|Home | About | Journals | Submit | Contact Us | Français|
Fibronectin (Fn) plays an important part in the branching morphogenesis of salivary gland, lung, and kidney. Here, we examine the effect of the conditional knockout of Fn in the mammary epithelium [FnMEp−/−] on postnatal mammary gland development, using Cre-loxP mediated gene knockout technology. Our data show that Fn deletion causes a moderate retardation in outgrowth and branching of the ductal tree in 5-week old mice. These defects are partially compensated in virgin 16-week old mice. However, mammary glands consisting of Fn-deficient epithelial cells fail to undergo normal lobuloalveolar differentiation during pregnancy. The severity of lobuloalveolar impairment ranged from lobular hypoplasia to aplasia in some cases and was associated with the amount of Fn protein recovered from these glands. Decreased rates of mammary epithelial cell proliferation accounted for delayed ductal outgrowth in virgin and lack of alveologenesis in pregnant FnMEp−/− mice. Concomitant decreased expression of integrin β1 (Itgb1) and lack of autophosphorylation of focal adhesion kinase (Fak) suggest that this pathology might, at least in part, be mediated by disruption of the Fn/Itgb1/Fak signaling pathway.
Fibronectin (Fn) is an adhesive glycoprotein that is present in extracellular matrices (insoluble form) and in body fluids (soluble form; e.g., blood plasma). It exists as a dimer composed of nearly identical ~250 kDa subunits linked covalently near their C-termini by a pair of disulfide bonds. Each monomer consists of 3 types of repeats: 12 type I, 2 types II, and 15 type III repeats (Hynes, 1990; Pankov and Yamada, 2002). Although Fn molecules are the product of a single gene, the resulting protein can exist in multiple forms arising from alternative splicing of a single pre-mRNA (~20 variants in humans). Splice sites are EDA (Extra Domain A), EDB (Extra Domain B), and IIICS (type III connecting segment; also called variable [V] region) (ffrench-Constant, 1995). This multi-domain structure harboring binding sites for cell adhesion receptors (e.g., integrins, syndecans, CD44, CD26), matrix macromolecules (e.g., collagens, hyaluronic acid, proteoglycans, tenascin, thrombospondin, fibrinogen), secreted, biologically important molecules (e.g., factor XIIIa, tissue transglutaminase, fibrinogen, heparin), as well as for Fn-Fn self-association makes it a perfect candidate for a multi-task role (Danen and Yamada, 2001; Johansson et al., 1997; Leiss et al., 2008; Wierzbicka-Patynowski and Schwarzbauer, 2003). As such, it anchors the extracellular matrix to the cell cytoskeleton via adhesion receptors and, in doing so, relays signals from the milieu outside the cell to its inside. This so-called outside-in signaling accounts for the many functions, with which Fn has been associated including cell adhesion, differentiation, growth, and migration (Danen and Yamada, 2001; Geiger et al., 2001; Kumar, 1998; Leiss et al., 2008). Thus, Fn is critical in vertebrate development, tissue morphogenesis and homeostasis, wound healing, angiogenesis, and cancer (Cheng et al., 2003; ffrench-Constant, 1995; Geiger et al., 2001; Hynes, 1990; Kumar, 1998; Miyamoto et al., 1998; Pankov and Yamada, 2002). Not surprisingly, Fn knockout is embryologically lethal, with abnormalities becoming manifest first at the onset of gastrulation (for details see George et al., 1993; Watt and Hodivala, 1994).
Many organs, including glands, lung and kidney are formed during embryonic development by epithelial branching through a process of repetitive epithelial cleft and bud formations. In an elegant organ culture study, Yamada and associates (Sakai et al., 2003) showed that Fn is essential for cleft formation during the initiation of epithelial branching in the submaxillary gland. Fn messenger RNA and fibrils were expressed transiently and focally in forming epithelial cleft regions. Knockdown of Fn by small interfering RNA and by inhibition with anti-Fn or anti-integrin antibodies (anti-α5, -α6, and -β1) blocked cleft formation and branching, while exogenous Fn accelerated cleft formation and branching. Although a similar scenario has not been established for the mammary gland, the functional role of ECM proteins, including Fn, collagen type I, and laminin, in mammary gland postnatal development has been studied more recently (Haslam and Woodward, 2003). Among the three ECM proteins only the Fn levels changed appreciably, increasing threefold between puberty and sexual maturity and remaining high during pregnancy and lactation. This temporal expression of Fn coincided with increased expression of select growth factors and AP-1 transcription factors as well as estrogen and progesterone (Fendrick et al., 1998; Haslam and Woodward, 2003; Shen et al., 2006; Wang et al., 1994; Woodward et al., 1998; Woodward et al., 2001; Yang et al., 1995). Together these compounds are involved in ductal development during puberty and mammary epithelial cell proliferation, ductal branching and alveologenesis at sexual maturity and during pregnancy (Fendrick et al., 1998; Woodward et al., 1998), while AP-1 blockade suppresses ductal branching and budding and reduces gland tree size and fat pad occupancy in developing mammary glands by inhibiting expression of AP-1-dependent genes (e.g., Fn, cyclin D, c-myc, TIMP-1, vimentin) (Shen et al., 2006).
