Loss of heterozygosity and point mutation at Aprt locus in T ... - Nature

12 downloads 0 Views 162KB Size Report
been described (Baker et al., 1995, Engle et al., 1996). The targeted (mutant) Aprt .... Toft NJ, Winton DJ, Kelly J, Howard LA, Dekker M, te. Riele H, Arends MJ, ...
ã

Oncogene (2002) 21, 2840 ± 2845 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc

Loss of heterozygosity and point mutation at Aprt locus in T cells and ®broblasts of Pms27/7 mice Changshun Shao1, Moying Yin2, Li Deng1, Peter J Stambrook3, Thomas Doetschman2 and Jay A Tisch®eld*,1 1

Department of Genetics, Rutgers, the State University of New Jersey, Piscataway, New Jersey, NJ 08854-8082, USA; Department of Molecular Genetics, Biochemistry and Microbiology, University of Cincinnati College of Medicine, Cincinnati, Ohio, OH 45267-0521, USA; 3Department of Cell Biology, Neurobiology, and Anatomy, University of Cincinnati College of Medicine, Cincinnati, Ohio, OH 45267-0521, USA

2

Mice null for the Pms2 mismatch repair (MMR) gene exhibit a predisposition to lymphoma, microsatellite repeat instability, and failure of spermatogenesis. To study the role of Pms2 in the maintenance of in vivo genomic integrity in somatic cells, we characterized Aprt mutations in T cells and ®broblasts of 1296C3H Pms27/7Aprt+/7 mice. The spontaneous frequency of DAP-resistant T lymphocytes, as a consequence of APRT-de®ciency, was increased threefold. Point mutation, which accounted for less than 20% of the DAPr mutant clones in Pms2+/+ mice, was predominant in the mutant T cell clones from Pms27/7 mice. These point mutations were predominantly TA to CG transitions. Fibroblasts of Pms27/7 mice exhibited only a modest increase in the frequency of clones with point mutations, such that mitotic recombination was still the primary cause of APRT de®ciency. Thus, the mutator phenotype as a consequence of PMS2 de®ciency is tissue-dependent, which may be related to the tissue-speci®c tumor proneness of Pms27/7 mice. Oncogene (2002) 21, 2840 ± 2845. DOI: 10.1038/sj/ onc/1205358 Keywords: mismatch repair; PMS2; APRT; mouse model; somatic mutation; loss of heterozygosity Introduction DNA mismatch repair (MMR) proteins contribute to the maintenance of genomic integrity in multiple pathways (Modrich and Lahue, 1996, Kolodner and Marsischky, 1999; Buermeyer et al., 1999, Harfe and Jinks-Robertson, 2000). First, MMR corrects mispaired or extrahelical nucleotides that are infrequently introduced during DNA replication, by mis-incorporation or by slippage. Second, MMR-mediated suppression of recombination is believed to reduce the incidence of both translocation and mitotic recombina-

*Correspondence: JA Tisch®eld; E-mail: [email protected] Received 3 December 2001; revised 21 January 2002; accepted 22 February 2002

tion (Harfe and Jinks-Robertson, 2000). Third, MMR plays a role in the regulation of apoptosis (Gong et al., 1999; Toft et al., 1999; Zhang et al., 1999; Zeng et al., 2000). In the absence of MMR, apoptosis is compromised in cells with DNA damage, and the increased chance of survival and proliferation of those cells may allow further accumulation of mutations. Organisms and cells that are MMR-de®cient typically exhibit an increased level of genetic instability, or a mutator phenotype. The mutator phenotype includes instability of microsatellite repeats, elevation of point mutation and increase of recombination between divergent sequences (homeologous recombination). In humans, MMR-de®ciency can lead to several types of cancer. For example, inherited mutations in a subset of MMR genes, including MSH2, MLH1 and PMS2, are responsible for hereditary non-polyposis colorectal cancer (HNPCC). To elucidate the biological functions of the MMR genes in vivo, mouse mutants of various MMR homologs have been generated through targeted gene disruption in embryonic stem cells. Depending on the MMR gene, the homozygous mutant mice displayed one, two or all of the following three phenotypes (Buermeyer et al., 1999): (i) tumor predisposition, (ii) mutator phenotype, and (iii) meiotic defect(s). For example, Mlh17/7 mice exhibit all three phenotypes, whereas Pms17/7 mice only show a mild mutator phenotype. Most studies of the mutator phenotype in MMRde®cient mice were based on observations of microsatellite repeat markers (Yao et al., 1999) or bacterial transgenes, such as supF (Narayanan et al., 1997), lacI and cII (Andrew et al., 1997, 2000; Baross-Francis et al., 2001). However, microsatellite repeat markers only re¯ect frameshift changes in sequences containing mono- or dinucleotide repeats. While bacterial transgenes can show base substitutions, frameshifts and other intragenic alterations, there are concerns as to whether or not their behavior mimics that of endogenous genes since they usually reside as multiple copies in a host cell, are hypermethylated and transcriptionally inactive (Mirsalis et al., 1995). We have documented the use of an endogenous, and ubiquitously expressed, APRT (adenine phosphoribo-

