We have previously shown (K. Kas et al., Nat. Genet., 15: 170–174, 1997) that the developmentally regulated zinc finger gene pleomorphic adenoma gene 1 (PLAG1) is the target gene in 8q12 in pleomorphic adenomas of the salivary glands with t(3;8)(p21;q12) translocations. The t(3;8) results in promoter swapping between PLAG1 and the constitutively expressed gene for β-catenin (CTNNB1), leading to activation of PLAG1 expression and reduced expression of CTNNB1. Here we have studied the expression of PLAG1 by Northern blot analysis in 47 primary benign and malignant human tumors with or without cytogenetic abnormalities of 8q12. Overexpression of PLAG1 was found in 23 tumors (49%). Thirteen of 17 pleomorphic adenomas with a normal karyotype and 5 of 10 with 12q13–15 abnormalities overexpressed PLAG1, which demonstrates that PLAG1 activation is a frequent event in adenomas irrespective of karyotype. In contrast, PLAG1 was overexpressed in only 2 of 11 malignant salivary gland tumors analyzed, which suggests that, at least in salivary gland tumors, PLAG1 activation preferentially occurs in benign tumors. PLAG1 overexpression was also found in three of nine mesenchymal tumors, i.e., in two uterine leiomyomas and one leiomyosarcoma. RNase protection, rapid amplification of 5′-cDNA ends (5′-RACE), and reverse transcription-PCR analyses of five adenomas with a normal karyotype revealed fusion transcripts in three tumors. Nucleotide sequence analysis of these showed that they contained fusions between PLAG1 and CTNNB1 (one case) or PLAG1 and a novel fusion partner gene, i.e., the gene encoding the transcription elongation factor SII (two cases). The fusions occurred in the 5′ noncoding region of PLAG1, leading to exchange of regulatory control elements and, as a consequence, activation of PLAG1 gene expression. Because all of the cases had grossly normal karyotypes, the rearrangements must result from cryptic rearrangements. The results suggest that in addition to chromosomal translocations and cryptic rearrangements, PLAG1 may also be activated by mutations or indirect mechanisms. Our findings establish a conserved mechanism of PLAG1 activation in salivary gland tumors with and without 8q12 aberrations, which indicates that such activation is a frequent event in these tumors.
We have previously identified (1 , 2) a new, developmentally regulated zinc finger gene, designated PLAG1, 4 as the target gene in 8q12 in pleomorphic adenomas of the salivary glands with t(3;8)(p21;q12) translocations. The translocation results in promoter swapping between PLAG1 and the constitutively expressed gene for β-catenin (CTNNB1) in 3p21, leading to activation of PLAG1 expression and reduced expression of CTNNB1. The breakpoints invariably occur in the 5′ noncoding regions of both genes. The deduced PLAG1 protein contains seven NH2-terminal C2H2 zinc finger domains and a serine-rich COOH terminus, which acts as a transcriptional activator (3) . PLAG1 is developmentally regulated with expression mainly restricted to certain fetal tissues. β-catenin is a protein functioning as an interface in adherens junctions and in the WG/WNT signaling pathway (4 , 5) . β-catenin has also been implicated in tumorigenesis (4) .
Recently (6) , we identified a second translocation partner gene of PLAG1 in pleomorphic adenomas with a recurrent t(5;8)(p13;q12) translocation, namely LIFR. LIFR encodes the ubiquitously expressed receptor for the leukemia inhibitory factor (7) . The translocation results in up-regulation of PLAG1 gene expression under control of the LIFR promoter, i.e., a mechanism similar to that seen in adenomas with 3;8-translocations.
In addition to the above-mentioned subgroup of pleomorphic adenomas with abnormalities involving 8q12 (39% of the cases), there are at least three other cytogenetic subgroups that are characterized by (i) rearrangements of 12q13–15 (8% of the cases); (ii) sporadic clonal changes not involving 3p21, 8q12 or 12q13–15 (23% of the cases); and (iii) an apparently normal karyotype (30% of the cases; Refs. 8, 9, 10 ). The gene consistently rearranged in adenomas with 12q13–15 abnormalities is the high mobility group protein gene, HMGIC (11, 12, 13) . This gene is also rearranged in a variety of benign mesenchymal tumors with 12q13–15 abnormalities (11 , 14, 15, 16) .
