Tranilast-induced stress alleviation in solid tumors improves the efficacy of chemo- and nanotherapeutics in a size-independent manner

Accumulation of mechanical stresses during cancer progression can induce blood and lymphatic vessel compression, creating hypo-perfusion, hypoxia and interstitial hypertension which decrease the efficacy of chemo- and nanotherapies. Stress alleviation treatment has been recently proposed to reduce mechanical stresses in order to decompress tumor vessels and improve perfusion and chemotherapy. However, it remains unclear if it improves the efficacy of nanomedicines, which present numerous advantages over traditional chemotherapeutic drugs. Furthermore, we need to identify safe and well-tolerated pharmaceutical agents that reduce stress levels and may be added to cancer patients’ treatment regimen. Here, we show mathematically and with a series of in vivo experiments that stress alleviation improves the delivery of drugs in a size-independent manner. Importantly, we propose the repurposing of tranilast, a clinically approved anti-fibrotic drug as stress-alleviating agent. Using two orthotopic mammary tumor models, we demonstrate that tranilast reduces mechanical stresses, decreases interstitial fluid pressure (IFP), improves tumor perfusion and significantly enhances the efficacy of different-sized drugs, doxorubicin, Abraxane and Doxil, by suppressing TGFβ signaling and expression of extracellular matrix components. Our findings strongly suggest that repurposing tranilast could be directly used as a promising strategy to enhance, not only chemotherapy, but also the efficacy of cancer nanomedicine.

Animal tumor models and treatment protocols. Orthotopic xenograft breast tumors were generated by implantation of 5 × 10 5 MCF10CA1a cells in 40 μ l of serum-free medium into the mammary fat pad of 6-week old female CD1 nude immunodeficient mice. Orthotopic syngeneic models for murine mammary tumors were generated by implantation of 10 5 4T1 mouse mammary cancer cells in 40 μ l of serum-free medium into the mammary fat pad of 6-week old BALB/c female mice. In both animal tumor models, tranilast (200 mg/kg) was administered orally once a day from day 4 post-implantation. Doxorubicin (2 mg/kg and 5 mg/kg for the MCF10CA1a and 4T1 models, respectively) was administered by intraperitoneal (i.p.) injection from day 11 post-implantation, every 72 hours. Doxil (3-6 mg/kg) and Abraxane (20 mg/kg) were administered by intravenous (i.v.) injection on days 14, 21 and 28 post-implantation 10,28,29,38 . During the course of each experiment, tumor growth was monitored daily and the planar dimensions (x, y) were measured using a digital caliper. Tumor volume was calculated using the volume of an ellipsoid and assuming that the third dimension, z, is equal to x y. All in vivo experiments were conducted in accordance with the animal welfare regulations and guidelines of the Experimental procedure. To measure alterations in the tumor microenvironment, right before the end of the experiment, animals were anesthetized by i.p. injection of Avertin (200 mg/kg) and interstitial fluid pressure was measured using the wick-in-needle technique 23,39,40 . Next, mice were injected intracardially with 100 μ l biotinylated lectin (1 mg/ml, Vector Labs) which was allowed to distribute throughout the body for 7 minutes 29 . Finally, mice were sacrificed via CO 2 inhalation and tumors were excised for measurement of mechanical properties and/or histological analysis.

Fluorescent immunohistochemistry and vessel perfusion histology. MCF10CA1a and 4T1
breast tumors were excised from mice, fixed and embedded in optimal cutting temperature compound (OCT). Transverse 40 μ m-thick tumor sections were produced using the Tissue-Tek Cryo3 (SAKURA) and immunostained with antibodies against collagen I, CD31 and hyaluronan. For Ki67 staining, tumors were fixed in paraformaldehyde and embedded in paraffin before sectioning 35 .
