Pleomorphic adenoma gene 1 (PLAG1) proto-oncogene overexpression is implicated in various human neoplasias, including salivary gland pleomorphic adenomas. To further assess the oncogenic capacity of PLAG1, two independent PLAG1 transgenic mouse strains were established, PTMS1 and PTMS2, in which activation of PLAG1 overexpression is Cre mediated. Crossbreeding of PTMS1 or PTMS2 mice with MMTV-Cre transgenic mice was done to target PLAG1 overexpression to salivary and mammary glands, in the P1-Mcre/P2-Mcre offspring. With a prevalence of 100% and 6%, respectively, P1-Mcre and P2-Mcre mice developed salivary gland tumors displaying various pleomorphic adenoma features. Moreover, histopathologic analysis of salivary glands of 1-week-old P1-Mcre mice pointed at early tumoral stages in epithelial structures. Malignant characteristics in the salivary gland tumors and frequent lung metastases were found in older tumor-bearing mice. PLAG1 overexpression was shown in all tumors, including early tumoral stages. The tumors revealed an up-regulation of the expression of two distinct, imprinted gene clusters (i.e., Igf2/H19 and Dlk1/Gtl2). With a latency period of about 1 year, 8% of the P2-Mcre mice developed mammary gland tumors displaying similar histopathologic features as the salivary gland tumors. In conclusion, our results establish the strong and apparently direct in vivo tumorigenic capacity of PLAG1 and indicate that the transgenic mice constitute a valuable model for pleomorphic salivary gland tumorigenesis and potentially for other glands as well.
- PLAG1 proto-oncogene
- transgenic mouse strains
- salivary and mammary gland tumorigenesis
- pleomorphic adenoma
- imprinted gene cluster
Pleomorphic adenoma gene 1 (PLAG1) is a proto-oncogene on human chromosome 8q12 whose oncogenic activation is a crucial event in the formation of pleomorphic adenomas of the salivary glands ( 1, 2) and lipoblastomas ( 3– 5). Recently, amplification and overexpression of PLAG1 was reported for hepatoblastoma, indicating that the PLAG1 gene might also be implicated in the molecular pathogenesis of this childhood neoplasia ( 6). Altogether, these observations seem to emphasize a more general importance of the PLAG1 gene in human benign solid tumor development. Recent studies have shown that PLAG1 and PLAGL2 seem also to be implicated in leukemia ( 7). Indeed, PLAG1 and PLAGL2 expression was increased in 20% of human acute myeloid leukemia (AML) samples, that contain in most cases the CBFB-MYH11 fusion gene ( 8, 9).
PLAG1 encodes for a zinc finger transcription factor and has two structurally related family members, PLAGL1 (PLAG-like 1) and PLAGL2 ( 10). The main mechanism of oncogenic activation of PLAG1 involves recurrent chromosome translocations. This leads to promoter exchange between PLAG1, a gene differentially expressed and primarily during fetal development and one of a variety of possible translocation partner genes, that are more broadly and constitutively expressed. Chromosome breakpoints invariably occur in the 5′ noncoding region of PLAG1, causing an exchange of regulatory control elements without affecting the coding sequences of the gene. This process, also called promoter swapping, leads to ectopic expression of PLAG1 in the genetically affected cells ( 1, 2). PLAG1 activation can also be caused by cryptic rearrangements involving the gene region ( 11). It is assumed that ectopic PLAG1 overexpression causes deregulation of expression of PLAG1 target genes leading to particular tumors in humans ( 1, 3).
Structurally, the PLAG1 transcription factor contains seven canonical C2H2 zinc fingers and a serine-rich COOH terminus with transactivation capacity ( 10). PLAG1 specifically recognizes a bipartite DNA-binding consensus sequence consisting of a core sequence, GRGGC, and a G cluster, GGG, which are separated by 6 to 8 nucleotides ( 12). A nuclear localization sequence was identified ( 13) and PLAG1 is post-translationally modified by SUMOylation ( 14). In recent microarray analyses ( 15), genes were identified that are consistently induced or repressed by PLAG1, and these were classified into various functional categories. Among the classes of up-regulated PLAG1 target genes, the one encoding growth factors was the largest and included the genes for insulin-like growth factor II (IGF-II), cytokine-like factor-1 (CLF-1), bone-derived growth factor (BPGF-1), choriogonadotropin β chain (CGB), vascular endothelial growth factor (VEGF), and placental growth factor (PIGF). From in silico evaluation of their promoter regions, it seemed that a large proportion of them harbor several copies of the specific bipartite DNA binding consensus sequence, suggesting that they constitute direct PLAG1 targets. Furthermore, the in silico studies indicated that PLAG1–down-regulated genes are likely to be regulated indirectly by PLAG1, because the typical PLAG1 DNA binding consensus sequences are not present in their corresponding promoter regions ( 15).
