Clopidogrel in a combined therapy with anticancer drugs—effect on tumor growth, metastasis, and treatment toxicity: Studies in animal models

Clopidogrel, a thienopyridine derivative with antiplatelet activity, is widely prescribed for patients with cardiovascular diseases. In addition to antiplatelet activity, antiplatelet agents possess anticancer and antimetastatic properties. Contrary to this, results of some studies have suggested that the use of clopidogrel and other thienopyridines accelerates the progression of breast, colorectal, and prostate cancer. Therefore, in this study, we aimed to evaluate the efficacy of clopidogrel and various anticancer agents as a combined treatment using mouse models of breast, colorectal, and prostate cancer. Metastatic dissemination, selected parameters of platelet morphology and biochemistry, as well as angiogenesis were assessed. In addition, body weight, blood morphology, and biochemistry were evaluated to test toxicity of the studied compounds. According to the results, clopidogrel increased antitumor and/or antimetastatic activity of chemotherapeutics such as 5-fluorouracil, cyclophosphamide, and mitoxantrone, whereas it decreased the anticancer activity of doxorubicin, cisplatin, and tamoxifen. The mechanisms of such divergent activities may be based on the modulation of tumor vasculature via factors, such as transforming growth factor β1 released from platelets. Moreover, clopidogrel increased the toxicity of docetaxel and protected against mitoxantrone-induced toxicity, which may be due to the modulation of hepatic enzymes and protection of the vasculature, respectively. These results demonstrate that antiplatelet agents can be useful but also dangerous in anticancer treatment and therefore use of thienopyridines in patients undergoing chemotherapy should be carefully evaluated.


Introduction
In cancer, a number of interactions between the activated vascular endothelial cells, cancer cells, and platelets might cause an imbalance in hemostasis. Platelets can be activated by cancer cells in a process called tumor cell-induced platelet aggregation (TCIP). Studies  ). Animal care staff performed the monitoring of animals daily: the changes in behavior, posture, and appearance. Mice were humanely euthanized by cervical dislocation if their body weights dropped more than 20% of the original weight for two consecutive measurements, if more than 30% of the original weight remained for one measurement, if their tumor grew larger than 2000 mm 3 or were necrotic, or if mice were in a state of prostration. Remaining animals were euthanized as indicated in the Table 1 on the day of termination of the experiment. The summary of the number of animals used and euthanized prior to the day of termination of experiments is shown in appropriate figure legends. No animals were found dead in this study. Data on tumor growth and body weights during experiment course for individual mice are included in Supporting Information S1 File or in suitable figures.
All culture media were supplemented with 100 U/mL penicillin and 100 μg/mL streptomycin (both from Polfa Tarchomin S.A. Warsaw, Poland). Cell cultures were maintained at 37˚C in a humidified atmosphere with 5% CO 2 .

Experimental design
Cell transplantation. For an orthotopic model of mammary gland cancer, 4T1 cells were suspended in Hank's Balanced Salt Solution (HBSS; IIET, Poland) such that a suspension of 3 × 10 5 4T1 cells in 0.05 mL of HBSS was inoculated into the mammary fat pad of female BALB/c mice.
The MC38/EGFP colon cancer cells derived from in vitro culture were inoculated subcutaneously (s.c.) in the right flank region with 1 × 10 6 cells suspended in 0.2 mL saline per C57BL/ 6 mouse.
For surgical procedures, mice were intraperitoneally injected with ketamine at a dose of 50 mg/kg (VET-AGRO Sp. z o.o., Poland) and then anesthetized with the mixture of air and isoflurane (3% v/v). For an orthotopic model of prostate tumor, 1.0 cm wide abdominal wall incision was made just above the bladder of male BALB/c nude mice, and the prostate gland was exposed for the injection. Then, 5 × 10 6 PC-3M-luc2 cells in 0.05 mL of HBSS were inoculated into the dorsal prostate lobes of mice. Immediately after transplantation, incised abdominal wall and skin were sewed with soluble surgical suture (Dexon-"S," Polfa, Poznań, Poland). For the model of colon cancer metastasis, HT-29-luc2 colon cancer cell suspension (4.5 × 10 6 / 50 μL) was inoculated intrasplenically (i.s.) into female BALB/c nude mice. In brief, anesthetized mouse was placed on a wooden board in the right lateral position, and the abdomen wall was incised at the left subcostal region. The lower pole of the spleen was exposed, and 0.33 × 13 mm needle was inserted at the upper splenic pole. After implantation, the peritoneum and the abdominal wall were sutured with soluble surgical sutures (Dexon-"S," Polfa, Poznań, Poland). Drug administration. Clopidogrel at a dose of 10 mg/kg/day was administered in sterile drinking water. Prior to treatment initiation, average daily water intake and mouse body weight were determined. The concentration of clopidogrel solution given to mice was calculated according to the formulas With an exception of DTX, all anticancer drugs were prepared in sterile water for injection. DTX was diluted in a mixture of cremophor (Sigma-Aldrich Hemie GmbH, Steinheim, Germany) and 96% ethanol (0.36:0.64 mL, respectively). The resulting 10× concentrated stock solution was further diluted in saline and immediately injected (0.9% NaCl, IIET, Wroclaw, Poland). All anticancer drugs were administered intraperitoneally in an injection volume of 10 μL/g of body weight and according to the dosages and regimens summarized in Table 1.

Estimation of the antitumor activity
When subcutaneous (MC38/EGFP) or mammary gland (4T1) tumors became palpable, their maximum length and width were measured thrice a week, and the tumor volume was calculated according to the formula: where TV is the tumor volume, a is the shorter diameter, and b is the longer diameter.