In order to determine the precise role of Fn in postnatal mammary gland development we have conditionally deleted this gene in mouse mammary epithelium, using the Cre-loxP mediated approach. Our whole-mount, histological, immunohistochemical, morphometric, biochemical, and functional data show that Fn deletion delays the branching morphogenesis and impairs lobuloalveolar differentiation during pregnancy. This pathology is duplicated in mice transplanted with mammary epithelial cells isolated from FnMEp−/− mice and is associated with significantly decreased rates of mammary epithelial proliferation. Preliminary evidence suggests that altered signaling via the β1 integrin (Itgb1) Fn receptor and focal adhesion kinase (Fak) may have ultimately accounted for the mammary gland pathology associated with Fn deletion.
Mice carrying loxP sites within the 5′ untranslated region at the Msc1 site and within the first intron at the Nhe1 site of the Fn gene were generated in Dr. Fässler’s laboratory (Sakai et al., 2001). Tg(MMTV-Cre)22Mam, Line F (MMTV-Cre) mice (Wagner et al., 1997) were used for expression of Cre recombinase in the mammary epithelium. These mice were obtained from the Mouse Models of Human Cancers Consortium (NIH/NCI). After backcrossing to the FVB/N background, male MMTV-Cre mice were paired with Fnfl/fl females in order to obtain MMTV-Cre Fnfl/fl mice. Depending on their homozygous or heterozygous status for the floxed Fn allele, these mice were designated FnMEp−/− or FnMEp+/−, respectively. Nonrecombinant littermates were used as a control. Rosa26StopflLacZ reporter mice (Gt(ROSA)26SorTM1sor) were purchased from the Jackson Laboratory (Bar Harbor, ME). In these mice, expression of bacterial β-galactosidase is possible only after deletion of a stop codon flanked by loxP sites (Jiang et al., 2000). All mice were maintained identically following recommendations of the institutional Laboratory Animal Use and Care Committee.
Offspring with the desired genotypes were identified by PCR analysis of tail DNA, using primer sets from regions flanking the Fn1 lox-P sites as described by Fässler and associates (Sakai et al., 2001). PCR of floxed Fn results in a 1,200 bp DNA fragment, while Fn deletion results in a 320 bp band. MMTV-Cre mice were identified by primers Cre-5′ (5′ GGA CAT GTT CAG GGA TCG CCA GGC G 3′) and Cre-3′ (5′ GCA TAA CCA GTG AAA CAG CAT TGC TG 3′). PCR amplification of Cre results in a 296 bp DNA fragment. The PCR temperature profile was 94°C for 30 seconds, 60°C for 1 min, and 72°C for 2 min with extension of the last cycle for 10 min at 72°C.
Frozen sections (5 μm thick) of the mammary glands from MMTV-Cre Gt(ROSA)26SorTM1sor mice and control nonrecombinant mates were stained with X-gal solution as described by Bonnerot and Nicolas (1993). Sections were counterstained with Nuclear Fast Red. Controls were mammary glands from Rosa26STOPloxPLacZ mice. To further assess organ-specificity of the MMTV-Cre-mediated Fn knockout in mammary glands, DNA was extracted from lung (Lu), heart (He), kidney (Ki), mammary gland (MG), salivary gland (SG), uterus (Ut) and liver (Li) of FnMEp−/− mice and subjected to PCR as described above.
The left 4th inguinal mammary fat pad (MFP) was excised essentially as described by Tiran and Elson (2003). In brief, excised MFPs were spread onto glass slides and fixed with Tellyesniczky’s fixative overnight. MFPs were then soaked in three changes of 100% acetone each for 6h, followed by 2h-treatments with each 100%, 95% and 70% ethanol, and stained with 0.2% carmine red overnight. After rinsing in water, MFPs were dehydrated sequentially in 50%, 70%, 95%, and 100% ethanol each for 2h, cleared with methyl salicylate overnight and examined under a dissecting microscope. Mammary gland whole mounts were evaluated as follows: (i) Extent of ductal outgrowth: maximal outgrowth beyond MFP lymph node measured in millimeters; (ii) number of ductal branch points per unit area of MFP; (iii) number of terminal end buds (TEBs) in mammary glands from 5-week old mice as described by Shen et al. (2006); and (iv) size and number of alveolar lobules from mammary glands of pregnant mice. A minimum of 5 mice per age-category and genotype were analyzed.
The right 4th inguinal MFP was excised as described above, spread onto glass slides and fixed with phosphate-buffered 4% paraformaldehyde overnight at 4°C. MFPs were then subjected to routine paraffin-embedding, and 4-μm thick, planar sections stained with hematoxylin & eosin (H.&E.) or 0.1% Sirius Red/0.1% Fast Green in 1.2% picric acid to depict the extent of collagenous extracellular matrix. Sections were scanned with the ScanScope (Aperio Technologies, Inc., Vista, CA) and evaluated as follows: (i) assessment of the tissue architecture and branching pattern; and (ii) estimation of the volume density (VMG) of the glandular structures occupying the MFP as described by Weibel and Bolender (1973) and Pauli et al. (1983), using computer-enlarged photographs from entire MFP sections and superimposed test lattices of 1-cm2 subunits: VMG=Pi (number of line intersection points falling on mammary gland components)/PT (total number of intersection points falling on both mammary gland components and MFP stroma) (Pauli et al., 1983). A minimum of 7 mice per age-category and genotype were analyzed.