7/7

Somatic mutations in Pms2 C Shao et al

syltransferase) gene as a reporter for loss of heterozygosity (LOH) in vivo (Shao et al., 1999, 2000, 2001). Because Aprt is autosomal, it can detect most genetic alterations at the chromosome level, such as mitotic recombination and multi-loci deletion, in addition to intragenic mutations and small deletions. In this study, we characterized somatic mutations at Aprt in normal T lymphocytes and ®broblasts of 1296C3H Pms27/7Aprt+/7 mice.

Results T cells of Pms27/7 mice exhibit increased point mutation at the Aprt and Hprt loci Though the frequency of DAP-resistant T cells varied greatly within each group, Pms27/7 mice exhibited a higher frequency than Pms2+/+ mice (P=0.017, Mann ± Whitney U-test). The median frequency was increased threefold in Pms27/7 mice, from 22.661076 to 73.361076 (Figure 1). We divided the DAPr clones into two classes according to the absence or retention of the untargeted Aprt allele (Shao et al., 1999). Class I variants, those exhibiting loss of the untargeted Aprt, were predominant in Pms2+/+ mice (40/46, 87%), which is consistent with our previous study (Shao et al., 2000). However, class I clones, those retaining the untargeted allele, replaced class I as the primary cause of APRTde®ciency in Pms27/7 mice (110/170, 65%) (Table 1).

The shift of the mutational spectrum between the two groups is highly signi®cant (P50.0001). Analysis of ¯anking SSLP showed that almost all of the class I variants in either Pms2+/+ or Pms27/7 mice were derived from mitotic recombination (Table 2). The frequency of such clones, as calculated by multiplying the median frequency of DAPr clones with the fraction of mitotic recombination in each group, was 19.261076 and 25.761076 in Pms2+/+ and Pms27/7 mice, respectively. This indicates that mitotic recombination in T cells are not a€ected by the loss of PMS2. The median frequency of class II T cells, on the other hand, di€ers dramatically between Pms2+/+ and Pms27/7 mice, 2.961076 vs 47.661076, a 16-fold di€erence. Class II variants can either be caused by point mutation or by epigenetic silencing of the wildtype (untargeted) Aprt allele (Shao et al., 1999, Rose et al., 2000). We sequenced all ®ve exons, part of the promoter region and three introns of the Aprt gene in 63 clones recovered from Pms27/7 mice and detected point mutations in 51. This indicates that the increase in the frequency of DAPr T cells was primarily the consequence of point mutation. Strikingly, more than half of the mutations in Pms27/7 mice were T to C transitions (Table 3). This predominance of TA to CG transitions is consistent with the mutational spectrum of Hprt in Pms27/7 mice (Shaddock et al., 2001), but is in contrast to the mutational spectrum observed in lacI of Pms27/7 mice (Andrew et al., 1997, BarossFrancis et al., 2001), in which CG to TA transitions predominate. In addition to base substitution, eight frameshift mutations at di- or mononucleotide runs were detected in class II clones recovered from Pms7/7 mice (Table 3, Figure 2). Four of the 51 (8%) clones that were con®rmed to have point mutation were putative sib clones (having the same mutation), indicating that most of mutants had originated independently. Thus, the higher frequency of T cell variants in Pms27/7 mice is caused by an increased mutation rate. We also estimated the frequency of 6-TG resistant (HPRT-de®cient) T cells in Pms27/7 and Pms2+/+ mice. Since Hprt is X-linked, HPRT de®ciency is presumably caused only by intragenic mutations. Consistent with the observations with Aprt, Pms27/7 mice also exhibited a dramatic increase in the frequency of the 6-TGr T cells, 2.3+ 1.461076 (s.e.m., n=9) in Pms2+/+ mice vs 109.8+16.361076 (n=12) in Pms27/7 mice. In comparison to the 16-fold increase in the median frequency of class II DAPr variants, which are primarily caused by point mutation, in Pms27/7 mice, the magnitude of increase for 6TGr variants is much greater (48-fold). The large size Table 1

