Losartan treatment enhances chemotherapy efficacy and reduces ascites in ovarian cancer models by normalizing the tumor stroma

Losartan Treatment Reduces ECM Content in Ovarian Tumors.

We first confirmed that AT1, the target for losartan, is expressed in our ovarian cancer cell lines (SI Appendix, Fig. S1A). Next, we determined if AT1 blockade by losartan treatment led to changes in ECM content. Mice were implanted with two human ovarian cancer cells, SKOV3ip1 and Hey-A8, orthotopically into the peritoneal cavity. To monitor peritoneal tumor growth, both cell lines were transduced with secretive Gaussia luciferase reporter gene (G-luc). Between 7 and 10 d after implantation when the blood G-luc value reached 2 × 10⁶ relative light units (RLU), mice were randomized into control and losartan (40 mg/kg, once daily) treatment groups. i.p. injection of cancer cells resulted in solid tumors growing on the surface of the peritoneal organs and invading into the diaphragm. Mice bearing SKOV3ip1 tumors also produced a large amount of ascites. Tumor tissues were collected on day 28 postimplantation and evaluated for ECM content. We found that losartan treatment significantly reduced collagen and hyaluronan levels in both SKOV3ip1 and Hey-A8 tumors as indicated by histological analysis and by cDNA array (Fig. 1 A and B and SI Appendix, Fig. S1B).

Fibroblasts are the primary source of ECM proteins in both normal and malignant tissues (23, 24). In both SKOV3ip1 and Hey-A8 models, we found that losartan treatment significantly decreased the number of intratumoral αSMA-positive stromal cells, the main cellular components among which are the fibrogenic fibroblasts (Fig. 1C). Losartan treatment significantly reduced the expression of matrix molecules, including collagen (Col)-I and III, alpha smooth muscle actin (Acta2), and integrin beta (Itgb)-3 and -6. However, it did not significantly reduce the expression of major fibrogenic genes, including CTGF, PDGF-β, TGF-β1, and Thbs-1 as analyzed by pathway-specific cDNA array (ECM and adhesion molecule array) (SI Appendix, Fig. S1B).

Losartan Treatment Lowers Solid Stress in Ovarian Cancer.

Since reduction in matrix is known to decrease solid stress (9, 25), we next investigated the effects of losartan treatment on solid stress in peritoneal ovarian tumors using the planar-cut technique (11, 25). In size-matched peritoneal SKOV3ip1 tumors, losartan treatment led to significant reduction in solid stress (Fig. 1D) and a trend toward decrease in matrix equilibrium modulus (stiffness) (Fig. 1E) and instantaneous modulus (SI Appendix, Fig. S2B). These results suggest that losartan treatment, via its antifibrotic effects, may be capable of decompressing vessels by reducing solid stress.

Losartan Treatment Improves Perfusion and Relieves Tumor Hypoxia.

To determine whether the decrease in ECM content translated into decompressed vessels and improved vessel perfusion, we measured the fraction of perfused vessels and the level of tumor hypoxia by immunohistology. Losartan treatment did not change VEGF levels or microvessel density (SI Appendix, Fig. S1C) but significantly increased the percentage of perfused blood vessels (Fig. 2A). Following improved vessel perfusion, we found that the hypoxic fraction (evaluated via pimonidazole) of the viable ovarian carcinoma tissue was significantly reduced in losartan-treated tumors (Fig. 2B).

Losartan Treatment Increases Delivery of Chemotherapeutics.

Decompression of blood vessels and improvements in vascular perfusion can enhance the delivery and intratumoral distribution of chemotherapeutic drugs in tumors (8–10). Pegylated liposomal doxorubicin is an FDA-approved treatment for patients with recurrent ovarian cancer which has demonstrated activity in both platinum-sensitive and platinum-resistant disease (26). Doxorubicin is autofluorescent; therefore, we used free doxorubicin as a tracer to study drug delivery in mouse models of ovarian cancer. In the control groups doxorubicin fluorescence signal was detected only proximal to blood vessels, whereas in losartan-treated mice fluorescence signal was broadly distributed throughout the tumor. Quantitative analysis confirmed that losartan treatment significantly increased the amount of intratumoral doxorubicin (red fluorescent signal) (Fig. 2C).

Mathematical Modeling Reproduces Losartan-Enhanced Drug Delivery and Predicts that Combined Losartan Treatment Will Improve Chemotherapy Efficacy.

