Midkine inhibition enhances anti-PD-1 immunotherapy in sorafenib-treated hepatocellular carcinoma via preventing immunosuppressive MDSCs infiltration

Lijuan Ding; Nanya Wang; Qiang Wang; Xia Fan; Yuning Xin; Shudong Wang

doi:10.1038/s41420-023-01392-3

Midkine inhibition enhances anti-PD-1 immunotherapy in sorafenib-treated hepatocellular carcinoma via preventing immunosuppressive MDSCs infiltration

Ding, Lijuan; Wang, Nanya; Wang, Qiang; Fan, Xia; Xin, Yuning; Wang, Shudong 2023-03-11 00:00:00 www.nature.com/cddiscovery ARTICLE OPEN Midkine inhibition enhances anti-PD-1 immunotherapy in sorafenib-treated hepatocellular carcinoma via preventing immunosuppressive MDSCs inﬁltration 1 2 1 1 1 3✉ Lijuan Ding , Nanya Wang , Qiang Wang , Xia Fan , Yuning Xin and Shudong Wang © The Author(s) 2023 Sorafenib, a multiple-target tyrosine kinase inhibitor, is the standard of care for patients with advanced hepatocellular carcinoma (HCC), but provides limited beneﬁts. Emerging evidences suggest that prolonged sorafenib treatment induces an immunosuppressive HCC microenvironment, but the underling mechanism is undetermined. In the present study, the potential function of midkine, a heparin-binding growth factor/cytokine, was evaluated in sorafenib-treated HCC tumors. Inﬁltrating immune cells of orthotopic HCC tumors were measured by ﬂow cytometry. Differentially expressed genes in sorafenib-treated HCC tumors were evaluated by transcriptome RNA sequencing. The potential function of midkine were evaluated by western blot, T cell suppression assay, immunohistochemistry (IHC) staining and tumor xenograft model. We found that sorafenib treatment increased intratumoral hypoxia and altered HCC microenvironment towards an immune-resistant state in orthotopic HCC tumors. Sorafenib treatment promoted midkine expression and secretion by HCC cells. Moreover, forced midkine expression stimulated immunosuppressive myeloid-derived suppressor cells (MDSCs) accumulation in HCC microenvironment, while knockdown of + + − midkine exhibited opposite effects. Furthermore, midkine overexpression promoted CD11b CD33 HLA-DR MDSCs expansion from human PBMCs, while midkine depletion suppressed this effect. PD-1 blockade showed no obvious inhibition on tumor growth of sorafenib-treated HCC tumors, but the inhibitory effect was greatly enhanced by midkine knockdown. Besides, midkine overexpression promoted multiple pathways activation and IL-10 production by MDSCs. Our data elucidated a novel role of midkine in the immunosuppressive microenvironment of sorafenib-treated HCC tumors. Mikdine might be a potential target for the combination of anti-PD-1 immunotherapy in HCC patients. Cell Death Discovery (2023) 9:92 ; https://doi.org/10.1038/s41420-023-01392-3 INTRODUCTION not superior to sorafenib when used in the ﬁrst-line setting [7]. Hepatocellular carcinoma (HCC) ranks the sixth most common Thus, there is urging needing to develop novel therapeutic cancer and second leading cause of cancer-related death world- strategies. wide [1]. In the last two decades, the incidence and mortality rates HCC is typically caused by chronic inﬂammation of a liver of HCC are increasing in most parts of the world, especially eastern diseases [8]. Therefore, there is a strong rationale for using of Asia and Africa [2]. Patients with HCC have a really poor prognosis, immunotherapy. Immune checkpoints include co-inhibitory recep- largely due to late diagnosis and high recurrence rate. The tors such as programmed cell death-1 (PD-1) and its ligand majority of HCCs are diagnosed at intermediate or advanced programmed death ligand 1 (PD-L1). PD-1 is expressed by an stages, with a 5-year survival rate as low as 16% [3]. Patients with extensive number of immune cells such as activated T cells and HCC are refractory to almost all conventional chemotherapeutic natural killer (NK) cells, while PD-L1 is expressed on tumor cells, drugs. Systemic therapy for advanced HCC consists only of stromal cells and myeloid cells [9]. In general, antiangiogenic antiangiogenic tyrosine kinase inhibitors until recently. Sorafenib, therapy with sorafenib, lenvatinib or VEGF antibodies remains the a multiple-target tyrosine kinase inhibitor with antiangiogenic and fundamental treatment for advanced HCC, whereas immunother- antiproliferation effects, is approved by the Food and Drug apy with checkpoint inhibitors becomes increasing important [10]. Administration (FDA) as the ﬁrst-line chemotherapeutic drug for Recently, immunotherapy with checkpoint inhibitors has been advanced HCC. In two large phase III trials, sorafenib is proved to tested in HCC patients and shows strong anti-tumor effect in a be effective in advanced HCC patients [4, 5]. However, less than subset of patients [11, 12]. The combination of PD-1/PD-L1 and 30% of HCC patients can beneﬁt from sorafenib treatment, and vascular endothelial growth factor (VEGF) antibodies results in a the median survival is extended only about 2.5 months [5]. better survival than sorafenib, which makes it as a new ﬁrst line Besides, these patients usually become resistant to sorafenib therapy [13]. Despite these improvements, only one in four within 6 months [6]. Another chemotherapeutic drug lenvatinib is patients respond to these immunotherapies, and the majority of 1 2 Department of Radiation Oncology, the First Hospital of Jilin University, Changchun 130021, China. Cancer Center, the First Hospital of Jilin University, Changchun 130021, China. Department of Cardiology, the First Hospital of Jilin University, Changchun 130021, China. email: wangshudongjlu@126.com Received: 24 August 2022 Revised: 28 February 2023 Accepted: 1 March 2023 Ofﬁcial journal of CDDpress 1234567890();,: L. Ding et al. Fig. 1 Sorafenib treatment increases intratumoral hypoxia and alters HCC microenvironment towards an immune-resistant state in mouse models. A, B orthotopic Hepa 1-6 and Hepa 1c1c7 tumors were treated with sorafenib or vehicle control for three weeks, then collected protein lysates for western blot (A) or tissue samples for Pimonidazole staining (B). C, D inﬁltrating immune cells in orthotopic Hepa + + + + 1-6 and Hepa 1c1c7 tumors treated with sorafenib or vehicle control were evaluated by ﬂow cytometry. CD3 CD4 CD25 FoxP3 Treg, + − − + + + int + + + high - + - + CD11b Gr-1 Ly6C F4/80 TAM, CD11b Gr-1 Ly6C Ly6G PMN-MDSC, CD11b Gr-1 Ly6C Ly6G M-MDSC and CD3 CD4 CD8 Cyto T cells were evaluated. E relative expression of PD-L1, TGFB1, IL10, and IL13 in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors was evaluated by qRT-PCR. F protein expression of PD-L1 in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors was evaluated by western blot. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. HCC patients do not respond due to unknown reasons. One Moreover, a number of immune cells are inﬂuenced by midkine, aspect is that prolonged antiangiogenic therapy increases including macrophage [18], polymorphonuclear (PMN) [19], B cells intratumoral hypoxia, which facilitates tumor recurrence and [20] and T cells [21]. In addition, midkine plays a vital role in fosters an immunosuppressive microenvironment [14, 15]. Since maintaining immunosuppressive tumor microenvironment of the efﬁciency of immune checkpoint inhibitors is greatly melanoma and gallbladder cancer [22, 23]. Midkine is upregulated inﬂuenced by immunosuppressive tumor microenvironment, it is in HCC patients and associated with poor prognosis [24]. However, necessary to elucidate the underlying mechanism. the potential role of midkine in HCC microenvironment is Midkine (MDK) is a heparin-binding growth factor/cytokine that undetermined. Myeloid-derived suppressor cells (MDSCs) are exhibits multiple functions and implicates in various physiological immature marrow-derived cell populations with potent immuno- process [16]. Midkine is rarely expressed by normal tissues, but suppressive activity. They are mainly consisted of granulocytic/ signiﬁcantly upregulated in inﬂammatory diseases and human polymorphonuclear MDSCs (PMN-MDSCs) and monocytic MDSCs malignant tumors [17]. Upregulation of midkine can promote (M-MDSCs), which are similar to neutrophils and monocytes, growth, survival, metastasis, and angiogenesis of cancer cells [16]. respectively. In the present study, orthotopic HCC models were Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 2 Sorafenib treatment promotes midkine expression and secretion by HCC cells. A differentially expressed genes in orthotopic Hepa 1-6 tumors treated with sorafenib (Sor) or vehicle control (Veh) were depicted in volcano map. B, C midkine expression in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors treated with sorafenib or vehicle control was evaluated by qRT-PCR (B) and western blot (C). D secreted mikdine in serum samples of orthotopic Hepa 1-6 and Hepa 1c1c7 tumors treated with sorafenib or vehicle control was measured by ELISA assay. E, F midkine expression in orthotopic HUH-7 and SNU-449 tumors treated with sorafenib or vehicle control was evaluated by qRT-PCR (E) and western blot (F). G secreted mikdine in serum samples of orthotopic HUH-7 and SNU-449 tumors treated with sorafenib or vehicle control was measured by ELISA assay. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. constructed to evaluate the potential role of midkine in sorafenib- expression and reduced vessel density in tumor samples induced immunosuppressive microenvironment. We found that (Supplementary Fig. 1A–C). Moreover, HIF1-α dependent genes midkine overexpression stimulated MDSCs inﬁltrating into HCC such as VEGF, GLUT-1, CA-9, CXCR4, and MDR1 were upregulated tumors, while knockdown of midkine exhibited opposite effects. In in sorafenib-treated patients (Supplementary Fig. 1D). These addition, midkine inhibition enhanced the inhibitory effects of results indicated that sorafenib increased intratumoral hypoxia anti-PD-1 immunotherapy in sorafenib-treated HCC tumors. Our in orthotopic HCC models and patient samples. To evaluate the results elucidated a novel role of midkine in HCC microenviron- immune microenvironment of orthotopic HCC tumors, inﬁltrating ment, and midkine might be a potential target for HCC treatment. immune cells were evaluated by ﬂow cytometry. The gating strategy and representative plots of inﬁltrating immune cell populations were depicted in Supplementary Fig. 2. The + + + + RESULTS percentages of regulator T cells (Treg, CD3 CD4 CD25 FoxP3 ), + − - + Sorafenib treatment increases intratumoral hypoxia and alters tumor-associated macrophages (TAM, CD11b Gr-1 Ly6C F4/80 ), HCC microenvironment towards an immune-resistant state in polymorphonuclear myeloid-derived suppressor cells (PMN-MDSC, + + int + mouse models CD11b Gr-1 Ly6C Ly6G ) and monocytic myeloid-derived sup- + + high - To determine whether sorafenib-induced intratumoral hypoxia pressor cells (M-MDSC, CD11b Gr-1 Ly6C Ly6G ) were may foster an immunosuppressive microenvironment, orthotopic obviously increased, while cytotoxic T cells (Cyto T, + − + HCC models were constructed using murine HCC cell lines Hepa CD3 CD4 CD8 ) were apparently decreased in sorafenib- 1-6 and Hepa 1c1c7 in C57BL/6 mice. In our study, sorafenib treated HCC tumors (Fig. 1C, D). PD-L1, TGFB1, IL10, and IL13 are treatment signiﬁcantly increased the protein expression of HIF1-α primary effectors for immunosuppressive tumor microenviron- in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors (Fig. 1A). ment [25, 26]. These genes were signiﬁcantly upregulated in Pimonidazole is a hypoxia-speciﬁc marker. Sorafenib treatment sorafenib-treated HCC tumors (Fig. 1E). The protein expression of evidently increased the number of pimonidazole positive cells in PD-L1 was also increased by sorafenib (Fig. 1F). Our results orthotopic Hepa 1-6 and Hepa 1c1c7 tumors (Fig. 1B). In HCC indicated that sorafenib treatment altered HCC microenvironment patients, sorafenib treatment signiﬁcantly increased HIF1-α toward an immune-resistant state in mouse models. Cell Death Discovery (2023) 9:92 L. Ding et al. Sorafenib treatment promotes midkine expression and Compared with monocytes or neutrophils, mouse MDSCs are secretion by HCC cells characterized by high level of arginase 1 (Arg1) [28]. Of note, Arg1 is To search for potential genes involved in sorafenib-induced vital for the immunosuppression function of MDSCs. In our study, immunosuppressive microenvironment, orthotopic Hepa 1-6 midkine overexpression dramatically increased the number of Arg1 + + tumors treated with sorafenib or vehicle control were subjected cells in CD11b Gr-1 subset of orthotopic HCC tumors with/without to transcriptome RNA sequencing. In our study, 71 genes were sorafenib treatment (Fig. 4A, B). Moreover, the percentages of Arg1 signiﬁcantly upregulated while 131 genes were downregulated in cells in PMN-MDSCs and M-MDSCs subsets were also increased by sorafenib-treated tumors compared with vehicle-treated tumors ectopic midkine expression (Fig. 4C). Arg1, Pdl1, Tgfb and Nos2 are (Fig. 2A, Supplementary Table 1). Among them, midkine ranked the primary effectors for the immunosuppression function of MDSCs. top ten upregulated genes (Fig. 2A, Supplementary Table 1). Forced midkine expression increased the expression of Arg1, Tgfb Midkine plays an important role in maintaining immunosuppres- and Nos2, and decreased the expression of Pdl1 in sorted PMN- sive microenvironment of melanoma and gallbladder cancer MDSCs and M-MDSCs (Fig. 4D). The immunosuppression activity of [22, 23]. Thus, we speculated that the upregulation of midkine MDSCs in midkine-overexpressing HCC tumors was further eval- + + might contribute to the immune-resistant state caused by uated by T-cell