Inhibition of the Stromal p38MAPK/MK2 Pathway Limits Breast Cancer Metastases and Chemotherapy-Induced Bone Loss

Breast cancer is one of the leading causes of cancer-related deaths in women in the United States (1). The mortality is largely attributed to metastasis of the disease from the primary site to other organs. Currently, there are limited therapeutic options for breast cancer metastases, and it remains a clinical challenge. For this reason, there is a continued and unmet need to identify novel therapeutic targets that increase disease-free survival while at the same time limit the morbidities associated with disease progression and therapy.

Tumor progression is a complex process that is governed by both cell-autonomous alterations within tumor cells and ongoing changes in the tumor microenvironment (TME). Importantly, work over the last decade has revealed that the TME plays a complex, active, and insidious role in tumor progression (2, 3). Further, it is now clear that tumor cells and stromal cells collaborate to facilitate proliferation, migration, invasion, immune evasion, and resistance to therapy (4, 5). Tumor-associated stromal cells promote progression by expressing a plethora of tumor-promoting factors, which display a high degree of overlap with the senescence-associated secretory phenotype (SASP; ref. 6). Interestingly, many of these factors are regulated by the stress-kinase p38MAPK that we previously revealed support primary tumor growth in a stromal-dependent manner (7). In addition, in earlier work using a novel genetic model, we demonstrated that induction of senescence in osteoblasts induced localized osteoclastogenesis through secretion of IL6, a SASP factor known to be responsive to p38MAPK signaling (8), leading to conditioning of the premetastatic niche and subsequent increase in bone metastatic outgrowth (9).

p38MAPK is a central regulator of the inflammatory response, and its activation is vital for expression of an array of inflammatory cytokines and chemokines (10). Orally administered, small-molecule inhibitors of p38MAPK have been evaluated as potential therapeutic targets in several chronic inflammatory diseases including rheumatoid arthritis (RA; ref. 11), chronic obstructive pulmonary disease (COPD; ref. 12), Crohn's disease (13), and cancer (14). In RA, several p38MAPK inhibitors were discontinued from clinical trials as a result of side effects, such as elevated liver enzymes and skin rash. Further, some inhibitors that advanced in trials displayed only weak clinical efficacy and transient suppression of inflammatory cytokines and systemic inflammation markers (11). Similar to their effect in RA, p38MAPK inhibitors have not shown promising results in Crohn's disease clinical trials (13). However, the p38 inhibitor that showed transient efficacy in RA, when implemented in a clinical trial for patients with COPD, displayed remarkable improvements in symptoms and advanced to phase III trials (12). These studies demonstrate that efficacy of p38MAPK inhibitors is disease-specific.

Mouse model studies also revealed that the role of p38MAPK signaling in tumor progression is complex and variable depending on cell type and tumor type (15). Further, p38MAPK inhibitors have also been investigated for oncologic indications such as multiple myeloma (14) and advanced metastatic disease (NIH Clinical Trial # NCT01463631). Although the outcomes of these clinical trials are currently unknown, these inhibitors have proven effective in animal models of human cancer. One study revealed a cell-autonomous role for p38MAPK in p53-deficient tumor cells (16), whereas others demonstrate that p38MAPK signaling within stromal cells leads to paracrine support of tumor growth (7, 17). A recent study demonstrated that conditional global deletion of p38MAPK reduced tumorigenesis in the PyMT breast cancer model. The degree to which deletion of p38MAPK in tumor cells versus the stromal compartment contributed to the reduced tumorigenesis was not investigated, but this study underscores the complexity of the p38MAPK pathway (18). In contrast, a study using the 4T1 triple-negative model demonstrated that knockdown of p38MAPK within tumor cells had no impact on orthotopic tumor growth but increased metastatic growth in the lung (19). Taken together, these findings underscore the complexity of the p38MAPK pathway in cancer and highlight the need to investigate the therapeutic potential of the pathway in metastatic settings.

Approximately 70% of patients with metastatic breast cancer develop bone metastases (20). Once in the bone, the disease is incurable and treatment options are limited or only palliative. Metastatic progression typically results in bone loss, leading to a variety of skeletal complications characterized by bone pain, hypercalcemia, and pathologic fractures (21). Chemotherapy is often used after surgical resection in an attempt to prevent relapse (22, 23). However, chemotherapy has many debilitating side effects including the induction of additional bone loss that severely affects quality of life of patients (24). Because 70% of patients with metastatic breast cancer harbor bone metastases that cause severe osteolytic bone destruction, it is imperative to explore therapies that can both reduce metastatic burden and prevent bone loss.

Interestingly, p38MAPK signaling (particularly p38α) plays a key role in regulating osteoclast differentiation mediated by Receptor Activator of NF-κB ligand (RANKL; ref. 25). Numerous studies have reported that p38MAPK inhibitors are effective at preventing bone loss via suppression of p38MAPK-induced cytokines. In addition, mitogen-activated protein kinase-activated protein kinase 2 (MAPKAPK2 or MK2), a kinase downstream of p38MAPK, also plays a central role in osteoclastogenesis as evidenced by increased bone mass and decreased osteoclast number and resorption in MK2^−/− mice (26). We therefore postulated that blocking the p38MAPK/MK2 pathway might limit the tumor-promoting activities of the bone stromal compartment while simultaneously preserving bone quality, something the current standard of care cannot achieve.

