Lactational competence and involution of the mouse mammary gland require plasminogen

The mammary gland undergoes extensive, but finely controlled tissue remodeling throughout its growth and development. During post-pubertal maturation of the ductal tree, a variety of proteinases, growth factors and integrins are expressed in a well-regulated spatial and temporal pattern (Ossowski et al., 1979; Busso et al., 1989; Robinson et al., 1991; Coleman-Kmacik and Rosen, 1994; Witty et al., 1995; Faraldo et al., 1998; Thomasset et al., 1998). Synthesis of most proteinases ceases during late pregnancy and lactation. After weaning, the mammary gland is again remodeled in preparation for the next pregnancy through a complex and well-regulated cellular program. This process of involution involves the collapse of alveolar structures, removal of secretory epithelial cells by programmed cell death, phagocytosis by macrophages, proteolytic degradation of basement membranes, and stromal remodeling. Consequently, most of the differentiated epithelial cells disappear and an adipocyte-rich stroma, in which the resting ductal system is embedded, reappears (Lascelles and Lee, 1978; Walker et al., 1989; Strange et al., 1992; Talhouk et al., 1992; Marti et al., 1994; Lund et al., 1996). After 10-15 days, the structure of involuting glands approaches that of resting virgin glands (Lascelles and Lee, 1978).

Based on these changes, post-lactational involution can be divided into two distinct phases. The initial phase is characterized by programmed cell death of the differentiated epithelial cells and induced expression of Bax, p53 and clusterin (Lund et al., 1996; Li et al., 1997; Jerry et al., 1998). This is followed by a second phase with extensive tissue remodeling and a characteristic spatial and temporal expression pattern of a number of extracellular proteinases (Lund et al., 1996). These include the matrix metalloproteinases (MMPs) stromelysin-1 (Str1), stromelysin-3 (Str3) and gelatinase A, the serine proteinases urokinase-type plasminogen activator (uPA) and tissue-type plasminogen activator (tPA), and the cysteine proteinase cathepsin B (Ossowski et al., 1979; Busso et al., 1989; Dickson and Warburton, 1992; Lefebvre et al., 1992; Strange et al., 1992; Talhouk et al., 1992; Guenette et al., 1994; Li et al., 1994; Lund et al., 1996). The spatial distribution of mRNAs for Str1, Str3, gelatinase A and uPA in fibroblast-like cells during involution (Lefebvre et al., 1992; Lund et al., 1996) points to an active role for the mesenchymal stroma during tissue remodeling.

In mice that overexpress an autoactivating Str1 transgene during pregnancy and in cultured mammary epithelial cells, the degradation of extracellular matrix (ECM) appears to initiate apoptosis (Boudreau et al., 1995; Alexander et al., 1996). Str1 also participates in the degradation of the basement membrane macromolecule entactin/nidogen-1. This degradation is attenuated in double transgenic mice that overexpress both Str1 and tissue inhibitor of matrix metalloproteinases-1, leading to a decrease in the overall proteolytic potential (Alexander et al., 1996). These experiments directly demonstrate a functional role for Str1 during mammary epithelial apoptosis (Boudreau et al., 1995; Alexander et al., 1996).

Plasmin produced from plasminogen (Plg) by uPA and tPA may also influence mammary gland development through activation of latent pro-MMPs, direct degradation of ECM substrates, or the release of bioactive ECM fragments and growth factors (see Sternlicht and Werb, 1999, for review). To explore the molecular and cellular mechanisms underlying the role of plasmin(ogen) in these tissue remodeling events, we have analyzed mammary gland morphology in the Plg^−/− mouse. We used stereological methods to obtain unbiased quantitative estimates of the structural composition and contents of the mammary gland, and biochemical and molecular analyses to determine the molecular mechanisms underlying the morphological changes.

Plg gene-targeted 129/Black Swiss mice (Bugge et al., 1995) were backcrossed into C57BL/6J (Panum Institute, Copenhagen) for 4 or 10 generations, and 8-to 12-week-old female mice were examined. All mice backcrossed for four generations appeared healthy and were of the same mass as Plg^+/+ controls until after 14 weeks of age. uPA gene-targeted 129 mice (Carmeliet et al., 1994) were backcrossed into C57Bl/6J for four generations. To normalize lactational pressure, the number of pups for each dam was adjusted to seven on post-partum day 1. Pups were removed after 10 days of lactation and mammary glands were collected immediately or after 5 days of involution. The mice were anesthetized by subcutaneous injection of 0.03 ml/10 g of a 1:1 mixture of Dormicum (Midazolam, 5 mg/ml) and Hypnorm (Fluanison, 5 mg/ml and Fentanyl, 0.1 mg/ml). The mice were perfused intracardially with 20 ml of ice-cold phosphate-buffered saline (PBS) and inguinal and thoracic mammary glands were removed for protein extraction as described below. Except where otherwise stated, all lactating and post-lactational mice studied were primiparous. Animal care at the University of Copenhagen and Copenhagen University Hospital, Copenhagen, Denmark was in accordance with national and institutional guidelines and all mice were found to be free of murine pathogens in accordance with the FELASSA recommendations for health monitoring of experimental units (Rehbinder et al., 1996).

