Increased Expression of PDGFA and RAF1 in Tumor-Derived Exosomes in Human Colorectal Cancer

Background: Overexpression of tumor markers in Extracellular vesicles (EV), especially in tumor-derived exosomes (TDEs), is implicated in metastasis. However, identifying the specic content of Ev's roles in CRC diagnosis or prognosis requires further validation by bioinformatics and clinical investigations. Methods: In the current study, we explored molecular markers shared between TDEs and circulating tumor cells (CTCs) in the blood of cancer patients to identify candidate genes involved in CRC metastasis. Common markers were analyzed in gene expression proles of two studies (GSE31023 and GSE72577). Results: In blood samples from 20 CRC patients, the expression of candidate genes was assessed by real-time PCR in CTC, TDEs, and microvesicles (MVs), and the expression levels were correlated with clinicopathological features. To further conrm, the expression of candidate genes was investigated in exosomes derived from the parental HT-29 colorectal cancer cell line (HT-29-EXOs), and CSC-enriched spheroids (CSC-EXOs) derived thereof. Gene ontology (GO) analysis suggested platelet-derived growth factor A (PDGFA) and proto-oncogene, Serine/Threonine kinase Raf-1 (RAF1) as new CRC candidate markers in CTCs and TDEs. According to real-time PCR, expression of PDGFA (P=0.0086) and RAF1 (P=0.048) were upregulated in TDEs but signicantly decreased (P=0.0001) in MVs. Furthermore, expression in CSC-EXOs (P=0.0004) was increased compared to HT-29-EXOs. Conclusion: PDGFA and RAF1 mRNA are higher in CSC-EXOs than in HT-29-EXOs, which correlates with higher expression in CSC than the primary tumor. Notably, as no increase was observed in MVs, PDGFA and RAF1 mRNA appear to be actively recruited into TDE. that there are no statistically signicant for expression in tumor of grades or stages these results should be considered with caution account small

subsequent RAS/MAPK pathways await further clari cation 31 . Van et al. showed that KRAS can recruit RAF kinase for activation at the membrane from the cytoplasm 32 . Furthermore, RAS-RAF complexation, and RAS-RAF interaction was proven in colorectal cancer 33 but it attracts much more debate in colorectal cancer exosomes. Ras family members have been found in a variety of vesicles 34,35 and Beckler et al were the rst to speci cally detect KRAS in exosomes 36 .
Proto-oncogene RAF1 transduces phosphorylation signals from the cell membrane to the nucleus in sequential activation of MAPK/ERK pathway (also known as the Ras-Raf-MEK-ERK pathway) 34 . Silencing or pharmacological inhibition of RAF1 impairs clonogenic and tumourigenic capacity of CRC cells and restores apicobasal polarity and formation of tight junctions in cancer cells 36 . RAS mutations are negative predictors of response to anti-EGFR antibodies. Hyperactivation of the RAS-RAF signaling pathway is associated with metastasis, angiogenesis, and poor outcomes of patients in CRC 37 . Where the oncogenic RAS activated MAPK and PI3K/AKT pathways are considered the main effectors in treatment resistance 38, 39 .
In the current research, we approached investigating potential common markers between TDEs and CTCs that are implicated in cancer development and metastasis. The mRNA expression levels of predicted target genes were evaluated by quantitative real-time polymerase chain reaction (qRT-PCR) in TDEs and MVs isolated from 20 plasma samples of CRC patients and 10 healthy donors as control. Expression levels of these markers were also examined in exosomes derived from the HT-29 colorectal cancer cell line (HT-29-EXOs) and HT-29 CSC-enriched spheroids (CSC-EXOs). We aimed to shed light on a possible correlation between candidate exosomal mRNAs expression levels and CRC progression and on the EVsassociated mRNA signature. Con rmation of our hypothesis, that TDEs re ect tumor state and aggressiveness, could hold tremendous potential as a minimally invasive screen and may provide hints towards new therapeutic options.