We sought to investigate the therapeutic benefit of stromal inhibition of the p38MAPK/MK2 pathway in limiting breast cancer metastasis and protecting against bone loss. To address this, we utilized an aggressive murine breast cancer cell line, PyMT Bo-1 (27), in an intracardiac (IC) model of bone metastasis. The PyMT-Bo1 cell line mimics the human luminal B subtype of breast cancer, for which there are few effective therapies (28). Although the IC injection model does not recapitulate every step in the metastatic cascade, it does allow us to examine tumor growth after seeding in the bone, which cannot be achieved in any other model. In this study, we utilized a p38MAPK inhibitor (p38i) and a novel drug (MK2Pi) that directly disrupts the p38MAPK-MK2 interface, and discovered that both approaches led to significant decreases in bone and visceral metastases, similar to that observed in mice treated with the chemotherapeutic agent, Paclitaxel (PTX). In addition, the inhibitors preserved bone density even in the presence of chemotherapy, which is known to drive bone loss independent of tumor growth. Our studies suggest that targeting the p38MAPK/MK2 pathway could have clinically meaningful antitumor and bone-preserving effects in breast cancer.

Wild-type, female B6(Cg)-Tyr^c-2J/J (B6-albino) mice (ages 6–8 weeks) were used in all experiments involving PyMT-Bo1 cell injections, and wild-type, female FVB/NJ mice (ages 6–8 weeks) were used in all experiments involving Met-1 cell injections. All mice were obtained from JAX laboratories and were housed in accordance with Washington University in St. Louis's Studies Committee and Institutional Animal Care and Use Committee.

MMTV PyMT-Bo1 mouse breast carcinoma cells were obtained through collaboration with Dr. Katherine Weilbaecher's laboratory (27). Met-1 mouse breast carcinoma cells were a kind gift from Dr. Sandra McAllister. Both PyMT-Bo1 cells and Met-1 cells were cultured in DMEM supplemented with 10% heat-inactivated FBS (cat# F2442; Sigma) and antibiotics (100 U/mL of penicillin and 100 μg/mL of streptomycin, cat# P0781; Sigma). Both cell lines were used at low passage and regularly tested by PCR for Mycoplasma.

On the day of IC injection, 6-week-old female mice were anesthetized with 100 μL/20 g body weight of Ketamine/xylazine cocktail (17.7 mg/mL of ketamine and 2.65 mg/mL of xylazine). When animals were completely anesthetized, cells were injected directly into the left cardiac ventricle: either 50 μL of PyMT-Bo1 (GFP/Luc) cells (5 × 10⁴ cells) into B6-albino mice or 50 μL of Met-1 (GFP/Luc) cells (1 × 10⁵) into FVB/NJ mice. For mammary gland injections, 10⁵ PyMT-Bo1 cells were injected into the fourth inguinal mammary gland, and tumor weights were assessed at sacrifice after dissection.

Bone marrow macrophages (BMM) were obtained by culturing mouse bone marrow cells isolated from C57BL6 mice in culture media containing a 1:10 dilution of supernatant from the fibroblastic cell line, CMG 14–12, as a source of M-CSF (29), a mitogenic factor for BMM, for approximately 5 days in a 10-cm dish as previously described (30). Nonadherent cells were removed by vigorous washes with PBS, and adherent BMM were detached with trypsin-EDTA and cultured in culture media containing a 1:10 dilution of CMG.

To induce osteoclast formation, BMM were plated at 5 × 10³ cells per well in a 96-well plate in culture media containing a 1:50 dilution of CMG and 100 ng/mL RANKL, a required cytokine for osteoclast differentiation. CDD-450 or CDD-110 resuspended in DMSO was added to cell cultures to yield 0.5% DMSO final concentration. Control cultures were exposed to 0.5% DMSO final concentration. Media with supplements were changed every other day and maintained for 4 days at 37°C in a humidified atmosphere of 5% CO₂ in air.

Cytochemical staining for tartrate-resistant acid phosphate (TRAP) was used to identify osteoclasts as described previously (30). Briefly, cells on a 96-well plate were fixed with 3.7% formaldehyde and 0.1% Triton X-100 for 10 minutes at room temperature. The cells were rinsed with water and incubated with the TRAP staining solution (Sigma leukocyte acid phosphatase kit) at room temperature for 30 minutes. Under light microscopy, multinuclear TRAP-positive cells with ≥ 3 nuclei were scored as osteoclasts. Mesenchymal stem cells were treated with 50 μg/mL ascorbic acid and 10 mmol/L β-glycerophosphate to differentiate into osteoblasts, and cells were stained on day 7 for alkaline phosphatase per the manufacturer's protocol (Sigma).

Bioluminescence imaging (BLI) was performed as previously described (9). In vivo imaging was performed on an IVIS100 or IVIS Lumina (PerkinElmer; Living Image 3.2, 1–60 second exposures, binning 4, 8, or 16, FOV 15 cm, f/stop 1, open filter). Mice were injected i.p. with D-luciferin (150 mg/kg in PBS; Gold Biotechnology) and imaged 10 minutes later under isoflurane anesthesia (2% vaporized in O2). Animals were sacrificed immediately following whole-body imaging, and both hind limbs were isolated and imaged for 10 seconds ex vivo. For analysis, total photon flux (photons/sec) was measured from a fixed region of interest (ROI) over the whole body or bones using Living Image 2.6 software.