Our previous studies of PLAG1 were restricted to pleomorphic adenomas with 8q12 abnormalities (1 , 6) . Here we present results showing that PLAG1 activation is not confined to adenomas with 8q12 abnormalities but is also found in tumors with normal karyotype and 12q13–15 abnormalities as well as in individual cases of malignant salivary gland tumors and mesenchymal tumors. In addition, we show that PLAG1 may also be activated by cryptic rearrangements in cases with normal karyotypes, leading to fusions between PLAG1 and CTNNB1 or a novel fusion partner gene.
MATERIALS AND METHODS
Tumor Material and Cytogenetic Analysis.
Fresh tumor tissue was obtained from patients at the time of surgery. Chromosome preparations were made from short-term primary cultures as described previously (17) . Forty-seven primary tumors were selected for molecular analysis, including 17 pleomorphic adenomas with a normal diploid karyotype, 10 pleomorphic adenomas with chromosome 12q13–15 aberrations, 11 malignant salivary gland tumors, 3 uterine leiomyomas, and 1 case each of soft-tissue chondroma, aggressive fibromatosis, infantile fibromatosis, fibrous dysplasia, malignant fibrous histiocytoma, and leiomyosarcoma. Karyotypic information was available from 40 of the 47 tumors. The diagnoses and relevant karyotypic data are presented in Table 1 ⇓ .
Preparation of RNA and Northern Blot Analysis.
Total RNA was extracted from frozen tumor samples using the TRIZOL (Life Technologies) method. Northern blot analysis was performed according to standard procedures (18) . Probes for filter hybridizations were radio-labeled with [α-32P]dCTP using the Megaprime DNA labeling system (Amersham). The PLAG1 probe used was a 3.7 kb EcoRI cDNA fragment consisting of 438 bp of the 5′-UTR, the complete coding region, and approximately 1800 bp of the 3′-UTR. This probe detects a 7.5-kb PLAG1 specific transcript. The actin probe used as an internal control was a 2.0-kb human β-actin cDNA probe (Clontech).
RNase Protection Assay.
The antisense riboprobes used for the RNase protection assay are schematically illustrated in Fig. 2 ⇓ . Preparation of probes and hybridization conditions were as described previously (6) . The protected riboprobes were purified by ethanol precipitation, resuspended in 80% (v/v) formamide, and analyzed by electrophoresis using a 5% polyacrylamide gel containing 7 m urea. End-labeled MspI fragments of pcDNA3 (Invitrogen) were used as single-strand molecular weight markers.
5′-RACE and Nucleotide Sequence Analysis.
5′-RACE was performed according to the protocol of the Marathon cDNA amplification kit (Clontech) with minor modifications. For first-strand cDNA synthesis, 5 μg of total RNA was used together with the PLAG1-specific cDNA synthesis primer MV2-low located in exon 5 of PLAG1 (Table 2) ⇓ . The double-stranded cDNA was then ligated to the cDNA adaptor and amplified using the adaptor primer AP1 and the PLAG1-specific MV5 primer also located in exon 5 of PLAG1 (Table 2) ⇓ . A second round of PCR was performed using the nested adaptor primer AP2 and the MV6 primer in exon 4 of PLAG1 (Table 2) ⇓ . The resulting PCR products were purified from agarose gels and cloned into the pCR2.1 vector (Invitrogen). Nucleotide sequence analysis was performed with an A.L.F. DNA sequencer (Pharmacia/LKB) using the T7 polymerase sequencing kit (Pharmacia/LKB). The resulting sequences were analyzed using Lasergene (DNASTAR) and basic local alignment search tool searches (National Center for Biotechnology Information).
