Hyperplasia and Spontaneous Tumor Development in the Gynecologic System in Mice Lacking the BRCA1-Δ11 Isoform

Abstract

Alternative splicing in the BRCA1 locus generates multiple protein products including BRCA1-Δ11, which is identical to the BRCA1 full-length isoform (BRCA1-FL) except for the absence of exon 11. Mutation analysis using gene targeting to create null mutations or disrupt BRCA-FL has provided much of our understanding of BRCA1 functions; however, targeted mutation of specific short forms of BRCA1 has not been reported. To understand the physiologic functions of BRCA1-Δ11, we used a knock-in approach that blocks alternative splicing between exons 10 and 12 to prevent the formation of this form of BRCA1. We showed that homozygous mutant mice (Brca1^FL/FL) were born at a Mendelian ratio without obvious developmental defects. However, the majority of Brca1^FL/FL female mice showed mammary gland abnormalities and uterine hyperplasia after one year of age with spontaneous tumor formation. Cultured Brca1^FL/FL cells exhibited abnormal centrosome amplification and reduction of G₁ population that was accompanied by accumulation of cyclin E and cyclin A. Accumulation of cyclin E was also found in epithelial layers of dilated ducts and hyperproliferative lobular regions in the mammary glands of Brca1^FL/FL mice. These observations provide evidence that BRCA1 splicing variants are involved in BRCA1 functions in modulating G₁/S transition, centrosome duplication, and repressing tumor formation.

Inherited mutations in the BRCA1 gene predispose women to early onset breast and ovarian cancers (1, 4). BRCA1 contains 24 exons that encode proteins of 1,863 and 1,812 amino acids in humans and mice, respectively (23, 27). Notably, over 60% of the protein is encoded by an unusually large exon, exon 11, which is 3.4 kb in length. In addition to the full-length BRCA1 protein (BRCA1-FL), BRCA1 also encodes at least two protein products of smaller size due to alternative splicing (12, 33, 38, 42). One of the variants, BRCA1-Δ11 (also termed BRCA1-Δ11b), arises from in-frame splicing between exon 10 and exon 12, and it retains the highly conserved amino-terminal RING finger and carboxyl-terminal BRCT domains found in full-length BRCA1. The other is BRCA1-IRIS, which is a 1,399-residue polypeptide encoded by an uninterrupted open reading frame that extends from codon 1 of the known BRCA1 open reading frame to a termination point 34 triplets into intron 11 (12). To study functions of BRCA1 and create animal models for BRCA1-assocated breast cancer, mice carrying various mutations of BRCA1 have been generated by gene targeting (reviewed in reference 8). Analyses of these animals revealed that BRCA1 is involved in controlling genetic stability, DNA damage repair, centrosome duplication, apoptosis, and cell cycle control (reviewed in references 9, 10, 35, and 44). However, due to technical difficulties, no mutations have been introduced specifically into any short splicing forms of BRCA1.

Several lines of evidence indicate that the BRCA1-Δ11 isoform is functional. It has been shown that BRCA1-null embryos die at embryonic day 6.5 (E6.5) to E8.5 (24, 26, 30), while embryos that lack BRCA1-FL due to a targeted deletion of exon 11 (Brca1^Δ11/Δ11) but still express the BRCA1-Δ11 isoform die between E12 and E18.5 (40). The significantly prolonged survival of Brca1^Δ11/Δ11 embryos suggests that BRCA1-Δ11 partially compensates for BRCA1-FL functions during early embryogenesis. In addition, BRCA1-null embryos exhibited profound growth defects during early postimplantation stages, while Brca1^Δ11/Δ11 embryos developed and proliferated normally before E10, although their proliferation rate gradually slowed down after E12.5 (5, 37, 40). Moreover, BRCA1-null embryos, when placed in a p53^−/− or p21^−/− background, only survived 1 to 2 days longer (15, 26, 30), while Brca1^Δ11/Δ11 embryos survived to adulthood when one or both p53 alleles were mutated (40). The surviving animals exhibited premature aging phenotypes, tumorigenesis, and developmental abnormalities in multiple tissues/organs (2, 5, 34, 39, 40), suggesting that the expression of BRCA1-Δ11 alone cannot support normal postnatal development.

To address the potential functions of BRCA1-Δ11, we employed a cDNA knock-in approach that specifically blocks the alternative splicing from exon 10 to exon 12. Our analysis of the mutant (Brca1^FL/FL) mice and cultured mutant cells indicates that BRCA1-Δ11 is dispensable for murine development and DNA damage response upon irradiation in the presence of BRCA1-FL. However, it mediates a part of BRCA1 functions in cell cycle regulation, and consequently, its absence resulted in hyperplasia of the mammary gland and uterus and spontaneous tumor formation in old populations of mutant mice.

MATERIALS AND METHODS

Targeting and generation of mice.

