Platelets Lead Morphological, Phenotypic and Functional Changes in Tumor Cells.

Background Platelets are active players in tumorigenesis, although the exact interactive mechanisms and their direct impact on tumor cells remain largely unknown. Methods Bidirectional transference of lipids, proteins and RNA between platelets and tumor cells and its impact on tumor cell behavior and tumor process are analyzed in this work. Phenotypic, genetic and functional modications induced by platelets were analyzed both in tumor cell lines and in circulating tumor cells (CTCs). Results Besides cross-talk between platelets and tumor cells, we demonstrated an ecient exchange of lipids, proteins and RNA. Remarkably, we observed that platelets transferred structural components to tumor cells with higher eciency than tumor cells to platelets (p=0.001). This biological interplay occurred by direct contact, internalization or via extracellular vesicles. As a result, tumor cells acquired platelet markers (CD61 and CD42), showed decreased EpCAM, expressed epithelial-to-mesenchymal transition markers, and increased proliferation rates. Analysis of platelets marker CD61 in CTCs showed intra and inter patient heterogeneity. Our results demonstrated, for the rst time, that platelets educate tumor cells by highly ecient transference of lipids, proteins and RNA through different mechanisms. These results suggest that tumor cells and CTCs might acquire highly dynamic and aggressive phenotypes due to platelets interaction including EMT, stem-like phenotype and high proliferative rates.


Introduction
The tumor dissemination process involves release of tumor cells from the primary site to the bloodstream or the lymphatic system, known as circulating tumor cells (CTCs) [1]. Albeit, it is widely accepted that the presence of CTCs in peripheral blood of cancer patients is strongly associated with poor survival outcomes [2], little is known about their biological complexity. Each day millions of tumor cells are released from the tumor site into the blood, but only few of them survive [3]. CTCs are subjected to a combination of physical stress (shear forces) [4], anoikis (a form of cell death that occurs in anchoragedependent cells when they detach from the surrounding extracellular matrix) [5], and are exposed to the immune system activity [6]. CTC survival may depend on their physical and molecular adapting ability which involves, presence of different CTCs subpopulations with markedly distinct characteristics [7].
The immune system has a dual role in cancer progression with both repressive and promoting actions.
Besides their physiological function in homeostasis, platelets have been shown to serve as active players during carcinogenesis, through mechanisms that affect both CTC migration and survival in circulation [9,11,12]. A complex crosstalk between cancer cells and platelets exists and the exact underlying mechanisms are relatively poor understood. Platelets are able to sequester tumor RNAs, turning into tumor-educated platelets (TEPs) [13]. Since TEPs discovery, numerous studies have analyzed the modalities for which tumor cells can modify platelets converting them into a potential predictive and prognostic biomarker in cancer through the evaluation of liquid biopsies [14]. However, few works have focused on investigating the mechanisms by which tumor cells are directly modi ed by platelets [15] and how this interaction affects their structural composition and therefore their phenotype and consequently their functionality [16]. The most widely studied consequence of the interaction between tumor cells and the immune system is the acquisition of mesenchymal phenotypes [17] through induction of the epithelial mesenchymal transition process (EMT) [18,19].
Herein, we explored the tangled bidirectional interactions between platelets and tumor cells, revealing that platelets can actively modify tumor cells phenotype, their genetic content and their functional abilities.

Sample collection
Peripheral blood samples from both healthy donors (with no history of malignant disease) and prostate cancer patients were obtained from Virgen de las Nieves University Hospital (Granada) after approval by the ethical Committee of this Hospital, in accordance with the Declaration of Helsinki. Written informed consent was signed from every cancer patient and healthy volunteer prior sample collection.
Samples were processed in the Liquid Biopsies & Cancer Interception laboratory (LiqBiopCI) at GENYO Centre (Granada). All prostate cancer patients were diagnosed and followed-up in the Urology Department and in the Oncology Department of the University Hospital Virgen de las Nieves (Granada).