For cDNA synthesis, 5 μg of total RNA was reverse-transcribed using the SuperScript Preamplification System according to the manufacturer′s manual (Life Technologies). An aliquot of 0.25 μg of the resulting first-strand cDNA was amplified using the appropriate primer sets. Primer sequences and annealing temperatures for all of the PCR primers used are shown in Table 2 ⇓ . All of the PCR amplifications were performed using the GeneAmp PCR system 9600 (Perkin-Elmer). For detection of the CTNNB1/PLAG1 fusion transcript, the first round of PCR was carried out with the CTNNB1 primer CAT-UP and the PLAG1 primer MV5. The second round of PCR was performed on a 20-fold diluted sample with the CTNNB1 primer NECAT-UP and the PLAG1 primer MV6. For detection of the reciprocal fusion transcript PLAG1/CTNNB1, the following nested primer sets were used: START-UP (PLAG1) and CAT3 (CTNNB1), and START-RACE (PLAG1) and CAT3NEST (CTNNB1). The SII/PLAG1 fusion transcript was detected after two rounds of PCR using the primer sets SII-UP (SII) and MV5 (PLAG1), and S2–764S (SII) and MV6 (PLAG1), respectively. For the reciprocal PLAG1/SII fusion transcript, the first round of PCR was performed using the SII primer S2-1105AS and the PLAG1 primer START-UP, and the second round was performed using the SII primer S2-972AS and the PLAG1 primer START-RACE. For detection of the LIFR/PLAG1 fusion transcript, the primer sets LIFR-NEUP (LIFR) and MV5 (PLAG1), and LIFR-CAU (LIFR) and MV6 (PLAG1) respectively, were used. For detection of the normal PLAG1 transcript, the first round of PCR was carried out with the PLAG1 primers START-UP and MV5 and the second round was carried out using the primers START-RACE and MV6. The normal SII transcript was detected using the primers SII-UP and S2-1105AS in the first round of PCR and S2-764S and S2-972AS in the second round. The normal CTNNB1 transcript was amplified using the primer sets: CAT-UP and CAT3, and NECAT-UP and CAT3NEST. The identities of the PCR products were confirmed by nucleotide sequence analysis. As control for intact RNA and cDNA, an RT-PCR reaction for expression of the housekeeping gene GAPDH was performed on all of the cDNAs used (19) .
Northern Blot Analysis of PLAG1 Gene Expression in Tumors.
The expression pattern of the PLAG1 gene in 47 benign and malignant salivary gland tumors and mesenchymal tumors was studied using Northern blot analysis (Table 1) ⇓ . In normal salivary gland tissue, PLAG1 expression was detectable only by RT-PCR and RNase protection analyses and not by conventional Northern blot analysis (Figs. 1 ⇓ and 2 ⇓ ). In tumors, however, PLAG1 overexpression was detected by Northern blot analysis in 49% (23 of 47) of the cases. In pleomorphic adenomas, PLAG1 overexpression was observed in 13 of 17 tumors with a normal karyotype (Fig. 1) ⇓ and in 5 of 10 tumors with rearrangements of 12q13–15. In contrast, only 2 of 11 malignant salivary gland tumors overexpressed PLAG1, i.e., 1 adenoid cystic carcinoma and 1 carcinoma ex pleomorphic adenoma. Among the mesenchymal tumors, two of three uterine leiomyomas and one leiomyosarcoma overexpressed PLAG1. The karyotypes of these tumors are not known. None of the four mesenchymal tumors with known cytogenetic aberrations affecting 8q11–13 overexpressed PLAG1.
RNase Protection Analysis of the PLAG1 Transcript in Five Tumors with a Normal Karyotype.