In vivo visualizations of PC-3M-luc2 or HT-29-luc2 tumors growing in prostate gland or spleen of BALB/c nude mice were performed using an In-vivo MS FX PRO system (Carestream Health INC., USA), no more often than every 4 days starting from the 15th or 13th day of the experiment, respectively. In brief, about 10 min before imaging, D-luciferin potassium salt (Synchem INC., Germany) was intraperitoneally administered to each mouse at a dose of 150 mg/kg. Then, animals were anesthetized with a 3-5% (v/v) mixture of isoflurane (Forane, Abbott Laboratories, USA) in synthetic air (200 mL/min). Anesthesia was maintained with 1.5-2% (v/v) mixture of isoflurane and synthetic air delivered via individual masks. Visualization was performed using the following settings: for X-ray-t = 2 min, f-stop = 5.57, FOV = 198.6; for luminescence capture-t = 3 min, binning 2×2, f-stop = 5.57, FOV = 198.6. Images were analyzed with Carestream MI SE software (Carestream Health INC., USA). The intensity of the luminescent signal is presented as the sum intensity of the region of interest and expressed in arbitrary units (a.u.).
Tumor tissue was also excised and weighed on the last day of the experiment ( Table 1).

Evaluation of the antimetastatic activity
To determine the antimetastatic activity of the compounds in 4T1 mammary gland model, the lungs of tumor-bearing mice were excised, weighed, and transferred into 5% solution of buffered formalin. After tissue fixation, metastatic foci were visually counted. To detect metastases in the mice with prostate cancer, their livers, lungs, kidneys, bones and axillary, as well as inguinal lymph nodes were isolated and fixed in buffered formalin on the day of the necropsy. All isolated tissues were paraffin embedded and cut into 4 μm thick sections that were later dewaxed with xylene and rehydrated in a gradient of ethanol. Finally, sections were washed in distilled water and cytoplasm was stained with eosin while nuclei were counterstained in hematoxylin. Such preparations were next dehydrated in an alcohol gradient and coverslip mounted. The number of metastases in isolated tissues was counted at 50× or 400× magnitude.

Evaluation of tumor angiogenesis
Tumor angiogenesis was analyzed by imaging of tumor blood vessel permeability or immunohistochemistry staining of platelet endothelial cell adhesion molecule (PECAM-1; CD31). For imaging, an IRDye 800CW PEG Contrast Agent (LI-COR, Lincoln, USA) was prepared according to the manufacturer's instructions. The contrast agent was injected intravenously (i. v.) at a dose of 1 nmol/100 μL/mouse. After 24 h, animals were anesthetized with the 3-5% (v/ v) mixture of isoflurane (Forane, Abbott Laboratories, USA) in synthetic air (200 mL/min) and placed on an animal handling station equipped with an individual mask providing with a 1.5-2% (v/v) mixture of isoflurane and synthetic air (In Vivo Ms FX Pro, Carestream Health INC., USA). Visualization was performed using the following settings: for X-Ray t = 2 min, fstop = 5.57, FOV = 198.6; for fluorescence capture t = 30 s, f-stop = 2.8, FOV = 141 (induction: 760 nm, emission: 830 nm). Images were analyzed with Carestream MI SE software (Carestream Health INC., USA). The recorded fluorescence signal is presented as the sum intensity for each region of interest and expressed in a.u.
Immunohistochemistry was performed in tumors isolated from mice on the last day of experiment. In brief, tissues were fixed in buffered formalin and then cut into 4 μm thick sections that were later dewaxed with xylene and rehydrated in a gradient of ethanol. For antigen retrieval, sections were heated in a water bath at 96˚C for 20 min with EnVision™ FLEX Target Retrieval Solution, High pH (Dako 1 , Carpinteria, USA). Endogenous peroxidase activity was quenched in EnVision™ FLEX Peroxidase-Blocking Reagent for 5 min. Thereafter, sections were incubated with primary antibody against PECAM-1 (diluted in 1:50 ratio) and washed in EnVision™ FLEX Wash Buffer. Then, EnVision™ FLEX/HR SM802 detection reagent was added for 30 min at room temperature. The reaction was developed using 3,3-diaminobenzidine tetrahydrochloride solution (DAB, EnVision™ FLEX DAB+Chromogen (DAKO 1 , Carpinteria, USA). Finally, sections were washed in distilled water and nuclei were counterstained in hematoxylin. Then, preparations were dehydrated in an alcohol gradient and coverslip mounted. Mean vessel density was evaluated by counting PECAM-1-positive vessels in five different fields of view at 200× magnitude.

Platelet activation status
Blood samples were collected during necropsy into tubes containing 0.05 mL of 5% ethylenediaminetetraacetic acid (EDTA) solution (Sigma-Aldrich, Germany). Platelet morphology was analyzed using Mythic 18 analyzer (C2 Diagnostics, France). Blood plasma was obtained by centrifugation (2000 × g, 15 min, 4˚C) and stored at −80˚C until further analysis. Prostacyclin generation was determined by the quantification of plasma 6-keto-prostaglandin F1α (6-keto-PGF1α) levels. Based on thromboxane B 2 (TXB2), von Willebrand factor (vWF), and soluble P-selectin plasma concentrations, platelet activation status was estimated. Using commercial kits available from Cusabio Biotech Co. Ltd. (Wuhan, China), all analyzes were conducted by using ELISA technique. In addition, plasma concentration of TGF-β1 was determined with ELISA kit from Boster Biological Technology (USA). All ELISA-based analyzes were conducted according to the manufacturer's instructions.