The oncogenic potential of PLAG1 has been established in vitro by transforming cultured cells via overexpression of PLAG1 ( 16). The results of these in vitro studies are consistent with the in vivo observation that increased PLAG1 expression is important in the development of particular tumors, such as those mentioned before. To explore the role of PLAG1 in neoplastic transformation in the context of a complex organism, we have initially generated transgenic mice overexpressing PLAG1 early and ubiquitously. 6 These studies revealed that such early and ubiquitous overexpression of PLAG1 leads to embryonic lethality. Now, we have developed PLAG1 transgenic mouse strains, in which activation of overexpression of the transgene as well as the tissue distribution of such overexpression can be manipulated (e.g., by Cre-mediated activation and targeted expression, respectively). Two independent PLAG1 transgenic mouse strains were obtained and used to target overexpression of the PLAG1 gene to a restricted number of tissues, including salivary gland and mammary gland tissue. PLAG1 overexpression in the salivary glands leads to tumor formation. These PLAG1-induced salivary gland neoplasias were studied in detail histopathologically and molecularly. Finally, induction of mammary gland tumors, with similar morphologic features, was also studied.
Materials and Methods
Generation of transgenic mouse strains with Cre-mediated activation of PLAG1 expression. A KpnI/XhoI cDNA fragment, comprising the complete open reading frame of human PLAG1 followed by the coding sequences for an HA-tag, was cloned into the EcoRI site of the pCAGGS vector DNA ( 17) using blunt end ligation ( Fig. 1A ). At the multiple cloning site between the CAG promoter and the sequences encoding HA-tagged PLAG1, a loxP/PGK-Neo/loxP DNA fragment (a kind gift of P. Kastner, Institut de Chimie Biologique, Université Louis Pasteur, Strasbourg, France; ref. 18) was inserted as a Cre-dependent removable stop cassette ( Fig. 1A). This resulted in a DNA construct conditionally allowing targeted activation of PLAG1 expression. The resulting DNA was digested with restriction endonucleases SalI and SfiI, and upon size fractionation by electrophoresis, a 6.0-kb SalI/SfiI DNA fragment ( Fig. 1A) was purified using the QIAquick extraction kit (Qiagen, Chatsworth, CA). This purified DNA was used for microinjection into the male pronucleus of fertilized eggs of (FVB × FVB) mice (Charles River Laboratories, Brussels, Belgium). Upon oviductal implantation of these eggs in pseudopregnant foster females, offspring was obtained and these mice were bred to select PLAG1 transgenic founders. To target PLAG1 expression to the salivary gland and mammary gland, these founders were crossed with B6129-Tgn(MMTV-LTR/Cre)1Mam transgenic mice (The Jackson Laboratories, Bar Harbor, ME; ref. 19).
Genotyping by PCR analysis. Genotyping of candidate PLAG1 founders was done by PCR analysis of tail DNA using oligonucleotide primers POS-1599 (5′-TTCTCAAGCATCGTCATCAT-3′) and β-globin (5′-AAAATTCCAACACACTATTGC-3′) at an annealing temperature of 58°C. PCR analysis was also used to evaluate Cre-mediated excision of the floxed stop cassette from the PLAG1 DNA construct in the transgenic mouse strains. For that purpose, genomic DNA was isolated from various tissues of transgenic mice using the DNeasy tissue kit (Qiagen) according to the manufacturer's instructions. PCR analysis to detect Cre-mediated excision of the floxed stop cassette was done using the following primers: P1 forward 5′-CTACAGCTCCTGGGCAA-CGTGCTGG-3′ or P2 forward 5′-GCCTGAAGAACGAGATCAGCAGCC-3′ in combination with P2 reverse 5′-CACCACTTGTTGCGGCATGCAAGGCC-3′. The P1 forward/P2 reverse and P2 forward/P2 reverse set of primers amplified, respectively, a DNA fragment of 500 or 600 bp. The selected annealing temperature was 65°C.
