Targeted Inhibition of Osteopontin Expression in the Mammary Gland Causes Abnormal Morphogenesis and Lactation Deficiency*

Osteopontin (OPN) is a sialic acid-rich, adhesive, extracellular matrix (ECM) protein with Arg-Gly-Asp cell-binding sequence that interacts with several integrins, including αvβ3. Since the ECM is a key regulator of mammary gland morphogenesis, and mammary epithelial cells express OPN at elevated levels, we sought to determine whether this protein plays a role in the postnatal mammary gland development. By generating transgenic mice that express OPN antisense-RNA (AS-OPN mice) in the mammary epithelia we achieved suppression of OPN production in this organ. The pregnant AS-OPN mice displayed a lack of mammary alveolar structures, a drastic reduction in the synthesis of β-casein, whey acidic milk protein, and lactation deficiency. In agreement with these findings, we uncovered that a mammary cell line, NMuMG, which undergoes both structural and functional differentiation on ECM-coated plates, when transfected with an antisense OPN-cDNA construct, failed to undergo such differentiation. Furthermore, the results of gel-invasion assays demonstrated that these cells manifest elevated matrix metalloproteinase (MMP) activity when OPN expression is significantly reduced. The identity of this proteinase as MMP-2 is confirmed by Western blotting, zymography, and inhibition of its activity by a specific inhibitor, TIMP-2. Taken together, our results demonstrate, for the first time, an essential role of OPN in mammary gland differentiation and that the molecular mechanism(s) of its action, at least in part, involves down-regulation of MMP-2.

Osteopontin (OPN) 1 is a secreted, sialic acid-rich, adhesive, glycophosphoprotein, that contains the arginine-glycine-aspartic acid (RGD) cell-binding sequence found in many extracellu-lar matrix (ECM) proteins. Molecular cloning and characterization of a cDNA encoding rat OPN revealed that this protein is composed of 317 amino acid residues, with a predicted molecular mass of approximately 32 kDa for the peptide backbone (reviewed in Ref. 1). However, OPNs isolated from or secreted by various tissues exhibit electrophoretic mobilities consistent with a protein of apparent molecular mass between 44 and 75 kDa (reviewed in Ref. 2). This variability is most likely due to the post-translational modifications (3). The murine OPN is encoded by a single copy gene (4) that is composed of seven exons and spans more than 8 kb of nucleotide sequence (5). It maps to the ric (Rickettsia resistance gene) locus on mouse chromosome 5, and a possible allelism between the opn and ric has been reported (6). In several cell types, OPN interacts with specific integrins, namely ␣ v ␤ 3, ␣ v ␤ 5 , and ␣ v ␤ 1 , and also with the CD44 receptor in RGD-dependent manner (reviewed in Ref. 7). The OPN-integrin interaction mediates important cell functions, most of which are RGD-dependent, including the promotion of cell adhesion and spreading on substrata, chemotactic and haptotactic activities, and cellular signaling (8 -13). Furthermore, the expression of OPN-mRNA and protein is suppressed in cells derived from mice generated by targeted disruption of the protooncogene, src (14). It has also been reported that src regulation of OPN production is due to the presence of a src response element in the promoter region of the OPN gene (15).
Although OPN is expressed in several organs, the epithelium of the mammary gland overexpresses this protein specifically during pregnancy and lactation, and is abundantly present in milk (16). It has now been well established that ECM at the stromal-epithelial interface plays a pivotal role in growth, survival, and differentiation of mammary epithelial cells, and unlike most tissues and organs, a major part of the mammary gland development occurs post-natally (17). Since OPN is an ECM protein, and mammary epithelial cells express this protein at elevated levels (18), we sought to determine whether this protein plays a role in the post-natal development of the mammary gland, including ductal branching, lobuloalveolar structure formation, and lactation. We rationalized that a lack of or significantly reduced levels of OPN synthesis may impair such processes. To test this hypothesis, we generated transgenic mice expressing OPN antisense RNA, specifically in the mammary epithelia, under the regulation of MMTV-LTR promoter/enhancer. This promoter was chosen because it directs mammary tissue-specific expression of transgenes at the onset of puberty without exogenous hormone supplement (19). We show here that suppression of OPN synthesis caused a virtual absence of lobuloalveolar structures, drastic reduction of ␤-casein and whey acidic protein (WAP) synthesis in mammary epithelia, and lactation deficiency. Similarly, in in vitro exper-iments, a mammary epithelial cell line, NMuMG, that undergoes structural and functional differentiation in culture, including the formation of lobuloalveolar structures and synthesis of milk proteins, failed to do so when they were transfected with an OPN antisense cDNA construct. Furthermore, these transfected cells showed enhanced gel invasion, and synthesized elevated levels of metalloproteinase-2 (MMP-2). Taken together, these results define an essential role of OPN in mammary gland development and differentiation, and raise the possibility that the molecular mechanism(s) of OPN action, at least in part, involves the down-regulation of MMP-2.

