α,1,6-Fucosyltransferase (FUT8) regulates cancer-promoting capacities of cancer-associated fibroblasts (CAFs) through the modification of EGFR core fucosylation in non-small cell lung cancer


 BackgroundCancer-associated fibroblasts (CAFs), the main component of the tumor microenvironment (TME) of NSCLC, are activated by phenotypic transformation into myofibroblasts. α,1,6-fucosyltransferase (FUT8), the key enzyme catalyzing core α,1,6-fucosylation (CF), plays important roles in multiple malignancies. In the current study, we investigated the functions and mechanism of CF mediated by FUT8 in CAFs in NSCLC through bioinformatics analysis, retrospective clinical studies and in vitro/in vivo laboratory experiments.MethodsBioinformatics was used to reveal the relationship between FUT8 and CAFs. Resected specimens, clinical data and prognostic information from 135 NSCLC patients were analyzed to assess the prognostic value of FUT8 in CAFs. Primary CAFs and normal lung fibroblasts were extracted from NSCLC patients and cocultured with NSCLC cell lines in a novel noncontact coculture device produced by 3D printing. In vivo, CAF/NSCLC coinjection tumorigenesis assay was performed in nude mice to study the function of FUT8/CF in TME. The mechanisms of FUT8/CF in CAFs regulating the cocultured NSCLC cells were investigated in cell and molecular experiments. ResultsFUT8 in CAFs is an independent risk factor for prognosis. FUT8/CF in CAFs is essential for CAFs to maintain their ability to promote NSCLC. FUT8/CF in CAFs is responsible for the cancer-promoting capacities of CAFs and lead to a malignant tumor microenvironment. CF modification enhances tyrosine phosphorylation of EGFR in CAFs, which causes activation of downstream signalings of EGFR and maintains cancer-promoting properties of CAFs.ConclusionFUT8 regulates cancer-promoting capacities of CAFs via the modification of EGFR CF in non-small cell lung cancer.


FUT8 regulates cancer-promoting capacities of CAFs via the modification of EGFR CF
in non-small cell lung cancer.