DNA transfection, SDS-PAGE, and Western blot analysis. Selected DNAs (1 μg) were (co)transfected into appropriate mammalian cells using FuGENE 6 transfection reagent (Boehringer Mannheim, Mannheim, Germany) according to the protocol of the manufacturer. About 24 hours after transfection, cells were lysed in SDS-PAGE sample buffer [50 mmol/L Tris-HCl (pH 6.8), 10% glycerol, 2% SDS, 4% β-mercaptoethanol], boiled, and loaded on a 10% SDS-polyacrylamide gel. After size separation, proteins were transferred electrophoretically to PROTEAN nitrocellulose membranes (Schleicher and Schuell, Keene, NH). Expression of HA-tagged PLAG1 was detected using mouse anti-HA monoclonal antibody (1:5,000, Roche, Nutley, NJ). Enhanced chemiluminescence Western blotting was done using Renaissance Western blotting detection reagents (Perkin-Elmer Life Sciences, Norwalk, CT; ref. 20).
Histopathology, immunohistochemistry, and immunofluorescence. For histopathology, organs and tumors of mice were removed carefully, fixed overnight in 4% paraformaldehyde, and embedded in paraffin using routine procedures. For each tumor, two to three different areas were selected to analyse histologically. Paraffin sections (5 μm) were stained with H&E, Alcian Blue (pH 0.2, 1, or 2.5), or periodic acid-Schiff (PAS).
Immunostainings were done on 5 μm paraffin sections according to standard procedures. Briefly, after antigen retrieval in citrate buffer (10 mmol/L citric acid, pH 6), the mouse anti-human smooth muscle actin (SMA; 1:500 DAKO, Carpinteria, CA), and the guinea pig anti-human cytokeratin ( 8– 18) antibody (1:50, Progen, Heidelberg, Germany) were used as primary antibodies. Such antibodies are frequently used to stain myoepithelial and epithelial cells, respectively ( 21). As a secondary reagent to detect smooth muscle actin, the avidin-biotin complex system was used. For detection of cytokeratin 8/18, a secondary rabbit anti-guinea pig antibody (DAKO) was used, followed by the application of the anti-rabbit Envision+ kit (DAKO) and development with 3,3′-diaminobenzidine (Sigma FastTM DAB Tablet Set, Sigma, St. Louis, MO). Proliferation was assessed in salivary gland tumors by bromodeoxyuridine (BrdUrd) incorporation. Briefly, mice received a single i.p. injection of BrdUrd (cell proliferation labeling reagent, 1 ml/100 g, GE Healthcare, Roosendaal, the Netherlands) 1 hour before sacrification. Paraffin-embedded salivary gland samples were subjected to antigen retrieval in citrate buffer (10 mmol/L citric acid, pH 6) and immunohistochemistry using anti-BrdUrd-POD Fab (1/100, Roche). Tyramide signal amplification was obtained with the TSATM Biotin System (Perkin-Elmer Life Sciences) and STP-POD (1/100, DAKO).
Immunofluorescence analysis was used to assay PLAG1 protein expression and nuclear staining [4′,6-diamidino-2-phenylindole (DAPI), blue] in mouse salivary gland tumor tissue of P1-Mcre and P2-Mcre mice. Five-micrometer cryostat sections were fixed in buffered 4.0% paraformaldehyde for 10 minutes, blocked in swine serum for 7 minutes (dilution 1:20), and incubated first with the primary rabbit polyclonal anti-PLAG1 antibody, PEM 190 or PEM195 (ref. 22; dilution 1:60) for 60 minutes at room temperature and, secondarily, with fluorescein-labeled swine anti-rabbit antibody (dilution 1:20; TRITC, red, DAKO) for 45 minutes. Sections were mounted in Citifluor containing 0.5 μg/μL of DAPI. The slides were finally analyzed with a Zeiss Axiophot microscope equipped with UV optics (Carl Zeiss, Inc., Thornwood, NY). Images were recorded with a CE200A CCD camera (Photometrics, Tucson, AZ).
Northern blot analysis and DNA probes. Total RNA from mouse tissues or tumors was isolated according to the method described by Chomczynski et al. ( 23). Northern blot analysis was done as described previously ( 24). The following probes were radiolabeled with [α-32P]dCTP (MP Biomedicals, Asse-Relegem, Belgium) using the Megaprime DNA Labeling kit (GE Healthcare): for PLAG1, a 1.5-kb cDNA fragment encompassing the complete PLAG1 open reading frame; for Igf2, a 674-bp DNA fragment containing the sequence between nucleotides 1253 and 1927 in NM_010514; for H19, a 1.5-kb EcoRI/XbaI DNA fragment of the AA408602 cDNA clone; for Dlk1, a 629-bp PCR fragment containing the sequence between the nucleotides 64 and 1270 of NM_003836.3; for Gtl2, a PCR fragment containing the sequence between the nucleotides 284 and 778 of AJ320506; and for the β-actin gene, a PCR fragment containing the sequence between nucleotides 825 and 1112 of mouse β-actin (X03672).
