Depletion of BRCA1 impairs differentiation but enhances proliferation of mammary epithelial cells

Materials and Methods

Adenovirus-Based RNAi Vector Construction. The adenovirus-based RNAi vector was generated by subcloning the transcriptional unit of U6 promoter-BRCA1 or -retinoblastoma (RB) short-hairpin RNAi (shRNAi) (0.4 kb) into pAdTrack plasmid up-stream of the CMV-GFP cassette (1.6 kb) (19, 20). 293T cells were transfected with the recombinant adenoviral plasmid using lipofectin (Invitrogen), and adenovirus with the titer of 10¹⁰-10¹²/ml was collected.

Cell Cultures. Human normal MEC MCF10A was cultured in DMEM/F12 medium (Invitrogen), as described (16), whereas breast cancer cell lines HCC1937 and SKBR3 were cultured in high-glucose DMEM (Invitrogen), as described (16).

3D Morphogenesis Assay. MCF10A cells were infected with adenovirus-RNAi at 20 multiplicities of infection (moi) for 24 h. Approximately 3,500 infected cells per well were seeded in eight-well chamber slides coated with Growth Factor Reduced Matrigel (BD Biosciences) and covered with growth medium supplemented with 2% Matrigel as described (16). The 3D morphogenesis was monitored by fluorescence microscopy/confocal sectioning at 15-h intervals for 2-3 weeks. For experiments using the conditioned medium, medium collected from MCF10A cells infected with luciferase-RNAi virus was used to feed MCF10A cells infected with BRCA1-RNAi or two carcinoma cell lines, HCC1937 and SKBR3. Collection/application of the conditioned medium was performed every 12-15 h for 2 weeks. For cell number counting, cells were recovered from Matrigel after 1 week by digestion with dispase (BD Biosciences), following the manufacturer's instructions. The number of viable cells was measured by using the trypan blue exclusion method.

Fluorescence Imaging. Fluorescence imaging was performed on a Zeiss Axiovert 200 M equipped with Hamamatsu Photonics (Hamamatsu City, Japan) K.K. Deep Cooled Digital Camera (model C4742-80-12AG) by using axiovision 4.2 software (Zeiss). The images of the hollow acinus structures of live or fixed/DAPI-stained cells were captured by the Z-stacking function for serial confocal sectioning at 2-μm intervals.

RNAi-Resistant BRCA1 Plasmid Construction. Cancer-linked point mutations of BRCA1 (Q356R and M1775R) were introduced into the cDNA by using a QuickChange site-directed mutagenesis kit (Stratagene). Within the wild-type BRCA1, the expression of which was driven by a CMV promoter in the CHpL vector (8), the nucleotides targeted by RNAi (nucleotides 385-405, 5′-GGCTACAGAAACCGAGCCAAA-3′) were partially substituted without changing the amino acid sequence (nucleotides 385-405, 5′-GGCTACCGGAATAGGGCCAAA-3′) by site-directed mutagenesis. The RNAi-resistant region was excised at NotI-EcoR and subcloned into each point mutant construct to replace the original sequence.

Microarray. MCF10A cells were infected with luciferase- or BRCA1-RNAi adenoviruses in duplicate at 20 moi for 24 h and seeded at 0.5 × 10⁶ cells in each 60-mm plate coated with Growth Factor Reduced Matrigel (BD Biosciences) and covered with the growth medium supplemented with 2% Matrigel. After 15 h, cells were harvested from Matrigel by digestion with dispase, and RNA was extracted with TRIzol reagent (Invitrogen). cDNA was synthesized and labeled with biotin from 10 μg of the collected RNA and hybridized onto Affymetrix array (54,676 genes), stained with streptavidin-phycoerythrin and analyzed by using gcos 1.2 software (Affymetrix) for multiplex pair-wise comparison provided by the University of California, Irvine, microarray core service. The statistical significance for each gene was evaluated by ANOVA single-factor analysis by using Microsoft excel 2000, and the fold difference >2.3, as well as P value <0.05, was considered significant.