Recombinant phage-containing overlapping genomic DNA of the Brca1 locus was isolated from a 129/SVJ mouse library (Stratagene, La Jolla, CA). To generate the deletion construct for BRCA1-Δ11, a 4.9-kb XhoI-XhoI fragment containing exon 11 and exon 12 was cloned in pBS Bluescript (Stratagene, La Jolla, CA). A restriction site, HindIII, is uniquely located in exon 11, which corresponds to the 3,878th nucleotide in Brca1 cDNA. The HindIII-XhoI fragment of genomic DNA was replaced by the 3′ HindIII digestion product of Brca1 cDNA. To ensure the proper expression and transcriptional termination, a simian virus 40 (SV40) poly(A) was also cloned at the end of the Brca1 cDNA. The resulting chimeric construct of genomic DNA and cDNA was cleaved with XhoI and XbaI and cloned into the SalI and XbaI sites of a pLoxPneo plasmid (43). The resulting construct was digested with XhoI and NotI followed by the insertion of the other arm of the targeting construct. The final targeting construct was designated pLoxPneo BRCA1-full length (pBrca1-FL). TC1 embryonic stem (ES) cells (7) were transfected with NotI-linearized targeting vector DNA and selected with G418 and FIAU [1-(2′-deoxy-2′-fluoro-β-d-arabinofuranosyl)-5′-indouracil] as described previously (11). ES cells heterozygous for the targeted mutation were microinjected into C57BL/6 blastocysts to generate chimerical mice. Germ line transmission was confirmed by agouti coat color in F₁ animals, and the offspring were genotyped for the Brca1 mutant allele by PCR and Southern blot analysis. The genetic background of these mice is 50% 129/SV and 50% NIH Black Swiss. Wild-type littermates were used as controls in all our studies.

Mouse analysis.

A total of 32 wild-type and 49 mutant mice at ages between 2 and 23 months were used for phenotype analysis. Mammary cycle-related studies and aging-related studies were performed according to methods described previously (5, 41). Mice used for long-term tumorigenesis study were all virgin. Ten mutant mice and ten wild-type littermates were monitored twice a week up to 23 month of age. At autopsy, tumors and surrounding tissues were dissected, frozen in liquid nitrogen, and stored at −80°C or fixed in 10% buffered formalin and embedded in paraffin for sectioning and hematoxylin and eosin staining. When each of the mutant mice was killed, its littermate wild-type mouse was also sacrificed to take the same tissues as control.

RT-PCR and Northern blot analysis.

Total RNAs were extracted from tissues and mouse embryonic fibroblasts (MEFs) using RNA STAT-60 (Tel-test, Friendswood, TX). Reverse transcriptase (RT) reactions were carried out using a first-strand cDNA synthesis kit (Roche, Indianapolis, IN). One microgram of RNA from each sample was used as the template for each reaction, and 1 μl of cDNA from each sample was used for PCR. The optimal number of cycles for amplification was chosen according to the cycle number that yielded the strongest band while staying within the linear range. The primers used in this study were as follows: 10F, 5′-AAG GCA AGC TGC ACT CTG CAG-3′; 11F, 5′-GGA AAT GGC AAC TTG CCT AG-3′; 14R, 5′-GCT TAC AGG ATT CTC ATT AAT G-3′; 15R, 5′-GAA GAT GCC TAG AGC AGC CAT G-3′; 16R, 5′-AGC TGA TTC CAG ATC CCA GG-3′; GAPDH-F, 5′-ACA GCC GCA TCT TCT TGT GC-3′; and GAPDH-R, 5′-TTT GAT GTT AGA GGG GTC TGC-3′. The numbers in the primers represent the numbers of exons, from which the sequences were originated. Northern blot analysis was performed using a P³²-labeled cDNA probe containing exons 10, 12, and 13 of the Brca1 gene as described previously (41).

MEF cells and cell cycle analysis.

MEF cells were derived from E14.5 embryos generated from intercrosses between Brca1^+/FL mice. All comparisons between wild-type and mutant cells were performed between littermates. For irradiation-induced G₁/S checkpoint analysis, MEFs at passage 1 that were plated 24 h earlier were synchronized at G₀ by incubation for 4 days in starvation medium containing 0.5% fetal bovine serum. The synchronized cells were irradiated after changing to complete medium containing 15% fetal bovine serum. After harvesting, the cells were analyzed by using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA). G₂/M and S checkpoint analyses were performed as described previously (22). The DNA contents and the cell cycle of MEFs were analyzed by the CellQuest (BD Biosciences, San Jose, CA) and ModiFit (Verity, Topsham, ME) programs.

Histology, immunohistochemical staining, and Western blotting.