Five of the pleomorphic adenomas with a normal karyotype that overexpressed the 7.5-kb PLAG1 transcript (C954, C974, C1067, C1102, and CG568) were selected for further analysis using the RNase protection assay (Fig. 2) ⇓ . To detect the normal 7.5-kb PLAG1 transcript, we used a riboprobe corresponding to the noncoding exons 1 and 3 of PLAG1 (riboprobe b) because exon 2 is often missing as a result of alternative splicing. To detect the CTNNB1/PLAG1 fusion transcript, we used riboprobe c which contains exon 1 of CTNNB1 and exon 3 of PLAG1. In normal salivary gland tissue, a protected fragment of 207 nucleotides was obtained with riboprobe b (Lane 5), indicating the presence of a normal PLAG1 transcript containing exons 1 and 3. A protected fragment of 207 nucleotides was also detected in adenomas C974, C1102, and CG568 (Lanes 11, 17, and 20), which suggests that these tumors contain a normal PLAG1 transcript. In contrast, similar analysis of RNA from adenomas C954 and C1067 failed to detect this fragment (Lanes 8 and 14). Instead, protected fragments of 67 bp were found in both tumors suggestive of the presence of chimeric mRNAs containing ectopic sequences fused to exon 3 of PLAG1. In C1067 this chimeric mRNA corresponds to a CTNNB1/PLAG1 fusion transcript because riboprobe c gave a protected fragment of 219 bp (Lane 15) indicative of such a transcript. In adenoma C954, a CTNNB1/PLAG1 fusion transcript could not be detected (Lane 9), indicating that in this case, novel ectopic sequences are fused to exon 3 of PLAG1.
Identification of SII as a New Fusion Partner Gene of PLAG1.
Adenomas C954, C1067, and CG568 were selected for further characterization of the putative chimeric PLAG1 transcripts using 5′-RACE analysis. Nucleotide sequence analysis of PCR products from adenoma C954 confirmed the presence of ectopic sequences fused to exon 3 of PLAG1. Basic local alignment search tool search of these sequences showed that they were identical to those of the gene encoding the transcription elongation factor SII (GenBank accession number X73534). The ectopic sequences were also assigned to chromosome 3 by PCR-analysis of National Institute of General Medical Sciences monochromosome hybrid Mapping Panel 2 (data not shown). This assignment is in agreement with the published localization of SII to 3p21.3–p22 (20) . The fusion point in SII was at nucleotide position 910 resulting in a fusion of 5′ noncoding sequences as well as 63 nucleotides of the coding region to the acceptor splice site of PLAG1 exon 3 (Fig. 3A) ⇓ .
5′-RACE and nucleotide sequence analyses of PCR products from adenoma C1067 confirmed the findings of the RNase protection assay and showed that this tumor indeed contained a hybrid transcript consisting of exon 1 of CTNNB1 fused to exon 3 of PLAG1.
5′-RACE and nucleotide sequence analyses of amplified fragments from adenoma CG568 revealed that this tumor only expressed an apparently normal PLAG1 transcript and no fusion transcript.
RT-PCR Analysis of PLAG1 Gene Fusions.
Using RT-PCR, we also screened the 20 benign and malignant salivary gland tumors that expressed PLAG1 (Table 1) ⇓ to search for additional cases with hidden CTNNB1/PLAG1, SII/PLAG1, and LIFR/PLAG1 gene fusions. Amplification with primers specific for exon 1 of CTNNB1 and exon 4 of PLAG1 revealed CTNNB1/PLAG1 fusion transcripts only in adenoma C1067 and in control RNA from adenoma CG588, which has the classical t(3;8)(p21;q12) (1) . In both tumors, two products of 509 bp and 614 bp were observed, consistent with hybrid transcripts containing exon 1 of CTNNB1 fused to either exons 3 and 4 of PLAG1 or to exons 2 to 4 of PLAG1 (data not shown). The reciprocal PLAG1/CTNNB1 fusion transcript could only be detected in adenoma CG588. Amplification with primers specific for SII and PLAG1 revealed fusion transcripts of 557 bp and 662 bp, respectively, in two adenomas with normal karyotypes, C954 and C974 (Fig. 3B) ⇓ . Sequence analysis of these products showed that they consisted of hybrid transcripts containing 149 nucleotides (including 5′ noncoding sequences and 63 nucleotides of the coding region) of SII fused to either exons 3 and 4 of PLAG1 or to exons 2 to 4 of PLAG1. The reciprocal fusion transcript PLAG1/SII could only be detected in C954. Amplification with primers specific for the LIFR/PLAG1 fusion transcript resulted in two PCR products of 474 bp and 579 bp, respectively, only in control RNA from adenoma C895, which has a t(5;8)(p13;q12) with a known LIFR/PLAG1 fusion (6) , demonstrating that none of the tumors analyzed expressed this fusion transcript. The normal transcripts for PLAG1, CTNNB1, and SII were detected in C954, C974, and C1067 as well as in normal salivary gland tissue. A positive GAPDH RT-PCR, resulting in a 299 bp fragment, was obtained in all of the cases analyzed by RT-PCR.
