Frequent Abnormalities of TSG101 Transcripts in Human Prostate Cancer

Frequent abnormalities of TSG101 transcripts in human prostate cancer

Zijie Sun, Jing Pan, Glenn Bubley and Steven P Balk

Cancer Biology Program, Division of Hematology-Oncology, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA 02215, USA

TSG101 has been identi®ed as a candidate tumor gene also results in cellular transformation (Li and suppressor gene and abnormal transcripts have been Cohen, 1996). identi®ed in a substantial fraction of breast cancers. To The human TSG101 gene is 94% homologous to determine whether TSG101 expression is commonly the murine gene at the protein level and has been altered in other tumors, a series of 15 primary and localized to chromosome 11p15.1-15.2, a region metastatic prostate cancers were analysed by reverse showing loss of heterozygosity (LOH) in a variety of transcriptase-PCR ampli®cation. Abnormal transcripts human malignancies (Li et al., 1997). A recent analysis with extensive deletions in the coding region were found of TSG101 transcripts in primary human breast in nine of these tumors, while only the normal transcript carcinomas revealed a series of deletions, although was found in control and benign prostatic hypertrophy the normal transcript was also detected in all cases (Li tissues. More than one abnormal transcript was found in et al., 1997). The many biological similarities between four of these nine cases and distinct abnormal TSG101 breast cancer and PCa prompted us to examine transcripts were found in separate biopsies taken from TSG101 expression in PCa. In this study, TSG101 one tumor. Importantly, the normal TSG101 transcript transcripts from 15 cases of primary and metastatic was undetectable in two metastatic prostate cancers, PCa were analysed by RT ± PCR. Overall, deletions indicating the absence of TSG101 protein. Sequence were detected in nine cases, including both primary analysis demonstrated that there were at least six and metastatic prostate cancers. Two or more distinct deletions, with four of these deletions found in deletions were found in multiple samples and distinct more than one tumor sample. The most commonly deletions were observed in two separate biopsies from identi®ed deletion, from bp 153 to 1055, was identical to the same patient. In two biopsies there was no a deletion reported previously in breast cancer. These evidence of the normal TSG101 transcript. Sequence results demonstrate that TSG101 transcripts are fre- analysis demonstrated that there were at least six quently abnormal in prostate cancer and suggest that distinct deletions, with four of these deletions found in loss of TSG101 protein contributes to disease develop- more than one tumor sample. The deletions removed ment or progression. large portions of the coding region. These results indicate that loss of TSG101 protein is common in Keywords: prostate cancer; tumor suppressor; aberrant PCa and that this could represent a signi®cant step in RNA splicing; TSG101 tumor progression.