In vitro live-cell BLI was performed on an IVIS 50 (PerkinElmer; Living Image 4.3, 5-minute exposure, binning 8, FOV 12cm, f/stop 1, open filter) as previously described (31). D-luciferin (150 mg/mL; Gold Biotechnology) was added to black-walled plates 10 minutes prior to imaging. Total photon flux (photons/sec) was measured from fixed ROIs over the plate or tumors using Living Image 2.6.

Mouse femur bones were isolated and fixed in 10% neutral buffered formalin for 24 hours. Bones were decalcified in 14% EDTA for 14 days at 4°C, embedded in paraffin, and sectioned 5 μm thick at the histology core of the Washington University Musculoskeletal Research Center. Standard hematoxylin and eosin (H&E) technique was used for all bone sections. Images were collected using the Olympus NanoZoomer 2.0-HT System, Alafi Neuroimaging Laboratory. Immunohistochemical staining was carried out on formalin-fixed, paraffin-embedded slides as previously described (32). Slides were stained with the following antibodies: anti-IL6 primary antibody (ab6672, 1:100; AbCam), pMK2 primary antibody (3007, 1:50; Cell Signaling Technology), Total p38 primary antibody (9212, 1:100; Cell Signaling Technology), and Biotinylated Donkey anti-Rabbit IgG (H+L) cross-adsorbed secondary antibody (cat# 31821, 1:500, 2.2 μg/mL; Thermofisher).

As previously described (32), patient-derived samples from primary breast cancer were collected from patients without detectable bone metastases at diagnosis, and matching bone metastases were collected at a later date (at least 6 months after initial diagnosis). Patient samples were obtained in accordance with the guidelines established by the Washington University's Institutional Review Board (IRB #201102394; waiver of consent under this IRB#) and WAIVER of Elements of Consent as per 45 CFR 46.116 (d). All patient information was deidentified prior to investigator use. All of the human research activities and all activities of the IRBs designated in the Washington University Federal Wide Assurance, regardless of sponsorship, are guided by the ethical principles in "The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects Research of the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research."

All breast cancer and matched bone metastatic samples displayed tumor cells, as determined by analysis of serial tissue microarray (TMA) sections stained for H&E, E-cadherin, and pan-cytokeratin along with either IL6 or pMK2. Semiquantitative analysis of the stained TMAs was performed using the histoscore (H-score) system, which is a measure of extent and intensity of expression. Each sample was assigned a staining intensity on a scale of 0 to 3 along with the percentage of cells at that intensity level. The H-score was calculated as follows: H = [1 × (% cells 1+) + 2 × (% cells 2+) + 3 × (% cells 3+)].

Virus production was carried out as described previously (31). Briefly, HEK239T cells were transfected with Trans-IT LT1 (Mirus), and virus was collected 48 hours later. Infections were carried out in the presence of 1 μg/mL protamine sulfate. Forty-eight hours after infection, cells were selected with 1 μg/mL puromycin. Short hairpin RNA sequences targeting murine MK2 Mapkapk2: NM_008551 (5′-AGAAAGAGAAGCATTCCGAAAT-3′; 5′-CCGGGCATGAAGACTCGTATT-3′ or 5′-CCAGAGAATGACCATCACAGA-3′), p38MAPKa, and control RFP were obtained from the Children's Discovery Institute's viral vector-based RNAi core at Washington University in St. Louis, and were supplied in the pLKO.1-puro backbone.

The p38MAPK small-molecule inhibitor CDD111 (Aclaris Therapeutics, Inc.) was compounded as described previously (7). The p38MAPK/MK2 small-molecule pathway inhibitor ATI-450 was compounded at 1,000 ppm. Female B6 (Cg)-Tyr^c-2J/J (B6-albino) and FVB/NJ mice were fed ad libitum. Mice were randomized onto inhibitor-containing or regular chow, 24 hours after tumor cell injection.

All statistical analyses were carried out using Graphpad Prism. Numerical data are expressed as mean ± SEM. Mouse analyses were performed by the Student t test or one-way ANOVA as indicated in the figure legends. The Kaplan–Meier method was used to estimate empirical survival probability by treatment, and Kaplan–Meier curves were generated for visualization. The survival difference among/between treatment groups was compared by the log-rank test. HRs between two treatments were estimated from Cox proportional hazard model with 95% confidence interval. FDR-adjusted log-rank test P values were derived to adjust for multiple pairwise comparisons.