For histology, tissues were fixed in 10% formalin, blocked in paraffin, sectioned, stained with hematoxylin and eosin, and examined by light microscopy. Detections of antigenic proteins were performed using a ZYMED Histostain staining kit (ZYMED, San Francisco, CA) according to the manufacturer's instructions. Western blot analysis was carried out according to standard procedures using ECL detection (Amersham, Piscataway, NJ). The following primary antibodies were used: Brca1 (against exon 11) (18), p53 (Upstate Biotechnology, Charlottesville, VA), p21, Chk1, and Chk2 (BD Biosciences, San Jose, CA), cyclin A, cyclin B, cyclin D1, cyclin E, pRb, cdc25a, and cdc25c (Santa Cruz, Santa Cruz, CA), and β-actin (Sigma, St. Louis, MO). Horseradish peroxidase-conjugated donkey anti-rabbit or sheep anti-mouse antibodies (Amersham, Piscataway, NJ) were used as secondary antibodies.

Centrosome staining and analysis.

Cells grown on chamber slides (BD Biosciences, San Jose, CA) were fixed in 3% paraformaldehyde for 10 min at room temperature. The slides were then washed with phosphate-buffered saline (PBS), permeabilized in 0.2% Triton X-100 and PBS, and incubated overnight with an anti-pericentrin polyclonal antibody (Covance, Princeton, NJ) diluted 1:1,000 in 3% goat serum and 3% bovine serum albumin in PBS. The antibody complexes were detected by an Alexa Fluor 488-conjugated secondary antibody (Molecular Probes, Eugene, OR) and stained with DAPI (4′,6′-diamidino-2-phenylindole). All images were obtained by Leica fluorescence microscopy with an Olympus charge-coupled-device camera and manipulated by Magnafire (version 2.0) software. The centrosome/nucleus ratio was determined by counting centrosome numbers per nucleus.

RNA interference of cyclin E.

RNA interference (RNAi)-mediated knockdown of cyclin E was performed in MEFs immortalized using a method described previously (5). The RNAi specific for mouse cyclin E was obtained from Santa Cruz (Santa Cruz, CA) and transfected with Lipofectamine 2000 (Invitrogen, Carlsbad, CA) following the manufacturer's suggestions.

Statistical analyses.

Student's t test (http://www.physics.csbsju.edu/stats/ttest.html) was used to compare differences between Brca1^FL/FL mutants and wild-type littermate controls as specified above. A P of ≤0.01 was considered statistically significant.

RESULTS

Targeted deletion of BRCA1-Δ11 in mice.

To specifically delete BRCA1-Δ11, we used a cDNA knock-in approach to construct a targeting vector, pBrca1-FL, which joins all exons that are 3′ to exon 11 (i.e., exon 12 through exon 24) together (Fig. 1A). To ensure the proper expression and transcriptional termination, the SV40 poly(A) was also placed at the end of the Brca1 cDNA. The correct targeting event should knock in the cDNA and concomitantly replace a part of Brca1 exon 11, intron 11, and the entire exon 12 from a HindIII site in exon 11 to an XhoI site in intron 12 (Fig. 1A). The deletion of exon 12 is designed to block the in-frame splicing between exons 10 and 12, preventing the generation of BRCA1-Δ11 transcripts.

FIG. 1.

Targeted deletion of BRCA1-Δ11. (A) Targeting vector and final structure of the mutant allele. B, BamHI; H, HindIII; X, XhoI; Xb, XbaI. (B) Southern blot analysis of ES cell DNA digested by BamHI and hybridized with a flanking probe (probe 1) to identify the targeting event that is characterized by a fragment shift from 8 kb to 9 kb. (C) Southern blot analysis of XbaI-digested tail DNA and hybridized with probe 2, which detects a fragment shift down from 8 kb to 6.5 kb. (D) RT-PCR analysis using primers against several exons of Brca1 as indicated by the numbers of the primers. PCR products specific for BRCA1-Δ11 transcripts (10F/16R, 10F/15R, and 10F/14R) were detected in wild type (+/+) but not in Brca1^FL/FL mutant (FL/FL) MEFs, while PCR products specific for BRCA1-FL (11F/14R) were detected in both wild-type and mutant MEFs. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (E) Northern blot analysis using a cDNA probe containing exons 10, 12, and 13 of the Brca1 gene also did not detect the Brca1-Δ11 transcript in the Brca1^FL/FL mutant MEF cells. Please note that the transcripts of the Brca1 from Brca1^FL/FL MEFs are shorter than the endogenous Brca1. This difference is caused by the replacement of the 3′ untranslated region of the endogenous Brca1 with a SV40 poly(A). (F) Expression of BRCA1-FL in MEFs generated from wild-type and Brca1^FL/FL mice revealed by Western blot analysis.