For blood vessel perfusion analysis, mice were slowly injected with 100 μ l of 1 mg/ml biotinylated lycopersicon esculentum lectin (Vector Labs) via intracardiac injection 7 minutes prior to euthanization and tumor removal. Upon excision, tumors were fixed in paraformaldehyde, embedded in OCT and frozen. Transverse 60 μ m-thick tumor sections were produced and stained with an antibody CD31, streptavidin-conjugated and fluorescently-labeled secondary antibodies.
In addition, MCF10A1a tumors were immunostained with an anti-phosphorylated-Smad2 (Ser465/467)/ Smad3 (Ser423/425) antibody and counterstained with anti-β -tubulin. Images from anti-collagen I, anti-CD31, anti-hyaluronan and anti-biotin-stained sections were analysed based on the area fraction of positive staining. To avoid any bias, the analysis was performed automatically using a previously developed in-house code in MATLAB (MathWorks, Inc., Natick, MA, USA) 23 . Five different sections per tumor (from the interior and the periphery) at × 10 magnification were taken and analyzed keeping the analysis settings and thresholds the same for all tumors. More details can be found in the Supplementary Material.

Mechanical testing measurements for calculation of elastic modulus and hydraulic conductivity.
The elastic modulus was calculated using an unconfined compression experimental protocol. Following tumor excision, tumor specimens 3 × 3 × 2 mm (length × width × thickness) were loaded on a high precision mechanical testing system (Instron, 5944, Norwood, MA, USA) and compressed to a final strain of 30% with a strain rate of 0.05 mm/min, the minimum rate the system can apply in order to avoid any transient, poroelastic effects. The elastic modulus was calculated from the slope of the stress-strain curve (details in Supplementary Material).
For the calculation of the hydraulic conductivity, stress relaxation experiments were performed in compression. Specimens underwent four cycles of testing for each of which a 5% compressive strain was applied for 1 minute, followed by a 10 minute hold and the stress vs. time response of the tissue was recorded. Subsequently, a common biphasic model of soft tissue mechanics was employed 41 accounting for both the solid phase (cells and extracellular matrix) and the fluid phase (interstitial fluid) of the tumor. The hydraulic conductivity was calculated by fitting the model to the experimental stress-time data (Fig. S5, details in Supplementary Material).
A description of the mathematical model, IFP measurements, biodistribution analysis, in vitro cell culture experiments, gene expression analysis and statistical analysis can be found in the Supplementary Material and Supplementary Table S1.

Results
Mathematical modeling of stress alleviation mechanism. Initially, we performed mathematical modeling analysis to investigate the effect of stress alleviation strategy via ECM depletion on drug delivery. The purpose of using model simulations was to confirm whether the general principle indicating that stress alleviation can cause size-independent delivery of drugs is mathematically consistent. The tumor was modeled in two dimensions with its vasculature represented by a percolation network consisting of one inlet and one outlet ( Fig. 1A) 10,42 . Each vascular node was assigned a vessel diameter and vessel wall pore size. We solved the equations of fluid flow and nanoparticle transport in the vascular, transvascular and interstitial space of the tumor and calculated blood vessel velocity, IFP and drug delivery. The first effect of stress alleviation is decompression of tumor blood vessels to improve perfusion. Vessel decompression was modeled as an increase in vessel diameter. Interestingly, an unexpected insight of the model is that the increase in the vessel diameter improves the fraction of perfused vessels independently of the vessel wall pore size (Fig. 1B) making this strategy suitable for both moderately as well as highly permeable tumors. Furthermore, depletion of ECM components increases the hydraulic conductivity -the ease with which interstitial fluid percolates through the interstitial space of the tumor 43 (Supplementary Eq. S13) -which in turn reduces IFP (Fig. 1C). Therefore, contrary to vascular normalization, in stress alleviation therapy IFP reduction is achieved without affecting the pores of the tumor vessel wall (i.e., the EPR effect). Improved tumor perfusion and reduced IFP enhances drug transport from the vessels into the tumor interior (Fig. 1D). Finally, the creation of a larger interstitial space available for diffusion, owing to ECM depletion, improves the intratumoral penetration of small (1 nm, 10 nm) and large (100 nm) nanoparticles (Fig. 1E). Importantly, model predictions suggest that stress alleviation improves delivery of drugs of all sizes (Fig. 1D,E). The extent of improved delivery depends on the permeability of the tumor blood vessels but not on the lymphatic function at the tumor periphery (Supplementary Figs S1 and S2). Particularly, for 10 nm drugs and smaller, the benefit of stress alleviation appears to be optimal for tumors with moderately permeable vessels ( Supplementary Fig. S1), such as pancreatic ductal adenocarcinomas, because for these tumors the effect of vessel decompression on perfusion and thus, on the functional vessel density is optimized (Fig. 1B). The delivery of larger particles, however, depends not only on the functionality of the tumor vessels but also on the ability of the particles to cross the tumor vessel wall into the tumor interior. Therefore, vessel decompression is expected to optimize the delivery of the 100 nm particles for tumors with highly permeable vessels ( Supplementary Fig. S1), such as a subset of mammary carcinomas.

