Identification of NFIB As Recurrent Translocation Partner Gene of HMGIC in Pleomorphic Adenomas

Identi®cation of NFIB as recurrent translocation partner gene of HMGIC in pleomorphic adenomas

Jan MW Geurts1, Eric FPM Schoenmakers1, Eva RoÈ ijer2, Anna-Karin AstroÈ m2,GoÈran Stenman2 and Wim JM van de Ven1

1Laboratory for Molecular Oncology, Center for Human Genetics, University of Leuven & Flanders Interuniversity Institute for Biotechnology, B-3000 Leuven, Belgium; 2Laboratory of Cancer Genetics, Department of Pathology, GoÈteborg University, Sahlgrenska University Hospital, S-41345 GoÈteborg, Sweden

Approximately 12% of all pleomorphic adenomas of the guished, i.e. tumors with rearrangements of 8q12 or salivary glands are characterized by chromosome 12q13 ± 15. Whereas the t(3;8)(p21;q12) is the most aberrations involving the chromosome segment 12q13 ± frequently observed translocation, many dierent 15. Several chromosomes have been found as transloca- chromosome segments can act as translocation tion partners of chromosome 12, and some of these partners of both 8q12 and 12q13 ± 15 (Sandros et al., recurrently. Recently, the HMGIC gene was identi®ed as 1990; Bullerdiek et al., 1993). The short arm of the target gene aected by the 12q13 ± 15 aberrations. chromosome 9, with breakpoints in the segment Here, we report the identi®cation and characterization of 9p12 ± 24, is the preferential translocation partner of a new translocation partner gene of HMGIC in chromosome 12q13 ± 15 (Stenman et al., 1994). The pleomorphic adenomas. 3'-RACE analysis of a primary reason for the variability of the breakpoints in 9p is adenoma with an apparently normal karyotype revealed likely to be due to the limited resolution of the banding an HMGIC fusion transcript containing ectopic se- pattern in 9p12 ± 24 and 12q13 ± 15, making it almost quences derived from the human NFIB gene, previously impossible to determine the exact locations of the mapped to chromosome band 9p24.1. The HMGIC breakpoints with certainty. NFIB fusion transcript was also con®rmed by RT ± Recently, HMGIC was identi®ed as the gene PCR. Since the chromosome segment 9p12 ± 24 is consistently aected in pleomorphic adenomas of the repeatedly involved as translocation partner of chromo- salivary glands and in a variety of other benign some 12q13 ± 15 in pleomorphic adenomas, we tested mesenchymal tumor types characterized by 12q13 ± 15 whether NFIB might be a recurrent partner of HMGIC. aberrations (Schoenmakers et al., 1995b; Ashar et al., RT ± PCR analysis of a second adenoma with an 1995). The location of the breakpoints with respect to ins(9;12)(p23;q12q15) as the sole anomaly, revealed that HMGIC is variable and occur 5' and 3' to HMGIC as also in this tumor an HMGIC/NFIB hybrid transcript well as recurrently in the very large, third intron of the was present. The reciprocal NFIB/HMGIC fusion gene, resulting in the separation of the three predicted transcript, however, could not be detected in any of DNA binding domains in the amino-terminal region these tumors. Nucleotide sequence analysis of the fusion from the acidic, carboxy-terminal domain of HMGIC. transcripts indicated that the genetic aberration in both HMGIC is a member of the high mobility group tumors resulted in the replacement of a carboxy-terminal (HMG) protein gene family which comprises proteins segment of HMGIC by the last ®ve amino acids of that are heterogeneous, nonhistone components of NFIB. In conclusion, our results reveal the recurrent chromatin (Grosschedl et al., 1994). HMGIC expres- involvement of the NFIB gene as translocation partner sion is tightly regulated, with the highest expression in gene of HMGIC in pleomorphic adenomas. fetal tissues and undetectable expression by Northern blot analysis of the normal full-length transcript in Keywords: NFIB; pleomorphic adenoma of the salivary normal adult tissues. HMGIC binds in the minor glands; HMGIC groove of DNA and has been suggested to act as an architectural factor (Wole, 1994) in the nuclear scaold (Saitoh and Laemmli, 1994; Tjian and Maniatis, 1994). The causal relationship between Introduction genetic aberrations aecting HMGIC and aberrant growth control in benign tumors remains to be Approximately 3 ± 6% of the tumors of the head and established. Of interest to note in this context is the neck region occur in the major and minor salivary observation that inactivation of both HMGIC alleles in glands (Shah and Ihde, 1990) and of these, the benign mice results in the so-called Pygmy phenotype with as pleomorphic adenoma is the most common tumor most important feature, disrupted growth leading to (Waldron, 1991). Previous studies have shown that dwar®sm (Zhou et al., 1995). These data demonstrate about 50 ± 80% of the pleomorphic adenomas display the important role of the HMGI proteins in clonal chromosomal abnormalities (Sandros et al., mammalian growth and development. 