Protein expression in tumor tissue
Protein expression in prostate tumor tissue was analyzed according to the standard western blot procedure [25]. In brief, tumor tissue samples were collected and immediately frozen on the last day of the experiment. The samples were homogenized in RIPA buffer (Sigma-Aldrich, Germany) using a FastPrep 1 -24 MP Bio device (Mp Biomedicals LLC., USA) with the following settings: CP 24 × 2, 6 m/s, 40 s. Protein content in all samples was analyzed using a Bio-Rad Protein Assay (Bio-Rad Laboratories Inc., USA) according to the manufacturer's protocol. Samples containing 100 μg of protein were separated on the pre-cast 4-20% gradient gels (Bio-Rad Laboratories, Inc., USA) and transferred onto 0.45 μm polyvinylidene fluoride (PVDF) membranes (Merck Millipore, USA). Next, the membranes were probed with primary rabbit polyclonal anti-E-cadherin (1:1000) and anti-N-cadherin (1:1000) antibodies (all from Proteintech Group, USA) or mouse anti-β-actin (1:1000, Sigma-Aldrich, Germany) antibody. Finally, the analyzed proteins were detected with IRDye 1 800CW Goat anti-Rabbit IgG or IRDye 1 680RD Donkey anti-Mouse IgG (both from LI-COR, USA) according to the manufacturer's instructions. Blots were visualized in ODDYSEY 1 CLx Imager (LI-COR, USA) and analyzed with ImageJ Software as follows. The total E-cadherin cellular content comprising truncated and unprocessed E-cadherin (with a molecular weight of approximately 100 and 130 kDa, respectively) was calculated. Similarly, total cellular content of N-cadherin comprising mature and unprocessed N-cadherin (with a molecular weight of approximately 70 and 100 kDa, respectively) was determined. Then, E-cadherin and N-cadherin contents were normalized to β-actin. Finally, E-cadherin to N-cadherin ratios in individual samples were calculated and presented as mean ± standard error of mean (SEM) values.

Toxicity of the anticancer treatment
The toxicity of the proposed anticancer treatment approach and its effect on the overall health condition were estimated based on body weight changes as well as morphological and biochemical analysis. The body weight of experimental animals was measured thrice each week throughout the course of our experiments. The changes in body weight are presented in supplementary materials (S1, S2, S4, S7, S9, S11 and S13 Figs). The health of the mouse was monitored daily. In case of the observed toxicity, animals demonstrating an obvious deterioration in health status were euthanized immediately, and the survival curves were plotted.
Blood morphology was performed with Mythic 18 analyzer (C2 Diagnostics, France). Biochemical analyzes were performed in Cobas C 111 analyzer (Roche Diagnostics, Switzerland) using reagents and procedures provided by the manufacturer.

Statistical analysis
Data normality was estimated using the Shapiro-Wilk test with a predetermined value of p<0.05. The Tukey-Kramer multiple comparisons test for parametric data or the Kruskal-Wallis Test for multiple comparisons for nonparametric data was applied; p values lower than 0.05 were considered significant. All calculations were performed using GraphPad Prism 7 (GraphPad Software, Inc., USA) software.

Results
The effects of clopidogrel in the combination anticancer therapy in the model of 4T1 mouse mammary gland cancer Improvement of 5-FU and CP antitumor and antimetastatic activity by clopidogrel. Clopidogrel itself did not affect either the growth of primary 4T1 tumors or their metastatic potential. But, when the drug was used in combination therapy with 5-FU or CP, significant improvement of the antitumor activity was observed (Figs 1A and 2A). In addition, when clopidogrel was included in the treatment schedule, we observed a limited metastatic spread of 4T1 cells as compared to 5-FU ( Fig 1B) or CP (Fig 2B) alone.
To determine the possible underlying mechanisms of the enhanced anticancer activity of the combined therapy, we performed a number of subsequent analyses. First, vascular permeability of 4T1 primary tumors was estimated based on the intratumoral accumulation of IRDye 800CW PEG Contrast Agent. As shown in Fig 2C and 2D, clopidogrel alone did not affect tumor blood vessel permeability, whereas CP dosed in metronomic schedule tended to increase it.
Then, we measured the expression of E and N-cadherins in tumor tissue. Neither 5-FU, CP, nor clopidogrel, when used alone, affected the expression of any of these proteins ( Fig 1C); however, in mice treated with both clopidogrel and CP or, to a lesser extent with 5-FU, an increase in E:N-cadherin ratio was noticed ( Fig 2E).
Clopidogrel and CP when used alone, as well as in the combination therapy, did not affect the platelet count or any other morphological parameters of platelets (Figs 1D and 2F; data shown only for platelet count in the combined treatment with CP). While 5-FU significantly increased platelet count, mean platelet volume (MPV), and plateletcrit (PCT), it decreased platelet distribution width (PDW; Fig 1D). When clopidogrel was combined with 5-FU, these parameters were changed noticeably, leading to a significant decrease in platelet count as compared to 5-FU, accompanied by the reduced values of other platelet morphological parameters (p<0.05 as compared to control group but not to 5-FU) ( Fig 1D).
Analysis of the selected biochemical parameters related to platelet activity demonstrated that upon single-drug therapy with clopidogrel, plasma level of TGF-β1 significantly decreased only in case of 4T1 tumor-bearing mice. But, plasma levels of TGF-β1 were significantly increased in mice characterized by the lowest number of metastatic foci in animals treated with clopidogrel combined with 5-FU. TGF-β1 levels were increased significantly as compared to clopidogrel and 5-FU administered alone ( Fig 1E). When used as a single-drug therapy as well as when combined with clopidogrel, 5-FU (but not CP) decreased plasma level of von Willebrant factor (vWF) (Figs 1E and 2G). Thromboxane B2 (TXB2) level in mice treated with clopidogrel alone or when combined with 5-FU remained unchanged ( Fig 1F), but in mice treated with CP alone, the level of TXB2 tended to increase. In mice treated with clopidogrel and CP, plasma concentration of TXB2 was found to be the lowest (p<0.05 vs. CP and clopidogrel alone) ( Fig 2G). Finally, P-selectin concentration in plasma of mice treated with either clopidogrel or 5-FU given alone or in the combination remained at a level similar to those observed for control tumor-bearing mice ( Fig 1F).