Tranilast improves the efficacy of common anticancer drugs in a size-independent manner.
Based on these predictions, we hypothesized that tranilast can improve the efficacy of therapeutics via stress alleviation. To investigate our hypothesis, we developed two orthotopic mouse models for breast cancer; a xenograft model using the human MCF10CA1a cancer cell line and a syngeneic model using the mouse 4T1 cancer cell line and employed three commonly used drugs: doxorubicin, which has a size of less than 1 nm; Abraxane, a 130 nm albumin-bound paclitaxel that shrinks to 10 nm following dilution to plasma 10 ; and Doxil, a ~100 nm pegylated liposomal doxorubicin. Mice received orally mock treatment or tranilast (200 mg/kg) 4 days post-implantation of cancer cells in the mammary fat pad, whereas anti-tumor drug administration begun on day 11. Higher doses of tranilast have been previously associated with direct anti-tumor and anti-vascular effects, whereas lower doses did not cause any effect on ECM composition 36,37,44 . We found that tranilast alone did not influence primary tumor growth ( Fig. 2) or cell proliferation in vivo based on the number of Ki67 positive cells (Supplementary Figure S3). Similarly, doxorubicin (2 mg/kg to 5 mg/kg, i.p.) had no effect on tumor growth in either animal model. However, combinatorial treatment with tranilast and doxorubicin significantly delayed tumor growth ( Fig. 2A,B and Supplementary Fig. S4A,B). Next, we investigated the role of tranilast in the efficacy of nanotherapeutics. Our data indicate that tranilast dramatically improved the efficacy of i.v. administered Abraxane (20 mg/kg) as depicted by the substantial differences in breast tumor growth rates (Fig. 2C, Supplementary Fig. S4C). Similarly, while initial i.v. administration of 6 mg/kg Doxil had a similar effect in control and tranilast-treated mice, lowering the dose (3 mg/kg) resulted in a differential response between the two groups (Fig. 2C). In agreement with model predictions, our data suggest that tranilast improves the efficacy of drugs in a size-independent manner.