1990; Bullerdiek et al., 1993). Two major cytogenetic So far, four dierent fusion partner genes of subgroups with abnormal karyotypes can be distin- HMGIC have been identi®ed. The ®rst one is the LPP gene (Schoenmakers et al., 1995b; Ashar et al., Correspondence: WJM van de Ven 1995; Petit et al., 1996) which is located on Received 16 June 1997; revised 30 September 1997; accepted 30 chromosome 3q27 ± q28 and encodes a protein contain- September 1997 ing LIM domains. The second one is a postulated gene NFIB, recurrent translocation partner gene of HMGIC JMW Geurts et al 866 on chromosome 15q24 which has not yet been centromeric and telomeric of HMGIC, respectively. characterized in detail but which is predicted to Both YACs gave only signals on the two chromosomes encode a protein with a serine ± threonine-rich domain 12, demonstrating that the second chromosome 12 (Ashar et al., 1995). The third one is the mitochondrial breakpoint is distal to SAS ± CDK4. As expected from aldehyde dehydrogenase ALDH2 gene on chromosome cytogenetics, the SAS ± CDK4 signal in CG732 was 12q24.1, which was found as translocation partner gene transposed to 9p while the MDM2 signal remained on of HMGIC in a uterine leiomyoma (Kazmierczak et the der(12) (data not shown). al., 1995). In pleomorphic adenoma, the ®rst known fusion partner gene of HMGIC is the FHIT gene, Detection of an HMGIC fusion transcript which has been mapped to chromosome 3p14.2 (Geurts et al., 1997a). The pathogenetic relevance of the fusion In an attempt to determine whether the target gene on partners remains to be elucidated. The diversity in 12q is indeed the HMGIC gene, 3'-RACE analysis was chromosomal segments that have been found to performed. Analysis of adenoma CG575 resulted in the participate in translocations could indicate that the isolation of a 814 bp PCR product, 773 bp of which critical sequences are abundantly present in the human were deposited at GenBank (accession number genome. However, it has not yet been possible to AF022215), representing the 3'-RACE product devoid reliably identify a common structural or functional of non-speci®c cloning and PCR tails. Nucleotide denominator for the limited number of known sequence analysis of this product revealed that it translocation partner genes. contains 21 nucleotides of exon 4 of HMGIC Here, we present the results of 3'-RACE and RT ± (nucleotides 1073 ± 1093 in wild-type HMGIC acces- PCR experiments leading to the identi®cation and sion number U28749) and 752 nucleotides from the last subsequent characterization of a novel fusion partner exon of NFIB and part of its 3'-UTR (nucleotides gene of HMGIC in pleomorphic adenomas of the 1455 ± 2206 in wild-type NFIB as described in this salivary glands. manuscript and also available under accession number U85193). The sequence diversion point was thus found immediately downstream of HMGIC exon 4, suggesting a rearrangement within intron 4 of the HMGIC Results gene. As a result, the acidic tail of HMGIC was separated from the three DNA binding domains and Cytogenetic and FISH analysis the spacer domain. In phase with the HMGIC reading Conventional cytogenetic analysis of adenoma CG575 frame, the ectopic sequences appeared to contain a revealed an apparently normal karyotype (Figure 1a). stop codon after ®ve codons, adding only ®ve, ectopic However, detailed analysis of two metaphases with amino acid residues (SWYLG) to the truncated elongated chromosomes suggested that one of the two HMGIC protein. chromosome 12 homologs was somewhat shorter than the other (data not shown). The banding pattern of the Identi®cation of NFIB as fusion partner gene of HMGIC 12q-marker suggested an interstitial deletion of band 12q15. Hybridization with a chromosome 12 painting To establish the nature and origin of the ectopic probe did not reveal translocation of chromosome 12 sequences, genomic clones corresponding to these were material to any other chromosome. Adenoma CG732 isolated from a phage library using PCH203 as had a pseudodiploid karyotype with an ins(9;12) molecular probe. This probe was synthesized with the (p23;q12q15) as the sole anomaly (Figure 1a). PCR primer set PCH203-upper and -lower, which were Hybridization with painting probes for chromosomes developed on the basis of the ectopic sequences. Three 9 and 12 con®rmed the insertion of 12q material into 9p. clones were obtained which were characterized by To test whether both cases with 12q13 ± 15 restriction enzyme and partial nucleotide sequence abnormalities belonged to the group of pleomorphic analysis. A routine BLAST search by e-mail at adenomas with rearrangements of HMGIC we used [email protected] revealed that the ectopic se- two YAC clones, 341C1 and 452E1 (data not shown), quences, fused to HMGIC sequences, were 95% which contain the complete HMGIC gene. FISH identical to hamster clone pNFI ± Red1 which encodes analysis using these YACs revealed signals on the nuclear factor I (NFI) protein (Gil et al., 1988), normal chromosome 12 as well as on the der(9) and suggesting that the ectopic sequences were derived der(12) chromosomes, demonstrating that the HMGIC from the human NFIB gene. YACs spanned the 12q breakpoints in both tumors The chromosomal origin of the ectopic sequences (Figure 1b and c). The signals on the der(9) were in was