Mice used for immunohistochemistry were treated similarly, except that perfusion with cold PBS was followed by intracardial perfusion/fixation with 20 ml 4% (w/v) paraformaldehyde (PFA) in PBS. Abdominal (#4) mammary glands were removed, weighed and fixed for 16 hours in 4% PFA. The tissue was then rinsed in PBS, dehydrated and embedded in paraffin. Tissue sections were floated onto SuperFrost⁺ slides (Fisher Scientific, Pittsburgh, PA, USA) and stained with Hematoxylin and Eosin, and Gomori’s one-step trichrome stains. Apoptotic cells were identified using an in situ apoptosis detection kit (Boehringer-Mannheim, Penzberg, Germany). After counterstaining, at least 1000 nuclei per specimen were counted.

Whole-mount analysis of abdominal mammary glands was performed as previously described (Sympson et al., 1994).

Abdominal mammary glands were systematically sectioned at equidistant levels 20-200 μm apart and perpendicular to the long axis, depending on the size of the gland. At each level, 4-5 μm sections were cut and stained as described above. To obtain unbiased sampling, the first sections from at least three levels (top, middle and bottom) of each gland were analyzed using standard stereological methods (Gundersen et al., 1988). Morphometric measurements were carried out on coded slides by an observer (S.F.B.) who was unaware of the mouse category. Area fractions were estimated by point counting with a microscope that projected a 491× brightfield image together with a grid onto a computer screen. The grid had several sets of points with appropriate ratios (1:2, 1:4, 1:8, 1:18). ‘Fine’ points 92 μm apart were used for low-volume structures and ‘coarse’ points for high-volume structures and the total tissue reference volume. The specimen stage was moved in a meandering pattern by a DC-motor to ensure that the mammary tissue and grid were positioned independently. Using this procedure, about 10% of the total area of each section was evaluated. The overall mammary volume that was made up of a given structure (e.g. alveoli) was estimated as the product of its area fraction and the wet mass (mg) of the corresponding mammary gland, expressed as mm³, assuming a glandular density of 1 mg/mm³.

Mammary gland wet masses were compared using one-way analysis of variance (ANOVA). Comparisons of morphometric data for all three genotypes were obtained using the Kruskall-Wallis test. Approximate 95% confidence intervals were estimated for the ratios of mean involuting to lactating mammary gland wet masses. Tests on proportions (e.g. the fraction of mice that failed to feed their pups) were done using exact methods. P-values less than 0.05 were considered significant. All calculations were performed using the SAS System version 6.12 (SAS Institute, Cary, North Carolina, USA).

Tissue sections were deparaffinized in xylene and hydrated through graded ethanol/water dilutions. Antigen retrieval was by proteolytic digestion with 0.025% trypsin (Sigma T8128) in 50 mM Tris (pH 7.6) containing 0.1% CaCl₂ for 5-6 minutes at 37°C. Endogenous peroxidase activity was blocked using 1% hydrogen peroxide for 15 minutes at ambient temperature. Sections were then washed in Tris-buffered saline (TBS; 50 mM Tris, 150 mM NaCl, pH 7.6) containing 0.5% Triton X-100 (TBS-T). The slides were then mounted into Shandon racks with immunostaining coverplates (AX-LAB, Copenhagen, Denmark) for subsequent incubations. Rabbit anti-mouse fibrin IgG (10 mg/ml; Bugge et al., 1995) were incubated at 1:1000, a biotinylated rat monoclonal antibody against activated mouse macrophages (clone BM-8; BMA Biomedicals, Augst, Switzerland; Malorny et al., 1986) was incubated at 1:30. Primary antibodies were incubated overnight at 4°C. Rabbit anti-fibrin IgG were detected with biotinylated, affinity-purified swine anti-rabbit IgG (DAKO, Glostrop, Denmark) followed by streptavidin-HRP complexes (DAKO). Biotinylated BM-8 was detected with streptavidin-HRP complexes followed by tyramide signal amplification. All antibody incubations were followed by washes with TBS-T. Sections were developed with 0.25 mg/ml 3-amino-9-ethylcarbazole (AEC) in 0.05 M buffered acetic acid (pH 5.0) for 10 minutes, counterstained in Mayer’s Hematoxylin for 30 seconds, and mounted in Glycergel (DAKO).