Results Introduction TSG101 transcripts from a series of prostate cancers Prostate cancer (PCa) is currently the most common (Table 1) and prostate epithelial and stromal cells malignancy in males and causes more than 40 000 from benign prostatic hypertrophy (BPH) were deaths in the United States annually (Parker et al., analysed by RT ± PCR using primers that bracketed 1996). Androgens and a functional androgen receptor the TSG101 protein coding region (Figure 1a). In the are required for the development of PCa. However, three tumor samples derived from radical prostatec- despite considerable eort, the identity of additional tomies (PCa1-3), matched prostate tissue which was genes which contribute to the development of PCa microscopically free of tumor was also examined. In remains unclear. Tumor susceptibility gene 101 these later tumor free prostate samples and in the (TSG101) was identi®ed in a screen using homozygous BPH samples, the predicted full length 1145 bp disruption in mouse ®broblasts to identify candidate TSG101 transcript was the only signi®cant product tumor suppressors (Li and Cohen, 1996). The TSG101 observed (Figure 2a). The full length transcript was gene encodes a 43 kDa protein with a proline-rich also the only product in the PCa1 and two other domain and a leucine zipper containing coiled-coil tumors, but a minor second band of about 170 bp domain, suggesting that it functions as a transcription was observed in the PCa3 tumor sample. Sequence factor (Li and Cohen, 1996). Functional inactivation of analysis con®rmed that this second band was a TSG101 by antisense RNA in mouse ®broblasts leads TSG101 transcript with a large deletion (see below). to transformation and the ability to form metastatic Tumor tissue derived from transurethral prostatec- tumors in nude mice, although overexpression of this tomies (channel TURPs) in patients with more advanced disease (PCa4-6), yielded readily detectable abnormal bands (Figure 2a). These were also con- Correspondence: Z Sun ®rmed by sequencing to be derived from TSG101 and Received 23 May 1997; revised 27 August 1997; accepted 27 August contained large deletions. In PCa5, two biopsies where 1997 obtained and yielded distinct abnormal transcripts Tumor susceptability gene in prostate cancer ZSunet al 3122 Table 1 Summary of Abnormal Transcripts of TSG101 Gene in a Human Prostate Cancer RT-PCR Casesb Tissue sites Products cDNA Sequence (Form)a PCa1 PCa2 PCa3 PCa4 PCa5 PCa6 PB7 MW N T N T N T T T1 T2 T E S NC MW PCa3 Prostate Abnormal Deletion bp 249 ± 1223 (2) PCa4 Prostate Abnormal Deletion bp 153 ± 1072 (3) PCa5 Prostate (1) Abnormal Deletion bp 132 ± 1055 & 1.6 Kb — Normal 63 bp insertion (4) 1.0 Kb — transcripts (1145 bp) Prostate (2) Abnormal Deletion bp 153 ± 1055 (1) 0.5 Kb — PCa6 Prostate Abnormal Deletion bp 153 ± 1055 (1) 0.3 Kb — Abnormal Deletion bp 283 ± 1055 (5) 0.2 Kb — transcripts Deletion bp 132 ± 731 (6) 0.1 Kb — PCa8 Bone Marrow Abnormal Deletion bp 153 ± 1055 (1) PCa10 Bone Marrow Abnormal Deletion bp 153 ± 1055 (1) Deletion bp 249 ± 1223 (2) b PCa11 Bone Marrow Abnormal Deletion bp 153 ± 1055 (1) Deletion bp 132 ± 731 (6) PCa14 Bone Marrow Abnormal Deletion bp 153 ± 1055 (1) MW PCa8 PCa9 PCa10 PCa11 PCa12 PCa13 PCa14 PCa15 PCa16 NC MW Deletion bp 249 ± 1223 (2) PCa15 Bone Marrow Abnormal Deletion bp 283 ± 1055 (5) 1.6 Kb — Normal 1.0 Kb — transcripts aAbnormal transcript forms 1, 2, 4 and 6 are out of reading frame (1145 bp) and forms 3 and 5 are in-frame. bPCa3 was derived from a radical 0.5 Kb — prostatectomy and PCa4-6 were derived from transurethral 0.3 Kb — Abnormal 0.2 Kb — transcripts prostatectomies (channel TURPs) in patients with advanced 0.1 Kb — androgen independent prostate cancer Figure 2 Detection of abnormal transcripts of TSG101 gene in human prostate cancer. (a) PCR products after nested RT ± PCR ampli®cation from tumor (T) and matched normal (N) tissues in three primary tumors (PCa1-3), three channel TURPs (PCa4-6) as well as epithelial (E) and stromal (S) cells from one benign prostatic hypertrophy (BPH) patient (PB7) were analysed on 1.6% agarose gels. Arrowheads show the positions of ampli®ed abnormal transcripts which were excised and sequenced. Six abnormal transcripts, with molecular weights from 171 bp to 546 bp (also see Figure 1b), were observed only in tumor samples. A 1145 bp band representing the normal transcript is also indicated. No speci®c DNA band was detected in a reaction without template DNA used as a negative control (NC). Standard 1 Kb ladder (GIBCO) was used as a molecular weigh marker (MW). (b) RT ± PCR products from metastatic tumors. As described above, the tumor samples were analysed by RT ± PCR and electrophoresis

