Early detection and adjuvant therapies have significantly improved survival of patients with breast cancer over the past three decades. In contrast, management of metastatic disease remains unresolved. Brain metastasis is a late complication frequently observed among patients with metastatic breast cancer, whose poor prognosis calls for novel and more effective therapies. Here, we report that active hypoxia inducible factor-1 (HIF1) signaling and loss of the miRNA let-7d concur to promote brain metastasis in a recently established model of spontaneous breast cancer metastasis from the primary site to the brain (4T1-BM2), and additionally in murine and human experimental models of breast cancer brain metastasis (D2A1-BM2 and MDA231-BrM2). Active HIF1 and let-7d loss upregulated expression of platelet-derived growth factor (PDGF) B/A in murine and human brain metastatic cells, respectively, while either individual silencing of HIF1α and PDGF-A/B or let-7d overexpression suppressed brain metastasis formation in the tested models. Let-7d silencing upregulated HIF1α expression and HIF1 activity, indicating a regulatory hierarchy of the system. The clinical relevance of the identified targets was supported by human gene expression data analyses. Treatment of mice with nilotinib, a kinase inhibitor impinging on PDGF receptor (PDGFR) signaling, prevented formation of spontaneous brain metastases in the 4T1-BM2 model and reduced growth of established brain metastases in mouse and human models. These results identify active HIF1 signaling and let-7d loss as coordinated events promoting breast cancer brain metastasis through increased expression of PDGF-A/B. Moreover, they identify PDGFR inhibition as a potentially actionable therapeutic strategy for patients with brain metastatis.

Significance:

These findings show that loss of miRNA let-7d and active HIF1 signaling promotes breast cancer brain metastasis via PDGF and that pharmacologic inhibition of PDGFR suppresses brain metastasis, suggesting novel therapeutic opportunities.

Brain metastasis occurs in about one-quarter of patients with breast cancer, often as a late complication of an already metastatic disease (1). The number of patients developing brain metastases is on the rise because of the extended survival achieved by the improvements in adjuvant therapies and better control of metastatic disease affecting other distant organs (1, 2). Current approaches to treat brain metastases, including whole-brain radiotherapy, stereotactic radiosurgery, chemotherapy, targeted therapies, such as HER2 inhibitors in HER2-amplified tumors, and combinations thereof, show limited benefits as mirrored by the short survival time upon diagnosis (3). This poor prognosis highlights the outstanding clinical need to identify valuable therapeutic targets to effectively halt metastatic dissemination to the brain (3).

Over the past decade, there have been renewed efforts to unravel cellular and molecular mechanisms responsible of brain metastasis (4). Recent studies have provided important mechanistic insights and identified a number of effector molecules and candidate therapeutic targets, including heparin binding EGF-like growth factor HBEGF, cyclooxygenase 2, (5), plasmin (6), connexin 43/protocadherin 7 (6, 7), and N-methyl-D-aspartate receptors (8).

The preclinical models currently available to study brain metastasis are largely based on the direct intracardiac or intracarotid injection of tumor cells and are referred to as experimental models. These models bypass primary tumor development and, therefore, do not fully recapitulate the steps of the metastatic cascade as they occur in patients (8–11). To circumvent this essential limitation, we developed a novel preclinical model of spontaneous breast cancer metastasis to the brain based on basal-like 4T1 cells selected for brain tropism (4T1-BM2) and orthotopically implanted into the mammary gland of immunocompetent syngeneic BALB/c mice. By taking advantage of this model, we demonstrated that the critical events responsible for the brain metastatic capacity of 4T1-BM2 cells take place during the brain metastatic colonization step and unraveled a novel mechanism promoting cancer cell nesting, survival, and metastatic outgrowth in the brain involving Cx31/43-dependent focal adhesion kinase (FAK) activation and laminin 5/8 deposition (Lorusso and colleagues; unpublished data).

By testing this novel spontaneous model, we set-up to identify key transcriptional and posttranscriptional regulators of brain metastasis development. Here, we report gain of hypoxia inducible factor-1 (HIF1) activity and loss of miRNA let-7d in brain metastatic cells, which concur to mediate brain metastasis through transcriptional and posttranscriptional regulations of platelet-derived growth factor (Pdgf) b expression in 4T1-BM2 breast cancer cells. These results are corroborated by experimental evidences obtained from additional murine (D2A1-BM2) and human (MDA213-BrM2) preclinical experimental models of brain metastasis and from the analyses of human transcriptomic datasets and pathways. In vivo treatment with the small molecular PDGF receptor (PDGFR) kinase inhibitor, nilotinib, halted brain metastasis progression, pointing to its translational relevance in the clinical setting.

In vivo experiments

All animal procedures were performed in accordance with the Swiss legislations on animal experimentation and approved by the Cantonal Veterinary Service of the Cantons Vaud and Fribourg. For pharmacological treatment, tumor-bearing mice were randomly selected. In all other experiments, mice were assigned to defined groups at the start of the experiments. Size group was determined in accordance with the Cantonal Veterinary Service to detect difference of 40% or more with SD of 30% relative to the mean for an alpha of 0.05 and a beta of 0.8. The total number of animals used in each experiment and in each group is given in the figure legends.

Orthotopic breast tumor injection, tumor monitoring, and metastasis quantification

4T1 cells were injected (5 × 104 cells in 50 μL PBS and 10% Matrigel mixture per mouse) in the fourth right inguinal mammary gland of 5- to 6-week-old (20–22 g) BALB/c female mice (Harlan Laboratories; Charles River Laboratories; and University of Lausanne, Lausanne, Switzerland). Prior to surgery, animals were anesthetized by an intraperitoneal injection of ketamine (1.5 mg/kg) and xylazine (150 μg/kg; both from Graeub). Tumor growth was measured twice a week with caliper and tumor volume was calculated by the formula: volume = π/6(length × width2). At endpoint of the experiment, mice were sacrificed according to defined ethical criteria and tumors, lungs, and brains were resected, tumors weighted, and lungs and brains analyzed by ex vivo bioluminescence imaging (BLI). Tumors and lungs were fixed over-night in 4% paraformaldehyde (PFA) and paraffin-embedded for sectioning. Lung and brain metastases were quantified by ex vivo culture and colony counting as described previously (11). Removed lungs and brains were digested in SFM containing 1% collagenase I and 10 mg/mL DNase I for 45 minutes at 37°C by vigorous shaking. The cell suspension obtained was passed through 100- and 40-μm pore-size nylon mesh strainers and suspended cells were cultured in complete (10% FBS) DMEM. After 24 hours, metastatic colony cultures were washed with PBS, and further grown for 6–7 days in complete (10% FBS) DMEM. Quantification of grown colonies was performed under a microscope (individual, sparse tumor cells, or nontumor cells were not counted). Metastatic index is defined as the number of organ metastases (lung or brain colonies) normalized over primary tumor weight (12).

Intracarotid artery injection

Prior to surgery, 5- to 6-week-old BALB/c or NSG female mice (20–22 g; from Charles River Laboratories and University of Lausanne) were anesthetized by isoflurane inhalation during the procedure. Glucose infusion was administered to mice before and after surgery. The throat skin was carefully opened with a scalpel (1 cm). A polyurethane catheter (0.4 mm) was inserted into the right common carotid artery and CFSE-labeled (25 μmol/L at 37°C for 30 minutes) or luciferase-tagged 4T1, D2A1, or MDA231 tumor cells were slowly injected (4T1 and D2A1 in BALB/c mice and MDA231 in NSG mice) through the catheter in retrograde fashion toward the aortic arch. The artery was ligated both upstream and downstream of the site of injection before the catheter was removed and the wound was closed with a silk suture. After approximately 3, 7, or 10 days after injection, mice were analyzed by in vivo BLI. At endpoint, mice were sacrificed, and brains and lungs were resected and analyzed by ex vivo BLI. Brains were fixed in 4% PFA for 1 hour at 4°C and then embedded in Tissue-Tek OCT (Sakura) for cryo-sectioning.