We have previously combined animal model studies with mathematical modeling to quantitatively predict drug delivery and to provide deeper insight into how the physiological barriers affect drug delivery (27). To further support the robustness of our observation that losartan increases the delivery of chemotherapeutics and enhances their efficacy, we developed a mathematical model (description in SI Appendix, Materials and Methods and Fig. S3). Informed by the experimental data of losartan-induced changes in the ECM content and doxorubicin delivery, our mathematical model reproduced the experimentally observed losartan effects on (i) the reduction of solid stress, (ii) the improvement of vascular perfusion, and (iii) the increase in intratumoral distribution of doxorubicin (Fig. 3 A–D).

In some cases of ovarian cancer, chemotherapy is directly administered i.p. (28). Using our mathematical model, we next investigated whether by reducing the matrix content losartan can directly affect the delivery and homogeneous distribution of a peritoneally administered drug into the peritoneal tumors. Peritoneally administered drugs can reach the tumor both by penetrating the tumor from the periphery and by being absorbed by blood vessels of the peritoneum and reach the tumor through the tumor vasculature. Physiological barriers that can hinder drug transport are (i) the dense tumor ECM that slows the direct diffusion (penetration) of macromolecules into the tumor from the peritoneal fluid, (ii) the solid stress that is accumulated during tumor growth and affects tumor blood vessel function by compressing vessels, and (iii) the hydraulic conductivity of the tumor, which is a measure of the tissue resistance to interstitial fluid flow and affects interstitial fluid pressure (22). A low hydraulic conductivity increases the tumor interstitial fluid pressure, and thus macromolecules have to penetrate from the peritoneal cavity into the tumor against a pressure gradient. Our model predicted that losartan treatment (i) reduced interstitial fluid pressure, (ii) increased the intratumoral delivery of oxygen, (iii) improved the intratumoral distribution of the i.p. administered first-line ovarian cancer chemotherapy drug paclitaxel, and (iv) resulted in a lower tumor volume when combined with chemotherapy (Fig. 3 E–H). We further performed a parametric analysis, varying drug’s diffusivity, the elastic modulus of the tumor (which in the model regulates solid stress levels), and the tumor hydraulic conductivity to get insights about their relative contribution to peritoneal drug delivery. We found that an increase in the diffusivity of the drug or a decrease in tumor elastic modulus/solid stress levels can enhance considerably intratumoral drug distribution compared with the hydraulic conductivity (Fig. 4). The model also predicted that losartan can improve the delivery of both the i.v. administered paclitaxel as well as the i.p. administered doxorubicin (SI Appendix, Fig. S4).

Losartan Treatment Enhances Chemotherapeutic Efficacy in Ovarian Cancer Models.

Based on the mathematical modeling predictions, we next determined whether losartan-induced changes in the ECM and blood vessels would enhance efficacy of i.p. paclitaxel, which is part of the first-line therapy for ovarian cancer, used by both the i.v. and i.p. routes. In both SKOV3ip1 and Hey-A8 models, mice were randomized into four treatment groups receiving (i) control, (ii) losartan, (iii) paclitaxel, or (iv) losartan combined with paclitaxel. In paclitaxel-treated mice, we found that peritoneal tumors were ∼50% smaller compared with those in the control group. Losartan treatment alone did not affect the tumor growth; however, when it was combined with paclitaxel, it significantly enhanced the antitumor effect of i.p. paclitaxel in both SKOV3ip1 (Fig. 5A) and Hey-A8 (Fig. 5D) models. In the SKOV3ip1 model, which develops a significant amount of bloody ascites, losartan treatment significantly reduced the incidence and the amount of ascites (Fig. 5 B and C). In paclitaxel-treated tumors, we found that the number of proliferating tumor cells (PCNA⁺) decreased and the number of apoptotic tumor cells (TUNEL⁺) increased compared with the control group. In the combined treatment group, these changes were even more pronounced compared with the paclitaxel-alone group (Fig. 5 E and F and SI Appendix, Fig. S1 D and E).

Losartan Treatment Decreased Collagen Content and “Normalized” Lymphatic Vessel Networks in the Diaphragm.