Data Collection and Bioinformatics Analysis
To obtain candidate genes for CTC and TDE in colorectal cancer, two studies (GSE31023 and GSE72577) were selected (https://www.ncbi.nlm.nih.gov/geo/); the Barbazan et al study contained six cancer and three healthy samples 40 and the Dou et al study encompassed three colorectal cancer cell lines (DLD-1, DKO-1 and DKs-8) and the corresponding TDEs 41 . First, we compared markers expressed in all three TDE and CTC, but not the cell lines. The genes common to TDEs and CTCs were selected.
Characteristic biological features of selected genes were assigned according to gene ontology analysis (GO) 42,43 to molecular functions, biology processes and cellular components using EnrichR (amp.pharm.mssm.edu/Enrichr/), STRING (https://string-db.org/). DisGeNET RDF v7.0 (ref: https://academic.oup.com/nar/article/48/D1/D845/5611674) was used to search for diseaseassociated markers, with emphasis on tumor growth. Disease classes were ranked based on the gene-disease associated (GDA) score and tumor growth-associated markers signi cantly above the mean score (0.06198) were included in the network analysis, generated using Cytoscape Version 3.7.1 44 with ClueGO plugin (http://apps.cytoscape.org/apps/cluego) 45 48 . DisGeNET integrates both expert-curated databases with text-mined data, covering information on Mendelian and complex diseases 49 . All analyses were performed in the R programming language to reach common molecular markers betweenCTC and TDE.

Clinical Sample Collection
Peripheral blood samples were collected from 20 CRC patients and 10 healthy controls at Bahman and Firozgar hospitals from 2018-2020 under ethical committee approval. Healthy controls were enrolled from people who underwent a routine health checkup without disease detection. Cell-free plasma was isolated from all blood samples using 2000×g for 10min and suspended in Qiazole (Qiagen, Germany).
Samples were stored at -80°C. Patient information including gender, age, TNM stage, tumor differentiation was also recorded.

Exosome Isolation by Ultracentrifugation
Exosomes were isolated from culture media of the HT-29 cell line, HT-29-derived spheroids and plasma samples from patients and healthy controls using ultracentrifugation. In brief, culture supernatants and plasma samples were centrifuged at 350× g for 10 min and then at 3000×g for 10 min to remove cell debris. To separate MV from other extracellular vesicles, supernatants were centrifuged at 21000×g for 20 min and the pelleted microvesicles were resuspended in PBS and stored in -80°C. The supernatant was passed twice through ultracentrifugationat 110,000g for 120 min (45Ti rotor, Beckman Coulter). The exosome pellets were resuspended in 1ml PBS or lysis buffer 51 .

Scanning Electron Microscopy (SEM)
Isolated exosomes were xed in 2.5% (w/v) glutaraldehyde for 20 min, washed in PBS and were dehydrated using a gradient of ethanol (60%, 80%, 90% and 100%). The exosomes were dried at room temperature for 10 min on glass. To make surface conductive, a coating of 2-5 nm gold-palladium alloy was applied by sputtering (SPI-Module Sputtering, Argon as gas for plasma) before imaging by SEM (EM3200, KYKY and SEM, Seron Technology, AIS-2100, Korea).

Dynamic Light Scattering (DLS) and Protein Concentration Measurement of Exosomes
To determine the size distribution of isolated exosomes, 50 μl of exosome samples were added to 950 μl PBS and analyzed by dynamic light-scattering measurements (Malvern, UK). Protein quanti cation was performed by a Bicinchoninic acid assay (BCA) protein assay (Pierce BCA Protein Assay Kit, Thermo Fisher Scienti c).