Treatment of mice with pharmacologic tyrosine kinase inhibitor

Micrometastases were allowed to form for 3 days after injection of tumor cells via the common carotid artery. Tyrosine kinase inhibitor (TKI), nilotinib, was diluted in DMSO. A vehicle was prepared for the oral administration containing 0.5% hydroxypropyl methylcellulose and 0.05% Tween80. Nilotinib (30 mg/kg/mouse) or a vehicle control was administered daily per os (by gavage) from day 10 till endpoint for the orthotopic spontaneous 4T1-BM2 model, from day 3 till endpoint for the intracarotid 4T1-BM2 model, and from day 4 till endpoint for the intracarotid MDA231-BrM2 model. Tumor measurement was performed starting at day 5 for the orthotopic model, while for the intracarotid models, in vivo BLI was performed every 2–3 days, till endpoint when mice were sacrificed for ex vivo BLI and organ processing.

Ex vivo quantification of brain metastatic clones

Primary tumors and the resected organs, lungs and brains, were digested in serum-free medium containing 1% collagenase type I and 10 mg/mL DNase I for 45 minutes at 37°C. The digested organs were stirred with a sterile magnet. The cell suspension was passed through nylon mesh strainers of 100 and 40 μm pore size. After centrifugation, the resuspended pellet was put in culture in complete medium. After 24 hours, cells were washed twice with PBS and were grown to colonies in complete medium for 6–7 days for metastatic clone quantification ex vivo. Correlative analysis of the BLI signals obtained from the brain of different mice ex vivo, and the number of ex vivo colonies derived from the same brains demonstrated a significant linear correlation between these two methods (Lorusso and colleagues; unpublished data).

Breast cancer patient expression data

These data are publicly available and can be obtained from NCBI Gene Expression Omnibus, as well as from the supporting information (Bos, 2009). They consist of four datasets: NKI295, EMC286, EMC192, and MSK82, done on different platforms by different laboratories. For analysis, datasets were merged.

Survival analysis for the selected genes

For the PDGFB Kaplan–Meier plot, a cutoff was chosen according to the Cox model. Because the expression values are not necessarily comparable, they were converted into a normalized rank for each gene in each dataset. These values ranged between 0 and 1.

Integrated System for Motif Activity Response Analysis

The automated web-based tool called Integrated System for Motif Activity Response Analysis (ISMARA) was used to model gene expression in terms of genome-wide predictions of regulatory sites allowing the identification of transcription factors with elevated transcriptional activity (13). For the primary tumor clinical breast cancer patient data regulatory motif activity analysis, the following publicly available datasets were analyzed: NKI295, EMC286, EMC192, and MSK82. In addition, expression data from lung, bone, and brain human breast cancer metastases were analyzed: GSE14017 and GSE14018.

Statistical analyses

Except for the gene expression studies described above, in all other experiments, statistical comparisons were performed by a two-tailed Student t test or by a one-way ANOVA test (Bonferroni post-test), using Prism 6.0 GraphPad Software, Inc. Results were considered to be significant with *, P < 0.05; **, P < 0.005; ***, P < 0.001 values. Results were considered nonsignificant with P > 0.05.

Constitutively increased HIF1 activity promotes breast cancer brain metastasis

To identify key transcriptional and posttranscriptional regulators of brain metastasis, we performed an ISMARA to computationally predict the activity of transcription factors and miRNAs (13). We applied ISMARA to gene expression datasets derived from both murine and human brain metastatic cells, 4T1-BM2 and MDA231-BrM2, their non-brain metastatic counterparts, the 4T1-T2 and 4T1-LM2 lines (derived from primary tumors and lung metastases respectively; Lorusso and colleagues; unpublished data), and the MDA231 parental line (5). ISMARA revealed a constitutively increased transcriptional activity associated with both the HIF1α and HIF1β components of the HIF1 transcription complex in 4T1-BM2 and MDA231-BrM2 cells compared with their non-brain metastatic counterparts (Fig. 1A; Supplementary Fig. S1A). We confirmed the bioinformatics prediction by ISMARA of elevated HIF1 transcriptional activity with an in vitro luciferase-based HIF1 activity reporter assay in 4T1-BM2 and MDA231-BrM2 cells and in a second murine brain metastatic cell line, D2A1-BM2, which we recently generated (Lorusso and colleagues; unpublished data; Fig. 1B; Supplementary Fig. S1B). Consistently, we found elevated mRNA expression of the direct HIF1 targets, carbonic anhydrase 9 (Ca9/CA9; ref. 14), glucose transporter 1 (Glut1/GLUT1; ref. 15), and VEGFA (Vegfa/VEGFA; ref. 16), in all brain metastatic cells relative to their non-brain metastatic counterparts (Fig. 1C; Supplementary Fig. S1C), confirming increased HIF1 transcriptional activity. Consistent with elevated HIF1 activity, 4T1-BM2, D2A1-BM2, and MDA231-BrM2 brain metastatic cells survived better under severe hypoxic conditions (i.e., 48 hours in 0.1% O2) in vitro, compared with their corresponding parental cells (Fig. 1D).

Figure 1.

Constitutive HIF1 transcriptional activity contributes to brain metastasis. A, HIF1α and HIF1β activities in 4T1-BM2 compared with 4T1-T2 and 4T1-LM2 cells as predicted by ISMARA from transcriptomic datasets. B, HIF1 promoter activity detected by a luciferase-based reporter assay is higher in 4T1-BM2 cells compared with 4T1-T2 cells. C,Ca9, Glut1, and Vegfa mRNA expression shown as fold change in 4T1-BM2 versus 4T1-T2 cells. D, Survival of 4T1-BM2 versus 4T1-T2, D2A1-BM2 versus D2A1-T2, and MDA231-BrM2 versus MDA231 parental cells under hypoxic conditions (0.1% O2) at 24 and 48 hours in vitro. E, Fold change of Ca9/CA9, Glut1/GLUT1, and Vegfa/VEGFA mRNA expression in 4T1-BM2/MDA231-BrM2 cells upon HIF1α silencing relative to corresponding NS control cells. F, Decreased survival of 4T1-BM2 cells with stable HIF1α silencing under hypoxic conditions (0.1% O2) in vitro. G, HIF1α silencing (kd#1 and kd#2) in 4T1-BM2 cells suppresses brain metastases generated by intracarotid injection and monitored by BLI in vivo. H, HIF1α silencing (kd#1 and kd#2) in MDA231-BrM2 suppresses brain metastases generated by intracarotid injection and monitored by BLI in vivo. Representative BLI images of the brains are shown. NS, nonsilencing controls. Mice per group n = 9. Values represent mean ± SD (in vitro) or SEM (in vivo). P values were determined by a two-tailed Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

Figure 1.

Constitutive HIF1 transcriptional activity contributes to brain metastasis. A, HIF1α and HIF1β activities in 4T1-BM2 compared with 4T1-T2 and 4T1-LM2 cells as predicted by ISMARA from transcriptomic datasets. B, HIF1 promoter activity detected by a luciferase-based reporter assay is higher in 4T1-BM2 cells compared with 4T1-T2 cells. C,Ca9, Glut1, and Vegfa mRNA expression shown as fold change in 4T1-BM2 versus 4T1-T2 cells. D, Survival of 4T1-BM2 versus 4T1-T2, D2A1-BM2 versus D2A1-T2, and MDA231-BrM2 versus MDA231 parental cells under hypoxic conditions (0.1% O2) at 24 and 48 hours in vitro. E, Fold change of Ca9/CA9, Glut1/GLUT1, and Vegfa/VEGFA mRNA expression in 4T1-BM2/MDA231-BrM2 cells upon HIF1α silencing relative to corresponding NS control cells. F, Decreased survival of 4T1-BM2 cells with stable HIF1α silencing under hypoxic conditions (0.1% O2) in vitro. G, HIF1α silencing (kd#1 and kd#2) in 4T1-BM2 cells suppresses brain metastases generated by intracarotid injection and monitored by BLI in vivo. H, HIF1α silencing (kd#1 and kd#2) in MDA231-BrM2 suppresses brain metastases generated by intracarotid injection and monitored by BLI in vivo. Representative BLI images of the brains are shown. NS, nonsilencing controls. Mice per group n = 9. Values represent mean ± SD (in vitro) or SEM (in vivo). P values were determined by a two-tailed Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

Close modal

To functionally validate the role of active HIF1 in promoting brain metastasis, we stably silenced it in 4T1-BM2, D2A1-BM2, and MDA231-BrM2 cells (Supplementary Fig. S1D). The expressions of the direct transcriptional targets Ca9/CA9, Glut1/GLUT1, and Vegfa/VEGFA were reduced, thus confirming HIF1 decreased transcriptional activity (Fig. 1E; Supplementary Fig. S1E). Consistently, upon HIF1α silencing, the ability of brain metastatic cells to withstand low oxygen concentration (0.1%) significantly diminished in vitro (Fig. 1F; Supplementary Fig. S1F). Next, we tested the effect of HIF1α silencing on brain metastasis formation. HIF1α-silenced 4T1-BM2 and MDA231-BrM2 cells delivered to the brain via intracarotid injection formed significantly less brain metastases compared with the control cells, as demonstrated by in vivo and ex vivo BLI of the brains (Fig. 1G and H).