Figure 1 Frequency of DAPr T cells in individual mouse spleens. Each spot represents one spleen. Bars represent median values

mice

+/+

Pms2 Pms27/7

2841

Class II variants DAPr T cells in Pms27/7 mice No. mice

Class I

Class II

Total

% of class II

9 15

40 60

6 110

46 170

13 65 Oncogene

7/7

Somatic mutations in Pms2 mice C Shao et al

2842

Table 2 Mutational spectrum of class I variant clones Cell type

Genotype

Mitotic recombination

Deletion or gene conversion

Total no. of clones

T cells

Pms2+/+ Pms27/7 Pms2+/+ Pms27/7

35 43 14 25

1 0 1 2

36 43 15 27

Fibroblasts

Table 3 Point mutations in Aprt of Pms27/7 mice are primarily T4C transitions Transitions T4C A4G C4T G4A Transversions 2 bp deletions 1 bp deletions 1 bp deletions Total

Pms2+/+

Pms2+/7

Pms27/7

1

1

1

1

26 3 1 6 3 1 4 3

2 1

5

2

47

®broblasts in Pms27/7 mice (35/54), though the fraction of class I was lower in Pms27/7 mice (65%), than in Pms2+/+ mice (83%) (Table 4). As with the T cells, most of the class I ®broblast clones were caused by mitotic recombination in ®broblasts. We calculated that the frequency of mitotic recombination is similar between the two groups of mice, 62.361076 in Pms2+/+ vs 72.261076 in Pms27/7 mice. We also estimated the frequency of 6-TGr ®broblasts in Pms2+/+ and Pms27/7 mice (Table 5). The 6-TGr clones were recovered from about 50% of the ears of the Pms27/7 mice, in comparison to 16% in wild-type mice. Excluding one Pms27/7 outlier, the accumulative frequency of 6-TGr is about four times higher in Pms27/7 mice than in wild-type mice, 4.361076 in Pms2+/+ vs 17 x 1076 in Pms27/7. These ®ndings with ®broblasts indicate that although point mutations are more common in Pms27/7 than in Pms2+/+ mice, the magnitude of increase is smaller than that in T lymphocytes. Discussion We estimated the frequency of in vivo DAPr and 6-TGr mutants in normal T cells and ®broblasts of Pms27/7 mice and characterized the mutational spectrum of

Figure 2 Distribution of frameshift mutations in ®ve exons of Aprt in Pms27/7 mice. Upward arrow head, deletion; downward arrow head, insertion

of the Hprt gene, 33 kb (Melton et al., 1984) vs 3 kb for Aprt (Dush et al., 1985), and certain unique structures, such as the six G repeat, may make it more unstable than Aprt in the absence of MMR. Fibroblasts of Pms27/7 mice exhibit a modest increase in point mutations at the Aprt and Hprt loci In contrast to the signi®cant increase in the frequency of DAPr mutants in T cells of Pms27/7 mice, the frequency of DAPr ®broblasts was not signi®cantly di€erent between Pms27/7 and Pms2+/+ mice (P=0.16), although the median frequency was higher in Pms27/7 mice than in Pms2+/+ mice (11161076 versus 7561076) (Figure 3). Molecular characterization showed that class I variants still accounted for the majority of the DAPr Oncogene

Figure 3 Frequency of DAPr ®broblasts in individual mouse ears. Each spot represents one ear. Bars represent median values