Circulating Tumor Cells (CTCs) isolation
Peripheral blood samples (10 ml) from prostate cancer patients diagnosed of localized disease or advanced disease (metastatic castration resistant prostate cancer) were collected in EDTA tubes (Vacutainer), stored at room temperature and processed into 4 hours after collection. CTCs were isolated according to the previously established protocol by our group [20,21]. Brie y, blood samples were subjected to density gradient centrifugation and immunomagnetic selection of epithelial cells using the Carcinoma Cell Enrichment and Detection Kit (Miltenyi Biotec) based on multicytokeratin (CK3-11D5) microbeads. Each sample was spun down onto two slides in a cytocentrifuge (Hettich) and stained for confocal microscopy visualization. , for confocal microscopy analysis as previously described. Negative and single stained controls were performed to ensure no uorescence bleed-through between channels. CTCs were described as CK + nucleus + cells and CD61 expression in CTCs was classi ed into presence or absence.

Platelet isolation
Platelets were isolated from whole blood collected in EDTA tubes (Vacutainer) by a series of centrifugations at room temperature in a swing-bucket rotor centrifuge. First, leukocyte-rich platelet-rich plasma (L-PRP) was obtained by centrifugation at 120 x g for 10 min without break. Then, remaining white blood cells and erythrocytes were removed by centrifugation at 105 x g for 15 min to obtain pure platelet-rich plasma (P-PRP). Platelets were isolated from P-PRP by centrifugation at 1000 x g for 12 min. Isolated platelets were resuspended in RPMI 1640 at physiological concentration. Experiments were performed immediately after platelet isolation.

Platelet activation induction
Platelet activation was induced by incubation with 15µg/mL of Adenosine 5'-diphosphate sodium salt (ADP) (Sigma-Aldrich) and 1U/mL of thrombin from bovine plasma (Sigma-Aldrich) for 30 min. Activated platelets were washed with PBS-EDTA 2mM and transferred to the cell culture.

Cell tracking
Interactions between platelets and cells were evaluated by labeling either cells or isolated platelets using membrane cell tracker ( Most relevant information about Material and Methods was graphically represented in Fig. S1.

Co-culture
For all experiments including tumor cell and platelets co-culture, tumor cells were seeded the day before platelet isolation in order to reach 60-70% of con uence at the time of co-culture. In experiments avoiding direct contact between cells and platelets, 0.4μm membranes Transwell® inserts (Millipore) were placed on well plates and platelets were added onto them.

Flow cytometry
Flow cytometry experiments were performed using 24-wells plates and 500μl of platelets suspension. Experiments were run in triplicates and collected at different time points. After co-culture, platelets were harvested from cell media after centrifugation at 105 x g for 15 min to eliminate cell fragments. Cells were washed twice with PBS 1X to remove any remaining platelets and cell colonies were dissociated with Tryple Express 1X (Life Technologies). Cells and platelets suspensions were xed with 3.7% Between primary and secondary antibody incubations, cells and platelets were washed with FACS Buffer (PBS 1X, 5% FBS, EDTA 2Mm) and PBS-EDTA 2mM, respectively.
Time-Lapse Assay were performed in Glass Bottom 35mm µ-dish (Ibidi). Tumor cells were labeled with DiD cell tracker (red) and platelets with DiO cell tracker (green), as previously described. Platelets were added a 0 time point and set in the incubation chamber of the confocal microscope at 37°C and 5% CO 2 for time lapse monitoring. Five positions were analyzed and a total of 24 images were acquired with 10min/image-time interval for a total duration of 240 min.
Confocal images were obtained using a LSM 710 confocal laser scanning microscope (Carl Zeiss, Jena, Germany) equipped with an incubation chamber (Pecon, Germany

Cell Proliferation Assay
The effect of co-culturing platelets with tumor cells on cell proliferation was evaluated with the real-time cell monitoring assay (RTCA) (xCELLigence; ACEA Biosciences, Inc.). Cells were seeded in the RTCA plates and isolated platelets were added 24h afterwards. Cell growth was monitored from seeding in a 4 hour interval on the RTCA system for 72 hours and impedance was recorded as a measurement of Cell Index (CI). Experiments were performed in quadruplicates and outliers were removed.

