Interleukin 35 Activates Intratumor Neovascularization Via Enhanced Secretion of FGF2 in Hepatocellular Carcinoma Through The Recruitment of Neutrophils, and Blocking it Could Facilitate The Ecacy of The PD-1 Antibody

Background: Recently, more and more treatment strategies for Hepatocellular carcinoma (HCC) have emerged, but the therapeutic effect is still not satisfactory. This study is aimed to explore the mechanism of Interleukin 35 (IL-35) in promoting the progression of liver cancer and to explore the application value of IL-35 in the treatment of HCC. Methods: We used clinical tissue microarray (TMA) immunohistochemistry (IHC) to explore the prognostic value of IL-35 expression in patients with HCC. The effect of IL-35 on the function of HCC was explored by functional experiments including wound-healing assay, transwell, cell counting kit-8, cell adhesion assay and endothelial tube formation assay in vitro and mouse xenografts in vivo. And ow cytometry was used to study the effect of IL-35 on inltrating immune cells in tumor. The molecular mechanism of the function of IL-35 on the progression of HCC was explored by sequencing, ELISA, WB, PCR and other technical means. Finally, through in vivo tumor animal experiments to explore the value of anti-IL-35 antibody and combined with anti-PD-1 antibody in the treatment of liver cancer. Results: High expression of IL-35 in patients with HCC were identied to be associated with poor prognosis. And we have found that IL-35 facilitated tumor progression by affecting neutrophil inltration, angiogenesis, and CD8+ T-cell inltration in a mouse model. Additionally, on the one hand C-C motif chemokine ligand 3 (CCL3) has been found to be a key factor mediating the recruitment of neutrophils by IL-35, on the other hand broblast growth factor 2 (FGF2) secreting by neutrophil when stimulated by IL-35 was also found to be the core cytokine to promote intratumoral angiogenesis. And IL-35 was also discovered to facilitated the adhesion of tumor to endothelial cells, neutrophils further enhancing this in vitro and vivo. important, anti-IL-35 antibody was a valid for HCC in xenograft tumor model, and it could give full play to the curative effect of 1:1(cid:0)2 when combination therapy with PD-1 antibody. Conclusion: Our data show that the expression of IL-35 in patients with HCC is an important tumor promoting factor. The application of anti-IL-35 antibody and treatment combined anti-IL-35 antibody with anti-PD-1 antibody have potential therapeutic value in the treatment of liver cancer. loops in the tube-like quantied in 24 per


Background
Hepatocellular carcinoma (HCC), the most common type of primary liver cancer, is the fourth most common malignant tumor and third in mortality in China [1,2] . In recent years, radical treatments, such as surgery (hepatectomy and liver transplantation), radiofrequency or microwave ablation of liver tumors have made great progress, resulting in the signi cant improvement of the therapeutic effect. However, the rate of tumor recurrence and metastasis in the 5 years after operation has been shown to be as high as 60 to 70 %, with a lack of effective drugs [3][4][5] . Targeted drug therapy represented by the administration of sora nib, lovatinib, and rego nib was reported to provide an improvement of only 2.8 months in the overall survival of patients with advanced HCC [6,7] . At present, immunotherapy, including the use of programmed cell death protein 1 (PD1) and programmed cell death protein ligand 1 (PD-L1) antibodies, as well as other immune checkpoint antibodies has attracted much attention. However, the effective rate was shown to be only about 15 to 20 % in patients with HCC, and thus there is an urgent need to nd a more effective strategy for the treatment of liver cancer [8][9][10] . Fortunately, the combination of immune checkpoint inhibitors and antiangiogenic antibodies has been shown to be extremely effective in the treatment of liver cancer, resulting in revolutionary therapeutic results [11] . We thus aimed to explore more effective treatments for HCC. IL-35 which was rst reported in 1997, is a new member of the cytokine IL-12 family, consisting of 2 subunits, P35 and EBI3 [12,13] . It is known to be primarily expressed by Treg and Breg cells, and also secreted by macrophages, dendritic cells (DCs), monocytes, endothelial, and CD4+HLA-G+ cells. It was con rmed that IL-35 has a strong immunosuppressive function and plays an important role in in ammation, infection, autoimmune diseases, transplantation immunity, tumor, and other diseases [14][15][16][17][18] . It has also been reported that tumor cells, including those from pancreatic cancer, HCC, cholangiocarcinoma, nasopharyngeal carcinoma, breast cancer, and melanoma could also synthesize and secrete IL-35, playing an important role in carcinogenesis and malignant biology [14,[19][20][21][22][23] . Its speci c mechanisms of action have been demonstrated to be the following: (1) promoting tumor cell proliferation and inhibiting tumor cell apoptosis [24] ; (2) inducing tumor angiogenesis [23] ; (3) promoting tumor metastasis by increasing the adhesion of tumor cells to endothelial cells and transendothelial migration [25] ; (4) reversing the epithelial-mesenchymal transition (EMT) of tumor cells to promote tumor colonization in the target organ during metastasis [18] ; (5) upregulating the expression of immune checkpoint proteins, such as PD-1, Lymphocyte-activation gene 3 (LAG3), and TIM3 in the tumor microenvironment (TME) and promoting the apoptosis of cytotoxic T-lymphocytes (CTL) [26,27] ; (6) recruiting myeloid inhibitory cells to form an immunosuppressive network, and reducing the sensitivity of tumor cells to CTL cells [23] ; and nally (7) promoting the differentiation of neutrophils towards the N2 tumor phenotype by enhancing the secretion of interleukin 6 (IL-6)and Granulocyte Colony-Stimulating Factor (G-CSF) [28] . Generally speaking, IL-35 is known to directly enhance the proliferation and adhesion of tumor cells. In contrast, it has also been shown to promote tumor angiogenesis and maintain an inhibitory TME by reshaping the local tumor microenvironment. Overexpression of IL-35 has been reported to be closely related to poor prognosis in a variety of tumors, including HCC. Early intervention screening of high-risk patients for overexpression of IL-35 might become a new treatment strategy for the prevention and treatment of the postoperative recurrence and metastasis in patients with HCC. However, at present, the mechanism of IL-35 in HCC is not clear. Therefore, in this study we systematically explored the mechanism of the effect of IL-35 on the malignant biology of liver cancer.