Tranilast remodels the tumor microenvironment reducing collagen and hyaluronan levels. To investigate the ability of tranilast to remodel the tumor ECM, we initially performed histological analysis of tumor cryosections. Tranilast treatment decreased the amount of both collagen and hyaluronan, as represented by area fraction quantification (Fig. 3). Specifically, area fractions of collagen (Fig. 3C) were reduced by 20% and 25% and of the hyaluronan (Fig. 3D) by 40% and 63% in the 4T1 and MCF10CA1a tranilast-treated tumors, respectively compared to control tumors (p = 0.025, 4T1; p = 0.024, MCF10CA1a for collagen and p = 0.043, 4T1; p = 0.016, MCF10CA1a for hyaluronan). Similarly, while doxorubicin alone (5 mg/kg) had no effect on growth of 4T1 murine breast tumors, combination of tranilast and doxorubicin significantly reduced tumor growth rate (p = 0.015 on day 20, n = 6-8). (C) Tumor volume growth rates of orthotopic MCF10CA1a breast tumors implanted in CD1 nude mice that were either control-treated (0.9% NaCl for Abraxane, saline for Doxil and NaHCO 3 for tranilast), tranilast-treated alone (200 mg/kg daily via gavage), Doxil alone (6 mg/ kg on day 14 and 3 mg/kg on days 21 and 28 post-implantation, i.v.), Abraxane alone (20 mg/kg on days 14, 21 and 28 post-implantation, i.v.), or tranilast in combination with either Doxil or Abraxane, as described above. Abraxane alone had no effect on tumor volume, compared to administration of tranilast with Abraxane, which significantly delayed tumor growth (p = 0.003 on day 33, n = 7-9). Doxil monotherapy at 6 mg/kg decreased tumor volume whereas at 3 mg/kg it only delayed tumor growth compared to control. On the other hand, combination of 3 mg/kg Doxil and tranilast significantly reduced tumor volume (p = 0.0001 on day 33, n = 7-9) compared to Doxil (3 mg/kg) alone. Asterisks indicate a statistically significant difference between compared groups (p < 0.05).
Scientific RepoRts | 7:46140 | DOI: 10.1038/srep46140 Tranilast alleviates intratumoral solid and fluid stresses. Subsequently, to further explore the potential of tranilast as a stress-alleviating agent, we performed detailed analysis of the mechanical properties of the tumors. We first measured the growth-induced stresses by employing our previously developed technique, the tumor opening experiment 23 . Upon excision, we cut the tumors along their longest axis at approximately 80% of their thickness. We then allowed tumors to relax and measured the formed distance between the two hemispheres (Fig. 4A). Control tumors had significantly larger tumor openings, which corresponds to higher growth-induced solid stress compared to tranilast-treated tumors (Fig. 4B). We also performed ex-vivo stress-strain experiments, which revealed that tranilast decreased the elastic modulus of the tumors, making them less stiff (Fig. 4C,D). Collectively, our data from the tumor opening and stress-strain experiments clearly suggest that tranilast decreases solid stresses (i.e., stresses of the solid phase) in both MCF10CA1a and 4T1 tumor models. In addition, we examined the effect of tranilast on tumor fluid phase features and particularly on the interstitial hydraulic conductivity and fluid pressure. To calculate the hydraulic conductivity, we performed ex vivo stress-relaxation experiments and the data were fitted to a biomechanical mathematical model (Supplementary Fig. S5). Treatment with tranilast increased hydraulic conductivity (Fig. 4E), which was expected due to the reduction in ECM components (Supplementary Eq. S13). According to our mathematical analysis (Fig. 1C), the increase in the hydraulic conductivity should cause alleviation of the IFP. Consistent to our predictions, using the wick-in-needle technique 39 , we found that tranilast significantly decreased IFP in both tumor models (Fig. 4F).

Alleviation of stresses decompresses tumor blood vessels and improves perfusion. To investi-
gate the effect of stress alleviation on the functionality of tumor blood vessels, we calculated the vessel diameter and the percentage of perfused blood vessels after mock or tranilast treatment, respectively. Tissue cryosections were stained with antibodies against CD31 and biotinylated lectin. Based on the ratio of the number of biotinylated lectin positive (+ ) to the CD31 positive (+ ) vessels, the fraction of perfused vessels was calculated. Tranilast-induced stress alleviation resulted in a statistically significant increase in vessel diameter for both tumor models, which in turn caused a significant increase in tumor perfusion (Fig. 5A-D). Interestingly, despite the small increase in vessel diameter by 10-15%, the fraction of perfused vessels increased significantly by 50-60%. This is in full agreement with our previous studies 23,29 and could be explained by the fact that blood flow rate is proportional to the fourth power of the vessel diameter (Supplementary Eq. S1) so that small increases in diameter could improve significantly blood flow. Furthermore, it could be also possible a compressed upstream vessel to exclude from blood flow a large number of downstream vessels, which will become functional when the upstream vessel is decompressed. In contrast, the total area of vessels (CD31 + area) remained unaffected (Fig. 5E), suggesting that tranilast treatment improved perfusion without affecting tumor angiogenesis. Importantly, in vivo biodistribution analysis revealed that improved tumor perfusion via stress alleviation resulted in a significant increase in intratumoral drug delivery without affecting delivery in normal tissues (Fig. 5F).