Generation of two independent PLAG1+/− transgenic mouse strains with PLAG1 expression dependent on Cre-mediated removal of a stop cassette. Results of initial experiments to generate PLAG1 transgenic mice indicated that early and ubiquitous expression of the PLAG1 gene leads to embryonic lethality. Therefore, a PLAG1 DNA construct was designed from which a floxed “stop cassette” had to be removed via Cre-mediated excision to allow PLAG1 expression. The cloning strategy is described in Materials and Methods and the genetic composition of the generated DNA construct is shown in Fig. 1A. The integrity and functionality of the PLAG1 DNA construct was validated in vitro by Western blot analysis of HEK293T cells, transfected with DNA of the PLAG1 construct and this in the presence or absence of DNA coding for the Cre enzyme. Upon transfection of only DNA representing the PLAG1 construct, the PLAG1 protein could not be detected ( Fig. 1B, lane 1). Upon cotransfection of the PLAG1 DNA and DNA encoding the Cre enzyme, readily detectable levels of the PLAG1 protein were present ( Fig. 1B, lane 2), indicating efficient Cre-mediated removal of the stop cassette.
Upon independent zygote injection experiments using the SalI/SfiI linearized and purified DNA fragment of about 6 kb ( Fig. 1A) and subsequent breeding of the offspring, two independent hemizygous PLAG1+/− transgenic mouse strains (PTMS), PTMS1 and PTMS2, were obtained. To test the integrity and functionality of the PLAG1 DNA construct in vivo, PTMS1 and PTMS2 mice were crossed with PGK-Cre+/+ transgenic mice. It is known that the PGK-promoter is already expressed in the two-cell stage during embryogenesis ( 25). Therefore, early and widespread PLAG1 expression could be expected in the offspring. Intercrossing of PTMS1 mice with PGK-Cre+/+ transgenic mice did not result in any PGK-Cre+/−/PLAG1+/− offspring. This was most likely due to early embryonic lethality, as can be deduced from the fact that about 50% (39 of 70) of the embryos appeared embryonically resorbed ( Fig. 1C). The genotype of the resorbed embryos could not be determined, however, that of the normally developed embryos (31 of 70) was invariably PGK-Cre+/−/PLAG1−/−. In contrast, intercrossing of PTMS2 mice with PGK-Cre+/+ transgenic mice resulted in litters in which 50% of the mice were PGK-Cre+/−/PLAG1+/−. Although theoretically ubiquitous expression was predicted for these PGK-Cre+/−/PLAG1+/− mice, PLAG1 expression seems restricted mainly to the bladder on Northern blot analysis (data not shown). However, discriminating PCR analysis of DNA isolated from various tissues of these PGK-Cre+/−/PLAG1+/− mice, including stomach, small intestine, caecum, colon, heart, seminal vesicle, uterus, kidney, and bladder, clearly established that excision of the stop cassette had taken place, at least partially, in all these tissues (data not shown).
Targeting stop cassette excision to the salivary glands results in salivary gland tumor development. In previous in vitro studies, the tumorigenic capacity of PLAG1 was established ( 16). The aim of the present studies was to establish whether PLAG1 is capable to induce salivary gland tumors in mice and also to investigate in what way such tumors resemble human pleomorphic adenomas with PLAG1 overexpression. Therefore, PTMS1 and PTMS2 mice were intercrossed with B6129-Tgn(MMTV-LTR/Cre)1Mam transgenic mice ( 19) to target stop cassette excision and subsequent PLAG1 expression activation to particular tissues. This mouse strain was used because the expression pattern of the Cre enzyme has been well defined and includes the salivary and the mammary glands ( 19). Such intercrossing between PTMS1 and B6129-Tgn(MMTV-LTR/Cre)1Mam transgenic mice resulted in the MMTV-LTR-Cre+/−/PLAG1+/− (P1-Mcre) offspring. Within 5 weeks, 100% (37 of 37) of the P1-Mcre mice developed a large tumor mass ( Fig. 2A ) in the ventral neck region, which upon necropsy always seemed to consist of bilateral, multifocal salivary gland tumors ( Fig. 2B). These tumors seemed to preferentially originate from submandibulary salivary glands, but tumors originating from the sublingual salivary glands were also observed. In contrast, only about 6% (6 of 97) of the P2-Mcre mice developed salivary gland tumors, and all of these originated from the submandibulary salivary glands, arose unilaterally, and constituted a single tumor mass. Furthermore, the tumors in the P2-Mcre mice became apparent much later (i.e., in most cases only after a latency period of several months). Discriminating PCR analysis of DNA isolated from salivary gland tumors of these P1-Mcre and P2-Mcre mice clearly showed that excision of the stop cassette had taken place in all these tumors (data not shown).
