Telocytes Promote the Metastasis of Hepatocellular Carcinoma by Activating ERK Signaling Pathway and Sponging miR-942-3p to Impact MMP9

Background: Hepatocellular carcinoma(HCC) in China is considered as a familiar malignant tumor with poor prognosis, high metastasis and disease relapse. Telocytes(TCs) have been veried to participate in progresses of tumorigenesis, invasions and migrations by secreting functional proteins and transmitting cell-to-cell information. Extracellular signal-regulared protein kinase(ERK) signal pathway is a vital mechanism driving cell proliferation, metastasis and apoptosis, but whether this molecular signaling mechanism contributes to matrix metalloproteinase-9(MMP) expression of TCs remains unclear. Methods: Telocytes and MMP9 expression in the liver cancer tissues are measured by immunohistochemistry assay, Westen blot assay and RT-PCR technique, meanwhile primary telocytes from liver para-cancer tissues are cultured in vitro. To demonstrate the function of telocytes for hepatocellular carcinoma, the metastatic cancer animal model is established by three typs of liver cancer cell-lines in vivo. Results: In our study, we elucidate that TCs in the para-cancer tissue can promote the metastasis of HCC cells by MMP-9 expression, in vitro and in vivo. PDGF derived from HCC cells has a capacity to activate Ras/ERK signaling pathway of TC as a result of accelerating MMP-9 expression, but it’s no signicant for proliferative potential and apoptotic rate of TCs. While tyrosine kinase inhibitors and miR-942-3p suppress MMP-9 expression to make loss functions of TCs. Various mutations of TCs are also tested and single nucleotide polymorphisms of MMP-9 may be the potentially molecular mechanism of increasing protein expression in the invasive process of HCC. Conclusion: Our results demonstrate two potential mechanisms between HCC cells and TCs, suggesting that TC is a novel marker and target on deciphering reasons of cancer metastasis. Our study started from nding differences of TCs and MMP-9 expression in the HCC tissues, elucidated two potential mechanisms of TCs in promoting the metastasis of HCC cells and stimulating MMP-9 expression in vitro and in vivo: (cid:0) ) PDGF from HCC cells activates ERK signaling pathway of TC to enhance MMP-9 expression; (cid:0) ) miR-942-3p as a inhibitor suppresses MMP-9 expression. Furthermore, whole exon gene mutations of TCs are tested and analyzed, and MMP-9 related SNPs mutations are detailed detected to explore new targets of TCs. the hepatic cancer cells by the Transwell assay(B). Representative photos of xenograft tumor in vivo(C), growth curves of tumor weight(D) and volume(E) over time in HepG2 group, adding TCs group, adding TCs and MMP9 inhibitor group. Tumor growth curves of volumes with incremental injections of TCs(F). Knockdown MMP9 in TCs and knockdown PDGF in HCC inuenced migrations and invasions of hepatic cancer cells after 48 hours by Wound healing and Transwell assays(G,H). The number of migratory cells(I) and invasive cells(J) were calculated in shMMP9 and shPDGF groups. Representative photographs of metastatic lung tissues derived from HepG2 cancer cells with distinct frequencies of TCs’ injection by HE staining(K-a) and in vivo(K-c). The expression of MMP9 by IHC staining were shown in different groups(K-b). Contrast curves of the nude mice weight(L) and the weight of metastatic lung tissues(M) with distinct frequencies of TCs’ injections. ns: no signicance; ※ p<0.05, ※※ p<0.01, ※※※ p<0.005.