Preparation of mammary gland extracts for uPA ELISA was as described (Lund et al., 1996). Nunc 96-well immunoplates were coated overnight with 4 μg/ml MA-H77A10 (directed against mouse uPA) in 0.1 M Na₂CO₃, pH 9.8, at 4°C. The wells were then washed with PBS-T and remaining protein-binding sites were blocked with 1% BSA in PBS at 37°C. Samples and a mouse uPA standard (concentration determined by amino acid analysis) were diluted in dilution buffer (pH 7.4) containing 1% BSA and 0.1% Tween-20, and incubated for 1 hour at 37°C. Bound uPA was detected using a polyclonal rabbit anti-mouse uPA IgG diluted to 2 μg/ml in dilution buffer. The signal was measured as enzyme rate (kinetic ELISA) using an alkaline phosphatase-conjugated monoclonal antibody against rabbit IgG (Sigma).

Pieces of mammary gland were homogenized in RIPA lysis buffer (150 mM NaCl, 1% NP40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris-HCl, pH 8.0) at 0.25 mg wet mass/μl and homogenates were spun at 10,000 g for 15 minutes at 4°C. Soluble fractions were diluted in nonreducing SDS sample buffer for zymography (Herron et al., 1986), and insoluble fractions were washed once in RIPA buffer and boiled in reducing SDS sample buffer for western blot analysis. To visualize gelatinolytic enzymes, RIPA-soluble proteins were separated on 10% SDS-PAGE mini-gels containing 1 mg/ml gelatin. After electrophoresis, the gels were washed for 30 minutes at room temperature in renaturing buffer (10 mM Tris-HCl, pH 7.5, 2.5% Triton X-100), incubated for 24-48 hours at 37°C in enzyme buffer (50 mM Tris-HCl, pH 7.6, 0.2 M NaCl, 5 mM CaCl₂, 0.02% Brij-35), and stained with Coomassie Brilliant Blue R-250. Gelatinolytic enzymes were thus revealed after destaining as clear bands against a background of uniformly stained substrate. To visualize entactin and its degradation products, reduced RIPA-insoluble proteins were resolved on 8% SDS-PAGE mini-gels and transferred to nitrocellulose membranes by electrophoretic blotting. Membranes were blocked for 2 hours at room temperature with 5% BSA in TBS containing 0.5% Tween-20 and 0.1% Triton X-100 (TBSTT, pH 7.6), and incubated overnight at 4°C with 0.2 μg/ml rat anti-mouse entactin (Upstate Biotechnology, Inc. #05-208) in TBSTT containing 0.5% BSA. After washing in TBSTT, membranes were incubated for 2 hours at room temperature with species-specific, peroxidase-conjugated sheep anti-rat IgG (Amersham #NA9320) diluted 1:2000 in TBSTT containing 0.5% BSA. Peroxidase activity on washed blots was detected using enhanced chemiluminescence (ECL) reagents from Amersham.

By definition, the study of lactational competence in mutant mice requires that they successfully reproduce. We found that 26% of Plg^−/− female mice (n=32) on a mixed 129/Swiss Black genetic background backcrossed for four generations onto C57BL/6J failed to give birth, whereas only 6% of wild-type littermates (n=30) and 10% of Plg^+/− mice (n=27) failed to do so (P < 0.04). Even though Plg^−/− mice had apparently normal estrus cycles (as judged by vaginal smears) and had been caged with two separate male mice of proven fertility for 2 weeks each, remnants of embryos and implantation sites were uniformly absent in those Plg^−/− mice that failed to bear pups. This indicates that the mice had failed to become pregnant and suggests that Plg deficiency had compromised their fertility. When the mutant mice were further backcrossed to C57BL/6J for 10 generations, this phenotype became more severe. Fewer than 50% of the C57BL/6J Plg^−/− mice became pregnant. For this reason, lactational studies were performed using the mixed background that had been backcrossed to C57BL/6J for four generations, except where noted. Although a significantly lower proportion of Plg^−/− mice became pregnant than did Plg^+/− or Plg^+/+ mice, all groups produced the same size litters (n≈8). Second pregnancies were rare in Plg^−/− mice of all backgrounds.

We next examined lactational competence in primiparous Plg^−/− mice. To normalize the lactational pressure for all mice, the number of pups was adjusted to 7 on post-partum day 1. Wild-type and Plg^+/− mice rarely (7%) failed to lactate, whereas 28% of Plg^−/− mice backcrossed for four generations into C57BL/6J and >75% of Plg^−/− mice backcrossed for ten generations into the C57BL/6J background were unable to sustain lactation for 10 days (data not shown). The dams showed normal mothering behavior and, in most cases, some milk was initially seen in the stomachs of pups. Even so, the pups failed to thrive rapidly thereafter. This lactational failure was even more severe after a second pregnancy. Out of eight second pregnancies observed in the mixed background, only one litter survived without loss.