Body weight of animals transitorily decreased in animals treated with 5-FU or CP administered alone as well as in the combination with clopidogrel. Clopidogrel alone did not affect the body weight of mice (S1A and S2A Figs).
We also observed that in mice treated with clopidogrel combined with 5-FU, the total leukocyte count significantly increased (S1B Fig) primarily due to an increase in the number of lymphocytes and monocytes. The leukocyte count was also found to be significantly increased in animals treated with clopidogrel as a single-drug therapy or when combined with 5-FU In case of animals treated with clopidogrel combined with both 5-FU and CP, we found a decrease in erythrocyte cell count (p<0.05, S1C and S2C Figs). In addition, other morphological parameters of erythrocytes such as mean corpuscular hemoglobin (MCH) and mean corpuscular hemoglobin concentration (MCHC) were found to be significantly decreased in animals treated with clopidogrel, 5-FU, and CP alone or when used as a combination therapy. Moreover, 5-FU used alone or in combination with clopidogrel decreased the level of hemoglobin and increased mean corpuscular volume of erythrocytes (MCV) significantly (S1E Fig To further analyze the potential toxic effects of the proposed therapeutic regimen, we measured blood biochemical parameters. For the combined treatment, we did not observe any general signs of toxicity such as loss of body weight. However, we noticed an increase in plasma lactate dehydrogenase (LDH) level as compared to control or CP-treated mice ( Fig  3B). Such effect was not observed in the combined treatment with 5-FU ( Fig 3A). Other parameters, including ASP, ALT, albumin, creatinine, urea, and the weight of selected organs collected during necropsy did not change significantly (S3 Fig).
Deterioration of CDDP, DOX, and TMX anticancer activity by clopidogrel. When 4T1 tumor-bearing mice were treated with clopidogrel in combination with DOX, CDDP, or TMX, there was an overall disadvantageous effect. For example, simultaneous administration of DOX and clopidogrel resulted in the loss of significant antitumor and antimetastatic activity of the cytostatic drug. Therefore, we observed an insignificant tumor growth inhibition ( Fig  4A) and an increased number of lung metastases ( Fig 4B) (p<0.05) when compared to DOX alone. In case of both CDDP and TMX, their effect on primary tumor growth was not affected, but their antimetastatic activity was diminished when the drugs were administered with clopidogrel ( Fig 4B and 4C).
Treatment with DOX (either alone or in combination therapy) did not affect platelet count, but in combination therapy, DOX and clopidogrel decreased PCT value that led to a significant increase in PDW ( Fig 5A). In mice treated with CDDP alone or when combined with clopidogrel, platelet count, MPV, and PCT were found to be significantly decreased. PDW that was increased by CDDP treatment was found to be significantly decreased in mice treated with both agents as compared to CDDP alone ( Fig 5B). Clopidogrel combined with TMX increased platelet count. Similar to the combined treatment with CDDP or DOX, PDW was found to be significantly decreased in the combined treatment with clopidogrel and TMX. MPV (as in the combined treatment with CDDP) was found to be significantly decreased when mice were treated with combined clopidogrel and TMX ( Fig 5C). Plasma TGF-β1 levels from clopidogrel-treated mice was found to be unchanged or decreased. Among the anticancer drugs tested, only CDDP significantly increased the plasma concentration of TGF-β1. In the combined treatment with DOX, clopidogrel significantly increased the level of TGF-β1; individual animals with median line; the exception is panel D: mean with standard error of mean (SEM) and kinetics of tumor growth: mean with standard deviation (SD) is presented. N = 10 mice per group; some tests were performed on tissue or plasma from selected animals from each group (at least 3 in western blot tests; data for individual blots presented in S1 Table). All mice were euthanized on day 26  with standard error of mean (SEM) and kinetics of tumor growth: mean with standard deviation (SD) is presented. N = 10 mice per group; some tests were performed on tissue or plasma from selected animals from each group (at least 3 in western blot tests; data for individual blots presented in S1 Table). All mice were euthanized on day 31. Statistical analysis: Kruskal-Wallis test for multiple comparisons; *p<0.05.
https://doi.org/10.1371/journal.pone.0188740.g002 Anticancer activity and toxicity of clopidogrel combined with anticancer drugs Anticancer activity and toxicity of clopidogrel combined with anticancer drugs however, when combined with CDDP, it significantly decreased the level of TGF-β1 when compared to CDDP alone ( Fig 5E).
Combination therapy did not affect body weight of mice significantly as compared to the treatment with DOX, CDDP, or TMX alone (S4A Fig).
In mice treated with clopidogrel and DOX, we observed a significantly higher number of lymphocytes, monocytes, and granulocytes (p<0.05 vs. control and DOX-treated animals). This effect was also observed in mice treated with TMX. On the contrary, in mice treated with combination therapy with CDDP, we found the lowest total number of leukocytes resulting from the decreased number of all three leukocyte populations (S4B-S4E Fig).
DOX alone decreased erythrocyte count significantly, and this effect was found to be reversed in the combination therapy (S5A Fig). Similarly, the levels of hemoglobin, hematocrit, MCV, and MCH were found to be elevated in DOX-treated animals, whereas these parameters were found to be decreased when combined with clopidogrel. In addition, clopidogrel admin- LDH activity was either unchanged or increased in the plasma of mice treated with clopidogrel alone (Fig 3C-3E). CDDP did not induce any changes in LDH activity in plasma of the treated animals ( Fig  3C). Only in the combined treatment with clopidogrel and TMX, significant decrease in LDH level, as compared to each agent alone, was observed ( Fig 3E). In addition, in mice treated with DOX and TMX, we analyzed the level of creatine kinase (CK). Clopidogrel alone tended to increase the level of CK, and in combined treatment with DOX and with TMX, a significant increase in the levels of CK was observed (Fig 3F and 3G).