Five micron thick, paraformaldehyde-fixed paraffin sections from the right 4th inguinal MFP mounted on poly-L-lysine-coated glass slides were used for immunohistochemical analyzes. After standard deparaffinization, sections were treated sequentially with 1% pepsin in 10 mM HCl (pH 2.0; 30 min; 37°C; Sigma-Aldrich, St. Louis, MO) for antigen unmasking and M.O.M. Ig blocking reagent (Vector Laboratories, Burlingame, CA) for blocking endogenous immunoglobulin in cases using mouse primary antibodies. Staining was then performed using VECSTATIN Elite ABC kits (Vector Laboratories) according to the manufacturer’s instructions. Primary antibodies were rabbit anti-Fn (Sigma-Aldrich), mouse anti-cytokeratin 8 (CK8) (Santa Cruz Biotechnology, Santa Cruz, CA), and rat anti-mouse CD31 (BD Biosciences, San Jose, CA) antibodies. Secondary antibodies were biotinylated “Universal” anti-mouse/rabbit IgG made in horse and biotinylated anti-rat IgG made in rabbit (included in VECSTATIN Elite ABC kit). Sections were counterstained with hematoxylin. Control sections were stained with normal rabbit, mouse or rat IgG instead of primary antibody.
For immunofluorescence analyses, primary antibodies were rabbit anti-Fn and mouse anti-CK8 IgG and secondary antibodies were phycoerythrin-conjugated donkey anti-rabbit IgG and FITC-conjugated donkey anti-mouse IgG antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA). Control sections were stained with normal rabbit or mouse IgG instead of primary antibody.
The mammary epithelium proliferative rate was estimated by BrdU assay. In brief, mice were injected with 0.1 ml/10 gm body weight of BrdU solution (Invitrogen, Carlsbad, CA) 2 hours prior to sacrifice. BrdU-positive cells were detected by immunohistochemistry using anti-BrdU antibody (1:150 dilution; 1.5h; 37°C) (Invitrogen). The general proliferative rate was expressed as the percent of BrdU-positive cells of the total number of mammary epithelial cells counted (600 cells/mammary gland counted). Ductal epithelial cell proliferation was expressed as the number of BrdU-positive mammary epithelial cells per 500 μm of ductal circumference, measuring the long and short axes of a ductal cross-section and calculating the circumference according to the following website: http://www.csgnetwork.com/circumellipse.html.
The mammary epithelium apoptotic rate was determined by the cleaved caspase 3 immunohistochemical assay (Shen et al., 2006). The anti-cleaved caspase 3 antibody was from BD Biosciences (San Jose, CA) and was used at a dilution of 1:150 as described above. The percent of cleaved caspase 3-positive cells of the total number of mammary epithelial cells (~1,000 cells counted) was used as an expression of the mammary epithelium-apoptotic rate.
Twelve 8–12-week old virgin FnMEp−/− and Fnfl/fl mice served as mammary epithelium donors. Mammary epithelial cells were isolated essentially as described by DeOme et al. (1959) and Brill et al. (2008) with some modifications. Donor MFPs were excised as described above, minced into small fragments (~1 mm3), and digested with 0.3% collagenase type 3 (Worthington Biochemical, Lakewood, NJ) and 0.1% hyaluronidase (Sigma) in DMEM containing 0.2% bovine serum albumin (BSA) for 8 hours at 37°C in end-over-end suspension. After lyzing red blood cells with 125 mM ammonium chloride in DMEM, the digest was centrifuged (450×g; 5 min), and the pellet further digested in sequential solutions of trypsin-EDTA (0.25%/0.02%) for 1–3 min (gentle up-and-down pipetting), and 0.25% dispase (Worthington Biochemical) for 5 min, both in HBSS. The final digest was passed through a 40-μm nylon mesh (Tetko, Depew, NY). Cells were then cultured for 3 days in collagen-coated dishes in DMEM containing 5% fetal bovine serum (FBS), epidermal growth factor (10ng/ml; Sigma-Aldrich), and insulin (1μg/ml; Sigma-Aldrich). Cells were harvested by standard trypsin/EDTA treatment and adjusted to 5×104 cells/20 μl HBSS. Recipient mice were six, 10-day old female FVB mice, whose left and right, 4th MFPs had been cleared as described by Brill et al. (2008). The sites of the left, cleared MFPs were inoculated with mammary epithelial cells harvested from MMTV-Cre+-Fnfl/fl mice and the sites of the right, cleared MFPs with mammary epithelial cells harvested from MFPs of Fnfl/fl mice. Recipient mice were mated at 8 weeks of age and their MFPs removed after 15 days of pregnancy. Analysis of the MFPs was by whole mount and histological examinations. A total of 9 mice were analyzed.
Total RNA was extracted from the mammary glands of 4-month-old and 15-day pregnant Fnfl/fl and FnMEp−/− mice, using Trizol as described by the manufacturer (Invitrogen). RNA samples were quantified both spectrophotometrically and electrophoretically, and amounts adjusted so that 1μg was reverse-transcribed (SuperScript reverse transcriptase; Invitrogen). cDNA was subjected to PCR (93°C, 30″; 55°C, 30″; 72°C, 30″; 35 cycles) using Taq DNA polymerase (Invitrogen). Primers were 5′-GGAGAGCAGAAGTCCCATGA-3′ and 5′-ACTCCAGGGCTTCATCGTTA-3′. Controls were run in the absence of reverse transcriptase. Gapdh served as reference standard (primers: 5′-ACGACCCCTTCATTGACCTC-3′ and 5′-CTTTCCAGAGGGGCCATCCAC-3′).