Eight out of 72 G418/FIAU-resistant clones analyzed by Southern blot analysis were correctly targeted (Fig. 1B). ES cells from three colonies were injected into blastocysts and resulted in germ line transmission. Mice carrying heterozygous deletion of BRCA1-Δ11 transcripts (Brca1^+/FL) were phenotypically normal and were further crossed to produce Brca1^FL/FL mice (Fig. 1C). Our data indicated that Brca1^FL/FL mice were presented in a Mendelian ratio at weaning (see Table S1 in the supplemental material). RT-PCR analysis (Fig. 1D) and Northern blot analysis (Fig. 1E) did not detect transcripts for BRCA1-Δ11, indicating that this isoform was knocked out. Western blot analysis indicated that the BRCA1-FL was expressed in both mutant and wild-type cells at comparable levels (Fig. 1F).

Our analysis indicated that the Brca1^FL/FL mice did not show any obvious abnormalities in terms of physical appearance, activity, and fertility before 12 months of age. This observation indicated that the loss of BRCA1-Δ11 did not cause any obvious developmental defects.

Hyperplasia in the gynecologic system and spontaneous tumor formation in aging population of Brca1^FL/FL mutant mice.

Our recent studies on mice bearing hypomorphic mutations revealed several abnormalities including premature aging and tumorigenesis (5, 22). Therefore, we investigated whether Brca1^FL/FL mice would have similar phenotypes. Our study revealed no signs of premature aging in any of the mutant mice examined up to 23 months of age. Of 10 Brca1^FL/FL female virgin mice examined between 13 to 23 months of age, 9 exhibited several gynecologic lesions such as uterine hyperplasia (squamous epithelial hyperplasia of the cervix and endometrial hyperplasia) and mammary gland hyperplasia (Fig. 2 and Table 1). Five mice also exhibited ovarian abnormalities including hemorrhagic cysts, stromal hyperthecosis, and/or follicular granulosa-theca cell hyperplasia (Fig. 2A and C). Histological analysis also revealed a significantly increased thickness of the cervix epithelium comparing controls (Fig. 2D and E) (wild type, 101 ± 15 μm; mutant, 787 ± 149 μm; P = 0.0046). Overall, seven mutant mice showed the squamous epithelial hyperplasia and five mutant mice revealed endometrial hyperplasia (Fig. 2E and G and Table 1). Three mice also developed thymic lymphoma (Table 1).

FIG. 2.

Brca1^FL/FL mice develop a variety of gynecologic lesions. (A) Uteri and ovaries from wild type (WT) and Brca1^FL/FL mice. Examples of histological sections of ovaries (B and C), cervixes (D and E), and endometria (F and G) from Brca1^FL/FL mice (C, E, and G) and age-matched wild-type mice (B, D, and F). The epithelium of a mutant uterine cervix is significantly increased (E) compared with the control (D). Bars, 100 μm. WT, wild type.

TABLE 1.

Summary of mice statuses^a

Mouse	Age of death (mo)	Abnormal organ(s) and/or condition	Uterus hyperplasia condition	Condition of mammary gland
				Ductal hyperplasiac	Lobular hyperplasia	Increased cyclin E	Overall
8182	19	Kidney	SEH, EH	+++	Yes	No	MC, DCIS
8222	13		SEH, EH	++		Yes
8223	19	Ovary	SEH	++		Yes	MC
8229	13	Dermatitis	ND	++		Yes	MC, ADH
8231	23	Lymphoma, ovary	SEH	++	Yes	Yes
8245	16	Ovary	None	+++		Yes	DC
8256	23	Lymphoma, ovary	SEH, EH	++	Yes	Yes	MC
8286b	13	Lymph node
8344	17	Lymphoma, lung	SEH, EH	+		Yes
8356	19	Mammary gland, ovary	SEH, EH	++++	Yes	Yes	MC, DCIS, ADH, SM, DC

^aSEH, squamous epithelial hyperplasia; EH, endometrial hyperplasia; MC, microcalcification; DCIS, ductal carcinoma in situ; ADH, atypical ductal hyperplasia; SM, squamous metaplasia; DC, ductal carcinoma; ND, not determined.

^b8286 died due to a lymph node tumor and no fresh tissues could be collected for further analysis.

^cThe number of plus signs indicates the severeness of hyperplasia.

Eight out of nine mice showed mammary gland hyperplasia with very dense branches and hyperplasic and/or neoplastic foci (Fig. 3B and C). Histological analysis revealed that the mammary glands of mutant mice had several abnormalities (Table 1; Fig. 3 and 4). Mammary ducts were commonly dilated and filled with pasty materials such as amorphous debris, lipids, crystal materials, and milk, and the epithelial cells were detached from the ducts (Fig. 3E; also see Fig. S1 in the supplemental material). The diameters of mammary ducts near the lymph node were two to ten times more dilated than those of wild-type littermates (Fig. 3F). Moreover, the mammary glands of many mutant mice showed microcalcifications (Fig. 4A), atypical ductal hyperplasia (Fig. 4B), ductal carcinoma in situ (DCIS) (Fig. 4C), infiltrating ductal carcinoma (Fig. 4D), adenosis (Fig. 4E), and squamous metaplasia (Fig. 4F). In contrast, analysis of 10 age-matched female wild-type littermates did not reveal any of these abnormalities.