Methods
Cell lines, mouse xenografts, human samples, and collection of patient clinical data LO2 and the HepG2, Huh7, PLC, Hep3B, MHCC97L, MHCC97H, and HCCLM3 HCC cell lines were obtained from the Liver Cancer Institute, Zhongshan Hospital, Fudan University. All cell lines were cultured in a reasonable conditioned media at 37 °C in a humidi ed incubator containing 5 % CO 2 , as previously described.
Male C57BL/6 and BALB/c-nu/nu mice (4-6 weeks old, from Shanghai Institute of Material Medicine, Chinese Academy of Science) were housed under pathogen-free conditions. Animals were cared for in accordance with the guidelines established by the Shanghai Medical Experimental Animal Care Commission. All experimental protocols were approved by the Zhongshan Hospital Research Ethics Committee. All experimental procedures involving animals were approved by the Animal Care and Use Committee of Zhongshan Hospital, Fudan University, China.
A total of 360 patients with primary liver cancer who underwent radical resection at Zhongshan Hospital a liated to Fudan University in 2012 were selected for this study. The obtained samples were con rmed to be HCC based on pathological assessment. Clinicopathological data, such as liver function, alpha-fetoprotein (AFP), carbohydrate antigen 19-9 (CA19-9), carcinoembryonic antigen (CEA), tumor size, number and differentiation, lymph node metastasis, vascular invasion, cancer thrombus formation, and microvascular invasion (MVI) were collected from the medical history and pathological reports of enrolled patients.

IHC and immuno uorescence
The P35, EBI3, GP130, IL-12 Rβ2, P40, and P28 antibodies were used to detect the level of protein expression in tumor tissues using IHC, based on the experiment protocol used in our previous study.
A number of pathology teachers were invited to evaluate the results, and H-scores were used to quantify P35, EBI3, GP130, IL12Rβ2, and other indexes. According to the calculation score of intensity and area, the intensity of positive staining was divided into 4 grades: 0, no staining; 1, weak staining; 2, moderate staining; and 3, strong staining. This was divided into 4 levels according to the area of colored cells: 0, <1 %; 1, 1-25 %; 2, 26-50 %; 3, >50 %. Dyeing was divided into the score of the intensity area, with the score of uneven dyeing requiring to be accumulated. The comprehensive score was divided into 4 grades: negative (-), 0 points; weak positive (+), 1-2 points; medium positive (+ +), 3-4 points; and strong positive (+ +), 5-9 points. Immune cells, such as CD4+, CD8+, CD34+, CD68+, neutrophils (CD66b+), and FOXP3+ cells were used to evaluate the level of in ltration. Five typical visual elds were randomly selected on the patient chip, and the average value was calculated as the immune cell count of the patient. The cutoff value of the statistical score was calculated using the X-tile software. The results of the chip were independently evaluated by 2 professionals, and any contradictory results were discussed and a joint decision was made.
Immuno uorescence was performed on the HCC cell line, where we explored the expression level of EBI3 and P35. Slides were prepared in the same manner as for IHC before incubation with antibodies. After that, incubation with anti-mouse and anti-rabbit antibodies was performed.
Western blotting (WB) was performed as previously described. Brie y, we generated total cell lysates and proteins were separated on SDS-PAGE gels, and then transferred to polyvinylidene di uoride (PVDF) membranes. Membranes were washed and blocked. Next, membranes were incubated with primary and secondary antibodies, detected using enhanced chemiluminescent (ECL) substrate, and processed using the Image Lab software.
Coimmunoprecipitation (CO-IP) was performed as previously described [29] . Brie y, both input and IP samples were analyzed by WB using various antibodies at the indicated dilutions.
ELISA kits were used to detect the levels of secretory IL-35, CCL3, and FGF2 in the supernatants of cell cultures according to the manufacturer's instructions.
Quantitative reverse-transcription PCR was performed using a SYBR PrimeScript reverse-transcription PCR Kit (Takara Bio, Shiga, Japan) in accordance with the manufacturer's instructions. The RT2 pro ler array was probed using the Pro ler PCR Array System and SYBR green/ uorescein quantitative PCR Master Mix (SABiosciences) in an ABI 7900 sequence analyzer (Applied Biosystems, Carlsbad, CA, USA) in accordance with the manufacturer's protocol.
Wound-healing assay and Transwell Cell Counting Kit-8 Analyses of cell proliferation were performed using the Cell Counting Kit-8 (CCK-8; Dojindo). Wound-healing assay and noncoated transwell chambers (BD Pharmingen) were used to assess the migratory ability of cells. The invasion ability of cells was assessed using Matrigel-coated transwell chambers (BD Pharmingen).