Results

Generation of Adenovirus Expressing the BRCA1- or RB-RNAi. To test whether BRCA1 has a direct role in proliferation and differentiation of MEC, we generated an adenovirus-based vector that expressed shRNA to knockdown the expression of BRCA1 (Fig. 1B Left) under the control of the U6 promoter. Because the RB gene has an essential role in muscle, adipose, and neuronal tissue differentiation (21), RB expression depleted by shRNA (Fig. 1B Right) was used as a comparison. These adenoviral vectors coexpressed GFP as a reporter for infection efficiency (Fig. 1 A). The knockdown potencies of these RNAi vectors were confirmed by Western blot analyses on the target proteins by using HeLa cells. The protein levels of BRCA1 (Fig. 1C Left) and RB (Fig. 1C Right) became undetectable after 24 h of infection with the corresponding adenovirus-RNAi at 20 moi. These adenovirus-based RNAi were used to deplete the expressions of BRCA1 and RB in normal MCF10A cells.

BRCA1-Depleted MEC Failed to Form Acinus Structure. The in vitro acinus is the polarized spherical structure of a single epithelial cell layer surrounding the lumen (16, 17). To capture the events critical for the normal acinus formation, the control, BRCA1-, and RB-depleted cells were plated in the 3D cultures, and their growths were monitored (Fig. 2). After 15 h, control (Fig. 2 Aa1) and RB-depleted (Fig. 2 Ac1) cells exhibited active migratory behaviors, as characterized by outward projections from individual cells, possibly for cell-to-cell communication. In contrast, BRCA1-depleted cells appeared immotile, retaining the original spherical shape of single cells (Fig. 2 Ab1). After 4 days, control (Fig. 2 Aa2) and RB-depleted (Fig. 2 Ac2) cells had started to form localized primordial acinus structures, whereas BRCA1-depleted cells (Fig. 2 Ab2) had started to exhibit horizontal spreading on the surface of Matrigel. After 7 days, control (Fig. 2 Aa3) and RB-depleted (Fig. 2 Ac3) cells had formed almost complete acini with the appearance of the central hollow lumens, as confirmed by confocal sectioning. On the other hand, BRCA1-depleted cells (Fig. 2 Ab3) had formed irregular-shaped aggregates embedded in a lawn of cells. In 20 days, control (Fig. 2Ba) and RB-depleted (data not shown) cells formed the organized acinus structure of a single luminal layer surrounding a hollow lumen. Conversely, the BRCA1-depleted cells formed large irregular-shaped multilobular structures with the completely filled lumen (Fig. 2Bb), similar to that observed by activation of an oncogene ErbB2 (16, 18).

Normal MEC Acinus Formation Is Rescued by RNAi-Resistant Wild-Type BRCA1. To validate that the ablated acinus formation was solely due to the absence of the functional BRCA1, it was tested whether reintroduction of wild-type BRCA1, by means of a plasmid expressing the RNAi-resistant wild-type BRCA1, could rescue the normal phenotype. In this vector, silent mutations were generated within the RNAi target sequence to confer the resistance (Fig. 3 Aa and Ab). The efficiency of the RNAi resistance was confirmed by Western analysis, showing the expression of wild-type BRCA1 was maintained in HeLa cells transfected with the RNAi-resistant BRCA1 vector before the RNAi treatment (Fig. 3Ac). The 3D morphogenesis assay was conducted on MCF10A cells pretreated with the RNAi-resistant wild-type BRCA1 vector then infected with either control luciferase- or BRCA1-RNAi adenovirus (Fig. 3 Ba and Bb). The presence of the RNAi-resistant BRCA1 rescued the normal acinus formation, because the phenotypes of those treated with the control- (Fig. 3Ba1-3) and BRCA-RNAi (Fig. 3Bb1-3) were almost indistinguishable at all time points monitored. The central hollow lumens became evident in 1 week, as confirmed by confocal sectioning. Overall, these results provided evidence that BRCA1 is essential for the normal acinar morphogenesis of MEC.

RNAi-Resistant M1775R Mutant BRCA1 Failed to Rescue the Acinus Formation. To further test that wild-type BRCA1 is essential for acinus formation, we introduced two naturally occurring point mutations into the RNAi-resistant BRCA1 construct, Q356R in the RAD50/ZBRK1-binding region and M1775R in BRCT domain. The 3D morphogenesis assay was conducted on MCF10A cells pretreated with the RNAi-resistant point-mutant BRCA1 vector, then infected with BRCA1-RNAi adenovirus (Fig. 3 Bc and Bd). The RNAi-resistant BRCA1 carrying Q356R (Fig. 3Bc1-3) mutation rescued the normal acinus formation with the appearance of the central hollow lumen in 1 week of growth. On the other hand, the RNAi-resistant BRCA1 the carrying M1775R mutation (Fig. 3Bd1-3) failed to rescue, suggesting that BRCT domain of BRCA1 is important for mediating the acinus formation.