suppression assay. CD3 CD8 cytotoxic T cells were sorafenib treatment in HCC tumors. In our study, midkine stimulated with CD3e and CD28 antibodies and co-cultured with expression in pan-cancer was evaluated by TIMER database. We MDSCs sorted from midkine-overexpressing HCC tumors or EV- found that midkine was signiﬁcantly upregulated in a variety of transduced tumors for three days. Cytotoxic T cells co-cultured with cancers, including HCC (Supplementary Fig. 3A). Data from TCGA MDSCs sorted from midkine-overexpressing HCC tumors showed and GEO database (GSE39791 and GSE112790) also suggested that reduced Ki67 and Granzyme B expression compared with T cells co- midkine was overexpressed in HCC tumor samples (Supplementary cultured with MDSCs derived from EV-transduced tumors, indicating Fig. 3B and C). Besides, high midkine expression was positively that the proliferation and activation of T cells were repressed (Fig. correlated with advanced tumor stages and poor overall-survival of 4E, F). Moreover, MDSCs derived from sorafenib-treated tumors HCC patients (Supplementary Fig. 3D and E). In our study, midkine showed enhanced inhibition on T cell proliferation and activation, expression was increased by sorafenib in orthotopic HCC tumors especially those from midkine-overexpressing tumors (Fig. 4E, F). (Figs. 2B and 2C). Midkine is a secreted protein. Sorafenib These results indicated that MDSCs derived from midkine- treatment apparently increased the level of secreted midkine in overexpressing and/or sorafenib-treated HCC tumors showed more serum samples of mice bearing HCC tumors (Fig. 2D). This was also potent immunosuppressive activity. The potential function of validated in human HCC cell lines HUH-7 and SNU-449. Sorafenib midkine was also evaluated in MDSCs expanded from human treatment evidently augmented midkine level in tumor tissues and PBMCs. Midkine was ectopic overexpressed in HUH-7 and SNU-449 serum samples of Balb/c nude mice bearing orthotopic HUH-7 or cells (Supplementary Fig. 6A). The secreted midkine in conditional SNU-449 tumors (Fig. 2E, F). The promoter region of midkine has a medium of HUH-7 and SNU-449 was signiﬁcantly increased by hypoxia responsive element, thus hypoxia may induce midkine midkine overexpression (Supplementary Fig. 6B). Human MDSCs are + + − expression via binding with HIF-1α [27]. This was tested in HCC commonly marked as CD11b , CD33 ,and HLA-DR . In our study, + + - cells. Secreted midkine was signiﬁcantly elevated in culture CD11b CD33 HLA-DR MDSCs were expanded from human PBMCs medium of both murine and human HCC cells under hypoxia via culturing with conditional medium from HUH-7 or SNU-449 cells condition (1% O ) compared with normoxia condition (20% O ) for 5 days. In our study, conditional medium from midkine- 2 2 (Fig. 2G). The above results indicated that sorafenib treatment overexpressing HUH-7 and SNU-449 cells evidently expanded more + + − promoted midkine expression and secretion by HCC cells. CD11b CD33 HLA-DR MDSCs compared with conditional medium from EV-transduced cells (Supplementary Fig. 6C). The immunosup- Forced midkine expression stimulates immunosuppressive pression activity of those MDSCs were evaluated by T-cell MDSCs accumulation in HCC tumor microenvironment suppression assay. T cells co-cultured with MDSCs induced by To elucidate the potential effects of midkine in HCC microenviron- conditional medium from midkine-overexpressing HUH-7 and SNU- ment, the correlation between midkine and tumor immune 449 cells showed less Ki67 staining and IFN-γ production, indicating inﬁltration was evaluated. The 22 tumor-inﬁltrating immune cells these T cells were less proliferative and cytotoxic (Supplementary in HCC tissues were estimated by CIBERSORT algorithm. HCC Fig. 6D–F). Altogether, our data indicated that forced midkine patients with high midkine expression showed enriched scores for expression stimulated immunosuppressive MDSCs accumulation in Treg (Supplementary Fig. 4). Besides, high midkine expression was HCC tumor microenvironment. positively associated with immune checkpoint molecules such as CTLA4, HAVCR2, LAG3, PDCD1, and TIGIT (Supplementary Fig. 5A). Midkine inhibition prevents immunosuppressive MDSCs TISIDB was also used to investigate the correlation of midkine with inﬁltrating into HCC tumors inﬁltrating immune cells. High midkine expression was positively The inﬂuence of midkine on MDSCs inﬁltration was further correlated with MDSCs abundance, but not monocytes or neutro- evaluated by loss-of-function assays. Short hairpin RNAs targeting phils (Supplementary Fig. 5B). The data from TIMER demonstrated mouse midkine (Sh-MDK-1 and Sh-MDK-2) were designed and that immune makers of MDSCs (CD33, ITGAM and FUT4) were introduced into Hepa 1-6 cells. Midkine expression was depleted signiﬁcantly associated with midkine expression (Supplementary Fig. by these two shRNAs, despite exposing to sorafenib treatment 5C). These results suggested that midkine might involve in tumor (Fig. 5A). The percentage of inﬁltrating MDSCs was measured. inﬁltrating of MDSCs. To validate this, midkine was overexpressed by Midkine knockdown apparently decreased the percentage of transducing with midikine expression lentivirial particles (Fig. 3A). inﬁltrating PMN-MDSCs and M-MDSCs in orthotopic Hepa 1-6 + + CD11b and Gr-1 are myeloid differentiation markers for mouse tumors with/without sorafenib treatment (Fig. 5B, C). Depletion of MDSCs. Forced midkine expression apparently increased the midkine also reduced the percentage of Arg1 cells in inﬁltrating + + + + percentages of inﬁltrating CD11b Gr-1 cells in orthotopic Hepa CD11b Gr-1 subset (Fig. 5D, E) and MDSCs subset (Fig. 5F). The 1-6 and Hepa 1c1c7 tumors, and this effect was enforced by inﬂuence of midkine inhibition on immunosuppression activity of sorafenib treatment (Fig. 3B, C). There are two main subsets of inﬁltrating MDSCs was evaluated. Cytotoxic T cells co-cultured + + int + mouse MDSCs: PMN-MDSCs (CD11b Gr-1 Ly6C Ly6G )and with MDSCs from midkine-depleted Hepa 1-6 tumors showed + + high − M-MDSCs (CD11b Gr-1 Ly6C Ly6G ). Midkine overexpression increased Ki67 and Granzyme B staining compared with cells co- signiﬁcantly increased the percentages of both PMN-MDSCs and cultured with MDSCs from non-targeting control (sh-NC) trans- M-MDSCs in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors with/ duced tumors (Fig. 5G, E). This was also evaluated in MDSCs without sorafenib treatment (Fig. 3D, E). expanded from human PBMCs. Short hairpin RNAs targeting Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 3 Hepatocellular midkine overexpression stimulates MDSCs accumulation in tumor microenvironment. Midkine-overexpression or EV-transduced Hepa 1-6 and Hepa 1c1c7 cells were orthotopically implanted into the liver of C57BL/6 mice, then treated with sorafenib (Sor) + + or vehicle control (Veh) for three weeks. A protein expression of midkine was validated by western blot. B, C inﬁltrating CD11b Gr-1 cells int + high - + + were evaluated by ﬂow cytometry. D, E Ly6C Ly6G PMN-MDSCs and Ly6C Ly6G M-MDSCs in inﬁltrating CD11b Gr-1 subset were evaluated by ﬂow cytometry. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. human midkine (Sh-MDK-3 and Sh-MDK-4) were designed and control group (Fig. 6A–C). Compared with the IgG isotype control, introduced into HUH-6 and HepG2 cells which showed high tumor growth, volume and weight of subcutaneous Hepa 1-6 endogenous midkine expression. Midkine was successfully tumors were signiﬁcant reduced by anti-PD-1 antibody, however knocked down by these two shRNAs (Supplementary Fig. 7A). this effect was largely abolished by midkine overexpression (Fig. Knockdown of midkine signiﬁcantly reduced the level of secreted 6A–C). The intratumoral inﬁltrating immune cells were further midkine in conditional medium from HUH-6 and HepG2 cells evaluated. Forced midkine expression signiﬁcantly increased the (Supplementary Fig. 7B). Furthermore, reduced midkine secretion percentages of inﬁltrating MDSCs and Tregs, and anti-PD-1 + + − suppressed CD11b CD33 HLA-DR MDSCs expansion from treatment augmented this effect to some extent (Fig. 6D). In human PBMCs (Supplementary Fig. 7C). In T cell suppression addition, anti-PD-1 treatment dramatically increased the percen- assay, T cells co-cultured with MDSCs induced by conditional tage of inﬁltrating CD8 cytotoxic T cells, however this effect was medium from midkine depleted HUH-6 and HepG2 cells exhibited almost abrogated by midkine overexpression (Fig. 6D). Next, the more Ki67 staining and IFN-γ production, indicating these T cells inﬂuence of midkine inhibition on immunotherapy with anti-PD-1 were more proliferative and cytotoxic (Supplementary Fig. 7D–F). antibody in sorafenib-treated HCC tumors was evaluated. Knock- Above all, our results suggested that midkine inhibition prevented down of midkine showed no evident inﬂuence on tumor growth immunosuppressive MDSCs inﬁltrating into HCC tumors. of subcutaneous Hepa 1-6 tumors (Fig. 6E–G). However, midkine inhibition apparently increased the cytotoxicity of sorafenib, with Midkine inhibition enhances anti-PD-1 immunotherapy in reduced tumor growth, volume and weight (Fig. 6E–G). Besides, sorafenib-treated HCC tumors anti-PD-1 antibody showed no obvious suppression on tumor To assess whether mikine could impinge on immunotherapy with growth of sorafenib-treated tumors, but this was greatly checkpoint inhibitors, we tested anti-PD-1 antibody. Midkine strengthened by midkine knockdown (Fig. 6E–G). Furthermore, overexpression showed no signiﬁcant inﬂuence on growth of depletion of midkine signiﬁcantly reduced the percentage of subcutaneous Hepa 1-6 tumors, as observed in the IgG isotype inﬁltrating MDSCs and Tregs, and increased the percentage of Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 4 Hepatocellular midkine overexpression induces T cell-suppressive MDSC accumulation in tumor microenvironment. Midkine- overexpression or EV-transduced Hepa 1-6 and Hepa 1c1c7 cells were orthotopically implanted into the liver of C57BL/6 mice, then treated + + + with sorafenib (Sor) or vehicle control (Veh) for three weeks. A-C, Arg1 cells in inﬁltrating CD11b Gr-1 subset (A–B) and MDSCs subset (C)of orthotopic Hepa 1-6 and Hepa 1c1c7 tumors were evaluated by ﬂow cytometry. D Relative expression of Arg1, Pdl1, Tgfb and Nos2 in inﬁltrating PMN-MDSCs and M-MDSCs subsets of orthotopic Hepa 1-6 and Hepa 1c1c7 tumors was evaluated by qRT-PCR. E, F Inﬁltrating + + PMN-MDSCs and M-MDSCs of orthotopic Hepa 1-6 and Hepa 1c1c7 tumors were used for T-cell suppression assay. Ki67 (E) and Granzyme B (F) cytotoxic T cells were evaluated by ﬂow cytometry. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. inﬁltrating CD8 cytotoxic T cells in sorafenib-treated Hepa 1-6 from midkine-overexpressing Hepa 1-6 tumors showed increased tumors (Fig. 6H). Taken together, our results indicated that phosphorylation of p65, Akt, ERK, and STAT3 compared with midkine inhibition enhanced anti-PD-1 immunotherapy in MDSCs from EV-transduced Hepa 1-6 tumors, indicating the sorafenib-treated HCC tumors. activation of NF-кB, Akt, ERK, and STAT3 signaling by midkine- driven secretome (Fig. 7A). This was also validated in + + - Midkine overexpression drives multiple pathways activation CD11b CD33 HLA-DR MDSCs induced from human PBMCs. and IL-10 production by MDSCs Conditional medium from midkine-overexpression HUH-7 and Previous studies demonstrate that midkine promotes the activa- SNU-449 cells evidently promoted the phosphorylation of p65, + + − tion of various signaling pathways, including NF-кB, PI3K/Akt, ERK, Akt, ERK and STAT3 in CD11b CD33 HLA-DR MDSCs compared and Notch2/Jak2/STAT3 [16, 22, 29]. In our study, MDSCs sorted with conditional medium form EV-transduced cells Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 5 Midkine inhibition prevents immunosuppressive MDSCs inﬁltrating into HCC tumors. Hepa 1-6 cells transduced with Sh-MDK-1, Sh- MDK-2 or Sh-NC lentivirus were orthotopically implanted into the liver of C57BL/6 mice, then treated with sorafenib (Sor) or vehicle control (Veh) for three weeks. A protein expression of midkine in midkine-depleted orthotopic Hepa 1-6 tumors was validated by western blot. int + high - + + B, C Ly6C Ly6G PMN-MDSCs and Ly6C Ly6G M-MDSCs in intratumoral CD11b Gr-1 subset of midkine-depleted orthotopic Hepa 1-6 + + + tumors were evaluated by ﬂow cytometry. D–F Arg1 cells in inﬁltrating CD11b Gr-1 subset (D–E) and MDSCs subset (F) of midkine- depleted orthotopic Hepa 1-6 tumors were evaluated by ﬂow cytometry. G, H inﬁltrating PMN-MDSCs and M-MDSCs of midkine-depleted + + orthotopic Hepa 1-6 tumors were used for T-cell suppression assay. Ki67 (G) and Granzyme B (H) cytotoxic T cells were evaluated by ﬂow cytometry. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. (Supplementary Fig. 8A). The immunosuppression activity of were used in T-cell suppression assay. MDSCs derived from MDSCs is acted through various mechanisms. Among them, IL- midkine-overexpressing Hepa 1-6 tumors signiﬁcantly inhibited 10 secreted by MDSCs promotes immunosuppression via target- the proliferation and activation of cytotoxic T cells compared with ing a variety of immune cells [30]. In our study, IL-10 expression MDSCs derived from EV-transduced tumors, but the addition of was signiﬁcantly upregulated in sorafenib-treated HCC tumors neutralizing antibodies against IL-10 largely abrogated these + + - (Fig. 1E). Moreover, MDSCs sorted from midkine-overexpressing effects (Fig. 7D, E). Furthermore, CD11b CD33 HLA-DR MDSCs Hepa 1-6 tumors showed elevated IL-10 expression and produc- induced by conditional medium from midkine-overexpressing tion compared with MDSCs derived from EV-transduced tumors HUH-7 and SNU-449 cells showed increased IL-10 expression and (Fig. 7B, C). To clarify the potential role of IL-10 in immunosup- secretion compared with MDSCs induced by conditional medium pression activity of MDSCs, neutralizing antibodies against IL-10 from EV-transduced cells (Supplementary Fig. 8B and C). Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 6 Midkine inhibition enhances anti-PD-1 immunotherapy in Sorafenib-treated HCC tumors. A–D Hepa 1-6 cells transduced with midkine expression lentivirus or EV control were subcutaneously injected into C57BL/6 mice, then treated with antibody against murine PD-1 or IgG2b isotype control as indicated. Tumor growth curves (A), representative images (B), tumor weight (C), and inﬁltrating immune cells (D) were shown. E–H Hepa 1-6 cells transduced with Sh-MDK-1 or Sh-NC lentivirus were subcutaneously injected into C57BL/6 mice, then treated with sorafenib, vehicle, antibody against murine PD-1, or IgG2b isotype control as indicated. Tumor growth curves (E), representative images (F), tumor weight (G), and inﬁltrating immune cells (H) were shown. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. Moreover, midkine overexpression dramatically potentiated the density, augmented intratumoral hypoxia and activated HIF-1α + + - inhibitory functions of CD11b CD33 HLA-DR MDSCs on prolif- signaling in orthotopic HCC models. Emerging evidences indicate eration and activation of stimulated cytotoxic T cells, but this was that midkine is vital for hypoxia-driven tumor angiogenesis abolished by supplement of neutralizing antibodies against IL-10 [39, 40]. It is worth noting that the promoter region of midkine (Supplementary Fig. 8D and E). Collectively, our results proved that has a hypoxia responsive element which can binds with HIF-1α midkine overexpression promoted multiple pathways activation [27]. Thus, it was not surprising that midkine was signiﬁcantly and IL-10 production by MDSCs. upregulated by sorafenib-induced hypoxia in orthotopic HCC tumors. There are increasing evidences suggesting that persistent DISCUSSION sorafenib treatment gives rise to an immunosuppressive tumor Accumulated studies prove that midkine is involved in the microenvironment. For example, sorafenib treatment increases malignant progression of cancers, including HCC [16, 31]. Midkine intratumoral inﬁltrating of F4/80 tumor-associated macrophages, + + + + + is signiﬁcantly upregulated in tumor tissues and serum samples of CD4 CD25 FoxP3 regulatory T cells, CCL2 /CCL17 tumor- + + HCC patients, and associated with poor overall survival [24, 32, 33]. associated neutrophils and CD11b Gr-1 myeloid cells in human Midkine is reported to protect HCC cells against TRAIL-mediated and murine HCC models [37, 41]. Besides, immunosuppressive M2 apoptosis and cadmium-induced cellular damage [34, 35]. In macrophages accumulate more while activated Natural Killer cells addition, midkine overexpression facilitates metastasis of HCC fail to proliferate and produce effector molecules in sorafenib- cells via increasing anoikis resistance [36]. Upregulation of midkine resistant tumors [42, 43]. Similarly, our data demonstrated that the in HCC samples was also demonstrated by us via analyzing data percentages of Treg, TAM, and MDSC were increased, while from public database. What’s more, we found that sorafenib cytotoxic T cells were decreased in sorafenib-treated tumors, treatment promoted midkine expression and secretion by HCC indicating an immune-resistant HCC tumor microenvironment. To cells. Sorafenib is a multikinase inhibitor with antiangiogenic elucidate the potential molecular mechanism underlying this, we property. Previous studies demonstrate that sustained sorafenib found that midkine was signiﬁcantly upregulated by sorafenib treatment reduces microvascular density and increases intratu- treatment in HCC tumors. More importantly, forced midkine moral hypoxia of HCC tumors [37, 38]. Corresponding with this, expression stimulated immunosuppressive MDSCs accumulation our data proved that sorafenib treatment decreased microvessel in HCC microenvironment, while knockdown of midkine Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 7 Midkine overexpression promotes multiple pathways activation and IL-10 production by MDSCs. A–C PMN-MDSCs and M-MDSCs were sorted from midkine-overexpressing or EV-transduced orthotopical Hepa 1-6 tumors with/without sorafenib treatment, then collected cell lysates for western blot (A). Relative IL-10 expression (B) and production (C) were evaluated by qRT-PCR and ELISA assay. D, E PMN-MDSCs, and M-MDSCs sorted from midkine-overexpressing or EV-transduced orthotopic Hepa 1-6 tumors with/without sorafenib treatment were + + used for T-cell suppression assay. Ki67 (D) and Granzyme B (E) cytotoxic T cells were evaluated by ﬂow cytometry. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. suppressed immunosuppressive MDSCs inﬁltrating into HCC ﬁrst line therapy, as atezolizumab (anti-PD-L1) and bevacizumab tumors. These results suggested midkine might partially account (anti-VEGF) improves the median overall survival for more than for sorafenib-induced immunosuppressive microenvironment. 17 months [13]. This is consistent with the idea that checkpoint Indeed, midkine is well-characterized for its function in the blockade may be more efﬁcacious as combination therapy. immune system, including promoting immune cell chemotaxis Nevertheless, only one in four patients respond to the combina- and sculpting myeloid cell phenotype. For instance, midkine tion therapy, and the rest majority do not respond for unknown promotes neutrophils trafﬁcking and adhesion during acute reasons. Sorafenib shows antiangiogenic activity by targeting a inﬂammation via inducing high afﬁnity conformation of β2 variety of tyrosine kinase receptors, including VEGF receptor 2 and integrins [19]. During ischemic renal injury, midkine deﬁcient 3. In our study, we found that prolonged sorafenib exposure mice show reduced number of inﬁltrating neutrophils and altered HCC microenvironment towards an immune-resistant macrophages into the tubulointerstitium and impaired induction state, and anti-PD-1 antibody showed no obvious inﬂuence on of macrophage inﬂammatory protein-2 and macrophage chemo- tumor growth of sorafenib-treating tumors. Similarly, the response tactic protein-1 [44]. Midkine is also critical for fostering the rate of anti-PD-1 drug nivolumab is lower in sorafenib experienced immunosuppressive microenvironment. In melanoma, midkine patient than sorafenib-naïve patients (16–19% vs. 23%) [11]. In our overexpression gives rise to an immunosuppressive tumor study, midkine was signiﬁcantly induced by sorafenib in HCC microenvironment via promoting intratumoral recruitment of tumors. Knockdown of midkine dramatically increased the myeloid cells and tumor-associated macrophages [22]. In ErbB inhibitory effects of anti-PD-1 antibody on sorafenib-treating pathway-mutated gallbladder cancer, upregulation of secreted tumors, and this was partially due to the destruction of sorafenib- midkine facilitates differentiation of immunosuppressive macro- induced immunosuppressive microenvironment. In addition, phages through biding with its receptor LRP1 [23]. Correspond- forced midkine expression apparently attenuated the suppression ingly, we found that midkine-modulated secretome promoted of anti-PD-1 antibody on HCC tumor growth. These results + + - expansion of CD11b CD33 HLA-DR MDSCs from human PBMCs. suggested that the immunocompromised microenvironment In recent years, immune checkpoint inhibitors such as anti-PD-1 caused by VEGF inhibitors such as sorafenib might account for or anti-PD-L1 become an alternative therapeutic choice for HCC the low response rate of the combination therapy of anti-PD-1/ treatment, however the majority of patients do not respond to anti-PD-L1 antibodies and VEGF inhibitors. Meanwhile, breakdown them when use as single agent [11, 45]. In contrast, the of immune-tolerant state such as midkine inhibition in our study combination of anti-PD-1/anti-PD-L1 antibodies with vascular might increase the efﬁciency of the combination therapy. This is endothelial growth factor (VEGF) inhibitors has become a new also demonstrated by many other studies. For instance, inhibition Cell Death Discovery (2023) 9:92 L. Ding et al. of CXCR4 by AMD3100 signiﬁcantly increased the efﬁciency of 40 ng/mL IL-6 (Peprotech, Germany). Cells were cultured in a humidiﬁed atmosphere at 37 ˚Cwith 5% CO . Sorafenib (Selleck Chemicals, USA) was anti-PD-1 immunotherapy in sorfaenib-treated HCC tumors [37]. 2 dissolved in dimethylsulfoxide (DMSO), thus DMSO was used as vehicle MDSCs are essential components of the suppressive HCC tumor control. microenvironment. The frequency of inﬁltrating MDSCs is tightly correlated with the prognosis of HCC patients and efﬁciency of immune checkpoint inhibitors and tyrosine kinase inhibitors. For Plasmid constructs Human or mouse midkine expression lentivirus vector was constructed by example, Ly6G MDSCs are evidently increased in sorafenib- cloning the coding sequence of human or mouse midkine into the pCDH treated orthotopic liver tumors, and targeting Ly6G MDSCs by lentivirus vector (System Biosciences #CD510B). The empty pCDH lentivirus anti-Ly6G antibody apparently improves the inhibitory effects of vector was used as empty vector control (EV). To deplete midkine sorafenib via decreasing inﬁltrating MDSCs and enhancing expression, short hairpin RNAs targeting mouse midkine (Sh-MDK-1 and intratumoral cytotoxic T cells [46]. In HCC, CCRK promotes Sh-MDK-2) or human midkine (Sh-MDK-3 and Sh-MDK-4) were cloned into immunosuppressive MDSCs expansion and accumulation via NF- the pLKO.1 plasmid. The pLKO.1 plasmid inserted with a non-targeting кB/IL-6 signaling [47]. Depletion of CCRK enhances PD-L1 blockade sequences was used as non-targeting control (Sh-NC). The shRNA efﬁciency via reducing MDSC accumulation and increasing sequences were listed in Supplementary Table 2. intratumorous cytotoxic T cells. In our study, midkine over- expression signiﬁcantly increased the intratumoral accumulation Quantitative real-time polymerase chain reaction (qRT-PCR) of MDSCs, thus diminished the efﬁciency of PD-1 blockade in Total RNAs from tissue samples or cell lines were extracted by the TRIzol subcutaneous Hepa 1-6 tumors. On the contrary, midkine reagent (Takara, Japan). Complementary DNA strands were reverse- inhibition reduced the percentage of inﬁltrating MDSCs in transcribed by the ﬁrst-strand cDNA synthesis kit (Takara, Japan). SYBR sorafenib-treated Hepa 1-6 tumors, thus enhanced the inhibitory Green PCR Master Mix (Applied Biosystems, USA) were used for qRT-PCR analysis on the ViiATM7/QuantStudio 7 Flex Real Time PCR System (Applied effects of anti-PD-1 antibody. MDSCs are generated and activated Biosystems, USA). GAPDH was used as internal control. Relative gene through a complicated process involving multiple signaling −ΔΔCq expression was calculated by the 2 method. The primers used for pathways, such as STAT3, NF-кB and ERK [48]. In our study, we qRT-PCR analysis were listed in Supplementary Table 3. found that MDSCs from midkine-overexpression HCC tumors showed increasing activation of NF-кB, Akt, ERK, and Western blot STAT3 signaling. Moreover, IL-10 production by MDSCs was Protein lysates from tissues samples or culture cells were prepared using evidently increased by midkine overexpression. IL-10 is one of RIPA lysis buffer (Beyotime, China) supplemented with protease inhibitors the several MDSCs-secreted immunosuppressive effectors [49, 50]. (Beyotime, China). Bradford reagent (Sigma, USA) was used to measure In our study, neutralizing antibodies against IL-10 largely protein concentration. A total of 10–40 μg protein lysates were separated abrogated the inhibitory effects of MDSCs on proliferation and by 8–12% SDS-PAGE gels and transferred to polyvinylidene ﬂuoride activation of cytotoxic T cells. These results partially explained the membranes (GE Healthcare, UK). The membranes were blocked by 5% potential inﬂuence of midkine on generation and activation of non-fat milk, then incubated with speciﬁc ﬁrst antibodies at 4 ˚C overnight MDSCs in HCC tumors. and corresponding second antibodies at room temperature for 1 h. The In summary, we found that sorafenib treatment increased western bands were detected by chemiluminescence imaging (Biorad, USA) using ECL kit (GE Healthcare, UK). Image J was used to quantify the protein intratumoral hypoxia and altered HCC microenvironment towards expression. Antibodies used in our study were: HIF-1α Rabbit mAb (CST an immune-resistant state in mouse models. Sorafenib treatment #36169, 1: 1000), GAPDH Rabbit mAb (CST #5174, 1: 5000), PD-L1 Rabbit facilitated midkine expression and secretion by HCC cells. Moreover, mAb (CST #13684, 1: 1000), mouse midkine antibody (Abcam #ab281534, 1: midkine overexpression stimulated immunosuppressive MDSC accu- 1000), human midkine antibody (Abcam #ab52637, 1: 1000), Erk1/2 Rabbit mulation in HCC tumor microenvironment, while midkine inhibition mAb (CST #4695, 1: 1000); Phospho-Erk1/2 (Thr202/Tyr204) Rabbit mAb exhibited opposite effects. In addition, midkine inhibition enhanced (CST #4370, 1: 1000); STAT3 Mouse mAb (CST #9139, 1: 1000), Phospho- anti-PD-1 immunotherapy in sorafenib-treated HCC tumors. Midkine STAT3 (Tyr705) Rabbit mAb(CST #73533, 1: 1000), Akt Antibody (CST #9272, overexpression promoted multiple pathways activation and IL-10 1: 1000), Phospho-Akt (Ser473) Rabbit mAb (CST #4060, 1: 1000). secretion by MDSCs. Our data provided a