Background
Globally, liver cancers are the sixth most familiar cancers and rank fourth in terms of cancer related mortality [1], and the World Health Organization estimates that more than one million patients will die of it in 2030 [2]. Metastasis is an common lethal factor for most malignant carcinoma. The secondary germination in distant organs requires multiply steps process such as intruding into the blood circulation system after tumor angiogenesis, sustainable growth with inexhaustible viability, and high-intensively colonized power [3]. Cancer cells could overcome multitudinous obstacles by a way of alternating surrounding environment and regulating peripheral cells to establish appropriate conditions for metastasis. Although various efforts have been conferred to demonstrate the molecular mechanism of HCC, we are still accustomed to diagnose it at the clinical stage, distant metastatic phase and postsurgical recrudesces [4]. Imminently, a novel biomarkers related to metastasis and prognosis of HCC, and new target regulations of promoting invasion and migration to normal tissues are necessary.
Telocytes(TCs), deriving from interstitial Cajal-like stem cells(ICLC), are a novel type of mesenchymal cell with several long thin and beaded-like telopodes and extensively exist in the most of mammal animals' organ [5][6][7][8][9]. According to previous researches, TCs are con rmed positive for CD34, CD117, plateletderived growth factor receptor(PDGFR) and negative for CD28, Vimentin and NOS in the hepatic tissue, kidney and vascular tissues [10]. These bio-marks are utilized for TCs' identi cation [11][12][13]. Potential functions of TCs have been discovered to construct the basis support of arteries [14], participate in the pathological process of chronic wound healing [12,15], involve in transmit signaling information by autocrine and paracrine approaches [16,17], and result of cell steatosis in atherosclerotic disease [10]. The study of TCs in the tumorigenesis and development, the current exploration is only limited on the morphological and quantitative alterations of TC in tumors [18,19] and restricted to activate or inhibit signaling pathways of TC [20]. For instance, hyperplastic TCs were found as the physiological counterpart on the neoplasia of in ammatory broid polyps and PDGFRA mutant gastrointestinal stromal tumor [19]. Furthermore, TCs with steroid hormone receptors on membrane involved in uterine leiomyoma growth by changing their density and local homeostasis [18]. Nonetheless, present studies only concentrate on the morphology, magnitudes, super cial bio-markers and ultrastructure changes of TCs in distinct organs, but the mechanisms of why TCs have capabilities to promote tumor growth and metastasis and which molecular variants happen inner TCs, were still unknown and imminently investigated.
Extracellular matrix(ECM) procides mechanical and biochemical support to cells and constructs homeostasis of peripheral tissues, and they contain matrix metalloproteinases(MMPs), heparanases and aggrecanases, among others [21,22]. MMP-9, one of MMPs, not only participates in matrix remodeling of tissues [23,24] but also involves in migration, in vasion and tumorigenesis of cancer [25,26]. MMP-9 can be secreted by ICLC such as TCs through the way of homocellular junctions to interact with other surrounding cells. Downregulating the density of MMP-9 suppressed the metastasis of HCC [27,28].
Focusing on the mechanism of MMP-9 expression, more than one signaling pathways contribute to promote or inhibit it, including PI3K/AKT/NF-ĸB signaling pathway, transforming growth factor-beta(TGFβ)/SMAD signaling pathway [29,28,30]. What's more, inhibition of ERK led to reduce MMP-9 expression and gain-of -function reversed by eliminating MAPK related inhibitors in the brosarcoma disease [31]. In head and neck squamous cell carcinoma, iron exhibited to medicate MMP-9 by ERK1/2 signaling pathway [32].
The correlation and interaction of TCs and HCC remain undetermined. Therefore, the aim of our study is to explore a new target to disclose the mechanism of metastasis of HCC and discover the signal transduction between HCC and TCs. With preliminary experiments, we audaciously hypothesis that HCC cells secreted PDGF, binding with PDGFR on TCs and resulted in accelerating MMP9 expression, the latter played a pivotal role in tumor migration and invasion to distance. Additionally, molecular mechanism of MMP9 expression of TCs was an extremely complicated progress including microRNA(miRNA) regulation and a characteristic signaling pathway, so we also highly conjecture that downstream target gene of MMP9-related miRNA simultaneously participated in adjusting the expression of MMP9. Therefore, in order to verify these speculations, we tested TCs density and MMP9 expression by immunohistochemistry, immuno uorescence and Westen blot assays, and analyzed their relationship with the overall survival(OS) phase. Furthermore, we construct experiments in vivo and in vitro to explore the in uence and mechanism of TCs with MMP9 in metastasis of HCC.

Clinical samples
Between January 2018 and June 2020, Surgical tissues were collected at the rst a liated hospital of Shandong rst Medical University from 132 patients with hepatocellular carcinoma disease con rmed by fast pathology biopsy during the operation after patients' signed informed consents were obtained. All fresh tissues were restored immediately on the refrigerator of -80℃ and anonymized before transfer to the laboratory for further processing. All the demographic data, including age, sex, clinical classi cation, survival time, and relative follow-up visits were gathered. All control subjects were followed up for two and a half years and were free of liver malignancy.

Primary TCs Culture
Fresh samples from liver para-cancer tissues and hepatic hemangioma tissues(as the control group) after patients' operation were selected and cut into fragments, and incubated with 5 mg/ml collagenase type II (Sigma-Aldrich, St. Louis, MO, USA) for 10min. PBS without calcium and magnesium(pH 7.4 G0002 Servicebio, USA ) washing twice and centrifuged at 1000 r.p.m., 5min, then re suspended in DMEM with 10% foetal calf serum. The BJ-40 capillary glass tube (1.0 mm outer diameter, 0.8 mm inner diameter) was soaked in 1 mol/L hydrochloric acid for 24 h, then rinsed continuously with ultrapure water, dried at 65°C and autoclaved. Under 200x magni cation of the microscope and selected cells to 0.2 mL centrifuge tube containing 2μL of lysate according to the morphology of TCs [33,34 ].

Liver TCs isolation and identi cation
After mature C57BL/6 mice(No.4432, Weitonglihua animal company, Beijing, China) were killed with anaesthetic. The hepatic tissue were isolated under sterile conditions and collected into sterile tubes containing DMEM (Gibco-8120217, NY, USA), Suppl.ed with 100 UI/ml penicillin, 0.1 mg/ml streptomycin (20201013, Kaisu biology Co Ltd., Jiangsu, China), transported to the cell culture laboratory. Dispersed cells were separated by ltration through a 40-m-diameter cell strainer(CLS431751 Falcon, NJ, Germany), collected by centrifugation at 1000 r.p.m., 5min., and resuspended in DMEM, Suppl.ed with 10% foetal calf serum, 100UI/ml penicillin, 0.1 mg/ml streptomycin (Sigma-Aldrich). Cells were distributed in 25cm 2 plastic culture asks, at a density of 1×10 5 cells/cm 2 , and maintained at 37℃/5% CO 2 atmosphere until becoming semi-con uent. Culture medium was changed every 48hrs. The typical TC were photographed by auto-microscopy per 12hrs. After cell adhesion on the plate, Cells were selected, puri ed and further ampli ed to the next experiment. The TCs were identi ed according to the morphology and immuno uorescent staining assays. The protocol followed by the hereinafter.

Lentivirus production and infection
To build recombinant lentivirus, 293T cells were cotransfected with progresses of package, envelop, and expression. The virus-containing supernatant was harvested and concentrated by ultracentrifugation. The viral stock was Suppl.ed with 8 mg/mL of polybrene for infections .