In previous work, we coinjected tumor cells and activated fibroblasts expressing p38MAPK-dependent factors and showed that p38MAPK inhibition of stromal-secreted tumorigenic factors reduced subcutaneous tumor growth in immunocompromised mice (7). Further, we showed that many p38MAPK-dependent factors were expressed in the stromal compartment of primary breast cancer lesions (7). Because the stromal compartment is known to support metastatic growth (2), these findings raised the possibility that strategies that target stromal p38MAPK in the metastatic setting might similarly limit tumor growth. To evaluate the potential clinical significance of targeting stromal p38MAPK in the metastatic setting, we first examined tumor epithelial and stromal expression of IL6, which is regulated by p38MAPK, in bones of patients harboring metastatic lesions and compared this with expression in the primary tumors of the same patients. To carry out this analysis, we constructed a tissue microarray consisting of a panel of 58 human breast cancer cases and 38 patient-matched bone metastatic biopsies and then stained for IL6, E-cadherin, pan-cytokeratin, and phosphorylated MK2 (pMK2) in serial sections by immunohistochemistry (IHC). Semiquantitative analysis of the IHC staining revealed higher expression of IL6 in the stromal compartment within the primary and metastatic site relative to the tumor epithelial compartment (Fig. 1A–C), which was identified by expression of pan-cytokeratin and E-cadherin (Fig. 1A and B). Of note, the IL6 expression in both the primary and metastatic stroma was coincident with the presence of activated (i.e., phosphorylated) MK2 (Fig. 1B), a downstream target of p38MAPK that is responsible for stabilizing the mRNAs of many proteins including IL6 (33, 34). Further, robust IL6 expression in the stroma was observed across all molecular subtypes of breast cancer samples including triple-negative, luminal A, luminal B, and Her2⁺ (Fig. 1D). Together, these data suggest that therapeutically targeting the stromal compartment within metastatic lesions with p38MAPK or MK2 inhibitors might reduce stromal-derived tumor-promoting factors including IL6 and metastatic tumor growth.

IL6 is a pleiotropic cytokine with a predominantly protumorigenic role in the context of breast cancer and associated bone metastases (35, 36). However, there is some evidence, albeit incompletely understood, that IL6 transsignaling may mobilize T-cell responses and therefore display antitumorigenic properties (37, 38). Given the potential dual faces of IL6 in the TME, we wanted to identify other p38MAPK-dependent factors in the stroma of patients and investigate the putative role that these factors play in breast cancer metastasis. We used gene set variation analysis (GSVA) to examine gene signatures associated with p38MAPK-dependent factors on 53 human breast cancer samples spanning all molecular subtypes (39) and observed that not only IL6 (Fig. 1D) but also other p38-dependent tumor-promoting factors were more highly expressed in the stroma relative to the tumor epithelial compartment. Furthermore, when stromal-specific gene signatures identified in the studies of Finak and colleagues (40), Ma and colleagues (41), and Karnoub and colleagues (5) were overlaid with our p38MAPK-dependent gene signature, we found a significant number of p38MAPK-dependent factors were enriched. We identified these as the Finak/Ma/Karnoub overlap (Fig. 1E and F). These datasets were generated by microarray comparison of normal and cancer-associated stroma from human breast tissue following laser capture microdissection (LCM; refs. 5, 40, 41). As predicted, the three overlap gene signatures were highly expressed in the stroma relative to epithelium. Together with the IHC data, these results demonstrate that numerous p38MAPK-dependent factors are expressed in the stromal compartment of primary and bone metastatic lesions. This preferential expression suggests that p38MAPK plays an important tumor-promoting role in the stroma not only in the primary setting but also in the metastatic setting.

Expression of p38MAPK-dependent factors in human primary breast lesions and corresponding metastatic lesions, coupled with our previous findings that p38MAPK inhibition can reduce the growth-promoting activities of stromal cells in a primary site (7), led us to investigate whether inhibiting the p38MAPK pathway in the metastatic setting would also limit tumor growth. Further, because the bone is the predominant site of metastasis in breast cancer, we delivered a bone-tropic, murine breast cancer cell line, PyMT-Bo1 (27), into immunocompetent C57BL/6 mice by IC injection (Fig. 2A). The IC injection model synchronously delivers tumor cells to the bones and visceral organs of animals. One day after tumor inoculation, mice were randomized into a control or treatment group. A highly selective p38MAPKα inhibitor (p38i; also known as CDD111; refs. 7, 42), compounded into mouse chow, was administered ad libitum to mice in the treatment group. To evaluate the efficacy of single-agent p38i as compared with standard chemotherapeutic approaches, we administered PTX (10 mg/kg) at 3-day interval via retro-orbital injections, either alone or in combination with p38i. As expected, PTX treatment reduced tumor burden in the bone by 4-fold as measured by BLI on day 13 post-IC. Strikingly, p38i as a single agent reduced bone metastases (Fig. 2B) to the same extent as PTX alone. Histologic evaluation of bone metastases within femurs supported the BLI results (Fig. 2C). Furthermore, p38i's antimetastatic effect was not confined to the bone and resulted in systemic reduction of visceral metastases. Indeed, p38i reduced visceral metastases (nonbone; including lung, liver, and spleen) by 3-fold (Fig. 2D), similar to that obtained with PTX alone (4-fold). We failed to observe any synergistic effect of p38i and PTX presumably because each as a single agent dramatically diminished tumor cell growth in vivo.

To ensure that the effect of p38i on metastases was not limited to one cancer cell line, we carried out the above experiment using the cell line Met-1, originally isolated from an MMTV-PyMT primary mammary tumor in FVB/NJ mice (43). Following IC injection of Met-1 cells, mice were randomized onto p38i or control groups, and metastatic burdens were measured on day 12. Similar to PyMT-Bo1 cells, p38i reduced bone metastasis and visceral metastasis by 9-fold and 13-fold, respectively (Fig. 2E and F). Together, these findings indicate that p38i significantly reduces metastatic growth in the bone and visceral organs.