novel role of midkine in sorafenib-induced immunosuppressive microenvironment. Midkine Immunohistochemistry (IHC) staining might be a potential target for the combination of anti-PD-1 Tissue samples were ﬁxed by formalin and embedded by parafﬁn. Tissue immunotherapy in HCC patients. sections (5 μm) were deparafﬁnized, rehydrated, and rinsed in distilled water. Antigen retrieval was conducted using citrate buffer (pH 6.0). The endogenous peroxidase activity was quenched by 3% hydrogen peroxide. For pimonidazole staining, the sections were incubated with monoclonal MATERIALS AND METHODS mouse antibodies against pimonidazole (1:50, Hypoxyprobe Inc., USA) at Patient samples 37 ˚C for 30 min. For HIF-1α staining, the sections were incubated with HIF- The collection and use of human samples were approved by the Ethics 1α Rabbit mAb (CST #36169, 1: 100) at 4 ˚C overnight. Next, the sections Committee of the First Hospital of Jilin University. The study was were stained with horseradish peroxidase anti-rabbit antibody and performed according to the guidelines with the Declaration of Helsinki detected by DAB kit (Thermo Fisher, USA). The nuclei were stained with and Ethics Committee of the First Hospital of Jilin University. Informed DAPI (Sigma, USA) at room temperature for 10 min. Images were obtained consents were obtained from all enrolled participants. Four pairs of HCC by Olympus FV1000 confocal microscopy. specimens and adjacent normal tissues were collected from the First Hospital of Jilin University between March 2019 and June 2019. Ten blood samples from healthy donors were collected, then peripheral blood Flow cytometry mononuclear cells (PBMCs) were isolated by Ficoll reagents (Sigma, USA) The orthotopic HCC tumor tissues or subcutaneous xenografts were according to the manufacturers’ instructions. dissected out and minced. Then tissues were digested with 0.8 mg/mL Collagenase IV (Sigma, USA) at 37 ˚C for 1 h. The cell suspensions were ﬁltered through 70 μm strainer and resuspended in 36% Percoll (GE Cell culture and reagents Healthcare, UK). PBMCs of anonymous human healthy donors were Murine HCC cell lines Hepa 1-6 and Hepa 1c1c7 and Human HCC cell lines isolated by Ficoll reagent (Sigma, USA) according to manufacturers’ HUH-7, SNU-449, HUH-6, and HepG2 were maintained in Dulbecco’smodiﬁed instructions. For cell surface staining, 1 × 10 cells were incubated with Eagle’s medium (DMEM, Gibco, USA) medium supplemented with 10% fetal anti-Fc receptor blocking antibody (2.4G2) at 4 ˚C for 15 min. Murine bovine serum (FBS, Gibco, USA), 100 units/mL penicillin and streptomycin samples were stained with anti-mouse CD45 APC, CD3 FITC, Gr-1 V450, (Gibco,USA). SortedMDSCs were cultured andexpandedby RPMImedium CD4 PE, CD8a V450, CD25 APC-CY7, Ly6C FITC, Ly6G PECY7, CD11 PE, and (Gibco, USA) supplemented with 10% FBS, 100 units/mL penicillin and F4/80 APC-CY7 from BD Bioscience (USA). Human samples were stained streptomycin (Gibco, USA), 40 ng/mL GM-CSF (Peprotech, Germany) and Cell Death Discovery (2023) 9:92 L. Ding et al. with anti-human CD45 APC, CD11 PE, CD33 FITC, and HLA-DR V450 from constructed by subcutaneously injected Hepa 1-6 cells (1 × 10 )intothe left BD Bioscience (USA). For intracellular staining of Arg1, Foxp3, and Ki67, ﬂank of 8-week-old male C57BL/6 mice. Two weeks after tumor injection, cells were ﬁxed and permeabilized by Fixation/Permeabilization solution mice were gave daily gavage of 50 mg/kg sorafenib (in PBS/1% Tween80), (BD Biosciences, USA) at 4 ˚C for 15 min. Then cells were washed and antibody against murine PD-1 (100 mg intraperitoneally every 4 days for ﬁve stained with anti-mouse Arg1, anti-mouse Foxp3, and anti-Ki67 from BD times, BioXCell, USA) or the IgG2b control isotype (100 mg intraperitoneally Bioscience (USA). Flow cytometry was performed on a B.D. Inﬂux cell sorter every 4 days for ﬁve times, BioXCell, USA). Tumor length was measures by (BD Bioscience, USA). Flowjo software was used to analyze the data. caliper every 3 days. At the end of drug treatment, the mice were anaesthetized by 3% isoﬂurane and sacriﬁced by broking the neck. Tumor xenografts were dissected out and weighed. Transcriptome RNA-sequencing and TCGA data analysis Total RNAs from sorafenib-treated or vehicle-treated orthotopic Hepa 1-6 tumors were extracted by TRIzol reagent (Thermo Fisher, USA). The Statistical analysis sequencing library was prepared using the Illumina’s TruSeq Stranded Statistical analysis was performed by GraphPad Prism 8 (GraphPad mRNA Sample Preparation kit (Illumina, USA) and sequenced on the Software, La Jolla, CA). Two-tailed Student’s t test and One-way ANOVA Illumina HiSeq 2500 platform. Fifty-base-pair sequenced reads were (Tukey’s post-hoc test) were used to compare difference between two or evaluated by the nextpresso pipeline (http://bioinfo.cnio.es/nextpresso/). more groups respectively. Data were shown as mean ± standard deviation The RNA sequencing data for the TCGA hepatocellular carcinoma samples (x ± s.d). P < 0.05 was considered as statistically signiﬁcant. were downloaded from the GDC Data Portal (https:// gdac.broadinstitute.org/) and analyzed as described. FastQC v0.11.0 was used to check sequencing quality. RNA-BisSeq method was performed to DATA AVAILABILITY ﬁlter the raw reads. The raw reads was mapped to mouse genome The data that support the ﬁndings of this study are available on request from the (GRCm38/mm10) or human genome (GRCh37/hg19) using the TopHat- corresponding author. 2.0.10 and evaluated by HTSeq. Differentially expressed genes were analyzed by DESeq2 and deﬁned as |Log Fold Change| ≥ 1.5 and adjusted p < 0.05. Samples were sequenced with three repeats. REFERENCES 1. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Tumor immune estimation resource (TIMER) and tumor- Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality world- immune system interactions database (TISIDB) analysis wide for 36 cancers in 185 countries. CA. 2021;71:209–49. TIMER was used to investigate the interactions between genes and tumor 2. Kulik L, El-Serag HB. Epidemiology and management of hepatocellular carcinoma. immune interactions (https://cistrome.shinyapps.io/timer/). The TISIDB was Gastroenterology. 2019;156:477–491.e471. used to analyze the tumor and immune system interaction (http:// 3. Bruix J, Reig M, Sherman M. Evidence-based diagnosis, staging, and treatment of cis.hku.hk/TISIDB/). In the present study, the correlation of midkine patients with hepatocellular carcinoma. Gastroenterology. 2016;150:835–53. expression with tumor immune inﬁltration was investigated by TIMER 4. Cheng AL, Kang YK, Chen Z, Tsao CJ, Qin S, Kim JS, et al. Efﬁcacy and safety of and TISIDB with the HCC database. The 22 tumor-inﬁltrating immune cells sorafenib in patients in the Asia-Paciﬁc region with advanced hepatocellular in HCC tissues were estimated by CIBERSORT algorithm. carcinoma: a phase III randomised, double-blind, placebo-controlled trial. Lancet Oncol. 2009;10:25–34. 5. Llovet JM, Ricci S, Mazzaferro V, Hilgard P, Gane E, Blanc JF, et al. Sorafenib in ELISA assay advanced hepatocellular carcinoma. N. Engl J Med. 2008;359:378–90. The levels of midkine, IL-10, and IFN-γ in serum samples or conditional 6. Tang W, Chen Z, Zhang W, Cheng Y, Zhang B, Wu F, et al. The mechanisms of medium were evaluated by ELISA kit (Abcam # ab193761, # ab279416, # sorafenib resistance in hepatocellular carcinoma: theoretical basis and ther- ab185986, #ab255729 and #ab174443, USA) according to manufacturer’s apeutic aspects. Signal Transduct Target Ther. 2020;5:87. instructions. Brieﬂy, samples were incubated with antibody-coated plates 7. Kudo M, Finn RS, Qin S, Han KH, Ikeda K, Piscaglia F, et al. Lenvatinib versus for 1 h at room temperature and washed for ﬁve times. Then plates were sorafenib in ﬁrst-line treatment of patients with unresectable hepatocellular incubated with biotinylated antibody, streptavidin antibody, substrate carcinoma: a randomised phase 3 non-inferiority trial. Lancet. 2018;391:1163–73. solution, and stop solution consecutively. The absorbance at 450 nm was 8. Yang YM, Kim SY, Seki E. Inﬂammation and liver cancer: molecular mechanisms measured by a microplate reader. All samples were done in triplicates. and therapeutic targets. Semin Liver Dis. 2019;39:26–42. 9. Fujiwara M, Garo LP, Murugaiyan G. PD1 blockade in cancer: impact on myeloid cells. Trends Cancer. 2020;6:443–4. T cell suppression assay 5 + + 10. Kudo M. Scientiﬁc rationale for combined immunotherapy with PD-1/PD-L1 5×10 CD3 CD8 cytotoxic T cells were stimulated with CD3/CD28 antibodies and VEGF inhibitors in advanced hepatocellular carcinoma. Cancers. dynabeads (Invitrogen, USA) in 24-well plates. At the same time, 5 × 10 2020;12:1089. sorted MDSCs were added into each well in the presence of human 11. Zhu AX, Finn RS, Edeline J, Cattan S, Ogasawara S, Palmer D, et al. Pembrolizumab recombinant IL-2 (R&D, USA) for 3 days. In separated experiment, cells in patients with advanced hepatocellular carcinoma previously treated with were supplemented with IL-10 antibody (Abcam #ab133575 or #ab189392, sorafenib (KEYNOTE-224): a non-randomised, open-label phase 2 trial. Lancet USA) or IgG isotype control (Abcam, USA). The cells were then stained with Oncol. 2018;19:940–52. surface markers of CD3/CD4/CD8 and intracellular protein of Ki67 12. El-Khoueiry AB, Sangro B, Yau T, Crocenzi TS, Kudo M, Hsu C, et al. Nivolumab in (eBioscience, USA) and Granzyme B (eBioscience, USA) for ﬂow cytometry. patients with advanced hepatocellular carcinoma (CheckMate 040): an open- All samples were done with three repeats. label, non-comparative, phase 1/2 dose escalation and expansion trial. Lancet. 2017;389:2492–502. Murine HCC models 13. Finn RS, Qin S, Ikeda M, Galle PR, Ducreux M, Kim TY, et al. Atezolizumab plus Hepa 1-6 or Hepa 1c1c7 cells were stably transduced with midkine, EV, sh- bevacizumab in unresectable hepatocellular carcinoma. N Engl J Med. MDK-1, sh-MDK-2, or sh-NC lentiviral particles as indicated for animal 2020;382:1894–905. experiments. Orthotopic HCC models were constructed as previously 14. Jain RK. Antiangiogenesis strategies revisited: from starving tumors to alleviating described [37]. 1 × 10 Hepa 1-6 or Hepa 1c1c7 cells were orthotopically hypoxia. Cancer Cell. 2014;26:605–22. implanted in the liver of 8-week-old male C57BL/6 mice. Brieﬂy, mice were 15. Mo Z, Liu D, Rong D, Zhang S. Hypoxic characteristic in the immunosuppressive anaesthetized by ketamine/xylazine, then median liver lobe was exposed by microenvironment of hepatocellular carcinoma. Front Immunol. 2021;12:611058. median laparotomy. 1 × 10 cells (suspended in 100 μL 1: 1 PBS and matrigel 16. Filippou PS, Karagiannis GS, Constantinidou A. Midkine (MDK) growth factor: a mixture) were orthotopically injected in the subcapsular region of liver by a key player in cancer progression and a promising therapeutic target. Oncogene. 28 G needle. One week after implantation, these mice were randomly divided 2020;39:2040–54. into sorafenib treatment group and vehicle group. Sorafenib treatment 17. Muramatsu T. Structure and function of midkine as the basis of its pharmaco- group received daily gavage of 50 mg/kg sorafenib (in PBS/1% Tween80), logical effects. Br J Pharm. 2014;171:814–26. while vehicle group received equal volume of PBS/1% Tween80. Mice were 18. Takemoto Y, Horiba M, Harada M, Sakamoto K, Takeshita K, Murohara T, et al. sacriﬁced 4 weeks post tumor implantation. Orthotopic HCC tumors were Midkine promotes atherosclerotic plaque formation through its pro-inﬂamma- dissected out, and inﬁltrating immune cells were isolated and evaluated by tory, angiogenic and anti-apoptotic functions in apolipoprotein E-knockout mice. ﬂow cytometry. In separated experiments, HCC tumor xenograft model were Circ J. 2017;82:19–27. Cell Death Discovery (2023) 9:92 L. Ding et al. 19. Weckbach LT, Gola A, Winkelmann M, Jakob SM, Groesser L, Borgolte J, et al. The 44. Sato W, Kadomatsu K, Yuzawa Y, Muramatsu H, Hotta N, Matsuo S, et al. Midkine cytokine midkine supports neutrophil trafﬁcking during acute inﬂammation by is involved in neutrophil inﬁltration into the tubulointerstitium in ischemic renal promoting adhesion via beta2 integrins (CD11/CD18). Blood. 2014;123:1887–96. injury. J Immunol. 2001;167:3463–9. 20. Cohen S, Shoshana OY, Zelman-Toister E, Maharshak N, Binsky-Ehrenreich I, 45. Finn RS, Ryoo BY, Merle P, Kudo M, Bouattour M, Lim HY, et al. Pembrolizumab as Gordin M, et al. The cytokine midkine and its receptor RPTPzeta regulate B cell second-line therapy in patients with advanced hepatocellular carcinoma in survival in a pathway induced by CD74. J Immunol. 2012;188:259–69. KEYNOTE-240: a randomized, double-blind, phase III trial. J Clin Oncol. 21. Guo X, Pan Y, Xiong M, Sanapala S, Anastasaki C, Cobb O, et al. Midkine activation 2020;38:193–202. of CD8(+) T cells establishes a neuron-immune-cancer axis responsible for low- 46. Chang CJ, Yang YH, Chiu CJ, Lu LC, Liao CC, Liang CW, et al. Targeting tumor- grade glioma growth. Nat Commun. 2020;11:2177. inﬁltrating Ly6G(+) myeloid cells improves sorafenib efﬁcacy in mouse ortho- 22. Cerezo-Wallis D, Contreras-Alcalde M, Troule K, Catena X, Mucientes C, Calvo TG, topic hepatocellular carcinoma. Int J Cancer. 2018;142:1878–89. et al. Midkine rewires the melanoma microenvironment toward a tolerogenic and 47. Zhou J, Liu M, Sun H, Feng Y, Xu L, Chan AWH, et al. Hepatoma-intrinsic CCRK immune-resistant state. Nat Med. 2020;26:1865–77. inhibition diminishes myeloid-derived suppressor cell immunosuppression and 23. Zhang Y, Zuo C, Liu L, Hu Y, Yang B, Qiu S, et al. Single-cell RNA-sequencing atlas enhances immune-checkpoint blockade efﬁcacy. Gut. 2018;67:931–44. reveals an MDK-dependent immunosuppressive environment in ErbB pathway- 48. Wang Y, Zhang T, Sun M, Ji X, Xie M, Huang W, et al. Therapeutic values of mutated gallbladder cancer. J Hepatol. 