Clopidogrel in combination with MTX and DTX in orthotopic murine model of human prostate cancer Protection against MTX-induced toxicity. The growth of orthotopically transplanted PC-3M-luc2 tumors was significantly inhibited in mice treated with MTX (Fig 6A-6C). This high chemotherapeutic effect was accompanied by an increased mortality of the treated animals (Fig 6E). In the combined therapy with clopidogrel, not only the antitumor activity decreased but also the toxic effect of MTX decreased. Platelets' morphological parameters did not change because of therapy. The tumor progression of mice inoculated with PC-3M-luc2 was accompanied by the loss in body weight that was further enhanced by the treatment with MTX ( S7A Fig). Mice treated with both clopidogrel and MTX had significantly lower total leukocyte and lymphocyte count (S7B and S7D Fig), whereas erythrocyte count, hemoglobin, hematocrit, and MCV were not found to be affected by the treatment. Clopidogrel increased MCH and MCHC significantly. RDW was found to be increased in mice treated with combination therapy (S7C and S7E Fig).
In mice treated with clopidogrel and MTX, we found a significant decrease in the weight of liver, kidneys, spleen, and lungs (S7F Fig). We also performed the experiment with low dose of MTX (1 mg/kg/dose thrice every week). In this experiment, while we did not observe any symptoms of treatment-related Anticancer activity and toxicity of clopidogrel combined with anticancer drugs toxicity, we also did not observe any antitumor activity of the drug. But, MTX either alone or in combination with clopidogrel, decreased the number of organs with diagnosed metastases significantly (Fig 6F and 6G).
We also analyzed the effect of treatment on tumor angiogenesis; however, we did not observe any significant changes in microvessel density (MVD) (S7G Fig). Plasma level of TGF-β1 was found to be decreased by clopidogrel treatment, but MTX alone and in combination with clopidogrel did not change the level of TGF-β1 significantly (Fig 6H). In this study, MTX was found to increase only urea level significantly (among tested LDH, CK, ALT, AST, creatinine, and urea; S8 Fig); clopidogrel decreased urea level as compared to MTX (S8F Fig). Moreover, in combination therapy with MTX, we found a significant increase in the level of CK (S8B Fig) and creatinine (S8E Fig) (p<0.05 vs. clopidogrel).
Clopidogrel increases the toxicity of DTX. Combined treatment with clopidogrel and DTX significantly decreased the growth of primary tumors (Fig 7A). Because of an increased toxicity of such combined therapy (as presented in survival curves, Fig 7E), the antimetastatic activity of the proposed therapeutic regimen could not been assessed. Clopidogrel did not affect the metastatic dissemination of PC-3M-luc2 cells, but in mice treated with DTX, metastases were detected in less number of organs (Fig 7F and 7G).
Platelet count and PCT decreased in response to the treatment with DTX ( Fig 8A). MPV was found to be diminished significantly by clopidogrel and was not affected by DTX, whereas PDW was found to be decreased by clopidogrel and increased by DTX (Fig 8A). Plasma concentration of vWF, TXB2, and P-selectin were significantly decreased by DXT. Clopidogrel significantly decreased plasma level of TGF-β1. Prostacyclin metabolite 6-keto-PGF1α was found to be unchanged (Fig 8B).
Combined treatment did not affect the body weight of mice significantly as compared to the treatment with DOX, CDDP, or TMX alone (S4A Fig). Among blood biochemical parameters (LDH, AST, ALT, creatinine, urea, and glucose), only glucose was found to be significantly increased by DTX. The changes in creatinine level were not statistically significant, but the tendency to lower this parameter was observed after treatment with both agents (S10A Fig). A pronounced decrease in the weight of kidneys and lungs were noticed in mice treated with clopidogrel or DTX. Moreover, DTX decreased the weight of spleen significantly. The weight of liver was unaffected by the treatment (S10B Fig). Combination therapy with clopidogrel and 5-FU in the subcutaneous model of mouse colon cancer and murine model of experimental metastasis of human colon cancer Clopidogrel reveals no effect on the efficacy of 5-FU in the subcutaneous model of mouse colon cancer. The volume of MC38/EFGP tumors decreased significantly in response to the treatment with 5-FU. The observed antitumor activity of chemotherapeutic agent was sustained in mice receiving the combined therapy with clopidogrel (Fig 9A and 9B). Accordingly, in the tumor tissue of the treated animals, we observed a decreased ratio of E:N-cadherin level (Fig 9E). Moreover, clopidogrel given alone as well as in combination with 5-FU tended to increase MVD in MC38/EGFP tumors (Fig 9C and 9D). Similar to the results obtained for BALB/c mice bearing 4T1 tumors, platelet count was increased in mice treated with 5-FU; however, clopidogrel did not affect their count. PDW, PCT, and MPV showed analogous changes such as in BALB/c mice (Fig 9F). However, the plasma level of vWF factor remained unchanged, whereas TXB2 level in MC38/EGFP bearing mice increased after clopidogrel treatment and decreased in combined treatment with 5-FU (p>0.05, Fig 9G). Anticancer activity and toxicity of clopidogrel combined with anticancer drugs Neither tumor growth nor the treatment applied affected the body weight of mice bearing MC38/EGFP tumors (S11A Fig). Similarly, MC38/EGFP tumors did not affect lymphocyte count, thus the effect of the proposed combined therapy on leukocyte count was less pronounced than in 4T1-tumor bearing mice. As shown in S11B and S11D Fig, 5-FU increased total leukocyte and granulocyte count (p<0.05). In general, morphological parameters of Anticancer activity and toxicity of clopidogrel combined with anticancer drugs erythrocytes remained similar between all experimental groups, with the only exceptions for hematocrit that was increased by clopidogrel and MCHC decreased by 5-FU (S11C and S11E Fig).