Mammary epithelial cells were isolated from MFPs of 15-day pregnant MMTV-Cre+Fnfl/fl and Fnfl/fl mice as described above (Brill et al., 2008; DeOme et al., 1959). Cells were used for biochemical analyses after a 3-day culture period on Matrigel-coated culture dishes in DMEM containing 5% FBS, 25 ng/ml EGF, and 1 μg/ml insulin (Sigma-Aldrich). To determine Fn protein levels and assess the role of the downstream Fn-signaling partners in the lobuloalveolar differentiation of the mammary gland, cells were extracted in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM benzamidine chloride, 1 mM PMSF, 2 μg/ml leupeptin, 0.27 TIU/ml aprotinin, 0.1 mM sodium vanadate, and 1% Triton X-100) for 1h at 4°C (Abdel-Ghany et al., 2002). Total cell lysates were subjected to SDS-PAGE (~20–50μg protein) and Western blotting, using anti-Fn (Sigma-Aldrich), anti-Itgb1 (Santa Cruz Biotechnology), anti-focal adhesion kinase (FAK) and anti-Fak(pY397) primary antibodies (both from Enzo Life Sciences International, Plymouth Meeting, PA). Secondary antibodies were HRP-conjugated donkey anti-rabbit or goat anti-mouse antibodies (Jackson ImmunoResearch Laboratories). ECL for detection of bound antibody was as described previously (Abdel-Ghany et al., 2002).
The significance of the mean differences among the three mouse groups (Fnfl/fl; FnMEp+/−; FnMEp−/−) was evaluated using the analysis of variance statistical technique (ANOVA). If we rejected the null hypothesis of no significant differences among the groups, we performed a post-hoc test using the Tukey’s multiple comparison test to identify which two groups were significantly different from each other. All statistical analyses were performed using the statistical software Statistix 9.0 (Analytical Software, Tallahassee, FL) at p<0.05. For two-group (Fnfl/fl; FnMEp−/−)-comparisons Student’s t-test for unpaired data was used (p<0.05). All data are displayed as mean ± standard error (S.E.).
To assure selectivity of Cre-recombinase activity in the mammary epithelium several approaches were used. First, Cre-mediated deletion of floxed Fn was assessed by PCR of DNA isolated from the lung, heart, kidney, liver, uterus, mammary gland, and salivary gland of FnMEp−/− and FnMEp+/− mice and their non-recombinant littermates (Fnfl/fl). In agreement with Wagner et al. (1997), among the tested organs only the mammary and the salivary gland of FnMEp−/− mice exhibited strong Cre activity as visualized by deletion of the floxed Fn-segment (320 bp band) in these organs (Fig. 1A). In contrast, heterozygous FnMEp+/− mice exhibited both Fn-species (i.e., deleted [320bp] and undeleted [1.2kb] floxed-Fn) and Fnfl/fl mice revealed only undeleted, floxed Fn as expected (data not shown). Second, to further assess the selectivity of Cre recombinase activity and expression in the mammary epithelium, MMTV-Cre+ male mice were crossed with Rosa26STOPflLacZ females, and frozen sections from the mammary glands of MMTV-Cre+- Rosa26STOPflLacZ and Rosa26STOPflLacZ (control) mice were stained with X-gal. X-gal staining was only observed in the mammary epithelium from MMTV-Cre+- Rosa26STOPflLacZ mice, where Cre recombinase activity removed a floxed stop colon to allow expression of LacZ (Fig. 1B). The extent of Cre expression in mammary epithelium ranged from 50 to 100%, which is consistent to that reported by Wagner et al. (1997). No X-gal staining was observed in non-epithelial tissues. Age-matched, Cre-negative control mice revealed no X-gal staining of the mammary epithelium (Fig. 1C).
To examine Fn protein expression we selected mammary glands (or sections thereof) from 15-day pregnant Fnfl/fl and FnMEp−/− mice since at this stage mammary glands of normal mice express relatively high levels of Fn (Haslam and Woodward, 2003). As expected, Fnfl/fl mice exhibit strong Fn protein expression, while the amount of Fn protein in FnMEp−/− mice varied from approximately 0 (non-detectable) to 50–60% of that measured in MFPs from Fnfl/fl mice (Fig. 1D). This variation is due to the mosaic nature of Cre expression.
Carmine red-stained MFPs from 5- and 16-week old, virgin Fnfl/fl, FnMEp+/−, and FnMEp−/− mice were analyzed qualitatively and quantitatively. At five weeks of age, mammary glands of FnMEp−/− mice exhibited a statistically significant retardation in ductal outgrowth (Fig. 2A,C) as well as decreased numbers of branch points (Fig. 2A,D) and terminal end buds (TEBs) (Fig. 2B,E) relative to Fnfl/fl and, with exception of ductal outgrowth, to heterozygous FnMEp+/− mice (Fig. 2A-E). These abnormalities were somewhat compensated at 16 weeks of age, as no statistical differences in the length of ductal outgrowth (p=0.07652) and number of branch points were observed between Fnfl/fl and FnMEp−/− mice (Fig. 2A,F,G).
Histological examination of planar sections prepared from mammary glands of mice of the same groups complemented our findings from mammary gland whole mounts. At both 5 and 16 weeks of age, there was a decrease in the mean number of ductal cross-sections in FnMEp−/− relative to Fnfl/fl mice, that was most prominent in micrographs taken from areas distal of the MFP lymph node (Ln) (Fig. 3A-E). However, the retarded outgrowth of delicate (linear) ductal strands did not have a significant effect on the volume percent of glandular tissue occupying the MFP (Fig. 3F,G).