FIG. 3.

Brca1^FL/FL mice develop mammary gland hyperplasia and enlarged ducts. (A to C) Whole-mount staining of mammary glands from wild-type (A) and Brca1^FL/FL (B, C) mice. The right panels represent magnifications of the boxed areas. (D, E) Sections of mammary glands from wild-type (D) and Brca1^FL/FL (E) mice. The mutant mammary gland demonstrates cystic dilated status. (F) Diameters of the largest ducts near the distal ends of the lymph nodes in three wild-type glands and nine mutant glands. Cross lines in panels A to C indicate the cutting positions. One of the largest ducts was measured from each gland. There is a significant difference between wild-type and mutant glands: wild type, 68.6 ± 14.8 μm; mutant, 437 ± 300 μm; P = 0.019. Bars, 100 μm. WT, wild type.

FIG. 4.

Brca1^FL/FL mice develop mammary gland hyperplasia and tumors. (A) Mutant mammary glands displayed enlarged ducts containing crystalline bodies in the lumen (arrows). (B to D) Mutant glands displayed enlarged ducts with increased atypical ductal epithelial cells (arrowhead, B), ductal carcinoma in situ (arrowhead, C), and infiltrating ductal carcinoma (D). Arrows indicate microcalcifications (A, B, C). (E, F) Mutant mammary glands also displayed adenosis (E) and squamous metaplasia (F). Bars, 50 μm.

Absence of BRCA1-Δ11 does not affect cell cycle checkpoints activated by DNA damaging upon γ-irradiation.

We next investigated possible causes for the defects displayed by Brca1^FL/FL mice. Several lines of evidence indicated that BRCA1 mediates a dramatic DNA damage response upon γ-irradiation. To determine whether Brca1^FL/FL mice are hypersensitive to γ-irradiation, we irradiated the Brca1^FL/FL mice. Our data revealed no difference in survival rate between wild-type and Brca1^FL/FL mice (Fig. 5A), suggesting that both types of mice had similar responses to DNA damaging induced by γ-irradiation. To understand this at the cellular level, we performed cell cycle analysis using MEFs derived from Brca1^FL/FL and control embryos. After treating these cells with γ-irradiation, we found that Brca1^FL/FL MEFs did not show any obvious defects in several cell cycle checkpoints analyzed, including the intra-S, G₂/M, and G₁/S cell cycle checkpoints (Fig. 5B to D). There were also no significant differences in the expression patterns of several irradiation-regulated proteins between mutants and wild-type controls (Fig. 5E). These results suggest that the BRCA1-Δ11 is not involved in acute response to DNA damage induced by γ-irradiation both in vitro and in vivo.

FIG. 5.

Absence of the BRCA1-Δ11 isoform does not affect response to γ-irradiation in vivo and in vitro. (A) Kaplan-Meier survival curve of 4-week-old Brca1^FL/FL mice (n = 13) and wild-type littermates (n = 12) after exposure to 10 Gy of irradiation. (B) DNA synthesis of wild-type and Brca1^FL/FL MEFs 1 h after irradiation. (C) The mitotic index of irradiated or unirradiated wild-type and Brca1^FL/FL MEFs was determined 2 h after 10-Gy irradiation. Data (B, C) are expressed as the percentage of unirradiated wild-type controls; means ± standard deviations are from triplicate experiments. (D) Representative histograms showing DNA content of wild-type or Brca1^FL/FL MEFs at P1 24 h after 10-Gy irradiation. Cells were treated with nocodazole immediately after irradiation, and they were harvested by trypsin and analyzed by flow cytometry with DAPI staining. (E) Protein expression patterns of wild-type and Brca1^FL/FL MEFs upon 10-Gy irradiation. WT, wild type; MT, mutant; IR, irradiation; CTL, control.

Brca1^Fl/FL cells displayed decreased G₁/S accumulation after prolonged culture.

