Expression of Autoactivated Stromelysin-1 in Mammary Glands of Transgenic Mice Leads to a Reactive Stroma During Early Development

Tissue Collection

Mammary tissue was obtained from normal CD-1 mice (Charles River, Wilmington, MA) and their transgenic counterparts. Transgenic mice expressing an autoactivating Val⁹²-to-Gly⁹² mutation of the rat SL-1 gene under the control of a whey acidic protein promoter were generated previously. 14 Mammary glands from two different transgenic lines (M2-5 and M2-21) were used interchangeably. The inguinal and abdominal mammary glands were surgically excised from female mice at various stages of development and immediately frozen in liquid nitrogen for RNA extraction. A small portion of the gland was fixed in 4% paraformaldehyde for in situ hybridization analysis. All experiments were performed under protocols approved by the Animal Welfare and Research Committee, Lawrence Berkeley National Laboratory, and the Committee on Animal Research, University of California, San Francisco.

Northern Blot Analysis

RNA was prepared by the technique of Chomczynski and Sacchi. 17 Total RNA (15 μg) was separated on denaturing formaldehyde agarose gels, transferred to Hybond N+ membranes (Amersham, Arlington Heights, IL), and hybridized at high stringency with a riboprobe generated with T7 polymerase (New England Biolabs, Beverly, MA) from the mouse SL-1 cDNA pmTRM11 18 that was radiolabeled with ³²P-UTP (Amersham) to a specific activity of 1 × 10⁸ cpm/μg. Sequences corresponding to the 3′ untranslated region of the rat SL-1 cDNA (nucleotides 1514 to 1772) 19 were used to identify rat SL-1 transgene mRNA. Sequences corresponding to nucleotides 83 to 2069 of the full-length mouse tenascin-C cDNA 20 were used to identify tenascin-C mRNA. The cDNA probes for platelet endothelial cell adhesion molecule-1 (PECAM-1) 21 and tenascin-C were radiolabeled by random priming (Rediprime kit, Amersham) according to the manufacturer’s instructions. For normalization, blots were boiled in water and reprobed with a cDNA probe for ribosomal 28S RNA.

Reverse Transcription-Polymerase Chain Reaction and Southern Hybridization

Total RNA was resuspended in diethyl pyrocarbonate-pretreated water and reverse transcribed with 10 U/μl Moloney murine leukemia virus reverse transcriptase (Life Technologies, Gaithersburg, MD) in 50 mmol/L Tris-HCl (pH 8.3), 75 mmol/L KCl, 3 mmol/L MgCl₂, 10 mmol/L dithiothreitol, 0.5 mmol/L dATP, 0.5 mmol/L dCTP, 0.5 mmol/L dTTP, 0.5 mmol/L dGTP, and 12.5 mg/μl oligo(dT)_12–18 (Life Technologies) for 30 minutes at 37°C. Polymerase chain reaction (PCR) amplification was performed with 2 ng/μl reverse-transcribed RNA, 0.025 U/μl Taq DNA polymerase (Life Technologies), 1 μmol/L 5′ primer, 1 μmol/L 3′ primer, 20 mmol/L Tris-HCl (pH 8.4), 50 mmol/L KCl, 2 mmol/L MgCl₂, 0.2 mmol/L dATP, 0.2 mmol/L dCTP, 0.2 mmol/L dGTP, and 0.2 mmol/L dTTP with cycle numbers indicated in the figure legends. Each PCR cycle was performed at 94°C for 1 minute, 55°C for 1 minute, and 72°C for 1 minute. Reverse-transcribed RNA (100 ng) was amplified with the following primer pairs (all from Biosynthesis, Lewisville, TX): GCAGCCATTTCTTTAAAGGC as 5′ primer and CCACTTCAGTGCGCCAAGTT as 3′ primer for amplification of rat SL-1; CTATGCCTACTTCCTTCGTGGC as 5′ primer and ATCTCATTACCAACACCACTCC as 3′ primer for mouse stromelysin-3 (SL-3); TTGAGAAGGATGGCAAGTATGG as 5′ primer and ACACCTTGCCATCGTTGC as 3′ primer for mouse gelatinase A; TTGAAGGATGGCAAGTATGG as 5′ primer and CGAAGGCATGACCTAGAGTGT as 3′ primer for mouse matrilysin. PCR amplification products were resolved on 1.5% agarose gels. To verify the identity of the amplified sequences, Southern hybridizations were performed according to published procedures 22 with oligonucleotides complementary to the mRNA sequence of the gene examined. The following oligonucleotides were used: GAAACCCAAATGCTTCAAAGACAGCATCCA for rat SL-1, TATGGCTGGGTCTCTTACATGATCTAAG for SL-3, GTCCATCAGCATTGCCACCCATGGTAAACA for gelatinase A, and TGTCTCCATGATCTCTCCTTGCGAAGCCAA for matrilysin.