From these experiments we conclude that, constitutively elevated HIF1 transcriptional activity in breast cancer cells plays an essential role in brain metastasis formation.

Reduced miRNA let-7d activity promotes colonization of the brain

Computational prediction by ISMARA revealed that transcripts containing binding motifs for let-7 miRNA in their 3′ untranslated region (UTR) were highly abundant in 4T1-BM2 and MDA231-BrM2 brain metastatic cells (Fig. 2A; Supplementary Fig. S2A). Let-7 family members are negative regulator of translation (17). Consistent with this in silico prediction, let-7d was the most downregulated let-7 family member from the gene expression analysis, in brain metastatic 4T1-BM2 cells, compared with 4T1-T2 and 4T1-LM2 primary tumor and lung metastasis–derived cells (Lorusso and colleagues; unpublished data). PCR analysis confirmed that let-7d miRNA expression in 4T1-BM2, D2A1-BM2, and MDA231-BrM2 cells was reduced by 30%–40% compared with their non-brain metastatic counterparts (Fig. 2B; Supplementary Fig. S2B). Bioinformatics analysis of breast cancer patient datasets revealed increased expression of miRNA let-7d targets and reduced miRNA let-7d levels in primary tumors of patients that developed brain metastases compared with patients that did not (Fig. 2C). Importantly, increased expression of miRNA let-7d targets and reduced levels of miRNA let-7d were observed in brain metastases compared with metastases to other organs, such as lung/bone (Fig. 2D), consistent with brain-specific effects. Interestingly, expression of Lin28a and Lin28b, two variants of Lin28, a previously reported natural inhibitor of miRNA let-7d (18), was elevated in all brain metastatic cells compared with the parental lines (Fig. 2E; Supplementary Fig. S2C). This provided additional evidence for dysregulated expression of let-7 miRNA in brain metastatic cells.

Figure 2.

Repression of miRNA let-7d in brain metastatic cells promotes brain metastasis formation. A, Let-7 activity in 4T1-BM2 compared with 4T1-T2 and 4T1-LM2 cells as predicted by ISMARA from transcriptomic datasets. B, Relative expression of miRNA let-7d in 4T1-BM2 cells compared with 4T1-T2 cells as assessed by TaqMan qPCR. C, Let-7d activity as predicted by ISMARA in primary tumor samples of patients without (0; 806 patients) or with (1; 49 patients) brain metastases (left) and relative Let-7d expression (transcriptomic datasets, right). D, Let-7d activity in brain (49 patients) versus lung/bone (297 patients) metastatic samples (left) and relative expression of let-7d miRNA (right). E, Relative mRNA expression of Lin28b mRNA in 4T1-BM2 and 4T1-T2 cells. F, Induced expression of miRNA let-7d in 4T1-BM2 cells. G, Brain metastasis derived from 4T1-BM2 cells overexpressing miRNA let-7d (miRlet-7d) or miR000 control after intracarotid artery injection, monitored by ex vivo BLI. H, Brain metastasis derived from MDA231-BrM2 cells overexpressing miRNA let-7d (miRlet-7d) or miR000 control after intracarotid artery injection, monitored by ex vivo BLI. Representative BLI pictures are shown. I, Primary tumor growth and weight (left) and brain and lung metastatic indexes of mice orthotopically injected with 4T1-BM2 cells overexpressing miRNA let-7d (miRlet-7d) or control cells (miR-000). Mice per group was n = 8–10. Values represent mean ± SD (in vitro) or SEM (in vivo). P values were determined by a two-tailed Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

Figure 2.

Repression of miRNA let-7d in brain metastatic cells promotes brain metastasis formation. A, Let-7 activity in 4T1-BM2 compared with 4T1-T2 and 4T1-LM2 cells as predicted by ISMARA from transcriptomic datasets. B, Relative expression of miRNA let-7d in 4T1-BM2 cells compared with 4T1-T2 cells as assessed by TaqMan qPCR. C, Let-7d activity as predicted by ISMARA in primary tumor samples of patients without (0; 806 patients) or with (1; 49 patients) brain metastases (left) and relative Let-7d expression (transcriptomic datasets, right). D, Let-7d activity in brain (49 patients) versus lung/bone (297 patients) metastatic samples (left) and relative expression of let-7d miRNA (right). E, Relative mRNA expression of Lin28b mRNA in 4T1-BM2 and 4T1-T2 cells. F, Induced expression of miRNA let-7d in 4T1-BM2 cells. G, Brain metastasis derived from 4T1-BM2 cells overexpressing miRNA let-7d (miRlet-7d) or miR000 control after intracarotid artery injection, monitored by ex vivo BLI. H, Brain metastasis derived from MDA231-BrM2 cells overexpressing miRNA let-7d (miRlet-7d) or miR000 control after intracarotid artery injection, monitored by ex vivo BLI. Representative BLI pictures are shown. I, Primary tumor growth and weight (left) and brain and lung metastatic indexes of mice orthotopically injected with 4T1-BM2 cells overexpressing miRNA let-7d (miRlet-7d) or control cells (miR-000). Mice per group was n = 8–10. Values represent mean ± SD (in vitro) or SEM (in vivo). P values were determined by a two-tailed Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.005.

Close modal

To assess the functional role of let-7d miRNA loss in brain metastasis formation, we stably overexpressed let-7d in 4T1-BM2 and MDA231-BrM2 cells (Fig. 2F; Supplementary Fig. S2D). Induction of miRNA let-7d had no effect on cell proliferation in vitro (Supplementary Fig. S2E), while it significantly reduced brain metastatic capacities of 4T1-BM2 and MDA231-BrM2 cells upon intracarotid injection (Fig. 2G and H). To test the role of let-7d overexpression on brain metastasis formation in a more physiologic and clinically relevant model, we injected miRNA let-7d–overexpressing 4T1-BM2 cells (and miR-000 control cells) in the mammary fad pad and monitored primary breast tumor growth and metastatic dissemination. miRNA let-7d overexpression led to slightly slower primary tumor growth, but it nearly completely abolished (more than 95% inhibition) brain metastasis formation, while only partially reducing (∼40% inhibition) lung metastasis formation (Fig. 2I).

From these results, we conclude that brain metastatic breast cancer cells have reduced miRNA let-7d expression and its loss significantly contributes to their brain metastatic capacity.

Increased HIF1 activity and repressed miRNA let-7d induce Pdgfb/PDGFA expression in brain metastatic cells

By mining human breast cancer patient databases in combination with gene expression datasets obtained from the 4T1-BM2 spontaneous brain metastasis model, we identified 18 genes potentially clinically relevant to human brain metastasis as they were associated with short brain metastasis–free survival (BMFS) in patients with metastatic breast cancer (Lorusso and colleagues; unpublished data). We combined the predicted regulatory matrix obtained from ISMARA for HIF1α and let-7d and applied it to this 18-gene set. Strikingly, Pdgfb was the only target gene whose expression was predicted to be under the control of both HIF1 and let-7d (Fig. 3A).

Figure 3.