The observation that mammary gland abnormalities appeared in the aged population of Brca1^FL/FL mice suggests that the absence of BRCA1-Δ11 may only have a subtle influence on mice of younger age and that the accumulated effect of BRCA1-Δ11 deficiency eventually results in morphologically distinct phenotypes. To assess the possible impacts of the BRCA1-Δ11 deficiency on later passages of cultured cells, we cultured Brca1^FL/FL and the wild-type MEFs by following an NIH 3T3 procedure for more than 10 passages and determined their DNA content by fluorescence-activated cell sorting analysis. Our analysis detected a statistically significant reduction of 2n population in later passages of mutant MEFs (Fig. 6A). The reduction of the G₁ population was seen in the P5 generation of MEFs and became greater with the number of passage increase (Fig. 6B). The reduction of G₁ accumulation in Brca1^FL/FL mutant cells might be caused by a weak arrest at G₁ or a strong arrest at G₂/M that reduces the number of cells returning to the G₁ phase in the cell cycle. To investigate this, we treated mutant and control MEFs at passage 1 with inhibitors that arrest cells at different phases of the cell cycle (Fig. 6C and D). We showed that the untreated MEFs, irrespective of their genotypes, exhibited no obvious differences in their distribution within the cell cycle. After treatment with mimosine and nocodazole, which prevents cells from escaping from the G₁ and G₂/M phases, respectively, both types of cells exhibited no differences in their distribution in the cell cycle (data not shown). In contrast, upon treatment with aphidicolin, an early S checkpoint-arresting reagent, Brca1^FL/FL MEFs showed a faster reduction of G₁ population of cells (Fig. 6C) than did control MEFs, indicating that Brca1^FL/FL MEFs progressed through the G₁ phase significantly faster than wild-type cells and stayed in early S checkpoint (Fig. 6C, and summarized in Fig. 6D).

FIG. 6.

Absence of BRCA1-Δ11 showed reduced G₁ accumulation and centrosome amplification with increased levels of cell cycle regulatory proteins. (A) Representative histograms of fluorescence-activated cell sorting showing DNA content of passage 1 (P1) and P10 of wild-type or Brca1^FL/FL (FL) MEFs 3 days after plating. 2n, diploid genome; 4n, tetraploid genome; 8n, octaploid genome. (B) DNA contents of wild-type and Brca1^FL/FL MEFs at P5 and P10. (C) The DNA contents of P1 wild-type and Brca1^FL/FL MEFs upon the treatment of aphidicolin (1 μg/ml). Cells at the G₁/S border and early S phase are marked by blue and red arrows, respectively. (D) Percent cell distribution in the cell cycle after the treatment. (E) Western blot analysis of cell cycle regulatory proteins in P1 and P10 of MEFs. (F-I) Number of centrosomes per nucleus in wild-type (G) and Brca1^FL/FL (H, I) MEFs at P10. The DAPI (blue)- and pericentrin (green)-stained images are merged. The centrosome numbers from at least three MEFs for each genotype are summarized in panel F. Small arrowheads point to centrosomes in panels G to I. (J) Acute suppression of cyclin E by RNAi in three immortalized Brca1^FL/FL MEF cell lines. Numbers indicate the individual cell lines we used. C and E represent mock and cyclin E-specific small interfering RNA transfections, respectively. (K) The centrosome number was counted from three cell lines in the absence or presence of cyclin E knockdown. “3<” indicates cells containing three or more centrosomes. Statistically significant differences (P > 0.01) are indicated by asterisks. WT, wild type; CTL, control.

Because Brca1^FL/FL cells exhibited partial loss of their ability to accumulate cells in the G₁ phase, we next tested the possibility that the deletion of BRCA1-Δ11 would affect expression of cell cycle regulatory proteins. As shown in Fig. 6E, there was no difference in protein levels in the early passage. However, mutant MEFs showed significantly higher levels of cyclin A and cyclin E in the later passage than did wild-type controls. In normal dividing cells, cyclin A and cyclin E play essential roles in regulating the transition from the G₁ phase to the S phase. In this regard, it was recently reported that a high level of cyclin E protein accelerates transition through the G₁ phase and increases tetraploid population in a passage-dependent manner (25).

These results suggest that decreased accumulation of the G₁ population in Brca1^FL/FL MEFs may be caused by increased levels of cyclin E. In addition, there is increasing evidence that links cyclin E to the centrosome duplication via its substrates nucleophosmin B23 and CP110 (28, 31). We therefore counted the number of centrosomes in control and Brca1^FL/FL MEF cells after staining these cells with an antibody against pericentrin. We found that 45.5% of the Brca1^FL/FL MEF cells contained more than one centrosome per nucleus, in contrast to 13.7% of the wild-type MEF cells (Fig. 6F to I). In addition, there were 20-fold more Brca1^FL/FL MEFs than control cells that contained more than two centrosomes per nucleus (5.5% versus 0.29%). To further study the relationship between cyclin E and centrosome amplification associated with the loss of Brca1-Δ11, we performed RNA interference-mediated knockdown of cyclin E in Brca1^FL/FL cells and counted the centrosome numbers. Our data indicated that after transfection of RNAi that is specific for mouse cyclin E, levels of fast-migrating cyclin E were significantly reduced (Fig. 6J) and the population of cells containing more than two centrosomes was reduced from 24% to 12% (Fig. 6K). These results suggested that alteration of cyclin E expression upon loss of Brca1-Δ11, at least in part, affects centrosome duplication.