Activation of expression of the PLAG1 transgene in P1-Mcre and P2-Mcre mice. To study activation of expression of the PLAG1 transgene after Cre-mediated excision of the stop cassette resulting from intercrossing PTMS1 or PTMS2 mice with B6129-Tgn(MMTV-LTR/Cre)1Mam transgenic mice, Northern blot analysis was done on RNA from tissues from the offspring using the open reading frame sequences of the human PLAG1 cDNA as molecular probe ( Fig. 2C). In salivary gland specimens of PTMS1 mice (PLAG1+/−), no expression of the transgene could be detected in line with the inhibition of expression by the stop cassette ( Fig. 2C, lane 1). Furthermore, no expression of endogenous Plag1 could be detected either ( Fig. 2C, lane 1). Analysis of RNAs isolated from a variety of tissue samples of salivary gland tumor-bearing male and female 5-week-old P1-Mcre mice revealed that expression of the PLAG1 transgene was very high in the samples of salivary gland tumors ( Fig. 2C, lane 2). Furthermore, weak PLAG1 expression was found in samples of mammary glands, ovary, and seminal vesicles ( Fig. 2C, lanes 3, 9, and 11), and very weak expression in testis and lung ( Fig. 2C, lanes 12 and 13). This correlates with the expression pattern of the Cre enzyme in the B6129-Tgn(MMTV-LTR/Cre)1Mam transgenic mice as described by Wagner et al. ( 19). Expression of the endogenous mouse Plag1 gene was also detected in some of the tissues tested, including mammary gland, adrenal gland, kidney, stomach, ovary, testis, lung, and brain but at low and varying levels ( Fig. 2C, lanes 3-6, 9, 12-14). Similar studies with tissues from salivary gland tumor bearing P2-Mcre mice (data not shown) also revealed high levels of PLAG1 expression in these tumors.
Time course of activation of PLAG1 overexpression. Salivary gland specimens from P1-Mcre mice were studied by Northern blot analysis at different ages after birth. For each stage, at least two mice were analyzed, and 10 in the case of 5-week-old mice. As shown in Fig. 3A , PLAG1 transcripts could be observed in salivary gland tissues at all ages analyzed, including mice ages 1 week (lane 2), 2 weeks (lane 3), 4 weeks (lane 4), or 5 weeks (lanes 5-7). This points towards activated expression of the PLAG1 transgene already at the very first measuring point. It should be noted that tumor nodules were not macroscopically visible in salivary gland specimens of 1- or 2-week-old mice; even not at necropsy. Salivary gland tumor nodules were macroscopically visible in 4-week-old mice but only at necropsy. Between 4 and 5 weeks, large salivary gland tumors became clearly visible as large protruding tumor masses that grew rapidly ( Fig. 2). Upon analysis of a salivary gland tumor specimen of a P2-Mcre mouse of about 11 weeks old, similar levels of PLAG1 transcripts were detected ( Fig. 3A, lane 8). Northern blot analysis of tissue specimens of salivary glands of PTMS1 mice, respectively, 1 and 3 weeks after birth, revealed no PLAG1 expression ( Fig. 3A, lanes 1 and 9). The transgene transcript was obviously also not detectable in salivary gland tissue of B6129-Tgn(MMTV-LTR/Cre)1Mam transgenic mice (MMTV-LTR/Cre+/+; Fig. 3A, lane 10). Immunofluorescence experiments using the well-characterized PLAG1-specific antibody PEM-195 revealed a widespread nuclear staining in the salivary gland tumors of about 5-week-old P1-Mcre mice. Mostly, such staining comprised whole lobules ( Fig. 3, B1 and B2). Similar staining was seen in tumors of P2-Mcre mice (data not shown).