HIF1 and miRNA let-7d regulate expression of Pdgfb/PDGFA in brain metastatic cells. A, Tabular view of the 18 coexpressed genes combined with HIF1α/HIF1β and let-7 regulation as predicted by ISMARA. The ranking was performed according to z values provided by the analysis. Only PDGFB gene was predicted to be regulated by both HIF1 and let-7. B, Kaplan–Meier plot shows BMFS of patients with breast cancer stratified for low and high PDGFB expression levels. The cutoff used is the 67th percentile: 570 patients have PDGFB levels below (PDGFB low) and 285 above (PDGFB high) this percentile. C,PDGFB expression levels in primary tumors of patients without (0; 806 patients) or with (1; 49 patients) brain metastases (left) and in brain (49 patients) versus lung/bone metastasis (297 patients) samples from patients with breast cancer (right). D, Relative Pdgfb mRNA and protein expression in 4T1-BM2, 4T1-LM2, and 4T1-T2 cells. E, Relative Pdgfb mRNA and protein expression upon stable silencing of HIF1α in 4T1-BM2 cells (HIF1α_kd#1 and kd#2). F, Relative Pdgfb mRNA levels upon overexpression of miRNA let-7d in 4T1-BM2 cells. G, Representation of the proposed model of regulation of the Pdgfb transcription and translation as predicted by ISMARA. HIF1α binds transcription factor (TF) motif in the promoter region, whereas miRNA let-7d recognizes an 8-mer binding site in the 3′ UTR of Pdgfb mRNA. Experimental silencing of HIF1α and overexpression of let-7d results in reduced Pdgfb/PDGFA expression. Values represent mean ± SD. P values were determined by a two-tailed Student t test. **, P < 0.01; ***, P < 0.005. Box plot shows median, quartiles, and extreme values.

Figure 3.

HIF1 and miRNA let-7d regulate expression of Pdgfb/PDGFA in brain metastatic cells. A, Tabular view of the 18 coexpressed genes combined with HIF1α/HIF1β and let-7 regulation as predicted by ISMARA. The ranking was performed according to z values provided by the analysis. Only PDGFB gene was predicted to be regulated by both HIF1 and let-7. B, Kaplan–Meier plot shows BMFS of patients with breast cancer stratified for low and high PDGFB expression levels. The cutoff used is the 67th percentile: 570 patients have PDGFB levels below (PDGFB low) and 285 above (PDGFB high) this percentile. C,PDGFB expression levels in primary tumors of patients without (0; 806 patients) or with (1; 49 patients) brain metastases (left) and in brain (49 patients) versus lung/bone metastasis (297 patients) samples from patients with breast cancer (right). D, Relative Pdgfb mRNA and protein expression in 4T1-BM2, 4T1-LM2, and 4T1-T2 cells. E, Relative Pdgfb mRNA and protein expression upon stable silencing of HIF1α in 4T1-BM2 cells (HIF1α_kd#1 and kd#2). F, Relative Pdgfb mRNA levels upon overexpression of miRNA let-7d in 4T1-BM2 cells. G, Representation of the proposed model of regulation of the Pdgfb transcription and translation as predicted by ISMARA. HIF1α binds transcription factor (TF) motif in the promoter region, whereas miRNA let-7d recognizes an 8-mer binding site in the 3′ UTR of Pdgfb mRNA. Experimental silencing of HIF1α and overexpression of let-7d results in reduced Pdgfb/PDGFA expression. Values represent mean ± SD. P values were determined by a two-tailed Student t test. **, P < 0.01; ***, P < 0.005. Box plot shows median, quartiles, and extreme values.

Close modal

Importantly, elevated PDGFB expression in the primary tumors of patients with breast cancer correlated with shorter BMFS (Fig. 3B). Furthermore, primary tumors of brain metastatic patients (Fig. 3C, left), as well as brain metastases compared with metastases to lung or bone (Fig. 3C, right), expressed higher levels of PDGFB. Consistent with these bioinformatics predictions, Pdgfb mRNA and protein expressions in 4T1-BM2 and D2A1-BM2 cells were significantly elevated compared with their respective non-brain metastatic counterparts (Fig. 3D; Supplementary Fig. S3A). PDGFA mRNA has been reported previously to be upregulated in the gene signature from the human MDA231-BrM2 brain metastatic model (5). We confirmed here elevated levels of PDGFA mRNA and protein in MDA231-BrM2 cells relative to MDA231 parental cells (Supplementary Fig. S3B). Pdgfa mRNA was also upregulated in 4T1-BM2 compared with 4T1-T2 cells, while PDGFB mRNA was expressed at equivalent levels in MDA231-BrM2 compared with MDA231 parental cells (Supplementary Fig. S3C). Increased PDGFA expression in patients with breast cancer was also associated with a reduced BMFS (Supplementary Fig. S3D).

HIF1 was reported to induce PDGFB gene expression under hypoxia (19), and according to bioinformatics prediction, the promoter region of the Pdgfb gene contains multiple HIF1-binding motifs, consistent with direct transcriptional regulation by HIF1. On the other side, the 3′ UTR of the Pdgfb mRNA was predicted to contain an miRNA let-7d conserved binding site consistent with posttranscriptional repression of Pdgfb protein translation. Consistently, stable silencing of HIF1α reduced Pdgfb mRNA and protein expressions in 4T1-BM2 and D2A1-BM2 murine brain metastatic cells (Fig. 3E; Supplementary Fig. S3E), as well as PDGFA expression in human brain metastatic MDA231-BrM cells (Supplementary Fig. S3F). Conversely, induced expression of miRNA let-7d led to decreased Pdgfb and PDGFA expression in the murine and human brain metastatic cells, respectively (Fig. 3F; Supplementary Fig. S3G and S3H). These experimental data are consistent with the in silico ISMARA prediction (Fig. 3G).

While these results demonstrate that HIF1 and let-7d individually regulate PDGF expression, they do not demonstrate a possible functional cross-talk between these two regulatory pathways. To address this question, we tested the effect of HIF1α silencing on let-7d (and also Let-7f belonging to the same miRNA family) expression, and contrariwise, the effect of let-7d overexpression on HIF1α mRNA expression and HIF transcriptional activity by monitoring expression of the target genes VEGF, CA9, and GLUT1 in both 4T1-BM2 and MDA231-BrM2 mouse and human models. We found that Let-7d overexpression reduced VEGF, CA9, and GLUT1 target gene expressions and HIF1α in MDA231-BrM2 cells and Vegf and Glut1 target gene expressions in 4T1-BM2 cells (Supplementary Fig. S4A and S4B). In contrast, we did not observe any significant effect of HIF1α silencing on let-7d or let-7f expressions (Supplementary Fig. S4C).

We conclude that HIF1 activation and repression of miRNA let-7d expression result in increased expression of Pdgfb and PDGFA in both the murine and human brain metastatic models, respectively, and that let-7d itself regulates HIF1α mRNA expression and HIF1 activity. These results also demonstrate a coordinated regulation of PDGF expression by HIF1 and let-7d.

Pdgfb contributes to the metastatic outgrowth of breast cancer cells in the brain

To reveal a potential role of tumor cell–derived Pdgfb or PDGFA in promoting cell growth under condition of limited growth factors availability, as it may occur during initial cell seeding in the brain, we silenced Pdgfb in 4T1-BM2 cells and tested the effect on their proliferation and survival under condition of growth factor deprivation in vitro (i.e., 1% FBS). Stable silencing of Pdgfb in 4T1-BM2 and D2A1-BM2 cells (Supplementary Fig. S5A and S5B) had no direct effect on cell survival, as assessed by Annexin V/propidium iodide staining (Supplementary Fig. S5C), while it decreased proliferation rate as detected by cell-cycle analyses (Supplementary Fig. S5D, at 48-hour timepoint), compared with nonsilencing control, resulting in reduced cell number in time (Fig. 4A; Supplementary Fig. S5E). This effect was reversed by supplementing recombinant Pdgfbb to the Pdgfb-silenced 4T1-BM2 and D2A1-BM2 cells (Fig. 4B; Supplementary Fig. S5F), thereby confirming the critical role of Pdgfb in 4T1-BM2 and D2A1-BM2 cell growth under condition of limited growth factors availability in vitro.