7/7

Somatic mutations in Pms2 C Shao et al

DAPr cell variants. We observed a dramatic elevation of point mutation in T cells of Pms27/7 mice, which signi®cantly shifted the mutational spectrum. While APRT-de®ciency in T cells is predominantly caused by mitotic recombination in Pms2+/+ mice, it is largely due to point mutation in Pms27/7 mice. However, skin ®broblasts only exhibit a modest increase in point mutation, as re¯ected by the relatively unchanged mutational spectrum of DAPr ®broblasts of Pms27/7 mice. In general, the absolute frequency of clones derived from mitotic recombination is not signi®cantly di€erent between Pms2+/+ and Pms27/7 mice. Since the assay using Aprt as a reporter is based on the clonal proliferation of mutant cells in vitro, we were unable to extend our observations to other tissues, such as intestinal epithelium. However, with the two types of tissue we examined, we can conclude that the Pms27/7 mutator phenotype is tissue-dependent. Our ®ndings are in contrast to a study using supF or supFG1 transgenic mice, in which spleen and skin tissues of Pms27/7 mice exhibited equal elevation in mutation frequency (Narayanan et al., 1997), but are consistent with a study of Msh27/7lacI transgenic mice in which lacI showed a greater magnitude of increase in mutation frequency in thymus than in brain compared to Msh2+/+ mice, 15.2-fold and 4.8-fold respectively (Andrew et al., 1997). It is unclear as to why point mutation is di€erentially elevated in di€erent tissues of Pms27/7 mice. One possible explanation may relate to the proliferation history of each tissue. A recent study showed that quiescent mouse cells do not repair premutatgenic damage and that mutation frequency is not changed during quiescence (Bielas and Heddle, 2000). DNA repair and increase of mutation frequency were observed only when quiescent cells were induced to proliferate. Thus, proliferation is required for both

Table 4 Class I variants predominate in DAPr ®broblasts of Pms2+/+ and Pms27/7 mice +/+

Pms2 * Pms2+/+ Pms27/7

Class I

Class II

Total

% of class I

96 15 35

22 3 19

118 18 54

81 83 65

*From Shao et al. (2000)

Table 5

CFU 20 000 ± 40 000 40 001 ± 60 000 60 001 ± 80 000 480 000 Total

No. ears

mice

repair and mutation ®xation. Since mismatches in DNA are primarily introduced during DNA replication, via misincorporation or slippage, the accumulation of DNA replication errors in a cell may be a function of the number of prior cell divisions. Possibly, ear ®broblasts undergo fewer cell divisions than peripheral T cells in vivo, so that their chances of accumulating replication errors are reduced in comparison. The primary function of MMR is to maintain genomic integrity. Thus, it is likely that tumor proneness displayed by the MMR-de®cient mice is driven by increased genetic instability, which can render key tumor suppressor genes more mutable. However, previous studies using microsatellite repeat and transgenic reporters showed that tumor spectrum and genetic instability are not necessarily correlated. For example, although Pms27/7 mice are predominantly predisposed to lymphomas, the mutation frequencies of supF and supFG1 are equally elevated in the skin and spleen tissues of those mice (Narayanan et al., 1997). Also, whereas both Pms27/7 and Mlh17/7 mice exhibit microsatellite instability in intestinal tissues, only Mlh17/7 mice are predisposed to intestinal carcinoma (Prolla et al., 1998). Thus, it is possible that the tissue-speci®c tumor proneness in MMR-de®cient mice is not solely determined by an increased level of genomic instability. Alternatively, the transgenic and microsatellite reporters may have not adequately re¯ected genetic changes at endogenous loci. Our study showed that Aprt exhibits a di€erent spectrum of point mutations in comparison to that of lacI. While more than half (55%) of the point mutations are T to C transitions at Aprt of Pms27/7 mice, only 14% are caused by TA to CG transitions at lacI (Baross-Francis et al., 2001). Also, deletions and insertions at mononucleotide runs account for 15% of the point mutations at Aprt, but such frameshift mutations account for 34% at lacI (Baross-Francis et al., 2001). Interestingly, another endogenous reporter gene, Hprt, displays a similar spectrum as Aprt in Pms27/7 mice (Shaddock et al., 2001). MMR proteins in yeast and possibly mice also have anti-recombination activity, which suppresses recombination between divergent DNA sequences (Harfe and Jinks-Robertson, 2000, Shao et al., 2001). Loss of MMR in yeast usually results in increased homologous

2843

Distribution of 6-TGr ®bbroblasts in ears of Pms2+/+ and Pms27/7 mice Pms2+/+ No. ears with 6-TGr

