High-grade serous carcinomas arise in the mouse oviduct via defects linked to the human disease

Genetically modified mice and animal care

The development and characterization of Ovgp1-iCreER^T2 transgenic mice have been described in detail [22]. Ovgp1-iCreER^T2 transgenic mice were crossed with mice carrying engineered Rb1, Trp53, Brca1, Nf1, and Pten alleles, to generate transgenic mice with various combinations of TSG alterations. In addition to floxed Rb1, Trp53, Brca1, Nf1, and Pten alleles in which Cre-mediated recombination generates null alleles, we employed mice carrying a Cre-inducible Trp53 mutation (LSL-R172H) and mice with constitutional inactivation of one Brca1 allele (Brca1^del). Rb1^fl, Trp53^fl, Trp53^LSL-R172H, and Brca1^fl mice were obtained from the National Cancer Institute’s mouse repository. Nf1^fl/fl mice were a gift from Yuan Zhu (currently at Children’s National Medical Center, Washington, DC, USA). Pten^fl/fl mice were a gift from Tak Mak (University Health Network, Toronto, Ontario, Canada). Mice with the Brca1^del allele were generated by crossing E2a-Cre mice (B6.FVB-Tg[EIIa-cre]C5379Lmgd/J, Jackson Laboratory, Bar Harbor, Maine, USA) with Brca1^fl/+ mice to generate mice carrying an inactivated Brca1 allele in the germline. All strains were maintained on a mixed genetic background. Mouse genotypes were confirmed by PCR analysis of tail DNA (primer sequences in Supplementary Table S1). Procedures involving mice were approved by the University of Michigan’s Institutional Animal Care and Use Committee (PRO00006370).

In vivo induction of oviductal tumours

Ovgp1-iCre-ER^T2 female mice carrying various engineered TSG alleles were given intraperitoneal injections of tamoxifen (TAM, T5648, Sigma-Aldrich, Indianapolis, IN, USA) at 100 mg/kg of body weight on three consecutive days, usually when mice were 6 to 8 weeks old. Initially, while monitoring mice for tumour development, selected mice were euthanized at various time points following TAM injection and inspected for tumour location and extent at necropsy. Thirteen mice underwent survival surgery, during which the right oviduct and ovary were removed at two (n=11), three (n=1) or twelve (n=1) months after TAM, and the mice were then monitored for tumour development in the remaining oviduct. Once the presence of a tumour was established, most mice were monitored to humane endpoints. Mice were grossly evaluated for tumour extent, and in each case, ovaries, oviducts, uteri, lungs, multiple abdominal organs, omentum, and mesentery were examined microscopically.

Histopathology and immunohistochemistry

Tissues were fixed in 10% buffered formalin and paraffin-embedded. Sections (5 µm) were H&E-stained for evaluation by light microscopy. To identify microscopic oviductal lesions, each oviduct and ovary were serially sectioned in their entirety for microscopic examination. Alternate sections were either stained with H&E or retained for immunohistochemical (IHC) staining, performed using standard methods as previously described [22]. Antigen retrieval was performed by microwaving the slides in citrate buffer, pH 6.0 (Biogenex, San Ramon, CA, USA) for 15 min. Antigen–antibody complexes were detected using the avidin–biotin peroxidase method, with diaminobenzidine (DAB) as the chromogenic substrate (Vectastain ABC kit, Vector Laboratories, Burlingame, CA, USA). Primary antibodies used were: rat anti-cytokeratin 8 (Developmental Studies Hybridoma Bank, University of Iowa), rabbit anti-p53 and rabbit anti-PAX8 (Proteintech, Chicago, IL, USA), rat anti-Ki67 (DAKO, Carpinteria, CA, USA) and goat anti-OVGP1 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA).

STICs, HGSCs, and MMMTs develop in the mouse oviduct following conditional inactivation of various combinations of Rb1, Brca1, Trp53 and/or Nf1 in the FTE

We crossed Ovgp1-iCreER^T2 mice with mice carrying Brca1, Trp53, Rb1, and Nf1 alleles, to generate 80 transgenic mice of several different genotypes (Figure 1 and Supplementary Table S2). Mice with conditionally modified Brca1, Trp53 and Rb1 alleles are hereafter referred to collectively as BPR mice (n=29). BPN mice (n=3) are those with modified Brca1, Trp53 and Nf1 alleles, while BPRN mice (n=48) carry modified alleles for all four TSGs. Data summarizing the oviductal lesions developing in these mice are shown in Figure 1, in which animals with generally similar genotypes are grouped together.