Tranilast inhibits TGFβ pathway and downregulates COL1A1, CTGF and HAS gene expression.
Finally, based on previous work related to the biological mechanisms underlying tranilast's mode of action 32-34 , we wanted to investigate whether similar effects could also be exerted during tranilast-induced remodeling of the tumor microenvironment. To elucidate these molecular events, we first performed immunofluorescence analysis in MCF10CA1a tumors, which indicated that tranilast suppressed Smad2/3 phosphorylation and nuclear translocation in vivo (Fig. 6A), suggesting that inhibition of TGFβ signaling pathway could be responsible, at least in part, for the observed reduction of extracellular matrix components, stress alleviation and increased drug delivery in tumors. We then quantified the in vivo expression of critical TGFβ target genes that are involved in collagen and hyaluronan synthesis as well as other encoding for extracellular matrix components or collagen crosslinking enzymes. RNA extraction from human MCF10CA1a tumors followed by real-time PCR using human-specific primers showed that tranilast suppressed collagen I (COL1A1), connective tissue growth factor (CTGF), hyaluronan synthase 2 (HAS2), hyaluronan synthase 3 (HAS3), collagen 3 (COL3A1) and lysyl oxidase (LOX) gene expression in breast cancer cells (Fig. 6B). To investigate the effects of tranilast on tumor stromal cells, which include mouse cancer-associated fibroblasts and immune cells, we designed mouse-specific primers and performed real-time PCR analysis. We found that while tranilast did not have a major effect on most genes of the stromal cells, it suppressed periostin (POSTN) and induced hyaluronan synthase 3 expression (Suppl. Fig. S6). To further confirm that these genes are regulated by the TGFβ pathway, we performed in vitro cell culture experiments which showed that while TGFβ is able to activate COL1A1, CTGF, HAS1 and HAS2 gene expression in 4T1 mammary adenocarcinoma cells, pre-treatment with tranilast significantly inhibited TGFβ -mediated upregulation of COL1A1, CTGF and HAS2, but not HAS1 (Supplementary Fig. S7). Finally, since tranilast can directly alter gene expression, we wanted to investigate the possibility whether gene expression changes could also alter the sensitivity of the cancer cells to doxorubicin. Therefore, we performed an in vitro assay to calculate the percentage of viable 4T1 breast cancer cells that were treated either with tranilast alone, doxorubicin alone or combination of tranilast with doxorubicin. These data indicated that while doxorubicin alone significantly reduced breast cancer cell viability, combination of tranilast with doxorubicin did not further decrease the number of viable breast cancer cells (Supplementary Fig. S8). Conclusively, our in vitro and in vivo evidence suggest that tranilast primarily enhances drug delivery and efficacy of doxorubicin by remodeling the tumor microenvironment, whereas it does not seem to significantly sensitize cancer cells to chemotherapy.