Although most of the lactating Plg^−/− mice produced enough milk to sustain 7 pups for 10 days, their mammary glands were histologically abnormal when compared to those of wild-type mice (Fig. 1A-D). After 10 days of lactation, the mammary glands of Plg^−/− and Plg^+/− mice were lighter and contained less total secretory alveolar tissue than the glands of their Plg^+/+ littermates (Fig. 1G,H). The mean mammary gland wet masses for each genotype were significantly different (P<0.0001, ANOVA), with the Plg^+/− being higher than Plg^−/− (P<0.01) and lower than Plg^+/+ (P<0.01). The hypothesis of equal total alveolar volumes was rejected (P<0.002, Kruskall-Wallis test), Plg^+/− being significantly lower than Plg^+/+ (P<0.003), whereas there was no significant difference between Plg^+/− and Plg^−/−. Plg^+/− mice exhibited an intermediate phenotype, suggesting haploinsufficiency. Not surprisingly, we also observed histological abnormalities in mice that had failed to lactate. The mammary glands from 7 of the 9 Plg^−/− dams that had been unable to nurse their pups contained almost no secretory alveolar epithelium when examined immediately after the death of their litters. Instead, they had regions of prominent fibrosis, scattered areas of atypical (multilocular, immature) adipose tissue, and, in two cases, abundant inflammatory cells (Fig. 1E,F). The mammary glands from the two Plg^+/+ and two Plg^+/− mice that failed to lactate for 10 days also exhibited minimal alveolar development, but in an adipocyte-rich stroma without fibrosis.

Fibrin did not accumulate in vascular or extravascular tissues of lactating mammary glands in Plg^−/− mice (Fig. 2A,B). The distribution of immunoreactive α-smooth muscle actin in myoepithelial and vascular smooth muscle cells was unchanged in Plg^−/− and Plg^+/− mice (data not shown). However, numerous activated macrophages were found in the stroma of lactating Plg^−/− glands by staining with a monoclonal antibody (BM-8), whereas there were few immunoreactive cells in wild-type mammary glands (Fig. 2C,D).

The expression of uPA is highly upregulated during post-lactational mammary gland involution (Lund et al., 1996), thus its physiological target, Plg, may play a critical role in involution. After 10 days of successful lactation and 5 days of involution, the median glandular wet masses for Plg^+/+, Plg^+/− and Plg^−/− dams were significantly different from one another (P<0.0001, ANOVA) with the mammary glands of Plg^+/− mice weighing less than those of Plg^−/− mice (P<0.05) and more than those of Plg^+/+ mice (P<0.01) (Fig. 3G). Thus, although the mammary glands from lactating Plg^−/− mice were initially smaller than those of their Plg^+/− and Plg^+/+ littermates (Fig. 1G), they ended up significantly larger than those of their heterozygous and wild-type littermates after 5 days of involution. As a result, the proportionate reduction in the median gland mass after 5 days of involution was dramatically different for the three genotypes; namely an 86% reduction in mammary gland size in wild-type mice, a 67% reduction in heterozygous mice, and only a 27% reduction in Plg^−/− mice.

The differences in mass of the involuting mammary glands were accompanied by profound histological differences (Fig. 3A-I). The alveolar tissue in Plg^−/− mice revealed only minimal regression (Fig. 3K). Residual alveolar and ductal structures each occupied significantly larger volume fractions in Plg^−/− and Plg^+/− animals than in wild-type animals. Collapsed alveoli with epithelial cells shed into the ducts and uncollapsed necrotic alveoli were seen in 7 of the 13 Plg^−/− glands, but were never seen in Plg^+/+ or Plg^+/− glands. In addition, total volumes for alveolar, ductal and combined onnective and vascular tissue compartments were each significantly larger in Plg^−/− and Plg^+/− glands than in Plg^+/+ glands, with heterozygous mice exhibiting intermediate values that were generally closer to those seen in Plg^−/− mice than wild-type mice. Thus, only 1 mm³ of alveolar tissue remained in Plg^+/+ mammary glands following 5 days of involution, whereas 8.6 and 8.9 mm³ of residual alveolar tissue persisted in Plg^+/− and Plg^−/− glands, respectively. These differences are even more remarkable when one considers that 5 days earlier, when pups were removed from their nursing dams, the total alveolar volume was significantly smaller in lactating Plg^+/− and Plg^−/− glands than in wild-type glands. Thus, the median alveolar volume was reduced 306-fold (95% confidence interval=140-471) over 5 days of involution in Plg^+/+ glands, but only 32-fold (13-50) and 11-fold (3-20) in Plg^+/− and Plg^−/− glands, respectively.