Cell adhesion assay and endothelial tube formation assay Human umbilical vein endothelial cells (HUVECs) were cultured with ECM, in slices, and allowed to reach full con uence. Then, the monolayers were stained with a CytoPainter Cell Tracking Staining Kit (Abcam) and inserted into a ow chamber. HCC cells were labeled with Calcein AM (Donjindo) according to the manufacturer' s instructions. Brie y, 1 × 10 5 labeled HCC cells were added to HUVEC cells and cocultured for 2 h at 37 °C in a cell incubator. After that, the culture medium was removed, and the unattached tumor cells were washed twice with PBS before being replaced with complete culture medium. Adhered tumor cells on the monolayers were imaged using a uorescence microscope and their average number was calculated in 5 random elds.
In vitro endothelial tube formation assays were performed using collagen gel. First, Matrigel was kept melt on ice, and then 250 mL of melted matrix glue was gently poured into each hole of a precooled 24-well plate to avoid bubbles. The plate was then placed into a 37 °C incubator for 35 min for solidi cation. HUVECs (8 × 10 4 cells/well in a 24-well plate) were cultured on the surface of collagen gel using endothelial cell medium (ECM) and other conditioned media. The total tube length, the number of branch points, and the number of loops in the tube-like structures were quanti ed in 24 elds per group.

Statistical analysis
Statistical analyses were performed using the IBM SPSS and GraphPad Prism. Each experiment was performed in triplicate, and values are presented as the mean ± SD, unless otherwise stated. The variance between the groups was statistically compared. The distributions of both the overall survival (OS) and recurrence-free survival (RFS) were depicted using the Kaplan-Meier method and analyzed by the log-rank test. Univariate and multivariate analyses for prognostic factors were based on the Cox proportional hazard model. P < 0.05 was considered signi cant.

Results
Co-expression analysis of P35 and EBI3 subunits in hepatocellular carcinoma could be used as a proxy for the level of IL-35.
At present, there is no high-quality speci c antibody for IL-35. So in order to investigate the level of IL-35, we detected the expression of P35 and EBI3 in continuous tissue sections of HCC samples, which has been reported to be a reliable method in other research studies [22,25] . As IL-12, IL-27, and IL-35 are known to share common subunits, we detected all 4 subunits, namely P35, EBI3, P40, and P28 in TMA of patients with HCC. The P35 and EBI3 proteins were observed to be mainly located in the cytoplasm, widely expressed in tumor cells, as well as in stromal cells. Whereas the expression of P35 and EBI3 was demonstrated to be lower in paracancerous compared with cancerous tissues of the same patient, we found that the expression level of both P28 and P40 in HCC was very low, and mainly located in stromal cells FIG1.A,B .
The proportional structure expression map of EBI3 and P35 was shown to be highly similar, but obviously different from that of P28 and P40. We further observed that the staining scores of EBI3 and P35 were strongly correlated (r = 0.698, P < 0.001). However, there was a poor association between EBI3 and P28 (r = 0.08, P < 0.05), as well as between P35 and P40 (r = 0.042, P < 0.001) (FIG1.D).
Furthermore, we used the HCCLM3, MHCC97H, SMMC-7721, and HUH7 cell lines to carry out cellular immuno uorescence experiments. We accordingly found that P35 and EBI3 were mainly expressed in the cytoplasm, and not in the nucleus. The expression intensity of P35 and EBI3 was shown to be basically the same, with the spatial expression sites basically overlapping. The expression intensity of these 2 proteins was found to be higher in HCCLM3 and MHCC97H, but low in SMMC-7721 and HUH7 (FIG1.C). In order to eliminate the effect of the cosubunits of IL-12 and IL-27 on the detection of IL-35, we used the CO-IP technique to study the structural relationship of the 4 subunits in 4 cases of HCC. The P35 antibody was demonstrated to successfully immunoprecipitate EBI3 but rarely bound to P40 and P28. Similarly, the same phenomenon was observed in CO-IP experiments in the MHCC97H and HCCLM3 cell lines (FIGS1.B).
Overexpression of IL-35 in hepatocellular carcinoma was an independent risk factor for prognosis.
We then analyzed the baseline characteristics of the 360 patients with HCC. Our study included 202 men (56.1 %) and 158 women (43.9 %), with an average age of 54.36 ± 11.038 (26-85) years and a median age of 54 years. The average overall survival (OS) time was 51.87 ± 0.996 months, and the median OS was 49 ± 7.19 months. The average recurrence-free survival (RFS) time was 41.91 ± 1.27 months, and the median RFS was 22.0 ± 1.837 months (Table1.).
The IHC staining score ++/+++ was strati ed as high expression. The high expression of both P35 and EBI3 was de ned as the IL-35 high expression group (41.6 %), whereas others were classi ed as the IL-35 low expression group (58.4 %) (FIG1.D). Signi cant correlations were found between the high expression of IL-35 and advanced Barcelona clinic liver cancer (BCLC) stage. Moreover, we also found that the level of serum alpha-fetoprotein (AFP) was signi cantly increased in the IL-35 overexpression group (637.45 ± 32.8 vs 212.47 ± 18.9 ng/mL, P < 0.05). In addition, we noted that overexpression of IL-35 was closely related to an increased prevalence of portal vein tumor thrombus (PVTT), microvascular invasion (MVI), and large tumor size (P < 0.001). Multivariate analysis showed that overexpression of IL-35 was an independent risk factor for both OS (HR = 1.947; 95 % CI, 1.046-3.624, P = 0.035) and RFS (HR = 2.442; 95 % CI, 1.459-4.088) (Table2.). Therefore, the expression of IL-35 in HCC was considered an important reference index for judging prognosis.
