Cyclophilin A Function in Mammary Epithelium Impacts Jak2/Stat5 Signaling, Morphogenesis, Differentiation, and Tumorigenesis in the Mammary Gland

Identified initially as a cyclosporine A (CsA)-interacting protein (1), cyclophilin A (CypA) was subsequently recognized as a cis-trans prolyl isomerase (2). Although classically thought to assist in protein folding, a significant body of evidence has implicated cyclophilins as signaling “switches” (3). In essence, the process of isomerization of the proline imide bond induces a conformational change in the peptide substrate backbone, enabling activation/inactivation of its intrinsic enzymatic and/or functional activity. Numerous examples of the switching activity of cyclophilins have been presented, including regulatory roles in cell surface receptor (4–6), kinase (6, 7), and transcription factor function (8).

Current data implicate cyclophilins both in physiologic and pathophysiologic processes in humans. Cyclophilin B contributes to collagen I folding and its loss is associated with the phenotype of osteogenesis imperfecta (9), while cyclophilin A contributes to vascular endothelial proliferation and remodeling (10). In malignancy, CypA is overexpressed in lung, pancreatic, and oral squamous cancers (11, 12). At the cellular level, inhibition of CypA by pharmacologic or loss-of-function approaches resulted in decreased in vitro growth of hepatocellular, breast, and non–small cell lung cancer cells (12, 13). Of note, highly significant reductions in the incidence of breast cancer were noted in a cohort of 25,000 female patients that had received CsA as a mainstay of their immunosuppressive therapy following renal or cardiac allograft (14).

The recognition of the role of CypA in the pathobiology of breast cancer arose from our observation of a critical function served by this foldase during prolactin receptor (PRLr) signaling. Classically associated with the terminal maturation of the mammary gland during pregnancy (15, 16), prolactin has also been shown to contribute to mammary oncogenesis (4, 17). This is supported by the following observations: (i) breast cancer risk is significantly elevated in women who demonstrate high levels of serum PRL (18), (ii) mining of expression array databases and IHC analyses reveal marked increases in PRLr expression in malignant breast tissues across ER⁺/ER⁻/Her2⁺ phenotypes (19–21), and (iii) in vivo gain- and loss-of-function studies targeting prolactin significantly alter mammary tumor latency (22, 23).

Given the significance of PRLr signaling during mammary development and the pathogenesis of breast cancer, our lab has sought to identify mechanisms that regulate the activation of PRLr-associated signaling pathways. In this context, our studies revealed that CypA, specifically its isomerase activity, was required for the phosphorylation and activation of the tyrosine kinase Jak2 following ligand binding (4, 24). Like the PRLr, Jak2 has also been implicated at the in vitro and in vivo level as significantly contributing to mammary development, lactation, and oncogenesis (25–28). Mechanistically, mutagenic approaches revealed that a proline residue in the intracellular domain (P334), near the “Box 1” Jak2-binding motif, was necessary for CypA engagement. As in vitro loss-of-function approaches further confirmed the relationship between CypA and Jak2, a natural extension of these studies was to examine the effect of small molecular prolyl isomerase inhibitors such as CsA and NIM811 (29, 30). Treatment with these agents blocked Jak2/Stat5 activation and resulted in the in vitro inhibition of ER⁺ and ER⁻ breast cancer cell proliferation, motility, invasion, and anchorage-independent growth (24). Parallel in vivo studies utilizing murine xenograft models revealed that CsA treatment inhibited metastatic progression of both ER⁺ and ER⁻ breast cancer cells (24). Collectively, these studies support the hypothesis that prolyl isomerase inhibition may represent a novel therapeutic strategy for human breast cancer, in part through its inhibition of Jak2 kinase activation in mammary epithelium.

The 129S6 x SvEv Ppia^−/− mice described previously (6, 31) were obtained from the Jackson Laboratory. Given that most mouse mammary gain-/loss-of-function models are bred into the FVB background, marker-assisted accelerated backcrossing (MAX-BAX) was performed at Charles River Laboratories to generate congenic FVB ppia^+/− mice (>99.9% FVB background), which were used in all studies described herein. Immunoblot analysis of mouse tissues from ppia^−/− confirmed the loss of CypA expression, as documented previously (see Supplementary Fig. S1; refs. 6, 31). To evaluate differences in growth, ppia^+/+, ppia^+/−, and ppia^−/− mice were weighed twice a week for 3 weeks, and then once a week for 24 weeks with a minimum of 5 mice per cohort.