2021;75:1128–41. myeloid-derived suppressor cells in hepatocellular carcinoma: facts and hopes. 24. Zhu WW, Guo JJ, Guo L, Jia HL, Zhu M, Zhang JB, et al. Evaluation of midkine as a Cancers. 2021;13:5127. diagnostic serum biomarker in hepatocellular carcinoma. Clin Cancer Res. 49. Zheng W, Song H, Luo Z, Wu H, Chen L, Wang Y, et al. Acetylcholine ameliorates 2013;19:3944–54. colitis by promoting IL-10 secretion of monocytic myeloid-derived suppressor 25. Batlle E, Massague J. Transforming growth factor-beta signaling in immunity and cells through the nAChR/ERK pathway. Proc Natl Acad Sci USA. cancer. Immunity. 2019;50:924–40. 2021;118:e2017762118. 26. Grivennikov SI, Greten FR, Karin M. Immunity, inﬂammation, and cancer. Cell. 50. Yaseen MM, Abuharfeil NM, Darmani H, Daoud A. Mechanisms of immune sup- 2010;140:883–99. pression by myeloid-derived suppressor cells: the role of interleukin-10 as a key 27. Reynolds PR, Mucenski ML, Le Cras TD, Nichols WC, Whitsett JA. Midkine is immunoregulatory cytokine. Open Biol. 2020;10:200111. regulated by hypoxia and causes pulmonary vascular remodeling. J Biol Chem. 2004;279:37124–32. 28. Wu H, Zhen Y, Ma Z, Li H, Yu J, Xu ZG, et al. Arginase-1-dependent promotion of AUTHOR CONTRIBUTIONS TH17 differentiation and disease progression by MDSCs in systemic lupus ery- All authors guaranteed the integrity of the entire study. The experiments were thematosus. Sci Transl Med. 2016;8:331ra340. conducted by LD, NW, and QW Clinical studies were conducted by XF. Data were 29. Lopez-Valero I, Davila D, Gonzalez-Martinez J, Salvador-Tormo N, Lorente M, Saiz- analyzed by YX. The manuscript was prepared and reviewed by SW. All authors have Ladera C, et al. Midkine signaling maintains the self-renewal and tumorigenic read and approved the manuscript. capacity of glioma initiating cells. Theranostics. 2020;10:5120–36. 30. Saraiva M, Vieira P, O’Garra A. Biology and therapeutic potential of interleukin-10. J Exp Med. 2020;217:e20190418. ETHICAL APPROVAL AND CONSENT TO PARTICIPATE 31. Cai YQ, Lv Y, Mo ZC, Lei J, Zhu JL, Zhong QQ. Multiple pathophysiological roles of The experiments involved human subjects were followed with the ethical standards midkine in human disease. Cytokine 2020;135:155242. of the institutional committee and with the 1964 Helsinki declaration and its later 32. Zheng L, Li H, Huang J, Shin J, Luo S, Guo C, et al. Serum midkine levels for the amendments or comparable ethical standards. The animal studies were followed diagnosis and assessment of response to interventional therapy in patients with with the ethical standards of the ethics committee of the First Hospital of Jilin hepatocellular carcinoma. J Inter Med. 2021;4:39–45. University. 33. Omran MM, Farid K, Omar MA, Emran TM, El-Taweel FM, Tabll AA. A combination of alpha-fetoprotein, midkine, thioredoxin and a metabolite for predicting hepatocellular carcinoma. Ann Hepatol. 2020;19:179–85. 34. Yazihan N, Ataoglu H, Akcil E, Yener B, Salman B, Aydin C. Midkine secretion COMPETING INTERESTS protects Hep3B cells from cadmium induced cellular damage. World J Gastro- The authors declare no competing interests. enterol. 2008;14:76–80. 35. Ohuchida T, Okamoto K, Akahane K, Higure A, Todoroki H, Abe Y, et al. Midkine protects hepatocellular carcinoma cells against TRAIL-mediated apoptosis ADDITIONAL INFORMATION through down-regulation of caspase-3 activity. Cancer. 2004;100:2430–6. Supplementary information The online version contains supplementary material 36. Sun B, Hu C, Yang Z, Zhang X, Zhao L, Xiong J, et al. Midkine promotes hepa- available at https://doi.org/10.1038/s41420-023-01392-3. tocellular carcinoma metastasis by elevating anoikis resistance of circulating tumor cells. Oncotarget. 2017;8:32523–35. Correspondence and requests for materials should be addressed to Shudong Wang. 37. Chen Y, Ramjiawan RR, Reiberger T, Ng MR, Hato T, Huang Y, et al. CXCR4 inhibition in tumor microenvironment facilitates anti-programmed death Reprints and permission information is available at http://www.nature.com/ receptor-1 immunotherapy in sorafenib-treated hepatocellular carcinoma in reprints mice. Hepatology. 2015;61:1591–602. 38. Liang Y, Zheng T, Song R, Wang J, Yin D, Wang L, et al. Hypoxia-mediated Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims sorafenib resistance can be overcome by EF24 through Von Hippel-Lindau tumor in published maps and institutional afﬁliations. suppressor-dependent HIF-1alpha inhibition in hepatocellular carcinoma. Hepa- tology. 2013;57:1847–57. 39. Choudhuri R, Zhang HT, Donnini S, Ziche M, Bicknell R. An angiogenic role for the neurokines midkine and pleiotrophin in tumorigenesis. Cancer Res. 1997;57:1814–9. Open Access This article is licensed under a Creative Commons 40. Muramaki M, Miyake H, Hara I, Kamidono S. Introduction of midkine gene into Attribution 4.0 International License, which permits use, sharing, human bladder cancer cells enhances their malignant phenotype but increases adaptation, distribution and reproduction in any medium or format, as long as you give their sensitivity to antiangiogenic therapy. Clin Cancer Res. 2003;9:5152–60. appropriate credit to the original author(s) and the source, provide a link to the Creative 41. Zhou SL, Zhou ZJ, Hu ZQ, Huang XW, Wang Z, Chen EB, et al. Tumor-associated Commons license, and indicate if changes were made. The images or other third party neutrophils recruit macrophages and T-Regulatory cells to promote progression material in this article are included in the article’s Creative Commons license, unless of hepatocellular carcinoma and resistance to Sorafenib. Gastroenterology. indicated otherwise in a credit line to the material. If material is not included in the 2016;150:1646–1658.e1617. article’s Creative Commons license and your intended use is not permitted by statutory 42. Dong N, Shi X, Wang S, Gao Y, Kuang Z, Xie Q, et al. M2 macrophages mediate regulation or exceeds the permitted use, you will need to obtain permission directly sorafenib resistance by secreting HGF in a feed-forward manner in hepatocellular from the copyright holder. To view a copy of this license, visit http:// carcinoma. Br J Cancer. 2019;121:22–33. creativecommons.org/licenses/by/4.0/. 43. Li C, Wei S, Xu X, Jiang Y, Xue L, Jiang P, et al. Sorafenib attenuated the function of natural killer cells inﬁltrated in HCC through inhibiting ERK1/2. Int Immuno- pharmacol. 2019;76:105855. © The Author(s) 2023 Cell Death Discovery (2023) 9:92 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Cell Death Discovery Springer Journals http://www.deepdyve.com/lp/springer-journals/midkine-inhibition-enhances-anti-pd-1-immunotherapy-in-sorafenib-vUiLvIDTg2

Loading next page...

References (52)

Y. Takemoto, M. Horiba, M. Harada, K. Sakamoto, Kyosuke Takeshita, T. Murohara, K. Kadomatsu, K. Kamiya (2017)
Midkine Promotes Atherosclerotic Plaque Formation Through Its Pro-Inflammatory, Angiogenic and Anti-Apoptotic Functions in Apolipoprotein E-Knockout Mice.
Circulation journal : official journal of the Japanese Circulation Society, 82 1
M. Fujiwara, Lucien Garo, G. Murugaiyan (2020)
PD1 Blockade in Cancer: Impact on Myeloid Cells.
Trends in cancer, 6 6
Shaolai Zhou, Zheng-jun Zhou, Zhi-qiang Hu, Xiaowu Huang, Zheng Wang, E. Chen, Jia Fan, Ya Cao, Z. Dai, Jian Zhou (2016)
Tumor-Associated Neutrophils Recruit Macrophages and T-Regulatory Cells to Promote Progression of Hepatocellular Carcinoma and Resistance to Sorafenib.
Gastroenterology, 150 7
M. Kudo (2020)
Scientific Rationale for Combined Immunotherapy with PD-1/PD-L1 Antibodies and VEGF Inhibitors in Advanced Hepatocellular Carcinoma
Cancers, 12
Sivan Cohen, Or-yam Shoshana, Einat Zelman-Toister, N. Maharshak, I. Binsky-Ehrenreich, Maya Gordin, I. Hazan-Halevy, Y. Herishanu, L. Shvidel, M. Haran, L. Leng, R. Bucala, S. Harroch, I. Shachar (2012)
The Cytokine Midkine and Its Receptor RPTPζ Regulate B Cell Survival in a Pathway Induced by CD74
The Journal of Immunology, 188
JM Llovet, S Ricci, V Mazzaferro, P Hilgard, E Gane, JF Blanc (2008)
Sorafenib in advanced hepatocellular carcinoma
N. Engl J Med, 359
Yijian Zhang, Chunman Zuo, Liguo Liu, Yun-ping Hu, Boyi Yang, Shimei Qiu, Yang Li, Dongyan Cao, Zheng Ju, Jing Ge, Qiu Wang, Ting Wang, L. Bai, Yang Yang, Guoqiang Li, Z. Shao, Yuan Gao, Yongsheng Li, R. Bian, H. Miao, Lin Li, Xuechuan Li, Chengkai Jiang, Siyuan Yan, Ziyi Wang, Zeyu Wang, Xuya Cui, Wen Huang, Dongxi Xiang, Congjun Wang, Qi-yun Li, Xiangsong Wu, W. Gong, Yun Liu, R. Shao, Fatao Liu, Maolan Li, Luonan Chen, Yingbin Liu (2021)
Single-cell RNA-sequencing atlas reveals an MDK-dependent immunosuppressive environment in ErbB pathway-mutated gallbladder cancer.
Journal of hepatology
Ying-jian Liang, T. Zheng, Rui-peng Song, Jiabei Wang, Dalong Yin, Luoluo Wang, Haitao Liu, Lan-tian Tian, Xiang Fang, Xianzhi Meng, Hongchi Jiang, Jia-Ren Liu, Lianxin Liu (2013)
Hypoxia‐mediated sorafenib resistance can be overcome by EF24 through Von Hippel‐Lindau tumor suppressor‐dependent HIF‐1α inhibition in hepatocellular carcinoma
Hepatology, 57
A. Zhu, R. Finn, J. Edeline, S. Cattan, S. Ogasawara, D. Palmer, C. Verslype, V. Zagonel, L. Fartoux, A. Vogel, D. Sarker, G. Verset, S. Chan, J. Knox, B. Daniele, A. Webber, S. Ebbinghaus, Junshui Ma, A. Siegel, A. Cheng, M. Kudo (2018)
Pembrolizumab in patients with advanced hepatocellular carcinoma previously treated with sorafenib (KEYNOTE-224): a non-randomised, open-label phase 2 trial.
The Lancet. Oncology, 19 7
M. Muramaki, H. Miyake, I. Hara, S. Kamidono (2003)
Introduction of midkine gene into human bladder cancer cells enhances their malignant phenotype but increases their sensitivity to antiangiogenic therapy.
Clinical cancer research : an official journal of the American Association for Cancer Research, 9 14
LT Weckbach, A Gola, M Winkelmann, SM Jakob, L Groesser, J Borgolte (2014)
The cytokine midkine supports neutrophil trafficking during acute inflammation by promoting adhesion via beta2 integrins (CD11/CD18)
Blood., 123
Ningning Dong, Xiangyi Shi, Suihai Wang, Yan-jun Gao, Zhenzhan Kuang, Q. Xie, Yonglong Li, H. Deng, Yingsong Wu, Ming Li, Ji-Liang Li (2019)
M2 macrophages mediate sorafenib resistance by secreting HGF in a feed-forward manner in hepatocellular carcinoma
British Journal of Cancer, 121
Lin Zheng, Hailiang Li, Jinhua Huang, J. Shin, S. Luo, Chen-Yang Guo, Yan Zhao, Fangkun Li (2020)
Serum midkine levels for the diagnosis and assessment of response to interventional therapy in patients with hepatocellular carcinoma
Journal of Interventional Medicine, 4
R. Finn, B. Ryoo, P. Merle, M. Kudo, M. Bouattour, Ho Lim, V. Breder, J. Edeline, Y. Chao, S. Ogasawara, T. Yau, M. Garrido, S. Chan, J. Knox, B. Daniele, S. Ebbinghaus, E. Chen, A. Siegel, A. Zhu, A. Cheng (2019)
Pembrolizumab As Second-Line Therapy in Patients With Advanced Hepatocellular Carcinoma in KEYNOTE-240: A Randomized, Double-Blind, Phase III Trial.
Journal of clinical oncology : official journal of the American Society of Clinical Oncology
M. Saraiva, P. Vieira, A. O’Garra (2019)
Biology and therapeutic potential of interleukin-10
The Journal of Experimental Medicine, 217
Chunxiao Li, Shuhua Wei, Xiaofei Xu, Yuliang Jiang, L. Xue, P. Jiang, Junjie Wang (2019)
Sorafenib attenuated the function of natural killer cells infiltrated in HCC through inhibiting ERK1/2.
International immunopharmacology, 76
Y. Yang, S. Kim, E. Seki (2019)
Inflammation and Liver Cancer: Molecular Mechanisms and Therapeutic Targets
Seminars in Liver Disease, 39
Panagiota Filippou, G. Karagiannis, A. Constantinidou (2019)
Midkine (MDK) growth factor: a key player in cancer progression and a promising therapeutic target
Oncogene, 39
A. El-Khoueiry, B. Sangro, T. Yau, T. Crocenzi, M. Kudo, Chiun Hsu, Tae-You Kim, S. Choo, J. Trojan, T. Welling, T. Meyer, Yoon-Koo Kang, W. Yeo, A. Chopra, Jeffrey Anderson, C. Cruz, Lixin Lang, J. Neely, Hao Tang, H. Dastani, I. Melero (2017)
Nivolumab in patients with advanced hepatocellular carcinoma (CheckMate 040): an open-label, non-comparative, phase 1/2 dose escalation and expansion trial
The Lancet, 389
Xiaofan Guo, Yuan Pan, M. Xiong, Shilpa Sanapala, C. Anastasaki, Olivia Cobb, S. Dahiya, D. Gutmann (2020)
Midkine activation of CD8+ T cells establishes a neuron–immune–cancer axis responsible for low-grade glioma growth
Nature Communications, 11
Bin Sun, Cong-li Hu, Zhi-bin Yang, Xiaofeng Zhang, Linlin Zhao, Junye Xiong, Junyong Ma, Lei Chen, H. Qian, Xiang-ji Luo, Le-hua Shi, Jun Li, Xianshuo Cheng, Z. Yin (2017)
Midkine promotes hepatocellular carcinoma metastasis by elevating anoikis resistance of circulating tumor cells
Oncotarget, 8
N. Yazıhan, H. Ataŏglu, Ethem Akçıl, Burcu Yener, B. Salman, C. Aydın (2008)
Midkine secretion protects Hep3B cells from cadmium induced cellular damage.
World journal of gastroenterology, 14 1
R. Choudhuri, Hua-Tang Zhang, S. Donnini, M. Ziche, Roy Bicknell (1997)
An angiogenic role for the neurokines midkine and pleiotrophin in tumorigenesis.
Cancer research, 57 9
R. Finn, S. Qin, K. Han, F. Piscaglia, A. Baron, Joong-Won Park, G. Han, J. Jassem, Jean Blanc, A. Vogel, D. Komov, T. Evans, C. Lopez, C. Dutcus, M. Guo, Kenichi Saito, S. Kraljevic, T. Tamai, M. Ren, A. Cheng (2018)
Lenvatinib versus sorafenib in first-line treatment of patients with unresectable hepatocellular carcinoma: a randomised phase 3 non-inferiority trial
The Lancet, 391
R. Jain (2014)
Antiangiogenesis strategies revisited: from starving tumors to alleviating hypoxia.