RNA Isolation and qRT-PCR
RNeasy Micro Kit from Qiagene (Qiagen Cat No. /ID: 74004) was used to isolate the total RNA from exosomes. To remove genomic DNA contamination, RNA samples were treated with DNase I and then Nanodrop (ThermoFisher Scienti c, USA) was used for RNA quanti cation and the purity was checked by the A260/A280 ratio. Twenty nanogram of total RNAs were used for reverse transcription (RT) to generate cDNAs using PrimeScript RT Reagent Kit (Takara, Japan). SYBR Green real-time master mix was used for qRT-PCR. The corresponding primers were acquired from SinaClon company (Iran) as follow: GAPDH: 5′-AACTTTGGCATTGTGGAAGG-3′ F and 5′-CACATTGGGGGTAGGAACAC-3′ R. PDGFA: 5′-GCC CAT TCG GAG GAA GAG AA-3′ F and 5′-CAG ATC AGG AAG TTG GCG GA -3′ R. RAF1: 5′-GGT GAT AGT GGA GTC CCA GC -3′ F and 5′-GGT GAA GGC GTG AGG TGT AG -3′ R. The expression levels of PDGFA and RAF1 mRNAs were normalized by GAPDH mRNA levels based on the 2 −ΔΔCt approach 52 .
Statistical Analysis SPSS software version 22.0 (IBM Corp, USA) was utilized to analyze the data. GraphPad Prism version 8.0 for Windows (GraphPad Software, La Jolla, CA, USA, www.graphpad.com)was used to determine the differences between tumor and normal blood samples. Pearson's χ2 and Spearman's correlation tests were used to analyze the signi cance of associations and correlations between PDGFA as well as RAF1 expression and clinicopathological parameters. Kruskal-Wallis and Mann-Whitney U tests were applied for pairwise comparisons between groups. In all parts, quanti ed data are derived from 20 tumors, 10 healthy and cell line-derived exosomes samples, a p-value of <0.05 was considered statistically signi cant. As noted, in the rst step, all quanti ed data was replicated an average of three times.

Ethical Approval
The research ethics committee of Iran University of Medical Sciences issued (IR.IUMS.REC 1395.9221513203) for this study. All procedures including informed consent before surgery from all participants were in accordance with the abovementioned ethical standards. The Ethics Committee of the Bahman and Firozgar Hospitals approved the use of clinical samples. Besides, the patients/participants provided their written informed consent to participate in this study.

Bioinformatics Approach to Select Appropriate Genes Involved in Metastasis
We started searching for upregulated genes expressed in 3 CRC-CTCs 40 and exosomes derived from these CRC lines (TDEs) 41 There have been 410 genes upregulated in CTC and exosomes of these lines (Table S1), with 56 markers overexpressed in TDEs from all three cell lines ( Figure 1A and Table S2). These 56 genes included a considerable number of non-coding RNAs, where particular long non-coding RNA (lncRNA) TPTE pseudogene 1 (TPTEP1) was signi cantly downregulated. For the remaining genes, network and enrichment analysis of the corresponding proteins according to clustered genes (kmeans) are shown in Table S3. However, only 4 of the genes expressed in TDEs of the 3 lines were recovered in CTCs, an additional 6 were overexpressed in CTCs and at least two TDEs derived from these lines ( Figure 1B and Table S2). Network and enrichment analysis of the corresponding ten proteins are shown in Table1 and Figure 1C and 1D. Next, we searched for tumor growth-associated genes in CTCs and the TDEs of the 3 CRC lines (DisGeNET RDF v7.0 database). From all 1046 diseases-related genes, 425 were tumor growth-associated genes (Table S4) and 92 displayed an above average GDA (genedisease associated) score. Network and enrichment analysis of the corresponding proteins uncovered 4 proteins, PDGFA, UBEH2, TPTE and YWHAZ that were also shared between CTC and tumor line exosomes and, in addition, RAF1, which clustered with TPTE and YWHAZ. In fact, RAF1, PDGFA and YWHAZ had the most shared tumor growth nodes ( Figure S1). Based on the connectivity and its central importance in colorectal cancer [53][54][55] , RAF1 was included in downstream analysis.
Next, we examined by EnrichR reactome, molecular functions, biological processes and KEGG pathway analysis in CRC TDE-and CSC-including oncogenesis-related genes (Table S3).The analysis of cellular component, molecular functions and biological processes of our selected genes were associated with apoptotic, PI3K-Akt signaling pathway and cell cycle sequentially, substantiating that we had depicted the most relevant biomarkers particularly for metastasis and stemness in CTCs and EVs. Besides, by using the above-mentioned tools, we generated the signi cant related go-terms for RAF1 and PDGFA genes shown in Table S5 and Figure 2. To con rm this hypothesis, the expression of these two candidate genes, where one should keep in mind the association between RAF1 with TPTE and YWHAZ, was assessed and compared in exosomes derived from the serum of 20 CRC patients and 10 healthy donors.