Tumors invading the diaphragm collapse local lymphatic vessels, leading to impaired abdominal fluid drainage and accumulation of ascites fluid (29). Although losartan monotherapy did not reduce tumor burden, it significantly decreased collagen content in size-matched diaphragm tumors (Fig. 6 A and B), suggesting a decreased solid stress. To assess effects on the diaphragm lymphatic vessels, we injected a fluorescent tracer (FITC-dextran) i.p. and directly visualized diaphragm lymphatic vessels by lymphangiography (29). In non-tumor-bearing mice, the normal lymphatic vessels displayed a network with organized branching on both the pleural and peritoneal sides of the diaphragm (Fig. 6C). In mice bearing SKOV3ip1 tumors (i) on the pleural side of the diaphragm, an enlarged network of lymphatic vessels was seen, indicating lymphatic network responding to increased fluid burden in the abdomen and potentially blockade of lymph flow, and (ii) on the peritoneal side the classical structure of lymphatic strips was completely disrupted. Strikingly, in mice treated with losartan, on both pleural and peritoneal sides of the diaphragm we found that the lymphatic vessel morphology was closer to that in normal non-tumor-bearing mice. The effects of losartan treatment on decreasing lymphatic vessel diameter on the pleural side were confirmed by image quantification (Fig. 6D). These lymphatic vessel morphological changes were further confirmed using double immunofluorescent staining for LYVE-1 and CD31 in whole-mount diaphragms (Fig. 6E).

Losartan Treatment Improves Lymphatic Vessel Function.

Next, we used two tests to study the lymphatic drainage function. First, we injected fluorescent beads (1 μm in diameter) i.p. Two hours postinjection (i) in non-tumor-bearing mice very few fluorescent beads were found in the diaphragm, indicating clearance by functional lymphatics, (ii) in mice with SKOV3ip1 tumors a significant amount of fluorescent beads were retained in the diaphragm, suggesting impaired lymphatic drainage, and (iii) in losartan-treated mice some beads were retained within the lymphatic vessels, but significantly less than in control tumors, indicating improved lymphatic drainage (Fig. 7A).

Second, as diaphragm lymphatic vessels drain into the caudal mediastinal lymph nodes (CMLN), we collected the CMLNs and evaluated the amount of fluorescent beads drained to CMLN. Compared with non-tumor-bearing mice with normal drainage, CMLNs from SKOV3ip1 tumor-bearing mice accumulated fewer fluorescent beads and showed lower fluorescence intensity, indicating decreased drainage. CMLNs from losartan-treated mice showed higher fluorescence intensity, closer to the level in normal non-tumor-bearing mice (Fig. 7 B and C), indicating losartan treatment improved drainage to lymph node.

Losartan Treatment Induces the Expression of Antifibrotic miRNAs.

Because miRNAs have emerged as major regulators of fibrosis in several fibrotic diseases (30), we used an miRNA array to evaluate how losartan altered the miRNA expression profile in ovarian cancer models. We found that losartan treatment significantly up-regulated the expression of miR-1-3p, miR-133a-3p (miR-133), miR-29b, and miR-26b-5p and down-regulated the expression of seven other miRNAs (Fig. 8A). As miRNAs function by binding to the target sites in the 3′UTR in target genes to repress their expression, we studied the miRNAs that were up-regulated by losartan treatment, to investigate if they contribute to the losartan-mediated reduction of matrix molecules in ovarian cancer models.

We screened the potential targets of these miRNAs using computational target-predicting software (www.microrna.org/microrna/home.do and www.targetscan.org/vert_72/). We found that miR-133 potentially targets collagen IA1 (COL1A1), collagen VA3 (COL5A3), and collagen VIA3 (COL6A3) genes, and its binding sequence is conserved across species (Fig. 8B), suggesting functional importance of the binding site. miR-1 is clustered on the same chromosomal locus as miR-133 (18q11.2); although it is up-regulated by losartan, it has no binding sites on any of the matrix genes. miR-29 is a master regulator of fibrosis that has been shown to target at least 20 different ECM-related genes, including collagens, laminins, and integrins in RAS-mediated hypertensive cardiovascular diseases and renal disease (31–33). miR-29 is down-regulated in fibrotic conditions in multiple organs, including the heart, liver, kidney, and skin (32–34), and the fibrogenic TGF-β/Smad3 axis has been shown to suppress miR-29 expression (33, 35). miR-26 has no binding sites on any of the matrix genes. Thus, as miR-133 is most significantly induced by losartan treatment (>500-fold) in ovarian cancer model, we focused our study on miR-133.

Losartan Treatment Up-Regulates miR-133 Expression, Which Reduces Collagen I Levels in Human Ovarian Cancer Cells.