Binding of cytokines to receptors is known to be an important link for the functional role cytokines play in cells. After binding to receptors, IL-35 is known to activate the intracellular signal transduction pathway. Therefore, it is of great signi cance to explore the expression of the GP130 and IL-12 Rβ2 receptors of IL-35 in HCC. At present, there has been no report on the expression of receptors of IL-35 in HCC. We found that the expression of GP130 and IL-12 Rβ2 in HCC was closely related (r = 0.39, P = 0.023)(FIG2B). Based on the expression of IL-35 and its receptors in HCC, we divided the 360 patients into 4 groups: IL-35R (+) IL-35 (high); IL-35R (+) IL-35 (low); IL-35R (-) IL-35 (high); and IL-35R (-) IL-35 (low). We found that patients in the IL-35R (+) IL-35 group (high) had the worst prognosis (P < 0.001), thus supporting the hypothesis that IL-35 facilitated the progression of HCC by directly acting on tumor cells in an autocrine or paracrine manner(FIG2C).
Overexpression of IL-35 in patients with hepatocellular carcinoma was closely related to the in ltration of neutrophils and CD8 + in tumor microenvironment.
We further explored the relationship between the level of IL-35 and TME in HCC. Our results showed that the in ltration of neutrophils in HCC tissues with high levels of IL-35 was signi cantly higher than that in the low expression group (5.31 vs. 14.80 ± 1.34, P < 0.001). In addition, we found that the number of in ltrated CD8+ Tcells in tissues with high levels of IL-35 was signi cantly decreased (34.55 ± 2.758 vs. 56.61 ± 3.53, P < 0.001). Importantly, the number of microvessel density (MVD) labeling by CD34 in patients with overexpression of IL-35 was demonstrated to be signi cantly increased (86.63 ± 4.789 vs. 56.54 ± 2.308, P < 0.001). Additionally, the number of neutrophils in ltrating the tumor was shown to be positively correlated with MVD (r = 0.301, P < 0.001), suggesting that neutrophil in ltration might be an important factor in tumor angiogenesis(FIG2A-C).
IL-35 facilitated tumor progression by affecting neutrophil in ltration, angiogenesis, and CD8+ T-cell in ltration in a mouse model.
Based on the expression of IL-35 in HCC cell lines, we selected the HCCLM3 and MHCC97H cell lines, which exhibited a high expression level to construct IL-35 knocked-down cell lines. Whereas the HUH7 and SMMC-7721 cell lines, which showed a low expression level of IL-35 were selected to construct IL-35 overexpressing cell lines. We also constructed IL-35 overexpressing and knocked-down Hepa1-6 cells. These HCC cell lines were also used to detect the expression of the 2 subunits of the IL-35 receptor, GP130 and IL-12 Rβ2. These were observed to be structural foundations for liver cancer cells in affecting the function of tumor cells through autocrine IL-35(FIGS1A).
First, we used the CCK8 transwell and wound healing to test the effect of IL-35 on HCC cells. Our results showed that IL-35 had no signi cant direct effects on the proliferation and migration of HCC cells in vitro. Surprisingly, we found that the formation rate of the subcutaneous tumor in the IL-35 overexpressing cell line in nude mice was signi cantly faster than that in the control group (volume mm 3 : 1186.81 ± 83. 53 vs. 612.82 ± 73. 49, P < 0. 001, tumor mass g: 1.17 ± 0.11 vs. 0.63 ± 0.09, P < 0.001), with the tumor growth rate being signi cantly slowed down in the knocked-down group (volume mm 3 : 495.48 ± 53.42 vs. 882.61 ± 73.25, P < 0.001, tumor mass g: 0.62 ± 0.07 vs. 1.11 ± 0.18, P < 0.001)(FIG3). We also observed the same phenomenon in the immunocompetent C57BL/6 mouse model when exploring the effect of different expression levels of IL-35 on the growth of subcutaneous tumors(FIG4). Therefore, we speculated that IL-35 might promote the progression of liver cancer by affecting TME.
Meanwhile, we found that the in ltration of neutrophils (169 ± 47 vs. 62 ± 15, P < 0.001), as well as the number of MVD (149 ± 39 vs. 33 ± 21, P < 0.001) were both signi cantly increased in immunode cient or immunocompetent mouse models injected with tumor cells overexpressing IL-35. In the IL-35 knocked-down group, the number of neutrophils in ltrating the tumor (31 ± 12 vs. 189 ± 43, P < 0.001), and the amount of MVD (43 ±21 vs. 125 ±34, P < 0.001) were shown to be signi cantly decreased(FIG3B). Thus, we assumed that IL-35 could promote intratumoral neovascularization by recruiting the in ltration of neutrophils in tumors. To test this hypothesis, we used a LY6G neutrophil antibody to deplete neutrophils in an IL-35-overexpressing subcutaneous tumor, and found that depleting neutrophils could signi cantly reverse the promoting effect of the overexpression of IL-35(FIG4D).
In addition, we also found that the number of in ltrating CD8+ T-cells was signi cantly decreased in the overexpression group, whereas it was signi cantly increased in the knocked-down group.