The analysis of plasma biochemistry showed a significant change in urea concentration decreasing upon the treatment with 5-FU alone or in combination with clopidogrel (S12A Fig). In addition, in mice bearing MC38/EGFP tumors, the treatment with clopidogrel or 5-FU alone or in combination did not affect the weight of liver or spleen (S12B Fig). Clopidogrel enhanced antitumor activity of 5-FU in the model of experimental metastasis of human colon cancer. As evidenced by the measurements of luminescent signal originating from the intrasplenically transplanted HT-29-luc2 cells, combination therapy with clopidogrel and 5-FU resulted initially in the decrease in the tumor volume in the treated animals. But, this effect was not as pronounced on the latter days of the study (Fig 10A-10C). The weight of tumors growing within the spleen tissue measured on day 51 after transplantation was significantly lower in mice treated with combined treatment. The number of liver metastases was significantly lower only in mice treated with 5-FU (p<0.05 vs. control, Fig 10D). Furthermore, 5-FU alone or in combination with clopidogrel, significantly increased platelet count and consequently increased PCT, whereas it decreased PDV (Fig 10E). Moreover, 5-FU alone significantly increased the plasma level of TGF-β1, but when combined with clopidogrel, TGF-β1 levels were restored to the control values ( Fig 10F).
We did not observe any adverse effects of the treatment as assessed based on the body weight changes as well as leukocyte morphology (S13A, S13B and S13D Fig). In mice treated with the combination therapy, fewer erythrocytes were observed (S13C Fig). Other RBC parameters such as hemoglobin level, MCHC as well as MCV, MCH, and RDW were decreased as result of treatment with 5-FU alone or when combined with clopidogrel (p<0.05) (S13E Fig). In addition, in mice treated with combination therapy, LDH, ALT, and bilirubin were significantly decreased (S14B Fig).

Discussion
Clopidogrel, a thienopyridine derivative with antiplatelet activity, is widely used in patients with cardiovascular disease. The randomized controlled trial (TRITON-TIMI 38) comparing the efficacy of a novel thienopyridine, prasugrel with clinical outcome of clopidogrel in patients with acute coronary syndrome showed accelerated cancer progression (breast, colorectal, and prostate) in prasugrel and, to a lesser extent, clopidogrel arm [26,27]. However, further large population-based studies on cohorts of patients with colorectal, breast, and prostate cancer did not provide any evidence of an increased risk of cancer-specific mortality among patients using clopidogrel [28]. Other systematic review and meta-analysis studies have shown that thienopyridine exposure is not significantly associated with malignancy [29]. But, some studies suggest that clopidogrel even reduces cancer incidence [30]. Moreover, Sitia et al. in mouse model of chronic hepatitis reported the preventive activity of clopidogrel against hepatocellular carcinoma [31]. These findings were further supported by clinical trial data indicating that patients with hepatitis B virus-related hepatocellular carcinoma that were treated with plateletcrit (PCT). (G) ELISA measurements of plasma proteins connected with platelets activity: von Willebrant factor (vWF) and thromboxane B2 (TXB2). All graphs show values for individual animals with median line; the exception is panel E: mean with standard error of mean (SEM) and B: mean with standard deviation (SD) is presented. All mice were euthanized on day 53. N = 6-9 mice per group; some tests were performed on tissue or plasma from selected animals from each group (at least 3 for western blot; data for individual blots presented in S1 Table)  clopidogrel showed improved recurrence-free survival and overall survival after liver resection [32].
In this study, we focused on three tumor models: breast, colon, and prostate cancer in which the use of thienopyridines was shown to be related with an increased cancer progression [26,27]. Mezouar et al. clearly showed that clopidogrel used alone can significantly inhibit primary tumor growth and metastasis of pancreatic cancer [21]. In the tumor models used in this study, clopidogrel used alone affected neither primary tumor growth nor metastasis or only tended to inhibit both parameters (Table 2). However, interesting observations were shown when clopidogrel was used in the combination therapy with various anticancer agents. As summarized in Table 2, in the combined treatment of 4T1 mouse mammary gland cancer with CP or with 5-FU, an increase in the anticancer and antimetastatic activity of both drugs was observed. Although clopidogrel alone tended to reduce the growth of MC38/EGFP mouse subcutaneous tumors, when the drug was administered simultaneously with 5-FU no improvement of cytostatic activity was observed. But, in human colon cancer model (HT-29-luc2) the growth of tumors transplanted intrasplenically was significantly diminished by the addition of clopidogrel to the treatment with 5-FU.
The observed enhanced antitumor activity of the combined therapy, most pronounced in case of the treatment comprising CP, could be related to the increase in E:N-cadherin ratio observed in tumor tissue. Such enhanced expression of E-cadherin versus N-cadherin may be one of the signs of diminished EMT. EMT is a process known to endow the tumor cells with invasive mesenchymal-like phenotype and to be driven by platelet-derived TGF-β, among others. Acceleration of EMT and direct platelet-tumor cell contact synergistically activate the TGF-β/Smad and NF-κB pathways in cancer cells, resulting in their enhanced invasion and consequently in vivo metastasis [12]. The mechanism of antimetastatic activity observed in the combined treatment with 5-FU in 4T1 cancer model can be explained by the decrease in invasive potential of tumor cells (reflected in the increased E:N-cadherin ratio) and by the decrease in the accumulation of platelets in the tumor [21]. In addition, the decrease in platelet activation, through diminished vWF plasma concentration [33] in mice treated with both 5-FU and clopidogrel can contribute to antimetastatic effect of the combined treatment.
However, in the discussed 4T1 mammary gland cancer model, we also observed some undesirable effects of clopidogrel used in combination treatment with CDDP, TMX, and DOX leading to a significant increase in the formation of lung metastatic foci in the treated mice. This is contrary with the work of Pandey et al. who reported an enhanced antitumor activity of CDDP or CDDP nanoparticles administered with clopidogrel. Other authors also showed that 4T1 cells expressed P2Y12 receptors and when co-incubated in vitro with CDDP and P2Y12 inhibitor, significantly higher cytotoxic response in 4T1 cancer cell line was observed. However, CDDP in these studies increased P2Y12 receptor expression on cancer cells [34], which was found to be disadvantageous, because P2Y12 receptors agonists were shown to stimulate the proliferation of glioma cells [35,36].