For explant cultures, the fourth set of mammary glands was dissected from 14-week-old CypA^+/+ and Cyp A^−/− mice under aseptic conditions. Mammary glands were cut into 3 mm³ pieces and placed into prewarmed 37°C DMEM/F12 media with HEPES and 1% penicillin/streptomycin and incubated for 18 hours at 37°C to starve cells of prolactin. These tissues were then incubated in the presence or absence of 250 ng/mL prolactin for 15 minutes, followed by harvesting in ice-cold PBS, prior to homogenization, lysis in Laemmli buffer, and Western blot analysis.

For tumorigenesis studies, the well-recognized mouse model of transgene-driven tumorigenesis (32) utilizing the MMTV-driven rat erbB2 receptor containing an activating mutation (Val664Glu) was employed. Mice expressing this transgene were obtained from Jackson Laboratories (strain #005038) and were bred successively to obtain erbB2 x ppia^+/+, x ppia^+/−, and x ppia^−/− offspring on the FVB background. The progeny of these crossings were palpated weekly until tumors were noted; subsequently the mice were palpated thrice weekly. Mice were sacrificed when the tumor approximated 10% of body weight, or lethargy or tumor excoriation was noted, in accordance with Northwestern University Institutional Animal Care and Use Committee (IACUC) guidelines. Following sacrifice, all tumors were promptly dissected and weighed postmortem. Following fixation with 10% buffered neutral-buffered formalin, tumors were paraffin-embedded and processed for histologic and histochemical analyses. No significant differences in morphology in the resultant tumors were noted between the ppia^+/+, ppia^+/−, or ppia^−/− x erbB2 crosses. All studies described herein were approved by the Northwestern University IACUC. For in vivo pharmacologic inhibition, Nu^−/− mice were gavaged with the nonimmunosuppressive, cyclophilin prolyl isomerase inhibitor NIM811 twice daily for a total dose of 100 mg/kg/day, prior to prolactin stimulation.

Mammary gland cross-transplants were performed as described previously (33). Anesthetized, 3-week old female recipients underwent surgical excision of their fourth mammary tree. The resected tissue was reserved for whole mount processing to verify that the entire ductal tree was removed from the fat pad. In turn, the fourth mammary gland of 5- to 10-week-old virgin female donors was dissected and the 1-mm piece of mammary tissue obtained was inserted into the cleared fat pad of the recipient. Recipient mice were sacrificed at 14 weeks of age for analysis of the transplanted mammary gland by whole mount analysis.

Mouse liver homogenates were analyzed by Western blot analysis as described previously (24). Antibodies utilized for these studies were obtained from the following sources and used as described: anti-CypA (EMD Millipore), anti-tubulin (Life Technologies). Mouse mammary gland homogenates were also subjected to Western blot analysis. Antibodies utilized for these studies were obtained from the following sources and used as described: anti-pStat5 (Cell Signaling Technology, 1:1,000), anti-Stat5 (Santa Cruz Biotechnology, 1:1,500) diluted in TBS-T with 3% soy protein isolate (NOW Sports).

Dissected fourth mammary gland pairs were split with one-gland formalin-fixed/paraffin embedded and the other processed for whole mount staining (34). Whole mount glands were fixed in 10% formalin overnight, washed, and then placed into acetone overnight to remove intramammary fat. Following extensive washing, the gland was digested with 1 mg/mL collagenase II (Sigma) in PBS for 4 hours at 37°C. After washing, the processed gland was stained with Carmine Alum (Sigma) overnight. After rinsing, remaining connective tissue was microdissected to expose the mammary tree. Following dehydration, the mounted tissues were covered with Entellan New (Electron Microscopy Sciences).