Above we demonstrated that p38i drastically reduces the metastatic growth of luminal B breast cancer cells. Previously in a subcutaneous xenograft setting, we demonstrated that p38i reduced tumor growth in a cell-nonautonomous fashion by inhibiting secretion of protumorigenic factors from stromal cells (7). To determine whether our treatment was attenuating tumor growth by directly targeting the proliferative ability of tumor cells, we treated luciferase-expressing PyMT-Bo1 tumor cells for 3 days with p38i (1 μmol/L) or PTX (25 nmol/L) in vitro and assessed tumor cell growth by BLI. Our tumor cells express luciferase, and we have previously demonstrated that relative luciferase expression is a reliable surrogate for cell number (31). As expected, PTX treatment significantly reduced the growth of PyMT-Bo1 cells by 2.5-fold. In contrast, p38i as a single agent had no impact on the growth of PyMT-Bo1 tumor cells (Fig. 2G). Similarly, the growth of the Met-1 tumor cells was also unaffected by treatment with p38i (Supplementary Fig. S1). These data demonstrated that the growth of our luminal B breast cancer cell lines is not directly sensitive to p38i.

To determine if p38MAPK is required within tumor cells for metastatic growth, we transduced the PyMT-Bo1 tumor line with a p38MAPKα-specific shRNA. p38MAPKα shRNA expression led to a significant reduction in p38MAPKα protein levels, as observed by Western blot analysis (Supplementary Fig. S2A), and had no impact on the in vitro growth of the cells (Supplementary Fig. S2B). In addition, there was no difference in the ability of these cells to form tumors in the mammary gland (Supplementary Fig. S2C). We introduced these p38MAPKα-depleted cells into mice via IC injection to examine the ability of these cells to grow in metastatic sites. Upon IC delivery of control (shRFP) or sh-p38MAPKα–expressing PyMT-Bo1 cells into C57BL/6 mice, we measured tumor growth in the overall body and bones 13 days after IC injection. We found that tumor burdens were similar in mice injected with sh-p38MAPKα versus shRFP-expressing control cells (Fig. 2H). Histologic evaluation of bone lesions in femurs from shRFP- and sh-p38 tumor-bearing mice confirmed the BLI results (Fig. 2I). To establish whether tumor cell expression of p38MAPK was requisite for metastatic growth, we stained lesions for p38MAPK. We found that p38MAPK expression within tumor cells was relatively heterogeneous, and there were lesions that lacked p38MAPK staining (Fig. 2J). Together, these data support the hypothesis that p38i primarily targets the stromal compartment to reduce metastatic growth. This finding was not limited to the bone because we also failed to observe a reduction in visceral metastasis in animals injected with sh-p38MAPKα–expressing PyMT-Bo1 cells (Fig. 2K). In fact upon analysis of the visceral metastasis, we found that tumor growth was increased in animals bearing sh-p38MAPKα cells. Although the reason for this is unclear, there are reports that p38MAPK plays an active role in tumor cell dormancy (44, 45), raising the possibility that in vivo the reduction of p38MAPK within breast tumor cells increases tumor cell growth. Together, these findings indicate that our p38i strategy does not directly target tumor cells, but rather it is the stromal compartment that is the target of p38i.

p38MAPK targets a large number of downstream factors, and more recent work suggests that it plays a role in maintaining tumor cell dormancy in some models (44, 45). Further, the clinical trials using p38MAPK inhibitors in chronic inflammatory diseases have had mixed results in regards to the durability of the treatment, and the efficacy of p38i treatment in patients with breast cancer remains unknown. For these reasons, we next asked if we could target the p38MAPK-MK2 axis to limit breast cancer metastasis, because MK2 is downstream of p38MAPK and stabilizes protumorigenic cytokine mRNAs including IL6 (46). For this purpose, we used a recently discovered p38MAPK-MK2 inhibitor, ATI-450 (MK2Pi; ref. 30). To first establish whether MK2Pi would target tumor cells, we treated PyMT-Bo1 cells in vitro with MK2Pi and measured their growth relative to vehicle control cells via BLI. PyMT-Bo1 cells were grown for 3 days in the presence of vehicle or MK2Pi (100 nmol/L), and growth was measured by BLI. Upon treatment with MK2Pi, tumor cell growth did not decrease compared with vehicle, indicating that MK2 inhibition alone does not directly affect tumor cell growth (Fig. 3A). To demonstrate this in vivo, we used shRNA to deplete MK2 in PyMT-Bo1 cells. Knockdown was confirmed by Western blot revealing a 93% reduction in MK2 protein levels in PyMT-Bo1-shMK2 cells relative to control cells (shRFP; Supplementary Fig. S2D). In addition, there was no difference in proliferative ability between control (shRFP) cells and shMK2-expressing PyMT-Bo1 cells in vitro (Supplementary Fig. S2B) nor was there a difference in the ability of these cells to form tumors in the mammary gland (Supplementary Fig. S2C). To examine the impact of MK2 depletion on tumor cell growth in vivo, control or shMK2 tumor cells were delivered into mice by IC injection, and metastatic tumor burden was measured on day 13 after IC injection. Similar to what we found with PyMT-Bo1 cells expressing sh-p38MAPKα, the shMK2 tumor cell–bearing mice developed bone and visceral metastases comparable with those injected with control PyMT-Bo1 cells (Fig. 3B and C), and analysis of lesions from mice injected with shMK2 tumor cells revealed lesions that lacked MK2 staining (Supplementary Fig. S2E), indicating that metastatic lesions can grow when tumor cells have significantly reduced levels of MK2. These data demonstrate that inhibiting MK2 in the tumor cells has no effect on their metastatic potential.