mutations or deletions were observed in the full length transcripts from the three matched normal prostate tissues (PCa1-3, data not shown). Sequence analyses of 15 abnormal bands revealed six dierent deletions, Figure 1 Outline of wild-type and abnormal TSG101 transcripts. all of which removed most of the TSG101 coding (a) The top line shows the intact TSG101 cDNA sequence. The closed bar represents the coding sequence within the untranslated region. As indicated in Table 1, four of these six regions. Positions of the primers used and sizes of PCR products deletions also changed the reading frame. The for full-length TSG101 cDNA are shown. (b) Structure of six sequences around these deletions are shown in Figure abnormal TSG101 transcripts isolated from tumor tissues. The 3 and the deletions are outlined in Figure 1. The deletions are marked according to the nucleotide numbers. Size of abnormal transcript designated form four included a each abnormal transcript is indicated. Form 4 contains a 42 bp insertion between nt 132 and 133. The detailed sequence for this 42 bp insertion between nt 132 and 133 (Figures 1 and abnormal tanscript is shown in Figure 3b 3b). The transcript containing a deletion between bp 153 to 1055 (form 1) was most commonly detected, (PCa5, T1 and T2). Analyses of distant metastases in occurring in six of the PCa tumors (Table 1). This nine additional cases (PCa8-16) yielded abnormal deletion was identical to one described previously as a TSG101 transcripts in ®ve tumors (Figure 2b). 153 to 1053 deletion in human breast cancer (Li et al., Signi®cantly, the full length TSG101 transcript was 1997). Forms 2, 5, and 6 were also identi®ed in more undetectable in two of these latter cases (PCa8 and 10) than one tumor. Form 4, which contained a 42 bp and was only a minor product in two other cases insertion between nt 132 and 133 and a deletion from (PCa11 and 14). Overall, tumors from eight of twelve nt 153 to 1055, was only found in one sample patients with advanced diseases had products consis- (PCa5T1). However, form 4 was similar to an tent with abnormal transcripts. abnormal transcript described in breast cancer, which To con®rm the identity of these RT ± PCR products was reported as a 63 bp insertion between nt 133 and as TSG101 transcripts and to map the deletions, 1054 (sequence data in that report indicates that the multiple full length and abnormally sized PCR 63 bp insertion includes the region between nt 133 and fragments were isolated and sequenced. No point 153, as in form 4) (Li et al., 1997). Tumor susceptability gene in prostate cancer ZSunet al 3123 a a A C G T A C G T A CG T PCa3 PCa4 PCa5 PCa6 PCa8 PCa10 PCa11 PCa13 PCa14 PCa15 PCa16 MW N T T T1T2 T T T T T T T T NC MW

1353 bp — FHIT Normal 872 bp — transcripts 603 bp — (615 bp)

310 bp —

1353 bp — 872 bp — β2 Microglobulin Form 1 Form 2 Form 3 603 bp — normal transcripts (658 bp) A C G T A C G T A C G T 310 bp —

Figure 4 Detection of FHIT and b2-microglobulin transcripts by RT ± PCR. Thirteen cDNA samples from tumor (T) and a matched normal (N) tissue sample of prostate cancer patients as described above were ampli®ed to detect FHIT and b2- microglobulin transcripts. PCR products were analysed on 1.6% agarose gels. The 615 bp and 658 bp bands representing the normal transcripts of FHIT (a) and b2-microglobulin (b), respectively, are indicated. The jw 174/HaeIII (New England Biolab) was used as a molecular weight marker (MW). A negative control was indicated as NC