The data presented above indicate that mammary involution is abnormal in Plg^−/− mice. This raises the question of the mechanisms responsible for this failure in mammary gland remodeling. The inability to efficiently degrade fibrin, which acts as a provisional matrix, accounts for the impaired wound healing phenotype in Plg^−/− mice (Bugge et al., 1996; Rømer et al., 1996; Lund et al., 1999). Mammary epithelial cells can use fibrin gels as an alternative to ECM for growth in culture (Alford et al., 1998). Thus, if excess fibrin accumulates in Plg^−/− mice, it could act as a provisional matrix for mammary epithelial cells, supporting their prolonged survival and function in the face of ECM degradation and other pro-apoptotic signals seen during involution. Nevertheless, we did not observe excess fibrin accumulation during involution in the Plg^−/− mice compared to Plg^+/+. Wild-type and Plg^−/− involuting mammary glands showed similar immunohistochemical staining for intra- and extravascular fibrin (Fig. 4A,B). In lactating Plg^−/− glands the lack of abnormal fibrin deposition may be due to fibrin clearance by increased macrophage infiltration (see Fig. 2C,D) via plasmin-independent mechanisms (Simon et al., 1993; Hiraoka et al., 1998). There was, however, no similar increase in the number of activated macrophages in involuting Plg^−/− glands as compared to wild-type glands (Fig. 4C,D).

Important paracrine crosstalk occurs between mammary epithelial cells and the adipocyte-rich stroma of the mammary gland. Indeed, a reciprocal relationship exists between adipocyte and epithelial differentiation, with one population increasing as the other wanes during pregnancy and involution (Topper and Freeman, 1980). In Plg^−/− mice, delayed epithelial regression was accompanied by diminished adipocyte differentiation during involution. The Plg^−/− glands had proportionally less adipose tissue than wild-type or Plg^+/− glands, and their stroma was instead filled with immature adipocytes and fibrotic tissue (Fig. 3A-I). The fraction of mammary tissue containing adipocytes was also statistically different for each genotype (P=0.0002; Kruskal-Wallis test) at 50% for Plg^+/+, 46% for Plg^+/− and 34% for Plg^−/− mice. These data suggest that an abnormal stromal environment may contribute to the altered epithelial phenotype.

uPA and matrix-degrading MMPs are highly regulated during mammary gland development (Busso et al., 1989; Lund et al., 1996). We next asked whether Plg deficiency alters their expression or enzymatic activation, because such alterations could clearly contribute to the mammary changes seen in lactating and involuting Plg^−/− mice. As seen previously (Lund et al., 1996), uPA levels were low during lactation and were upregulated five-to sixfold during involution in Plg^+/+ mice; similar levels were found in Plg^+/− mice (Fig. 5A). In Plg^−/− mice, however, mammary uPA levels were fourfold higher than in wild-type mice during lactation (P<0.02). Only a slight increase of this already high concentration of uPA was seen during involution, so that uPA levels in involuting Plg^−/− mammary glands were comparable to those seen in wild-type mice. Because mammary changes were observed during lactation, when uPA is normally low, they may arise during prior stages of mammary development when uPA is more abundant and Plg activation is more apt to have an impact. However, mammary glands from uPA^−/− mice showed no detectable delay in development or postlactational involution by whole-mount or histological analysis (data not shown). These data suggest that uPA is not essential and may serve as a marker of the macrophage infiltrate (Fig. 2C,D).

There were no pronounced differences between the three genotypes in the expression of gelatinase A during either lactation or involution, although its activation during involution appeared to be slightly lower in Plg^−/− glands as compared to Plg^+/− and Plg^+/+ glands (Fig. 5B,C). In the Plg^+/+ and Plg^+/− mice, gelatinase B levels were low in the lactating glands and were consistently higher on the fifth day of involution. Both latent and active gelatinase B levels were unusually high in most Plg^−/− mammary glands during lactation (Fig. 5B), but were similar to those seen in the Plg^+/+ and Plg^+/− mice during involution (Fig. 5C). Because uPA and gelatinase B are produced by macrophages, their unusually high expression may be attributed to the unusual abundance of activated, BM-8-immunoreactive macrophages that were also present in the mammary glands of lactating (Fig. 2C,D), but not involuting (Fig. 4C,D), Plg^−/− mice. The high levels of gelatinase B may compensate for the effect of Plg-deficiency (e.g. by preventing fibrin accumulation during lactation; Hiraoka et al., 1998). It is noteworthy that both gelatinase A and B continued to be activated in vivo despite the absence of Plg. A 75 kDa gelatinolytic activity was detected in lactating, but not involuting, glands from Plg^+/+, Plg^+/− and Plg^−/− mice (Fig. 5B,C). Inhibition experiments showed that the activity was completely abolished by aprotinin, but not by EDTA, calpain I and II, and E64, suggesting that the activity was probably caused by a serine protease (data not shown).