determined by analysis of the NIGMS human ± both cases located close to pter in band 9p23. The rodent somatic hybrid mapping panel 2 (Coriell Cell FISH data thus show that CG575 has a hidden Repositories) (data not shown). From these studies, it ins(9;12). This case may therefore be equivalent to appeared that the ectopic sequences were derived from CG732 in which a larger segment of 12q is inserted chromosome 9. For subsequent ®ne mapping, CH203- into the same part of 9p. Collectively, the FISH data speci®c YAC clones 42G9, 100C6, and 246C4 were from both tumors raise the possibility that the target isolated from the CEPH mark 1 library and YAC gene on 12q is HMGIC and that the insertions aect clones 783H10, 800A7, 806F7, 912E9, and 924D12 the same gene on 9p. from the CEPH mark 3 library using the ectopic- To further characterize the 12q breakpoints in speci®c amplimer set PCH203 (Figure 2). By internet- adenoma CG575 we also hybridized with YACs mediated YAC query at http://www-genome.wi.mit.edu containing the closely linked genes SAS ± CDK4 /cgi-bin/contig/YAC info followed by STS information (YAC 751A4) and MDM2 (YAC 975D8) which map retrieval at the Human Genome Database, the YAC NFIB, recurrent translocation partner gene of HMGIC JMW Geurts et al 867

Figure 1 (a) Partial G-banded karyotypes of adenoma CG575 and CG732 showing the der(9) and der(12) markers (to the right) as well as their normal homologs (to the left). (b and c) FISH-analysis of tumor chromosomes from CG575 and CG732 using the HMGIC containing YAC 341C1 as probe. Co-hybridization of YAC 341C1 (green signal) and alpha satellite probes for chromosomes 9 and 12 (red signal) resulted in green hybridization signals on the normal chromosome 12, the der(12), and the der(9), demonstrating that this YAC spans the 12q breakpoint in both tumors. Note that in (b), the hybridization signal on the der(12) is weaker than on the normal chromosome 12. (d) FISH mapping of the 9p insertion breakpoint in adenoma CG732: Co-hybridization of the NFIB containing YAC 912E9 (green signal) with alpha satellite probes for chromosomes 9 and 12 (red signals) resulted in NFIB signals only on the normal 9 and the der(9). Note a split signal at the insertion breakpoint on the der(9). Chromosomes were counterstained in blue with 4,6-diamino-2-phenylindole (DAPI). Fluorescence signals were digitalized, processed and analysed using the ProbeMaster FISH image analysis system (Perceptive Scienti®c Instruments, Houston, Texas)

clones were found to map on the short arm of con®rm that in both cases there are breakpoints in the chromosome 9, at 9p23, i.e. the region preferentially NFIB containing YAC. aected in pleomorphic adenomas with 9p involve- Since only the incomplete open reading frame of the ment. This was also con®rmed by FISH using several human NFIB gene had been published (Qian et al., YACs, including YAC 912E9 which contains the entire 1995), overlapping cDNA clones were isolated to NFIB gene. FISH mapping of YAC 912E9 on obtain the complete coding region. In order to isolate metaphase chromosomes from both CG575 and these overlapping cDNA clones, two dierent ap- CG732 resulted in signals only on the chromosome 9 proaches were followed. As a ®rst approach, homologs. There was no evidence of insertion of sequences derived from the ends of the published chromosome 9 material into chromosome 12 in any incomplete reading frame were used as starting points of the tumors. In CG732, there was a split signal on for the isolation of overlapping clones by repeated the der(9) chromosome caused by the insertion of 12q BLAST searches. After each round of BLAST screen- sequences into 9p23 (Figure 1d), whereas in CG575 the ing, the ends of the composite contig constructed with insertion was at the border of visibility and was only the retrieved anonymous EST (Expressed Sequence possible to detect in three metaphases with elongated Tagged) clones were used as new starting points for chromosomes (data not shown). FISH analysis thus additional rounds of BLAST screening. As a second NFIB, recurrent translocation partner gene of HMGIC JMW Geurts et al 868 the phage genomic clones described above, it was established that the ectopic sequences (cf. PCH203) were derived from the last coding exon (LCE) of NFIB.

Expression pattern of the NFIB gene Commercially obtained, polyA+ multiple tissue North- ern blots were screened with the PCR probe PCH203. A single transcript of approximately 8.5 kb was detected in heart, colon, small intestine, ovary, prostate, and, at low levels, in peripheral blood leukocyte, testis, and spleen, whereas expression levels in pancreas, kidney, skeletal muscle, liver, lung, placenta, brain, and thymus were below the detection limit (Figure 4). Furthermore, no transcripts were detected that could represent naturally occurring, alternatively spliced mRNA isoforms. Since no specific NFIB signal could be detected on a salivary gland tissue derived total RNA Northern blot, more sensitive RT ± PCR analysis was performed to investigate the possible wild type NFIB expression in this tissue. These experiments revealed