Background
Lung cancer is the most common malignancy in the world [1] , and non-small cell lung cancer (NSCLC) accounts for approximately 85% of lung cancer cases [2] . Despite advances in the diagnosis and treatment of lung cancer, the five-year survival rate is still below 30% [3] . Due to the presence of micrometastasis, even early-stage patients are at risk of metastasis and relapse after radical treatment [4] . Therefore, the mechanism of NSCLC proliferation and metastasis still needs to be further explored, and more accurate prognostic indicators and more effective molecular therapeutic targets need to be identified.
In recent years, crosstalk between the tumor microenvironment (TME) and cancer cells has received increasing attention [5] . The TME is the internal environment in which cancer cells grow in association with "nontumoral" components [6] . Cancerassociated fibroblasts (CAFs), the main component of the TME of NSCLC, account for 70% of the cells in solid tumors [7] and are most often identified by the expression of specific biomarkers such as α-smooth muscle actin (α-SMA) [8] , fibroblast activating protein (FAP) and vimentin [9] . CAFs can promote tumor progression through direct interaction and the secretion of various cytokines [10] . Major studies have 5 demonstrated that CAFs are transformed from normal lung fibroblasts to a myofibroblast phenotype and then acquire cancer-associated properties. Thus, elucidation of the mechanisms of the activation and maintenance of CAFs could lead to novel diagnostic and therapeutic approaches.
Glycosylation is a common type of posttranslational modification of proteins. It is important for the folding, stability, and many of the biologic functions of glycoproteins [11] . Glycosylation can be divided into O-linked glycosylation and Nlinked glycosylation according to the position of the sugar modification, and the latter is more common [12] . Fucosylation catalyzed by fucosyltransferases is one of the most common forms of glycosylation [13] . Among the 13 fucosyltransferases known to be encoded by the human genome [14] , α,1,6-fucosyltransferase(FUT8), which is also known as α1,6-fucosyltransferase, is the only enzyme that catalyzes the core α1,6-fucosylation of asparagine-linked oligosaccharides in mammals [15] .
FUT8 is responsible for the transfer of a fucosyl moiety from guanosine diphosphate (GDP)-fucose to the innermost GlcNAc residue of hybrid and complex asparaginelinked oligosaccharides in glycoproteins via α1,6 linkage to generate the core fucosylation (CF) [16] .This modification was essential for the development and progression of cancer [11,14,[17][18][19][20] , including NSCLC. However, whether FUT8/CF plays an important role in the TME remains unknown.
Epidermal growth factor receptor (EGFR), a member of the ERBB family of receptor tyrosine kinase super-family of cell surface receptors, have received much attention given their strong association with malignant proliferation [21] . EGFR serves as a mediator of cell signaling by extra-cellular growth factors to activate intracellular 6 pathways such as PI3K/AKT/mTOR, JAK/STAT and MAPK/ERK [22] . Deregulation of EGFR signaling is also common in CAFs [23][24][25] . It was reported that the modification of EGFR fucosylation is essential for the activation of EGFR by tyrosine phosphorylation [26][27][28] . Extracellular ligands of EGFR such as epidermal growth factor (EGF) [29,30] , transforming growth factor -alpha (TGF-α) [31] , neuregulin1 (NRG1) [32] and G protein estrogen receptor (GPER) [23] were identified to be essential for the activation and maintenance of CAFs. Moreover, EGFR signaling were reported to be important for the cancer-promoting ability of CAFs in breast cancer [23] . However, whether CF mediated by FUT8 regulates the cancer-promoting ability of CAFs in NSCLC is not elucidated.
Our previous study showed that FUT8/CF is involved in the transformation of pericytes and fibroblasts into myofibroblasts in pulmonary fibrosis [33] , which is very similar to the process in which normal lung fibroblasts are activated to become CAFs. Therefore, we begin with the hypothesis that FUT8/CF in CAFs plays a key role in the activation and maintenance of the cancer-promoting capacities of CAFs in NSCLC. During the study, we found that the activity of the ERBB family and its downstream signaling pathways in CAFs was regulated by FUT8, and further, we identified EGFR as a key protein that was regulated by FUT8/CF. Subsequently, we confirmed the effect of FUT8 by regulating the activity EGFR on CAFs' cancerpromoting capacities through molecular and cell biology experiments. yIn addition, we aimed to explore whether the FUT8/CF levels in the TME serve as a prognostic indicator or potential therapeutic target for NSCLC.  [34][35][36] . Box charts were drawn to show the expression of FUT8 in each dataset. The t-test was used to assess the differences between groups. Committee on Cancer TNM classification [37] was applied to all enrolled patients.

Patient characteristics and tissue preparation
Baseline data and clinicopathological information were collected from the medical Zhe Sun). The total follow-up period was 61 months, and any patient who was lost to follow-up or died of causes other than lung cancer was recorded as censored at the time of the most recent follow-up. The study was approved by the Medical Ethical Committees of the First Hospital of Dalian Medical University. All specimens were 8 obtained from primary lesions, fixed with formalin, embedded with paraffins, and continuously sliced at a thickness of 4 μm.

Immunohistochemistry staining and evaluation of expression level
IHC staining for FUT8 and α-SMA was performed according to the manufacturer's instructions of the products used in our previous studies [38,39] . A streptavidinperoxidase staining kit was purchased from ZSGB (Beijing, China) BIO. The paraffinembedded tissues were dewaxed and rehydrated. The tissues were incubated with primary antibodies diluted to the recommended concentration overnight at 4°C.
3,3′-diaminobenzidine (DAB) staining was performed, and the results were observed under a microscope. Pathological diagnosis was performed according to the 2015 World Health Organization classification of lung tumors [40] and the International Association for the Study of Lung Cancer/ American Thoracic Society/European Respiratory Society International multidisciplinary classification of lung adenocarcinoma [41] . The evaluation of expression was conducted by three professional pathologists separately (Dr. Ling Song, Mr. Sheng Hu and Dr. Jiaqi Qiang). Any dispute was settled by majority rule. The sites and extents of the tumor stroma (mainly consisting of CAFs) were clearly marked by α-SMA staining. In another slice of the same tissue, the expression levels of FUT8 in CAFs were then evaluated according to the methodologies indicated in previous reports for evaluating protein levels in CAFs [42][43][44][45][46] .