In Situ Hybridization

Deparaffinized sections (5 μm) were treated with proteinase K (5 μg/ml) and hybridized overnight with ³⁵S-labeled transcripts from a mouse SL-1 cDNA 18 subcloned in Bluescript KS (+) (Strategene, La Jolla, CA). For antisense probe, pmTRM12 fragment (3115 to 4051) was used, and for sense probe pmTRM12 fragment (2205 to 2918) was used. After hybridization at 47°C, sections were treated with RNase A (20 μg/ml; Boehringer Mannheim, Indianapolis, IN) and RNase T1 (2 U/ml; Boehringer Mannheim) for 10 minutes at 37°C, followed by washes in 2× standard saline citrate, 50% formamide at 50°C for 2 hours and in 0.1× standard saline citrate at 25°C for 30 minutes, before autoradiography with NTB2 emulsion (Kodak, Rochester, NY). Slides were exposed for 10 days and counterstained with hematoxylin after developing.

Histochemistry

Paraffin-embedded tissue sections (5 μm) were deparaffinized in xylene, treated with ethanol, rinsed in phosphate-buffered saline, and incubated for 10 minutes at 37°C with 0.1% trypsin in 0.05 mol/L Tris-HCl and 0.1% CaCl₂, pH 7.4. Sections were then incubated with antibodies against von Willebrand factor (vWF) (A 0082, 1:200 dilution; DAKO, Carpinteria, CA) in blocking buffer (phosphate-buffered saline with 2% bovine serum albumin) overnight at 4°C. After washing with phosphate-buffered saline, the specimens were incubated with biotinylated goat anti-F(ab′) mouse antibody (1:100 dilution; Amersham) for 30 minutes at ambient temperature. The specimens were then washed and incubated for 30 minutes at ambient temperature with Texas Red-conjugated streptavidin (1:100 dilution; Amersham). Nuclei were visualized by brief incubation of sections with 4′,6′-diamidino-2-phenyl indole (DAPI, 0.5 μg/ml) (Sigma Chemical Co., St. Louis, MO) before mounting with Vectashield (Vector Laboratories, Burlingame, CA).

For the detection of tenascin-C, frozen tissue sections (5 μm) were fixed in 2% paraformaldehyde for 20 minutes, washed with 0.1 mol/L glycine in phosphate-buffered saline for 15 minutes, and incubated with a rabbit polyclonal antiserum to mouse tenascin-C (1:500 dilution, pK7, gift from Dr. Melitta Schachner). Indirect immunofluorescence was performed as described previously. 23

Gomori’s trichrome staining was used to stain total collagens. 24

Endogenous SL-1 Is Expressed in the Mammary Stroma and Is Up-Regulated in Transgenic Mice

In a previous study, we showed that the mammary glands of virgin SL-1 transgenic animals resembled those of glands from early pregnant normal mice and that during lactation the mammary glands of transgenic mice had smaller alveoli than wild-type mice and exhibited reduced milk expression. 13,14 Thus, the transition to loss of function in these animals must have begun during pregnancy or even earlier, rather than during lactation. To gain insight into the underlying mechanism of this phenomenon, we analyzed when and how expression of the SL-1 transgene led to alteration of gene expression in the developing mammary gland.