The effect of Fn deletion on mammary gland postnatal development is most dramatic during pregnancy. Mammary gland whole mounts of 15-day pregnant FnMEp−/− mice exhibited near normal ductal development and branching, extending from the nipple up to the terminal ductules, but lobuloalveolar differentiation, typically observed at this stage of pregnancy in normal and Fnfl/fl mice, was seriously compromised (Fig. 4A). Alterations varied greatly and ranged from the assembly of rudimentary lobuloalveolar structures (hypoplasia) in select lobules to a total absence of the lobuloalveolar compartment (aplasia). In general, the degree of lobuloalveolar hypoplasia correlated well with the amount of Fn protein recovered from these glands (compare Fig. 1D, lanes 2–4 and Fig. 4B of the same FnMEp−/− mice).
Histopathologically, there was normal lobuloalveolar differentiation in pregnant Fnfl/fl mice (Fig. 5A,D,F). Alveoli were budding out from patent terminal ductules (Fig. 5D) and associated closely with each other to form lobules (Fig. 5A). They were separated by scant vascularized, fibrous connective tissue (Fig. 5F) and their epithelia contained typical lipid globules (Fig. 5A). Associated small ducts and terminal ductules were patent and often filled with secretory products and were enveloped with a finely fibrillar collagenous extracellular matrix (Fig. 5F). In contrast, FnMEp−/− mice either failed to undergo lobuloalveolar development entirely (Fig. 5B) or formed loose clusters of alveoli that varied in number between lobules and that were often totally filled with epithelial cells (Fig. 5C). In the absence of alveoli, small ducts and terminal ductules were almost completely collapsed and were surrounded with abundant, collagenous connective tissue (Fig. 5G). Consistent with the degree of lobuloalveolar hyoplasia is a corresponding decrease in the glandular volume percent occupying the MFP in FnMEp−/− mice that is statistically significantly different from that in both Fnfl/fl and heterozygous FnMEp+/− mice (Fig. 5H).
To examine whether failure of generating the lobuloalveolar compartment in pregnant FnMEp−/− mice was due to decreased proliferation and/or increased apoptosis of mammary epithelial cells in FnMEp−/− mice, we performed BrdU incorporation and cleaved caspase-3 assays, respectively. Results show a significant decrease in the overall percentile of BrdU-positive mammary epithelial cells in pregnant FnMEp−/− mice relative to pregnant Fnfl/fl mice (Fig. 6A-C). Since this result might be skewed by the presence of a normal lobuloalveolar compartment in Fnfl/fl mice and a moderately to severely compromised lobuloalveolar compartment in FnMEp−/− mice, we also counted the number of BrdU-positive epithelial cells per 500 μm of ductual circumference in the same mice. Again, these counts were significantly lower in ductal epithelial cells from FnMEp−/− relative to those from Fnfl/fl mice (Fig. 6D–F). Decreased numbers of BrdU-positive ductal epithelial cells were also observed in 16-week old, virgin FnMEp−/− mice relative to age-matched Fnfl/fl mice (Suppl. Fig. 1). Interestingly, values of ductal epithelial cell incorporation of BrdU in Fnfl/fl and FnMEp−/− mice during pregnancy (Fig. 6F) were higher than in virgin, 4-month old mice of the same genotypes (Suppl. Fig. 1), correlating with increased Fn expression by pregnancy-associated hormonal conditions (Woodward et al., 2001) during alveologenesis (Williams et al., 2008). Together, our data indicate that Fn-deletion caused a significant decrease in the rate of mammary epithelial cell proliferation that accounted for the delay in ductal outgrowth in adult, virgin and impaired alveologenesis in pregnant FnMEp−/− mice.
No significant differences in cleaved caspase-3 were observed (data not shown), albeit a few shrunken or fragmented nuclei were noticeable in mammary epithelial cells from pregnant FnMEp−/− mice.
To exclude the possibility that MMTV-Cre transgene expression in other tissues, most notably in the hematopoietic system (Wagner et al., 1997), may have affected our data, we examined whether a similar pathology to that depicted in 15-day pregnant FnMEp−/− mice (Fig. 4A,B) could be reproduced in transplantation experiments. To do this, we injected the cleared 4th MFP of recipient FVB mice at 3-weeks of age with mammary epithelial cells isolated from FnMEp−/− or Fnfl/fl donor mice. Carmine red-stained mammary gland whole mounts were then analyzed at day 15 of pregnancy (Fig. 7A,B). Data show a degree of lobuloalveolar hypoplasia that paralleled the loss of Fn protein in mammary epithelial cell isolates from the combined group of FnMEp−/− donor mice (Fig. 7B). In contrast, mammary epithelial cell isolates from Fnfl/fl donor mice generated mammary gland transplants with normal lobuloalveolar development (Fig. 7A).
Normal alveologenesisis is thought to proceed by repetitive epithelial cleft and bud formation (Sakai et al., 2003). This process appears to be initiated by the deposition of wedges of Fn-positive material in clefts in the CK8-positive epithelial lining of a budding alveolus as shown in the mammary gland of a 15-day pregnant Fnfl/fl mouse and is followed by the ingrowth of vascularized mesenchyme to further separate budding alveoli (Fig. 8A–C, arrows). At the same time, new alveolar structures appear also to be formed by the outgrowth of epithelial cells from defined sectors of existing alveoli (Fig. 8D–F). These cells are polygonal in shape, are weakly CK8-positive, and their surfaces are decorated with a delicate rim of Fn-positive material (Fig. 8E). In contrast, deletion of Fn in mammary epithelial cells is associated with the sparse deposition of finely fibrillar Fn-positive material in a dense collagenous extracellular matrix that surrounds small ducts and ductules (Figs. 5G & 8G–I). Interestingly, many of these ductular cross-sections lack a lumen and instead are filled entirely with CK8-positve epithelial cells (Fig. 8G).