Our data indicate that the stromal ECM is abnormal in the Plg^−/− glands, showing an increase in collagen deposition (Fig. 3). However, MMP expression and activation were not substantially altered in Plg^−/− mice. Therefore, we sought to determine if basement membrane remodeling was altered. Following weaning, a characteristic pattern of proteolysis of various components of the extracellular matrix is seen (Alexander et al., 1996). We found that the ECM component entactin/nidogen-1 was degraded in Plg^+/+, Plg^+/− and Plg^−/−glands. By immunoblotting analysis, the level of 112 and 110 kDa degradation products of entactin were virtually the same in the three genotypes both at day 10 of lactation and at day 5 of involution (data not shown). However, bands at about 58 and 55 kDa, that were weakly detectable in wild-type mice, were strongly increased in Plg^+/− and Plg^−/− mice during both lactation (data not shown) and involution (Fig. 5D), suggesting that plasmin is required to prevent accumulation of lower molecular weight degradation products. We found that laminins 1 and 5 were degraded in involuting mammary glands; however, we observed no differences in the degradation patterns for the β;1 or γ1 chain of laminin-1 or the γ2 chain of laminin-5 for the various Plg genotypes (data not shown). These results indicate that although ECM degradation does take place, there is an accumulation of specific basement membrane protein fragments that probably require plasmin for further degradation. Taken together with the accumulation of a fibrotic stromal matrix, they indicate that mammary ECM homeostasis is severely compromised in Plg^−/− mice. Moreover, such fragments of ECM could have new biological activities in involution (Pujuguet et al., 2000).

Because ECM remodeling is coupled to epithelial apoptosis during involution (Boudreau et al., 1995), one consequence of altered ECM homeostasis might be altered apoptosis. Indeed, we observed decreased apoptosis in Plg^−/− mammary glands at 5 days of involution. Apoptotic cells were detected in mammary glands of each genotype at 5 days post-weaning (Fig. 6A,B); however, the proportion of epithelial cells undergoing apoptosis in Plg^−/− glands (2.0±0.5%; n=3) was significantly lower than in Plg^+/+ glands (5.2±0.5%; n=3, P<0.001). These data suggest that alveolar regression is delayed, but not entirely blocked. Indeed, histological analysis of mammary glands from Plg^−/− mice 7 and 14 days post-weaning showed continued but incompleted remodeling compared to wild-type glands (Fig. 6C-F). Interestingly, mammary glands harvested from a Plg^−/− mouse 21 days after its second parturition and unsuccessful lactation had not involuted and were still engorged with milk (data not shown). Thus the delay in alveolar regression was even more severe after a second pregnancy.

The tight regulation of uPA expression during mammary gland development (Busso et al., 1989; Lund et al., 1996) suggests that its physiological target, Plg, may play an important role in these processes. By examining Plg^−/− mice, we have shown that Plg is indeed required for adequate lactation and post-lactational involution. Moreover, a haploinsufficiency phenotype was observed in Plg^+/− mice, suggesting that these processes require a threshold level of Plg that is not achieved by one allele alone. In accordance with indications that Plg may play a permissive role in ovulation (Hägglund et al., 1996) and embryonic implantation (Sappino et al., 1989), we also observed diminished fertility in Plg^−/− female mice. In addition, it is likely that modifiers of Plg activity exist, based on our observations of strain-dependent differences in phenotypic severity.

Plg does not appear to be required during lactation, when uPA expression is low (Busso et al., 1989), but prior to lactation, when uPA expression is high and the lactational phenotype is acquired. Indeed, between 28% and more than 75% of the primiparous Plg^−/− mice were unable to lactate successfully, depending on the strain. In addition, the mammary glands of those Plg^−/− mice that were able to lactate were significantly smaller and had significantly smaller secretory alveolar volumes than those of their Plg^+/+ and Plg^+/− littermates. The Plg^−/− dependent defects seen in lactation probably arose during mammary development. Indeed, whole-mount preparations from 6-to 8-week-old virgin Plg^−/− mice in estrus revealed delayed ductal branching during development, even though they appeared to initiate puberty at the same time as Plg^+/+ mice, cycled normally, and were healthy and of normal body mass (our unpublished observations). These data demonstrate that Plg plays a significant role in achieving lactational competence during pregnancy. In support of the notion that lactation itself does not require Plg, the existing alveolar epithelium required for milk production and the myoepithelial cells required for milk ejection remained functional in Plg^−/− mice, so that some were able to nurse their pups despite a severe shortage of secretory epithelium.