Cell Transfection
In order to gain stable experimental cell lines, the primary TCs from hepatic tissues(from mice) transfected with SV40 large and small T antigen to constructed TC SV40 . The culture medium of TC SV40 cells was Dulbecco's modi ed Eagle's medium/F12 with 10% fetal calf serum (Gibco-8120330, NY, USA) Suppl.ed. HepG2, SNU182 and SK-HEP-1 cell-Lines were cultured in DMEM(GIBCO Beijing, China) Suppl.ed with 10% FBS, 100 U/ml of penicillin, and 100 μg/ml of streptomycin and then Cells were placed in a humidi ed atmosphere containing 5% CO2 at 37°C. When cells reached 60-80% con uence, positive and stable transfectants were selected for further study.

RNA extraction and qRT-PCR analysis
To verify the mRNA expression level of MMP2, MMP3, MMP9, MMP11 and MMP14 in HCC tissues and para-cancer tissues, qRT-PCR analysis was performed. Total RNA was extracted from tissues using Trizol reagent (CW0581, Kangweishiji company, China) according to the manufacturer's protocol. Reverse transcription and cDNA ampli cation were performed using the SYBR Master Mixture (CW0957, Kangweishiji company, China), respectively, according to the manufacturer's guidelines. The β-actin genes were used as endogenous controls. A HiFiScript Primer Assay (CW2569, Kangweishiji company, China) was used to assay MMPs and β-actin(Suppl. Table 1).

RNA interference
For miR-942-3p test, mature miRNA sequence was found by miRBase database, shRNAs and mimics of indicated miRNAs were obtained by RiboBio Company (Shanghai and Wuhan, China). Transfection with shRNAs and miRNAs was completed using riboFECT™ CP (RiboBio) according to the manufacturer's instructions.

Western blot analysis
Human hepatocellular cancer tissues with para-cancer tissues were fetched out from the -80℃ condition, thawed and resuspended using lysis buffer (20% Glycerol, 4% SDS in 100 mM Tris Buffer, pH 6.8). Cell extracts were boiled for 10 min in loading buffer and then equal amounts of cell extracts were separated on 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis gels. Separated protein bands were transferred onto polyvinylidene uoride membranes. The primary antibodies against MMP2, MMP3, MMP11, MMP9, MMP14 and GAPDH were diluted according to the instructions for each of the antibodies and incubated overnight at 4°C. Then, horseradish peroxidase-linked secondary antibodies were added and samples were incubated at room temperature for 2 h. The membranes were washed with phosphatebuffered saline (PBS), and the immune-reactive bands were colored using an ECL-PLUS/TM (Amersham company, UK) according to the manufacturer's instructions. The relative protein levels between cancer tissues and para-cancer tissues were normalized to GAPDH concentrations. Three separate experiments were performed for each clone. Furthermore, the primary antibodies against Bax and Cleaved-caspase-3 were incubated for cell apoptosis test(Suppl. Table 6).

Immunohistochemistry (IHC) staining
Formalin xed para n-embedded primary tumor tissue with para-cancer tissue were utilized for IHC. For heat-induced antigen retrieval, slides were soaked in citric acid buffer and heated keeping 1300w for 2 min. After quenching endogenous peroxidase activity with 3% H 2 O 2. Specimens were incubated with antibody at 4°C overnight. Speci c signals were developed with second-antibody using diaminobenzidine as chromogen(Suppl. Table 6). Sections were then counterstained with hematoxylin and observed under light Microscope(XSP-C204, CIC, China). Slides were scanned using laster scanning confocal microscope (Eclipse Ti-E, Nikon, Japan) with 40×magni cation. Datum were quanti ed in immunohistochemistry digital slides with Leica Aperio positive pixel count algorithm using whole slide analysis (PANNORAMIC DESK/MIDI/250/1000, 3DHISTECH, Hungary).

Immuno uorescence(IF) staining
All sections were incubated in 2 changes of xylene at 15 min each. A dehydrator was used to dehydrate the sections in 2 changes of pure ethanol and were immersed in EDTA antigen retrieval buffer (pH 8.0, G1206/G1203, Servicebio, USA). The slides were incubated with primary antibodies to double stain for CD34, CD117, PDGFR-α and MMP9 overnight at 4℃ and then with secondary antibody after washing three times with PBS(Suppl. Table 6). Under microscopy, images were collected using uorescence microscopy (NIKON ECLIPSE C1, Tokyo, Japan) with an imaging system (NIKON DS-U3, Tokyo, Japan). Images were captured at 1 to 400 magni cation (Microscope Camera XSP-C204, Olympus Europa GmbH, Hamburg, Germany).

Transwell assay
For invasion assays, Transwell migration chambers and Matrigel coated chambers(Becton Dickinson, Waltham, MA) were used. Brie y, 5 × 10 4 HCC cells were seeded into the upper chamber in serum-free culture medium. The lower chamber was lled with 5 × 10 4 TCs completed medium with 10% FBS. After 48 h for the invasion assay, cells that have invaded through the membrane were stained with 1% crystal violet and counted in the microscopy(CKX-51,OLYMPUS company, Japan).

CCK-8 cell counting assay
TCs in logarithmic phase were digested and made into cell suspensions. After uniform spreading, the cells were incubated and then 10 μ L of 5 mg/mL cell counting kit-8(CCK-8)(HY K0301, MCE, Shanghai, China) was added to the wells starting from the second day after spreading and 4 h before the termination of the incubation. The OD value was measured at 450 nm by enzyme marker after 4 hours.

Wound healing assay
Use a marker pen on the back of the 6-well plate uniformly with horizontal lines. Add approximately 2.5 × 10 5 HCC cells to the wells and overnight to reach 100% fusion rate. Prepare co-culture cells-TC SV40 , wash the cells 2-3 times with PBS after digestion, resuspending with serum-free medium and add co-culture chambers. MMP9 inhibitor group was a concentration of 3uM, adding each group of co-culture chambers into the corresponding wells. Every chambers incubated in 37℃ 5% CO 2 incubator for 48hr. The area was counted using Image J software.