IL-35 promoted neutrophil in ltration by increasing the expression of CCL3 in vitro
We used a transwell assay to verify the effect of the HCC-related expression of IL-35 on neutrophil chemotaxis. The chemotactic effect of the IL-35-KD conditioned medium (CM) on neutrophils was shown to be decreased by 64.5 % and 56.3 % (P < 0.05), whereas the overexpression in CM increased by 3.97 and 4.67 times, respectively (P < 0.05). However, we noted that the recombinant IL-35 (rIL-35) had no signi cant effect on neutrophil chemotaxis (P > 0 05). These results showed that HCC-related IL-35 did not directly affect neutrophil in ltration(FIG5A).
We then aimed to further explore the pathway through which IL-35 affects the chemotaxis of neutrophils. After comparing the results of IL-35 positive related genes and sequencing in the TCGA database, we found that the expression of neutrophil-related chemokine genes was signi cantly increased after overexpression of IL-35(FIGS2A). We further found that following overexpression of IL-35, the intracellular levels of the CCL3 protein were signi cantly increased, whereas after knocking down IL-35, the intracellular levels of the CCL3 protein were shown to be signi cantly decreased. We further con rmed that CCL3 was signi cantly increased in IL-35-overexpressing patients (P < 0.012, r=0.431) (FIG5B). In order to verify whether IL-35 chemotactically affected neutrophils through the expression of CCL3, we carried out a CCL3 antibody block test. We accordingly discovered that CCL3 could signi cantly enhance the chemotactic effect on neutrophils, as the CCL3 antibody intervention test was demonstrated to signi cantly reduce the chemotactic effect of CM on neutrophils(FIG5A). In summary, we found that IL-35 could promote the chemotactic effect of neutrophils by promoting the expression of CCL3 in HCC.

IL-35 stimulated neutrophil secretion of FGF2 to promote angiogenesis
To illustrate the roles and underlying mechanism of IL-35 in tumor angiogenesis, we carried out a tube formation experiment in vitro. First, we stimulated HUVEC endothelial cells with rIL-35 or CM from IL-35overexpressing or knocked-down cells, and found that the tube formation rate did not signi cantly change. This suggested that IL-35 did not directly stimulate the formation of vascular endothelium. Considering that accumulation of neutrophils in HCC tissues has been reported to increase the production of angiogenesis factors and facilitate microvessel formation, we speculated that IL-35 might indirectly affect tumor angiogenesis by stimulating neutrophils. To further assess this hypothesis, we stimulated HUVECs with CM from the coculturing of neutrophils and HCC cells. We found that CM from neutrophils cocultured with IL-35overexpressing HCC cells could enhance tube formation (tube density: 212 ± 31 vs. 141 ± 19, P < 0.0024; tube branch: 365 ± 27 vs. 238 ± 24, P < 0.001). Conversely, the CM from IL-35 knocked-down HCC cells cocultured with neutrophils could signi cantly inhibit the tubule formation of endothelial cells (tube density: 119 ± 19 vs. 169 ± 23, P < 0.0056; tubule branch: 229 ± 24 vs. 315 ± 32, P < 0.0013).(FIG5C) These results demonstrated that IL-35 stimulated neutrophils to produce angiogenesis factors.
To further explore this, we isolated neutrophils from patients with HCC, stimulated them with human IL-35, and then sequenced them. Our sequencing results were subjected to GO and KEGG analysis, where it was revealed that the expression of genes related to angiogenesis and adhesion factors in neutrophils was signi cantly increased. The KEGG pathway enrichment map showed that after neutrophils were stimulated by IL-35, the pathways of epidermal growth factor receptor (EGFR) and vascular endothelial growth factor (VEGF) were signi cantly activated. The FGF2 protein was demonstrated to be the most signi cantly elevated angiogenic factor, with the expression of the FGFR3 and FGFR4 receptors of FGF2 being also increased by 574 and 65 times, respectively (FIGS2B,FIG6A). This nding suggested that there might be a mechanism by which IL-35 stimulates the positive feedback secretion of FGF2 by neutrophils. This was further con rmed by WB analysis.
To explore whether FGF2 plays a decisive role in mediating IL-35 to promote angiogenesis, we performed blocking experiments. When anti-IL-35 and FGF2 neutralizing antibodies were used, the tube formation rate was shown to be signi cantly abrogated. Furthermore, after IL-35 knocked-down HCC cells were cocultured with neutrophils in the presence of rIL-35, the tube formation rate was demonstrated to be signi cantly elevated; however, when anti-FGF2 neutralizing antibody was added to the above CM, tubule formation was blocked.(FIG6B) IL-35 facilitated the adhesion of tumor to endothelial cells, with neutrophils further enhancing this effect in vitro.
We also found that IL-35 could enhance the adhesion of HCC cells to HUVECs in vitro. Following the overexpression of IL-35, the adhesion rate of SMMC-7721 or HUH7 cells to endothelial cells was demonstrated to be signi cantly increased. In contrast, knocking down IL-35 strongly inhibited the number of HCCLM3 or MHCC97H cells adhered to endothelial cells. At the same time, we also found that when neutrophils were added to overexpressing IL-35 trials, the number of SMMC-7721 or HUH7 cells adhered to endothelial cells was increased by 46 % and 54 %, respectively (P < 0.001). However, there was no signi cant change observed in the adhesion of tumor cells cocultured with neutrophils in the knocked-down group (P > 0.05). These results showed that IL-35 could increase the adhesion of hepatocellular carcinoma cells to endothelial cells, and neutrophils could further enhance this effect. In addition, we further veri ed this nding using the Hepa1-6 mouse HCC cell line (FIG7A).