In order to determine whether Brca1, Trp53, Rb1, and/or Nf1 inactivation in the FTE results in an oviductal tumour phenotype, ten mice with a range of defined genotypes were euthanized 1.5 to 6.5 months after TAM. One BPRN mouse (#2820) showed a microscopic lesion closely resembling human STIC in one oviduct 2.5 months after TAM injection. Oviductal lesions were not identified in the remaining nine mice. In an attempt to reduce the number of mice needed to determine if, and how often, mice with different genotypes acquire oviductal tumours, the right oviduct and ovary were removed from thirteen mice in a survival surgery procedure. Eleven of the thirteen (four BPRN, six BPR, and one BPN) had the right oviduct and ovary removed two months after TAM injection. STIC was found in one BPRN mouse (#3867, with homozygous floxed alleles for all four TSGs). Two additional BPRN mice underwent survival surgery at 3 and 12 months after TAM (#3559 and #8234), and STIC was identified in the latter. Photomicrographs of two representative mouse STIC lesions are shown in Figure 2. As in humans, the mouse STICs have a predilection for arising in the distal portion of the oviduct (Figures 2A and 2C, equivalent to human fallopian tube fimbriae). The lesional cells display enlarged, hyperchromatic and pleomorphic nuclei, prominent nucleoli, increased mitotic activity, and loss of polarization (Figures 2B and 2D).

All thirteen mice that underwent survival surgery were monitored for tumour development in the remaining oviduct for 5 to 23 months following TAM injection. Lesions were identified in the remaining oviduct in five of six BPRN mice, two of six BPR mice, and in the BPN mouse. The oviductal lesions identified in the remaining oviduct included two STICs, two early HGSCs (confined to oviduct), two HGSCs extending beyond the oviduct, and two MMMTs, one of which metastasized widely.

Oviductal lesions (STIC, early HGSC, HGSC and/or MMMT) were identified in 50 of 57 additional mice (12 BPR, 36 BPRN, and 2 BPN mice) not included in the groups described above. These mice were monitored for tumour development for time periods ranging from 7 to 24 months after TAM (Figure 1). In the 59 of 80 mice with oviductal lesions, tumour involved one or both ovaries in 24 (41%), obliterating the ovary in all but four cases. Widespread peritoneal metastases were observed in 8 mice (14%) and ascites in 7 (12%). In similar fashion to the human disease, metastases were often observed in the omentum. Representative photomicrographs showing tumour progression from STIC to early HGSC, HGSC/MMMT, and metastatic disease are shown in Figure 3, and IHC staining of primary tumours at various stages of tumour progression, and metastases to omentum and ovary are shown in Figure 4 and Supplementary Figure S1, respectively. The tumour cells expressed CK8 and PAX8 and showed increased proliferation, based on IHC staining for Ki67. Tumours with the conditional Trp53 R172H mutant allele only occasionally showed focal stabilization of p53 protein (Figure 4, #3724). While the neoplastic cells in STICs and HGSCs retain expression of PAX8, a marker of secretory cells, they do not express tubulin, which marks ciliated cells (Supplementary Figure S2). The absence of tubulin expression is not surprising, as human HGSCs and STICs do not typically display ciliated morphology.

Only 6 of 48 (12.5%) BPRN mice, compared to 15 of 29 (52%) BPR, failed to develop lesions. Some of the failures are likely attributable to euthanasia at relatively early time points (≤6.5 months after TAM), particularly in the BPR mice. All three BPN mice included in the study developed oviductal HGSC and/or MMMT. On a per oviduct basis, and excluding those removed at the time of survival surgery or evaluated earlier than 7 months after TAM, 87 oviducts had at least one floxed Nf1 allele. In these oviducts, 11% had no neoplastic lesions, 21% had STIC, 25% had early HGSC, and 43% had HGSC/MMMT. In contrast, for the 39 oviducts without floxed Nf1, the percentages were 36%, 41%, 13% and 10%, respectively (p = 3.8 × 10⁻⁵, chi-squared test; p = 2.7 × 10⁻⁶, Mantel-Haenszel chi-squared test of association). Based on our analysis of the data shown in Figure 1 and in Supplementary Table S2, we conclude that the HGSC phenotype in our model is highly penetrant in BPRN and BPN mice, especially those carrying homozygous floxed alleles for all TSGs. Furthermore, disease progression is generally more rapid in BPRN than in BPR or BPN mice.