Patients' Characteristics
The study comprised blood samples from twenty patients and ten healthy volunteers. Clinical data from all patients were recorded. The mean age of patients whom TDEs were isolated were 59 years (SD = 12.48, range 29-81) and whom MVs were isolated were 59 years (SD = 13.08, range 31-81). The mean age of healthy volunteers were 60 years (SD = 14.79, range 25-90); 6 (60%) of them were aged ≤60 years and 4 (40%) were aged >60.All of the clinicopathological details were indicated in Table 2.

Characterization of Isolated Exosomes
TDEs and MVs were isolated from plasma samples of CRC patients and healthy donors through ultracentrifugation. SEM revealed TDEs and MVs presenting a homogeneous, round morphology ( Figure  3A and B) with a Z-average of 133.7 and 920, and a size range of 88.26 ± 10.76 nm and 644 ± 49 nm, respectively, as determined by dynamic light scattering ( Figure 3C and D). In addition, HT-29-EXOs and CSC-EXOs also displayed round morphology in SEM imaging ( Figure 3E and F) and consisted of homogeneous vesicles with a Z-average of 131.2 and 102.8, and a size range of 91.53 ± 9.65 and 79.53 ± 8.46 for HT-29-EXOs and CSC-EXOs, respectively ( Figure 3G and H). Western blot revealed TSG101, CD81 and CD9 exosomes surface markers expression in all three exosome preparations, whereas in the negative control; calnexin (CANX) was not detected (Figure 4A and B).

Increased mRNA Expression levels of PDGFA and RAF1 in CRC Patients' TDEs
The nonparametric Kruskal-Wallis and Mann-Whitney U tests were utilized to measure the differences between the median mRNA PDGFA and RAF1 expressionlevels in TDEs from CRC patients compared to the healthy control group. The results showed statistically signi cant differences of PDGFA and RAF1 between CRC patients and healthy controls (P=0.0086 and P=0.048, respectively) ( Figure 5A and B). The median mRNA expression level of PDGFA and RAF1 was 5.95 and 1.50, respectively. In addition, the median mRNA expression level of their healthy controls was 0.80 and 0.85, respectively. Moreover, Spearman's correlation analysis revealed that the expression pattern of PDGFA and RAF1 genes was strongly positively correlated with each other's (Spearman's rho, P=0.0084) in TDEs ( Figure 5C). The results of the Mann-Whitney U test also showed a statistically signi cant difference in the median level of RAF1 mRNA expression between patients' ages (p=0.05). Pearson's χ2 test also revealed no statistically signi cant association between the mRNA expression levels of PDGFA and RAF1 in TDEs and clinicopathological characteristics, summarized in Table 3.
Decreased mRNA Expression levels of PDGFA and RAF1 in CRC Patients'-Derived MVs Our ndings revealed that PDGFA and RAF1 expression levels were decreased in MVs derived from CRC patients compared to the healthy control group ones(all, P=0.0001) ( Figure 6A and B). The median mRNA expression level of PDGFA and RAF1 in CRC Patient and healthy controls-derived MVs were as follows: 0.22 and 0.10 (tumor), 0.95 and 1.0 (healthy), respectively. As described for exosomes, downregulated expression of PDGFA and RAF1 genes in MVs correlated to each other (Spearman's rho,P=0.0029) ( Figure 6C). Pearson's χ2 test also exhibited no statistically signi cant association between mRNA expression levels of PDGFA and RAF1 in MVs and clinicopathological parameters that are summarized in Table 4 Besides CTC/CSC, TDEs contribute to EMT, angiogenesis and tumor progression. In addition, TDEs carry selected mRNAs that are shuttled in the CRC microenvironment 63 , potentiating cancer progression and affecting CRC patient's prognosis 64,65 . TDEs also suppress immune responses, promote immune evasion and increase drug/ chemotherapy resistance 66, 67 . The steadily increasing interest in CTC and TDE biomarkers in relation to tumor metastasis is forced by high-throughput analyses, which allow for simultaneous identi cation and quanti cation of multiple mRNAs. Novel bioinformatic tools facilitate integration of this multitude of genes into functional networks 68 .