No. 6-TGr colonies

No. ears

Pms27/7 No. ears with 6-TGr

No. 6-TGr colonies

4 6 3 5

0 1 1 1

0 4 1 1

6 12 8 3

1 8 4 2

5 9 11 448

18

3 (16.7%)

6

29

15 (52%)

473

CFU, colony-forming unit Oncogene

7/7

Somatic mutations in Pms2 mice C Shao et al

2844

recombination (Harfe and Jinks-Robertson, 2000). It appears from the present study that mitotic recombination between the two homologues of chromosome 8, derived from strain 129 and strain C3H, respectively, is not a€ected in Pms27/7 mice. Probably, PMS2 is not involved in the recognition and processing of the recombination heteroduplex during mitotic recombination. Alternatively, the sequence dissimilarity between the two strains may not be sucient to produce a signi®cant inhibitory e€ect on recombination. Studies with other MMR gene knockout mice, such as Msh27/7 and Mlh17/7 mice, will address these questions.

Materials and methods Mice Production of Pms2-de®cient and Aprt-de®cient mice (both using D3 embryonic stem cells of 129/Sv mouse origin) has been described (Baker et al., 1995, Engle et al., 1996). The targeted (mutant) Aprt allele contains a neo insertion in exon 3, making it easily distinguishable from the wild-type allele. Pms2+/7 mice in N5C57 background were backcrossed to 129/Sv and C3H/HeJ, respectively, for four generations to produce (N4)129Pms2+/7 and (N4)C3HPms2+/7 mice. 1296C3HAprt+/7 hybrid mice with di€erent Pms2 genotypes were generated by crossing (N4)129Pms2+/7Aprt+/7 mice to (N4)C3HPms2+/7mice. Aprt and Pms2 genotyping were done as described (Shao et al., 1999; Baker et al., 1995). The hybrid mice were 2 to 3 months old when sacri®ced.

(2 mg/ml) or 2,6-diaminopurine (DAP) (50 mg/ml). Their colony-forming eciency (CFE) was estimated by seeding four cells per well in the presence of feeder cells, which are stimulated splenocytes irradiated with 6000 rads of X-rays. Mutant T cell colonies were scored on day 10. The mutant frequency was calculated by applying Poisson statistics. Skin ®broblasts were prepared as described (Shao et al., 1999). They were seeded in 100 mm plates (16106 cells/plate) containing 6-TG (5 mg/ml) or DAP (50 mg/ml) for recovery of drug-resistant cell colonies. Their CFE was estimated by plating 5 000 to 10 000 cells in the presence of 16106 irradiated ®broblasts. Mutant ®broblast colonies were scored on day 12. Molecular analysis Characterization of DAPr clones was as described (Shao et al., 1999). Brie¯y, DAPr clones were ®rst divided into two classes by the absence (class I) or retention (class II) of the untargeted allele. Class I clones were further characterized with polymorphic microsatellite markers along chromosome 8 to determine their mechanism(s) of origin. Clones that exhibited LOH at loci distal to Aprt but remained heterozygous at loci proximal to Aprt were interpreted to be derivatives of mitotic recombination, as was previously demonstrated more rigorously (Shao et al., 1999). Clones that did not exhibit LOH at the most distal microsatellite marker locus, D8Mit56, were interpreted to be the outcome of gene conversion or interstitial deletion. Class II clones were sequenced for all ®ve exons, introns 1, 3 and 4, and part of the promoter region of the untargeted Aprt allele for detection of point mutation.

Isolation of mutant clones Splenocytes were isolated as described (Meng et al., 1998). After being stimulated with conconavalin A overnight, the splenocytes were seeded in 96-well plates at the density of 26104 cells/well in the presence of 6-thioguanine (6-TG)

Acknowledgments We thank RM Liskay for generously providing us with Pms2 knockout mice. This work was supported by grants from NIH (R01DK38185, P01ES05652 and P30ES05022).