In vivo models
For building metastatic lung cancer, 1 × 10 7 HepG2 cells in 100 μl of PBS were injected into the tail veins of BALB/c-nu mice(No.4272, Weitonglihua animal company, Beijing, China). 6 × 10 4 TCs in 50 μl of 0.9% normal saline were injected into mouse tail vein per 7 days after HepG2 cells injection. The mice were sacri ced on day 42. Lung tissues were resected, photoed and xed with 4% paraformaldehyde, and then strained with hematoxylin and eosin (HE). For subaxillary transplantation, HepG2 cells were injected into the right axilla of nude mice and 6 × 10 4 TCs in 50 μl of 0.9% normal saline were injected around local tumors per 3 days and TCs with MMP9 inhibitor group built as the compared group. After 28 days, these mice were sacri ced to gain axilla transplanted tumors which were measured weights and volumes(maximum axis × minimum axis 2 × 1/2).

Whole-exome Sequencing Technique
Experimental Flow A cohort of 23 patients with HCC who underwent surgical resection between 2019 and 2020. Primary culture TCs were selected from fresh para-cancer tissues of Genomic DNA samples of acceptable quality were randomly interrupted by ultrasonic high performance sample processing system (Covaris) into fragments with a major peak of about 200bp-300bp. Subsequent DNA fragments were then end-repaired by adding an "A" base at the 3' end and a library splice at both ends. An appropriate amount of hybridization library was captured and enriched with the exome chip, and the unenriched fragments were eluted and ampli ed, and the whole exome was captured. The ampli cation products were subjected to Agilent 2100 bioanalyzer instrument (Agilent DNA 1000 Reagents) and QPCR quality control. We used the Illumina HiSeq family of platforms to perform high-throughput sequencing of each quali ed library and to ensure that the data volume of each sample was up to standard. The raw image data obtained from sequencing was converted into raw reads (raw reads) by Illumina Base Calling software, i.e. double-end reads (paired-end reads). The data were stored in FASTQ le format, called raw data.

Variant Calling and Bioinformatics Analysis
The information analysis started with the sequenced downstream data. The raw data contained adapter sequences, bases of low sequencing quality, and undetected bases (expressed as N). Next, the clean data of each sample was compared to the reference genome using the alignment software BWA(Burrows-Wheeler Aligner) [35] to obtain the initial alignment result le in BAM format. To ensure the accuracy of variant detection, we followed the optimal variant detection analysis procedure recommended by the o cial website of GATK. Based on the alignment results, the evaluation indexes such as sequencing depth, coverage, and alignment rate of each sample were counted. In the process, we wield HaplotypeCaller of GATK v3.4.0, including SNPs(single nucleotide polymorphism) and InDels, and ltered the raw variant detection results with high con dence(Suppl. table 3,4). Next, the variant results were annotated and impact predicted using the in-house software AnnoDB, as well comparison of different types of sample sets using the genome visualization software IGV(Integrative Genomics Viewer; Suppl. gure 3-F ).
Statistical analyses SPSS20.0 was utilized for the statistical analyses. Quantitative value was recorded as the mean ± standard(SD). Two-tailed Student's t-tests, paired t-tests, chi-square tests, and multivariate analysis were used to assess the differences among groups. Pearson correlation analysis was used to analyze MMP9 expression with TCs number in HCC tissues. Survival curves were drawn by the Kaplan-Meier analysis and statistical signi cance was considered as P < 0.05.

Results
The identi cation of TCs The diversity and correlation analysis between Telocytes and MMP-9 in HCC tissues.
In order to verify the number changes of telocytes in HCC tissues, we chose 132 specimen by immuno uorescence assay to calculate the quantity of CD34-positive telocytes in the HCC tissues and para-cancer tissues (Fig2 A-a,e). Visibly, the number of telocytes in the HCC tissues were rare compared with them in the para-cancer tissues, and it had signi cant by paired t test statistic assay(t value =57.640, p <0.0001, Fig2 B). MMP-9 protein was also detected by immunohistochemistry and immuno uorescence assays both in the HCC and para-cancer tissues, the MMP-9 protein expression index of former tissues was lower than it of latter tissues(t value =138.600, p <0.0001, Fig2 A,B). Furthermore, MMPs family including MMP2, MMP3, MMP9, MMP11and MMP14 were also utilized to represent the differences between HCC tissue and para-cancer tissue by Wesen blot assay, qRT-PCR analysis and immunohistochemistry staining. Interestingly, not all proteins of MMPs family had similar expressions in HCC and para-cancer tissues(Fig2 C,D,E) : there was scarce discrepancy of MMP2 and MMP14 expression; MMP3 protein expressed lowly in the para-cancer tissue but highly in the HCC tissue; MMP11 protein expression was more in the HCC tissue, but the result was adverse by qRT-PCR analysis. Additionally, the correlation of MMP9 protein and telocytes was demonstrated through Pearson's correlation statistic analysis both HCC tissues and para-cancer tissues (Table 3), and this positive correlation between MMP9 protein and telocytes was a critical and high-level TC precondition for following studies. In order to clarify potential clinical correlations among MMP9 protein, telocytes and HCC metastasis patients, we analyzed quantities of telocytes from 93 cases with HCC metastasizing patients and 39 cases without metastasizing patients. The count of telocytes and MMP9 protein expression in the para-cancer tissue were obviously higher in metastasizing samples than them in nonmetastasizing samples (Fig 3 B), and MMP9 protein expression allocated surrounding telocytes on scans of immuno uorescence staining (Fig 3 A). Whether high number of telocytes were relative with cancer metastasis in the HCC para-cancer tissues , we factitiously made two hierarchical classi cations of telocytes: low-count ≤ 8/μm 2 high-count > 8/μm 2 (  (Table 2).