EXPERIMENTAL PROCEDURES
Production and Characterization of Transgenic Mice-Mice were housed under 10-h dark and 14-h light cycles, and were handled in accordance with approved protocols. To generate the transgene construct, a full-length murine OPN cDNA (20) was ligated in the antisense orientation into SalI site of the eukaryotic expression vector pMAMneo (CLONTECH). The linearized 7.4-kb NdeI-ApaI (AS-OPN) fragment was injected into single-cell embryos of B6XSJL mice, which were then introduced into the oviducts of pseudopregnant CD1 mice (DNX Inc.). Transgenic mice were identified by Southern blot analysis of BamHI-digested genomic DNA from tail biopsy using mouse OPN cDNA, and neomycin gene-specific probes, respectively. In some instances, mice were also genotyped by PCR amplification using murine OPN cDNA-specific primers (MOP-L, 5Ј-TCA CCA TTC GGA TGA GTC TG-3Ј; MOP-R, 5Ј-ACT TGT GGC TCT GAT GTT CC-3Ј).
Whole-mount and Histological Analysis of the Mammary Glands-For whole-mount analysis (21), the second inguinal mammary glands were removed, kept flat in tissue cassettes, and fixed overnight in 4% buffered paraformaldehyde at 4°C. The tissues were then dehydrated, delipidated, stained with iron hematoxylin, cleared in xylene, mounted, and photographed. For histological analyses, 5-7-m sections of paraffin-embedded tissues were stained with hematoxylin/eosin and photographed.
RNA Extraction and Analysis-Total RNA was purified from freshly isolated mouse tissues and from actively growing cultured cells as described (22). RNA (10 g/lane) was electrophoresed on 1.2% agaroseformaldehyde gels, transferred to nylon membranes, and hybridized to 32 P-labeled sense or antisense OPN riboprobes transcribed from pGEM4 -2ar plasmid (20) using T7/SP6 RNA labeling kit following the instructions of the manufacturer (Promega). Hybridization was carried out overnight at 55°C in hybridization solution (50% formamide, 6ϫ SSPE, 5ϫ Denhardt's solution, 0.5% SDS, and 100 mg/ml salmon sperm DNA) containing 1 ϫ 10 6 cpm/ml of the appropriate probe. After hybridization, membranes were washed twice, 5 min each, at room temperature with 2ϫ SSC containing 0.5% SDS, twice at room temperature for 20 min with 2ϫ SSC containing 0.1% SDS, twice at 65°C for 30 min, with 0.1ϫ SSC containing 0.5% SDS, and once for 30 min at room temperature with 0.1ϫ SSC and autoradiographed. For RT-PCR analysis, 200 ng of total RNA were used for the reverse transcription and PCR amplification using GenAmp RT-PCR kit (Perkin Elmer) and the primers MOP-L and MOP-R to reverse-transcribe antisense and sense RNA, respectively. The amplified cDNA fragments were resolved by electrophoresis on agarose gels, transferred to nylon membranes, and probed with digoxigenin-labeled OPN oligonucleotide probe MOP-P (5Ј-AGA GCG GTG AGT CTA AGG AG-3Ј) and detected by using alkaline phosphatase-conjugated anti-digoxigenin antibody and chemilumigraphy.