Expression up-regulation of other genes resulting from activation of PLAG1 overexpression. Previously, we showed that the IGF-II gene is strongly up-regulated in human pleomorphic salivary gland adenomas with PLAG1 overexpression, ( 12). Moreover, recent microarray studies revealed that the human IGF-II gene is a genuine target gene of PLAG1 ( 15). To test whether PLAG1 overexpression in the salivary glands of P1-Mcre and P2-Mcre mice also led to induction of Igf2 expression, Northern blot analysis of the same salivary gland tissue specimens was done with an Igf2-specific probe. In all cases in which PLAG1 transcripts were detectable, the multiple Igf2 transcripts (4.8, 3.8, and 3.6 kb in size) were clearly expressed ( Fig. 3A, lanes 2-8). Furthermore, we found that expression of another gene within the imprinted Igf2 gene cluster on mouse chromosome 7 (i.e., the H19 gene that encodes an untranslated RNA) was also strongly up-regulated in the salivary gland tumors of these mice but not in control salivary gland specimens ( Fig. 3A, lanes 1 and 9). Of interest to note is that expression of another but similarly imprinted gene cluster is also affected by PLAG1. This cluster, located on mouse chromosome 12, contains the genes Dlk1 and Gtl2, of which Gtl2 encodes an untranslated RNA, like H19. As can be seen in Fig. 3A, a 1.8-kb Dlk1 and a 7-kb Gtl2 transcript are expressed in the salivary gland tumors ( Fig. 3A, lanes 6-8) but not in the control glands ( Fig. 3A, lanes 9 and 10). These results clearly establish the up-regulation of expression of several genes in two distinct, but structurally somewhat similar organized, imprinted gene clusters.
Early tumoral stages of the salivary glands in P1-Mcre mice. Submandibulary glands of P1-Mcre mice and those of the wild-type controls were examined microscopically at weekly intervals starting at 1 week after birth. Histologic analysis of salivary glands of P1-Mcre mice ages 1 week (one case), 2 weeks (one case), and 4 weeks (four cases) revealed multiple foci most likely representing early tumoral stages. Most of the lobules show these abnormalities. They contain epithelial cells with a large basal oval nucleus and basophilic cytoplasm that are densely packed around a small empty lumen. Dilated ducts with a flat epithelium, sometimes multilayered, are also visible. Mitotic figures were regularly observed ( Fig. 4A ) and BrdUrd labeling revealed active proliferation in early as well as in later tumoral stages ( Fig. 4B).
Pleomorphic adenoma in P1-Mcre and P2-Mcre mice. In P1-Mcre mice, the tumors were clearly visible macroscopically in the ventral neck region after 5 weeks. Lobulated tumors were observed bilaterally at necropsy, consisting of a large white, encapsulated mass. The largest tumors possessed hemorrhagic parts. Similar observations were made at dissection of salivary gland tumors of six P2-Mcre mice (tumor 1-6, respectively, 11, 16, 23, 38, 36, and 70 weeks old at sacrifice), but it should be emphasized that those tumors arose after a much longer latency period (several months) and not bilaterally.
Histologically, all salivary gland tumors from the P1-Mcre (12 tumors studied) as well as from the P2-M-Cre mice (six tumors) emerged from the submandibulary gland. The tumors show histologic features that are reminiscent of those of human pleomorphic adenomas. A thin capsule was usually visible, sometimes with capsular ingrowth of tumoral lobes. The tumors were composed of epithelial structures and myoepithelial cells ( Fig. 4C). Epithelial components of the tumors showed a high morphologic diversity, because within the same tumor, ductal structures lined by one or more cellular rows, cysts filled with fluid, papillary patterns, solid sheets of cells, sebaceous differentiation, and frequently squamous areas with keratinization ( Fig. 4D) could be observed. Spindle and stellate cells, more or less numerous, that have the morphologic characteristics of myoepitelial cells were closely associated with epithelial islands. Immunohistochemical analysis showed that the epithelial parts of the tumors were strongly positive for anti-cytokeratin 8/18 staining, as illustrated in Fig. 5A . A high proportion of the spindle or stellate cells were positive after anti-SMA staining, as illustrated in Fig. 5B. The myxoid stroma ( Fig. 5C) seemed alcianophilic, after Alcian Blue staining at pH 0.2, 1, and 2.5 ( Fig. 5D) and mucus secreting, PAS-positive cells, were numerous. In older tumors of the P1-Mcre mice (8 weeks to 4 months, eight tumors studied) and of the P2-Mcre mice (six tumors), malignant features were observed. These included large areas of necrosis ( Fig. 6A ), hemorrhages and cellular polymorphism. Besides some regions with typical characteristics of pleomorphic adenomas, some of the older tumors of the P1-Mcre and P2-Mcre mice contained regions resembling carcinomas. In 3 tumors of the P1-Mcre mice (9, 14, and 15 weeks old) and one tumor of the P2-Mcre mice (23 weeks old), lung metastases were observed. These showed the same histologic characteristics of the primary salivary gland tumor as illustrated in Fig. 6B and C.