Telocytes promoted HCC migration and invasion by MMP9 expression
To investigate the effects of TCs on HCC migration and invasion, we performed Wound healing migration (Fig 4-A) and Transwell invasion assays (Fig 4-B), which indicated that TCs promoted cell migratory and invasive abilities of HepG2, SNU182 and SK-HEP-1 cells, meanwhile inhibiting MMP9 expression attenuated these impacts. To functionally validate roles of TCs to HCC cells, we performed in vivo assays using xenograft mouse models by injecting HepG2 cells and TCs to mice axilla(Fir 4-C). After 21days, the weight and volume of HepG2 tumor with fth TCs injections were obviously biggest than only HepG2 tumor and tumor with MMP9 inhibitor(p<0.01; Fig 4-D,E). There were no signi cant of migration and tumor growth between HepG2 tumor group and tumor with MMP9 inhibitor group(p>0.05), so TCs impacted HCC activity depending on MMP9 expression. Within a certain range(from once to fth TCs infections), the higher the number of TCs the larger the volume of tumor growth (Fig 4-F). To verify the interaction mechanism between TCs and HCC cells we designed lentivirus bonded with shRNA to knockdown MMP9 in TCs and knockdown PDGF in HCC cell lines, respectively. The migrations and invasions of HepG2 SNU182 and SK-HEP-1 cells with shMMP9 TCs were apparently receded compared against these cells with shNC group (Fig 4-I,J) whereas shPDGF cell-lines represented more strong mobility and invasiveness because of compounding with normal MMP9 expression of TCs (Fig 4-G,H). Without MMP9 protein, TCs lost the role of promoting HCC cells' migratory and invasive abilities, of which might be irritated by PDGF protein. To identify our conjecture that TCs played an critical role in distant metastasis of HCC cells, we utilized nude mice to establish lung metastatic lesions from liver cancer and distinguished subgroups according to times of TCs injection from the mouse tail vein (Fig 4-K). We found that the volume of lung metastatic cancer lesions increased with the frequency of TCs injected by HE staining assay (Fig 4-K-a), and meanwhile MMP9 protein expression were more higher in TCs' injection group as contrasted with control group by IHC staining assay (Fig 4-K-b). The body weight of mice in more than twice TCs infection groups decreased after 28 days compared with control group and once group (Fig 4-L). Anatomical observation and the weight of lung rose evidently along with TCs injections compared with control group (Fig 4-M; Fig 4-K-c).
The potential mechanism of PDGF to Telocytes To explore how PDGF impacted MMP9 expression of TCs, we rst tested whether HCC cells secreted MMP9 protein by IHF staining assay, and found that HepG2 cells could only express PDGF protein instead of MMP9 protein (Fig 5-A). In order to explore the concentration gradient of PDGF to facilitate the secretion of MMP9 protein by TCs, we detected the amount of MMP9 secretion by increasing the dose of PDGF and plotted the concentration curve, from which we extracted the most optimal number of TCs cells and PDGF concentration for the following experimental study, so the experiment showed that 5μl of PDGF acting on 6×10 4 TCs resulted in the maximum quantity of MMP9 (Fig 5-B). Through previous studies and published literature, we boldly hypothesized that PDGF stimulates TCs to secrete MMP9 by activating the mitrogen-activated protein kinase(MAPK) signaling pathway, so we drew a simulated diagram of extracellular signal-regulated kinase(ERK) signaling pathway to demonstrate the molecular mechanism [37]. In order to verify our conjectures, we utilized PDGF which was the one of agonists for MAPK signaling pathway to stimulate TCs, comparing with resting conditions(NC group) we found that PDGFR as one of TCs membrane receptors transmitted activating information by Son Of Sevenless(SOS) combining with growth-factor-bound-2(Grb2) to motivate downstream effector proteins -Ras/ERK signals. Grb2-sos compound activating, RAF activating and MEK/ERK phosphoylation were shown by Western Blot assay, and nally MMP9 protein was quantitatively expressed (Fig 5-D). Moreover, to demonstrate potent inhibitors of dimeric BRAF, we chose AZ-628[38] which belonged to RAF inhibitors and U0126 [39] which belonged to ERK inhibitors to interdict revitalites of MAPK singles, and we found that active RAF, phosphoylative MEK/ERK and MMP9 protein disappeared on the condition of these inhibitors regardless of PDGF (Fig 5-D).
Because miR-942-3p was one of direct targeted gene in regulating MMP9 expression by referring miRbase and TargetScan database, we nally chose miR-942-3p and validated its interaction with MMP9 by Luciferase assay. Differential expression of miR-942-3p in TCs of compared normal liver tissue and hepatocellular carcinoma paracancer tissue was the basis for future study, and we found that miR-942-3p of TCs separated from HCC para-cancer tissue was more in comparison with normal liver tissue (Fig 5-E). After generating reporter constructs in which 5' UTR mRNA of MMP9 was cloned downstream of the Luciferase open reading frame (Fig 5-F), miR-942-3p could directly combined with MMP9 mRN, in addition mutagenesisi of seed sequences of miR-942-3p binding sites lost the above-mentioned responsiveness (Fig 5-G). Overexpression of miR-942-3p reduced quantities of mRNA and protein of MMP9, whereas miR-942-3p inhibitors reversed these condition (Fig 5-H, J). In line with the result of Western Blot assay, miR-942-3p mimics suppressed TCs to express MMP9 protein, and this competitive effect was overturned by miR-942-3p inhibitors (Fig 5-I).
Additionally, we also detected the cell proliferation and apoptosis of TCs after PDGF stimulated in different phases. There was no signi cant in TCs' proliferation between with PDGF(1μL) stimulating group and without group(p 0.05) by CCK8 assay in vitro(Suppl. Fig 1-A,B). To determine whether PDGF affects the apoptosis of TCs, we observed apoptosis bodies of TCs during 120hrs after PDGF stimulating by scanning electron microscopy(Suppl. Fig 1-C). Apoptosis bodies occurred on 48hr and visibly appeared on 72hr. Moreover, the expression of cleaved-Caspas3 and Bax were also measured by Western blot assay because these two protein were special characteristics of cell apoptosis, their quantities could signify the level of apoptosis. TCs, stimulated by PDGF, arose apoptosis from 24hr to 96hr as the same time that TCs without PDGF stimulating expressed apoptotic protein(Suppl . Fig 1-D).