Among the target genes of miR-133, the collagen I expression was significantly reduced by losartan treatment (SI Appendix, Fig. S1B). The Col I gene encodes the two pro-a1(I) chains in type I collagen, which is the most abundant form of matrix molecule present in the tumor ECM. First, we confirmed that losartan increased the miR-133 level in SKOV3ip1 and Hey-A8 ovarian cancer cells, as well as in mouse fibroblast (10T1/2) and macrophage (Raw264.7) cell lines in vitro by TaqMan assay (SI Appendix, Fig. S5A). Next, to study if miR-133 reduces collagen level, we cloned pri-miR-133 by amplifying a 500-bp fragment spanning miR-133 stem–loop sequence from human genomic DNA. SKOV3ip1 cells were transduced to overexpress pri-miR-133 (529-fold compared with nontransfected cells). While overexpression of miR-133 did not change Col1A1 mRNA levels (0.9 ± 0.2-fold compared with control), it significantly decreased collagen I protein levels (Fig. 8C).

To confirm that miR-133 directly binds to the predicted target site in the collagen I gene and led to its down-regulation, we cloned the predicted miR-133 target site from the ColIA1 gene and inserted it into the 3′UTR of the firefly luciferase gene (pmiRGLO-ColI-wt; Fig. 8D). When miR-133 directly binds to the target sequence, it leads to mRNA destabilization or translational repression, resulting in reduced expression of firefly luciferase protein and low chemiluminescent signal. To confirm whether the reduction of luciferase activity is a direct consequence of miR-133 binding to the target sequence in the ColIA1 gene, we created a mutated target site, which causes decreased complementarity to miR-133 (pmiRGLO-ColI-mut; Fig. 8D). Parental, mock, and miR-133–overexpressing SKOV3ip1 cells were transiently transfected with these wild-type and mutated luciferase reporters and the Renilla luciferase reporter gene (as transfection efficiency control). In parental and mock-transfected SKOV3ip1 cells, the low level of endogenous miR-133 did not affect the luciferase activity. In cells that overexpress miR-133 (SKOV-miR133), large amounts of miR-133 presumably bind to the cloned wild-type ColIA1 target site, leading to significant reduction of the luciferase activity. Mutation of the target site abolished the miR-133–mediated repression of luciferase activity (Fig. 8E), confirming that miR-133 directly targets ColIA1 gene.

Finally, to study the biological role of miR-133 in vivo, we implanted (i) parental, (ii) mock, and (iii) pri-miR-133–overexpressing SKOV3ip1 cells into the peritoneal cavity of nude mice. miR-133 overexpression did not affect tumor growth but slightly decreased ascites (but did not reach statistical significance; SI Appendix, Fig. S5B). Histological analysis of the tumors revealed that miR-133 overexpression did not significantly change the collagen I content (15.4 ± 4.6% in SKOV3ip1 vs. 12.6 ± 2.8% in SKOV-miR133 tumors) (SI Appendix, Fig. S5C) or intratumoral doxorubicin distribution (24.4 ± 3.7% in SKOV3ip1 vs. 20.6 ± 4.4% in SKOV-miR133 tumors) (SI Appendix, Fig. S5D).

Use of an ACEi or ARB Is Associated with Increased Survival in Patients with Ovarian Cancer.

We next sought to test the hypothesis that angiotensin pathway modulation would improve survival in patients with advanced-stage ovarian cancer concomitantly receiving standard of care. We performed a retrospective analysis of patients with stage IIIC or IV ovarian cancer treated at Massachusetts General Hospital (MGH) and Brigham and Women’s Hospital (BWH) between January 1, 2010 and December 31, 2014. We used inverse probability of treatment weighting based on propensity score to create a weighted cohort of patients who differed with respect to blood pressure medication use (ACEi/ARB versus betablocker/calcium channel blocker/diuretic) but were similar with respect to prognostic factors (SI Appendix, Table S1) such as age (P = 0.88), comorbidity index (P = 0.95), histology (P = 0.92), treatment approach (P = 0.97), and residual disease status (P = 1). In the weighted cohort, compared with patients using betablockers, calcium channel blockers, or diuretics, use of ACEi/ARB was associated with a significant reduction in hazard of death (hazard ratio 0.55, 95% CI interval 0.36–0.95; P = 0.004). Women taking an ACEi/ARB had a median survival of 63 mo compared with 33 mo among women taking another type of blood pressure medication (Fig. 9A). The robustness of the main analysis was assessed in several sensitivity analyses. To ensure that the main effect was not due to the survival effects from other antihypertensive medications, the main analysis was repeated after excluding patients using each of the following categories of antihypertensive: betablockers, calcium channel blockers, or diuretics (SI Appendix, Table S1). Furthermore, we assessed whether the effect of angiotensin blockade was evident among patients taking ACEi or ARB medications (SI Appendix, Table S2). Finally, we evaluated whether survival differed between patients using ACE or ARB medications. As shown in Fig. 9B, treatment with ARB was superior to ACEi treatment, consistent with data from our preclinical mouse models with losartan (9).