Protein Extraction and Western Blot Analysis-For Western blot analysis, mammary glands were first flash-frozen in liquid nitrogen, pulverized, and homogenized in five volumes of extraction buffer (100 mM KCl, 10 mM Hepes, pH 7.5, 5% glycerol, 10 mM EDTA, 1% Triton X-100, 0.1% SDS, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 20 g/l aprotinin, 2 g/l leupeptin, and 1 g/l pepstatin). After clearing, aliquots of the homogenates containing ϳ20 g of protein were resolved by SDS-PAGE and transferred to nitrocellulose membranes, which were incubated with anti-OPN antiserum (OST-1; 1:200; Ref. 23) for 2 h at room temperature. The membranes were then washed three times with Tris-buffered saline containing 0.25% Tween 20 for 15 min each at room temperature. Bound primary antibody was detected with alkaline phosphatase-conjugated goat anti-rabbit antiserum (Sigma) using nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate substrate. The membranes were then washed thoroughly with water and photographed. For the detection of MMP-2, semiconfluent cultures of non-transfected, mock-transfected, and OPN antisense cDNA-transfected NMuMG cells were lysed in radioimmune precipitation buffer, and aliquots containing equal amounts of protein (20 g) were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membranes. The membranes were blocked by placing overnight with Tris-buffered saline with Tween 20 containing 1% bovine serum albumin at 4 C, followed by incubation with rabbit anti-MMP-2 antibody (1:1000; Chemicon) for 1 h at room temperature. Antibody detection was carried out as described above.
Differentiation of NMuMG Cells on Matrigel-Matrigel TM (basement membrane from Engelbreth-Holm-Swarm tumors) was obtained from Collaborative Biomedical Products and used to coat 35-mm culture dishes (400 l/dish). Matrigel was allowed to gel by incubation for 1 h at 37°C. The cells were then plated (3.5 ϫ 10 5 cells/dish) in DMEM containing 10% fetal calf serum, 3 g/ml bovine prolactin (Sigma), 5 g/ml insulin (Sigma), 1 g/ml hydrocortisone (Sigma), and hepatocyte growth factor at concentration of 20 ng/ml. Cells were grown for 6 days to allow morphological differentiation to occur.
Reverse Transcription PCR-For the analysis of expression of milk proteins, an RT-PCR kit from Promega was used. Ten g of total RNA was used for each reaction. For ␤-casein, the primers used are: CAGT-GAGGAATCTGTTGAA (forward primer) and GGTCTTGGACAGA-CATATC (reverse primer). For WAP, the primers used are: 5Ј-TAG CAG CAG ATT GAA AGC ATT ATG-3Ј (RT primer) and 5Ј-GAC ACC GGT ACC ATG CGT TG-3Ј (PCR primer). The first strand cDNA synthesis was carried out with avian myeloblastosis virus RT (avian mammary tumor virus reverse transcriptase) at 48°C for 45 min. The avian myeloblastosis virus RT inactivation was carried out by heat denaturation at 94°C for 2 min. Amplification was carried out in 40 PCR cycles consisting of a denaturation step of 30 s at 94°C, an annealing step of 1 min at 60°C, and an extension step of 2 min at 68°C.
Gel Invasion Assay-For gel invasion assay, invasion chambers containing inserts coated with Matrigel (Biocoat) were used as suggested by the manufacturer (Becton Dickinson). Cells were serum-starved for 4 h, harvested by trypsinization, and plated in the upper chambers at a density of 3.5 ϫ 10 5 cells/well, in serum-free medium. Growth medium containing 10% fetal bovine serum was placed in the lower chamber to serve as a chemoattractant. After 16 h of incubation at 37°C, the cells attached to the upper surface of the membrane along with the layer of Matrigel coating were removed using a cotton-tipped swab. The cells that migrated through the filter and attached to the lower surface of the membrane were fixed with 10% methanol, stained with Giemsa, and counted under a phase contrast microscope. For invasion assays in presence of the MMP-2 inhibitor TIMP-2, the upper surface of the membranes of the Transwell inserts (Costar) were coated with 10 l/ cm 2 Matrigel containing 10 g/ml human TIMP-2 (Calbiochem). After hardening of the Matrigel layer at 37°C, the cells were plated and the assays were carried out as described above.