that the NFIB gene is also expressed in normal salivary gland tissue (data not shown). Figure 2 (a) STS content overview of ®ve CEPH mark 3 YACs containing 3'-UTR sequences of NFIB. YAC clone 912E9 contains the entire NFIB gene. (b) Long-range physical Diversity of HMGIC/NFIB fusion transcripts expressed map of a 230 kb genomic DNA region located at human in pleomorphic adenomas chromosome segment 9p24.1. Data have been deduced from a YAC contig consisting of three overlapping CEPH YAC To investigate whether NFIB and HMGIC are also clones. The long-range physical map of the composite genomic aected in adenomas with cytogenetically detectable DNA covered by the YAC inserts is represented by a solid line 9;12-rearrangements, we analysed a second adenoma, with the relative positions of the various restriction sites of rare-cutting enzymes indicated. Below the long-range physical CG732, with an ins(9;12)(p23;q12q15) by RT ± PCR. map, the sizes and relative positions of the overlapping YAC Using the appropriate HMGIC and NFIB-speci®c clones are represented by solid lines. Unique CEPH microtiter primers, RT ± PCR resulted in a 450 bp HMGIC/NFIB plate addresses of the YAC clones are listed. DNA fragments product in tumor CG575 (Figure 3b, lane 3) and a containing the 5'-UTR of NFIB or the last coding exon (LCE) 417 bp HMGIC/NFIB product in tumor CG732 (Figure of NFIB are indicated. PCH203 maps to the same DNA fragment as L.C.E restriction sites: B=BssHII; K=KspI; 3b, lane 7). No reciprocal, NFIB/HMGIC, fusion M=MluI; Sal=SalI; S®=S®I transcripts were detected in either of the tumors (Figure 3b, lanes 2 and 6). Nucleotide sequence analysis of both HMGIC/NFIB products revealed that in adenoma CG575, a fusion had occurred between the approach, an oligo-(dT)- and randomly primed cDNA fourth exon of HMGIC and the last coding exon of library (Clontech # HL1133b) was screened. Eight NFIB (Figure 3c). These results con®rm the data overlapping cDNA clones were retrieved, i.e. obtained in the 3'-RACE experiments. In adenoma HFBCP66, U07810, 130995, 133643, 133980, 145106, CG732, a fusion had occurred between the third exon of 178178, and C6A07. Screening of a cDNA library with HMGIC and the last coding exon of NFIB (Figure 3d). a probe derived from C6A07 resulted in the isolation The detection of HMGIC/NFIB transcripts in both of a cDNA (clone pCH294) and nucleotide sequence tumors raised the question as to whether normal analysis of it indicated that it contained the 5'-coding transcripts of these genes were also expressed. In part of the NFIB gene (®rst ®ve codons and ¯anking RT ± PCR experiments, using the appropriate primer 5'-UTR). Using the nucleotide sequence data of the sets (Figure 3), expression of normal HMGIC nine cDNA clones, a consensus composite NFIB transcripts could not be demonstrated in adenoma cDNA of about 2.5 kb could be established (Figure CG575 (Figure 3b, lane 1) whereas in tumor CG732, 3a). The cDNA appeared to contain an open reading an RT ± PCR product of 508 bp was detected, frame of 1260 nucleotides coding for a protein of 420 indicating that normal HMGIC transcripts are amino acid residues. It should be noted that the NFIB expressed (Figure 3b, lane 5). In both tumors, a transcript is about 8.5 kb in size but since only the 1340 bp RT ± PCR product was detected with the coding sequences are relevant to the content of the appropriate NFIB primers, suggesting expression of manuscript, characterization of the NFIB mRNA has normal NFIB transcripts (Figure 3b, lanes 4 and 8). been limited to only that region. The composite human Nucleotide sequence analysis of the coding regions of NFIB cDNA shows a high sequence homology to the the normal HMGIC and NFIB transcripts revealed that chicken NFIB2 splice variant (Rupp et al., 1990) and, the reading frames were devoid of mutations (data not possibly constitutes a similar splice variant of the shown). Since the HMGIC and NFIB genes were not human NFIB gene. The nucleotide sequence of the sequenced in their entirety, it remains to be established composite cDNA of NFIB has been deposited at whether changes have occurred in the noncoding GenBank (Accession number: #U85193). Using data of sequences. NFIB, recurrent translocation partner gene of HMGIC JMW Geurts et al 869 Discussion NFIB fusion transcripts that were detected in the two tumors studied revealed that in each case, the last In the present study, we have identi®ed NFIB as a coding exon of the NFIB gene was fused to sequences novel translocation partner gene of HMGIC in derived from HMGIC. In tumor CG732, the rearrange- pleomorphic adenomas with chromosome 12q13 ± 15 ment had occurred within the very large, third intron aberrations. The NFIB gene was mapped to chromo- of HMGIC, thus separating the three amino-terminal some band 9p23, a region known to be the most DNA-binding domains from the spacer domain and frequent translocation partner of chromosome 12q13 ± the carboxy-terminal acidic tail. In tumor CG575, 15 (Stenman et al., 1994). Analysis of the HMGIC/ however, the rearrangement in the HMGIC gene was in

b CG575 CG732

M12345678 a Size in bp