Culture of primary fibroblasts
Tissue samples from 9 patients with a definitive p-stage IIIA~IIIB NSCLC were collected at the First Affiliated Hospital of Dalian Medical University from February to May in 2018. The tissues were resected and soaked in DMEM at 0℃ and digested 9 within 2 h in the lab. Tumor tissues and paired normal lung tissues (collected 4~5 cm from the incisal margin) were homogenized and digested for 2.5~4 h at 37℃ in DMEM containing 0.1 mg/mL Roche DNase I (Basel, Switzerland) and 1 mg/mL Roche collagenase A. The cells were filtered with a 75 μm filter and then resuspended and plated with DMEM containing 1% penicillin/streptomycin and 15% GIBCO fetal bovine serum (FBS, Massachusetts, US). The cultures were maintained at 37℃ in 5% CO 2 . After 5 passages, cell purity was tested by RT-PCR. In general, primary fibroblasts extracted in this way can survive for 10 to 14 consecutive passages in vitro. Then, CAFs were immortalized with the SV40-large T antigen (EX-SV40T-Lv105, GeneCopoeia/Funeng, Guangdong Province, China) following the recommended protocol. Finally, 5 paired primary fibroblast cell lines were successfully extracted and cultured and were subsequently cultured in DMEM containing 1% penicillin/streptomycin and 10% FBS.

Cell lines and in vitro cell culture
We selected cell lines for in vitro experiments that could grow in the same medium (Dulbecco's modified eagle medium, DMEM) to observe their interaction in the coculture system. The human NSCLC cell lines A549 and H322 were obtained from the American Type Culture Collection (ATCC, Virginia, US). The human lung fibroblast cell lines HLF, MRC5 and HFL1 were gifts from the Institute of Cancer Stem Cells of Dalian Medical University. All the cell lines were cultured in DMEM containing 1% penicillin/streptomycin and 10% FBS at 37°C under 5% CO 2 .

Noncontact coculture system
A noncontact coculture device was designed (already in the patent examination and approval process in China) and generated by 3D printing (Wanwan 3D, Guangdong Province, China) with highly transparent nontoxic resin to simulate the internal environment. The reliability of the device was verified in our recent studies [47] . The schematic diagram of the device used for 3D printing is shown in Figure 3C, and a modified version that was easier to produce was used in this study ( Figure 3D). The device was a vessel consisting of two culture wells and a precipitation well. The culture wells and the precipitation well were separated by two 2 mm-high partitions.
The efficiency of solute exchange was tested with a standard protein solution by measuring the relationship between concentration changes and time in the two culture wells.
During cell seeding, the levels of the liquid in both culture wells should be kept below the height of the partition. Fibroblasts and NSCLC cells were seeded into the two culture wells, at a 2:1 ratio. After cell adherence, DMEM containing 10% FBS was added to the vessel until the level was above the partition. Based on the design of the coculture device, floating cells do not go directly into the contralateral culture well but are deposited into the precipitation well. Before the cells were manipulated in any way, the medium in the precipitation well was drained in case any floating cells had entered the contralateral culture well. Then, the two cell lines were isolated again, and 0.5% trypsin was used for the separate passage/collection of the two cell lines.

Conditioned medium
Fibroblasts were cultured in DMEM with 10% FBS and routine antibiotics. The medium was changed to FBS-free medium when the cells had reached 80% confluence. After 48 h, the supernatant was extracted and centrifuged at 1600×g for 10 min. The conditioned medium (CM) was stored at -79°C and used for the colony formation assay.

Reagents and antibodies
Primary antibodies against α-SMA and FUT8 were purchased from Abcam Lens culinaris agglutinin, a lectin that can specifically identify fucose, was purchased from Vector Labs (California, USA). NSC 228155, an activator of EGFR which binds to the extracellular region of EGFR and enhance tyrosine phosphorylation of EGFR [48] , was purchased from MedChemExpress (New Jersey, USA). Other chemicals and reagents were purchased from Sigma (St. Louis, USA) and Vetec (St. Louis, USA).

Plasmids and lentivirus
Four mCherry labeled shRNA fragments targeting the FUT8 gene, one GFP labeled expression clone fragment encoding the full length of the FUT8 ORF sequence and the corresponding negative control fragments were purchased from GeneCopoeia.