The lung metastasis model of the tail vein in nude mice further showed the promoting ability of IL-35 and neutrophils.
In the in vivo experiment, we rst labeled tumor cells with a uorescent dye. Following the intravenous injection of HCC cells into the tail vein of nude mice, the remaining tumor cells in the lung tissue were observed using a uorescence tracer. We found that there was no signi cant difference in the retention of tumor cells in each group at 30 min; however we observed a signi cant difference in the number of tumor cells stranded in lung tissue 24 h later. The number of retained tumor cells in the coinjection group (279 ± 53)was signi cantly higher than that in the overexpression (103 ± 31) and control (62 ±21) groups (P < 0.001), whereas the number of tumor cells retained in the overexpression group was also signi cantly higher than that in the control group. We observed that in the coinjection group, a large number of tumor cells adhered directly to neutrophils. We collected gross specimens of lung tissue, and found that the number of metastatic nodules after injecting HUH7-OE was signi cantly higher than that in the control group (FIG7A). After simultaneous injection of mixed cells of neutrophils and OE strains, the number of pulmonary metastatic nodules was demonstrated to be further increased. This number was signi cantly higher than that of the OE stable lung metastasis model, consistent with the results of HE staining of lung metastasis nodules. Meanwhile, we found that the rate of lung metastasis was signi cantly decreased when IL-35 was knocked down (FIG7B).

Anti-IL-35 neutralizing antibody enhanced the e cacy of PD-1 antibody
The combined use of drugs is an important way to explore better treatments of liver and other cancers. Therefore, we aimed to explore whether an IL-35 antibody could enhance the effect of the administration of the PD-1 antibody in the treatment of HCC.
We found that in the subcutaneous tumor model of immunocompetent mice recipients bearing Hepa1-6 cells, tumor growth was slightly suppressed after treatment with either an anti-IL-35 or a PD-1 neutralizing antibody. However, we observed more dramatic and durable responses, compared with the responses in the control treatment, when the anti-PD-1 antibody was combined with the anti-IL-35 neutralizing antibody(FIG8A). We also constructed an in vivo model using the IL-35-OE Hepa1-6 cell line, followed by the administration of both the anti-PD-1 and anti-IL-35 antibodies. Our results showed that the anti-PD-1 antibody combined with treatment with the anti-IL-35 antibody could reverse the increased tumor growth induced by IL-35OE Hepa1-6 cells in C57BL/6 mice (FIGS3).
As shown in Figure8B-C, the in ltration of CD8+ T-cells was increased after treatment with the IL-35 and PD-1 antibodies (P < 0.001). In contrast, we found that neutrophil in ltration was decreased after treatment with the IL-35 antibody (P < 0.001), whereas no effect was observed in neutrophil in ltration after administration of the PD-1 antibody. Nevertheless, we did not observe any signi cant difference in other cells, including macrophages and Treg cells. Immunohistochemical analysis showed that treatment with the PD-1 and IL-35 antibodies could increase the in ltration of CD8+ T-cells in the tumor, with the combined treatment group being shown to further increase the in ltration of CD8+ T-cells compared with the single treatment group (P < 0.001). The neutrophil in ltration in the IL-35 antibody group and the combined treatment group was observed to be signi cantly lower than that in the control group (P < 0.001); however, there was no signi cant difference shown in neutrophil in ltration between the 2 groups. The PD-1 antibody treatment group was also shown to have no effect on neutrophil in ltration.

Discussion
Cytokines, including interleukin, chemokine, and growth factors, are small molecules synthesized and secreted by immune cells (monocytes, macrophages, NK cells, and other cells) and some nonimmune cells (endothelial cells, ber cells, etc.). Cytokines are known to directly act and participate in the proliferation and metastasis of tumor cells, and to indirectly participate in the occurrence and development of tumors through the regulation of TME. Functionally, cytokines can be divided into 2 categories: antitumor and tumor-promoting. The former is represented by IL-2, tumor necrosis factor (TNF), and interferon (IFN), whereas the latter is represented by transforming growth factor (TGF), IL-10, VEGF, and FGF. As a special member of the IL-12 family, IL-35 has been reported to play an important role in tumor immunosuppression. We found that IL-35 promoted intratumoral neovascularization by recruiting neutrophils and reshaping the immune microenvironment to promote the progression of liver cancer. In addition, the combination of anti-IL-35 and anti-PD-1 antibodies was demonstrated to play a synergistic role in anticancer activity.
We found that overexpression of IL-35 in HCC was an independent risk factor for overall survival and recurrence. The GP130 and IL-12 Rβ2 receptors were also found to be expressed in HCC tissues, indicating a structural basis for the paracrine and autocrine function of IL-35, consistent with previous reports [22,30] . We further combined the expression levels of IL-35 and its receptors in HCC tissues, and found that patients with high expression of IL-35 and its receptors in tumor tissues had the worst prognosis, indicating that IL-35 expressed in HCC might activate the receptors of tumor cells through an autocrine pathway to promote the progression of liver cancer. However, no signi cant correlations were noted between the overexpression of IL-35 and advanced BCLC stage, higher levels of serum AFP, larger tumors, and higher risk of PVTT and MVI, thus suggesting a signi cant role of IL-35 in the progression and metastasis of HCC.