For hematoxylin/eosin and IHC labeling, harvested mammary glands were fixed in 10% formalin, dehydrated, and paraffin embedded. Five-micron sections were obtained and stained with hematoxylin and eosin (Sigma). IHC labeling was performed as described previously (20) All primary antibodies were obtained from the following sources and utilized at the indicated titers: cleaved-caspase-3 (Cell Signaling Technology, 1:1,000), Ki-67 (Cell Signaling Technologies, 1:300), CISH (Abcam, 1:800), phospho-Stat5a (Cell Signaling Technology, 1:25), lactalbumin (Santa Cruz Biotechnology, 1:300), whey acidic protein (Santa Cruz Biotechnology, 1:800), β-casein (Abbiotec, 1:300). After washing, slides were incubated with SignalStain Boost IHC Detection Reagent (Cell Signaling Technology) for 30 minutes, followed by incubation with SignalStain DAB Substrate (Cell Signaling Technology) for 10 minutes. Washed slides were counterstained with hematoxylin (Gill No. 3, Sigma) and incubated with 1× Scott Tap Water Substitute (Sigma) for additional bluing. Slides were dehydrated and coverslipped using Permount mounting medium (Electron Microscopy Sciences). Sections from at least three harvested mammary glands from independent mice from each time point were scored. Cleaved caspase-3 and Ki-67 staining were expressed as percent positive cells; all others as an Allred score, namely as a function of % positive cells × mean intensity (0–3 scale).

For enzymatic dissociation, dissected fourth mammary gland pairs from 14-week-old CypA^+/+ or CypA ^−/− female mice were incubated for 16 hours at 37°C in 5 mL of 1× gentle collagenase/hyaluronidase (StemCell Technologies) in DMEM/F12 (1:1) medium (Life Technologies), supplemented with 5% FBS (Atlanta Biologicals) and 1% penicillin/streptomycin (Life Technologies). After incubation and centrifugation, the resultant cell pellet was resuspended in 5 mL of a 4:1 mixture of cold ammonium chloride (StemCell Technologies) and cold Hank Balanced Salt Solution Modified (HBSS, StemCell Technologies), supplemented with 2% FBS. Following centrifugation, the cell pellet was resuspended in 5 mL prewarmed 0.25% Trypsin-EDTA (Life Technologies), and the cell suspension was mixed by gentle pipetting with a P1000 disposable tip for 3 minutes. After washing, the cell pellet was resuspended in 2 mL of prewarmed dispase (5 mg/mL, StemCell Technologies), supplemented with 200 μL DNase I (1 mg/mL, StemCell Technologies). The resulting suspension was incubated for 5 minutes at 37°C, diluted with 10-mL cold HBSS supplemented with 2% FBS, and filtered through a 40-μm filter. The resuspended cells were enriched for epithelial populations (i.e., Lin⁻ cells) using the EasySep Mouse Epithelial Cell Enrichment Kit (StemCell Technologies), according to the manufacturer's protocol. For fluorescent labeling, cells were preincubated with Mouse BD Fc Block (BD Biosciences) at 2 μg per 100 μL cell suspension for 5 minutes at 4°C. The cells were then incubated with BV510 rat anti-mouse CD24 (BD Biosciences), Alexa Fluor 488 rat anti-mouse CD29 (BD Biosciences), and Alexa Fluor BV421 hamster anti-mouse CD61 (BD Biosciences) at 0.25 μg, 1 μg, and 0.25 μg per 100 μL cell suspension, respectively, for 30 minutes at 4°C. After washing, propidium iodide was added to the labeled cells at 50 ng per 100 μL cell suspension. Cells were analyzed on the BD LSRFortessa-X20 (BD Biosciences), using BD FACSDiva software for data acquisition and analysis at the VCU Massey Cancer Center Flow Cytometry Shared Resource Core. To characterize CD61⁺/29⁺ populations as described previously (35), intact cells were gated from debris and aggregates using forward and side scatter; dead cells were eliminated using propidium iodide exclusion. Dual parameter CD29 versus CD61 analysis was then performed on cells identified as CD24⁺.

Differences between group means were determined by unpaired Student t test assuming equal variances, with the exception of the immunoblot quantitation, for which a one-way ANOVA with the Holm–Sidak correction for multiple comparisons was performed. For all, and statistical difference defined as P < 0.05, using GraphPad Prism V6.0g (GraphPad Software, Inc.).