Cre-mediated recombination of Brca1, Trp53, Rb1 and Nf1 alleles in representative tumours was confirmed by PCR (Supplementary Figure S3). Interestingly, in mice carrying one floxed and one wild type Rb1 allele, loss of the wild type Rb1 allele was seen in seven of eight tumours analyzed (e.g. Supplementary Figure S3, #s 13085, 13576, 13179, and 2117). Hence, as in human HGSC pathogenesis, the mouse tumours acquire additional genetic alterations during tumour progression. We evaluated oviductal tumour DNA for activation of the conditional Trp53^LSL-R172H mutant allele in three representative tumours, two of which are shown in Supplementary Figure S3C. In all three tumours, recombination of the conditional mutant Trp53 allele was not observed. Interestingly, of mice with both oviducts intact for 7 months or more, 3 of 6 with the Trp53^LSL-R172H/+ genotype had no detectable oviductal lesions, while only 4 of 51 mice with at least one floxed Trp53 allele lacked lesions (p = 0.020, two-sided Fisher’s exact test). These findings suggest that in this HGSC model system (i) there is selection for biallelic inactivation of Rb1 and (ii) the Trp53^fl allele, generating a null Trp53 mutation via Cre-recombination, is preferable to the Trp53^LSL-R172H mutant allele, perhaps related to inefficient Cre recombination or to the absence of selective pressure for recombination and/or selection against recombination. Of note, in the absence of Cre-mediated activation of the R172H mutation, the Trp53^LSL-R172H allele is null [23].

In addition to the oviductal tumours, non-Müllerian tumours (e.g. lymphomas, soft tissue sarcomas, and adenomas/carcinomas of the thyroid, lung, skin, or breast) were identified in a subset of mice. Lymphomas were the most common type of non-oviductal tumour (17/80 – 21%). Interestingly, three of five mammary carcinomas arose in mice carrying one constitutional inactivated Brca1 allele. Development of lymphomas, sarcomas, carcinomas and other tumour types has been reported in various mouse strains, including some with high incidence [24–26]. Hence, development of non-oviductal tumours is not unexpected, given the advanced age of many of the mice and the presence of mutant Brca1 and/or Trp53 alleles in the germline of a sizeable subset. In order to determine if Cre-mediated recombination occurred in the non-Müllerian tumour cells, we evaluated DNA from representative tumours for the presence of recombined Brca1, Trp53, Rb1, and Nf1 alleles (Supplementary Figure S4). We found Cre-mediated recombination of engineered TSG alleles in one endometrial carcinoma (not shown) and in the presumptive thyroid carcinomas, which had morphological features reminiscent of human medullary carcinoma. The remaining tumours appear to arise spontaneously. We do not observe OVGP1 protein expression in the thyroid based on IHC staining, and hence allelic recombination in this site appears to be independent of endogenous Ovgp1 expression.

TAM-treated Ovgp1-iCreER^T2;Brca1^fl/fl;Trp53^fl/fl;Pten^flfl (BPP) mice develop STICs, HGSCs, and MMMTs, and also display oviductal epithelial hyperplasia and mucinous metaplasia

We crossed Ovgp1-iCreER^T2 mice with animals carrying floxed Brca1, Trp53, and Pten alleles, to generate Ovgp1-iCreER^T2;Brca1^fl/fl;Trp53^fl/fl;Pten^flfl (BPP) mice. Ten mice were treated with TAM, and two were euthanized at each of five time points (1, 2, 3, 4, and 6 months after TAM). STICs or early HGSC-like lesions were already present in treated mice one month after TAM. All ten mice developed bilateral oviductal lesions, including oviductal carcinomas at the later time points. One BPP mouse, examined 6 months after TAM, developed carcinosarcoma with metastasis to the ovary and ascites. Endometrial hyperplasia near the junction of the uterine horn with oviduct was noted in two of the ten BPP mice. Data from the BPP mice are summarized in Supplementary Table S3, and representative photomicrographs of H&E and immunostained lesions arising in the oviducts of BPP mice at 1 month, 2 months, and 6 months after TAM are shown in Figure 5. Like tumours arising in BPR, BPRN, and BPN mice, oviductal STICs (Figure 5A and B), HGSCs (Figure 5E, F, I, and J) and MMMTs (Figure 5M and N) arising in BPP mice express CK8 (Figure 5C, G, K and O) and PAX8 (Figure 5D, H, L, and P). As expected, they lack expression of PTEN and have an increased Ki-67 proliferative index (not shown). In addition, the oviducts of the TAM-treated BPP mice show rather diffuse epithelial hyperplasia characterized by areas with increased epithelial stratification, as well as mucinous metaplasia (Figure 6) in which the cells display prominent cytoplasmic accumulation of both acidic (Alcian blue-positive) and neutral (periodic acid Schiff-positive) mucin compared to normal oviductal epithelium. These latter findings were not observed in the BPR, BPN or BPRN mice. None of the BPP mice developed non-oviductal malignancies during the 6-month surveillance period.