According to Pandey et al., higher efficacy of chemotherapy with CDDP was caused by the increased vessel permeability induced by clopidogrel and the resulting improved drug delivery [10]. But, according to our results, the blood vessel permeability in 4T1 tumors in mice treated with clopidogrel was similar to that observed in control tumors. Furthermore, similar permeability was observed for tumors treated with CP and CP combined with clopidogrel. Moreover, MVD estimated by immunohistological staining of CD31 molecule showed that upon treatment with clopidogrel, MVD was found to be increased in MC38/EGFP or contrarily slightly decreased in PC-3M-luc2 tumors. The intratumoral vessels' network might also be built up by tumor cells [37,38]. Such vasculogenic mimicry was shown in small cell lung cancer xenografts to decrease tumor latency, and despite an increased intra-tumor delivery of CDDP, it decreased CDDP efficacy [38]. Moreover, in mice bearing orthotopic 4T1 tumors vascularized by vessels built up by tumor cells that were treated with anticoagulant warfarin, an increased metastatic potential was observed [37]. Therefore, increased leakage of tumor vasculature can lead not only to a better drug penetration but also to the decreased drug efficacy and increased metastatic potential of cancer cells.
However, Yang et al. have shown that clopidogrel protects endothelium against oxidative stress and consequently diminish endothelium dysfunction [39]. Therefore, depending on tumor model, clopidogrel can affect tumor vasculature by influencing platelet activity [7,8,40] or can directly interfere with endothelium by modulating its activation status [39]. These effects, depending on the changes in tumor microenvironment induced by anticancer drug, may be one of the mechanisms of the observed improvement or deterioration of the antitumor activity of antineoplastic agents given in the combination therapy with clopidogrel.
Platelet granules contain a variety of factors, such as VEGF, TGF-β, and so on that are secreted upon platelet activation. Many of these factors have been implicated in various steps of tumor progression and metastasis. Indeed, due to the tumor vasculature leakage, platelets may come in direct contact with tumor cells and deposit several of angiogenic factors in situ, thus promoting tumor vascularization [7]. But, our data show that in mice treated with clopidogrel, the level of serum proteins related to platelet activity changed between cancer models and anticancer drugs applied in the combined treatment. Only TGF-β1 plasma concentration was diminished by clopidogrel alone in mice bearing mammary gland as well as prostate cancer. Elevated TGF-β plasma level is proposed as a marker of metastatic breast cancer with poor prognosis [41]. Moreover, TGF-β is known to modulate regulatory T-cell (Treg) differentiation, which suppress antitumor T cell response [42]. TGF-β also promotes the differentiation Table 2. Summarized effects of the combination therapy with clopidogrel and anticancer drugs on tumor growth and metastasis, as well as platelets condition. Drug  Activity of combined treatment as compare to anticancer drug used alone a   Primary tumor size  Metastases  Platelets morphology  Blood biochemical parameters   PLT  PDW  MPV  PCT  vWF  TXB2  TGFβ   4T1 CLO

Tumor model
In case of CLO-as compared to control mice; NT-not tested, NT'-not tested because of animals death; "-an increase (e.g., an increase in number of metastases, tumor size, or platelet count); #-a decrease; -/# or -/"-tendency observed; "" or ##-statistically significant changes;-no effect. *data from experiment with lowered doses of MTX.
Abbreviations: PLT, platelet count; PDW, platelet distribution width; MPV, mean platelet volume; PCT, plateletcrit; vWF, von Willebrand factor; TXB2, thromboxane B2; TGFβ, transforming growth factor β; CLO, clopidogrel; MTX, mitoxantrone. https://doi.org/10.1371/journal.pone.0188740.t002 of tumor-promoting tumor-associated macrophages (M2), and tumor-associated neutrophils (N2) [43,44]. Significant increase in TGF-β1 plasma level was observed in our study in mice bearing 4T1 mammary gland tumors treated with clopidogrel combined with DOX or with 5-FU. It correlates with the significant increase in metastases number observed in mice treated with clopidogrel combined with DOX and can explain such unfavorable effect. But, it has been reported that in women with an early stage, pre-metastatic cancer, elevated levels of TGF-β1 are observed [45]. This is in agreement with our observations on 4T1 tumor-bearing mice that were treated with clopidogrel and 5-FU, which indicates that in these animals, tumor progression was markedly but probably reversibly inhibited. In human colon cancer model, clopidogrel decreased the level of TGF-β1 that was elevated by 5-FU which in turn contributed to the antitumor effect of combined treatment. While analyzing the adverse effects resulting from the use of the combination therapies in mice bearing 4T1 tumors, we only observed the decrease in body weight in the animals treated with the combined therapy with TMX. However, LDH, alanine aminotransaminase (ALT), and aspartate aminotransaminase (AST) levels in these animals decreased suggesting that overall the use of clopidogrel in the combination therapies does not impair treatment safety. Significant increase in treatment toxicity was observed when clopidogrel was used simultaneously with DTX in nude mice bearing prostate tumors. The metabolism of taxanes, such as paclitaxel or DTX is related to the hepatic cytochrome (CYP) P450 3A isoform CYP3A4 [46]. DTX can be metabolized by the CYP3A5 [47], whereas paclitaxel can be metabolized by CYP2C8 [46]. Similarly, clopidogrel is considered to be primarily the substrate for CYP3A4/5 [48], whereas its metabolite acyl-β-D-glucuronide is a potent inhibitor of CYP2C8 [49]. A published case report as well as larger epidemiological studies indicate that clopidogrel is associated with a clinically relevant increased risk of neuropathy in patients treated with paclitaxel [49,50]. In the cited report, authors suggested that clinically used doses of clopidogrel can reduce the CYP2C8-mediated systemic clearance of paclitaxel, leading to an increased risk of paclitaxel toxicity [50]. Since TMX, DTX, and clopidogrel are all CYP3A4 substrates [47,48,51] and are characterized by a mild to substantial toxicity, their concurrent use with clopidogrel may lead to the development of severe toxicities, as observed in our studies. Such hypothesis may be further supported by the results of the studies showing that clopidogrel incubation with CYP3A4 led to the formation of metabolites that are toxic for hepatocytes and can be trapped by glutathione. However, clopidogrel depletes the cellular glutathione pool. Thus, high CYP3A4 activity and low cellular glutathione supplies may be the risk factors for clopidogrel-associated hepatocellular toxicity [52]. Low glutathione level may also increase the toxicity of DTX or other anticancer drugs being substrates of this detoxification system [53]. Likewise, metabolic pathway and toxicity of CP, even when used in low doses, are also related to the liver function [54]. Although we did not observe any clinical signs of an increased toxicity of the combined treatment with clopidogrel and CP, significant increase in LDH level as compared to CP alone was observed in serum of mice treated with both agents, suggesting increased hepatic toxicity of such treatment.