Lactational failure in the absence of sufficient Plg could also reflect inadequate mammary development during puberty, when uPA levels are high (Busso et al., 1989). Indeed, whole-mount analysis of mammary glands from virgin Plg^−/− mice showed a delay in early ductal development (data not shown). Thus, Plg may play an important role in branching morphogenesis, which, if lacking, could limit lactational competence. Lactational failure was even more prevalent following the rare occurance of a second pregnancy. Thus, although Plg^−/− mammary glands may involute to some extent, their phenotype becomes progressively more severe due to an inability to dedifferentiate and redifferentiate to a normal lactating state. These data also suggest that abnormal or incomplete involution can compromise lactational development during a subsequent pregnancy.

Mammary gland involution in mice occurs in two distinct phases: an initial wave of apoptosis that begins in the absence of high proteinase expression 2 days after weaning, and a proteinase-dependent phase of glandular regression and ECM remodeling that starts 2 days later (Lund et al., 1996). This latter phase is associated with strong upregulation of uPA and MMPs (Lund et al., 1996) and can be suppressed by administering inhibitors of both uPA and MMPs, but not by administering either inhibitor alone (unpublished results). Nevertheless, our results indicate that post-lactational involution is profoundly compromised in the absence of Plg. The reduction in overall mammary gland size after 5 days of involution in Plg^−/− mice was less than one-third the reduction seen in Plg^+/+ mice, and the reduction in secretory alveolar volume was almost one-thirtieth the reduction seen in Plg^+/+ mice. In fact, although Plg^−/− mice had significantly less alveolar tissue than Plg^+/+ mice after 10 days of lactation, they ended up with significantly more alveolar tissue than their wild-type littermates after 5 days of involution.

Interestingly, heterozygous Plg^+/− mice exhibited an intermediate phenotype in both lactation and involution, suggesting that a strict requirement for adequate levels of circulating Plg is not met by a single functional allele. Likewise, we have found that Plg^+/− mice show a slight, but significant delay in incisional wound repair as compared to Plg^+/+ controls (our unpublished results), while Plg^−/− mice show a more substantial delay (Rømer et al., 1996, Lund et al., 1999). Some cases of lactational failure in women, with its attendant potential for dire consequences in the breast-fed infant, have been attributed to insufficient glandular development of the breast (Neifert et al., 1985). Whether a subset of these cases is due to or exacerbated by low circulating Plg remains untested. Recently the inherited disease ligneous conjunctivitis has been linked with mutations in the Plg gene causing Plg deficiency (Schuster et al., 1997). Our results predict that some women with homozygous or heterozygous Plg deficiency would be unable to lactate, and that others would exhibit compromised lactation and involution.

Because little is known concerning the role of ECM-degrading proteinases in establishing a normal lactational phenotype, it is difficult to speculate on why Plg is required. We can, however, rule out the possibility that Plg is required for the activation of other proteinases that are required for normal mammary gland development. Indeed, both gelatinase A and B were activated in lactating and involuting Plg^−/− mice. Although activation of gelatinases A and B failed to occur in one study using cultured cells isolated from Plg^−/− mice (Carmeliet et al., 1997), both enzymes were activated in another such study (Lijnen et al., 1998). Thus MMP-activating enzymes other than plasmin are likely to be present in vivo, but may not be accounted for in all cell culture experiments. Indeed, mast cell chymase has been shown to activate gelatinase B during skin tumor progression in vivo (Coussens et al., 1999). Thus, the proteolytic environment other than that provided by Plg appears intact in Plg^−/− mice. What then are the potential mechanisms leading to abnormal involution in Plg^−/− mice? One possibility is that involution is not sensed. If this were the case, upregulation of the molecular program required for involution should be absent or diminished. We, however, observed that uPA and MMPs were upregulated normally. Thus, regulation of their expression during involution is independent of Plg concentration, and the lack of excess expression suggests an absence of compensatory mechanisms. A second possibility is that although MMPs are upregulated, they are not activated in the absence of Plg; however, MMPs were activated. It is likely that Plg and MMPs play distinct roles in mammary development. One result of Plg deficiency is a change in proteolytic balance to one dominated by MMPs. Moreover, some aspects of the Plg^−/− phenotype resemble those of excess MMP activity, such as fibrosis and adipocyte abnormalities (Thomasset et al., 1998; Sternlicht et al., 1999).