As the pathways regulating mammary gland development can also provide understanding of the mechanisms utilized during oncogenesis, it was reasoned that genetic ablation of the CypA locus might provide additional insight into the in vivo function of this isomerase in the normal and malignant development of the mammary gland. While a nonconditional CypA^−/− mouse had been previously described to be resistant to the effects of CsA (31), no study to date had examined the loss of CypA function during mammary development and oncogenesis. Given our precedent data, we hypothesized that this CypA^−/− mouse should demonstrate decreased in vivo Jak2/Stat5 signaling and delayed/altered mammary development and tumorigenesis. To facilitate its use with standard genetic mouse models of mammary neoplasia, the nonconditional CypA^−/− mouse was back-bred to congenicity onto the FVB background. Confirmation of the loss of CypA in the resultant progeny was confirmed at the level of protein expression by immunoblot analysis (Supplementary Fig. S1A). CypA^−/− × FVB mice (hereafter referred to as CypA^−/− mice) were viable, demonstrating normal estrous cycles, fertility, and litter size with expected Mendelian ratios, consistent with prior reports (31).

The effect of CypA loss on mammary gland development, however, was significant, as observed by both whole mount and histologic analysis. Demonstrated in Figs. 1A–C, these analyses reveal a marked loss of alveolar budding in the mammary glands of 14-week-old virgin CypA^−/− mice, and was comparable with that seen with the PRLr^−/− (36) and Stat5^−/− (37, 38) mice. Morphometric analysis of these glands, quantitatively confirmed significant reductions in the number of alveolar buds (Fig. 1C). In parallel, a persistence of terminal end buds was noted (Fig. 1D), while their penetration into the mammary fat pad was not altered (Fig. 1E). Interestingly, while alveolar budding was profoundly decreased by CypA deletion (Fig. 1c), the number of secondary ductal branches was not altered (not shown). Occasional duct ectasia was also noted in the CypA^−/− females. Altered development of the mammary gland of the CypA^−/− mice persisted through pregnancy and lactation. Alveologenesis proper was also significantly delayed in the Cyp^−/− females both at day 15 of pregnancy (P15) and day 2 of lactation (L2). The lack of luminal clearing and epithelial secretory granules in the CypA^−/− glands at P15 was also readily evident (Fig. 1A and B). In addition, while all CypA^+/+ mice at L2 demonstrated a 90%–100% effacement of the mammary fat with alveolar epithelium, all CypA^−/− females revealed only a 50%–80% replacement (Fig. 1A and B; Supplementary Fig. S1B). Finally, a far faster involution of the mammary epithelium was noted in CypA^−/− females (Fig. 1A and B) at both involution days 2 and 6 (I2 and I6). As a note, the loss of only a single allele, that is, the CypA^+/− mouse, did not result in any discernable change in mammary morphology, as the mammary development of the heterozygous females paralleled that of the homozygous littermates.

Given the significant morphologic alterations noted in the mammary glands of CypA^−/− females, it was reasoned that the functional differentiation of the lactating mammary gland may also be altered. Indeed, both the PRLr^−/− and Stat5^−/− mice were incapable of supporting lactation (15, 39). To that end, quantitative and qualitative assessments of milk production were made. To assess for quantitative alteration in milk production, litters of entirely CypA^+/− progeny were bred and nursed from their corresponding CypA^+/+ or CypA^−/− mothers. Litters were normalized down to n = 5 and weighed twice a week. No statistically significant differences in litter weight or ability to thrive were detected (not shown). Qualitative assessment of milk protein constituents in mammary glands from P15 and L2 females utilized IHC analysis. There was no statistically significant alteration of either beta casein or whey acidic protein at the level of IHC (Supplementary Fig. S2). As seen in Fig. 2A and Supplementary Fig. S2A, however, a highly significant decrease the expression of lactalbumin in the mammary glands of CypA^−/− females was observed at both P15 and L2. Thus, the effects of loss of CypA on both quantitative and qualitative milk production do not appear as severe as noted for both the PRLr^−/− and Stat5^−/− mice.