Gene Expression
Total RNA from cancer cells cultured alone or in co-culture with resting platelets was extracted with TRIzol TM Reagent (Invitrogen), according to manufacturer's instructions. RNA concentration and purity were determined using NanoDrop 2000c Spectrophotometer (ThermoFisher Scienti c) and 1µg of total RNA was converted to complementary DNA (cDNA) using the Transcriptor First Strand cDNA Synthesis Kit (Roche) for subsequent RNA expression.
qRT-PCR primers previously described elsewhere were used (Sigma-Aldrich), details in Table S1. Gene expression was measured using iTaq™ Universal SYBR® Green Supermix (Biorad) on a 7900 HT Real- were assessed as a continuous (number of CTCs) and CD61 expression was de ned as a dichotomous variable (positive or negative). Dynamics of tumor cell growth after platelets addition were studied using non-linear regression (second order polynomial, quadratic) best-t modeling, moreover tumor cell growth with or without platelets were compared at 24h and 48h. P values less than 0.05 were considered statistically signi cant.
Second, platelets activation in different co-culture conditions with cells (C) was analyzed. We observed a signi cant increase in PAC-1+ platelets in all conditions: unlabeled co-culture (P+C) (p<0.001), labeled coculture independently on the labeled cell type (P+CCT and PCT+C) (p<0.01 and p<0.001, respectively), and when ADP+Thrombin-treated platelets were used (p<0.05) (Fig. S3A).

Communication between platelets and tumor cells through lipid membrane components
Cancer cells were pre-labeled using a lipophilic DiO cell tracker and the percentage of CD42 + platelets acquiring uorescence after 1h and 24h of co-culture is presented as tumor-educated platelets (TEPs) (Fig. 1A). Alternatively, platelets were pre-labeled with DiO cell tracker and the percentage of EpCAM + tumor cells that acquired uorescence is shown as platelets-educated tumor cells (PETs) (Fig. 1B). Although, transference of tumor cell material to platelets has been widely studied, we observed using eleven cell lines from four different tumor types (Fig. 1A) that only 9.69±2.3% and 32.95±6.33%of platelets showed lipid cell tracker at 1h and 24 h, respectively; whereas 26.54±7.64% and 78.77±10.17% of tumor cells presented lipid from platelets. These ndings demonstrated that transference from platelets to cancer cells is signi cantly more e cient than from cell to platelets at 1h (p=0.047) and at 24h (p=0.001). Individual results for each cell line are presented in Fig. 1B, all cell lines excluding SW620 and HCC70 showed higher lipid transference from platelets to cells than from cells to platelets at 24h of co-culture. Lipid transference through transwell membrane was studied in LNCAP cells, no signi cant differences were found (p=0.079) from cell to platelets, suggesting cell contact is not mandatory; however, our results showed a signi cant reduction of platelet-to-cell lipid transference (83.8±5.02%) compared with the co-culture control (98.38±0.56%) (p=0.0068) (Fig. 1B). These results might suggest that transference from platelets to cancer cells is more dependent on cell contact than the one from cells to platelets.

Transference of lipid components from platelets to cells is mediated by different mechanisms
The mechanisms involved in transference from platelets to tumor cells were exhaustively analyzed through confocal and electron microscopy. As visualized, platelets were able to modify tumor cell membrane through direct contact, observing that the tumor cell membrane became green after acquisition of the lipid DiO green marker. In particular, platelets were able to fuse with the tumor cell membrane ( Fig. 2A and C, Fig. S4), to be integrally internalized ( Fig. 2A and 2B) and then, to fuse with the cell nuclear envelope (Fig. 2A, middle and Fig. 2B right). Additionally, platelets were able to transfer information through vesicles ( Fig. 2A right and 2B center).