Form 4 Form 5 Form 6 Discussion b This report indicates that abnormal TSG101 transcripts are common in prostate cancer. These abnormalities all involved deletion of most of the coding region, indicating that any protein derived from these transcripts would be Figure 3 TSG101 cDNA Sequences of abnormal transcripts at nonfunctional. The extensive deletions of the coding deletion junction regions. (a) Six dierent abnormal TSG101 transcripts were isolated and their sequences are shown. Based on region also make very unlikely the production of the wild-type TSG101 cDNA sequence (see c), the deletions in proteins which would have dominant negative eects. abnormal transcripts were determined and are shown. The These observations indicate that TSG101 may function nucleotides corresponding to both the end points and start as a tumor suppressor in prostate cancer. Further crucial points of the deletion junction are marked by * in the sequencing evidence for this hypothesis would be the demonstration gels and their numbers are indicated in the listed sequences. The arrowheads indicate the deletion junction. (b) Form 4 is indicated that both TSG101 genes were nonfunctional in tumor as having a 63 bp insertion, but as shown form 4 contains a 42 bp cells. Unfortunately, both in this study and in the insertion between bp 132 and 133. The detailed sequence for this previous study in breast cancer (Li et al., 1997), normal abnormal transcript is represented full length TSG101 transcripts were also ampli®ed from most tumor samples. Although these full length transcripts may have been derived from the substantial Due to the small amounts of tumor tissue numbers of normal cells contained in most fresh tumor available, these studies of TSG101 transcripts have samples, an origin from the tumor cells could not be necessarily been done by RT ± PCR and con®rmation excluded. Therefore, in this report the identi®cation of through Northern blot or nuclease protection experi- two prostate cancer samples with no detectable full ments has not been possible. The ®nding of only full length TSG101 transcripts, and presumably no normal length TSG101 transcripts in biopsies without tumor TSG101 protein, provides important support for the suggests that the deletions found in the tumor hypothesis that TSG101 can function as a tumor samples were not artifacts of the PCR ampli®cation. suppressor. The complete absence of full length TSG101 Although LOH at 11p15, the location of the human transcripts in two samples (PCa8 and 10) was also TSG101 gene, has been observed in many tumors, it is inconsistent with an ampli®cation artifact. The not yet clear whether deletions in the TSG101 gene possible contamination of the tumor samples with account for the precise deletions seen in TSG101 factors which could produce a bias towards truncated transcripts from multiple tumors. Sequence analysis PCR products was further addressed by an analysis of the abnormal TSG101 transcripts in this report of additional transcripts in the same samples which showed identical deletions in multiple tumors. The contained abnormal TSG101 transcripts. RT ± PCR most common abnormal transcript was form 1, a analyses were carried out to examine transcripts of deletion from bp 153 to 1055, which was identi®ed in FHIT (Ohta et al., 1996; Sozzi et al., 1996) and b2- six patients (six tumor samples). This identical deletion microglobulin (Gussow et al., 1987). In contrast to was also reported previously in tumor samples from the TSG101 results, only full length transcripts were breast cancers (Li et al., 1997). It is noteworthy that detected (Figure 4a and b). consensus sequences for RNA splicing are observed at Tumor susceptability gene in prostate cancer ZSunet al 3124 the 5' and 3' ends of deleted sequences in the abnormal The predicted TSG101 protein structure is suggestive forms 1, 3 and 5 (see Figure 5) (Kramer, 1996; le et of a transcription factor, with a leucine zipper al., 1986). Therefore, alternative splicing due to diverse containing coiled-coil domain and a proline rich and potentially subtle alterations in the TSG101 gene region consistent with a transactivation domain (deletions, insertions and/or point mutations) may (Mitchell and Tjian, 1989). A previous report showing account for the precise deletions seen in multiple that TSG101 may interact with stathmin provides a prostate and breast cancers. Direct structural studies clue as to the mechanism through which this protein of the TSG101 gene will require larger amounts of may function as a tumor suppressor (Johnson et al., tissue containing a high proportion of tumor cells or 1995). Stathmin, also termed Oncoprotein 18, phos- appropriate cell lines. phoprotein 18, and p19, is a highly conserved 19 kDa If TSG101 does function as a tumor suppressor, it is cytoplasmic protein which is phosphorylated in not clear whether it is lost as an early or late event in response to a wide variety of signals (Beretta et al., carcinogenesis. The frequency of abnormal transcripts 1993; Chneiweiss et al., 1989; Duraj et al., 1995; in more advanced prostate cancers from channel Leighton et al., 1993; Nakamura et al., 1995; Sobel, TURPs (PCa4-6) and from metastatic sites (PCa8-16), 1991). These observations suggest that TSG101 may appeared to be greater than in the radical prostatect- function as a transcription factor downstream of omy samples (PCa1-3). However, PCa samples derived stathmin to suppress cell growth and that stathmin from radical prostatectomies contain a large fraction of may stimulate cell growth in part through binding and normal epithelial and stromal cells, making it dicult sequestering of TSG101. Finally, since TSG101 to quantitatively assess abnormal TSG101 transcripts expression appears to be ubiquitous, it may contribute derived from the tumor. Several of the advanced to the development of progression of tumors in prostate cancers expressed more than one abnormal addition to breast and prostate cancers. transcripts. Analyses of additional minor bands in some of these advanced tumors suggest further heterogeneity (data not shown). If each of these Materials and methods abnormal transcripts represents a distinct genetic alteration, then these observations would suggest that Patient tissues loss of TSG101 is a later event. A previous report also A total of 16 tumor samples from 15 prostate cancer suggested that loss of TSG101 may not be an early patients were analysed in this study. Tumor and matched event (Li and Cohen, 1996). Eorts to analyse TSG101 normal tissues were obtained from radical prostatectomies expression directly in early and late stage tumors by in three cases. Tumor was also obtained from channel immunohistochemistry and in situ hybridization are transurethral prostatectomies (channel TURPs) and biop- underway. sies from metastatic sites in patients with more advanced