Endogenous SL-1 is the major MMP expressed in the stroma of mammary glands from virgin, early pregnant, and involuting mice. 9,12,15,25,26 We therefore examined expression of endogenous SL-1 as a function of the expression of the rat transgene. Reverse transcription-PCR amplification suggested that endogenous SL-1 mRNA was up-regulated as a consequence of SL-1 transgene expression (Figure 1A) ▶ . As a more quantitative measure of endogenous SL-1 mRNA, total RNA from mammary glands of normal and transgenic mice was analyzed by Northern hybridization. Endogenous SL-1 mRNA was up-regulated in mammary glands from 70-day virgin and 10- and 15-day pregnant transgenic mice, relative to their normal counterparts (Figure 1 ▶ , B and C). However, the difference in the expression of endogenous SL-1 in glands from transgenic and normal mice at 6 or 18 days of pregnancy was not statistically significant, and there was no difference during lactation and involution. Three other MMPs that have characteristic expression patterns during mammary gland development, matrilysin, gelatinase A, and SL-3, 9-15 were also examined by reverse transcription-PCR. The levels of these enzymes are too low to be detected by northern blots. Matrilysin, which is expressed only by epithelial cells, was expressed in glands from 70-day virgin transgenic mice, but was not found in glands from normal mice at the same developmental stage (Figure 1A) ▶ . SL-3, which is restricted to the mammary stroma, and gelatinase A were detected in glands from lactating transgenic, but not normal, mice (Figure 1A) ▶ .

SL-1 expression in glands from normal and transgenic animals was restricted to periductal and intralobular fibroblasts. However, in transgenic mice, the fibrous connective tissue was extended into the adipocyte compartment (Figure 2) ▶ . The pattern of endogenous SL-1 expression in transgenic animals during mid-pregnancy was reminiscent of the expression pattern of SL-1 observed in glands from normal involuting mice (Figure 2 ▶ , d and f). However, the pattern of expression in involution was similar in normal and transgenic mice (Figure 2 ▶ , f and h). In both normal and transgenic mice, endogenous SL-1 was expressed in fibroblasts surrounding epithelial cords during involution. In all cases, the endogenous SL-1 mRNA was undetectable in the epithelial compartment (Figure 2) ▶ .

Both Epithelial and Adipocyte BMs Are Degraded in Glands from Pregnant Transgenic Mice

During normal mammary gland involution, BM disruption is paralleled by increased expression of a number of MMPs in the stromal compartment. In our previous study, we documented loss of BM around epithelial components of the mammary glands from lactating transgenic mice. 14 When mammary glands from virgin (data not shown) and 6-day pregnant (Figure 3) ▶ transgenic animals were examined by immunostaining for laminin and type IV collagen, BMs appeared intact, despite the observed increase in endogenous SL-1 mRNA in the virgin animals. However, BM fragmentation was obvious in glands from 15-day pregnant mice (Figure 3) ▶ . Interestingly, not only the BMs around alveoli, but also those around adipocytes were affected (Figure 3 ▶ , d and f), whereas the BMs of blood vessels remained intact (Figure 3d) ▶ , indicating the significance of locally secreted SL-1, as opposed to systemic changes.