Zymography-For establishing the proteolytic profiles of NMuMG cells by zymography (24), cell lysates were prepared in substrate gel sample buffer (0.1% SDS, 4% sucrose, 0.25 M Tris-HCl, pH 6.8, and 0.1% bromphenol blue), cleared by centrifugation, and electrophoresed in 10% SDS-PAGE gels containing 1 mg/ml gelatin. After electrophoresis the gels were soaked twice for 30 min each at room temperature in 2.5% Triton X-100, with gentle shaking. The gels were rinsed in distilled water and incubated overnight at 37°C in substrate buffer (50 mM Tris-HCl, pH 8.0, 5 mM CaCl 2 , and 0.02% NaN 3 ). After incubation, gels were stained with 0.5% Coomassie Blue, destained with water, and photographed. The areas of gelatinolytic activity appeared as light translucent bands over a blue background.

Generation of AS-OPN Mice-
The construct used for the generation of AS-OPN transgenic mice contained full-length OPN cDNA in the antisense orientation downstream from the MMTV-LTR promoter of the eukaryotic expression vector pM-AMneo (Fig. 1A). The biological activity of this transgene construct was first tested by transfecting the murine cell lines MC3T3E1 (25) and JB6 (26). Transfected cells expressed OPN antisense RNA which efficiently suppressed OPN-protein synthesis (Ref. 27; data not shown). This construct was, therefore, used to generate transgenic mice. Six founder mice were obtained, two of which were used to establish two independent transgenic mouse lines. Southern blot and PCR analyses confirmed the genomic integration and germline transmission of the transgene to the progeny of both founders. Southern blot analysis of BamHI-digested genomic DNA of one line of AS-OPN transgenic mice using OPN cDNA and neomycin gene probes showed the presence of 1.3-and 2.6-kb bands, respectively ( Fig. 1, C and B). The Southern hybridization signal patterns were found to be different for the two lines when genomic DNA from both were digested with ClaI and hybridized with the same OPN cDNA probe (data not shown), confirming different integration site of the transgene in the two mouse lines.
Expression of OPN Antisense RNA in the Mammary Glands of AS-OPN Mice-To detect transgene expression, total RNA isolated from mammary glands of 65-day-old virgin mice was subjected to RT-PCR analysis. Fig. 2A shows that the OPN antisense RNA is expressed in the mammary glands of transgenic, but not in those of the normal control mice, although OPN-mRNA is expressed in both normal and transgenic mice. To determine whether the expression of the transgene is more pronounced in mammary tissue due to the presence of MMTV-LTR promoter, total RNA was isolated from several organs including the mammary gland, kidney, spleen, and liver of the transgenic mice and analyzed by Northern blotting, using OPN antisense RNA probe. The results showed that the OPN antisense RNA expression is significantly higher in the mammary gland, compared with that of other tissues ( Fig. 2B and data not shown). To examine transgene expression in different stages of mammary gland development, RNA was extracted from glands of virgin and pregnant AS-OPN mice at early and late (17 days) gestation, and Northern blot analyses were performed. As shown in Fig. 2C, OPN antisense-RNA is expressed in the mammary glands of virgin transgenic mice, and its expression is increased during pregnancy, consistent with previous reports on the pattern of MMTV-LTR promoter-driven transgene expression in the mammary gland (28). These data confirm the coexpression of endogenous OPN gene and transgene in the mammary glands of AS-OPN mice. To determine whether OPN-protein production is suppressed as a consequence of the expression of OPN antisense RNA in transgenic mice, mammary glands from both 65-day-old virgin normal and AS-OPN mice were homogenized, and protein extracts were analyzed by Western blotting using anti-OPN serum (OST-1). The results show that the OPN levels are drastically reduced in the mammary glands of both lines of AS-OPN mice derived from two independent founders as compared with those of the normal controls (Fig. 2D). These results clearly demonstrate that elevated expression of OPN antisense RNA in the mammary epithelium effectively suppressed the production of OPN protein in this tissue.
AS-OPN Mice Show Lactation Deficiency-The founder transgenic mice did not show any detectable phenotypic abnormality, and developed to the adult stage as normally as their non-transgenic littermates. The female progeny gave birth to normal size litters; the newborn pups appeared normal at birth and did not show any obvious morphological abnormalities. All pups born to approximately 68% of AS-OPN mice (n ϭ 21; Table I) died within 24 h after birth, and approximately 32% of them produced enough milk to support 1-4 pups but not fullsize litters. Close morphological examination of the pups that died after birth revealed the absence of milk in their stomachs. This indicated that death of pups occurred due to starvation, and suggested that either most of the AS-OPN mothers were lactation-deficient or the pups were incapable of suckling. To examine these possibilities, pups born to AS-OPN mice, whose previous litters had died after birth, were allowed to be nursed by actively lactating foster mothers. Eighty-eight percent of these pups survived and grew normally (Table I). These results clearly show that all AS-OPN mothers have lactation deficiency and a vast majority of them are clearly non-lactating.