Our lab has previously shown that CypA facilitates PRL-driven Jak2/Stat5 signaling in vitro in breast cancer cells via its prolyl isomerase activity on the PRLr/Jak2 complex (24). To examine whether the developmental morphologic and qualitative lactation changes noted in the CypA^−/− mammary gland was related to alterations in Jak2/Stat5 signaling, IHC analysis of CypA^+/+ and CypA^−/− mammary glands with anti-phospho-Jak2 and anti-phospho-Stat5 (anti-pJak2 and anti-pStat5, respectively) was performed at P15 and L2 (Fig. 2B and C; Supplementary Fig. S2B and S2C). At both time points, significant reductions in pJak2 and pStat5 expression in the CypA^−/− mice were noted. Taken as a whole, these data revealed that the developmental alterations noted in the CypA^−/− mammary glands were associated with significant decreases in the phosphorylation/activation status of Jak2 and Stat5.

CypA^−/− females demonstrated normal estrous cycles and fecundity, suggesting an intact hypothalamic–pituitary–ovarian hormonal axis, as well as normal serum prolactin, growth hormone, and IGF1 levels (as determined by the National Hormone and Pituitary Program; not shown). Nevertheless, the changes found in the nonconditional CypA^−/− mouse could be secondary to its impact on either or both of the endocrine and stromal axes. To exclude these possibilities, mammary gland cross-transplants were harvested at 14 weeks of age, following clearing of the mammary fat pad of recipients of their native mammary gland at 3 weeks of age. Analysis of the cross-transplanted mammary glands by whole mount analysis (Fig. 3A) was striking. Similar levels of alveolar budding were noted when comparing the control transplant (e.g., CypA^+/+ epithelium into a CypA^+/+ stromal fat pad) to the heterologous transplant of CypA^+/+ epithelium into CypA^−/− stromal fat pad. In contrast, a highly significant reduction in the number of alveolar buds was observed when the CypA^−/− epithelium was transplanted into the CypA^+/+ stromal fad. These data confirmed that the developmental defects noted in the mammary glands of CypA^−/− females was the result of a primary epithelial defect, and not the result of endocrine hormones or the stromal microenvironment.

To further confirm that the prolyl isomerase activity per se of CypA was necessary for normal mammary development 8-week-old virgin mice were treated for 6 weeks with the nonimmunosuppressive inhibitor of cyclophilin prolyl isomerase activity, NIM811 (30). As clearly seen in Fig. 3B, NIM811-treated mice demonstrated a marked paucity of alveolar budding in the mammary gland, comparable with that seen in comparably aged Cyp^−/− females. Taken together with the cross-transplant data, these findings argue that it is the prolyl isomerase activity of CypA within the mammary epithelium that is necessary for appropriate mammary alveologenesis.

Given the reduction in endogenous levels of pJak2 and pStat5 in CypA^−/− mice (Fig. 2; Supplementary Fig. S2) during pregnancy and lactation, periods in the reproductive cycle demonstrating high levels of serum prolactin, it was conjectured that prolactin-induced signaling in a nonlactating mouse mammary gland should also be reduced. To test this hypothesis, 20-week virgin females were injected intraperitoneally with PRL, and after 30 minutes the mammary glands were harvested and prepared for IHC labeling and immunoblot analysis with anti-phospho-Jak2 or pStat5 antibodies. As hypothesized, the PRL-induced expression of both pJak2 and pStat5 in the CypA^−/− mammary glands was profoundly reduced (Fig. 4A). Further confirmation of the reduction of prolactin-induced Stat5 phosphorylation at the biochemical level was obtained by anti-pStat5 analysis of PRL-stimulated CypA^+/+ and CypA^−/− mammary explants (Fig. 4B and C).