We next asked if MK2Pi could limit metastasis. PyMT-Bo1 tumor cells were injected IC, and 24 hours later, mice were randomized into either an MK2Pi or control treatment group. The drug was administered ad libitum for 12 days (Fig. 3D). We found that MK2Pi significantly reduced metastases in the bone (5-fold) and visceral organs (2.6-fold) compared with mice receiving control chow (Fig. 3E and F). Histology of tumor-bearing femurs confirmed reduction of metastases seen via BLI (Fig. 3G). In addition, mice injected with an alternate tumor cell line, Met-1, showed similar reduction in both bone and visceral metastatic outgrowth when treated with MK2Pi (Fig. 3H and I). Taken together, these results uncover a cell-nonautonomous action of MK2 inhibitor in limiting overall metastases.

To assess the impact of p38i versus MK2Pi on overall survival, PyMT-Bo-1 cells were delivered to mice by IC injection, and 24 hours later, mice were enrolled into a single- or dual-arm treatment strategy, and overall survival was assessed. As shown in Fig. 4, PTX, p38i, and MK2Pi significantly extended survival compared with animals receiving vehicle alone. When p38i was combined with PTX, we failed to observe a combinatorial effect. In contrast, the combination of MK2Pi and PTX significantly extended survival compared with the single-arm treatments (Fig. 4 and Supplementary Fig. S3). The reason for this extension is not clear, but we did find that MK2Pi provided enhanced bone protection relative to p38i treatment (Fig. 5, below), thereby preventing paralysis of hind limb and allowing the mice to remain active and mobile for longer. Alternatively, MK2Pi may be more effective at limiting the p38MAPK-MK2 pathway. Indeed, we found that Hsp27 phosphorylation, which plays a role in stabilizing numerous protumorigenic cytokines including IL6 (47), was lower in the lungs of tumor bearing mice treated with MK2Pi relative to p38i (Supplementary Fig. S4A and S4B).

Bone metastasis often leads to increased osteoclastogenesis leading to osteolytic-driven bone destruction (21), and chemotherapy is known to exacerbate bone loss in metastatic patients. The p38MAPK-MK2 pathway plays an important role in bone homeostasis, particularly RANKL-induced osteoclast differentiation (25). Thus, it is not surprising that MK2-deficient mice have increased trabecular and cortical bone mass and decreased osteoclast number and function (26). Based on these reports, we tested the effects of p38i or MK2Pi on in vitro RANKL-induced osteoclast differentiation of bone marrow–derived macrophages. We found that both p38i and MK2Pi inhibited osteoclast differentiation in a concentration-dependent manner (Fig. 5A). We also found that these inhibitors decreased osteoclast bone-resorbing activity in vitro (Fig. 5B). Because p38MAPK has been implicated in osteoblast function (48), we also examined the impact of our drugs on the ability of osteoblast to differentiate. In contrast to what we observed with osteoclasts, these inhibitors had no effect on osteoblastogenesis as measured by alkaline phosphatase activity (Supplementary Fig. S5).

The well-established role of the p38MAPK-MK2 pathway in osteoclastogenesis, coupled with our finding that the inhibitors limited osteoclastogenesis (Fig. 5A and B), led us to ask if p38i and MK2Pi could attenuate the devastating bone loss observed in the metastatic setting. To test this, we delivered PyMT-Bo1 cells by IC injection and, 24 hours later, randomized the mice into the following groups: Vehicle, PTX, p38i, MK2Pi, and zoledronic acid (ZOL). ZOL is a widely used bisphosphonate that can effectively limit bone loss by inhibiting osteoclast activity. To compare the activities of p38i and MK2Pi to ZOL, we administered two doses of 0.75 μg ZOL (or vehicle) subcutaneously, once each week (Fig. 5C). On day 13, tumor burden in the bones was measured by BLI. Following that, bones were processed for density analysis using microcomputed tomography (μCT). We measured trabecular bone volume of tumor-bearing mice in all groups except vehicle because the femurs of vehicle-treated mice had large, invasive tumors that destroyed nearly all measureable bones, making them unsuitable for μCT analysis. Instead, the bone volume of the other four groups was compared with femurs from non–tumor-bearing mice. We observed that although ZOL effectively limited tumor-induced bone loss (Fig. 5D), it did not impede tumor growth in the bone (Supplementary Fig. S6). Given that chemotherapy can induce bone loss in mice and patients, it was not surprising to find that PTX exacerbated bone loss by 2.5-fold relative to untreated, non–tumor-bearing mice. Importantly, treatment with either p38i or MK2Pi reduced tumor burdens (Fig. 2B–E, respectively) and preserved bone density to the same extent as ZOL in tumor-bearing mouse bones (Fig. 5D). Three-dimensional reconstructions of tumor-bearing femurs from each of the groups corroborated the μCT results (Fig. 5E). Together, these findings demonstrate that p38i and MK2Pi provide a dual benefit, in that they not only attenuate disease progression by limiting stromal support of tumor growth but they also effectively protect against bone loss likely by inhibiting osteoclastogenesis, even in the face of chemotherapy, making them attractive, stromal-targeted therapies to pursue.