Interestingly, although the results of CCK8, transwell, and wound healing assays were negative in vitro, we found that in vivo, elevated IL-35 could signi cantly promote tumor growth, whereas the downregulation of the expression of IL-35 was signi cantly inhibited in both immunocompetent and immunode cient mouse models. It has also been previously reported that low expression of miR-28-5p in hepatocellular carcinoma cell lines did not affect the proliferation and invasion of tumor cells in vitro, but could promote the tumor secretion of IL-34 known to lead to the recruitment and transformation of tumor-related macrophages to promote the metastasis of hepatocellular carcinoma cells [31] . Wang et al. reported that melanoma-derived IL-35 could promote the growth of subcutaneous tumors in immunocompetent mice by recruiting myeloid inhibitory cells to in ltrate the TME [23] . A study of pancreatic cancer also found that tumor-derived IL-35 could promote tumor growth by recruiting monocytes to in ltrate the tumor [20] . Therefore, we speculated that IL-35 might also indirectly affect and promote tumor progression by modifying TME.
By studying the TME of HCC tissue samples, we found that the local in ltration of neutrophils and MVD were signi cantly increased, whereas that of CD8+ T-lymphocytes was signi cantly decreased in tumors with high expression of IL-35. In in vivo experiments, we also found that IL-35 was closely related to neutrophil in ltration and intratumoral angiogenesis, suggesting that the increased expression of IL-35 was related to local immune tolerance and tumor angiogenesis in HCC. Although neutrophils are known to play an important role in innate immunity, following their recruitment by tumor cells, N2 neutrophils have been reported to promote tumor growth and metastasis by stimulating tumor angiogenesis. Our study also showed that there was a signi cant positive correlation between neutrophils and angiogenesis in HCC, and there was a signi cant increase in angiogenesis and neutrophil in ltration in HCC tissues with high expression of IL-35. These ndings suggested that there might be a complex interaction between the level of IL-35 and neutrophil in ltration and angiogenesis in hepatocellular carcinoma. The factors that affect the formation of blood vessels in tumors are complex. For instance, angiogenic factors could directly stimulate tumors to form neovascularization. However, tumor angiogenesis could also be indirectly affected by the recruitment of neutrophils, macrophages, Treg cells, stellate cells, and other interstitial cells known to affect TME. It has been reported that IL-35 could also affect the secretion of CXCL1 and CXCL8 by monocytes and promote angiogenesis in pancreatic cancer [20] . Wang et al. reported that IL-35 promoted tumor angiogenesis through macrophages in melanoma [23] . We found that both direct stimulation of rIL-35 and IL-35-overexpressing or knocked-down CM had no signi cant effect on microvessel formation, indicating that the tumor did not directly stimulate the formation of vascular endothelium by secreting IL-35. Therefore, we considered that tumor secretion of IL-35 might promote intratumoral angiogenesis through indirect mechanisms.
Tumor-related neutrophils are known to play an important role in the occurrence and development of liver cancer and have been shown to be closely related to tumor angiogenesis. It has been reported that neutrophils (mainly N2) could participate in the reconstruction of the tumor extracellular matrix by secreting matrix metalloproteinases (MMPs), NE, and cathepsin G, thus promoting tumor metastasis [32,33] . Tumor cells have been shown to recruit a large number of neutrophils to in ltrate tumor tissues by secreting the CXCL8 chemokine. These in ltrating neutrophils are known to interact with adhesion molecules, such as ICAM1 on the surface of tumor cells to enhance the adhesion and metastasis of tumor cells to endothelial cells [25] .
Neutrophils in breast cancer were demonstrated to bind to circulating tumor cells (CTCs) entering the circulatory system through vascular cell adhesion protein 1 (VCAM1) to promote metastasis of breast cancer cells [32] . Neutrophils have also been shown to directly stimulate tumor proliferation and metastasis by releasing the extracellular capture network (NET), containing chromatin, granule proteins, matrix metalloproteins, and NE. In particular, NET is rich in active protein components, and thus can provide a favorable local microenvironment for tumor cells with transvascular metastasis, recruit tumor cells to form cell clusters, and inhibit the apoptosis of tumor cells. At the same time, it could enhance the adhesion between tumor and vascular endothelial cells [34,35] . Zou et al. found that IL-35 could indirectly promote neutrophil polarization to N2 [28] . We found that IL-35 was associated with neutrophil in ltration, indicating that IL-35 might promote intratumoral angiogenesis through neutrophils. We further found that IL-35 could signi cantly affect the expression of CCL3 and promote neutrophil in ltration. It should be mentioned that CCL3 is an important chemokine in neutrophils. We further found that IL-35 could promote vascular catheterization and intratumoral angiogenesis by promoting the secretion of the FGF2 angiogenic factor by neutrophils. Moreover, IL-35 was demonstrated to stimulate the expression of the FGFR3 and FGFR4 neutrophil FGF2 receptors, indicating that there might be a positive feedback loop in TME regarding the IL-35-stimulated secretion of neutrophil FGF2. Brie y, FGF2, secreted by neutrophils, is known to be an important cytokine in the promotion of vascular growth. It has been reported that neutrophils in ltrated at metastasis sites could promote intratumor angiogenesis by secreting FGF2, and could thus also promote tumor metastasis [36] .
It is interesting that IL-35 could signi cantly reduce the in ltration of CD8+ T-cells in tumor tissues in an immunocompetent mouse model, a nding that was in accordance with the phenomenon observed in the TMA of HCC patients, suggesting that IL-35 might have an inhibitory effect on tumor immunity. Liver cancer cells might also secrete IL-35 to reduce the in ltration of CD8+ T-cells in order to create an immunosuppressive microenvironment, thus inducing tumor immune escape. It has also been reported that IL-35 secreted by tumors did not directly affect the differentiation and function of CD8+ T-cells in melanoma. However, tumor cells have also been reported to stimulate the enhanced expression of the GP130 receptor of IL-35 to reduce their sensitivity to CD8+ T-cell killing, so as to escape immune surveillance and promote tumor growth [23] .