1200 — 800 — 540 460 — 320 —

Figure 3 (a) Nucleotide sequence of a composite cDNA of the NFIB transcript and the deduced amino acid sequence of the putative NFIB protein. The predicted amino acid sequence of the NFIB-encoded protein is shown in single-letter amino acid code. Numbering of the amino acid residues and the nucleotides is indicated at the end of each line. The stop codon is marked with an asterisk (*). (b) RT ± PCR analysis of pleomorphic adenomas CG575 and CG732. PCR products obtained in RT ± PCR analysis of total RNA corresponding to a normal HMGIC transcript (lanes 1 and 5; ®rst round PCR, primers 31970-004 [5- CCCAGCCCTATCACCTCA-3'] and 32627-001 [5'-AAGACCATGGCAATACAG-3']; second round PCR, primers 31970-005 [5'-CTCATCTCCCGAAAGGTG-3'] and 32627-001), a NFIB/HMGIC fusion transcript (lanes 2 and 6; ®rst round PCR, primers 41475-001 [5'-GCATCTCCCGGAAAGTGCGTT-3'] and 41475-002 [5'-AAAAGATCCAACTGCTGCTGAGGTAGAA-3']; second round PCR, primers 41475-001 and 41475-003 [5'-GTCCTCTTCGGCAGACTCTTGTGA-3'], a HMGIC/NFIB fusion transcript (lanes 3 and 7; ®rst round PCR, primers 41475-004 [5'-GGCGTCCCCAGCCCTATCAC-3'] and 41475-006 [5'-TGGA- CATTGGCGGGTAAGATGG-3'] second round PCR, primers 41475-004 and 41475-005 [5'-TGGCCGGTAA- GATGGGTGTCCT-3'], and a normal NFIB transcript (lanes 4 and 8; ®rst round PCR, primers 41475-001 and 41475-006; second round PCR, primers 41475-001 and 41475-005. Lane M: DNA molecular weight markers. (c) Nucleotide sequence of the HMGIC NFIB fusion transcript detected in tumor CG575. The ®rst four HMGIC exons are fused to the last coding exon of the NFIB gene. The relative positions of the various exons are indicated by arrows above the sequence and those of primers used in nested RT ± PCR reactions are indicated by arrows labelled with the primer number, both above and below the sequence. As shown the ®rst four exons of HMGIC are fused to the last exon of the NFIB gene. The stop codon in the HMGIC/NFIB fusion transcript is indicated by an asterisk. (d) Nucleotide sequence of the HMGIC/NFIB fusion transcript detected in tumor CG732. The relative positions of the various exons and primers are indicated as above. As shown, the ®rst three exons of the HMGIC gene are fused to the last coding exon of the NFIB gene. The stop codon in the HMGIC/NFIB fusion transcript is indicated by an asterisk NFIB, recurrent translocation partner gene of HMGIC JMW Geurts et al 870 (Schoenmakers et al., 1995b; Ashar et al., 1995; 12345678 9 10 11 12 13 14 15 16 Schoenberg Fejzo et al., 1996). kb The breakpoints within the NFIB gene were found to occur in the 3'-region of the gene for both tumors, i.e. in the intron preceding the last coding exon of — 9.5 — 7.5 NFIB. This exon encodes only for the last ®ve amino — 4.4 acids (SWYLG) of the transactivating domain of the NFIB protein. It has been reported (Roulet et al., — 2.4 1995), that these ®ve amino acids are critical for the proper functioning of the transcription factor. — 1.35 The observation that the NFIB gene is a recurrent translocation partner gene of HMGIC in pleomorphic — 2.4 adenomas with chromosome 12q13 ± 15 involvement and that in both tumors studied only HMGIC/NFIB — 1.35 fusion transcripts could be detected but not their Figure 4 (Upper panel) commercially obtained Northern blots reciprocal forms, may be indicative for a causal containing poly A+ selected mRNA were hybridized with the 105 bp PCH203 PCR probe. 1: pancreas; 2: kidney; 3: skeletal relationship between the corresponding fusion proteins muscle; 4: liver; 5: lung; 6: placenta; 7: brain; 8: heart; 9: and pleomorphic adenoma pathogenesis. Although peripheral blood leukocyte; 10: colon; 11: small intestine; 12: RT ± PCR results exclude a possible pathogenetical ovary; 13: testis; 14: prostate; 15: thymus; 16: spleen. Molecular role for NFIB/HMGIC fusion transcripts, one can not weight markers (0.24 ± 9.5 kb RNA ladder, GIBCO BRL exclude the possibility that an otherwise truncated #5620SA) are indicated. (Lower panel) b-actin hybridization; in muscle tissue, two naturally occurring forms of b-actin mRNA NFIB gene plays such a role in pleomorphic adenomas. are present with sizes of about 2 kb and 1.7 kb Recently Tkachenko et al. (1997) proposed a model for lipoma formation in which the absence of expression of wild type HMGIC depends on the gain of a transcriptional regulatory domain as a result of a intron 4, resulting in the separation of the acidic tail chromosomal rearrangement. Since the presence vs from the three amino-terminal DNA-binding domains absence of the HMGIC encoded spacer domain is the and the spacer domain. The fact that a HMGIC/NFIB only molecular dierence between the two tumor fusion gene is present and expressed in a benign tumor specimens characterized here, our observations con- indicates that the translocation is not sucient for full cerning wild type HMGIC expression within these oncogenesis. tumors only ®ts this model in case the spacer domain NFIB is a member of the human nuclear factor I contains such transcriptional regulatory domains. Since (NFI) gene family, which has been mapped to this