To further explore the role of neovascularization in mammary gland alveologenesis in pregnant Fnfl/fl mice, we stained sections from mammary gland lobules with anti CD31 and anti-Fn antibodies. Blood vessels form an intricate CD31-positive capillary network that surrounds alveoli in a honeycomb pattern (Fig. 9A). They are closely associated with Fn-deposits and are actively involved in modulating alveologenesis (Fig. 9B). In contrast, blood vessels in pregnant FnMEp−/− mice with severely compromised lobuloalveolar development are restricted to the fibrous connective capsule that envelops ductal cross-sections (Figs. 9C and and5G).5G). Accordingly, Vegf expression, which is primarily expressed by epithelial and myoepithelial cells and in only minor amounts in stromal cells (Pepper et al., 2000), was significantly lower in pregnant FnMEp−/− than Fnfl/fl mice (Fig. 9D), but was similar in adult, virgin mice of the same genotypes (Suppl. Fig. 2), in which the volume percent of glandular tissue occupying the fat pad was similar. Albeit these data may suggest that low Vegf expression levels merely reflect the loss of alveolar epithelial cells in the mammary glands from pregnant FnMEp−/− mice, its role as a Fn-dependent factor in alveologenesis (Khan et al., 2004, 2005; Rossiter et al., 2007; Qiu et al., 2008; Zhou et al., 2008) can not be excluded, given the close association between Fn-deposits and capillary endothelial cells at the base of emerging alveoli (Fig. 9B).
Conditional knockout of Fn, Itgb1, and Fak in mammary epithelium all compromise the normal development of the lobuloalveolar compartment during pregnancy (Li et al., 2005; Nagy et al., 2007; Naylor et al., 2005; White et al., 2004). To test whether Itgb1 and Fak might act downstream of Fn, we examined expression of these proteins and in the case of Fak, the activation status, in the mammary glands of Fnfl/fl and FnMEp−/− mice. In pregnant mice, Fn deletion appeared to be associated with decreased Itgb1 expression and suppressed phosphorylation of FakY397 (Fig. 10). Since this result might be affected by a compromised lobular compartment in FnMEp−/− mice, we also measured Itgb1 expression levels in adult, virgin Fnfl/fl and FnMEp−/− mice. Albeit there was no statistical difference in the glandular volume percent occupying the MFP in these mice, Itgb1 message levels were significantly decreased in 8-week-ol, virgin and pregnant FnMEp−/− mice (Suppl. Figs. 2 and 3), suggesting a co-regulation of Fn and Itgb1 during mammary gland postnatal development.
Fibronectin has been introduced as a multi-domain glycoprotein that regulates a variety of cellular functions including differentiation, proliferation, apoptosis, and migration (Danen and Yamada, 2001; ffrench-Constant, 1995; Geiger et al., 2001; Hynes, 1990; Johansson et al., 1997; Kumar, 1998; Leiss et al., 2008; Pankov and Yamada, 2002; Wierzbicka-Patynowski and Schwarzbauer, 2003). As such it is involved in embryologic and postnatal development (George et al., 1993; Sakai et al., 2003; Watt and Hodivala, 1994), tissue homeostasis, wound healing, angiogenesis, and cancer (Cheng et al., 2003; Danen and Yamada, 2001; ffrench-Constant, 1995; Geiger et al., 2001; Hynes, 1990; Johansson et al., 1997; Kumar, 1998; Leiss et al., 2008; Pankov and Yamada, 2002; Wierzbicka-Patynowski and Schwarzbauer, 2003). Here, we have analyzed the effects of the conditional knockout of Fn in mammary epithelial cells on postnatal development of the mammary gland. Studies of adolescent (5 weeks) and adult virgin (16 weeks) mice of the three phenotypes Fnfl/fl, FnMEp+/−, and FnMEp−/− yielded quantitative differences in the expansion of the ductal tree within the MFPs of the three groups of mice. Noteworthy are statistically decreased distances of branch outgrowth and decreased numbers of branch points and TEBs in 5-week old FnMEp−/− relative to Fnfl/fl mice. These differences were still there in 16-week old FnMEp−/− mice, but were less pronounced. Statistically sound differences were observed in pregnant mice, where the mammary glands of FnMEp−/− mice exhibited deficiencies of various degrees in the typical lobuloalveolar differentiation observed in Fnfl/fl and heterozygous FnMEp+/− mice (see Figs. 4 and and5).5). The degree of lobuloalveolar impairment in FnMEp−/−mice ranged from segmental hypoplasia to aplasia and was closely associated with the amount of Fn protein recovered from these glands.