RNA extraction and RT-PCR.
Total RNA was extracted from CAFs and their paired normal lung fibroblasts (NLFs) using the TRIzol method according to previous reports [51] . A TransScript One-step gDNA Removal and cDNA Synthesis Supermix purchased from Transgene (Beijing, China) was used to synthesize first-strand cDNA. Taq Master Mix (Vazyme , Nanjing, China) was used for quantitative PCR according to the manufacturer's instructions.
The reaction program was set according to the Tm values of the primers. Primer sequences were as follows: vimentin-F 5'-TGCCGTTGAAGCTGCTAACTA-3', vimentin-R

Western blot analysis and lectin blot analysis
Total protein was collected according to the methods described in our previous report [38] . The protein concentration of the cell lysis solution was assessed using a 13 Thermo Fisher BCA kit. Then, 20~30 μg of total protein was run in 10% SDS-PAGE and transferred to a PVDF membrane. The samples were incubated with primary antibodies diluted to the recommended concentration overnight at 4°C. The samples were incubated with an HRP-conjugated secondary antibody at room temperature for 2 h and protein bands were detected with a chemiluminescence device. The lectin blot protocol was similar to that of the western blot, except biotinylated Lens culinaris agglutinin was used instead of primary antibodies.

Colony formation assay
Briefly, A549 cells or H322 cells were seeded into six-well plates (2 × 10 3 per well) and incubated in complete DMEM for 24 h. Then, the medium was replaced by CM with 10% FBS, and the cells were cultured in a 37°C incubator with 5% CO 2 for 2 weeks until they grew into macroscopic colonies. Finally, the medium was removed, and the cell colonies were stained with 0.1% crystal violet and counted. were stained with 0.1% crystal violet and the numbers of cells were recorded using a microscope.

Wound-healing assay
After 5 continuous passages of coculture, a wound-healing assay was used to test cell migration ability. NSCLC cells and fibroblasts were grown to 80%-90% confluence in the coculture plates and incubated overnight in FBS free DMEM.
NSCLC cell monolayers were wounded with a sterile 100 µL pipette tip and washed with FBS free DMEM to remove detached cells from the plates. After 48 h, the medium was replaced with PBS. The wound gap was observed and photographed using a microscope, and the distance of the wound gap was measured.

Immunofluorescence staining
For immunofluorescence staining, cells grown on chamber slides were washed with PBS and fixed with 4% paraformaldehyde for 15 minutes at room temperature. The samples were pretreated with Triton X-100 and incubated with 10% BSA in PBS for 30 minutes. The primary antibody against FUT8 was added to the samples, followed by incubation overnight at 4°C. Fluorescein isothiocyanate-and rhodamineconjugated secondary antibodies and red luminous LCA were added and followed by incubation of 1 h at room temperature. DIPI was added to stain the nuclei. The samples were examined with a fluorescence microscope.

Immunoprecipitation (IP) assay
Lysates of fibroblasts were mixed with EGFR or IgG antibodies and placed on a 4℃ table shaker for overnight. Then, 40 ug protein A/G magnetic nanospheres (Santa Cruz, Texas, USA) were added in the mixture and were shaken at 4℃ for 4 h. EGFR proteins were extracted following the recommended protocols and analyzed by lectin blotting described above.

In vitro coculture assay
To obtain an appropriate initial ratio at which combined CAFs/NSCLC cells could grow continuously and maintain a proper constitution that best simulates the TME in direct contact coculture, an in vitro coculture assay was conducted. GTP (green)labeled CAFs and mCherry (red)-labeled A549 cells were combined in test tubes and seeded in 10 cm dishes for direct contact coculture. After continuous culture for 5 passages, the cells were observed under a fluorescence microscope. The stable cell ratios were calculated as a reference to select the initial cell ratio for in vivo coinjection.

Bioinformatics and gene enrichment analysis
The data used in the analysis came from the GSE22862 gene chip from the NCBI Gene Expression Omnibus (GEO) and were obtained from CAFs from the tumor tissues of 15 primary NSCLC cases. The differential expression of genes was calculated using the R pack limma. Enrichment analysis was conducted for the genes that were significantly coexpressed with the FUT8 gene (P<0.05) the using R package Cluster-Profiler.