putative function of the spacer domain seems chromosome 9p24.1 (Qian et al., 1995). Structurally, rather unlikely, we expect that the observed dierences members of this family contain DNA binding and in wild type HMGIC expression most likely re¯ects dimerization domains in their amino-terminal regions dierences in tumor sampling. and proline-rich, transactivating domains in their The contribution of the NFIB sequences to carboxy-termini. Human nuclear factor I (NFI) was pleomorphic adenoma pathogenesis is expected to be ®rst identi®ed and puri®ed as a 47 kDa stimulatory limited since only ®ve amino acid residues replace a protein required for ecient initiation of adenovirus major carboxy-terminal portion of HMGIC, i.e. the DNA replication (Nagata et al., 1982). It was later region including either the acidic domain or, alterna- shown to be involved in the transcriptional regulation of tively, the acidic domain and the spacer domain. Our a variety of viral and cellular genes. While the ®ndings are somewhat similar to those recently observed mechanism of transcriptional activation is still un- in a pleomorphic adenoma in which the FHIT gene known, NFI binding sites have been found in a wide (which maps to 3p14.2 (Ohta et al., 1996)) was found to variety of genes (Raymondjean et al., 1988; Inoue et al., be fused to HMGIC as a result of a complex t(3;10;12) 1990; Zorbas et al., 1992), and NFI genes are expressed (Geurts et al., 1997a). In this case, a reciprocal fusion in many dierent tissues (Cereghini et al., 1987; between HMGIC and the FHIT gene results in the Paonessa et al., 1988; Apt et al., 1994). At present, the exchange of 26 amino acids of the acidic tail of HMGIC NFI gene family consists of four members, i.e. NFIA, by 31 carboxy-terminal amino acids of the FHIT NFIB, NFIC, and NFIX. The human NFIA and NFIB protein, which are encoded by exon 9 of FHIT.A genes have been mapped to positions 1p31.2 ± p31.3 and similar observation has been made in a case of uterine 9p24.1, respectively, whereas the NFIC and NFIX genes leiomyoma, in which the mitochondrial aldehyde both map to 19p13.3 (Qian et al., 1995). dehydrogenase gene (ALDH2) acted as fusion partner In adenomas, breakpoints have thusfar been gene of HMGIC (Kazmierczak et al., 1995). Since no assigned either upstream of HMGIC (Ad-302 SV40), common denominator can be identi®ed in the coding within intron 3 (Ad-211 SV40, Ad-295 SV40 and sequences of the fusion partner genes (FHIT, NFIB, and CG592), within intron 4 (CG575), within the 3'-UTR ALDH2), it is tempting to speculate that the importance (Ad-312 SV40), or downstream of HMGIC (Ad-263 of the fused sequences is limited and merely restricted to SV40 and Ad-386 SV40) by FISH and/or 3'-RACE providing the necessary elements for proper translation experiments (Schoenmakers et al., 1995b; Geurts et al., of the hybrid transcripts, e.g. a translational stop codon 1997a). The alternative locations for the chromosome and a polyadenylation signal. Based on these results, breakpoints within the HMGIC gene found in the two one could argue that the separation of the DNA binding adenomas studied are consistent with previous studies domains from the carboxy-terminal acidic domain of of benign tumors with chromosome 12 aberrations HMGIC is the critical event. Fusion of the DNA NFIB, recurrent translocation partner gene of HMGIC JMW Geurts et al 871 binding domains of the HMGIC protein to well-de®ned CCG TCG ACA TC(T)17 was used. For both initial and functional domains, as described for instance for secondary rounds of PCR, the universal ampli®cation lipomas with t(3;12) (Petit et al., 1996), may have to primer (UAP2; 95L736) CUACUACUACUAAAG- be explained in this manner as well. On the other hand, GATCCGTCGACATC was used as `reverse primer'. In however, one should not exclude the possibility of the ®rst PCR round the following HMGIC-speci®c `forward primers' were used: (i), 94N1501) 5'-CTTCA- alternative scenarios. Recently, we characterized a GCCCAGGGACAAC-3' (exon 1); (ii), 94N701) 5'-CAA- pleomorphic adenoma cell line in which the HMGIC GAGGCAGACCTAGGA-3' (exon 3); or (iii), 94M1892) rearrangement occurred within the 3' noncoding region 5'-AACAATGCAACTTTTAATTACTG-3' (3'-UTR). In of the gene, leaving the complete coding region of the second round the following HMGIC-speci®c forward HMGIC intact (Geurts et al., 1997b). In conclusion, the primers (nested primers as compared to those used in the identi®cation of the NFIB gene as recurrent transloca- ®rst round) were used: (i), 94N1502) 5'-CAUCAUCAUCA- tion partner gene of HMGIC in adenomas with UCGCCTCAGAAGAGAGGAC-3' (exon 1); (ii), 94N700) involvement of chromosomes 9 and 12 further 5' - CAUCAUCAUCAUGT TCAGAAGAAGC CTGCT - 3' establishes the diversity in translocation partner (exon 4); or (iii), 94M1908) 5'-CAUCAUCAUCAUTT- sequences of HMGIC in benign tumors and adds GATCTGATAAGCAAGAGTGGG-3' (3'-UTR). CUA ± CAU-tailing of the nested, speci®c primers allowed the use support to the hypothesis that the HMGIC gene plays of the directional CloneAmp cloning system (GIBCO BRL). a central role in benign tumor development.