Tumor immunotherapy represented by therapy with PD-1/PD-L1 antibodies is a hot spot of tumor immunotherapy in recent years. So far, it has been reported to achieve good results in a variety of tumors, including hepatocellular carcinoma, and has received extensive attention. At present, there are more than 3000 clinical trials evaluating single drugs and combined treatments of immune checkpoints. Several studies have shown that the curative effect of the PD-1 antibody combined with chemotherapy, targeted drugs, and multiple immune checkpoint drugs was better than that of the administration of a single drug. The effect of getting twice the result with half the effort was obtained [37][38][39][40][41][42] . It should be noted that the therapy with PD-1 or PD-L1 antibody is known to be effective mainly through the immunological mechanism playing the antitumor role; however, its e cacy has certain limitations and therefore needs to be combined with other antitumor drugs in order to produce improved results. Recently, a number of studies have shown that immune checkpoint inhibitors combined with antivascular-targeting drugs or antibodies could achieve a breakthrough effect, giving full play to the curative effect of 1:1 > 2. More optimistically, we found that the IL-35 antibody could enhance the e cacy of PD-1. The possible reasons behind this might be the following: (1) the IL-35 antibody might reduce the inhibitory effect of IL-35 on lymphocytes; (2) the IL-35 antibody could enhance the in ltration of lymphocytes and reduce the in ltration of neutrophils. Related studies have shown that the number of neutrophils in ltrating tumors could promote the progression of liver cancer by secreting C-C motif chemokine ligand 2 (CCL2) and CCL17 to recruit macrophages and Treg cells [31] . Blocking neutrophils with the IL-35 antibody could also reduce the recruitment of macrophages and Treg cells and reduce the inhibitory effect on CD8+ T-cells; (3) blocking IL-35 could block neutrophil-mediated intratumoral angiogenesis and further promote the e cacy of the PD-1 antibody. In related studies of hepatocellular carcinoma, it was found that the PD-L1 of neutrophils near the cancer was signi cantly increased [43] ; (4)

Conclusions
Our data show that the expression of IL-35 in patients with HCC is an important tumor promoting factor. And we have found that IL-35 facilitated tumor progression by affecting neutrophil in ltration, angiogenesis, and CD8+ T-cell in ltration in a mouse model. Additionally, on the one hand CCL3 has been found to be a key factor mediating the recruitment of neutrophils by IL-35, on the other hand FGF2 secreting by neutrophil when stimulated by IL-35 was also found to be the core cytokine to promote intratumoral angiogenesis. The   Figure 1 High expression of IL-35 confers a poor prognosis in HCC patients. Serial sections of HCC tissue microarray (TMA) were used to explore the expression levels of the two subunits of the IL-35 ligand: EBI3 and P35. As EBI3 forms IL-27 with P28 and P35 forms IL-12 with P40, we also examined the expression levels of P28 and P40 in consecutive tumor tissue. A. a signi cant higher expression of P35 and EBI3 in HCC tumor were found than that of adjacent normal tissue. And the expression levels were much higher than that of P40 and P28 In consecutive sections of HCC tissue. B.The expression levels of P35 and EBI3 are highly consistent. And these are the representative pictures of P35 and EBI3 from low to high depending on the expression level (-,+,++,+++).
C. Cell uorescent microscopy images demonstrating that the expression of P35 (red), EBI3 (green) were higher in HCC tumor cells HCC-LM3 and MHCC-97H than that of SMMC-7721 and HUH7. D. The distribution bar chart of P35, EBI3, P28, P40 in 360 HCC patients. and, staining extent correlation among them. We have also illustrated the Kaplan-Meier survival analysis of OS and RFS according to different IL-35 levels.    The tuber formation capable of neutrophils were activated in a FGF2 depended manner while it was stimulated by IL-35. A. Neutrophils's proangiogenetic capability were activated by increasing the secretion of FGF2. B.Anti-FGF2 antibody were used to blocked IL-35 mediated neutrophils' proangiogenetic capability. And blocking IL-35 with IL-35 antibody also could decrease the vessel density and branch points in co-culture system of HCC and neutrophiles. IL-35 a liated the adhesion between HCC tumor cell and HUVEC, and promoted the pulmonary metastasis in vivo. A. More HCC were found to adhere the surface of HUVEC. In vivo experiment the remaining tumor cells in lung tissue were observed by uorescence tracer. We found that there was no signi cant difference in the retention of tumor cells in each group at 30 minutes, but there was a signi cant difference in the number of tumor cells stranded in lung tissue 24 hours later. B. HE staining were used to count the number of pulmonary metastatic nodules in lung metastasis model. Signi cantly increased metastatic nodules were found in overexpression co-inject group. Meanwhile, we found that the lung metastasis rate decreased signi cantly when IL-35 was knocking down. plot were utilized to explore the in ltration of immune cells of these tumors tissue. Anti-PD-1 antibody has no effect on the in ltration of Ly6G+ neutrophiles, but signi cant difference were catch in combined treatment group and antiIL-35 group. There were less in ltration of neutrophils among these two groups. C. Interesting we also found that CD8+T cell were also signi cant increase in both PD-1 antidy group and combined treatment group.

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