Mutation analysis of Telocytes in the HCC para-cancer tissue
Twenty-three specimens from HCC para-cancer tissues and nine normal specimens from perihepatic hemangioma tissues(as somatic gene control analysis) were enrolled in this study. TCs, which were primarily cultured from these samples, were puri ed to the exon sequence test and analyzed. Low quality reads, junction contamination and reads with high content of unknown base N were removed from the original sequencing data before data analysis(Suppl. Table 3;), and the sample sequencing quality was good as visible by the base quality distribution map(Suppl. Fig 2-A,B ). Approximately 60.46 Mb initial bases on target region was captured by the microarray in our experiment, and the mutative test was performed on this target region. The clean reads of each sample were aligned to the reference genome using BWA [Li,2009], and on average 99.9% of the reads were aligned to the reference genome. After removing duplex reads, an average of 54929003 effective reads (8201.91 Mb, effective bases) were obtained(Suppl. Table 4). The average sequencing depth on target was about 86.82X, and 99.71% target regions were covered by at least 1 reads and 96.74% of the target regions were covered by at least 10 reads per sample(Suppl. Table 4; Suppl. Fig 2-C,D ). The copy number variants(CNV) in telocytes were also shown with suitable fraction of per reads(Suppl . Fig 2-E).
Each mutation was annotated and predicted function in Human Genome Variation Society (HGVS) nomenclature and pathogenic relationship in COSMIC for somatic mutations. According to exon gene sequences, there were seven types of exon mutations in the somatic TCs compared with TCs in the normal hepatic tissues: 3-prime-UTR mutation(15.25%), 5-prime-UTR mutation(1.69%), splice variants(5.08%), missense mutation(38.98%), frameshift mutation(3.39%), upstream-gene variant(8.47%), synonymous mutation(27.12%), respectively (Fig 6-A). There were only four sorts of exon mutations of MMPs family in TCs target regain contrast with the control group: SNP-splice mutation(3.33%), SNPmissense(53.33%), SNP-synonymous variant(36.67%), Indel-frameshift mutation(6.67%), respectively (Fig   6-B). were two exonic biotypes of missense variants, which betided in Chr20 stating at 44642406 and ending at 44642406, of which codon changes were cGg/cCg as well as cAg/cGg(Suppl. Table 2). So we speculated that cAg/cGg SNP mutation in MMP9 sequence was the reason why TCs from HCC paracancer tissue could express more MMP9 protein. One synonymous alteration in MMP9 was recorded as "silent biotype" with a low impact of mutative function and wasn't considered as a promoter involved in cancer metastasis progress by impacting MMP9 expression(Suppl. Table 2). Additionally, three specimens of all CNV(copy number variants) tests had mutative changes in exome gene(Suppl. Table 5; Suppl. Fig 2-F).