Statistical analysis
Statistical data were analyzed using IBM Statistical Product and Service Solutions 25.0. The differences in clinical characteristics associated with different FUT8 expression levels were evaluated using the Chi-square Test and the Fisher Exact Probability Test. To adjust the heterogeneity of the follow-up period, survival curves were plotted on the basis of Kaplan-Meier Survival Analysis [52] , and the differences between the curves were analyzed by the Log-Rank Test [53] . The Cox's Proportional Hazards Model was applied to analyze the significance and independence of the influence of all the studied clinical and pathological factors on both tumor recurrence and cancer death risk during the follow-up period [54] . As dummy variables were established to analyze polytomous variables, the method whereby variables entered the Cox's equation was limited to the 'Enter' method, which means that all the variables entered the equation at the same time. The difference was considered to be significant at P<0.05.  Figure 1B~1H. Totally 963 samples were analyzed in the 7 datasets, showing that FUT8 was significantly overexpressed in adenocarcinoma rather than in any other types of lung cancer or normal lung tissues. The 7 datasets showed that FUT8 was expressed at 1.626~2.569 folds higher in lung adenocarcinomas than that in normal lung tissues.

FUT8 was overexpressed in the
To confirm these findings, FUT8 was tested by western blotting in 5 lung adenocarcinoma tissues and paired normal lung tissues. FUT8 was overexpressed in the tumors compared with the normal tissues in all 5 cases ( Figure 1I). In the IHC staining analysis of the 135 NSCLC cases, 20 cases involving solid predominant, mucinous and micropapillary predominant lung adenocarcinomas were excluded because of failure to separate the tumor stroma from the cancerous epithelium 18 (representative images are shown in Figure 1J). FUT8 expression was successfully evaluated in 115 cases, and representative images are shown in Figure 1L. It was found that FUT8 was overexpressed in tumor cells in almost every case, and in tumor stroma of 62 out of the 115 cases.

FUT8 expression in CAFs correlates with long-term prognosis
The baseline information and characteristics of the 115 patients are shown in Table   1. The relationship between FUT8 expression and clinicopathological factors is shown in Table 2. We found that the ratio of  (Table 3). Furthermore, Cox's multivariate analysis showed that the T stage, N stage and FUT8 expression of CAFs can all serve as independent risk predicting factors for both overall and disease-free survival (Table 4).  [55][56][57] , no significant differences were observed in the current study. The isolated primary fibroblasts were identified on the basis of specific markers of myofibroblasts ( Figure 3A),

FUT8 is overexpressed in CAFs and mediates high levels of CF
showing that phenotypic differences existed between CAFs and NLFs. Western blotting showed that FUT8 was overexpressed in 4 out of 5 CAFs (P01, P03, P07, P08) compared with the corresponding NLFs ( Figure 3B). lectin blotting (LCA) showed coincident results of increased CF in the 4 pairs of primary fibroblasts ( Figure 3C). As a result, The CAFs of P03, which probably showed the most significant difference in CF between CAFs and NLFs, were selected for the following study.
FUT8 and ST6Gal1 (key enzyme for sialic acid glycosylation) were tested by western blotting in CAFs and three frequently-used lab-grown human normal lung fibroblast cell lines (MRC5, HLF1 and HLF, Figure 3D, 3E). FUT8 and ST6Gal1 together catalyze most glycosylation modifications in mammals [58] . These results indicated that the general glycosylation level in CAF is higher than that in normal lung fibroblasts.