We used bioinformatic tools to search for molecular markers shared between CTCs and TDEs in CRC. For
EVs isolation we choose ultracentrifugation, considered the gold standard for exosome isolation 69 . Expression levels of markers, including PDGFA and RAF1, shared by either CTC and TDE or serum-derived TDEs and MVs were evaluated by RT-PCR. To further validate the engagement of TDE-derived PDGFA and RAF1 in the CSC phenotype and characteristics, their expressions were investigated in HT-29-EXOs compared to CSC-EXOs.
Our results indicate increased expression of both PDGFA and RAF1 in TDEs of CRC patients compared to healthy controls. Unexpectedly, expression of both PDGFA and RAF1 was decreased in MVs from CRC patients compared to healthy controls. Notably, PDGFA and RAF1 expressions were correlated in both TDEs and MVs. Low recovery of PDGFA and RAF1 in MVs is in line with MVs originating from budding membrane domains. However, this does not explain lower recovery than in healthy donors' MV. One possible mechanism could be regulation by non-coding RNA, which requires further exploration as we excluded noncoding RNAs from our analysis. Instead, molecular packaging named selective cargo may well contribute to higher PDGFA and RAF1 recovery in TDEs, where besides active recruitment, silencing of counter regulatory elements can presently not be excluded. Finally, PDGFA and RAF1 mRNA is higher in CSC-EXOs than in HT-29-EXOs indicating their potential correlation with the CSC phenotype.
Thus, with all caution, we assume a possible correlation between increased RAF1 and PDGFA expression in exosomes with tumor progression. This study is the rst report touching the increased expression of RAF1 and PDGFA in exosomes derived from CRC patients and CSC-EXOs compared to healthy donor and HT-29 parental-derived exosomes (HT-29-EXOs). Since exosomes content re ects the condition of cells of origin and can lead to modulation in recipient cells by transferring their cargo, as well as based on accumulating evidence of the role of increased expression of RAF1 and PDGFA in tumor progression (invasion, angiogenesis and EMT), we assume some possible correlation between increased RAF1 and PDGFA expression in exosomes with tumor progression.
Accumulating evidence supports that PDGFA is an important mediator of EMT that contributes to cancer invasion and angiogenesis. Overexpression of PDGF-D showed EMT promotion in prostate cancer cells 70 . The crosstalk between PDGF and EMT-related signaling pathways, such as the nuclear factor κlight-chain-enhancer of activated B cells (NF-κB) and chemokine (C-X-C motif) receptor 4 (CXCR4), further strengthens PDGF playing an important role in EMT. Interestingly, in a study on hepatocellular carcinoma (HCC), PDGF was hypothesized to be involved in TGF-β-induced EMT of metastasizing cancer cells 71 . Additional studies on the TGF-β enhanced expression of PDGF and PDGFR via activation of β-catenin and the signal transducer and activator of transcription 3 (STAT3) 72 . However, it should be mentioned that unsupervised use of anti-PDGF could potentially promote tumor invasion and metastasis, stressing dosedependence of the molecular mechanism 73 . A study on rectal cancer indicated that mRNA expression of PDGFA was decreased following anti-PDGF treatment and pointed towards PDGFA expression being also associated with drug resistance 74 . Moreover, in line with our study, expression was up-regulated under hypoxia as well as in the CSCs and high PDGFA expression was associated with poor overall survival 75,76 . Signaling mostly proceeded through the EGF-STAT3 pathway with increased levels of LGR5, and participation of the Wnt signaling pathway in EGFR-positive CRCs. Also, in line to our study, RT-PCR con rmed increased PDGFA levels in both HCT-116-CSCs and HT-29-CSCs 27 . Our study extends those ndings towards the important recovery in TDEs and even more pronounced in CSC-EXOs, which allows for the transfer into target cells.
RAF1 play a carcinogenic role especially in angiogenesis process 38 . Moreover, targeting of RAF1 by miR-7-5p inhibits endothelial cell proliferation. Inhibition of RAF1 kinase activity impairs CRC growth. Furthermore, down-regulation of miR-431-5p as well as up-regulation of FBXL19-AS1 increases RAF1 expression. Thus, FBXL19-AS1 knockdown-mediated inhibition of lung cancer progression and the expression of angiogenesis associated proteins could be rescued by RAF1 overexpression 77,78 .
In summary, in line with these studies, we observed overexpression of PDGFA and RAF1 in exosomes from CRC patients, where overexpressions were associated with poor clinicopathological features. These ndings suggest a possible role of TDEs with prometastatic factors, including speci c mRNAs, as messengers for primary tumor growth and microenvironment preparation for metastasis.
Additional advantages of mRNA biomarkers deserve discussion. In comparison to DNA, mRNAs provide a solid base for signaling network connectivities. Uncovering signaling networks makes RNAs most important in unraveling protein-RNA complexes and allows identifying potential candidates for follow-up work at the protein level including functional studies. Furthermore, due to translational modulation and post-translational modi cation, protein levels do not necessarily re ect gene expression levels, a problem that can be circumvented by elaborating RNA levels 79,80 . Several publications also support the importance of mRNA in CRC. Furthermore, additional publications support the view that both mRNA and protein analysis can con rm each other 81-83 . Finally, again in CRC, liquid biopsy-based mRNA evaluation was suggested providing new insights into potential mRNA indicators that may allow avoiding invasive diagnostic operations.
In spite of that, current study had a few limitations, Firstly, our patients' sample size was small requiring future validation in large cohorts of patients' serum-derived TDEs and CSC-EXOs. Secondly, our ndings need con rmation at the protein level. This accounts for clinical serum samples as well as for cell culturederived exosomes and for in vivo controls in mouse models. Though our exosome sample sizes, particularly in the clinical cohort, were too small for a comprehensive protein analysis, this will be possible by the restriction to PDGFA and RAF1. Thirdly, the mode of functional activity of PDGFA and RAF1 in CRC requires further elaboration including the clari cation of preferential actions at the mRNA or the protein level. This is important as to our knowledge we were the rst describing high PDGFA and RAF1 mRNA expression in TDEs from CRC patients and exosomes derived from CSC-enriched spheroids, exosomes being a prerequisite for transfer into host cells, may play potential roles in tumor growth and progression.
Additionally, serum-derived TDEs would present an easily accessible, non-invasive tool for early diagnosis, prognosis and therapy control.