Discussion
To conclude, the objective of our study was two-fold. The first objective was to show the potential of stress alleviation treatment to improve the efficacy of chemo-and nano-therapeutics. Stress alleviation is a recently developed therapeutic strategy, which has shown promising results in improving chemotherapy in preclinical tumor models [29][30][31][45][46][47] but there was no mathematical/theoretical basis for the effects of this strategy and also there was no study to explore its potential use for nanomedicines, where compromised drug delivery is a major barrier to their Tumor sections from control-treated or tranilast-mice were stained with anti-phospho-Smad2/3 (red) antibody and counterstained with anti-β -tubulin (green) antibody for detection of the cytoskeleton, indicated that tranilast inhibits Smad2/3 phosphorylation and nuclear translocation. Scale bar: 100 μ m. (B) Quantification of Smad2/3 phosphorylation and nuclear translocation was performed by counting of the p-Smad2/3-stained nuclei (red) and expressed as percentage of red positive nuclei compared to the total number of cells using confocal images from 3 independent experiments. (C) Real-time PCR gene expression analysis and quantification of hCOL1A1, connective tissue growth factor hCTGF, hHAS1, hHAS2, hHAS3, hPOSTN, hTGFβ , hCOL3A1, hCOL4A1, hLOX, hFN1, hVIM and hTSC mRNA levels extracted from control-treated compared to tranilast-treated MCF10CA1a tumors, indicated that tranilast suppresses COL1A1, CTGF, HAS2 and HAS3 gene expression. Relative expression for all genes in both groups was normalized based on the expression of beta-actin. Data represent the average of at least 3 independent experiments from 5 control and 5 tranilast-treated tumors ± S.E. values and asterisks indicate statistically significant differences between compared groups (p < 0.05).
efficacy. Here, we demonstrated both mathematically as well as experimentally that stress alleviation improves the efficacy of common chemotherapeutics and nanomedicines in a size-independent manner. The second objective was to introduce a safe, well-tolerated drug as a stress-alleviating agent. Tranilast is an inexpensive, clinically approved drug that has already been tested in humans for toxicity and tolerance, rendering the results of our study highly transferable to the clinic. Therefore, tranilast-induced stress alleviation combined with common anticancer drugs could directly lead to Phase II clinical trials to test the efficacy of this therapeutic strategy in humans.
Stress alleviation strategy is expected to be beneficial for tumors that have abundant compressed vessels. Vessel compression in turn is caused by mechanical stresses exerted by the cells and the extracellular matrix. Therefore, desmoplastic tumors (e.g. pancreatic and breast cancers as well as sarcomas), which contain high levels of stromal cells and ECM components should have a large amount of compressed vessels. In principle, judicious depletion of either stromal cells or ECM components or both will have the ability to alleviate stresses in order to decompress blood vessels and improve perfusion in these tumors [23][24][25] . Furthermore, as shown here, stress alleviation strategy is beneficial even for tumors with hyper-permeable vessels (e.g. a subset of breast cancers) where perfusion depends not only on vessel diameter but also on vessel permeability. In the case of tumors with compressed and hyper-permeable vessels, stress alleviation strategy could be combined with vascular normalization so that blood vessels will become decompressed and less permeable, which should optimize perfusion and delivery of small-sized drugs (Fig. S1) 48 . The degree of vessel compression and permeability, however, can vary considerably not only among tumor types but also within the same tumor type as well as between the primary tumor and its metastases. Therefore, it is difficult to choose an appropriate strategy until the state of that individual tumor is known. Emerging imaging approaches have the potential to help in this selection 49 .
Furthermore, it is reasonable to argue that improving tumor perfusion could also allow more oxygen and nutrients to be transported in the tumor and thus, increase its growth rate. Additionally, the decompressed vessels could potentially assist metastatic cells to leave the primary tumor and form metastases, which has been shown in pre-clinical studies 50,51 . Therefore, drugs that decompress vessels should only be given with concurrent cytotoxic treatments, such as chemotherapy, nanomedicine, radiation therapy or immune therapy. Specifically, as far as the use of tranilast is concerned, preclinical studies have indicated that it does not enhance metastatic dissemination 36,37,44 .
Finally, even though our mathematical model is general and the mechanism of tranilast action was not explicitly incorporated, this could be feasible through its involvement in the TGFβ pathway and could be used for the investigation of further interactions between TGFβ and tumor perfusion that go beyond the scope of the current study.