FUT8 in CAFs was necessary for the construction of an invasive tumor microenvironment in vivo
Four shRNA fragments were used for downregulating FUT8 in CAFs and the effects of gene intervention were tested by western blotting ( Figure 4A). Sh4 was used for producing the stably shFUT8-expressing CAFs. Lectin blotting showed that CF levels were downregulated in CAFs-shFUT8 ( Figure 4B).
For animal pre-experiments, an in vitro coculture assay was conducted. In previous articles, inappropriate coinjection ratios often resulted in the absence of fibroblasts or too few tumor cells in the generated tumors. Thus, it is important to investigate a proper ratio for the cell mixture. According to the results, a ratio of CAFs:A549 = 2:1, which stably mimics a TME similar to that of primary lung cancer, was selected as the initial ratio for subcutaneous coinjection into nude mice ( Figure 4C). After 28 days of observation, the tumors were removed ( Figure 4F) Furthermore, IHC staining showed the interior structure of the reconstructed tumors. The sites and extents of the stroma were clearly marked by α-SMA staining.
It was interesting to find that the cancerous epithelium was more mixed with the stroma in the CAF-NC/A549 tumors. In the CAF-shFUT8/A549 tumors and HLF/A549 tumors, the tumor cells preferred to clump together rather than invade the stroma ( Figure 4H). This finding was most obvious at the junctions between the epithelial tissues and the stroma indicated by white and black arrows in Figure 4H. Cancer cells rarely invaded the stroma in the CAF-shFUT8/A549 tumors. The invasion of cancer cells was slightly more obvious in the HLF/A549 tumors and was most obvious in the CAFs-NC/A549 tumors.

CF mediated by FUT8 is necessary for maintaining the protumorigenic capacity of CAFs
FUT8 was upregulated in the stably FUT8-overexpressing HLFs ( Figure 5A). Lectin blotting showed that CF levels were upregulated in CAF-FUT8 OE cells ( Figure 4B). Non-contact coculture system was successfully established. A concise 3D printing illustration of the prototype coculture device design is shown in Figure 4F, and the improved design of this device used in the current study is shown in Supplementary   Figure 1C. Evaluation of the efficiency of solute exchange showed that the diffusion time of the standard protein solution was 40 to 50 minutes (Supplementary Figure   1D).As the change in migration abilities was significant, the results regarding cell migration were reconfirmed in the wound healing assays using the non-contact coculture system ( Figure 4G). CAFs led to blockage of the G1/S checkpoint in NSCLC cells in the non-contact coculture system, as the ratio of cells in G0/G1 increased, while the ratio of cells in S phase decreased. However, no change in the regulation of cell cycle of NSCLC cells could be observed when FUT8 was upregulated in the cocultured HLFs ( Figure   5H). The results implied that G2/M blockage had probably occurred as diploids were observed in H322 cells and part of A549 cells (yellow peaks); however, this was not supported by enough molecular evidence (data not shown). The results implied that FUT8/CF was necessary but insufficient for lung fibroblasts to acquire the capacity to promote the cell cycle of NSCLC cells.
To explain these findings, signaling pathways of NF-kB, MAPK/ERK1/2, Biomarkers of EMT and cell cycle G1/S checkpoint in A549 cells in the non-contact coculture system were tested by western blotting. (Figure 5I~L). The results showed that the downregulation of FUT8 in CAFs led to dramatic inhibition of these signalings in NSCLC cells in the non-contact coculture system, while HLF-FUT8 OE cells didn't show a matching increase in these signalings. This result is in accord with the change in proliferation, migration, invasion, and cell cycle of NSCLC cells in the noncontact coculture system, and supports the fact that high-FUT8/CF CAFs help to construct a more invasive TME. However, this finding also suggested that only upregulation of CF modification in HLFs cannot make them to become as protumorigenic as CAFs.

Potential mechanisms of FUT8 maintaining the tumor-promoting phenotype of CAFs
Although it is believed that CAFs most likely promote the growth of tumors in a paracrine manner, we noted that it was also important that CAFs maintained their cancer-promoting phenotype in dynamic symbiosis with tumor cells. To explain the potential mechanisms by which FUT8 acts to maintain the tumor-promoting were increased significantly in FUT8 high-expressing CAFs ( Figure 6F).