Cancer cells and platelets transfer RNA components
Transference of other cellular components as RNA was also analyzed, by using a RNA labeling method (SytoRNA) in LNCAP cells. After 24h of co-culture, tumor cells had transferred SytoRNA-labeled RNA to 11.47±2.08% of the platelets while lipid cell tracker was transferred to 21.47±1.53%. In the opposite direction, platelets transferred SytoRNA-labeled RNA to 6.62±0.59% of tumor cells whereas 98.38±0.56% showed lipid uptake at 24h. RNA transference e ciency was signi cantly lower than lipid transference, in both TEPs (p<0.001) and PETs (p<0.001) (Fig. 3A) and there was no difference in RNA transference between cell to platelets and platelets to cells (Fig. A-C). Interestingly, pre-treatment of platelets with ADP+Trombin signi catively increased transference from cells to platelets of DiO cell tracker and SytoRNA-labeled RNA from (p<0.01 and p<0.001, respectively) (Fig. S3B). Pre-activation of platelets also increased RNA lipid transference from cells to platelets (p<0.001) while no differences were found in DiO cell tracker transference due to lipid transference is maximum at 24h with untreated platelets (Fig. S3B). After ADP+T pre-treatment of platelets, RNA transference were more e cient from platelets to cells than from cells to platelets (p<0.05) (Fig. 3).
SytoRNA-labeled RNA delivery was also analyzed using imaging techniques, showing RNA release in platelet-derived microparticles and platelet-labeled RNA inside tumor cells (Fig 3D and E). ImageStream pictures showed both SytoRNA-labeled RNA transference from tumor cells to platelets (TEPs) and from platelets to tumor (PETs) (Fig. 3F and Fig. S5).

Tumor cells and platelets exchange proteins
Protein transference analysis was observed on ImageStream experiments revealing that not only tumor cells are able to transfer a protein of epithelial origin (EpCAM) to platelets (Fig. 3G) but also platelets transfer a speci c protein (CD42) to tumor cells (Fig. 3H). After that, platelets speci c proteins (CD42 and CD61) were analyzed by ow cytometry in prostate cancer cell line alone (C), after co-culture with platelets (C+P), and after co-culture with activated platelets [C+P (ADP+T) ]. We observed CD61 and CD42 expression in >40% of EpCAM + tumor cells after 24h of co-culture with platelets while tumor cell culture alone did not show any expression of these markers. Interestingly, pre-activation of platelets did not enhance protein transference from platelet to tumor cells (Fig. 4A). Imaging of platelet to tumor cell transference con rmed cellular uptake of CD61 and incorporation to their membranes (Fig. 4B).
Platelets-speci c proteins are detected in a subpopulation of circulating tumor cells from prostate cancer patients In order to con rm CD61 transference, we isolated circulating tumor cells (CTCs) from peripheral blood samples from fteen early and advanced prostate cancer patients and analyzed the expression of CD61.
Our results showed the presence of two CTCs subpopulations according to CD61 expression (CK + /CD61 + and CK + /CD61 -), and intra and inter patient heterogeneity (Fig. 5). Interestingly, all CTCs analyzed in the advanced prostate cancer stage presented CD61 + expression while three localized stage patients showed CTCs with absence of CD61 expression (CK + /CD61 -) ( Table 1).

Tumor cells and platelets cross-talk induce tumor cell plasticity
Consequences in tumor cell behavior after interaction with platelets was studied in terms of EpCAM protein expression, gene expression of EMT and stemness markers and cell growth. Expression levels of EpCAM were analyzed in tumor cell lines after 1h and 24h after co-culture with platelets. There were a clear trend of EpCAM reduction after 1h of co-culture with platelets in all tumor types analyzed (prostate, lung, colorectal and breast), being signi cant in PC3, 22RV1, H1975, MCF7 and MDA-MB-231 cell lines. This reduction was greater after 24h of co-culture (Fig. 6A).

Induction of the EMT process and acquisition of stem-like features by tumor cells after platelet
interactions were further analyzed. We observed a signi cant induction of the EMT process (upregulation of VIMENTIN, SNAIL1, SNAIL2) at short time-point (1h) in LNCAP, PC3 and SW480 cells lines and also at 48h in LNCAP cells (for VIMENTIN and SNAIL1). In contrast, H1975 cell line from lung cancer only induced VIMENTIN expression at 48h of co-culture (Fig. 6A). Regarding stemness induction, expression of REX1 gene was increased after 1h of platelet co-culture in all cell lines, whereas OCT4 and NANOG were only increased in H1975 and SW480 cell lines at short term. OCT4 was induced at 48h of co-culture in LNCAP, PC3 and H1975 cell lines (Fig. 6B).
Moreover, we studied the effect of platelets addition to prostate cancer cell culture in tumor cell growth Seven of the eleven (63.6%) cell lines studied showed a signi cant increase in cell proliferation after 48h of co-culture with platelets suggesting that the interaction with platelets induced a more proliferative tumor cell phenotype in several tumor types (Fig. S5).