Figure 4.

Tumor-derived Pdgfb/PDGFA promotes brain metastasis. A, Relative cell number of 4T1-BM2 cells in 1% FBS upon stable Pdgfb silencing. B, Relative cell number of 4T1-BM2 Pdgfb-silenced cells in 1% FBS supplemented with recombinant Pdgfbb at the indicated concentrations. C, Growth curve (left) and weight (middle) of primary tumors and brain metastases (brain metastatic index; right) derived from 4T1-BM2 Pdgfb-silenced and control NS cells injected orthotopically in the fourth mammary fat pad (mfp). D, MVD and illustrative sections stained for CD31 of primary tumors of mice injected orthotopically with Pdgfb-silenced 4T1-BM2 or NS control cells at endpoint. E, Brain metastatic progression of 4T1-BM2 Pdgfb-silenced and control 4T1-BM2 NS cells after intracarotid (ica) injection, and monitored by in vivo (left) and ex vivo (middle and right) BLI. F, Brain metastatic progression of MDA231-BrM2_NS and _PDGFA-silenced (kd#1 and kd #2) cells after intracarotid injection, and monitored by in vivo (left) and ex vivo (middle and right) BLI. Representative ex vivo BLI brain pictures are shown. Mice per group was n = 9. Values represent mean ± SD (in vitro) or SEM (in vivo). P values were determined by a two-tailed Student t test (***, P < 0.005; **, P < 0.01; *, P < 0.05).

Figure 4.

Tumor-derived Pdgfb/PDGFA promotes brain metastasis. A, Relative cell number of 4T1-BM2 cells in 1% FBS upon stable Pdgfb silencing. B, Relative cell number of 4T1-BM2 Pdgfb-silenced cells in 1% FBS supplemented with recombinant Pdgfbb at the indicated concentrations. C, Growth curve (left) and weight (middle) of primary tumors and brain metastases (brain metastatic index; right) derived from 4T1-BM2 Pdgfb-silenced and control NS cells injected orthotopically in the fourth mammary fat pad (mfp). D, MVD and illustrative sections stained for CD31 of primary tumors of mice injected orthotopically with Pdgfb-silenced 4T1-BM2 or NS control cells at endpoint. E, Brain metastatic progression of 4T1-BM2 Pdgfb-silenced and control 4T1-BM2 NS cells after intracarotid (ica) injection, and monitored by in vivo (left) and ex vivo (middle and right) BLI. F, Brain metastatic progression of MDA231-BrM2_NS and _PDGFA-silenced (kd#1 and kd #2) cells after intracarotid injection, and monitored by in vivo (left) and ex vivo (middle and right) BLI. Representative ex vivo BLI brain pictures are shown. Mice per group was n = 9. Values represent mean ± SD (in vitro) or SEM (in vivo). P values were determined by a two-tailed Student t test (***, P < 0.005; **, P < 0.01; *, P < 0.05).

Close modal

Next, we investigated the in vivo functional role of cancer cell–derived Pdgfb in the formation of spontaneous brain metastasis in the 4T1-BM2 orthotopic breast cancer model. Primary tumor growth was not significantly altered upon Pdgfb silencing (Fig. 4C, left and middle). However, the spontaneous formation of 4T1-BM2 brain metastases was significantly decreased (Fig. 4C, right). Strikingly, Pdgfb silencing did not cause a decrease in microvascular density (MVD) in the primary tumor, but rather it induced a slight increased (∼20%) and a visible enlargement of the tumor vessels (Fig. 4D).

Furthermore, to address the functional role of Pdgfb and PDGFA in the late-metastatic steps of colonization and outgrowth in the brain, we directly injected 4T1-BM2 and MDA231-BrM2 cells into the common carotid artery. Indeed, stable silencing of Pdgfb in 4T1-BM2 hampered the formation of brain metastases, as monitored by in vivo and ex vivo BLI (Fig. 4E). Stable silencing of PDGFA expression in MDA231-BrM2 cells (Supplementary Fig. S5G) reduced proliferation in vitro (Supplementary Fig. S5H) and suppressed brain metastases formation in vivo in the human model (Fig. 4F), paralleling the effects observed in the mouse model.

We conclude that cancer cell–derived Pdgfb and PDGFA are essential mediators of brain metastasis formation in the murine and human brain metastasis models, respectively.

Pharmacologic inhibition of PDGFR prevents brain metastasis formation and suppresses growth of established brain metastases

The expression of PDGFRs in breast cancer cells correlates with tumor progression, invasion, and metastasis (20–23). Experimental studies demonstrated an important role for Pdgfr activation on pericytes and astrocytes in blood–brain barrier stability and brain metastasis formation (24, 25). We, therefore, decided to investigate the role of PDGFR in the progression of brain metastasis in our models. First, we confirmed the expression of Pdgfra/PDGFRA and Pdgfrb/PDGFRB in 4T1-BM2, D2A1-BM2, and MDA231-BrM2 cells, respectively. Both receptors were expressed in brain metastatic cells and in the respective non-brain metastatic counterparts at similar levels (Supplementary Fig. S6A–S6C). Second, we showed that treatment of brain metastatic cells in vitro with the TKI, nilotinib, a second-generation BCR-ABL TKI with potent PDGFR inhibitory activity (26), significantly inhibited in vitro cell autonomous proliferation across the three different breast cancer models (Supplementary Fig. S6D). Third, we systemically (per os) treated mice previously injected in the fourth mammary fat pad with 4T1-BM2 mammary tumors, by administrating nilotinib starting from day 10, a timepoint when tumors are well established, until endpoint. Primary tumor growth (Fig. 5A, left and middle), as well as the spontaneous formation of lung metastases (Supplementary Fig. S6E), were not significantly affected by the nilotinib treatment, while the spontaneous formation of brain metastases was strongly and significantly reduced (Fig. 5A, right). Fourth, to mimic the clinical situation, whereby patients are treated only when brain metastases are diagnosed, we injected 4T1-BM2 or MDA231-BrM2 cells into the right common carotid artery and started nilotinib treatment once metastases were detectable by in vivo BLI (days 3 and 4, respectively, after injection). Systemic treatment with nilotinib significantly reduced brain metastatic burden compared with untreated vehicle control mice, as detected by in vivo and ex vivo BLI in both murine and human models (Fig. 5B and C), while it did not affect lung metastasis formation (Supplementary Fig. S6F and S6G). Nilotinib treatment did not affect MVD in 4T1-BM2 primary tumors, while it induced a visible enlargement of the tumor vessels (Fig. 5D).

Figure 5.

Inhibition of Pdgfr/PDGFR signaling by nilotinib halts brain metastasis progression. A, Growth curve (left) and weight (middle) of breast primary tumors and brain metastases (brain metastatic index; right) derived from 4T1-BM2 cells injected orthotopically in the fourth mammary fat pad, and treated with nilotinib (30 mg/kg) or vehicle daily (orally) from day 10 till endpoint after tumor cell injection. Nilotinib inhibits spontaneous brain metastasis formation, but not primary tumor growth. B, Brain metastatic progression of 4T1-BM2-LV-Luc cells in mice injected via carotid artery and treated with nilotinib (30 mg/kg) or vehicle daily (orally) from day 3 till endpoint (left), and monitored by in vivo (middle) and ex vivo (right) BLI. C, Brain metastatic progression of MDA231-BrM2-LV-Luc cells in mice injected via carotid artery and treated with nilotinib (30 mg/kg) or vehicle daily (orally) from day 4 till endpoint (left), and monitored by in vivo (middle) and ex vivo (right) BLI. Representative BLI images are shown. Nilotinib inhibits brain metastasis progression. D, MVD and illustrative sections stained for CD31 of primary tumors of mice injected orthotopically with 4T1-BM2 cells and treated with nilotinib (30 mg/kg) or vehicle only, daily from day 10 after tumor cell injection till endpoint. Mice per group was n = 8. Values represent mean ± SD (in vitro) or SEM (in vivo). P values were determined by a two-tailed Student t test. *, P < 0.05; **, P < 0.01.