PLAG1 was originally identified as the gene consistently rearranged in pleomorphic adenomas with chromosome translocations involving 8q12 (1 , 2 , 6) . In this paper we show that PLAG1 is activated not only in adenomas with 8q12 abnormalities but also in tumors with a normal karyotype and 12q13–15 abnormalities as well as in individual cases of malignant salivary gland tumors and smooth muscle tumors. In addition, we show that PLAG1 may also be activated by cryptic rearrangements in cases with normal karyotypes, leading to fusions between PLAG1 and CTNNB1 or a novel fusion partner gene SII.
Overexpression of PLAG1 as determined by Northern blot analysis was found in 76% of pleomorphic adenomas with a normal karyotype. In none of these cases was there any cytogenetic evidence of rearrangements of band 8q12, which indicates that PLAG1 in these cases is activated by mechanisms other than gross chromosomal changes. Further analysis of these tumors using RNase protection, 5′-RACE, and RT-PCR analyses revealed chimeric transcripts with ectopic sequences fused to exon 3 of PLAG1 in three tumors. One of the tumors contained a chimeric transcript resulting from a fusion of exon 1 of CTNNB1 to exon 3 of PLAG1, i.e., the same fusion as in tumors with the classical t(3;8)(p21;q12) (1) . In two other adenomas we could demonstrate that the SII gene is a novel and recurrent fusion partner of PLAG1. The fusion points in SII were in both cases at nucleotide position 910, resulting in a fusion of 5′ noncoding sequences and 63 nucleotides of the coding region to exon 3 of PLAG1. Because all three of the tumors had a normal karyotype, these fusions must result from cryptic rearrangements. In the tumor with the CTNNB1/PLAG1 fusion, this rearrangement was, however, too small to be detected by fluorescence in situ hybridization using yeast artificial chromosomes and cosmids containing these genes (data not shown). In all of the three tumors, the rearrangements led to up-regulation of PLAG1 gene expression under control of the CTNNB1 and SII promoters, respectively. Together with CTNNB1 (1) and LIFR (6) , SII is the third fusion partner gene known for PLAG1.
The SII gene (also known as TCEA1) encodes the transcription elongation factor SII (21 , 22) . The SII locus has previously been assigned to human chromosome segment 3p21.3–22 (20) , i.e., to the same region as CTNNB1. The gene encodes a ubiquitously expressed protein with a predicted molecular mass of 38 kDa (22 , 23) . SII belongs to the group of RNA polymerase II general elongation factors that are proteins involved in the regulation of the transcription of most, if not all, eukaryotic protein-coding genes (reviewed in Ref. 22 ). SII facilitates elongation of transcripts by preventing RNA polymerase II from terminating transcription prematurely at transcriptional blockages. The SII promoter contains two putative GC-box type and two CCAAT-box consensus sequences as well as an Alu sequence (21) . There are also several potential binding sites for transcription factors such as Sp1, MEP-1, TCF-2, and E2A.