Mammary gland tumors in P2-Mcre mice. Expression of the PLAG1 transgene in the P1-Mcre mice was not restricted to the salivary glands, which is consistent with the Cre expression pattern reported for the B6129-Tgn(MMTV-LTR/Cre)1Mam transgenic mice ( 19). Normal mammary glands of the P2-Mcre mice also expressed PLAG1 transcript levels detectable by Northern blot analysis ( Fig. 2C, lane 3). About 8% of the P2-Mcre mice developed tumors of the mammary gland with a latency period of about 1 year. These tumors shared some histologic features with the salivary gland tumors ( Fig. 6D). They were also composed of epithelial and myoepithelial cells in a myxoid stroma. They also showed solid areas of poorly differentiated basophilic epithelial cells, with a high mitotic index. A sebaceous differentiation was often seen. Mammary gland tumors were not observed in P1-Mcre mice, because these mice had to be sacrificed for ethical reasons before the age of 4 months because of their enormous salivary gland tumors. Nevertheless, some histologic abnormalities were found in mammary glands of P1-Mcre mice, showing abnormal differentiation with sebaceous differentiation, multilayered cells around the ducts and cystic structures (data not shown).
It is well established that overexpression of the developmentally regulated PLAG1 gene occurs frequently in various benign human solid tumors, such as pleomorphic salivary gland adenomas ( 1, 2), lipoblastomas ( 3, 4), and also in the childhood neoplasia hepatoblastoma ( 6). In the first two tumor types, it has been suggested that overexpression of PLAG1 is directly responsible for the formation of this type of tumors. This is mainly based on the causal link between the cytogenetically and molecularly well-defined recurrent chromosome 8q12 aberrations and the invariably strong up-regulation of PLAG1 expression in such tumors. Direct solid proof, however, is still lacking ( 1). Recently, PLAG1 and PLAGL2 have also been found implicated in AML ( 7, 9). Furthermore, experiments have clearly established the in vitro oncogenic potential of PLAG1 ( 16). Altogether, these findings have demarcated the PLAG1 gene as a versatile proto-oncogene and attracted considerably attention to the study of its function. As an approach to assess in vivo the consequences of overexpression of PLAG1 in the complex genetic background of a mammalian model system, two independent PLAG1 transgenic mouse strains were developed and various features are reported here.
Our PLAG1 transgenic mouse strains constitute a valuable model to study pleomorphic salivary gland tumorigenesis. With the P1-Mcre model, tumor progression can easily be followed in the time. Indeed, already after 1 week, 100% of the salivary glands of these mice contain early tumoral stages in the ductal region. It should be noticed that the expression pattern of the Cre recombinase in the salivary glands of B6129-Tgn(MMTV-LTR/Cre)1Mam transgenic mice is also limited to these regions ( 19). Already after 5 weeks, P1-Mcre mice develop pleomorphic adenomas, which share many histopathologic features with human pleomorphic adenomas. Like the human tumors ( 26), they are mixed tumors containing epithelial and myoepithelial cells. Epithelial elements are highly polymorphic within the tumor, because a tubular, acinar, cystic, papillary, or solid pattern with frequent squamous differentiation can occur. Tumors are embedded in a pale myxoid stroma in which fusiform to stellate myoepithelial SMA-positive cells are seen. Usually, the tumor is encapsulated and epithelial structures are well differentiated. In some of the older P1-Mcre and P2-Mcre mice, besides typical features of pleomorphic adenoma also malignant characteristics were observed, including areas of necrosis, hemorrhages, and metastasis to the lungs. Some regions even resembled carcinomas. Consistently, it has been well described that also human pleomorphic adenomas can progress to malignancy. In humans, 2% to 17% of the pleomorphic adenomas can progress to carcinomas ex pleomorphic adenomas ( 27– 29). Sometimes, the pleomorphic part of the tumor is only very small. The malignant characteristics observed in our mouse model suggest a similar malignant progression.