References Andrew SE, Reitmair AH, Fox J, Hsiao L, Francis A, McKinnon M, Mak TW and Jirik FR. (1997). Oncogene, 15, 123 ± 129. Andrew SE, Xu XS, Baross-Francis A, Narayanan L, Milhausen K, Liskay RM, Jirik FR and Glazer PM. (2000). Carcinogenesis, 21, 1291 ± 1295. Baker SM, Bronner CE, Zhang L, Plug A W, Robatzek M, Warren G, Elliott EA, Yu J, Ashley T, Arnheim N, Flavell RA and Liskay RM. (1995). Cell, 82, 309 ± 319. Baross-Francis A, Makhani N, Liskay RM and Jirik FR. (2001). Oncogene, 20, 619 ± 625. Bielas JH and Heddle JA. (2000). Proc. Natl. Acad. Sci. USA, 97, 11391 ± 11396. Buermeyer AB, Deschenes SM, Baker SM and Liskay RM. (1999). Annu. Rev. Genet. 33, 533 ± 564. Dush MK, Sikela JM, Khan SA, Tisch®eld JA, Stambrook PJ. (1985). Proc. Natl. Acad. Sci. USA, 82, 2731 ± 2735. Engle SJ, Stockelman MG, Chen J, Boivin G, Yum MN, Davies PM, Ying MY, Sahota A, Simmonds HA, Stambrook PJ and Tisch®eld JA. (1996). Proc. Natl. Acad. Sci. USA, 93, 5307 ± 5312. Oncogene

Gong JG, Costanzo A, Yang HQ, Melino G, Kaelin Jr WG, Levrero M and Wang JY. (1999). Nature, 399, 806 ± 809. Harfe BD and Jinks-Robertson S. (2000). Annu. Rev. Genet., 34, 359 ± 399. Kolodner RD and Marsischky GT. (1999). Curr. Opin. Genet. Dev., 9, 89 ± 96. Melton DW, Konecki DS, Brennand J, Caskey CT. (1984). Proc. Natl. Acad. Sci. USA, 81, 2147 ± 2151. Meng Q, Skopek TR, Walker DM, Hurley-Leslie S, Chen T, Zimmer DM and Walker VE. (1998). Environ. Mol. Mutagen., 32, 236 ± 243. Mirsalis JC, Monforte JA and Wineger RA. (1995). Annu. Rev. Pharmacol. Toxicol., 35, 145 ± 164. Modrich P and Lahue R. (1996). Annu. Rev. Biochem., 65, 101 ± 133. Narayanan L, Fritzell JA, Baker SM, Liskay RM and Glazer PM. (1997). Proc. Natl. Acad. Sci. USA, 94, 3122 ± 3127. Prolla TA, Baker SM, Harris AC, Tsao JL, Yao X, Bronner CE, Zheng, B, Gordon M, Reneker, J, Arnheim N, Shibata D, Bradley A and Liskay RM. (1998). Nat. Genet., 18, 276 ± 279.

7/7

Somatic mutations in Pms2 C Shao et al

Rose JA, Yates PA, Simpson J, Tisch®eld JA, Stambrook PJ, Turker MS. (2000). Cancer Res., 60, 3404 ± 3408. Shaddock JG, Dobrovolsky VN, Mittelstaedt RA, Bendre S, He¯ich RH and Parsons BL. (2001). Environ. Mol. Mutagen., 37, 68. Shao C, Deng L, Henegariu O, Liang L, Raikwar N, Sahota A, Stambrook PJ and Tisch®eld JA. (1999). Proc. Natl. Acad. Sci. USA, 96, 9230 ± 9235. Shao C, Deng L, Henegariu O, Liang L, Stambrook PJ and Tisch®eld JA. (2000). Proc. Natl. Acad. Sci. USA, 97, 7405 ± 7410. Shao C, Stambrook PJ and Tisch®eld JA. (2001). Nat. Genet. 28,169 ± 172.

mice

Toft NJ, Winton DJ, Kelly J, Howard LA, Dekker M, te Riele H, Arends MJ, Wyllie AH, Margison GP and Clarke AR. (1999). Proc. Natl. Acad. Sci. USA, 96, 3911 ± 3915. Yao X, Buermeyer AB, Narayanan L, Tran D, Baker SM, Prolla TA, Glazer PM, Liskay RM and Arnheim N. (1999). Proc. Natl. Acad. Sci. USA, 96, 6850 ± 6855. Zeng M, Narayanan L, Xu XS, Prolla TA, Liskay RM and Glazer PM. (2000). Cancer Res., 60, 4889 ± 4893. Zhang H, Richards B, Wilson T, Lloyd M, Cranston A, Thorburn A, Fishel R and Meuth M. (1999). Cancer Res., 59, 3021 ± 3027.

2845

Oncogene