The fusion points in SII were at nucleotide position 910 in both tumors. In the corresponding gene in Xenopus, which contains 10 exons and 9 introns, this position coincides with the location of the first intron (24) . Similarly, the related human gene TCEA2 has also an intron at this position (25) . The fact that the fusions occurred at the same nucleotide position in both tumors suggests that the breakpoints might have occurred in an intron. This is at variance with the findings of Park et al. (21) , who reported that the human SII gene lacks introns. Further analysis of the genomic organization of SII will be necessary to resolve this discrepancy. To the best of our knowledge, this is the first time SII has been implicated in human neoplasia.
The SII/PLAG1 fusions are the first examples where the breakpoints in a PLAG1 fusion partner gene interrupt the coding sequence (the breakpoints in CTNNB1 and LIFR invariably occur in the 5′ noncoding regions). The resulting fusion transcripts encode a truncated SII/PLAG1 protein of 90 amino acids as well as an intact PLAG1 protein (Fig. 3A) ⇓ . Because the coding region of PLAG1 is not interrupted, the consequences of the fusion is the same as for those involving CTNNB1 and LIFR, i.e., exchange of the PLAG1 promoter by an ubiquitously expressed promoter, leading to ectopic expression of PLAG1. Recently, a similar observation was made in B-cell non-Hodgkin’s lymphomas with t(3;6)(q27;p21), in which, as a result of the translocation, the entire H4 gene or part of the coding region including 5′ regulatory sequences replaced the 5′ noncoding region of the zinc finger gene BCL6, resulting in transcriptional deregulation of BCL6 (26) .
Our finding of a reciprocal fusion transcript between SII and PLAG1 in one tumor demonstrates that the activation of PLAG1 by promoter swapping occurs not only in tumors with the classical t(3;8)(p21;q12) (1) but also as a result of cryptic rearrangements in tumors with a normal karyotype. In the two other tumors with CTNNB1/PLAG1 and SII/PLAG1 fusions, no reciprocal transcripts could be detected, not even after multiple rounds of PCR, whereas the normal PLAG1, CTNNB1, and SII transcripts were detected in both tumors. The mechanisms in these two cases may, therefore, be classified as promoter substitution, resulting from a nonreciprocal rearrangement such as, for example, an insertion. Indeed, this was recently observed in a pleomorphic adenoma with normal karyotype, which was shown to have a hidden insertion of the 12q15 segment into 9p23, resulting in a fusion of the last coding exon of the NF1B gene to exons 1 to 4 of the HMGIC gene (13) .
In adenoma CG568, 5′-RACE analysis revealed an apparently normal PLAG1 transcript, suggesting that PLAG1 in this case may be activated by a mechanism other than gene fusion. In C974, the RNase protection analysis revealed protected fragments of 207 bp (corresponding to a normal PLAG1 transcript) and 67 bp (corresponding to exon 3 in the SII/PLAG1 fusion transcript), respectively, whereas in C954 and C1067, only the 67 bp fragments (corresponding to the SII/PLAG1 and CTNNB1/PLAG1 fusion transcripts) were observed. These observations raise the question of whether normal PLAG1 alleles independently or concomitantly with gene fusions may be activated by e.g., mutations as previously described for BCL6 (27, 28, 29) .
Overexpression of PLAG1 was also observed in 50% of the pleomorphic adenomas with 12q13–15 abnormalities. Previous studies have shown that the gene consistently rearranged in adenomas with 12q13–15 involvement is HMGIC (11, 12, 13) . A crucial question was, therefore, to find out whether PLAG1 and HMGIC may be affected in the same tumors. In two of the five tumors with PLAG1 activation, the status of HMGIC was known. Both tumors contained HMGIC/NF1B fusion transcripts (13) , demonstrating that, indeed, both genes may be affected in the same tumor. The frequency and role of such coexpression remains, however, to be determined. The fact that PLAG1 is affected not only in adenomas with 8q12 abnormalities but also in adenomas with 12q13–15 abnormalities and normal karyotype further emphasize the importance of PLAG1 in salivary gland tumorigenesis.