Abnormal Development of Mammary Gland of AS-OPN Mice-To determine the possible cause of the apparent lactation deficiency in AS-OPN mice, we examined the mammary glands of post-partum animals morphologically and biochemically. The mammary glands of non-lactating AS-OPN mice were 2-5 times smaller in size than those of normal lactating mice (data not shown). Whole mount analysis of these glands showed that, although some alveolar lobules were formed (Fig.  3C), they were greatly reduced in size and number as compared with those of the lactating non-transgenic mice (Fig. 3A). Fig.  3B shows histological sections of the normal gland where the alveolar lobules almost completely filled the fat pad, leaving little stromal space. In addition, they were also arranged in well organized, tightly packed structures with large luminal spaces (Fig. 3B). In contrast, the mammary glands of nonlactating AS-OPN mice revealed only few alveolar structures, sparsely distributed in individual patches throughout the fat pad, some in collapsing state, and with little or no luminal spaces (Fig. 3D). In order to determine whether mammary glands of AS-OPN mice are capable of synthesizing milk proteins, we examined the expression of ␤-casein and WAP mRNAs by RT-PCR using total RNA from the mammary tissues of AS-OPN and non-transgenic mice at 16 days of gestation. Fig. 3E shows that while the RNAs of both proteins are abundantly expressed in mammary tissues of non-transgenic mice (N), they are virtually undetectable in the mammary tissues of the AS-OPN mice (T). These observations demonstrate that lack or drastic reduction of OPN synthesis in the mammary tissues of AS-OPN mice not only interferes with the development of normal lobuloalveolar structures but also affects the functional differentiation of the mammary epithelia, attested by a virtual lack of synthesis of milk proteins ␤-casein and WAP.
Impaired Morphogenesis and Functional Differentiation in Vitro of Mammary Epithelial Cells Transfected with Antisense OPN cDNA-It has been well established that morphological and functional differentiation of mammary epithelial cells can be induced in vitro (29). Therefore, an in vitro system, using mammary epithelial cells, offers a unique opportunity to further investigate the effects of down-regulation of OPN synthesis in these cells on their structural and functional differenti-ation. In the present study, we used a normal murine mammary epithelial cell line, NMuMG, that differentiate in vitro. NMuMG cells were transfected with the same construct used to generate the AS-OPN mice. The mock-transfected (vector only) and non-transfected cells served as controls. Several stably transfected clones were isolated, and two of these clones, AS6 and AS8, expressing the lowest levels of OPN were characterized and chosen for further analysis. As shown in Fig. 4A, both AS6 and AS8 clones stably incorporated the antisense OPN cDNA, which is absent in non-transfected and mocktransfected cells. RT-PCR analysis showed that OPN antisense RNA is expressed by both of these clones (Fig. 4B). As a result, there was a drastic reduction in the levels of OPN protein, as compared with non-transfected and mock-transfected NMuMG cells (Fig. 4C). These observations show that expression of antisense OPN-RNA effectively down-regulates the synthesis of OPN in antisense-transfected NMuMG cells.
When non-transfected, mock-transfected, AS6, and AS8 cells were cultured on Matrigel-coated dishes in presence of hormones, non-transfected and mock-transfected NMuMG cells formed compact spherical structures (Fig. 5, A and B). When hepatocyte growth factor was added to the cultures, many of those spheroids formed long branching tubules (Fig. 5E). Histological analysis showed that many of these spheroids or tubular structures also contained a lumen, resembling the alveoli of the mammary gland (Fig. 5F). However, AS-OPN-transfected AS6 and AS8 cells failed to form such spheroids or branching tubules and remained as monolayer cultures (Fig. 5,  C and D). Thus, NMuMG cells cultured on Matrigel-coated dishes undergo morphogenesis, including the formation of structures typical of mammary glands in vivo, but their counterparts, expressing antisense OPN RNA and showing drastically reduced OPN production, failed to undergo such differentiation. Similar results were also obtained using branching morphogenesis assays in collagen gels (data not shown).