Discussion
Globally, HCC ranks fourth in cancer-related mortality and is about to overtake breast, prostate and colorectal cancers as the third most common cancer worldwide [2]. In the Asia-Paci c region, the number of new cases of liver cancer in 2020 will be approximately 467,327 [1]. In China, the incidence of HCC remains high for a long time, with 50% of new cases occurring in China, and most patients are already in the progressive stage at the time of consultation. Notwithstanding, China has made great achievements in modern medical development, the incidence and mortality rates of hepatocellular cancer remain high, due to the disadvantageous clinical stage, low surgical resection rates, and high recurrence rates [2].
Metastasis of HCC, incorporating intrahepatic and extrahepatic metastasis, is another main culprit of lethal for patients, by the way to interdict malignant invasiveness into distant organs extensively improves the overall survival [3,4]. We play attention to demonstrate interactions of HCC behavior and surrounding cells, and Telocyte is a fantastic cell type in the progress of cancer metastasis.
In our study, TCs were veri ed by immunobiology markers CD34, CD117 and PDGFR as well as their morphology on the primary culture as the rst step [10]. On account of multiple sub-classi cations of MMPs, the expression of MMP-2, 3, 9, 11, 14 were tested by Westen blot and qRT-PCR techniques in the HCC tissue and para-cancer tissue, and then we found that MMP-9 expression was higher in the HCC para-cancer tissue. According to multivariate analysis, high density of TCs and MMP9 expression were correlated with poor overall survival. This result was consistent with previous reports that abnormal MMP-9 expression was associated with tumor malignancy [29] and lymphatic metastasis and clinical stages [40] in breast cancer, as well as close relation with the poor prognosis of primary HCC [41]. Combining with our study, high quantities of TCs and MMP-9 were undoubtedly correlated with HCC invasion and migration. To the contrary, the prognostic model of " high telocytes, low OS" is never absolutely appropriate for all types of malignant tumors: in the breast cancer with BRCA1/2 mutations, the phenomenon of TCs decrease were associated with poor prognosis [42].
Telocytes, belonging to a characteristic and novel type of ICLC, possess ability to interact with HCC cells by secreting signal molecule and various proteins. Telopodes originated from TC's body have ability to transmit cellular information to targeted cells by which was emitted from surrounding cells and circulatory system. These functions of telocytes are calculable basics to participate on changing microenvironment metabolism [43]. Therefore we nervily raise a presumption that TCs might product certain type of MMPs proteins to in uence behaviors of tumorigenesis. We found that MMP-9 expression was positive correlation with TCs by Spearman statistic analysis in the HCC para-cancer tissue. TCs were also con rmed to secret MMP-9 in vito. MMP-9 protein have capacity to decompose gelatin and type , and collagens which were the barrier of ECM [44], and takes part in the progress of cancer metastasis [45]. The function of TCs was identi ed in terms of migration and invasion of HCC cells(HepG2, SNU182 and SK-HEP-1 cell lines) by Transwell assay and Wound healing assay, and TCs lost the function with MMP9 inhibitor in vitro and in vivo. Until now, we found the authentic reason to explain that TCs could promote metastasis of HCC through producing and secreting MMP-9.
HCC cells can secret multiple chemokines, growth factors and in ammatory cytokines to impact surrounding cells and change tumor microenvironment(TME), such as vascular endothelial growth factor and PDGF. PDGF is associated with lymphangiogenesis and angiogenesis in gliomas, sarcomas, leukemias and epithelial cancer [46,47]. Chen B proved evidences that PDGF and VEGF expression were implicated with poor prognosis, because HCC cells could secreted them for facilitating cell proliferation, migration and invasion [48]. Based on these facts, we found that PDGF combined with PDGFR which was existed on the surface of TC's cytomembrane and then stimulated TCs to express MMP-9. To illustrate the mechanism of MMP-9 related signal pathway, we refered Kyoto Encyclopedia of Genes and Genomes(KEGG) database to search feasible MMP-9 signal pathways, and determined that PDGF affected MMP-9 expression by ERK signal pathway. By Westen blot assay, PDGF could activate GrB2-sos and Raf in TCs, and contributed to MEK1/2 and ERK1/2 phosphorylation, and nally resulted in MMP-9 expression. This process was obstructed by Raf/ERK diverse inhibitors of AZ-628 and U0126 as a mechanism of down-regulating MMP-9 protein synthesis. As a consequence, we drew a conclusion that TCs promoted the metastasis of HCC through activating ERK signal pathway to express MMP-9 by PDGF stimulating. Therefore, TCs and HCC cells had a consanguineous association, and TCs were considered as a counterpart of HCC in terms of facilitating invasion and migration of cancer cells. Coincidentally, TC hyperplasia comprising the submucosal thickening characteristic of PDGFRA mutant syndrome, was pathogenetically associated to in ammatory broid polyp, and physiological counterpart in the muscularis propria of gastrointestinal stromal tumour [19]. This conclusion and our study jointly elucidate that TCs play an crucial role in the course of oncogenicity and neoplastic circumjacent environment. Inversely, the decrease of TCs also contributed to a genuine brosis in Crohn's disease, Ulcerative colitis and liver brosis [49].
As is well-known that the ERK signaling pathway belonging to a branch of MAPK pathway is a major and fundamental regulator of human being cells proliferation, apoptosis(cell death) and survival [50]. Growth factor receptor activation induces a series of cascade reactions including phosphorylation and excitation.
TCs endlessly accept irritations of diverse growth factor deriving from HCC cells, sustaining sophisticatedly active level of ERK up-regulation, and result in sequential MMP-9 expression which is a vital impetus of invasion an migration of cancer cells. In consideration of multiple functions of ERK signaling for TCs, we detected cell proliferation by CCK-8 staining assay, cell apoptosis by observing apoptotic body under scanning electron microscopy and apoptotic related proteins(Bax and Cleavecaspase-3) by Westen blot method at different phases after PDGF disposing. We investigated that PDGF had no effective in uence on TCs' proliferation and apoptosis because amount of TCs with PDGF group in culture were similar to which without PDGF group, and emblematic apoptosis body of TCs still started to appear at 48hrs. Furthermore, the expression of Bax and Cleave-caspase-3 which belong to apoptosisassociated proteins profoundly con rmed aforementioned results. PDGF activation of Raf/ERK signaling pathway is e cient to accelerate MMP-9 expression of TC and promote HCC cell invasion and migration, whereas rarely effective to TC proliferation and apoptosis.
miRNAs pertain to classi catory gene of noncoding RNAs of 18 to 25 nucleotides that decompose mRNA or inhibit translation by the way to bind to the 3'UTR of their target RNAs [51]. Numerous of miRNAs are veri ed to regulate or control cell biology processes including tumor metastasis, cell division, proliferation and death, directly or indirectly. For instance, miR-128-3p as a tumour suppressor triggers cell cycle arrest by repressing LIMK1 in breast cancer [52], and downregulation of miR-34a reduces HCC metastasis [35].
miRNAs combining to MMP-9 mRNA of TC involve in miR-942-3p, miR-6792, miR-34, and miR-6734 by TargetScan database predicting. In our study, we detected that miR-942-3p in TCs cultured from HCC para-cancer tissue was obviously lower compared with it from normal liver tissue, so we notarized a conclusion that miR-942-3p repressed MMP-9 expression by Luciferase and Westen blot assays as a result of decrease of cell invasion, migration and cancer metastasis. miR-942-3p mimics signi cantly reduced the expression of MMP-9. Downregulation of mi-942-3p was the mechanism of MMP-9 foison and also the springhead for TCs to promote HCC metastasis.
Up to now, we found tow sorts of molecular mechanisms to decipher interactions between HCC cells and TCs, account for divers MMP-9 expressions and the way for TCs to accelerate cancer metastasis. For TCs, activation of Raf/ERK signaling pathway and downregulation of mi942-3p, respectively, played a pivotal role in aggrandizing MMP-9 expression and promoting HCC metastasis. Although miR-942 has been reported to regulate various informational pathways and protein expressions, whether miR-942-3p has the function to impress activation or inhibition of ERK signaling pathway and whether PDGF has ability to suppress miR-942-3p synchronously require more investigations in the further.
The discrepancy between annotated protein-coding genes and human polypetides shows an authentic hypothesis of "one-gene, one-polypeptide". Alternative splicing of gene plays a vital role in protein diversity and protein isoforms complexity [53]. For human beings, up to 95% of multiple exon genes exist mutation to encode proteins during cellular-functions implement [54]. Furthermore, numerous of human hereditary disease and cancers are reported to be correlated with various mutations have capability to change characteristics. MMPs are one of proteins which contribute to the breakdown of extracellular matrix, determining cell migration, proliferation, and metastasis, and MMP-9 is one member of MMPs that is controlled by ERK activity [31].  [55,56] and colon cancer [57]. The MMP-9 -1562CC heterozygous genotype was demonstrated a mark of increased genotype susceptibility to non-small cell lung cancer [58]. Another standpoint on multiple genotypes of MMP-9 SNP was that C/C, C/T and T/T splicing shifts in the pathogenesis and nosetiology of gastric cancer were no signi cant differences [59]. Furthermore, from dissident theory ecumenical genetic variants in MMP-9 were irrelevant with altered susceptibility of breast cancer in a shanghai breast cancer genetic study [60]. Notwithstanding diverse conceptions in regard to the function of MMP-9 genetic mutations, we persistently deemed that SNPs of MMP-9 in TCs played an crucial role in transforming proteinic expression and promoting metastasis of HCC. In our study, we didn't monitor -1562 position shift mutation of MMP-9, but -836A/G, -1721G/C and -1821A/C exon gene variants will be considered novel targets to demystify the mechanism of TCs to accelerate invasion and migration of HCC cells, depending on the MMP-9 expression. These preceding targets should be analyzed by the statistic method with age, sex, clinical stage, distant metastasis, OS and histological type parameters in a large cohort of HCC patients in order to support our theory.