MEC Proliferates in the Absence of Functional BRCA1. Next, to verify whether the loss of the functional BRCA1 leads to the cellular proliferation, cells were recovered from the Matrigel after 1 week of growth, and the number of viable cells was counted (Fig. 3C). The populations of cells that formed the normal acinus structures had cell numbers two to three times that originally seeded, suggesting that cells had divided at least once before committing themselves to differentiate, in a manner comparable to adipocyte differentiation mediated by RB (22). On the other hand, the populations that failed to form the acinus structures had cell numbers six to eight times the original, suggesting these cells continued dividing during the course of the experiment. This result reflects the fact that cellular differentiation and proliferation are opposing signaling events (17), and the presence of the functional BRCA1 seems essential for driving cells into the differentiation pathway.

Loss of BRCA1 at the Initial Stage of Acinar Morphogenesis Up-Regulates the Expression of Proliferation but Down-Regulates Differentiation Genes. To assess the differential gene expression pattern caused by the loss of BRCA1 at the initial stage of acinar morphogenesis, we performed pair-wise microarray analyses on MCF10A cells infected with the control luciferase- or BRCA1-siRNA adenovirus and grown in 3D cultures for 15 h (Table 1). The result confirmed that BRCA1-RNAi largely decreased the BRCA1 level (-7.9-fold). In BRCA1-depleted cells, a mammary differentiation factor STAT5B (23) and a caretaker/cancer susceptibility gene FANCA (24), as well several IFN- or caspase-associated proteins, were down-regulated. Concomitantly, in these cells, certain proliferation markers, including HMGA2 (25), angiopoietin-1 (26), and CaM kinase II β (27), were up-regulated. The data suggest that the gene expression pattern of BRCA1-depleted cells at the initial stage of acinar morphogenesis is overall directed toward proliferation rather than differentiation, supporting our phenotypic observation that the BRCA1-depleted MEC failed to enter the acinus-forming pathway but instead proceeded to proliferate.

Conditioned Medium from Differentiating MEC Rescue the Acinus-Forming Phenotype of Differentiation-Defective MEC. Because the transfection efficiency in MCF10A cells usually reach ≈30-50%, it was surprising to observe that cells transfected with wild-type, as well as Q356R point mutant, BRCA1 rescued the acinus formation phenotype (Fig. 3). Then, we speculated that certain factor(s) released from BRCA1-positive cells might have influenced the morphogenesis of the neighboring BRCA1-negative cells. To test this possibility, the conditioned medium collected from MCF10A cells, which has been infected with the luciferase-RNAi/GFP adenovirus, was used to plate and continually (every 15 h) feed BRCA1-RNAi/GFP adenovirus-infected MCF10A cells in 3D culture. In contrast to BRCA1-depleted cells with the fresh growth medium (Fig. 4Aa1-3), ≈20% of BRCA1-depleted cells fed with the conditioned medium were able to resume the acinus-forming phenotype (Fig. 4Ab1-3). This observation suggests that MCF10A cells secrete certain paracrine/autocrine factor(s) in response to ECM signals, whereas cells depleted of BRCA1 lack such activity.