Figure 5.

Inhibition of Pdgfr/PDGFR signaling by nilotinib halts brain metastasis progression. A, Growth curve (left) and weight (middle) of breast primary tumors and brain metastases (brain metastatic index; right) derived from 4T1-BM2 cells injected orthotopically in the fourth mammary fat pad, and treated with nilotinib (30 mg/kg) or vehicle daily (orally) from day 10 till endpoint after tumor cell injection. Nilotinib inhibits spontaneous brain metastasis formation, but not primary tumor growth. B, Brain metastatic progression of 4T1-BM2-LV-Luc cells in mice injected via carotid artery and treated with nilotinib (30 mg/kg) or vehicle daily (orally) from day 3 till endpoint (left), and monitored by in vivo (middle) and ex vivo (right) BLI. C, Brain metastatic progression of MDA231-BrM2-LV-Luc cells in mice injected via carotid artery and treated with nilotinib (30 mg/kg) or vehicle daily (orally) from day 4 till endpoint (left), and monitored by in vivo (middle) and ex vivo (right) BLI. Representative BLI images are shown. Nilotinib inhibits brain metastasis progression. D, MVD and illustrative sections stained for CD31 of primary tumors of mice injected orthotopically with 4T1-BM2 cells and treated with nilotinib (30 mg/kg) or vehicle only, daily from day 10 after tumor cell injection till endpoint. Mice per group was n = 8. Values represent mean ± SD (in vitro) or SEM (in vivo). P values were determined by a two-tailed Student t test. *, P < 0.05; **, P < 0.01.

Close modal

From these experiments we conclude that, pharmacologic inhibition of PDGFR signaling significantly reduces de novo formation of brain metastases (4T1-BM2 model), as well as the outgrowth of already established breast cancer brain metastases in both murine 4T1-BM2 and human MDA231-BrM2 models. Importantly, decreased 4T1-BM2 spontaneous brain metastasis formation following nilotinib treatment (or Pdgfb silencing) was not associated with decreased angiogenesis at the primary tumor site.

Metastatic relapse in the brain is a critical complication of many cancer types associated with rapid decline of quality of life and high mortality (1). Because of lack of efficient clinical treatments, there is an urgent need to better understand the cellular and molecular mechanisms responsible for this particularly dismal metastatic disease with the aim to identify novel therapeutic candidates (3). Several cellular and molecular mechanisms of breast cancer brain metastasis have been recently reported on the basis of experimental models of brain metastasis by systemic injection of cancer cells (4, 27).

We recently developed a spontaneous model of breast cancer metastasis to the brain and unraveled an adaptive molecular program initiated by Cx31/43 coordinating cancer cell survival via FAK and NF-κB and laminin 5/8 deposition in the brain during metastatic colonization (Lorusso and colleagues; unpublished data).

Here, we set-up to identify transcriptional and posttranscriptional regulators of brain metastasis formation and report that HIF1 and miRNA let-7d coordinate the upregulation of Pdgfb/PDGFA protein expressions in brain metastatic cells. We demonstrate that tumor-derived Pdgfb/PDGFA is critically involved in promoting breast cancer metastasis to the brain. Importantly, systemic treatment with the TKI, nilotinib, prevents spontaneous brain metastasis dissemination from primary tumor, as well as it reduces growth of already established brain metastases in the mouse and human models. Taken together, these data reveal that gain of HIF1 activity and loss of miRNA let-7d orchestrate breast cancer metastasis to the brain via the PDGF/PDGFR axis.

Elevated HIF1 transcriptional activity in cancer correlates with more aggressive and more metastatic behavior and it is associated with poor prognosis in patients, including in breast cancer (28, 29). HIF1 regulates multiple events promoting cancer progression, such as metabolic adaptation, cell motility, epithelial-to-mesenchymal transition (EMT), survival, and angiogenesis (30). Increased HIF1 index has been reported in brain metastases versus primary tumors in a variety of cancers, including melanoma, lung, breast, renal, and colorectal cancers, but the putative mechanism involved has not been investigated (28, 31, 32). Here, we show that in three independent models of brain metastasis, HIF1 is constitutively active under standard normoxic culture conditions, and HIF1α silencing suppresses brain metastasis formation. There is growing evidence that HIF1α regulation goes beyond oxygen levels. Mitochondria-derived reactive oxygen species, extracellular factors, such as angiotensin II, thrombin, and endothelin, and activation of PI3K and PKC pathways were reported to increase HIF1 activity independently of hypoxia. Increased HIF1α protein translation, rather than PHD-dependent stabilization, seems to be the main mechanism involved (33–36). As we found that let-7d regulates HIF1 activity (Supplementary Fig. S4A and S4B), but not vice versa, a possibility to consider is that deregulated Lin28/miRNA let-7d may be responsible for increased HIF1 activity independently of hypoxia.

Having identified Pdgfb/PDGFA as transcriptional targets positively regulated by HIF1 in brain metastatic cancer cells, this may suggest that HIF1 is a potential therapeutic target, however, while various small-molecular HIF1α inhibitors were developed and tested for the treatment of advanced cancers, none is approved for clinical use yet, due to lack of specificity and undesired side effects (37, 38).

Parallel to HIF1 activation, we observed decreased expression of miRNA let-7d (and increased expression of let-7d target genes) in brain metastatic breast cancer cells, as well as increased expression of its negative regulator, Lin28 (17, 39). An inverse correlation between the miRNA-processing protein, Lin28, and miRNA let-7d was reported previously. Gain of Lin28 expression and concomitant loss of miRNA let-7d maturation are associated with aggressive cancers. Interestingly, those cancers inherited high self-renewal capacity and were associated with poor prognosis in patients (18, 40). Here, we show that Lin28 is strongly expressed in brain metastases from patients, by performing dataset bioinformatics analyses. Thus, Lin28-controlled inhibition of miRNA let-7 might serve as a posttranscriptional mechanism further contributing to brain metastatic capacity of breast cancer cells. Importantly, breast cancer patients with brain metastasis show significantly lower expression of miRNA let-7 in the primary tumor and in brain metastatic lesions. Previous studies identified miRNA let-7 as a suppressor of tumorigenesis and metastasis (41, 42). Suppression of miRNA let-7 increased tumor angiogenesis, induction of EMT, invasion, and metastasis including of the breast (43–45). A direct role of miRNA let-7 in brain metastatic colonization, however, has not been reported yet. Here, we show that loss of miRNA let-7 promotes brain metastasis by stabilizing Pdgfb/PGDFA mRNAs and increasing their protein translation. Interestingly, we found that let-7d overexpression inhibits HIF1 activity (while HIF1 inhibition did not affect let-7 expression), suggesting a regulatory coordination between these two systems in modulating Pdgfb/PGDFA mRNA levels. Importantly, our results do not exclude that additional HIF1-regulated genes may contribute to the brain metastatic phenotype. Indeed, among the genes upregulated in 4T1-BM2 versus 4T1-T2/LM2 cells (Lorusso and colleagues; unpublished data), at least eight additional genes (i.e., Krt14, Mmp9, Car9, Slc11a2, Loxl4, Igfbp7, Edn1, and Ctgf) are known to be regulated by HIF1 and to promote cancer progression and metastasis. Their putative role in brain metastasis, however, remains unknown at this point.

PDGF signaling was reported to drive tumorigenesis, proliferation, invasion, and metastasis in various tumor models (22, 46). The activation of the PDGF pathway in breast cancer correlates with tumor aggressiveness and early recurrence, particularly in the triple-negative subtype (23, 47). PDGFRs are also critical mediators of breast cancer chemoresistance driven by Foxq1 expression (48). It has been reported that metastatic cells disseminating to the brain express PDGFs (7, 49). Through bioinformatics analysis we show that, PDGFB and PDGFA expression in primary tumors of patients with breast cancer correlates with shorter BMFS. Furthermore, patients' brain metastatic lesions express higher levels of PDGFB than metastatic lesions derived from other organs (lung or bone). Pdgfb or PDGFA expression by metastatic tumor cells provides an autocrine proliferation loop, which is critical for efficient metastatic colonization of the brain. In vitro experiments demonstrate that Pdgfb produced by tumor cells induces proliferation and promotes cell autonomous survival. Silencing of Pdgfb reduced tumor cell proliferation and brain metastasis formation. Inhibition of PDGFR signaling with the TKI, nilotinib (26), prevented the formation of spontaneous brain metastases from the primary tumor site and also halted the progression of established ones.