Conclusion
Our study started from nding differences of TCs and MMP-9 expression in the HCC tissues, elucidated two potential mechanisms of TCs in promoting the metastasis of HCC cells and stimulating MMP-9 expression in vitro and in vivo: ) PDGF from HCC cells activates ERK signaling pathway of TC to enhance MMP-9 expression; ) miR-942-3p as a inhibitor suppresses MMP-9 expression. Furthermore, whole exon gene mutations of TCs are tested and analyzed, and MMP-9 related SNPs mutations are detailed detected to explore new targets of TCs.  Table 3 Pearson's correlation of MMP-9 expression and telocyte number in HCC tissues. The analysis implies signi cant positive correlation between the two factors. The correlation between MMP-9 protein and telocytes is signi cant . a is in the cancer tissues(*p < 0.05), and b is in the para-cancer tissues(**p < 0.05). The difference of telocytes and MMP-9 protein expression in the hepatocellular cancer tissue and paracancer tissue. A: CD34 positive telocytes(red) in the HCC tissue and para-cancer tissue(a,e). MMP-9 protein expression in paired-tissues by immunohistochemistry and immuno uorescence assays(b,c,f,g).
The relationship and distribution of telocytes(CD34+,green) and MMP-9 protein(red) in the same pairedtissue by immuno uorescence assay(d,h). B: The accurately quantities of telocytes and MMP-9 protein Telocytes and MMP-9 protein expression in the HCC para-cancer tissues between metastasis group and non-metastasis group. A: MMP-9 protein expression represents distinctly in the HCC tissue with nonmetastasis(a) and metastasis(b) by immunohistochemistry staining. The relationship and distribution between telocytes(green) and MMP-9 protein(red) in the para-cancer tissue are disparate with nonmetastasis(c) and metastasis(d) by immuno uorescence staining . B: The levels of telocytes and MMP-9 protein in the metastasis group(n=93) and non-metastasis group(n=39), and the differences have statistically signi cant. C: Kaplan-Meier analysis of overall survival in HCC patients(n=265) with a low(red line) or high(blue line) level of telcoyte(left panel) and MMP-9 protein(right panel), respectively(p<0.05, Log-rank test). ※※※※p<0.001, ※※※p=0.001 expression of miR-942-3p in TCs between HCC para-cancer tissue and normal liver tissue(E). Predicted binding site between miR-942-3p and the 5' UTR sequence of MMP9, as well as arti cial mutative sites of MMP9(F). The relative binding evident among miR-942-3p, 5' UTR of MMP9 mRNA and mutative type of MMP9 mRNA by luciferase reporter gene assay(G). Measurements of MMP9 incorporating mRNA and protein in TCs after miR-942-3p was overexpressed or inhibited(H, J). Differential expression of MMP9 with miR-942-3p mimics or inhibitors by Westen Blot assay(I). Figure 6