Next, we tested whether the administration of the conditioned medium from the differentiating MCF10A cells can alter the growth and morphology of breast cancer cells lain in 3D cultures (Fig. 4B). Two different breast carcinoma cell lines were tested: HCC1937 (Fig. 4 Ba and Bb), which expresses a truncated form of BRCA1 at one of the C-terminal BRCT domains (28) and SKBR3 (Fig. 4 Bc and Bd), which expresses a low basal level of BRCA1 (29). Both cancer cells lain in 3D culture with fresh growth medium formed irregular-shaped large aggregates (Fig. 4 Ba1-3 and Bc1-3). Conversely, when these cancer cells were continually (every 12 h) fed with the conditioned growth medium from the differentiating MCF10A cells, a large fraction of the cell population (>40%) formed acinus-like spherical structures (Fig. 4 Bb1-3 and Bd1-3) with the appearance of the central hollow lumen in 1 week. In addition, a significant number of dead cells were observed among those incapable of forming an acinus-like structure (≈60%). After 1 week of growth, the number of viable cells was counted (Fig. 4C). BRCA1-depleted MCF10A cells were ≈40% lower in those treated with the conditioned medium than untreated. Also, the two cancer cell lines tested were at least 60% lower in those treated with the conditioned medium than untreated. These observations suggest that the differentiating MCF10A cells, in response to ECM signals, secrete certain paracrine/autocrine factors, which promote differentiation for the nonmalignant/malignant MEC that have reduced BRCA1 functions.

Discussion

The processes of mammary acinus formation are comprised of a series of molecular events, including ECM signal response, cell migration/communication, aggregate formation, polarity establishment, and hollow lumen formation by luminal cell death (16-18). Although the completion of the entire cycle may take 2-3 weeks, all of the distinctive stages of acinus formation become evident within 1 week of growth in 3D cultures (16, 18). We took advantage of this rapid approach to study the acinar morphogenesis of MEC after BRCA1 was depleted. Our study shows that the loss of the functional BRCA1 leads to the aberrant acinus formation and enhanced proliferation of MEC. The role of BRCA1 in MEC differentiation manifests from the very early step (<15 h), where cells elongate and establish cell-to-cell communications in response, to the ECM signal, the point of the cell fate decision between differentiation and proliferation (17). MEC defective in this critical step at the initial stage of acinar morphogenesis fail to differentiate despite that the BRCA1 level is expected to gradually recover during the course of the experiment. BRCA1 must sit in a position to mediate the differentiation signals transcending from ECM and then respond to them through regulating the expression of specific genes involved in the very early stage of MEC differentiation. Importantly, this entire signal cascade requires an intact BRCT domain of BRCA1, which interacts with several cellular proteins, including CtIP, HDAC, SWI/SNF, and BACH1 (8, 30, 31), and is worth pursuing.

MEC depleted of BRCA1 fail to enter the differentiation pathway but instead proliferate in 3D culture. This result is at odds with the previous notion that cells lacking BRCA1 will fail to repair damaged DNA and accumulate genetic errors leading to cell death (10). However, certain fractions of BRCA1-depleted cells eventually continue proliferating and form a tumor (10). The 3D culture system may supply the best condition that allows MEC to respond to ECM signals for a decision about proliferation, differentiation, or death. It is possible that a failure to enter the acinus-forming pathway may trigger deregulated cellular proliferation, because cells committed to differentiate will be subjected to regulations on the cell cycle/division to define the organoid's polarity, size, and shape (17), whereas cells defective in this process will be precluded from such regulations. Based on this premise, we observed that in BRCA1-depleted cells, certain proliferation markers, including HMGA2, angiopoietin-1, and CaM kinase II β, were up-regulated, whereas a mammary differentiation factor STAT5B and a caretaker/cancer susceptibility gene FANCA were down-regulated. Nevertheless, it has yet to be examined whether BRCA1 is directly involved in the transcriptional regulations of these genes or whether their expression patterns change as the BRCA1 level gradually recovers during the course of the experiment.

Furthermore, surprisingly, the application of the conditioned medium from differentiating MEC can promote differentiation for the nonmalignant/malignant MEC that have reduced BRCA1 functions. This observation suggests that differentiating MEC, in response to ECM signals, secrete certain paracrine/autocrine factors to induce differentiation of the recipient cells. Several paracrine factors, such as Wnt, EGF, IGF, Rank, and TGF family members, are shown to regulate growth and differentiation of MEC (32, 33). Determination of the factors that can induce differentiation of malignant MEC and their link to the BRCA1 signaling pathway is currently under vigorous investigations.

Taken together, our results suggest BRCA1 is involved in the cellular response to the ECM stimuli that drive MEC into a proper differentiation pathway. Once this initial signaling is disrupted, cells proceed to proliferate. Such consequence may, at least in part, account for the malignancy experienced by BRCA1 mutation carriers as well as sporadic breast cancer in which BRCA1 expression is significantly reduced (4).