Cyclin E accumulation, increased cell proliferation, and apoptosis in hyperplastic mammary glands of Brca1^FL/FL mice.

Because BRCA1-Δ11-deficient MEFs exhibited higher levels of cyclin E in later passages, we tested whether the hyperplastic phenotype of the Brca1^FL/FL mammary gland is associated with cyclin E accumulation. As shown in Fig. 7A to D, hyperplastic mammary glands expressed cyclin E in both dilated epithelial layers of the duct (Fig. 7B and D), and the lobular region (Fig. 7C and D), while wild-type cells had very low levels of cyclin E (Fig. 7A and D). These results suggest that these cyclin E-positive regions were associated with extra cell activity leading to increased duct size and lobular hyperproliferation of the Brca1^FL/FL mammary gland. To investigate whether the expression of cyclin E was associated with increased cell proliferation, we checked the incorporation of bromodeoxyuridine (BrdU) in the tissues. Immunohistochemical staining with an antibody to BrdU revealed differential staining depending on the different regions. Lobular regions showed higher incorporation (Fig. 7G and H) than the ductal area (Fig. 7F and H). Interestingly, the fast-growing lobular region also exhibited increased levels of apoptosis as revealed by terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling assay (Fig. 7K and L) compared with ductal areas (Fig. 7J and L). These results suggest that the dilated structural change of Brca1^FL/FL mammary ducts may be a result of the overexpression of cyclin E. When ductal hyperplasia invaded the lobular region, it induced apoptosis. These cyclin E staining patterns were found in eight out of nine mutant mice but not in any of the five wild-type age-matched controls (Table 1). Of note, the mutant gland that did not show an increased level of cyclin E also exhibited mammary hyperplasia (Fig. 3B). This observation suggests that some other downstream events, instead of cyclin E, could contribute to the hyperplasia in this mouse.

FIG. 7.

Overexpression of cyclin E and increased cell proliferation in hyperplastic mammary glands of Brca1^FL/FL mice. Immunohistochemical staining of mammary gland sections from wild-type mice at 23 months of age (A, E, and I) and ductal (B, F, and J) and lobular (C, G, and K) areas of three Brca1^FL/FL mice at 13 (B), 23 (C), and 19 months of age (F, G, J, and K). Images of sections were detected by primary antibodies against cyclin E (A, B, and C), BrdU (E, F, and G), and terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) (I, J, and K). Bars, 100 μm. WT, wild type.

DISCUSSION

Alternative splicing occurs in at least 74% of multiexon human genes (20). Currently, three protein products encoded by the BRCA1 locus, i.e., BRCA1-FL, BRCA1-IRIS, and BRCA1-Δ11, have been identified in mammalian cells (12, 42). Although previous gene targeting studies have introduced many mutations into BRCA1 (reviewed in reference 9), targeted disruption of BRCA1 short isoforms has not been reported primarily due to technical limitations. In our continuous efforts to address the biological functions of BRCA1, we specifically deleted BRCA1-Δ11 while leaving BRCA1-FL intact. Our study on Brca1^FL/FL mice indicated that deletion of BRCA1-Δ11 did not interfere with the normal development or the responsiveness to acute DNA damage induced by γ-irradiation. Brca1^FL/FL cells also displayed no obvious defects in any cell cycle checkpoints analyzed. Our in vitro and in vivo data indicated that the prolonged effects of BRCA1-Δ11 deficiency resulted in a number of cellular changes, including increased expression of cyclin A and cyclin E, centrosome amplification, and faster G₁/S transition. Brca1^FL/FL mutant mice exhibited hyperplasia in the mammary gland, the uterus, and the ovary, and they were prone to spontaneous tumor formation.

A comparison between Brca1^FL/FL and Brca1^Δ11/Δ11 mice may provide clues to understanding the biological functions that BRCA1-Δ11 may play. Brca1^Δ11/Δ11 mice die at E12.5 to E18.5 and survive to adulthood in p53^−/− or p53^+/− backgrounds (40). Brca1^Δ11/Δ11 cells grow poorly and are defective in the G₂/M cell cycle checkpoint and the spindle checkpoint (36, 42). In contrast, Brca1^FL/FL cells grow normally and do not show any obvious defects in the several major cell cycle checkpoints analyzed, suggesting that BRCA1-FL may play an indispensable role for these processes while BRCA1-Δ11 does not. Of note, about 25% of Brca1^Δ11/Δ11 cells exhibit abnormal centrosome amplification, while only 5.5% of Brca1^FL/FL cells show this defect, suggesting that BRCA1-Δ11 is involved in, but is not necessarily indispensable for, centrosome duplication. It has been shown that BRCA1 is located in the centrosome and interacts with a variety of proteins that are involved in centrosome duplication, including BRCA2, CDK2-cyclin A, CDK2-cyclin E, Gadd45, p21, p53, Rb, nucleophosmin B23, γ-tubulin, and BARD1 (9). The interaction between BRCA1 and γ-tubulin, a major component of the centrosome, is mediated by an unspecified domain in exon 11 (16), providing a molecular basis for the differential effects of BRCA1-FL and BRCA1-Δ11 on centrosome duplication. On the other hand, it was shown that BRCA1/BARD1-dependent ubiquitination of γ-tubulin is essential for regulating centrosome number, and both the BRCA1-RING domain and the carboxyl terminus of BRCA1 are necessary for the ubiquitination of γ-tubulin (32). This observation provides a basis for the involvement of BRCA1-Δ11 in centrosome duplication, as it contains both of these domains.