In contrast to the high frequency of overexpression of PLAG1 in benign salivary gland tumors, only 2 of 11 malignant salivary gland tumors analyzed overexpressed PLAG1, i.e., 1 carcinoma ex pleomorphic adenoma and 1 adenoid cystic carcinoma. In the former case, it is likely that PLAG1 was already activated in the benign pleomorphic adenoma before malignant transformation. In the latter case, we cannot rule out that this may also represent a carcinoma ex pleomorphic adenoma because it is known that adenoid cystic carcinomas may develop from the epithelial components of preexisting pleomorphic adenomas (30) . It should be pointed out that none of the two tumors with PLAG1 activation had any cytogenetic evidence of rearrangements of 8q12. These observations suggest that the activation of PLAG1 in salivary gland tumors is largely confined to benign pleomorphic adenomas and rarely occur in malignant salivary gland tumors, with the exception of carcinoma ex pleomorphic adenoma.
To find out whether PLAG1 activation is restricted to tumors of epithelial origin, we also analyzed a series of mesenchymal tumors. Three of 9 tumors analyzed overexpressed PLAG1. All of the three cases were smooth muscle tumors. Whether PLAG1 overexpression in mesenchymal tumors is limited to smooth muscle tumors remains, however, to be determined. There is an interesting similarity between pleomorphic adenomas and uterine leiomyomas in that both tumor types contain subgroups with rearrangements of HMGIC (11 , 14 , 15) . In the present cases, it was, however, not possible to determine whether HMGIC was affected, but it is plausible that in uterine leiomyomas also, both PLAG1 and HMGIC may be affected in the same tumors. Because most pleomorphic adenomas with 8q12 abnormalities overexpress PLAG1, it was of interest to see whether mesenchymal tumors with similar abnormalities also have activation of PLAG1. However, in none of the four cases analyzed did we find any evidence of PLAG1 overexpression, which indicates that PLAG1 is not the target gene in proximal 8q in these cases.
The results of this investigation clearly demonstrate that activation of PLAG1 is a frequent genetic event occurring in all of the major cytogenetic subgroups of pleomorphic adenomas as well as in certain mesenchymal tumors. In addition to activation by chromosomal translocations or cryptic rearrangements in cases with normal karyotypes, our findings indicate that PLAG1 may also be activated by other mechanisms such as mutations or indirect mechanisms. These mechanisms may operate independently in different tumors or concomitantly in the same tumor. In recent functional studies of the PLAG1 gene (3) , it was established that PLAG1 possess transcriptional activation capacity, raising the possibility that benign salivary gland tumors may originate because of the activation of particular target genes by ectopically overexpressed PLAG1.
We thank Rigmor Dahlenfors for excellent technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by the Swedish Cancer Society; the IngaBritt and Arne Lundbergs Research Foundation; the Assar Gabrielssons Research Foundation; the EC through Biomed 1 program “Molecular Cytogenetics of Solid Tumours”; the “Geconcerteerde Onderzoekacties 1997–2001”; the “Fonds voor Wetenschappelijk Onderzoek Vlaanderen”; and the “ASLK-programma voor Kankeronderzoek.” K. K. is a postdoctoral scholar of the Fonds voor Wetenschappelijk Onderzoek Vlaanderen.
↵2 Both authors contributed equally to this work.
↵3 To whom requests for reprints should be addressed, at Laboratory of Cancer Genetics, Department of Pathology, Göteborg University, Sahlgrenska University Hospital, SE-413 45 Göteborg, Sweden. Phone: 46-31-3422922; Fax: 46-31-820525; E-mail:
↵4 The abbreviations used are: PLAG1, pleomorphic adenoma gene 1; UTR, untranslated region; GAPDH, glyceraldehyde-3-phosphate dehydrogenase gene; RT-PCR, reverse transcription-PCR; 5′-RACE, rapid amplification of 5′-cDNA ends.
- Received September 11, 1998.
- Accepted December 14, 1998.
- ©1999 American Association for Cancer Research.