insured dogs. J Small Anim Pract 43:240–246.[CrossRef][ISI][Medline]
Drayton S, Brookes S, Rowe J, and Peters G, 2004. The significance of p16INK4a in cell defenses against transformation. Cell Cycle 3:611–615.[ISI][Medline]
Ebner R, Chen R-H, Shum L, Lawler S, Zioncheck TF, Lee A, Lopez AR, and Derynck R, 1993. Cloning of a type I TGF-beta receptor and its effect on TGF-beta binding to the type II receptor. Science 260:1344–1348.
Graphodatsky AS, Yang F, O'Brien PC, Serdukova N, Milne BS, Trifonov V, and Ferguson-Smith MA, 2000. A comparative chromosome map of the Arctic fox, red fox and dog defined by chromosome painting and high resolution G-banding. Chromosome Res 8:253–263.[CrossRef][ISI][Medline]
Guyon R, Lorentzen TD, Hitte C, Kim L, Cadieu E, Parker HG, Quignon P, Lowe JK, Renier C, Gelfenbeyn B, Vignaux F, Defrance HB, Gloux S, Mahairas GG, Andre C, Galibert F, and Ostrander EA, 2003. A 1-Mb resolution radiation hybrid map of the canine genome. Proc Natl Acad Sci USA 100:5296–5301.
Hendrick MJ, Mahaffey EA, Moore FM, Vos JH, and Walder EJ, 1998. Mesenchymal tumors of skin and soft tissues of domestic animals. In: WHO histological classification of tumors of domestic animals. Washington, DC: Armed Forces Institute of Pathology.
Inman GJ, Nicolas FJ, and Hill CS, 2002. Nucleocytoplasmic shuttling of Smads 2, 3, and 4 permits sensing of TGF-beta receptor activity. Mol Cell 10:283–294.[CrossRef][ISI][Medline]
Langford CF, Fischer PE, Binns MM, Holmes NG, and Carter NP, 1996. Chromosome-specific paints from a high-resolution flow karyotype of the dog. Chromosome Res 4:115–123.[CrossRef][ISI][Medline]
Milne BS, Hoather T, O'Brien PCM, Yang F, Ferguson-Smith MA, Dobson J, and Sargan D, 2004. Karyotype of canine soft tissue sarcomas: a multi-colour, multi-species approach to canine chromosome painting. Chromosome Res 12:825–835.[CrossRef][ISI][Medline]
Panagopoulos I, Mertens F, Isaksson M, Limon J, Gustafson P, Skytting B, Akerman M, Sciot R, Dal Cin P, Samson I, Iliszko M, Ryoe J, Debiec-Rychter M, Szadowska A, Brosjo O, Larsson O, Rydholm A, and Mandahl N, 2001. Clinical impact of molecular and cytogenetic findings in synovial sarcoma. Genes Chromosomes Cancer 31:362–372.[CrossRef][ISI][Medline]
Preister WA and McKay FW, 1980. The occurrence of tumors in domestic animals. Natl Cancer Inst Monogr 54:178.
Sargan DR, Yang F, Squire M, Milne BS, O'Brien PC, and Ferguson-Smith MA, 2000. Use of flow-sorted canine chromosomes in the assignment of canine linkage, radiation hybrid, and syntenic groups to chromosomes: refinement and verification of the comparative chromosome map for dog and human. Genomics 69:182–195.[CrossRef][ISI][Medline]
Seoane J, Pouponnot C, Staller P, Schader M, Eilers M, and Massague J, 2001. TGFbeta influences Myc, Miz-1 and Smad to control the CDK inhibitor p15INK4b. Nat Cell Biol 3:400–408.[CrossRef][ISI][Medline]
Telenius H, Pelmear AH, Tunnacliffe A, Carter NP, Behmel A, Ferguson-Smith MA, Nordenskjold M, Pfragner R, and Ponder BA, 1992. Cytogenetic analysis by chromosome painting using DOP-PCR amplified flow-sorted chromosomes. Genes Chromosomes Cancer 4:257–263.[ISI][Medline]
Thomas R, Fiegler H, Ostrander EA, Galibert F, Carter NP, and Breen M, 2003. A canine cancer-gene microarray for CGH analysis of tumors. Cytogenet Genome Res 102:254–260.[CrossRef][ISI][Medline]
Vail DM and MacEwen EG, 2000. Spontaneously occurring tumors of companion animals as models for human cancer. Cancer Invest 18:781–792.[ISI][Medline]
Withrow SJ, Powers BE, Straw RC, and Wilkins RM, 1991. Comparative aspects of osteosarcoma. Dog versus man. Clin Orthop 270:159–168.
Yang F, O'Brien PC, Milne BS, Graphodatsky AS, Solanky N, Trifonov V, Rens W, Sargan D, and Ferguson-Smith MA, 1999. A complete comparative chromosome map for the dog, red fox, and human and its integration with canine genetic maps. Genomics 62:189–202.[CrossRef][ISI][Medline]
Zhou Y, Luoh SM, Zhang Y, Watanabe C, Wu TD, Ostland M, Wood WI, and Zhang Z, 2003. Genome-wide identification of chromosomal regions of increased tumor expression by transcriptome analysis. Cancer Res 63:5781–5784.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||