Several studies have established that the stroma plays a significant role in tumor progression, thereby establishing a rationale for developing stroma-targeted antitumor therapies. In this study, using a preclinical model of luminal B–like breast cancer, we demonstrated that inhibiting the p38MAPK-MK2 pathway limits visceral and bone metastases. Importantly, we show that depleting p38MAPK or MK2 in the tumor cells had no effect on metastatic outgrowth, providing evidence that the inhibitor's target is indeed the stroma and not the tumor cells directly. This is in contrast to chemotherapy, which directly targets tumor cells. Indeed, PTX, the chemotherapeutic agent used in this study, limited metastasis as effectively as the tested p38MAPK and MK2 inhibitors. Despite PTX's ability to limit tumor growth, it failed to provide any survival advantage to our mice, underscoring its overall toxicity and the effectiveness of the MK2Pi, which not only reduced tumor burden but also significantly extended survival. Given that chemotherapy directly targets tumor cells, which tend to be genetically malleable to imposed selective pressures, leading to drug resistance (49), our findings suggest that stromal-targeted therapies might provide a more durable response in patients. Therefore, targeting stromal cells could help circumvent the challenge of drug resistance. In addition, stromal status and composition of distal organs is implicated in determining the fate of disseminated tumor cells. Stromal-targeted therapies can be used to block the metastatic cascade at an early stage by impeding the development of fertile niches where tumor cells tend to thrive and eventually outgrow into macrometastases. In this way, stromal therapy has the potential to synergize with tumor-targeted therapies to ensure more effective and widespread killing of tumor cells.

Although many studies favor the development and use of p38 inhibitors in treating cancers, some reports provide contradictory evidence suggesting that blocking p38 may confer a growth advantage to tumor cells (50, 51). These divergent results should be evaluated in light of the fact that p38MAPK signaling is different across cell types and tumor types, thereby making it challenging to generalize findings. Our work demonstrates that limiting p38MAPK within tumor cells without simultaneously limiting it in the stromal compartment can increase metastatic tumor growth (Fig. 2K). Indeed, in contrast to the limited metastatic growth we observe upon p38i, we find that visceral metastases increase in mice injected with p38MAPKα-depleted tumor cells. This result may not be surprising given recent evidence demonstrating that inhibition of p38MAPKα within tumor cells can increase their invasiveness (50). In addition, work from Guiso and colleagues suggests that active p38MAPKα keeps tumor cells in a dormant state by phosphorylating a number of factors including ATF2 (52) that are not substrates for MK2. In light of this finding, inhibition of p38MAPK could be seen as deleterious as it may “awaken” cells out of dormancy that may have otherwise continued in an indolent state. If true, the use of our MK2Pi may be a better approach in patients with minimal residual disease rather than those with active metastatic lesions. Another potential advantage of p38i/MK2Pi is that if they were to drive nondividing tumor cells—be it dormant or otherwise—into the cell cycle, they may increase the killing potential of chemotherapies that rely on cell cycling. This is an important area that will require further investigation.

Reducing metastatic tumor burden is the goal of all cancer therapies. However, the devastating side effects of many of the therapies used negatively affect a patient's quality of life. In patients with breast cancer, the risk for skeletal-related events—pathologic fractures, hypercalcemia, and bone pain—due to tumor-induced as well as therapy-induced osteolysis (21) remains a significant problem. For this reason, bone-preserving therapies such as the bisphosphonate ZOL are now standard of care in the metastatic setting (53). Although ZOL effectively protects bone quality, there is conflicting evidence about its efficacy at limiting tumor growth in models of bone metastases, and in rare instances, it can result in significant toxicity. Studies suggest that the antitumor effects of ZOL depend on the size of bone lesions and whether the treatment is preventive (more effective) versus therapeutic (54, 55). Given the severity of the skeletal complications observed in many patients, there is a clear need for new breast cancer therapies that combat not only tumor growth but also the associated comorbidities. Strikingly, we show that blocking the p38MAPK-MK2 pathway with either inhibitor (p38i or MK2Pi) limited osteoclastogenesis and had a significant protective effect on the bone. The dual action of p38i and MK2Pi makes them promising candidates to pursue for clinical trials. However, given we found that sh-p38MAPK led to increased metastatic burden in the visceral organs and the fact that MK2Pi combined with PTX extended survival, our data suggest that MK2 may be a more viable target. Finally, given the potent inhibition of metastases observed with the MK2 inhibitor throughout the mouse, further studies are warranted to investigate the specific stromal cell types targeted by the drug beyond osteoclasts to gain a mechanistic understanding of its action that will help shed light on where and how best to employ it in patients with breast cancer.

R.M. Johnson is Associate Scientist at Genentech. B. Burnette is Senior Principal Scientist at Confluence Discovery Technologies. G. Mbalaviele is a consultant at and has ownership interest (including stock, patents, etc.) in Aclaris Therapeutics, Inc. J.B. Monahan has ownership interest (including stock, patents, etc.) in Aclaris Therapeutics, Inc. No potential conflicts of interest were disclosed by the other authors.

This publication is solely the responsibility of the authors and does not necessarily represent the official view of the National Center for Research Resources (NCRR) or NIH. Opinion, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the Department of Defense.