Another possible mechanism is a failure to sense apoptotic signals and/or the presence of excess survival signals. The most significant observation is that apoptosis was decreased in involuting Plg^−/− mammary glands, indicating that this mechanism is involved. Although we cannot rule out systemic effects of Plg deficiency in involution, the mice appeared normal based on mass until 14 weeks, well after the experiments were terminated. In light of the potent fibrinolytic activity of Plg, it is also possible that the mammary epithelial cells do not sense the altered microenvironment because of increased accumulation of fibrin, which would substitute for basement membrane. Indeed, fibrin has been shown to support the growth of cultured mammary epithelial cells (Alford et al., 1998). Surprisingly, however, there was no excess fibrin in the Plg^−/− mammary glands. This is probably the result of a significant, Plg-independent role for MMPs in fibrin degradation (Bini et al., 1996; Hiraoka et al., 1998; Lund et al., 1999) and the fact that fibrinolytic MMPs, such as Str-1 and MT1-MMP, are present during the late phase of mammary involution (Lund et al., 1996). In support of this premise, treatment of involuting Plg^−/− mice with a broad-spectrum MMP inhibitor (GM6001) resulted in the accumulation of excess mammary fibrin (unpublished results). Thus, although rescue of the mammary phenotype in fibrin-deficient mice could not be tested directly, since fibrinogen-deficient female mice uniformly fail to carry their offspring to term (Suh et al., 1995), it appears unlikely that the accumulation of a provisional fibrin matrix is a significant contributor to the failure of Plg^−/− mice to involute, whereas this is a major pathway for defective wound healing (Bugge et al., 1996; Rømer et al., 1996; Lund et al., 1999). Nevertheless, it is possible that subtle fibrin accumulation occurs in the Plg^−/− mammary glands when and where the expression of other fibrinolytic enzymes is low. In this case, the fibrin may act as a nidus for granulation-like tissue and the formation of a fibrous interstitial matrix into which mammary epithelium is unable to grow and differentiate. This, in turn, may account for the tendency towards fibrosis in the Plg^−/− mammary glands. Thus, unlike acute wound healing, fibrin accumulation may not be obvious in Plg^−/− mammary glands at any particular instant, yet its repeated removal and replacement may help create an environment that is unfit for mammary development and lactation.

Another possible reason for insufficient involution is that the cellular environment is not changed appropriately in the absence of adequate Plg. Although basement membrane degradation takes place during involution, as indicated by cleavage of entactin/nidogen-1, there is an abnormal accumulation of breakdown products of this molecule, which could contribute to altered development and involution (Pujuguet et al., 2000). Thus, Plg either directly or indirectly contributes to the clearance of low molecular weight entactin cleavage products. Moreover, insufficient clearance of entactin breakdown products or of other bioactive ECM fragments may support cell viability and protect against cell death, and may thereby defy the process of mammary involution. Another intriguing observation was the altered stromal and adipocyte differentiation seen in Plg-deficient mice. This stromal phenotype may also be the result of abnormal accumulation of ECM fragments with cellular signaling capabilities. Such an environment may defy proper lactational development and may harbor signals that impede involution. Clearly, the epithelial and stromal compartments are coordinately regulated (Thomasset et al., 1998). In the present model, as long as the stromal differentiation remained closer to the lactational phenotype, epithelial involution was delayed.

Plasmin may also activate and release sequestered growth factors from the stromal ECM or cell surface. For example, plasmin can cleave high-affinity IGF-binding proteins (IGFBPs), cause the release of active IGF (Campbell and Andress, 1997) and bFGF (Brunner et al., 1991), and activate latent TGF-β; (Munger et al., 1997). Thus, Plg may influence mammary function by regulating the availability or activation of important mammary growth and survival factors.

Our data demonstrate a functional role for plasmin(ogen) in mammary gland biology; particularly in postlactational involution. However, they also show that although involution is delayed by more than a week, it can ultimately proceed in the absence of Plg. There are strong similarities between the proteolytic mechanisms involved in mammary involution, wound healing and cancer invasion and metastasis (Johnsen et al., 1998), and the effect of Plg deficiency on these processes is also similar. Like mammary involution, skin wound healing and tumor metastasis are delayed but ultimately do occur in the absence of Plg (Rømer et al., 1996; Bugge et al., 1998; Lund et al., 1999). In the case of wound repair, there is functional overlap between Plg and one or more MMPs, such that a complete arrest of healing requires both Plg deficiency and MMP inhibition (Lund et al., 1999). Whether similar overlaps exist between Plg and other proteinases in mammary gland development and cancer invasion remain to be determined.

The excellent technical assistance of Ole Behrendtsen, Trine Hansen, Kirsten Lund Jakobsen, Pia Gottrup Knudsen, Pia Petersen and John Post is acknowledged. This work was supported by Fru Astrid Thaysens Legat for Lœgevidenskabelig Grundforskning, The Danish Cancer Society, The Danish Research Council, The Danish Cancer Research Foundation, the US Army Medical Research and Materiel Command (DAMD17-97-1-7246), the National Cancer Institute (CA 57621) and the Women’s Health Initiative, Office of the Director of the NIH (CA 57621S1).