We had proposed previously that loss of BM in culture leads to apoptosis of mammary epithelial cells. 27 It was also reported that apoptosis occurs in SL-1 transgenic animals during late pregnancy. 13 To determine when unscheduled apoptosis was initiated in the glands from pregnant transgenic animals, we examined apoptosis across all developmental stages. No difference in apoptosis between glands of wild-type and transgenic animals was seen at day 10 of pregnancy or earlier (not shown). However, by 15 days, when the glands of transgenic mice appeared morphologically underdeveloped, there were 37 cells, (±13% of the total number of cells) and a range of 2 to 12 (range per field) times the number of apoptotic cells per high-powered field in transgenic gland sections compared with the normal glands (Figure 4 ▶ ; Table 1 ▶ ). Therefore, the transgenic glands showed a peak increase of 7 to 10-fold times the apoptotic index (apoptotic cells/total cells) at 15 to 16 days of pregnancy, compared with normal glands. The enhanced apoptosis in glands from transgenic animals was suppressed during lactation (not shown). These results indicate that the transgenic phenotype of smaller alveoli and lower β-casein expression during the later stages of pregnancy can be attributed to unscheduled apoptosis. Our data also suggest that the lactational phenotype is strongly dominant, because it overrides the aberrant consequences of SL-1 transgene expression during pregnancy.

Tenascin-C Is Expressed Inappropriately in Glands from Pregnant and Lactating Transgenic Mice

Tenascin-C is up-regulated in the mammary gland when epithelial-mesenchymal communications are altered, such as in involution or in the stromal reaction in response to breast cancer. 23,28,29 Thus, we would expect inappropriate tenascin-C expression if the microenvironment of the mammary stroma were altered to resemble the involuting and/or the tumor stroma. Tenascin-C mRNA was not expressed in glands from virgin, pregnant, or lactating normal mice, but was abundant during involution (Figure 5A) ▶ . There was no tenascin-C expression in glands from virgin transgenic mice, consistent with the observation that BM was not degraded. During pregnancy and lactation, however, tenascin-C was expressed in transgenic mammary tissue (Figure 5A) ▶ . Interestingly, two tenascin-C mRNA species were found, with the larger transcript predominating in pregnancy and involution. In contrast, only the smaller mRNA was present during lactation in transgenic mice (Figure 5A) ▶ . Furthermore, the amount was very small, underscoring again the dominance of the lactational phenotype mentioned above. Mammary glands from normal mice expressed both tenascin-C transcripts at similar levels only during involution, although in contrast to transgenic animals, expression was much lower at day 2 than at day 4 of involution. Tenascin-C protein was localized to the proximity of the BM surrounding the alveoli and ducts in glands from lactating transgenic mice (Figure 5B) ▶ .

Collagen and Vascularization Are Increased in Glands from Transgenic Mice

The results described above suggest that in glands from pregnant transgenic mice, both stromal and epithelial compartments express characteristics usually found only during involution in normal mice, and in the stroma from breast cancer patients. 9,28 To further explore this possibility, we stained for collagen content throughout mammary gland development using Gomori’s trichrome method. There was a considerable increase in collagen deposition around the ducts and alveoli as well as in the adipose compartment, in both virgin and pregnant transgenic animals (Figure 6) ▶ . Collagen deposition increased approximately 8-fold in virgin and 2-fold in 6-day pregnant glands from transgenic mice compared with normal mice (Table 2) ▶ . At day 15 of pregnancy and during lactation, the number of ducts and alveoli were no longer significantly different in normal and transgenic mice; however, there was still a statistically significant increase in collagen content at day 15 of pregnancy, because of increased deposition around the ducts.

Reactive stroma is characterized by increased vascularization. We noted that the number of blood vessels in transgenic glands appeared more numerous than in normal glands in Gomori’s trichrome-stained sections. This observation was confirmed when we immunostained sections with antibodies against von Willebrand factor (vWF), a specific marker of endothelial cells (Figure 7A) ▶ . To quantify the blood vessels, we then immunoblotted gland extracts and found a 2-fold increase in vWF per unit weight of mammary gland (not shown). The increase in vascularization was observable also by measuring PECAM-1 mRNA. At all developmental stages examined, we found an increase in mRNA for the adhesion molecule PECAM-1, a constitutive marker of endothelial cells (Figure 7 ▶ , B and C). 21