CF mediated by FUT8 regulates cancer-promoting capacities of CAFs via the modification of EGFR fucosylation
As EGFR and its downstream signaling molecules are mostly activated by phosphorylation modifications, it was essential to reconfirm the regulation of FUT8/CF modification on EGFR and its downstream signalings. EGFR, p-EGFR and FUT8 in the 5 pairs of primary CAFs/NLFs were tested by western blotting ( Figure   7A). The expression of EGFR showed no change between CAFs and NLFs, while p-EGFR was consistent with the expression of FUT8. Four shRNA fragments were used for downregulating FUT8 in CAFs and the effects of gene intervention were tested by western blotting (Figure 7B). Sh2 was used for producing the stably shEGFR CAFs.
The binding of fucose to EGFR was tested by immunoprecipitation and lectin blotting ( Figure 7C). EGFR was successfully enriched and the fucose binding to EGFR was consistent with the expression of FUT8, which reconfirmed the fucoglycosylated modification of EGFR. Rescue experiments were conducted respectively in CAFs and HLFs, and the results are shown in Figure 7D, E, which strongly support that EGFR and its downstream signalings are regulated by FUT8/CF modification. Further, clone formation assay showed that downregulation of EGFR tyrosine phosphorylation led to reduced cancer-promoting capacity of CAFs( Figure 7F), and counteraction of the cancer-promoting capacity of FUT8 overexpressing HLFs ( Figure 7G). properties of CAFs [59] . In the current study, we aimed to investigate the effect of Classic models include the use of conditioned medium, the Transwell system [60] , the Boyden chamber system [61] , and xenoanimal coinjection models [62] . In recent years, microfluidic chip technology [63] and the patient-derived xenograft mouse model [64] (PDX) have also become common research tools. In the current study, we introduced a novel noncontact coculture device, that overcomes the limitations of previous models in terms of cell quantities and passages, and its practicality was experimentally verified.

Discussion
Consistent with previous reports, the current study showed that FUT8 was overexpressed in NSCLC tissues. Moreover, FUT8 was overexpressed not only in cancerous epithelial cells (NSCLC cells) but also in the tumor stroma (CAFs) in more than half of the NSCLC cases. Although the FUT8 overexpression in CAFs was not as high as that in NSCLC cells in these cases, it was clearly higher than in the paired normal lung tissues. Based on our finding in previous studies that CF mediated by FUT8 plays a key role in the transformation of pericytes and fibroblasts into myofibroblasts, we assumed that FUT8 functions similarly in the activation and maintenance of CAFs. More importantly, we confirmed FUT8 expression in CAFs as an independent risk factor for long-term prognosis in 115 patients who were retrospectively observed. These results implied that patients with high FUT8 in CAFs likely had tumors that were so aggressive and metastatic that they had already formed micrometastases that could not be detected by screening before surgical treatments.
To investigate the effects of FUT8 on CAFs, we extracted primary CAFs and NLFs 26 from very aggressive NSCLC with a clinical stage of at least IIIA. For example, the CAFs used in the functional and mechanism experiments (P03) were extracted from a female patient with stage IIIA lung adenocarcinoma. It was found that FUT8 and CF were upregulated in more than half of those CAFs compared with their paired NLFs.
In addition, the level of CF was largely consistent with the expression of FUT8 in these primary cell lines. It was interesting that secretory vesicles containing abundant FUT8 were observed in CAFs rather than in NLFs or lab-grown HLFs. This implied that the fucosylation of secretory proteins was much more active in CAFs compared with NLFs and HLF.
In the current study, the silencing of FUT8 in CAFs and the upregulation of It was showed by the GSEA that the low expression of FUT8/CF led to inhibition of both extrinsic and intrinsic apoptosis pathways. There has been a debate about the effect of FUT8/CF on cell apoptosis in different types of cell and tissue as some researchers reported promotion [66][67][68] , and one reported inhibition [69] . However, whether it was the reason of the inhibited capacity of FUT8 low-expressing CAFs remains unknown. Moreover, the inhibition of the ErbB pathway may be more important for the inhibited cancer-promoting capacities of FUT8 low-expressing CAFs, as it blocked multiple growth factor signals [70] that were previously proved to be necessary for the activation and maintenance of CAFs, including EGF, TGF-α, NRG1 and so on. The unchanged gene copy numbers of the ErbB genes implied that the activity of these genes was not regulated at the level of transcription and translation. As was reported in previous articles that the EGFR can only function if it is properly glycosylated [28,71] , it is considered that the downregulation of FUT8

Consent for publication
Not applicable

Data availability statements
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
No conflict of interest exits in the submission of this manuscript, and manuscript is approved by all authors for publication.The work described in this paper was original research that has not been published previously, and not under consideration for publication elsewhere, in whole or in part.

Funding
This work was supported by grants from the National Natural Science Foundation of China (81670063) .