Alterations in mammary gland morphology and differentiation have been linked to perturbations in the hierarchical organization of the mammary epithelium, consisting of identifiable stem, progenitor, and mature luminal cell populations. These relationships have been made evident in part through the use of loss-of-function mouse models for the transcription factors Gata-3 (35), Runx2 (40), and Elf5 (41). Given that loss of Jak2 (42) and Stat5 (38, 39, 43) function are also associated with alterations in mammary morphology, it was reasoned that the CypA^−/− mouse might also demonstrate alterations to its epithelial hierarchy. To that end, flow cytometry on freshly disassociated, lin⁻, CD24⁺ cells was utilized to quantitate stem (CD29^hi/CD61^hi), progenitor (CD29^lo/CD61^hi), and mature luminal (CD29^lo/CD61^lo) populations in 14-week-old virgin CypA^+/+ and CypA^−/− mice, using previously described and validated methodology (35). These analyses (Fig. 5A) revealed significant differences between the CypA^+/+ and CypA^−/− mice, as a 15-fold increase in the stem and a 34-fold increase in the progenitor populations were observed in the CypA^−/− mice. Correspondingly, the CypA^−/− mice also demonstrated a significant decrease in percentage of mature luminal cells. Disruption of the normal mammary development by genetic loss-of-function has frequently been associated with decreased cellular proliferation in the mammary gland (35, 40), and is associated with delayed morphologic development and a decrease in the volume occupied by the epithelial compartment. To assess whether a similar phenomenon was occurring in the CypA^−/− mice both anti-Ki67 and -cleaved caspase-3 IHC analyses were performed on the mammary glands from 14-week-old virgin mice (Fig. 5B; Supplementary Fig. S3). These analyses revealed that a profound decrease in proliferation occurred in the mammary epithelium of the CypA^−/− mice, whereas the rates of apoptosis between the CypA^+/+ and CypA^−/− mice were comparable. Taken together, these findings indicate that loss of CypA results in delayed maturation of stem and progenitor populations into mature luminal cells that is manifest in a decrease in observable proliferation in the mammary epithelium.

Many mouse loss-of-function models that produce altered mammary development, also demonstrate delayed or disrupted tumorigenesis when crossed into the appropriate line expressing a transgenic mammary oncogene (44–46). We hypothesized that mammary tumorigenesis driven by a mammary oncogene, such as ErbB2/neu, would be significantly delayed by its introduction into the CypA^−/− background. We based this hypothesis on the following observations: (i) Significant delays in alveolar budding with the CypA^−/− mouse were noted (Fig. 1A–C); (ii) Jak2-mediated signaling has been found to influence erbB2 transduction and erbB2/neu-driven mammary tumorigenesis (27); and (iii) our data revealing that CypA regulates Jak2 activation in breast cancer cell lines (24), in the developing mammary gland (see Figs. 2 and 4), and in human breast cancer xenografts (24). To that end, CypA^+/− x FVB and erbB2 x FVB mice were mated, and after remating of select progeny, CypA^+/+ and CypA^−/− (xFVB) female littermates were followed for mammary tumor development for 86 weeks. As presented in the Kaplan–Meier curves in Fig. 6A, a highly significant, 2-fold delay in tumor latency was noted in the CypA^−/− females (CypA^+/+ median latency = 190 days vs. CypA^−/− = 475 days). In parallel, survival of the CypA^−/− females was significantly enhanced (CypA^+/+ = 245 days vs. CypA^−/− = 480 days). Parallel studies with CypA^+/− x erbB2 females showed no increase in tumor latency or survival, paralleling the curves of the CypA^+/+ mice. Interestingly, the time from observation of the initial primary tumor to euthanasia was only modestly increased between CypA^+/+ and CypA^−/− mice (CypA^+/+ = 51 days versus CypA^−/− = 70 days), suggesting that cyclophilins primarily contribute to the progression of the precursor lesion into the primary invasive tumor, and secondarily promote the growth of the primary tumor per se. This was further substantiated by analysis of mammary tumor multiplicity in the mouse cohorts at time of euthanasia/death; CypA^+/+ mice (9.75 ± 1.65; mean tumors/mouse ± SEM) demonstrated significantly more (P < 0.005) discrete carcinomatous foci than did CypA^−/− mice (1.75 ± 0.25)

To assess whether parallels between the developing mammary gland and erbB2-driven tumorigenesis existed in the CypA^−/− mice, the levels of proliferation, apoptosis, Stat5 phosphorylation, and expression of the Stat5-responsive CISH gene were assessed by IHC in primary tumors from CypA^+/+ and Cyp^−/− mice (Fig. 6B; Supplementary Fig. S4). As was seen in the 14-week-old mammary gland, erbB2 tumors from CypA^−/− mice demonstrated decreased Stat5 activation, CISH expression, and Ki67 labeling, but unchanged levels of cleaved caspase-3 labeling, when compared with CypA^+/+ mice. Taken together, these data suggest that the function of signaling pathways altered by CypA loss contribute to both mammary development and tumorigenesis.