Conception and design: B. Murali, Q. Ren, K.C. Flanagan, J.B. Monahan, S.A. Stewart

Development of methodology: B. Murali, Q. Ren, X. Luo, D.V. Faget, E. Alspach, X. Su, K.N. Weilbaecher, G. Mbalaviele

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B. Murali, Q. Ren, D.V. Faget, C. Wang, K.C. Flanagan, Y. Fu, K. Leahy, X. Su, M.H. Ross, B. Burnette, K.N. Weilbaecher, G. Mbalaviele, S.A. Stewart

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B. Murali, Q. Ren, D.V. Faget, R.M. Johnson, T. Gruosso, K.C. Flanagan, M.H. Ross, B. Burnette, M. Park, S.A. Stewart

Writing, review, and/or revision of the manuscript: B. Murali, Q. Ren, D.V. Faget, R.M. Johnson, Y. Fu, K. Leahy, X. Su, K.N. Weilbaecher, M. Park, J.B. Monahan, S.A. Stewart

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Murali, K. Leahy, M. Park, J.B. Monahan, S.A. Stewart

Study supervision: G. Mbalaviele, S.A. Stewart

We thank Deborah (Novack) Veis, Joshua Rubin, Daniel Link, Roberta Faccio, and David DeNardo for their valuable suggestions. We thank Lynne Marsala, Julie Prior, and the ICCE Institute at Washington University School of Medicine for live cell and live animal imaging. We thank Crystal Idleburg and Samantha Coleman at the Musculoskeletal Histology core for their expert technical assistance with bone tissue sectioning and staining and Deborah Veis, Thomas Walsh, and Graham Colditz for assistance in constructing the human TMA. In addition, we thank Daniel Leib and the Structure and Strength Musculoskeletal core for μCT imaging. shRNA constructs were obtained from the Children's Discovery Institute's viral vector-based RNAi core at Washington University in St. Louis. We thank the Genome Technology Access Center in the Department of Genetics at Washington University School of Medicine for help with genomic analysis. Finally, we thank Lorry Blath and Judy Johnson for their constant support, enthusiasm, and critical assessment of our work and its impact on breast cancer patients. The Center is partially supported by NCI Cancer Center Support Grant #P30 CA91842 to the Siteman Cancer Center and by ICTS/CTSA Grant# UL1TR000448 from the NCRR, a component of the NIH, and NIH Roadmap for Medical Research.

This work was supported by the Cancer Biology Pathway Molecular Oncology Training Grant NIH T32CA113275 (B. Murali), NIH grants NIH 5 R01 CA130919 (S.A. Stewart), NIH Cellular Biochemical and Molecular Sciences Pre-doctoral Training Grant T32 GM007067 (K.C. Flanagan and E. Alspach), NIH F31 CA189669 (K.C. Flanagan), American Cancer Society Research Scholar Award (S.A. Stewart), CA100730 (K.N. Weilbaecher), CA097250 (K.N. Weilbaecher), and training grants 5T32GM007067-39 (M.H. Ross) and T32AR060719 (M.H. Ross). The work was supported in part by the Siteman Investment Program [supported by The Foundation for Barnes-Jewish Hospital Cancer Frontier Fund (FBJH CFF 3773); Barnard Trust; Washington University Musculoskeletal Research Center (NIH P30 AR057235); Fashion Footwear Charitable Foundation of New York, Inc.; and the National Cancer Institute Cancer Center Support Grant P30CA091842, Eberlein, PI; S.A. Stewart] and the St. Louis Breast Tissue Registry (funded by The Department of Surgery at Washington University School of Medicine). T. Gruosso has been supported by the Charlotte and Leo Karassik Foundation oncology postdoctoral fellowship. The study involving LCM followed by gene expression was supported by grants to M. Park from the Québec Breast Cancer Foundation, Genome Canada–Génome Québec, NIH, and CIHR (Canadian Institutes of Health Research). Research supported by SU2C Canada-Canadian Cancer Society Breast Cancer Dream Team Research Funding (SU2C-AACR-DT-18-15), with supplemental support from the Ontario Institute for Cancer Research, through funding provided by the Government of Ontario. Stand Up To Cancer Canada is a Canadian Registered Charity (Reg. # 80550 6730 RR0001). Research Funding is administered by the American Association for Cancer Research International-Canada, the Scientific Partner of SU2C Canada (to M. Park). The breast tissue and data bank at McGill University is supported by funding from the Database and Tissue Bank Axis of the Réseau de Recherche en Cancer of the Fonds de Recherche du Québec-Santé and the Quebec Breast Cancer Foundation (to M. Park). G. Mbalaviele is supported by NIH/NIAMS AR064755 and AR068972 grants. Luminescent imaging was supported by NIH P50 CA094056. Imaging and analysis of human breast cancer and bone biopsy slides were performed using Zeiss Axio ScanZ.1 through the use of Washington University Center for Cellular Imaging supported by Washington University School of Medicine, The Children's Discovery Institute of Washington University and St. Louis Children's Hospital (CDI-CORE-2015-505), and the National Institute for Neurological Disorders and Stroke (NS086741). Finally, the U.S. Army Medical Research Acquisition Activity, 820 Chandler Street, Fort Detrick MD 21702-5014, is the awarding and administrating acquisition office, and this was supported in part by the Office of the Assistant Secretary of Defense for Health Affairs, through the Breast Cancer Research Program, under award No. W81XWH-16-1-0728.

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.