RT ± PCR analysis RT ± PCR (reverse transcriptase reaction combined with Materials and methods polymerase chain reaction) was performed following the procedure described by Schoenmakers et al. (1995b). Tumor material and cytogenetic and FISH analyses Brie¯y,cDNAwassynthesizedfrom5mg of total RNA. Short-term, primary cultures were established from two Reverse transcription was performed according to the ®rst primary pleomorphic adenomas, CG575 and CG732, as strand cDNA synthesis protocol from GIBCO BRL using previously described (Nordkvist et al., 1994). Cells from NFIB-speci®c cDNA synthesis primer (35978 ± 001) 5'- primary cultures were harvested after Colcemid exposure AGTGCTGCAATTGCTGGTCT-3' and HMGIC-speci®c followed by hypotonic treatment and ®xation in metha- cDNA synthesis primer (95C3362) 5' TACAG- nol:acetic acid. Slides were subsequently G-banded and CAGTTTTTCACTA-3'. The primers used in the various analysed according to the recommendations of the ISCN experiments are listed in the legend to Figure 3. The RT ± (1995). FISH analysis of metaphase chromosomes was PCR products were directly subcloned into the pCR II performed as previously described (RoÈ ijer et al., 1996). vector (TA-cloning kit, Invitrogen #K2000-40) and subsequently used for sequence analysis. The following PCR reactions have been performed for each tumor: two Isolation of genomic phage and YAC clones PCR reactions to detect expression of normal HMGIC, Phage clones were isolated from a genomic library HMGIC NFIB, NFIB HMGIC, and normal NFIB (in total constructed with DNA from cell line LM-30.1/SV40 eight PCR reactions, ®rst and second round of PCR). (Schoenmakers et al., 1995a) using lambdaFIX II as Three subclones of each reaction were sequenced. vector according to standard procedures. YAC clones were isolated from the CEPH human genomic YAC PCR ampli®cation libraries mark 1 and 3 using a combination of PCR-based screening and colony hybridization analysis as previously All PCR ampli®cations were carried out in ®nal volumes of described (Schoenmakers et al., 1995a). YAC clones are 100 ml containing 10 mM Tris-HCl pH 8.3, 50 mM KCl, indicated by their unique microtiter plate addresses. The 1.5 mM MgCl2, 0.01% gelatine, 0.2 mM dNTP's, 20 pmole experimental procedures for isolation and subsequent of each amplimer, 2.5 units of AmpliTaq (Perkin-Elmer characterization of the YAC clones by pulsed-®eld gel Cetus), and template DNA, using a Pharmacia ± LKB Gene electrophoresis (Chu et al., 1986), Southern blot, hybridi- ATAQ controller or a Perkin Elmer DNA thermal cycler. zation and restriction mapping analysis, were exactly as For isolation of YACs, 100 ng (for YAC superpools) or described by Schoenmakers et al. (1994). 20 ng (for YAC pools) of template DNA was used together with 20 pmole of each amplimer (PCH203-upper (5'- GGGAAAAAGAAACACAACAA-3') and PCH203-lower RNA isolation and Northern blot analysis (5'-TCAGTTTCTT-TCCTTTCTGC-3')). After initial de- Total RNA was isolated from primary adenomas CG575 naturation for 5 min at 948C, 35 ampli®cation cycles were and CG732 using TRIzol LS Reagent (GIBCO BRL) performed each consisting of denaturation for 1 min at according to the supplier's instructions. Commercially 948C, annealing for 1 min at 608C and extension for 1 min obtained Northern blots (Clontech #7759 ± 1 and #7760 ± at 728C. The PCR reaction was completed by a ®nal 1) of poly A+ selected RNA from dierent human tissues extension at 728C for 5 min. The results were evalulated by (pancreas, kidney, skeletal muscle, liver, lung, placenta, analysis of 10 ml of the reaction product on 6% brain, heart, peripheral blood leukocytes, colon, small polyacrylamide minigels. intestine, ovary, testis, prostate, thymus, spleen) were screened with the NFIB-speci®c probe PCH203 according Nucleotide sequence analysis to the supplier's instructions. For normalization, human b- actincDNAwasusedasaprobe. Nucleotide sequences were determined essentially according to the dideoxy chain termination method (Sanger et al., 1977), using a T7 polymerase sequencing kit (Pharmacia 3'-RACE analysis LKB) or the dsDNA Cycle Sequencing System (GIBCO Rapid ampli®cation of 3' cDNA-ends (3'-RACE) was BRL). Sequencing results were obtained using an Auto- performed using the 3'-exon trapping protocol of GIBCO mated Laser Fluorescent (ALF) DNA sequencer (Phar- BRL with slight modi®cations. For ®rst strand cDNA macia Biotech) and standard 30 cm, 6% Hydrolink, Long synthesis, adapter primer (AP2; 94M2363) AAG GAT Range gels (AT Biochem). The nucleotide sequences were NFIB, recurrent translocation partner gene of HMGIC JMW Geurts et al 872 analysed using the sequence analysis software Genepro YAC clones. This work was supported in part by the (Riverside Scienti®c) and Lasergene (DNASTAR, Inc.). All `Fonds voor Wetenschappelijk Onderzoek Vlaanderen' oligonucleotides were purchased from Pharmacia Biotech. (FWO), by the `Geconcerteerde Onderzoekacties 1997 ± 2001', the `ASLK Programma voor Kankeronderzoek', the Swedish Cancer Society, the Assar Gabrielssons Research Acknowledgements Foundation, and the IngaBritt and Arne Lundbergs The authors thank M Behrendt for cytogenetic analysis Foundation. Jan MW Geurts is a research assistant of and C Huysmans for isolation and characterization of the the FWO (Kom op tegen Kanker, Belgium).