The developmental delays in mammary gland differentiation in the CypA^−/− mouse demonstrate many similarities to other loss-of-function mouse models of the PRL–PRLr–Jak2–Stat5 signaling pathway (37, 38, 42). The requirement for this signaling axis is required for alveolar budding and terminal end bud regression. This phenotype was also observed in the CypA^−/− mouse (Fig. 1) and is consistent with the role of CypA in PRLr-driven activation of Jak2 (24). Correspondingly, it is not surprising that secondary ductal branching in CypA^−/− mice was preserved, as this process appears to be principally driven by the actions of progesterone, as has been noted in PRL^−/− females supplemented with progesterone (36, 47).

Defective alveolar development proper persisted during pregnancy in the CypA^−/− mouse, as a lack of luminal clearing of the alveoli and secretory granule formation in the epithelium was observed. This too has been noted in genetic deletions in the PRL–PRLr–Jak2–Stat5 pathway (37–39, 42, 47). What distinguishes the CypA^−/− mouse from these other genetic deletion models, however, was their ability to successfully lactate and nurse their progeny, as measured by the viability and weights of CypA^−/− litters. A trivial explanation for this is that the CypA^−/− mouse was on a congenic FVB background (i.e., a strain-related phenomenon), unlike many of the other noted genetic models. We view this possibility as unlikely, given that no difficulties in lactation or litter viability were noted in the parental line of mixed background, CypA^−/− 129S6 x SvEv, either by our laboratory or the laboratory where this strain was generated (31). A more likely basis for the mammary phenotype of the CypA^−/− mouse, may relate to a significant, but not complete, loss of Jak2/Stat5 signaling (Fig. 2; Supplementary Fig. S2). Precedent in vitro studies by our laboratory have revealed a significant (70%–90%) inhibition of the Jak2/Stat5 pathway in vitro following CypA loss or inhibition in breast cancer cells (24). While there were no significant differences in comparably sized litters from either CypA^−/− or CypA^+/+ dams, lactation in the CypA^−/− mouse was not entirely normal. Morphologically, full expansion of the alveolar epithelium into the mammary fat pad during lactation did not occur in CypA^−/− females (Fig. 1). Similarly, unlike the PRLr^−/− mouse that demonstrates a global deficiency of milk protein constituent synthesis, the loss of milk protein synthesis in the CypA^−/− mouse was selective with only a significant decrease of lactalbumin expression noted in CypA^−/− alveolar epithelium (Fig. 2; Supplementary Fig. S2). The selective differences in milk protein expression observed may suggest that proteins such as lactalbumin require a higher level of induced Stat5 activity, not achieved in the CypA^−/− mouse. Taken together, these data suggest that titration of Jak2/Stat5 function by CypA has discrete impacts on mammary development.

Precedent studies (36, 47) using the nonconditional PRLr^−/− mouse have revealed that the defect in alveologenesis and lactation in this model is epithelial and not stromal (or endocrine) in origin. Mammary gland cross-transplants of CypA^+/+ and CypA^−/− donors and recipients demonstrated a similar phenomenon, namely the developmental defects observed in CypA^−/− mice, are epithelial in nature. The in vivo role of CypA in epithelial differentiation was also confirmed by treatment of CypA^+/+ mice with the nonimmunosuppressive analogue of cyclosporine A (CsA), NIM811. The nonimmunosuppressive properties of NIM811 are the result of a single amino acid side-chain replacement, resulting in its inability to engage calcineurin A. As a consequence, NIM811 is not immunosuppressive and has none of the other toxic side effects of CsA (30, 48). Treatment of 14-week-old virgin females, resulted in the inhibition of alveolar budding as was observed with the CypA^−/− females (Figs. 1 and 4). While the effect of NIM811 on lactation proper was not examined here, given the data presented here with the CypA^−/− mouse and a report of successful lactation in a female patient receiving cyclosporine (49), we would predict that NIM811-treated mice would successfully lactate, but with alterations in milk composition and fat pad effacement as noted above. Alterations in mammary gland development, as seen in the CypA^−/− mouse, have been frequently associated with delayed/inhibited maturation of the epithelial hierarchy of stem, progenitor, and mature luminal cells in the mammary gland, as determined by flow cytometric and functional approaches (35, 40, 43). Significant expansion of the CD61^hi populations were noted for both stem (CD29^hi/CD61^hi; 15-fold increase) and progenitor (CD29^lo/CD61^hi; 34-fold increase) populations. Correspondingly, a decrease in mature luminal cells (CD29^lo/CD61^lo) was also noted. Taken together, these observations are consistent with inhibited differentiation in CypA^−/− mammary epithelium (Fig. 5), particularly at the progenitor to mature luminal level, paralleling what was seen in the above cited loss-of-function mouse models.