Our results do not exclude that PDGFB may also initiate paracrine communication with resident cells within the brain parenchyma, such as astrocytes or microglia, further promoting and supporting metastatic outgrowth. Indeed, Gril and colleagues demonstrated that abrogation of PDGFR signaling in astrocytes with the small-molecular kinase inhibitor, pazopanib, prevented outgrowth of experimental brain metastases derived from HER2+ breast cancer (25). More recently, Thies and colleagues reported that stromal PDGFR-beta signaling promotes breast cancer brain metastasis in response to tumor-derived PDGFB (50).

In conclusion, we provide original evidence for a novel regulatory and effector pathway involving HIF1, Let-7d/Lin28, and PDGFA/B in promoting metastasis to the brain. Pharmacologic PDGFR inhibition may be an attractive approach to considered for clinical testing in the treatment of patients with breast cancer presenting with brain metastases given the current lack of effective treatments and the availability of several clinically approved PDGFR TKIs (i.e., imatinib, pazopanib, nilotinib, and crenolanib).

No disclosures were reported.

C.B. Wyss: Conceptualization, formal analysis, supervision, investigation, visualization, methodology, writing-original draft, writing-review and editing. N. Duffey: Investigation, visualization, methodology. S. Peyvandi: Investigation, visualization, methodology. D. Barras: Data curation, investigation, visualization, methodology. A. Martinez Usatorre: Investigation, methodology. O. Coquoz: Investigation, methodology. P. Romero: Supervision, investigation, methodology. M. Delorenzi: Supervision, funding acquisition, investigation, visualization, methodology. G. Lorusso: Conceptualization, formal analysis, supervision, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. C. Rüegg: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, writing-original draft, project administration, writing-review and editing.

The authors wish to thank Prof. Dr. Joan Massagué (Memorial Sloan Kettering Cancer Center, New York, NY) for providing MDA-231 and MDA-231-BrM2 human breast cancer cell lines. We thank Prof. Erik van Nimwegen and Dr. Mikhail Pachkov (Biozentrum, University of Basel, Basel, Switzerland) for assistance with ISMARA. We thank Dr. L. Borsig (Institute of Physiology, University of Zurich, Zurich, Switzerland) for providing murine HIF1α shRNA sequences. This work was supported by grants from the Swiss National Science Foundation (PDFMP3_137079, 31003A_159824, and 31003A_179248), the Swiss Cancer League (KFS-2814-08-2011, KFS-3513-08-2014, and KFS 4400-02-2018), and the Medic Foundation (to C. Rüegg).

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.