Figure 5 Sequences of human TSG101 cDNA and deletions. Full length cDNA sequence of TSG101 gene is represented and the predicted translation start site and stop codon are underlined. The ®rst nucleotides of each exon are marked with an arrowhead, based on the previously published sequence (Li et al., 1997). The detailed sequences of the six abnormal transcripts identi®ed here in human prostate cancer are shown. The nucleotides immediately 5' and 3' of the deletions were bolded and underlined as well as marked by arrowheads above and below, respectively. The sequence of the 42 bp insertion in abnormal form 4 marked by * is shown in Figure 3b. The potential sequences for RNA splicing are boxed at the 5' and 3' ends of deleted sequences in the abnormal forms 1, 3 and 5 Tumor susceptability gene in prostate cancer ZSunet al 3125 androgen independent PCa. Among 16 tumor samples, volume for 30 cycles of 958C for 45 s, 648C for 40 s and 728C seven were from prostate and nine from metastatic sites for 1 min. The PCR products were analysed on 1.6% agarose (seven from bone marrow, one from a skin nodule, and one gels and stained with ethidium bromide. from a malignant pleural eusion). In patient 5 (PCa5), Ampli®cation of FHIT gene was performed exactly as two tumor samples from the prostate were isolated and described previously (Ohta et al., 1996; Sozzi et al., 1996). 1ml used in this study. The presence of tumor in all samples of cDNA was used in primary ampli®cations with 5U2 and was con®rmed microscopically, although the ratio of 3D2 primers. Subsequently, 1ml of 20-fold diluted primary tumour to normal tissue varied. Stromal and epithelial PCR product was ampli®ed using nested 5U1 and 3D1 cells fractions were prepared by collagenase digestion primers. Human b2 microglobulin transcripts were ampli®ed followed by dierential centrifugation. for 37 cycles of 30 s at 958C, 40 s at 568C, 30 s at 728C and ®nal extension of 7 min at 728C. The sequences of the primers were 5'-ATCCAGCGTACTCCAAAGATTCAG-3' RNA extraction and RT ± PCR and 5'-AAATTGAAAGTTAACTTATGCACGC-3' (Gus- Both tumor and normal specimens were frozen immedi- sow et al., 1987). ately after surgical resection or biopsy and stored in liquid nitrogen. Total RNA was extracted from frozen tissues cDNA subcloning and sequencing using RNAzol B (TEL-TEST Inc, Friendswood, TX). cDNA was synthesized from 3mg of total RNA in a 10ml To further analyze potential abnormal TSG101 transcripts, volume with 10 mM DTT, 500 mM dNTPs, 50 ng/oligo(dT) ampli®ed PCR samples were resolved on a series of 1.6% (Pharmacia), 20 unit of RNasin (Promega) and 25 units of low melting agarose gels. Each DNA band which could be AMV reverse transcriptase (Promega). The total RNA clearly visualized as distinct from wild-type was excised samples were ®rst heated for 5 min at 728C, before enzyme from the gels. The agarose was digested using b-agarase addition, and were then incubated at 428C for 90 min. The (New England, Biolabs) and DNA was precipitated with reaction was stopped by inactivating the enzyme at 958C 100% ethanol. The DNA was dissolved in 10 ± 15mlTE for 5 min. The reaction was diluted to 100ml, and 1ml buer and 20 ± 40 ng of DNA was used for ligation with aliquots were used for subsequent PCR ampli®cations. the pCR II T-vector (Invitrogen). The ligation mixtures Primary PCR ampli®cations were performed in a 20ml were transformed into DH5a competent cells and at least volume containing 1ml of cDNA, 0.8 mM of P1 primer (5'- two plasmids derived from each DNA band in each patient CGGGTGTCGGAGAGCCAGCTCAAGAAA-3') and P2 sample were sequenced. In all cases, the insert size of the primer (5'-CCTCCAGCTGGTAGCAGAGAAGTCGT-3'), plasmids matched the products isolated from the gels. 50 mM dNTP and 0.5 unit of Taq polymerase (Promega) Sequencing was carried out in both directions by the (see Figure 1). The PCR reaction consisted of an initial dideoxynucleotide termination reaction using Sequenase denaturation at 958C for 2 min and 25 cycles of 30 s at 958C, (US Biologicals). 40 s at 628C and 1 min at 728C, and ®nal extension of 7 min at 728C using a Perkin-Elmer PCR themocycler. The ampli®ed product was diluted 20-fold in TE buer, and 1ml of the diluted reaction product was subjected to a second Acknowledgements round of PCR ampli®cation using two nested primers P3 (5'- The authors thank Dr Daniel Morganstern for preparing AGCCAGCTCAAGAAAATGGTGTCCAAG-3') and P4 epithelial and stromal cell fractions and Dr Michael Lu for (5'-TCACTGAGACCGGCAGTCTTTCTTGCTT-3'). The kindly providing several of the human prostate tissues. primers were identical to those described previously (Li et This work was supported by NIH research grants CA70297 al., 1997). The PCR reaction was also carried out in a 20 ml and CA65647 to ZJS and SPB, respectively.