When crossed into an erbB2 model of transgene-driven mammary carcinogenesis (32), CypA^−/− x erbB2 mice demonstrated significantly prolonged tumor latency/survival and decreased tumor multiplicity in comparison with the Cyp^+/+x erbB2 cohorts (Fig. 6). This may in part result from the decreased levels of Stat5 activation, Stat5-driven gene expression, and proliferation noted in CypA^−/− x erbB2 tumors (Fig. 6; Supplementary Fig. S4). However, only a modest difference in the time-to-death from first tumor observation between the CypA^−/− and Cyp^+/+ mice was noted, as was seen with the Jak2^−/− x erbB2 mice (27). Thus, the CypA^−/− and Jak2^−/− mice demonstrate some similarities in their effects on tumor initiation and primary tumor growth (24, 27, 45). Treatment of ER⁺ and ER⁻ xenografts with the prolyl isomerase inhibitor cyclosporine A (CsA), following initial tumor establishment in the absence of CsA, also had only a modest effect on primary tumor growth; however, metastasis in these treated models was profoundly inhibited (24). A significant difference, however, between the CypA^−/− x erbB2 and Jak2^−/− x erbB2 (45) mouse models is observed in their respective differential delays in median tumor latency in comparison with wild-type x erbB2 mice (delays of 235d for CypA^−/− x erbB2 vs. wt mice versus 42d for Jak2^−/− x erbB2 vs. wt mice). This would suggest with respect to mammary tumorigenesis, the loss of CypA function may have a broader action, perhaps by affecting relevant signaling pathways other than those directly associated with Jak2. Indeed, while our working model of CypA action during PRLr signaling is focused on the activation of Jak2 by CypA, many downstream Jak2 targets such as AKT and MAPK (24), as well as the Tec tyrosine kinase family (7, 50), are regulated by CypA activity. Taken together, our findings indicate that CypA significantly contributes to in vivo PRLr-mediated signaling, mammary gland development/differentiation and tumorigenesis, representing a novel therapeutic target at the levels of chemoprevention and the inhibition of metastatic progression.

No potential conflicts of interest were disclosed by the authors.

Conception and design: C.V. Clevenger, S.E. Volker

Development of methodology: C.V. Clevenger, S.E. Volker, S.E. Hedrick

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.V. Clevenger, S.E. Volker, S.E. Hedrick, Y.B. Feeney

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.V. Clevenger, S.E. Volker, S.E. Hedrick, Y.B. Feeney

Writing, review, and/or revision of the manuscript: C.V. Clevenger, S.E. Volker, S.E. Hedrick

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.V. Clevenger, S.E. Volker, Y.B. Feeney

Study supervision: C.V. Clevenger

This work was supported in part by grants from the NIH, R01 CA173305 and the American Association for Cancer Research/Breast Cancer Research Foundation Research grant (grant number: 10-60-26-CLEV; to C. Clevenger).

Services and products in support of the research project were generated by the VCU Massey Cancer Center Flow Cytometry Shared Resource, Transgenic/ Knockout Mouse Core, and the VCU Microscopy Facility, supported in part with funding from NIH-NCI Cancer Center Support Grant P30 CA016059; and by the Department of Pathology, Anatomic Pathology Research Services, at VCU Health. The assistance of Alicia Wook and Jacqueline Gribble in immunoblot optimizations is acknowledged.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.