Functionally, delays in ductal outgrowth and lack of alveologenesis in FnMEp−/− mice were associated with significantly decreased rates of proliferation of Fn-deficient mammary epithelial cells (Fig. 6; Suppl. Fig. 1). The near compensation of ductal outgrowth from 5- to 16-week old FnMEp−/− mice can be explained by the mosaic nature of Cre expression and the ensuing growth advantage of Fn-positive over Fn-negative epithelial cells. Similarly, the extent of Cre expression can also account for the broad range in the degree of impaired alveologenesis in pregnant FnMEp−/− mice; i.e., mammary glands with a high percentile of Cre-expressing epithelial cells fail to develop a lobuloalveolar department, while mammary glands, in which only half the epithelial cells are Cre-positive, exhibit only a mildly hypoplastic lobuloalveolar compartment. This effect is enhanced by high levels of Fn expression during pregnancy and alveologenesis (Woodward et al., 2001) and by a growth advantage of Fn-positive over Fn-negative mammary epithelial cells. Irrespective of the problems associated with MMTV-Cre mediated knockout models (Wagner et al., 1997; 2001), our data are consistent with experiments in 3-D culture showing that Fn promotes mammary epithelial cell growth and alveologenesis (Williams et al., 2008). Moreover, Fn knockdown in the salivary gland epithelium by siRNA as well as functional anti-Fn and anti-Itgb1 antibodies prevented alveologenesis in organ culture (Sakai et al., 2003).
Our data mimic those reported for mice with the MMTV-Cre mediated deletion of Fak in the mammary epithelium (Nagy et al., 2007). Fak-knockout in mammary epithelium, like Fn-knockout, did not affect the pattern of ductal outgrowth and branching, albeit both were slowed down in FnMEp−/− and FakMEp−/− mice. In pregnant mice, however, total deletion of either Fn or Fak left only a skeleton of ducts and ductules (Fig. 4) and was associated with failure in generating the lobuloalveolar compartment (Nagy et al., 2007). These data show that neither Fn nor Fak played a major role during early stages of postnatal mammary gland development (i.e., ductal outgrowth and branching), but were principals in alveologenesis (Nagy et al., 2007). A similar pathology, consisting of a decreased number of alveoli that were disorganized and contained clumps of epithelial cells bulging into what would normally be luminal space during pregnancy, has been reported in mice with conditional knockout of Itgb1 in the mammary epithelium (Naylor et al., 2005). These alveolar aberrations are consistent with those observed in FnMEp−/− mice, where alveolar lumina were often filled entirely with epithelial cells. In this comparison between Fn-, Itgb1-, and Fak-conditional knockout in mammary epithelium, it is important to recall that mammary epithelial deletion of Fn and Fak were mediated by Cre recombinase under control of the MMTV promoter, while Itgb1 deletion was mediated by Cre under control of the β-lactoglobulin (Blg) and whey acidic protein (Wap) promoters, respectively (Li et al., 2005; Naylor et al., 2005). MMTV-Cre expression in mammary epithelium, as monitored in ROSA reporter mice, was recorded as early as 6 days after parturition, while Bgl-Cre was detected in ductal epithelium of nulliparous mice from the age of 12 weeks and was extended throughout pregnancy and lactation and Wap-Cre was specifically expressed in secretory luminal epithelial cells starting at mid pregnancy and culminating at day 3 of lactation (Li et al., 2005; Wagner et al., 2001). Accordingly, the delay in Bgl-mediated Cre expression relative to that mediated by MMTV may have been responsible for the less severe effects on the lobuloalveolar compartment in Itgb1-knockout mice, while Wap-Cre-mediated Itgb1, as expected, had no effect on the generation of the lobuloalveolar compartment during mid-pregnancy.
In our study, loss of Fn appears to be closely associated with loss of Itgb1. Based on studies by Woodward et al (2001), both Fn and Itga5b1 increase significantly from adolescent to virgin adult mice and remain high during pregnancy in mammary epithelia and myopepithelial cells, while their expression levels in the stromal compartment remain more or less constant throughout mammary gland development (3-week old mice to lactating mice). Thus, it is possible that decreased Itgb1 expression is secondary to epithelial and myopepithelial loss in FnMEp−/− mice. However, the fact that Itgb1 is also significantly downregulated in virgin FnMEp−/− mice, in which we did not observe a statistical difference in the glandular volume percent occupying the MFP can not exclude a co-regulation of Fn and Itgb1. To this effect, Woodward et al. (2001) showed that ovariectomy of virgin adult mice caused a proportional decrease in both Fn and Itga5b1 and that this effect was reversed by estrogen treatment, suggesting a hormonal regulation of these two genes. Moreover, placement of mammary epithelial cells on various ECM substrates in vitro caused a significant increase in DNA synthesis only when placed on Fn, which among others might also have promoted synthesis and expression of its principal cell surface receptor Itga5b1 (Woodward et al., 1998). The co-expression of Fn and Itga5b1 has also been recognized in developing and regenerating chick peripheral nerve (Lefcort et al., 1992). Both Fn and Itga5b1 were expressed at high levels in developing nerve, but only at low levels in mature nerves. Following injury of mature nerve, both Fn and Itga5b1 were induced. These observations correlated with those obtained in mammary gland development in that both Fn and Itgb1 are upregulated only during states of high cellular proliferation, i.e. alveologenesis in the mammary gland and development and regeneration of nerve tissue.
The fact that Fn-, Itgb1-, and Fak-deletion in mammary epithelium have similar effects on the mammary gland development, combined with decreased Itgb1 expression and lack of phosphorylation of FakY397 in mice with Fn conditional knockout suggest that the observed pathology may be mediated, at least in part, by a disruption in the Fn/Itgb1/Fak signaling pathway. Decreased FakY397 phosphorylation, which is essential for its full activation (Schlaepfer and Hunter, 1998), is also noticed in mice with conditional knockout of Itgb1 in the mammary epithelium (Naylor et al., 2005; White et al., 2004) and in mice that expressed a dominant-negative chimeric form of Itgb1 (Faraldo et al., 1998). Thus, disruption of Fn/Itgb1/Fak signaling and associated malformation of focal adhesions (Naylor et al., 2005) may have accounted for the observed, dramatically decreased rates of proliferation (Abbi and Guan, 2002; Li et al., 2005; Mitra and Schlaepfer, 2006; Nagy et al., 2007; Naylor et al., 2005; White et al., 2004; Woodward et al., 2001), loss of polarity (Tomar et al., 2009), and epithelial invasion of mammary gland stroma (Hsia et al., 2003) during alveologenesis in pregnant FnMEp−/−, Itgb1MEp−/− and FakMEp−/− mice.