Conclusion
Based on our results, PDGFA and RAF1 mRNA are higher in CSC-EXOs than in HT-29-EXOs which correlates with higher expression in CSC than the primary tumor. Notably, as no increase was observed in MVs, PDGFA and RAF1 mRNA appear being actively recruited into TDE. In view of the suggested importance of PDGFA and RAF1 expression in CRC prognosis, further validation has high priority in future plan. 1395.9221513203 for this study. All procedures, including obtaining informed consent from each human participant before surgery, were in accordance with the above-mentioned ethical standards.

Consent for publication
The signed consent ensured that the publisher has theparticipate and author's permission to publish the relevant Contribution. Besides, all of the included studies data were used by reference citation and all of the authors consent to publication.

Availability of Data and Materials
Data sharing is not applicable to this article as no new data were created or analyzed in this study. Analyzed data are openly available in [repository name at http://doi.org/[doi] and reference number.

Competing Interests
The authors whose names are listed certify that they have NO a liations in any organization or entity with any nancial interest and non-nancial interest in the subject matter or materials discussed in this manuscript.

Funding
This study is part of a PhD thesis in Department of Molecular Medicine and mainly was funded by Iran University of Medical Sciences (number: 96-01-87-30129).
Author's Contributions ZM, ME and SV, conceived the presented idea. SV collected the blood samples, run the experimental laboratory tests, analyzed and interpreted data and wrote the rst draft. ZM and ME developed, revised and approved the theory. MN performed cell culture and spheroid formation assay. SV performed critical revisions of the rst draft and veri ed the concept. YG and RK helped in bioinformatics data analysis. HA, colorectal surgery specialists who provided patients' tumor sample and their data. All authors read, approved and discussed the results and contributed to the nal manuscript.