Functional differentiation of NMUMG cells in culture was evaluated by their capacity to synthesize major milk proteins. a The pups were counted on the morning of delivery. The numbers in parentheses indicate the number of litters.
b On the morning of delivery, pups born to non-lactating mothers were transferred to actively lactating mothers. Not all pups born were transferred; those that appeared too weak were not. c The number of surviving pups was determined 2-4 days after birth. d 20 surviving pups were born to 8 AS-OPN mice; the remaining 13 out of 21 were totally non-lactating. It has been reported that synthesis of milk proteins such as ␤-casein and WAP can be induced in mammary epithelial cells in culture (29). Therefore, we examined by RT-PCR whether AS6 and AS8 cells, grown on Matrigel-coated dishes in presence of appropriate hormones, have the ability to express these milk proteins. Non-transfected and mock-transfected NMuMG cells were used as controls. As shown in Fig. 6, while nontransfected and mock-transfected control cells expressed significant amounts of both ␤-casein and WAP mRNAs, the AS6 and AS8 cells did not. These in vitro data, therefore, support our observations in vivo that mammary glands of AS-OPN mice, expressing significantly reduced levels of OPN, fail to form alveolar structures and do not synthesize ␤-casein and WAP mRNAs.
Increased Proteolytic Activity and Invasiveness of Antisense OPN cDNA-transfected NMuMG Cells-It has been well established that ECM at the stromal-epithelial interface is a key regulator of the structural and functional differentiation of the mammary epithelium (16). Extracellular matrix remodeling of the basement membrane by ECM-degrading metalloproteinases are implicated as major determinants for the loss of mammary epithelial function during involution (21). It has been shown that the mammary gland expresses several matrix proteinases and their inhibitors during different stages of its development, and a delicate balance between the expression of these proteinases and their inhibitors controls mammary gland morphogenesis (30). To investigate whether OPN may have a role in regulating metalloproteinases and whether such regulation may explain the observed abnormality of the mammary glands of AS-OPN mice, we determined the metalloproteinase activity in AS-OPN-transfected, mock-, and non-transfected cells by gelatin zymography. The results show that AS6 and AS8 cells express a protease with a molecular mass of 72 kDa, which is present at significantly reduced levels in mock-transfected and non-transfected NMuMG cells (Fig. 7A). The molecular weight of this band closely resembles that of pro-collagenase-A (MMP-2). To determine whether this protein is indeed MMP-2, cell-lysates were subjected to Western blot analysis using an anti-MMP-2 antibody. As shown in Fig. 7B, this antibody detects a protein, the molecular weight and abundance of which resemble those of the band observed in the zymogram (Fig. 7A), indicating an elevated expression of MMP-2 in both AS6 and AS8 cells as compared with mocktransfected and non-transfected cells.
Invasion of the ECM is an important property manifested during mammary gland development and differentiation. Therefore, it is possible that the abnormal in vivo and in vitro morphogenesis of mammary epithelial cells expressing low levels of OPN is due to their increased invasiveness. To test this possibility, AS-OPN cDNA-transfected, mock-, and non-transfected parental NMuMG cells were subjected to gel-invasion assays (31). Fig. 7C shows that AS6 and AS8 clonal cells are more invasive than non-transfected and mock-transfected NMuMG cells. To determine whether the MMP-2 is responsible for the invasive phenotype, the above assays were carried out in presence and absence of TIMP-2, a specific inhibitor of MMP-2. These results show that addition of TIMP-2 results in complete inhibition of invasiveness of AS6 and AS8 cells (Fig.  7C). Taken together, these results demonstrate that increased expression of a metalloproteinase observed in AS6 and AS8 Arrows in E and F show a spheroid structure similar to that shown in A and B, which also contains lumen. Magnification for A and B, ϫ300; for C and D, ϫ100; for E, ϫ30; for F, ϫ250. cells is MMP-2. Further, these data also suggest that suppression of OPN synthesis in mammary epithelial cells leads to increased MMP-2 activity and invasiveness, which in turn may be responsible for the abnormal structural and functional differentiation of the mammary glands of AS-OPN mice, and those of the AS-OPN cDNA-transfected NMuMG cells. DISCUSSION In the present study, we have examined the role of OPN in mammary gland morphogenesis by utilizing transgenic mice with targeted expression of OPN antisense RNA (AS-OPN) and an established normal murine mammary epithelial cell line, NMuMG, in which OPN production is suppressed by transfection