1.
Takei
H
,
Rouah
E
,
Ishida
Y
. 
Brain metastasis: clinical characteristics, pathological findings and molecular subtyping for therapeutic implications
.
Brain Tumor Pathol
2016
;
33
:
1
12
.
2.
Ponde
NF
,
Zardavas
D
,
Piccart
M
. 
Progress in adjuvant systemic therapy for breast cancer
.
Nat Rev Clin Oncol
2019
;
16
:
27
44
.
3.
Franchino
F
,
Ruda
R
,
Soffietti
R
. 
Mechanisms and therapy for cancer metastasis to the brain
.
Front Oncol
2018
;
8
:
161
.
4.
Ni
W
,
Chen
W
,
Lu
Y
. 
Emerging findings into molecular mechanism of brain metastasis
.
Cancer Med
2018
;
7
:
3820
33
.
5.
Bos
PD
,
Zhang
XH
,
Nadal
C
,
Shu
W
,
Gomis
RR
,
Nguyen
DX
, et al
Genes that mediate breast cancer metastasis to the brain
.
Nature
2009
;
459
:
1005
9
.
6.
Valiente
M
,
Obenauf
AC
,
Jin
X
,
Chen
Q
,
Zhang
XH
,
Lee
DJ
, et al
Serpins promote cancer cell survival and vascular co-option in brain metastasis
.
Cell
2014
;
156
:
1002
16
.
7.
Chen
Q
,
Boire
A
,
Jin
X
,
Valiente
M
,
Er
EE
,
Lopez-Soto
A
, et al
Carcinoma-astrocyte gap junctions promote brain metastasis by cGAMP transfer
.
Nature
2016
;
533
:
493
8
.
8.
Zeng
Q
,
Michael
IP
,
Zhang
P
,
Saghafinia
S
,
Knott
G
,
Jiao
W
, et al
Synaptic proximity enables NMDAR signalling to promote brain metastasis
.
Nature
2019
;
573
:
526
31
.
9.
Cruz-Munoz
W
,
Kerbel
RS
. 
Preclinical approaches to study the biology and treatment of brain metastases
.
Semin Cancer Biol
2011
;
21
:
123
30
.
10.
Daphu
I
,
Sundstrom
T
,
Horn
S
,
Huszthy
PC
,
Niclou
SP
,
Sakariassen
PO
, et al
In vivo animal models for studying brain metastasis: value and limitations
.
Clin Exp Metastasis
2013
;
30
:
695
710
.
11.
Bos
PD
,
Nguyen
DX
,
Massague
J
. 
Modeling metastasis in the mouse
.
Curr Opin Pharmacol
2010
;
10
:
571
7
.
12.
Aslakson
CJ
,
Miller
FR
. 
Selective events in the metastatic process defined by analysis of the sequential dissemination of subpopulations of a mouse mammary tumor
.
Cancer Res
1992
;
52
:
1399
405
.
13.
Balwierz
PJ
,
Pachkov
M
,
Arnold
P
,
Gruber
AJ
,
Zavolan
M
,
van Nimwegen
E
. 
ISMARA: automated modeling of genomic signals as a democracy of regulatory motifs
.
Genome Res
2014
;
24
:
869
84
.
14.
Kaluz
S
,
Kaluzova
M
,
Liao
SY
,
Lerman
M
,
Stanbridge
EJ
. 
Transcriptional control of the tumor- and hypoxia-marker carbonic anhydrase 9: a one transcription factor (HIF-1) show?
Biochim Biophys Acta
2009
;
1795
:
162
72
.
15.
Chen
CH
,
Pore
N
,
Behrooz
A
,
Ismail
BF
,
Maity
A
. 
Regulation of GLUT1 mRNA by hypoxia-inducible factor-1: interaction between H-ras and hypoxia
.
J Biol Chem
2001
;
276
:
9519
25
.
16.
Bilton
RL
,
Booker
GW
. 
The subtle side to hypoxia inducible factor (HIFalpha) regulation
.
Eur J Biochem
2003
;
270
:
791
8
.
17.
Rybak
A
,
Fuchs
H
,
Smirnova
L
,
Brandt
C
,
Pohl
EE
,
Nitsch
R
, et al
A feedback loop comprising lin-28 and let-7 controls pre-let-7 maturation during neural stem-cell commitment
.
Nat Cell Biol
2008
;
10
:
987
93
.
18.
Viswanathan
SR
,
Daley
GQ
. 
Lin28: a microRNA regulator with a macro role
.
Cell
2010
;
140
:
445
9
.
19.
Schito
L
,
Rey
S
,
Tafani
M
,
Zhang
H
,
Wong
CC
,
Russo
A
, et al
Hypoxia-inducible factor 1-dependent expression of platelet-derived growth factor B promotes lymphatic metastasis of hypoxic breast cancer cells
.
Proc Natl Acad Sci U S A
2012
;
109
:
E2707
16
.
20.
Seymour
L
,
Bezwoda
WR
. 
Positive immunostaining for platelet derived growth factor (PDGF) is an adverse prognostic factor in patients with advanced breast cancer
.
Breast Cancer Res Treat
1994
;
32
:
229
33
.
21.
Jechlinger
M
,
Sommer
A
,
Moriggl
R
,
Seither
P
,
Kraut
N
,
Capodiecci
P
, et al
Autocrine PDGFR signaling promotes mammary cancer metastasis
.
J Clin Invest
2006
;
116
:
1561
70
.
22.
Carvalho
I
,
Milanezi
F
,
Martins
A
,
Reis
RM
,
Schmitt
F
. 
Overexpression of platelet-derived growth factor receptor alpha in breast cancer is associated with tumour progression
.
Breast Cancer Res
2005
;
7
:
R788
95
.
23.
Roswall
P
,
Bocci
M
,
Bartoschek
M
,
Li
H
,
Kristiansen
G
,
Jansson
S
, et al
Microenvironmental control of breast cancer subtype elicited through paracrine platelet-derived growth factor-CC signaling
.
Nat Med
2018
;
24
:
463
73
.
24.
Winkler
EA
,
Bell
RD
,
Zlokovic
BV
. 
Pericyte-specific expression of PDGF beta receptor in mouse models with normal and deficient PDGF beta receptor signaling
.
Mol Neurodegener
2010
;
5
:
32
.
25.
Gril
B
,
Palmieri
D
,
Qian
Y
,
Anwar
T
,
Liewehr
DJ
,
Steinberg
SM
, et al
Pazopanib inhibits the activation of PDGFRbeta-expressing astrocytes in the brain metastatic microenvironment of breast cancer cells
.
Am J Pathol
2013
;
182
:
2368
79
.
26.
Blay
JY
,
von Mehren
M
. 
Nilotinib: a novel, selective tyrosine kinase inhibitor
.
Semin Oncol
2011
;
38
:
S3
9
.
27.
Yuzhalin
AE
,
Yu
D
. 
Brain metastasis organotropism
.
Cold Spring Harb Perspect Med
2020
;
10
:
a037242
.
28.
Gilkes
DM
,
Semenza
GL
. 
Role of hypoxia-inducible factors in breast cancer metastasis
.
Future Oncol
2013
;
9
:
1623
36
.
29.
Pezzuto
A
,
Carico
E
. 
Role of HIF-1 in cancer progression: novel insights. a review.
Curr Mol Med
2018
;
18
:
343
51
.
30.
Schito
L
,
Semenza
GL
. 
Hypoxia-inducible factors: master regulators of cancer progression
.
Trends Cancer
2016
;
2
:
758
70
.
31.
Crowder
SW
,
Balikov
DA
,
Hwang
YS
,
Sung
HJ
. 
Cancer stem cells under hypoxia as a chemoresistance factor in breast and brain
.
Curr Pathobiol Rep
2014
;
2
:
33
40
.
32.
Berghoff
AS
,
Ilhan-Mutlu
A
,
Dinhof
C
,
Magerle
M
,
Hackl
M
,
Widhalm
G
, et al
Differential role of angiogenesis and tumour cell proliferation in brain metastases according to primary tumour type: analysis of 639 cases
.
Neuropathol Appl Neurobiol
2015
;
41
:
e41
55
.
33.
Patten
DA
,
Lafleur
VN
,
Robitaille
GA
,
Chan
DA
,
Giaccia
AJ
,
Richard
DE
. 
Hypoxia-inducible factor-1 activation in nonhypoxic conditions: the essential role of mitochondrial-derived reactive oxygen species
.
Mol Biol Cell
2010
;
21
:
3247
57
.
34.
Iommarini
L
,
Porcelli
AM
,
Gasparre
G
,
Kurelac
I
. 
Non-canonical mechanisms regulating hypoxia-inducible factor 1 alpha in cancer
.
Front Oncol
2017
;
7
:
286
.
35.
Dery
MA
,
Michaud
MD
,
Richard
DE
. 
Hypoxia-inducible factor 1: regulation by hypoxic and non-hypoxic activators
.
Int J Biochem Cell Biol
2005
;
37
:
535
40
.
36.
Richard
DE
,
Berra
E
,
Pouyssegur
J
. 
Nonhypoxic pathway mediates the induction of hypoxia-inducible factor 1alpha in vascular smooth muscle cells
.
J Biol Chem
2000
;
275
:
26765
71
.
37.
Semenza
GL
. 
Hypoxia-inducible factors: mediators of cancer progression and targets for cancer therapy
.
Trends Pharmacol Sci
2012
;
33
:
207
14
.
38.
Fallah
J
,
Rini
BI
. 
HIF inhibitors: status of current clinical development
.
Curr Oncol Rep
2019
;
21
:
6
.
39.
Newman
MA
,
Thomson
JM
,
Hammond
SM
. 
Lin-28 interaction with the Let-7 precursor loop mediates regulated microRNA processing
.
RNA
2008
;
14
:
1539
49
.
40.
Heo
I
,
Joo
C
,
Cho
J
,
Ha
M
,
Han
J
,
Kim
VN
. 
Lin28 mediates the terminal uridylation of let-7 precursor MicroRNA
.
Mol Cell
2008
;
32
:
276
84
.
41.
Yu
F
,
Yao
H
,
Zhu
P
,
Zhang
X
,
Pan
Q
,
Gong
C
, et al
let-7 regulates self renewal and tumorigenicity of breast cancer cells
.
Cell
2007
;
131
:
1109
23
.
42.
Sempere
LF
,
Christensen
M
,
Silahtaroglu
A
,
Bak
M
,
Heath
CV
,
Schwartz
G
, et al
Altered microRNA expression confined to specific epithelial cell subpopulations in breast cancer
.
Cancer Res
2007
;
67
:
11612
20
.
43.
Zhao
B
,
Han
H
,
Chen
J
,
Zhang
Z
,
Li
S
,
Fang
F
, et al
MicroRNA let-7c inhibits migration and invasion of human non-small cell lung cancer by targeting ITGB3 and MAP4K3
.
Cancer Lett
2014
;
342
:
43
51
.
44.
Yan
LX
,
Huang
XF
,
Shao
Q
,
Huang
MY
,
Deng
L
,
Wu
QL
, et al
MicroRNA miR-21 overexpression in human breast cancer is associated with advanced clinical stage, lymph node metastasis and patient poor prognosis
.
RNA
2008
;
14
:
2348
60
.
45.
Hu
X
,
Guo
J
,
Zheng
L
,
Li
C
,
Zheng
TM
,
Tanyi
JL
, et al
The heterochronic microRNA let-7 inhibits cell motility by regulating the genes in the actin cytoskeleton pathway in breast cancer
.
Mol Cancer Res
2013
;
11
:
240
50
.
46.
Pietras
K
,
Sjoblom
T
,
Rubin
K
,
Heldin
CH
,
Ostman
A
. 
PDGF receptors as cancer drug targets
.
Cancer Cell
2003
;
3
:
439
43
.
47.
Jansson
S
,
Aaltonen
K
,
Bendahl
PO
,
Falck
AK
,
Karlsson
M
,
Pietras
K
, et al
The PDGF pathway in breast cancer is linked to tumour aggressiveness, triple-negative subtype and early recurrence
.
Breast Cancer Res Treat
2018
;
169
:
231
41
.
48.
Meng
F
,
Speyer
CL
,
Zhang
B
,
Zhao
Y
,
Chen
W
,
Gorski
DH
, et al
PDGFRalpha and beta play critical roles in mediating Foxq1-driven breast cancer stemness and chemoresistance
.
Cancer Res
2015
;
75
:
584
93
.
49.
Boire
A
,
Zou
Y
,
Shieh
J
,
Macalinao
DG
,
Pentsova
E
,
Massague
J
. 
Complement component 3 adapts the cerebrospinal fluid for leptomeningeal metastasis
.
Cell
2017
;
168
:
1101
13 e13
.
50.
Thies
KA
,
Hammer
AM
,
Hildreth
BE
 III
,
Steck
SA
,
Spehar
JM
,
Kladney
RD
, et al
Stromal platelet–derived growth factor receptor-β signaling promotes breast cancer metastasis in the brain
.
Cancer Res
2021
;
81
:
606
18
.