Leukopenia, in addition to toxic neurological effects, is the primary toxic effect of DTX [47]. But, in human neutrophils, granulocytes, and lymphocytes, clopidogrel has been shown to decrease the membrane potential of the inner mitochondrial membrane, increase ROS production, and induce cytochrome-c release into the cytoplasm ultimately leading to apoptosis [55]. Therefore, leukocytes affected by anticancer drugs may be more susceptible to apoptotic process during combined treatment with clopidogrel. Accordingly, in mice bearing PC-3M-luc2 prostate tumors, clopidogrel significantly decreased leukocyte count, which was caused by the reduction of lymphocyte and granulocyte count (summarized in Table 3). Similar tendency after clopidogrel treatment was observed in the same strain of mice (nude mice) bearing HT29-luc2 tumors. Therefore, we can speculate that in mice treated with both clopidogrel and DTX, this effect was further accelerated, contributing to the observed toxic effects of the combined treatment. However, in mice bearing HT-29-luc2 or MC38/EGFP colon tumors that were treated with 5-FU, blood parameters did not change significantly; therefore, the effect of clopidogrel on blood cell count in the combined treatment with 5-FU was also not significant.
Different effects of clopidogrel on leukocyte were observed in 4T1 tumor model, a highly metastatic and highly immunogenic mammary gland cancer. Its rapid progress is followed by leukocytosis caused by the expansion of all leukocyte populations and accompanying splenomegalia [56]. In this tumor model, clopidogrel alone increased the leukocyte count significantly, but its effects in the combined treatment varied according to the anticancer drug used. In the combined treatment of clopidogrel with CP and with CDDP, a decrease in lymphocyte and monocyte count was observed, whereas with DOX, 5-FU, and to a lesser extent with TMX, significant increase in leukocyte was observed as compared to the treatment with anticancer drug alone and/or to control 4T1 tumor-bearing mice. Although leukocytes can contribute to cancer growth and metastasis [57], there was no correlation between the effect on leukocyte and the pro-or antimetastatic effect of the combined treatment applied. In various diseases, the direct interaction of platelets and leukocytes lead to the targeted release of soluble mediators and reciprocal activation of both cell types. Depending on the environment, platelet-leukocyte interplay can boost and dampen inflammatory responses [58]. For example, different effects of aspirin on platelet-inflammatory biomarkers are observed in healthy volunteers when compared to patients with acute ischemic stroke [59].
The activity of the combined treatment is shown as compared to anticancer drug used alone or in the case of CLO-as compared to control mice.
## a in case of MCH and MCHC; "" a in case of hematocrit; NT-not tested, NT'-not tested because of animals death; "-an increase (e.g. an increase in number of metastases, tumor size, or platelet count); #-a decrease; -/# or -/"-tendency observed; "" or ##-statistically significant changes;-no effect. * data from experiment with lowered doses of MTX. In this study with prostate cancer model, clopidogrel was found to protect against the MTX-induced toxicity. MTX and DOX affect the heart by impairing mitochondrial function. Both drugs share similarities in their modes of toxicity as redox-interfering drugs and inducers of energy imbalance [60]. The decreased toxicity of MTX due to clopidogrel may be related to the observed antioxidant properties of clopidogrel in patients with coronary heart disease [61]. Moreover, clopidogrel improves endothelial nitric oxide bioavailability [62] as well as activate defensive mechanisms of endothelium involving NO-and prostacyclin PGI 2 -mediated actions [63]. These activities may protect vascular wall against toxic effects of MTX.

Conclusion
Enhanced platelet activity that accompanies the development of malignant tumor is not only a risk factor for cardiovascular events but also contributes to metastases formation. Therefore, the use of antiplatelet agents in the combined treatment of metastatic cancer may become an important strategy in controlling both the venous thromboembolism and metastasis in patients with cancer. As we have shown in this study, clopidogrel can increase anticancer and antimetastatic activity of some of the known anticancer drugs, such as 5-FU, CP, or MTX, whereas it can decrease the efficacy of the other anticancer drugs, such as DOX, CDDP, and TMX. Moreover, clopidogrel increases the toxicity of DTX, whereas it protects against toxicity of MTX. Observed effects may be related to various aspects of clopidogrel activities including its primary target-platelets and others such as endothelial cells, leukocytes, and cancer cells; however, further studies are needed to exactly explain the role of each one. These observations suggest that antiplatelet agents are an important player in the anticancer treatment, but their use in patients should be carefully reviewed to avoid undesirable drug interactions.