with the same AS-OPN construct. The antisense RNA technology has been employed successfully to suppress the expression of various endogenous genes in both prokaryotes and eukaryotes (32). Although the versatility of the antisense RNA has been demonstrated in various studies, there are subtleties inherent to the optimal antisense RNA structure for the inhibition of specific gene expression. Our construct appears to have effectively interfered with OPN expression, specifically in the mammary epithelial cells of AS-OPN mice originating from two independent founders. The antisense-transgenic as opposed to gene-targeting strategy was conceived because of the possibility that OPN null mice may be embryonically lethal or other RGD-containing proteins may functionally compensate for OPN in these mice, consequently obscuring any abnormal phenotype that may develop due to the lack of OPN in situ, in a specific organ, such as the mammary gland. Furthermore, the endogenous OPN gene remains functional in AS-OPN mice during early stages of mammary gland development as the MMTV-LTR promoter-directed antisense transgene expression in mammary epithelium does not occur before puberty without exogenous hormone treatment. Therefore, it may be too advanced a stage for such compensatory mechanism to offset the effect of OPN deficiency in the development of the mammary gland of the transgenic mice.
Several cDNA constructs using the MMTV-LTR promoter/ enhancer to target the expression of transgenes in the mammary gland have been reported (reviewed in Ref. 18). In each of these studies, transgene expression was observed solely or predominantly in the mammary glands of mature virgin females during pregnancy and at the post-partum stage (33). Our present observations are in agreement with those of the previous studies. Although dexamethasone treatment can further enhance transgene expression in the mammary gland and induce its expression in other tissues (34), such treatment was avoided in this study because dexamethasone profoundly affects mammary epithelial cell differentiation in vivo as well as in vitro (35). Therefore, such treatment would have obscured any effect of the AS-OPN transgene on mammary gland development. The possibility that the abnormal development of the mammary gland observed in the present study may have been caused by a cis-acting effect of the MMTV-LTR promoter on mammary specific transcription factors is highly unlikely because the results of each of the studies mentioned above show that the use of the MMTV-LTR promoter results in distinct transgene-specific phenotypes. For example, the transgenic mice showing MMTV-LTR-driven expression of TGF-␤ had hypoplastic mammary epithelia (36), as opposed to hyperplasias or adenocarcinomas developed by the mice overexpressing TGF-␣ (33) or cyclin D1 gene (37), respectively, under the control of the same promoter. Finally, the possibility exists that the observed phenotype of AS-OPN mice is due to a position effect of the transgene integration, causing insertional mutagenesis or activation of adjacent gene(s) involved in mammary gland development. However, this is highly unlikely since an identical phenotype is observed in two transgenic mouse lines independently derived from two founders. These transgenic mouse lines, as well as the NMuMG cells transfected with AS-OPN cDNA, harbor the transgene at different sites in the genome. Thus, the anomalies of the mammary gland development in AS-OPN transgenic mice, and the abnormality of AS-OPN-transfected NMuMG cells observed in the present study, are most likely caused by the down-regulation of OPN synthesis in mammary epithelial cells due to transgene activity.
Using in situ hybridization analysis, we found that transgene expression is confined to the ductal epithelium of the mammary gland, but not in the stromal adipose tissues (data not shown). These results are in agreement with the previous reports (33,36). In addition, our observations show that the transgene is expressed in the same tissue component of the mammary gland that normally expresses OPN mRNA. This is important because co-expression of the transgene with the endogenous gene is a pre-requisite for the antisense RNAmediated inhibition of opn gene expression. Antisense RNA is believed to interfere with gene expression at the level of transcription, processing of transcripts, stability of mature mRNA, or at the level of protein synthesis (38). Our Northern blot analysis shows that the mammary glands of AS-OPN mice produce a normal-sized 1.6-kb OPN transcript. However, Western blot analysis of mammary gland extracts show that OPN protein levels are drastically reduced in AS-OPN mice compared with those of the normal controls. Taken together, these results suggest that expression of AS-OPN RNA interferes with the OPN mRNA translation, and not with its transcription, maturation, or stability.