References

Apt D, Liu Y and Bernard H-U. (1994). Nucleic Acids Res., Rupp RAW, Kruse U, Multhaup G, GoÈ bel U, Beyreuther K 22, 3825 ± 3833. and Sippel AE. (1990). Nucleic Acids Res., 18, 2607 ± 2616. Ashar HR, Schoenberg Fejzo M, Tkachenko A, Zhou X, RoÈ ijer E, Kas K, Klawitz I, Bullerdiek J, van de Ven W and Fletcher JA, Weremowicz S, Morton CC and Chada K. Stenman G. (1996). Genes Chromosom. Cancer, 17, 166 ± (1995). Cell, 82, 57 ± 65. 171. Bullerdiek J, Wobst G, Meyer-Bolte K, Chilla R, Haubrich J, Saitoh Y and Laemmli UK. (1994). Cell, 76, 609 ± 622. Thode B and Bartnitzke S. (1993). Cancer Genet. Sandros J, Stenman G and Mark J. (1990). Cancer Genet. Cytogenet., 65, 27 ± 31. Cytogenet., 44, 153 ± 167. Cereghini S, Raymondjean M, Carranca AG, Herbomel P Sanger F, Nicklen S and Couslon A. (1977). Proc. Natl. and Yaniv M. (1987). Cell, 50, 627 ± 638. Acad. Sci. USA, 74, 5463 ± 5467. Chu G, Vollrath D and Davis RW. (1986). Science, 234, Schoenberg Fejzo M, Ashar HR, Krauter KS, Powell WL, 1582 ± 1585. Rein MS, Weremowicz S, Yoon SJ, Kucherlapati RS, Geurts JMW, Schoenmakers EFPM, RoÈ ijer E, Stenman G Chada K and Morton CC. (1996). Genes Chromosom. and van de Ven WJM. (1997a). Cancer Res., 57, 13 ± 17. Cancer, 17, 1±6. Geurts JMW, Schoenmakers EFPM and van de Ven WJM. Schoenmakers EFPM, Kools PFJ, Mols R, Kazmierczak B, (1997b). Cancer Genet. Cytogenet., 95, 198 ± 205. Bartnitzke S, Bullerdiek J, Dal Cin P, De Jong PJ, van den Gil G, Smith JR, Goldstein JL, Slaughter CA, Orth K, Berghe H and van de Ven WJM. (1994). Genomics, 20, Brown MS and Osborne TF. (1988). Proc. Natl. Acad. Sci. 210 ± 222. USA, 85, 8963 ± 8967. Schoenmakers EFPM, Geurts JMW, Kools PFJ, Mols R, Grosschedl R, Giese K and Pagel J. (1994). Trends Genet., Huysmans C, Bullerdiek J, van den Berghe H and van de 10, 94 ± 100. Ven WJM. (1995a). Genomics, 29, 665 ± 678. Inoue T, Tamura T, Furuichi T and Mikoshiba K. (1990). J. Schoenmakers EFPM, Wanschura S, Mols R, Bullerdiek J, Biol. Chem., 265, 19065 ± 19070. van den Berghe H and van de Ven WJM. (1995b). Nature Kazmierczak B, Hennig Y, Wanschura S, Rogalla P, Genetics, 10, 436 ± 444. Bartnitzke S, van de Ven WJM and Bullerdiek J. (1995). Shah JP and Ihde JK. (1990). Curr. Probl. Surg., 27, 779 ± Cancer Res., 55, 6038 ± 6039. 843. Nagata K, Guggenheimer RA, Enomoto T, Lichy JH and Stenman G, Bullerdiek J, Bartnitzke S, Sahlin P, RoÈ ijer E Hurwitz J. (1982). Proc. Natl. Acad. Sci. USA, 21, 6438 ± and Mark J. (1994). In: Chromosome 12 Aberrations in 6442. Human Solid Tumors: Cytogenetics and Molecular Nordkvist A, Mark J, Gustavsson H, Bang G and Stenman Genetics. Eds. J Bullerdiek and S Bartnitzke. Springer G. (1994). Genes Chromosom. Cancer, 10, 115 ± 121. Verlag, Berlin, pp 3 ± 15. Ohta M, Inoue H, Cotticelli MG, Kastury K, Baa R, Tijan R and Maniatis T. (1994). Cell, 77, 5±8. Palazzo J, Siprashvili Z, Mori M, McCue P, Druck T, Tkachenko A, Ashar H, Meloni AM, Sandberg AA and Croce CM and Huebner K. (1996). Cell, 84, 587 ± 597. Chada K. (1997). Cancer Res., 57, 2276 ± 2280. Paonessa G, Gounari F, Frank R and Cortese R. (1988). Waldron CA. (1991). Adenoid cystic carcinoma, In Ellis GL, EMBO J., 7, 3115 ± 3123. Auclair PL and Gnopp DR (eds): Surgical Pathology of Petit MMR, Mols R, Schoenmakers EFPM, Mandahl N and the Salivary Glands. Philadelphia: W.B. Saunders van de Ven WJM. (1996). Genomics, 36, 118 ± 129. Company, pp 165 ± 186. Qian F, Kruse U, Lichter P and Sippel AE. (1995). Genomics, Wole AP. (1994). Science, 264, 1100 ± 1101. 28, 66 ± 73. Zhou X, Benson KF, Ashar HR and Chada K. (1995). Raymondjean M, Cereghini S and Yaniv M. (1988). Proc. Nature, 376, 771 ± 774. Natl. Acad. Sci. USA, 85, 757 ± 761. Zorbas H, Rein T, Krause A, Homann K and Winnacker E- Roulet E, Armentero M-T, Krey G, CortheÂ sy B, Dreyer C, L. (1992). J. Biol. Chem., 267, 8478 ± 8484. Mermod N and Wahli W. (1995). Mol. Cell. Biol., 15, 5552 ± 5562.