Although the PLAG1-induced salivary gland tumors in the transgenic mouse strains possess the basic hallmarks of human pleomorphic adenoma, a major difference with the human tumors pertains to the aggressive growth of the tumors, at least for those of the P1-Mcre strain. The engineered genetic intervention and/or the different genetic background of the mouse can most likely explain this difference. Human pleomorphic adenomas of the salivary glands are assumed to arise following a genetic accident, presumably in a single progenitor cell, and leading to a clonal activation of PLAG1 expression long after birth. In contrast, in the PLAG1 transgenic mouse model system, PLAG1 activation starts already during embryonic development and occurs multiclonally and at an early phase in the epithelial duct cells of the developing salivary glands. We cannot exclude that some of the alterations observed in the early stages are developmental changes. In any case, a major phenotypic effect of overexpression of the PLAG1 transgene is the onset of fast growing salivary gland tumors.
There are some differences between our two PLAG1 transgenic mouse strains. First, in P1-Mcre mice lesions occur bilaterally and in most of the lobules. Moreover, 100% of the mice develop tumors. In contrast, however, in the P2-Mcre strain, the tumor frequency is low and the tumorigenic process has a long latency period (several months). Moreover, such tumors arose invariably locally and unilaterally. Zhao et al. reported a similar variation in tumor incidence in PLAG1 transgenic mouse strains ( 30). Second, crossing of the PTMS1 with PGK-cre mice to generate PGK-Cre+/−/Plag1+/− mice resulted in embryonic lethality, due to early and ubiquitous expression of the transgene. Embryonic lethality was not observed in similar experiments with PTMS2, probably because there is almost no expression of the PLAG1 transgene in the offspring. We assume that the differences in expression of the PLAG1 transgene in the two founder strains are due to the effect of the integration site of the transgene ( 31).
Of interest to note, finally, is that after a latency period of about 1 year, 8% of the P2-Mcre mice developed mammary gland tumors, which possessed histologic features similar to those observed in the salivary gland tumors. Similar observations were not made with the P1-Mcre mice, because they developed fast growing salivary gland tumors and had to be sacrificed before mammary gland tumors could arise. For this strain, more direct and specific targeting of expression of the PLAG1 transgene to the mammary gland, excluding the salivary gland, seems the strategy to follow. It should be mentioned that PLAG1 expression has never been described in primary human mammary gland tumors.
In conclusion, our studies clearly establish first that ubiquitous overexpression of the PLAG1 proto-oncogene leads to lethality during embryonic life. Second, a direct link was shown between overexpression of the PLAG1 proto-oncogene and tumorigenesis. Third, activation of genes from independent imprinted gene clusters (Igf2, H19, Dlk1, and Gtl2; refs. 32– 34) may play a role in this tumorigenic process, although the precise nature of it remains to be established. In human salivary gland pleomorphic adenomas, the IGF-II gene, which is an established important element in tumor cell growth ( 35), has been identified as a genuine target gene of PLAG1 ( 12, 15). On the other hand, it is possible that the mechanism of imprinting is involved. Finally, the PTMS1 and PTMS2 strains described here have been designed in such a way that activation of expression of the PLAG1 transgene is Cre mediated. With the availability of mouse strains with different tissue-specific Cre expression ( 36), valuable models can now be obtained to study the contribution of PLAG1 to the development of other types of tumors.
Grant support: Geconcerteerde Onderzoekacties 2002/010, the “Fonds voor Wetenschappelijk Onderzoek Vlaanderen” grant G.0099.02, and the Cancer Research Program of Fortis Bank Insurance 2002-2005.
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.
We thank C. Van den Broeck for technical assistance.
Note: L. Fiette and W.J.M. Van de Ven should both be considered as last authors.
M. Voz is currently at the Laboratoire de Biologie et de Génie Génétique, Institut de Chimie, allée du 16ième Août, 4000 Sart-Tilman, Belgium.
L. Fiette is currently at the Platform of Veterinary Diagnosis, CMU, 1 rue Michel Servet, CH 1211, Geneva 4, Switzerland.
C.V. Braem, I. Van Valckenborgh, and J. Declercq are aspirant fellows of the “Fonds voor Wetenschappelijk Onderzoek Vlaanderen,” whereas F. Van Dyck is an aspirant fellow of the “Instituut voor de aanmoediging van Innovatie door Wetenschap en Technologie in Vlaanderen”.
↵6 Unpublished results.
- Received November 16, 2004.
- Revision received March 18, 2005.
- Accepted March 22, 2005.
- ©2005 American Association for Cancer Research.