In addition to its mayor cell growth promoting role in generating the lobuloalveolar compartment, Fn appears also to be involved in alveolar budding in pregnant Fnfl/fl mice, inducing cleft formation within the epithelial lining of an existing alveolus as described in salivary gland morphogenesis in organ culture (Sakai et al., 2003). Accumulation of Fn in clefts supports the ingrowth of capillary blood vessels, which separate the new alveolus from the parental one and eventually surround each alveolus with vascular honeycombs in preparation for lactation (Djonov et al., 2003). Alternatively, new alveoli may also form from a cohort of weakly CK8- and Fn-positive mammary epithelial cells that grow out from the epithelial lining of an existing alveolus in a polarized fashion and penetrate the surrounding mammary stroma (see Fig. 8A-F). This epithelial outgrowth involves stromal invasion, matrix remodeling, and cell proliferation (Fata et al., 2004; Larsen et al., 2006; Williams et al., 2008) that may well be regulated by Fn and its putative signaling via Itgb1 and Fak (Li et al., 2005; Nagy et al., 2007; Naylor et al., 2005; White et al., 2004). The following differentiation into mature alveoli during lactation may then be achieved in an Fn-independent manner; e.g., Jak2/Stat5a-signaling, which has been reported to be accompanied by increased expression of epithelial differentiation genes such as E-cadherin, zonula occludens-1 and CK8 and CK18 (Sultan et al., 2008). It is paralleled by a downregulation of Fn (Williams et al., 2008; Adams and Watt, 1989).
High levels of Vegf expression combined with anti-CD31 immunohistochemical analyses underscore the importance of neovascularization in alveologenesis in Fnfl/fl mice. It is therefore not surprising that deletion of Vegf in mammary epithelium (Rossiter et al., 2007) and expression of the anti-angiogenic splice isoform of Vegf, Vegf165b, in transgenic mice (Qiu et al., 2008) compromised lobuloalveolar expansion into the fat pad during pregnancy and lactation and failed to upregulate genes involved in milk secretion (Rossiter et al., 2007). Albeit Vegf loss in FnMEp−/− mice could well be associated with a missing lobuloalveolar compartment, there is increasing evidence of a regulation of Vegf by Fn, particularly by its EDB+ splice variant, in a variety of tissues in both physiologic and pathologic conditions (Jiang et al., 1994; Rongish et al., 1996; Kahn et al., 2004 and 2005) as well as in vitro (Zhou et al., 2008). The deposition of Fn at the base of alveoli and its close association with capillary blood vessels in our study (Fig. 9B) suggests that Fn may provide a scaffolding for the migration of endothelial cells into areas of alveologenesis.
In conclusion, we have presented novel evidence of the role of Fn in postnatal mammary gland morphogenesis. Conditional knockout of Fn in mammary epithelium leads to lobuloalveolar hypoplasia/aplasia. This pathology is caused by a significantly decreased rate of proliferation of Fn-deficient mammary epithelial cells (Fig. 6), resulting in delayed ductal outgrowth and deficient alveologenesis (Fig. 5D&E). Preliminary data showing the concomitant downregulation of Itgb1 and decreased phosphorylation of FakY397 suggest that this pathology may, at least in part, be regulated by the disruption of Fn/Itgb1/Fak signaling. This notion is supported by the fact that conditional knockouts of Fn, Itgb1 and Fak all affect alveologenesis during pregnancy (Li et al., 2005; Nagy et al., 2007; Naylor et al., 2005; White et al., 2004). Further unraveling of the spatial and temporal expression patterns of Fn and other matrix macromolecules and their cell surface receptors as well as growth promoting factors (e.g., cyclin D, c-myc, HGF, EGF, VEGF) and steroid hormones (e.g., estrogen, progesterone) and their associated, interacting signaling pathways, which have all been implicated individually in mammary epithelial cell proliferation and glandular morphogenesis (reviewed in Hennighausen and Robinson, 2001; Hennighausen and Robinson, 2005; Shillingford and Hennighausen, 2001), will be needed to understand the complex picture of mammary gland morphogenesis. These efforts could ultimately harbor new clues of the role that these factors, particularly Fn, may play in mammary tumor progression (Haslam and Woodward, 2003; White et al., 2004; Williams et al., 2008; Huang et al., 2008).
We would like to thank Dr. Fässler for the kind gift of Fnfl/fl mice and Dr. Hynes for sending us these mice. We gratefully acknowledge the advice and assistance of Dr. Hussni Mohammed in the statistical analyses, the administrative assistance of Ms. Cindy Westmiller, and the technical help of Mr. Joseph Druso. This work was supported by a grant from the US Department of Defense Breast Cancer Program BC 031992 and a grant from the Rochester Breast Cancer Coalition (BUP) and by NIH grants R01 CA96823, R01 CA107013, and R01 CA116583 (AYN).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.