One of the striking features of most of the AS-OPN mice is their non-lactating phenotype, which in most cases caused death of the newborn pups. Morphological analysis showed that the mammary glands of the non-lactating mice, 24 h post-partum, were mostly rudimentary, and normal alveolar structures were absent. A few that could be detected had a convoluted appearance, were small in size, and appeared to have no central lumen. A small proportion of the transgenic mice produced enough milk to feed only small size litters (1-4 pups) and the glands of these animals showed only a few normal sized alveoli. Variability in OPN levels were observed in glands of AS-OPN mice, which could be due to differential activity of the transgene in individual mice or due to their genetic differences. Such variation may explain why a small percentage of AS-OPN mice could support 1-4 pups.
It has been reported that coordinated expression of matrix metalloproteinases such as MMP-2 and their inhibitors regulate mammary gland development (30). In the present investigation, we observed that antisense OPN cDNA-transfected NMuMG cells have increased MMP-2 activity associated with increased Matrigel invasiveness, although our data do not exclude the role(s) of other proteases nor do they explain how reduced levels of OPN may increase cellular MMP-2 activity. One possibility is that, as MMP-2 and OPN interact with the same ␣ v ␤ 3 integrin (vitronectin receptor) on the cell surface (39,40), regulation of MMP-2 activity by OPN may well be achieved through competition between these two ligands for the same receptor. Since OPN antisense-transfected cells synthesize low levels of OPN, more MMP-2 may be bound to the cell surface receptor where activation of the enzyme is reported to take place (41), resulting in the observed structural abnormalities of the mammary glands in AS-OPN mice. Furthermore, higher levels of MMP-2 detected in cell lysates of transfected cells may also be due to enhanced pericellular binding of this enzyme.
It has been observed in the present study that mammary epithelial cells of AS-OPN mice, and transfected NMuMG cells synthesize significantly reduced levels of ␤-casein and WAP, suggesting that reduced levels of OPN synthesis interferes with the functional differentiation of the mammary epithelia. It has been reported that the interaction of OPN with its receptor initiates cellular signaling as indicated by the transient increase in the cytosolic free calcium (42), and activation of cellular tyrosine kinases such as pp60 src and p129 FAK (12). OPN signaling leads to changes in gene expression as shown by the suppression of inducible nitric-oxide synthase gene expression (43,44), and by enhancement of immunoglobulin gene expression (45). We propose, therefore, that lack of interaction between OPN and its receptor ␣ v ␤ 3 , due to the decreased levels of OPN may lead to suppression of milk protein expression either directly by blocking the activation of signaling pathway, or by increased localization of MMP-2 on the cell surface, causing an extensive degradation of the ECM that interfered with the functional differentiation of the mammary gland. This latter possibility is strongly supported by our in vitro data showing increased invasiveness of AS-OPN cells and their increased proteolytic activity and lack of milk protein expression.
It has been reported that OPN null mice manifest normal embryonic development and the adults are phenotypically normal, but have abnormal wound healing due to an altered assembly of the collagenous ECM (46). It has also been suggested that in OPN-null mice other RGD-containing protein(s) may functionally compensate for OPN during early development and obscure any effect that may arise due to the lack of OPN on the phenotype of these animals. Such compensatory mecha-nisms are known to exist for several classes of proteins, including other RGD-containing proteins (47). In the present study, we targeted the expression of OPN antisense RNA specifically to the mammary gland, the major developmental activity of which occurs post-natally; this organ also expresses OPN at a very high level. The AS-OPN mice are expected to have normal OPN levels during the embryonic to the prepubertal stages because of the presence of functional endogenous opn gene. Since, in the AS-OPN mice, the MMTV-LTR promoter-directed antisense transgene expression in mammary epithelial cells does not start until puberty, even if the expression of the compensatory proteins continues at the post-partum stage, it may not be able to offset the effects of OPN deficiency at a later stage, resulting in the observed abnormal development and impaired functional morphogenesis of the mammary gland of AS-OPN mice. This, we believe, is the possible cause of the discrepancy between our observations and those of Liaw et al. (46). The present study, therefore, demonstrates, for the first time, that down-regulation of OPN synthesis in mammary epithelia is correlated with abnormal mammary gland differentiation, and our AS-OPN mice provide a unique and valid model to further explore the role of this protein in mammary gland development, differentiation, and tumorigenesis.