Discussion
Over the last several years accumulating evidence demonstrated that platelets exert several additional biological functions beyond limiting blood loss and promoting wound healing recognizing their role on tumorigenesis [12]. To describe the interaction between tumor cells and blood platelets the term tumoreducated platelets (TEPs) was coined identifying a novel biomarker that enables blood-based cancer diagnostics and treatment monitoring [23]. However, complex bidirectional interactions essential for cancer progression, occur between tumor cells and platelets and involve direct contact trough the formation of tumor-platelet aggregates and release of soluble factors [8,24]. Our study shows that platelets transfer lipids, proteins and RNA to tumor cells inducing structural, genetic and functional modi cations to the tumor cells (Fig. S6).
A large body of literature related with the role of platelets in cancer has been particularly focused on their ability to facilitate circulating tumor cells (CTCs) extravasation and to protect them from shear forces and assault of natural killer (NK) [25,26]. However, most of these studies do not describe structural (lipids and protein) transference promoting tumor cell membrane modi cations and not only platelet cloaking [27].
Here, for the rst time, we demonstrated that platelets are able not only to modify lipid composition of cell membrane but also to introduce themselves inside tumor cells and to modify the lipids of the nuclear envelope.
Some recently published works [28][29][30] showed that microparticles (or microvesicles) derived from platelets in ltrate the tumor being able to transfer miRNAs to tumor cells. In this context, we observed that platelets were able to transfer RNA to tumor cells by direct contact and by the release of microparticles containing RNA. Our results coincide with results obtained by Risitano et al. who observed the ability of platelets to transfer RNA to leukocytes in mouse models and to vascular cells in culture [31].
The EMT is another of the key processes analyzed in this study as changes in the lipid composition of the cell membrane might induce this mechanism. The EMT involves loss of EpCAM expression together with an increase of mesenchymal-associated genes expression [34,35]. In accordance to that, we found an increase of VIMENTIN and SNAIL gene family expression in tumor cell lines after co-culture with platelets. Importantly, these modi cations varied regarding tumor type, most likely due to constitutive expression of EMT markers in some tumor types (such as lung cancer cell lines, that present a semimesenchymal phenotype) [36]. In addition, EMT is known to be related with stem-like phenotype, which is also associated with drug resistance and disease progression [37]. In our work, we found a similar induction pattern of progenitor genes expression (REX1, NANOG and OCT4) and EMT gene expression patterns.
Therefore, alterations in cell and nuclear membranes may affect processes as relevant as cell cycle and genome regulation, cell signaling, or migration and metastasis [38,39].
Interestingly, our results showed alterations of cancer cell membranes after platelet-cell interaction, promoting changes in cell functionality. We observed that co-culture with platelets induced alterations in the cell growth compared with naïve tumor cells.
Likewise, several studies described platelets interactions with many blood cell types, including CTCs, leukocytes, and endothelial cells [6,40]. However, the mechanisms by which platelets promote CTC survival in the blood stream are not fully understood yet. Thus, some studies analyzed the role of platelets on metastasis demonstrating that platelets activation promotes tumor cells survival through thrombin expression, increasing their metastatic potential [25]. Packle T et al. demonstrated that platelet coating may cause transference of MHC class I to tumor cell surface resulting in high expression levels of platelet-derived normal MHC class I, which in turn, mimics host cells and helps them escaping immune surveillance [15]. Despite our results agree with these previous results, our work extensively describes the mechanisms and quanti es the transference of three different groups of molecules (lipids, proteins and RNA) between platelets and tumor cells from several tumor types.

Conclusions
In conclusion, to the best of our knowledge this is the rst work that tumor cells and platelets cross-talk (focusing on tumor cells modi cations induced by platelets) is described and analyzed in such a comprehensive way. We demonstrated that the role of platelets goes far beyond indirect activation of pathways associated with the microenvironment to support CTC dissemination from the primary tumor.
Our data suggest that platelets confer cell plasticity, modifying tumor cell behavior, promoting cell growth and CTC survival, allowing them to evade the immune system and probably chemotherapy.      Representative pictures of CD61+ and CD61-CTCs.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.