References

Beretta L, Dobransky T and Sobel A. (1993). J. Biol. Chem., Li L and Cohen SN. (1996). Cell, 85, 319 ± 329. 268, 20076 ± 20084. Li L, Li X, Francke U and Cohen SN. (1997). Cell, 88, 143 ± Chneiweiss H, Beretta L, Cordier J, Boutterin MC, 154. Glowinski J and Sobel A. (1989). J. Neurochem, 53, Mitchell PJ and Tjian R. (1989). Science, 245, 371 ± 378. 856 ± 863. Nakamura K, Fujimoto M, Tanaka T and Fujikura Y. DurajJ,KovacikovaM,SedlakJ,KoppelJ,SobelAand (1995). Electrophoresis, 16, 1530 ± 1535. Chorvath B. (1995). Leuk. Res., 19, 457 ± 461. Ohta M, Inoue H, Cotticelli MG, Kastury K, Baa R, Gussow D, Rein R, Ginjaar I, Hochstenbach F, Seemann G, Palazzo J, Siprashvili Z, Mori M, McCue P, Druck T and Kottman A and Ploegh HL. (1987). J. Immunol., 139, et al. (1996). Cell, 84, 587 ± 597. 3132 ± 3138. Parker SL, Tong T, Bolden S and Wingo PA. (1996). CA. Johnson WE, Jones NA, Rowlands DC, Williams A, Guest Cancer J. Clin., 46, 5±27. SS and Brown G. (1995). FEBS Lett., 3645, 309 ± 313. Sobel A. (1991). Trends Biochem. Sci., 16, 301 ± 305. Kramer A. (1996). Annu. Rev. Biochem., 65, 367 ± 409. Sozzi G, Veronese ML, Negrini M, Baa R, Cotticelli MG, Le SE, Rosenfeld MG and Evans RM. (1986). Ann. Rev. InoueH,TornielliS,PilottiS,DeGregorioL,Pastorino Biochem., 55, 1091 ± 1117. U, Pierotti MA, Ohta M, Huebner K and Croce CM. Leighton IA, Curmi P, Campbell DG, Cohen P and Sobel A. (1996). Cell, 85, 17 ± 26. (1993). Mol. Cell. Biochem., 127, 151 ± 156.