In therapy, our knock-in strategy should also abolish the BRCA1-IRIS, which was recently found in human cells (12). Conceivably, the phenotypes observed in Brca1^FL/FL mice could be due to a combined loss of BRCA1-Δ11 and BRCA1-IRIS. A unique feature of BRCA1-IRIS is its 34 triplets in the beginning of intron 11, which can be specifically targeted. Unless this is done, the involvement of BRCA1-IRIS in the BRCA1-FL/FL phenotype cannot be reliably assessed. Interestingly, when examining expression of full-length BRCA1 in three immortalized MEF cell lines and a lymphoma from a Brca1^FL/FL mouse by Western blot analysis, we found that the full-length BRCA1 was not detected in the tumor of the Brca1^FL/FL mouse, while it was presented in the immortalized cells as a number of weaker, yet highly phosphorylated, forms (see Fig. S2 in the supplemental material). BRCA1 is known to be phosphorylated upon DNA damage. These results suggest that loss of BRCA1-Δ11 may eventually affect full-length BRCA1, which also contributes to tumorigenesis in Brca1^FL/FL mice.

Brca1^FL/FL mutant mice displayed accumulation of cyclin E on the epithelial cells of hyperplastic ducts. Our data also revealed a correlation between the accumulation of cyclin E and the faster progression of cells through the G₁ phase, suggesting a role for BRCA1-Δ11 in this process through its affect on cyclin E expression or stability. It has been demonstrated that constitutive cyclin E overexpression in both immortalized rat embryonic fibroblasts and human breast epithelial cells results in chromosome instability with abnormal G₁/S-phase transition (31). A strong correlation has been established between elevated cyclin E expression and mammary cancer. Transgenic mice that overexpress human cyclin E in the mammary gland under the control of the ovine β-lactoglobulin promoter developed mammary carcinomas with latencies ranging from 8 to 13 months (3). Thus, we believe that the accumulation of cyclin E in Brca1^FL/FL mutant mice may, at least in part, account for the hyperplasia of gynecologic glands and spontaneous tumor formation in these mice.

Cyclin E-positive staining in breast cancer is considered as one of the determinants of the virulence and metastatic potency. The hazard ratio for death from breast cancer for patients with high total cyclin E levels compared with those with low total cyclin E levels is 13.3, or about eight times as high as the hazard ratios associated with other independent clinical and pathological risk factors (21). In a recent study, Chappuis et al. (6) showed that a high level of cyclin E is characteristic of BRCA1-related breast cancer and serves as a marker of poor prognosis. Our data uncovers an important function of BRCA1-Δ11 in preventing cyclin E1 accumulation, and therefore repressing abnormal cell growth and tumorigenesis in aged populations.

More than half of aged Brca1^FL/FL mutant mice also exhibited mammary gland hyperplasia with microcalcification, which are specks of calcium. Microcalcifications are the most common mammographic sign of (DCIS) and it is associated with almost 90% of cases of DCIS (13). The precise reason for the accumulation of calcium and microcalcification in the Brca1^FL/FL mammary glands is not clear. Studies in MCF-7 breast cancer cells indicated that estrogen signaling could trigger the release of Ca²⁺ from intracellular stores (19). Previous investigations indicated that BRCA1 inhibits estrogen synthesis and estrogen/estrogen receptor alpha signaling (14, 17). Thus, the increased estrogen/estrogen receptor alpha signaling in Brca1 mutant mice could be a factor that contributes to the microcalcification, although it does not rule out the involvement of other hormones, such as prolactin, that could also induce secretory activity of mammary epithelial cells. Because at least 30% of human BRCA1 mutation carriers do not develop mammary tumors at 70 years of age (29), Brca1^FL/FL mutant mice, which have a normal life span and develop mammary hyperplasia and tumors at late life, should be a good model for studying the histopathological changes and long-term effects of risk factors in BRCA1-associated tumorigenesis. This may allow for the development of new and effective therapies for this disease. In addition, because alternative splicing occurs in most multiexon mammalian genes and is largely responsible for the functional complexity of vertebrates relative to invertebrates, our strategy to specifically disrupt splicing variants should have a wide application in future gene function studies.