Cancer cell, 26 5
Chun-Jung Chang, Yao-Hsu Yang, Chiao-Juno Chiu, Li‐Chun Lu, C. Liao, C. Liang, Chih-Hung Hsu, A. Cheng (2018)
Targeting tumor‐infiltrating Ly6G+ myeloid cells improves sorafenib efficacy in mouse orthotopic hepatocellular carcinoma
International Journal of Cancer, 142
A. Cheng, Yoon-Koo Kang, Zhendong Chen, C. Tsao, S. Qin, Jun Kim, R. Luo, Jifeng Feng, S. Ye, Tsai‐Sheng Yang, Jianming Xu, Yan Sun, Houjie Liang, Jiwei Liu, Jie-jun Wang, W. Tak, H. Pan, K. Burock, J. Zou, D. Voliotis, Z. Guan (2009)
Efficacy and safety of sorafenib in patients in the Asia-Pacific region with advanced hepatocellular carcinoma: a phase III randomised, double-blind, placebo-controlled trial.
The Lancet. Oncology, 10 1
L. Kulik, H. El‐Serag (2019)
Epidemiology and Management of Hepatocellular Carcinoma.
Gastroenterology, 156 2
Yunching Chen, R. Ramjiawan, T. Reiberger, M. Ng, T. Hato, Yuhui Huang, H. Ochiai, Shuji Kitahara, E. Unan, T. Reddy, C. Fan, Peigen Huang, N. Bardeesy, A. Zhu, R. Jain, D. Duda (2015)
CXCR4 inhibition in tumor microenvironment facilitates anti‐programmed death receptor‐1 immunotherapy in sorafenib‐treated hepatocellular carcinoma in mice
Hepatology, 61
R. Finn, S. Qin, M. Ikeda, P. Galle, M. Ducreux, Tae-You Kim, M. Kudo, V. Breder, P. Merle, A. Kaseb, Daneng Li, W. Verret, Derek-Zhen Xu, S. Hernandez, Juan Liu, Chen Huang, S. Mulla, Yulei Wang, Ho Lim, A. Zhu, A. Cheng (2020)
Atezolizumab plus Bevacizumab in Unresectable Hepatocellular Carcinoma.
The New England journal of medicine, 382 20
J. Bruix, M. Reig, M. Sherman (2016)
Evidence-Based Diagnosis, Staging, and Treatment of Patients With Hepatocellular Carcinoma.
Gastroenterology, 150 4
M. Yaseen, N. Abuharfeil, H. Darmani, Ammar Daoud (2020)
Mechanisms of immune suppression by myeloid-derived suppressor cells: the role of interleukin-10 as a key immunoregulatory cytokine
Open Biology, 10
E. Batlle, J. Massagué (2019)
Transforming Growth Factor-β Signaling in Immunity and Cancer.
Immunity, 50 4
Hao Wu, Y. Zhen, Zhanchuan Ma, Huimin Li, Jinyu Yu, Zhong-Gao Xu, Xiang-Yang Wang, H. Yi, Yong-Guang Yang (2016)
Arginase-1–dependent promotion of TH17 differentiation and disease progression by MDSCs in systemic lupus erythematosus
Science Translational Medicine, 8
H. Sung, J. Ferlay, R. Siegel, M. Laversanne, I. Soerjomataram, A. Jemal, F. Bray (2021)
Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries
CA: A Cancer Journal for Clinicians, 71
Yijun Wang, Tongyue Zhang, Mengyu Sun, Xiaoyu Ji, Meng Xie, Wenjie Huang, L. Xia (2021)
Therapeutic Values of Myeloid-Derived Suppressor Cells in Hepatocellular Carcinoma: Facts and Hopes
Cancers, 13
W. Sato, K. Kadomatsu, Y. Yuzawa, H. Muramatsu, N. Hotta, S. Matsuo, T. Muramatsu (2001)
Midkine Is Involved in Neutrophil Infiltration into the Tubulointerstitium in Ischemic Renal Injury1
The Journal of Immunology, 167
Wenwei Zhu, Jia-Jian Guo, Lei Guo, H. Jia, Ming-Yue Zhu, Ju-bo Zhang, C. Loffredo, M. Forgues, Hua Huang, Xu Xing, N. Ren, Q. Dong, Hai-jun Zhou, Zheng-Gang Ren, Nai-Qing Zhao, X. Wang, Zhao-You Tang, L. Qin, Q. Ye (2013)
Evaluation of Midkine as a Diagnostic Serum Biomarker in Hepatocellular Carcinoma
Clinical Cancer Research, 19
M. Omran, K. Farid, Mona Omar, T. Emran, F. El-Taweel, A. Tabll (2020)
A combination of α-fetoprotein, midkine, thioredoxin and a metabolite for predicting hepatocellular carcinoma.
Annals of hepatology
G. Spinzi, S. Paggi (2008)
Sorafenib in advanced hepatocellular carcinoma.
The New England journal of medicine, 359 23
T. Ohuchida, K. Okamoto, K. Akahane, A. Higure, H. Todoroki, Yukio Abe, M. Kikuchi, S. Ikematsu, T. Muramatsu, H. Itoh (2004)
Midkine protects hepatocellular carcinoma cells against TRAIL‐mediated apoptosis through down‐regulation of caspase‐3 activity
Cancer, 100
Takashi Muramatsu (2014)
Structure and function of midkine as the basis of its pharmacological effects
British Journal of Pharmacology, 171
Yaqin Cai, Yuncheng Lv, Z. Mo, J. Lei, Jing-Ling Zhu, Q. Zhong (2020)
Multiple pathophysiological roles of midkine in human disease.
Cytokine, 135
Weiwei Tang, Ziyi Chen, Wenling Zhang, Ye Cheng, Betty Zhang, Fan Wu, Qian Wang, Shouju Wang, D. Rong, F. Reiter, E. Toni, Xuehao Wang (2020)
The mechanisms of sorafenib resistance in hepatocellular carcinoma: theoretical basis and therapeutic aspects
Signal Transduction and Targeted Therapy, 5
Wanwei Zheng, Huan Song, Zhongguang Luo, Hao-Ran Wu, Lin Chen, Yuedi Wang, Haoshu Cui, Yufei Zhang, Bangting Wang, Wenshuai Li, Yao Liu, Jun Zhang, Y. Chu, Feifei Luo, Jie Liu (2021)
Acetylcholine ameliorates colitis by promoting IL-10 secretion of monocytic myeloid-derived suppressor cells through the nAChR/ERK pathway
Proceedings of the National Academy of Sciences, 118
S. Grivennikov, F. Greten, M. Karin (2010)
Immunity, Inflammation, and Cancer
Cell, 140
Jingying Zhou, Man Liu, Hanyong Sun, Yu Feng, Liangliang Xu, A. Chan, J. Tong, J. Wong, C. Chong, P. Lai, H. Wang, S. Tsang, Tyler Goodwin, Rihe Liu, Leaf Huang, Zhiwei Chen, J. Sung, K. Chow, K. To, A. Cheng (2017)
Hepatoma-intrinsic CCRK inhibition diminishes myeloid-derived suppressor cell immunosuppression and enhances immune-checkpoint blockade efficacy
Gut, 67
P. Reynolds, M. Mucenski, T. Cras, W. Nichols, J. Whitsett (2004)
Midkine Is Regulated by Hypoxia and Causes Pulmonary Vascular Remodeling*
Journal of Biological Chemistry, 279
Zhuomao Mo, Daiyuan Liu, Dade Rong, Shijun Zhang (2021)
Hypoxic Characteristic in the Immunosuppressive Microenvironment of Hepatocellular Carcinoma
Frontiers in Immunology, 12
L. Weckbach, Anita Gola, M. Winkelmann, S. Jakob, L. Groesser, J. Borgolte, Frank Pogoda, R. Pick, M. Pruenster, J. Müller‐Höcker, E. Deindl, M. Sperandio, B. Walzog (2014)
PHAGOCYTES , GRANULOCYTES , AND MYELOPOIESIS The cytokine midkine supports neutrophil traf fi cking during acute in fl ammation by promoting adhesion via b 2 integrins ( CD 11 / CD 18 )
I. López‐Valero, D. Dávila, J. González-Martínez, Nélida Salvador-Tormo, M. Lorente, Cristina Sáiz‐Ladera, Sofía Torres, Estibaliz Gabicagogeascoa, S. Hernández‐Tiedra, E. García‐Taboada, M. Mendiburu-Eliçabe, Fátima Rodríguez-Fornés, R. Sánchez-Domínguez, J. Segovia, P. Sánchez-Gómez, A. Matheu, J. Sepúlveda, G. Velasco (2020)
Midkine signaling maintains the self-renewal and tumorigenic capacity of glioma initiating cells
Theranostics, 10
Daniela Cerezo-Wallis, Marta Contreras-Alcalde, K. Troulé, Xavier Catena, Cynthia Mucientes, T. Calvo, Estela Cañón, Cristina Tejedo, P. Pennacchi, S. Hogan, P. Kölblinger, Héctor Tejero, Andrew Chen, Nuria Ibarz, Osvaldo Graña-Castro, Lola Martínez, J. Muñoz, P. Ortiz-Romero, J. Rodríguez-Peralto, G. Gómez-López, F. Al-Shahrour, R. Rabadán, M. Levesque, David Olmeda, M. Soengas (2020)
Midkine rewires the melanoma microenvironment toward a tolerogenic and immune-resistant state
Nature Medicine

Publisher: Springer Journals
Copyright: Copyright © The Author(s) 2023
eISSN: 2058-7716
DOI: 10.1038/s41420-023-01392-3
Publisher site: See Article on Publisher Site

Abstract

www.nature.com/cddiscovery ARTICLE OPEN Midkine inhibition enhances anti-PD-1 immunotherapy in sorafenib-treated hepatocellular carcinoma via preventing immunosuppressive MDSCs inﬁltration 1 2 1 1 1 3✉ Lijuan Ding , Nanya Wang , Qiang Wang , Xia Fan , Yuning Xin and Shudong Wang © The Author(s) 2023 Sorafenib, a multiple-target tyrosine kinase inhibitor, is the standard of care for patients with advanced hepatocellular carcinoma (HCC), but provides limited beneﬁts. Emerging evidences suggest that prolonged sorafenib treatment induces an immunosuppressive HCC microenvironment, but the underling mechanism is undetermined. In the present study, the potential function of midkine, a heparin-binding growth factor/cytokine, was evaluated in sorafenib-treated HCC tumors. Inﬁltrating immune cells of orthotopic HCC tumors were measured by ﬂow cytometry. Differentially expressed genes in sorafenib-treated HCC tumors were evaluated by transcriptome RNA sequencing. The potential function of midkine were evaluated by western blot, T cell suppression assay, immunohistochemistry (IHC) staining and tumor xenograft model. We found that sorafenib treatment increased intratumoral hypoxia and altered HCC microenvironment towards an immune-resistant state in orthotopic HCC tumors. Sorafenib treatment promoted midkine expression and secretion by HCC cells. Moreover, forced midkine expression stimulated immunosuppressive myeloid-derived suppressor cells (MDSCs) accumulation in HCC microenvironment, while knockdown of + + − midkine exhibited opposite effects. Furthermore, midkine overexpression promoted CD11b CD33 HLA-DR MDSCs expansion from human PBMCs, while midkine depletion suppressed this effect. PD-1 blockade showed no obvious inhibition on tumor growth of sorafenib-treated HCC tumors, but the inhibitory effect was greatly enhanced by midkine knockdown. Besides, midkine overexpression promoted multiple pathways activation and IL-10 production by MDSCs. Our data elucidated a novel role of midkine in the immunosuppressive microenvironment of sorafenib-treated HCC tumors. Mikdine might be a potential target for the combination of anti-PD-1 immunotherapy in HCC patients. Cell Death Discovery (2023) 9:92 ; https://doi.org/10.1038/s41420-023-01392-3 INTRODUCTION not superior to sorafenib when used in the ﬁrst-line setting [7]. Hepatocellular carcinoma (HCC) ranks the sixth most common Thus, there is urging needing to develop novel therapeutic cancer and second leading cause of cancer-related death world- strategies. wide [1]. In the last two decades, the incidence and mortality rates HCC is typically caused by chronic inﬂammation of a liver of HCC are increasing in most parts of the world, especially eastern diseases [8]. Therefore, there is a strong rationale for using of Asia and Africa [2]. Patients with HCC have a really poor prognosis, immunotherapy. Immune checkpoints include co-inhibitory recep- largely due to late diagnosis and high recurrence rate. The tors such as programmed cell death-1 (PD-1) and its ligand majority of HCCs are diagnosed at intermediate or advanced programmed death ligand 1 (PD-L1). PD-1 is expressed by an stages, with a 5-year survival rate as low as 16% [3]. Patients with extensive number of immune cells such as activated T cells and HCC are refractory to almost all conventional chemotherapeutic natural killer (NK) cells, while PD-L1 is expressed on tumor cells, drugs. Systemic therapy for advanced HCC consists only of stromal cells and myeloid cells [9]. In general, antiangiogenic antiangiogenic tyrosine kinase inhibitors until recently. Sorafenib, therapy with sorafenib, lenvatinib or VEGF antibodies remains the a multiple-target tyrosine kinase inhibitor with antiangiogenic and fundamental treatment for advanced HCC, whereas immunother- antiproliferation effects, is approved by the Food and Drug apy with checkpoint inhibitors becomes increasing important [10]. Administration (FDA) as the ﬁrst-line chemotherapeutic drug for Recently, immunotherapy with checkpoint inhibitors has been advanced HCC. In two large phase III trials, sorafenib is proved to tested in HCC patients and shows strong anti-tumor effect in a be effective in advanced HCC patients [4, 5]. However, less than subset of patients [11, 12]. The combination of PD-1/PD-L1 and 30% of HCC patients can beneﬁt from sorafenib treatment, and vascular endothelial growth factor (VEGF) antibodies results in a the median survival is extended only about 2.5 months [5]. better survival than sorafenib, which makes it as a new ﬁrst line Besides, these patients usually become resistant to sorafenib therapy [13]. Despite these improvements, only one in four within 6 months [6]. Another chemotherapeutic drug lenvatinib is patients respond to these immunotherapies, and the majority of 1 2 Department of Radiation Oncology, the First Hospital of Jilin University, Changchun 130021, China. Cancer Center, the First Hospital of Jilin University, Changchun 130021, China. Department of Cardiology, the First Hospital of Jilin University, Changchun 130021, China. email: wangshudongjlu@126.com Received: 24 August 2022 Revised: 28 February 2023 Accepted: 1 March 2023 Ofﬁcial journal of CDDpress 1234567890();,: L. Ding et al. Fig. 1 Sorafenib treatment increases intratumoral hypoxia and alters HCC microenvironment towards an immune-resistant state in mouse models. A, B orthotopic Hepa 1-6 and Hepa 1c1c7 tumors were treated with sorafenib or vehicle control for three weeks, then collected protein lysates for western blot (A) or tissue samples for Pimonidazole staining (B). C, D inﬁltrating immune cells in orthotopic Hepa + + + + 1-6 and Hepa 1c1c7 tumors treated with sorafenib or vehicle control were evaluated by ﬂow cytometry. CD3 CD4 CD25 FoxP3 Treg, + − − + + + int + + + high - + - + CD11b Gr-1 Ly6C F4/80 TAM, CD11b Gr-1 Ly6C Ly6G PMN-MDSC, CD11b Gr-1 Ly6C Ly6G M-MDSC and CD3 CD4 CD8 Cyto T cells were evaluated. E relative expression of PD-L1, TGFB1, IL10, and IL13 in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors was evaluated by qRT-PCR. F protein expression of PD-L1 in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors was evaluated by western blot. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. HCC patients do not respond due to unknown reasons. One Moreover, a number of immune cells are inﬂuenced by midkine, aspect is that prolonged antiangiogenic therapy increases including macrophage [18], polymorphonuclear (PMN) [19], B cells intratumoral hypoxia, which facilitates tumor recurrence and [20] and T cells [21]. In addition, midkine plays a vital role in fosters an immunosuppressive microenvironment [14, 15]. Since maintaining immunosuppressive tumor microenvironment of the efﬁciency of immune checkpoint inhibitors is greatly melanoma and gallbladder cancer [22, 23]. Midkine is upregulated inﬂuenced by immunosuppressive tumor microenvironment, it is in HCC patients and associated with poor prognosis [24]. However, necessary to elucidate the underlying mechanism. the potential role of midkine in HCC microenvironment is Midkine (MDK) is a heparin-binding growth factor/cytokine that undetermined. Myeloid-derived suppressor cells (MDSCs) are exhibits multiple functions and implicates in various physiological immature marrow-derived cell populations with potent immuno- process [16]. Midkine is rarely expressed by normal tissues, but suppressive activity. They are mainly consisted of granulocytic/ signiﬁcantly upregulated in inﬂammatory diseases and human polymorphonuclear MDSCs (PMN-MDSCs) and monocytic MDSCs malignant tumors [17]. Upregulation of midkine can promote (M-MDSCs), which are similar to neutrophils and monocytes, growth, survival, metastasis, and angiogenesis of cancer cells [16]. respectively. In the present study, orthotopic HCC models were Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 2 Sorafenib treatment promotes midkine expression and secretion by HCC cells. A differentially expressed genes in orthotopic Hepa 1-6 tumors treated with sorafenib (Sor) or vehicle control (Veh) were depicted in volcano map. B, C midkine expression in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors treated with sorafenib or vehicle control was evaluated by qRT-PCR (B) and western blot (C). D secreted mikdine in serum samples of orthotopic Hepa 1-6 and Hepa 1c1c7 tumors treated with sorafenib or vehicle control was measured by ELISA assay. E, F midkine expression in orthotopic HUH-7 and SNU-449 tumors treated with sorafenib or vehicle control was evaluated by qRT-PCR (E) and western blot (F). G secreted mikdine in serum samples of orthotopic HUH-7 and SNU-449 tumors treated with sorafenib or vehicle control was measured by ELISA assay. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. constructed to evaluate the potential role of midkine in sorafenib- expression and reduced vessel density in tumor samples induced immunosuppressive microenvironment. We found that (Supplementary Fig. 1A–C). Moreover, HIF1-α dependent genes midkine overexpression stimulated MDSCs inﬁltrating into HCC such as VEGF, GLUT-1, CA-9, CXCR4, and MDR1 were upregulated tumors, while knockdown of midkine exhibited opposite effects. In in sorafenib-treated patients (Supplementary Fig. 1D). These addition, midkine inhibition enhanced the inhibitory effects of results indicated that sorafenib increased intratumoral hypoxia anti-PD-1 immunotherapy in sorafenib-treated HCC tumors. Our in orthotopic HCC models and patient samples. To evaluate the results elucidated a novel role of midkine in HCC microenviron- immune microenvironment of orthotopic HCC tumors, inﬁltrating ment, and midkine might be a potential target for HCC treatment. immune cells were evaluated by ﬂow cytometry. The gating strategy and representative plots of inﬁltrating immune cell populations were depicted in Supplementary Fig. 2. The + + + + RESULTS percentages of regulator T cells (Treg, CD3 CD4 CD25 FoxP3 ), + − - + Sorafenib treatment increases intratumoral hypoxia and alters tumor-associated macrophages (TAM, CD11b Gr-1 Ly6C F4/80 ), HCC microenvironment towards an immune-resistant state in polymorphonuclear myeloid-derived suppressor cells (PMN-MDSC, + + int + mouse models CD11b Gr-1 Ly6C Ly6G ) and monocytic myeloid-derived sup- + + high - To determine whether sorafenib-induced intratumoral hypoxia pressor cells (M-MDSC, CD11b Gr-1 Ly6C Ly6G ) were may foster an immunosuppressive microenvironment, orthotopic obviously increased, while cytotoxic T cells (Cyto T, + − + HCC models were constructed using murine HCC cell lines Hepa CD3 CD4 CD8 ) were apparently decreased in sorafenib- 1-6 and Hepa 1c1c7 in C57BL/6 mice. In our study, sorafenib treated HCC tumors (Fig. 1C, D). PD-L1, TGFB1, IL10, and IL13 are treatment signiﬁcantly increased the protein expression of HIF1-α primary effectors for immunosuppressive tumor microenviron- in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors (Fig. 1A). ment [25, 26]. These genes were signiﬁcantly upregulated in Pimonidazole is a hypoxia-speciﬁc marker. Sorafenib treatment sorafenib-treated HCC tumors (Fig. 1E). The protein expression of evidently increased the number of pimonidazole positive cells in PD-L1 was also increased by sorafenib (Fig. 1F). Our results orthotopic Hepa 1-6 and Hepa 1c1c7 tumors (Fig. 1B). In HCC indicated that sorafenib treatment altered HCC microenvironment patients, sorafenib treatment signiﬁcantly increased HIF1-α toward an immune-resistant state in mouse models. Cell Death Discovery (2023) 9:92 L. Ding et al. Sorafenib treatment promotes midkine expression and Compared with monocytes or neutrophils, mouse MDSCs are secretion by HCC cells characterized by high level of arginase 1 (Arg1) [28]. Of note, Arg1 is To search for potential genes involved in sorafenib-induced vital for the immunosuppression function of MDSCs. In our study, immunosuppressive microenvironment, orthotopic Hepa 1-6 midkine overexpression dramatically increased the number of Arg1 + + tumors treated with sorafenib or vehicle control were subjected cells in CD11b Gr-1 subset of orthotopic HCC tumors with/without to transcriptome RNA sequencing. In our study, 71 genes were sorafenib treatment (Fig. 4A, B). Moreover, the percentages of Arg1 signiﬁcantly upregulated while 131 genes were downregulated in cells in PMN-MDSCs and M-MDSCs subsets were also increased by sorafenib-treated tumors compared with vehicle-treated tumors ectopic midkine expression (Fig. 4C). Arg1, Pdl1, Tgfb and Nos2 are (Fig. 2A, Supplementary Table 1). Among them, midkine ranked the primary effectors for the immunosuppression function of MDSCs. top ten upregulated genes (Fig. 2A, Supplementary Table 1). Forced midkine expression increased the expression of Arg1, Tgfb Midkine plays an important role in maintaining immunosuppres- and Nos2, and decreased the expression of Pdl1 in sorted PMN- sive microenvironment of melanoma and gallbladder cancer MDSCs and M-MDSCs (Fig. 4D). The immunosuppression activity of [22, 23]. Thus, we speculated that the upregulation of midkine MDSCs in midkine-overexpressing HCC tumors was further eval- + + might contribute to the immune-resistant state caused by uated by T-cell suppression assay. CD3 CD8 cytotoxic T cells were sorafenib treatment in HCC tumors. In our study, midkine stimulated with CD3e and CD28 antibodies and co-cultured with expression in pan-cancer was evaluated by TIMER database. We MDSCs sorted from midkine-overexpressing HCC tumors or EV- found that midkine was signiﬁcantly upregulated in a variety of transduced tumors for three days. Cytotoxic T cells co-cultured with cancers, including HCC (Supplementary Fig. 3A). Data from TCGA MDSCs sorted from midkine-overexpressing HCC tumors showed and GEO database (GSE39791 and GSE112790) also suggested that reduced Ki67 and Granzyme B expression compared with T cells co- midkine was overexpressed in HCC tumor samples (Supplementary cultured with MDSCs derived from EV-transduced tumors, indicating Fig. 3B and C). Besides, high midkine expression was positively that the proliferation and activation of T cells were repressed (Fig. correlated with advanced tumor stages and poor overall-survival of 4E, F). Moreover, MDSCs derived from sorafenib-treated tumors HCC patients (Supplementary Fig. 3D and E). In our study, midkine showed enhanced inhibition on T cell proliferation and activation, expression was increased by sorafenib in orthotopic HCC tumors especially those from midkine-overexpressing tumors (Fig. 4E, F). (Figs. 2B and 2C). Midkine is a secreted protein. Sorafenib These results indicated that MDSCs derived from midkine- treatment apparently increased the level of secreted midkine in overexpressing and/or sorafenib-treated HCC tumors showed more serum samples of mice bearing HCC tumors (Fig. 2D). This was also potent immunosuppressive activity. The potential function of validated in human HCC cell lines HUH-7 and SNU-449. Sorafenib midkine was also evaluated in MDSCs expanded from human treatment evidently augmented midkine level in tumor tissues and PBMCs. Midkine was ectopic overexpressed in HUH-7 and SNU-449 serum samples of Balb/c nude mice bearing orthotopic HUH-7 or cells (Supplementary Fig. 6A). The secreted midkine in conditional SNU-449 tumors (Fig. 2E, F). The promoter region of midkine has a medium of HUH-7 and SNU-449 was signiﬁcantly increased by hypoxia responsive element, thus hypoxia may induce midkine midkine overexpression (Supplementary Fig. 6B). Human MDSCs are + + − expression via binding with HIF-1α [27]. This was tested in HCC commonly marked as CD11b , CD33 ,and HLA-DR . In our study, + + - cells. Secreted midkine was signiﬁcantly elevated in culture CD11b CD33 HLA-DR MDSCs were expanded from human PBMCs medium of both murine and human HCC cells under hypoxia via culturing with conditional medium from HUH-7 or SNU-449 cells condition (1% O ) compared with normoxia condition (20% O ) for 5 days. In our study, conditional medium from midkine- 2 2 (Fig. 2G). The above results indicated that sorafenib treatment overexpressing HUH-7 and SNU-449 cells evidently expanded more + + − promoted midkine expression and secretion by HCC cells. CD11b CD33 HLA-DR MDSCs compared with conditional medium from EV-transduced cells (Supplementary Fig. 6C). The immunosup- Forced midkine expression stimulates immunosuppressive pression activity of those MDSCs were evaluated by T-cell MDSCs accumulation in HCC tumor microenvironment suppression assay. T cells co-cultured with MDSCs induced by To elucidate the potential effects of midkine in HCC microenviron- conditional medium from midkine-overexpressing HUH-7 and SNU- ment, the correlation between midkine and tumor immune 449 cells showed less Ki67 staining and IFN-γ production, indicating inﬁltration was evaluated. The 22 tumor-inﬁltrating immune cells these T cells were less proliferative and cytotoxic (Supplementary in HCC tissues were estimated by CIBERSORT algorithm. HCC Fig. 6D–F). Altogether, our data indicated that forced midkine patients with high midkine expression showed enriched scores for expression stimulated immunosuppressive MDSCs accumulation in Treg (Supplementary Fig. 4). Besides, high midkine expression was HCC tumor microenvironment. positively associated with immune checkpoint molecules such as CTLA4, HAVCR2, LAG3, PDCD1, and TIGIT (Supplementary Fig. 5A). Midkine inhibition prevents immunosuppressive MDSCs TISIDB was also used to investigate the correlation of midkine with inﬁltrating into HCC tumors inﬁltrating immune cells. High midkine expression was positively The inﬂuence of midkine on MDSCs inﬁltration was further correlated with MDSCs abundance, but not monocytes or neutro- evaluated by loss-of-function assays. Short hairpin RNAs targeting phils (Supplementary Fig. 5B). The data from TIMER demonstrated mouse midkine (Sh-MDK-1 and Sh-MDK-2) were designed and that immune makers of MDSCs (CD33, ITGAM and FUT4) were introduced into Hepa 1-6 cells. Midkine expression was depleted signiﬁcantly associated with midkine expression (Supplementary Fig. by these two shRNAs, despite exposing to sorafenib treatment 5C). These results suggested that midkine might involve in tumor (Fig. 5A). The percentage of inﬁltrating MDSCs was measured. inﬁltrating of MDSCs. To validate this, midkine was overexpressed by Midkine knockdown apparently decreased the percentage of transducing with midikine expression lentivirial particles (Fig. 3A). inﬁltrating PMN-MDSCs and M-MDSCs in orthotopic Hepa 1-6 + + CD11b and Gr-1 are myeloid differentiation markers for mouse tumors with/without sorafenib treatment (Fig. 5B, C). Depletion of MDSCs. Forced midkine expression apparently increased the midkine also reduced the percentage of Arg1 cells in inﬁltrating + + + + percentages of inﬁltrating CD11b Gr-1 cells in orthotopic Hepa CD11b Gr-1 subset (Fig. 5D, E) and MDSCs subset (Fig. 5F). The 1-6 and Hepa 1c1c7 tumors, and this effect was enforced by inﬂuence of midkine inhibition on immunosuppression activity of sorafenib treatment (Fig. 3B, C). There are two main subsets of inﬁltrating MDSCs was evaluated. Cytotoxic T cells co-cultured + + int + mouse MDSCs: PMN-MDSCs (CD11b Gr-1 Ly6C Ly6G )and with MDSCs from midkine-depleted Hepa 1-6 tumors showed + + high − M-MDSCs (CD11b Gr-1 Ly6C Ly6G ). Midkine overexpression increased Ki67 and Granzyme B staining compared with cells co- signiﬁcantly increased the percentages of both PMN-MDSCs and cultured with MDSCs from non-targeting control (sh-NC) trans- M-MDSCs in orthotopic Hepa 1-6 and Hepa 1c1c7 tumors with/ duced tumors (Fig. 5G, E). This was also evaluated in MDSCs without sorafenib treatment (Fig. 3D, E). expanded from human PBMCs. Short hairpin RNAs targeting Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 3 Hepatocellular midkine overexpression stimulates MDSCs accumulation in tumor microenvironment. Midkine-overexpression or EV-transduced Hepa 1-6 and Hepa 1c1c7 cells were orthotopically implanted into the liver of C57BL/6 mice, then treated with sorafenib (Sor) + + or vehicle control (Veh) for three weeks. A protein expression of midkine was validated by western blot. B, C inﬁltrating CD11b Gr-1 cells int + high - + + were evaluated by ﬂow cytometry. D, E Ly6C Ly6G PMN-MDSCs and Ly6C Ly6G M-MDSCs in inﬁltrating CD11b Gr-1 subset were evaluated by ﬂow cytometry. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. human midkine (Sh-MDK-3 and Sh-MDK-4) were designed and control group (Fig. 6A–C). Compared with the IgG isotype control, introduced into HUH-6 and HepG2 cells which showed high tumor growth, volume and weight of subcutaneous Hepa 1-6 endogenous midkine expression. Midkine was successfully tumors were signiﬁcant reduced by anti-PD-1 antibody, however knocked down by these two shRNAs (Supplementary Fig. 7A). this effect was largely abolished by midkine overexpression (Fig. Knockdown of midkine signiﬁcantly reduced the level of secreted 6A–C). The intratumoral inﬁltrating immune cells were further midkine in conditional medium from HUH-6 and HepG2 cells evaluated. Forced midkine expression signiﬁcantly increased the (Supplementary Fig. 7B). Furthermore, reduced midkine secretion percentages of inﬁltrating MDSCs and Tregs, and anti-PD-1 + + − suppressed CD11b CD33 HLA-DR MDSCs expansion from treatment augmented this effect to some extent (Fig. 6D). In human PBMCs (Supplementary Fig. 7C). In T cell suppression addition, anti-PD-1 treatment dramatically increased the percen- assay, T cells co-cultured with MDSCs induced by conditional tage of inﬁltrating CD8 cytotoxic T cells, however this effect was medium from midkine depleted HUH-6 and HepG2 cells exhibited almost abrogated by midkine overexpression (Fig. 6D). Next, the more Ki67 staining and IFN-γ production, indicating these T cells inﬂuence of midkine inhibition on immunotherapy with anti-PD-1 were more proliferative and cytotoxic (Supplementary Fig. 7D–F). antibody in sorafenib-treated HCC tumors was evaluated. Knock- Above all, our results suggested that midkine inhibition prevented down of midkine showed no evident inﬂuence on tumor growth immunosuppressive MDSCs inﬁltrating into HCC tumors. of subcutaneous Hepa 1-6 tumors (Fig. 6E–G). However, midkine inhibition apparently increased the cytotoxicity of sorafenib, with Midkine inhibition enhances anti-PD-1 immunotherapy in reduced tumor growth, volume and weight (Fig. 6E–G). Besides, sorafenib-treated HCC tumors anti-PD-1 antibody showed no obvious suppression on tumor To assess whether mikine could impinge on immunotherapy with growth of sorafenib-treated tumors, but this was greatly checkpoint inhibitors, we tested anti-PD-1 antibody. Midkine strengthened by midkine knockdown (Fig. 6E–G). Furthermore, overexpression showed no signiﬁcant inﬂuence on growth of depletion of midkine signiﬁcantly reduced the percentage of subcutaneous Hepa 1-6 tumors, as observed in the IgG isotype inﬁltrating MDSCs and Tregs, and increased the percentage of Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 4 Hepatocellular midkine overexpression induces T cell-suppressive MDSC accumulation in tumor microenvironment. Midkine- overexpression or EV-transduced Hepa 1-6 and Hepa 1c1c7 cells were orthotopically implanted into the liver of C57BL/6 mice, then treated + + + with sorafenib (Sor) or vehicle control (Veh) for three weeks. A-C, Arg1 cells in inﬁltrating CD11b Gr-1 subset (A–B) and MDSCs subset (C)of orthotopic Hepa 1-6 and Hepa 1c1c7 tumors were evaluated by ﬂow cytometry. D Relative expression of Arg1, Pdl1, Tgfb and Nos2 in inﬁltrating PMN-MDSCs and M-MDSCs subsets of orthotopic Hepa 1-6 and Hepa 1c1c7 tumors was evaluated by qRT-PCR. E, F Inﬁltrating + + PMN-MDSCs and M-MDSCs of orthotopic Hepa 1-6 and Hepa 1c1c7 tumors were used for T-cell suppression assay. Ki67 (E) and Granzyme B (F) cytotoxic T cells were evaluated by ﬂow cytometry. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. inﬁltrating CD8 cytotoxic T cells in sorafenib-treated Hepa 1-6 from midkine-overexpressing Hepa 1-6 tumors showed increased tumors (Fig. 6H). Taken together, our results indicated that phosphorylation of p65, Akt, ERK, and STAT3 compared with midkine inhibition enhanced anti-PD-1 immunotherapy in MDSCs from EV-transduced Hepa 1-6 tumors, indicating the sorafenib-treated HCC tumors. activation of NF-кB, Akt, ERK, and STAT3 signaling by midkine- driven secretome (Fig. 7A). This was also validated in + + - Midkine overexpression drives multiple pathways activation CD11b CD33 HLA-DR MDSCs induced from human PBMCs. and IL-10 production by MDSCs Conditional medium from midkine-overexpression HUH-7 and Previous studies demonstrate that midkine promotes the activa- SNU-449 cells evidently promoted the phosphorylation of p65, + + − tion of various signaling pathways, including NF-кB, PI3K/Akt, ERK, Akt, ERK and STAT3 in CD11b CD33 HLA-DR MDSCs compared and Notch2/Jak2/STAT3 [16, 22, 29]. In our study, MDSCs sorted with conditional medium form EV-transduced cells Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 5 Midkine inhibition prevents immunosuppressive MDSCs inﬁltrating into HCC tumors. Hepa 1-6 cells transduced with Sh-MDK-1, Sh- MDK-2 or Sh-NC lentivirus were orthotopically implanted into the liver of C57BL/6 mice, then treated with sorafenib (Sor) or vehicle control (Veh) for three weeks. A protein expression of midkine in midkine-depleted orthotopic Hepa 1-6 tumors was validated by western blot. int + high - + + B, C Ly6C Ly6G PMN-MDSCs and Ly6C Ly6G M-MDSCs in intratumoral CD11b Gr-1 subset of midkine-depleted orthotopic Hepa 1-6 + + + tumors were evaluated by ﬂow cytometry. D–F Arg1 cells in inﬁltrating CD11b Gr-1 subset (D–E) and MDSCs subset (F) of midkine- depleted orthotopic Hepa 1-6 tumors were evaluated by ﬂow cytometry. G, H inﬁltrating PMN-MDSCs and M-MDSCs of midkine-depleted + + orthotopic Hepa 1-6 tumors were used for T-cell suppression assay. Ki67 (G) and Granzyme B (H) cytotoxic T cells were evaluated by ﬂow cytometry. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. (Supplementary Fig. 8A). The immunosuppression activity of were used in T-cell suppression assay. MDSCs derived from MDSCs is acted through various mechanisms. Among them, IL- midkine-overexpressing Hepa 1-6 tumors signiﬁcantly inhibited 10 secreted by MDSCs promotes immunosuppression via target- the proliferation and activation of cytotoxic T cells compared with ing a variety of immune cells [30]. In our study, IL-10 expression MDSCs derived from EV-transduced tumors, but the addition of was signiﬁcantly upregulated in sorafenib-treated HCC tumors neutralizing antibodies against IL-10 largely abrogated these + + - (Fig. 1E). Moreover, MDSCs sorted from midkine-overexpressing effects (Fig. 7D, E). Furthermore, CD11b CD33 HLA-DR MDSCs Hepa 1-6 tumors showed elevated IL-10 expression and produc- induced by conditional medium from midkine-overexpressing tion compared with MDSCs derived from EV-transduced tumors HUH-7 and SNU-449 cells showed increased IL-10 expression and (Fig. 7B, C). To clarify the potential role of IL-10 in immunosup- secretion compared with MDSCs induced by conditional medium pression activity of MDSCs, neutralizing antibodies against IL-10 from EV-transduced cells (Supplementary Fig. 8B and C). Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 6 Midkine inhibition enhances anti-PD-1 immunotherapy in Sorafenib-treated HCC tumors. A–D Hepa 1-6 cells transduced with midkine expression lentivirus or EV control were subcutaneously injected into C57BL/6 mice, then treated with antibody against murine PD-1 or IgG2b isotype control as indicated. Tumor growth curves (A), representative images (B), tumor weight (C), and inﬁltrating immune cells (D) were shown. E–H Hepa 1-6 cells transduced with Sh-MDK-1 or Sh-NC lentivirus were subcutaneously injected into C57BL/6 mice, then treated with sorafenib, vehicle, antibody against murine PD-1, or IgG2b isotype control as indicated. Tumor growth curves (E), representative images (F), tumor weight (G), and inﬁltrating immune cells (H) were shown. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. Moreover, midkine overexpression dramatically potentiated the density, augmented intratumoral hypoxia and activated HIF-1α + + - inhibitory functions of CD11b CD33 HLA-DR MDSCs on prolif- signaling in orthotopic HCC models. Emerging evidences indicate eration and activation of stimulated cytotoxic T cells, but this was that midkine is vital for hypoxia-driven tumor angiogenesis abolished by supplement of neutralizing antibodies against IL-10 [39, 40]. It is worth noting that the promoter region of midkine (Supplementary Fig. 8D and E). Collectively, our results proved that has a hypoxia responsive element which can binds with HIF-1α midkine overexpression promoted multiple pathways activation [27]. Thus, it was not surprising that midkine was signiﬁcantly and IL-10 production by MDSCs. upregulated by sorafenib-induced hypoxia in orthotopic HCC tumors. There are increasing evidences suggesting that persistent DISCUSSION sorafenib treatment gives rise to an immunosuppressive tumor Accumulated studies prove that midkine is involved in the microenvironment. For example, sorafenib treatment increases malignant progression of cancers, including HCC [16, 31]. Midkine intratumoral inﬁltrating of F4/80 tumor-associated macrophages, + + + + + is signiﬁcantly upregulated in tumor tissues and serum samples of CD4 CD25 FoxP3 regulatory T cells, CCL2 /CCL17 tumor- + + HCC patients, and associated with poor overall survival [24, 32, 33]. associated neutrophils and CD11b Gr-1 myeloid cells in human Midkine is reported to protect HCC cells against TRAIL-mediated and murine HCC models [37, 41]. Besides, immunosuppressive M2 apoptosis and cadmium-induced cellular damage [34, 35]. In macrophages accumulate more while activated Natural Killer cells addition, midkine overexpression facilitates metastasis of HCC fail to proliferate and produce effector molecules in sorafenib- cells via increasing anoikis resistance [36]. Upregulation of midkine resistant tumors [42, 43]. Similarly, our data demonstrated that the in HCC samples was also demonstrated by us via analyzing data percentages of Treg, TAM, and MDSC were increased, while from public database. What’s more, we found that sorafenib cytotoxic T cells were decreased in sorafenib-treated tumors, treatment promoted midkine expression and secretion by HCC indicating an immune-resistant HCC tumor microenvironment. To cells. Sorafenib is a multikinase inhibitor with antiangiogenic elucidate the potential molecular mechanism underlying this, we property. Previous studies demonstrate that sustained sorafenib found that midkine was signiﬁcantly upregulated by sorafenib treatment reduces microvascular density and increases intratu- treatment in HCC tumors. More importantly, forced midkine moral hypoxia of HCC tumors [37, 38]. Corresponding with this, expression stimulated immunosuppressive MDSCs accumulation our data proved that sorafenib treatment decreased microvessel in HCC microenvironment, while knockdown of midkine Cell Death Discovery (2023) 9:92 L. Ding et al. Fig. 7 Midkine overexpression promotes multiple pathways activation and IL-10 production by MDSCs. A–C PMN-MDSCs and M-MDSCs were sorted from midkine-overexpressing or EV-transduced orthotopical Hepa 1-6 tumors with/without sorafenib treatment, then collected cell lysates for western blot (A). Relative IL-10 expression (B) and production (C) were evaluated by qRT-PCR and ELISA assay. D, E PMN-MDSCs, and M-MDSCs sorted from midkine-overexpressing or EV-transduced orthotopic Hepa 1-6 tumors with/without sorafenib treatment were + + used for T-cell suppression assay. Ki67 (D) and Granzyme B (E) cytotoxic T cells were evaluated by ﬂow cytometry. All assays were done with at least three repeats. Data were shown as mean ± s.d., *P < 0.05. suppressed immunosuppressive MDSCs inﬁltrating into HCC ﬁrst line therapy, as atezolizumab (anti-PD-L1) and bevacizumab tumors. These results suggested midkine might partially account (anti-VEGF) improves the median overall survival for more than for sorafenib-induced immunosuppressive microenvironment. 17 months [13]. This is consistent with the idea that checkpoint Indeed, midkine is well-characterized for its function in the blockade may be more efﬁcacious as combination therapy. immune system, including promoting immune cell chemotaxis Nevertheless, only one in four patients respond to the combina- and sculpting myeloid cell phenotype. For instance, midkine tion therapy, and the rest majority do not respond for unknown promotes neutrophils trafﬁcking and adhesion during acute reasons. Sorafenib shows antiangiogenic activity by targeting a inﬂammation via inducing high afﬁnity conformation of β2 variety of tyrosine kinase receptors, including VEGF receptor 2 and integrins [19]. During ischemic renal injury, midkine deﬁcient 3. In our study, we found that prolonged sorafenib exposure mice show reduced number of inﬁltrating neutrophils and altered HCC microenvironment towards an immune-resistant macrophages into the tubulointerstitium and impaired induction state, and anti-PD-1 antibody showed no obvious inﬂuence on of macrophage inﬂammatory protein-2 and macrophage chemo- tumor growth of sorafenib-treating tumors. Similarly, the response tactic protein-1 [44]. Midkine is also critical for fostering the rate of anti-PD-1 drug nivolumab is lower in sorafenib experienced immunosuppressive microenvironment. In melanoma, midkine patient than sorafenib-naïve patients (16–19% vs. 23%) [11]. In our overexpression gives rise to an immunosuppressive tumor study, midkine was signiﬁcantly induced by sorafenib in HCC microenvironment via promoting intratumoral recruitment of tumors. Knockdown of midkine dramatically increased the myeloid cells and tumor-associated macrophages [22]. In ErbB inhibitory effects of anti-PD-1 antibody on sorafenib-treating pathway-mutated gallbladder cancer, upregulation of secreted tumors, and this was partially due to the destruction of sorafenib- midkine facilitates differentiation of immunosuppressive macro- induced immunosuppressive microenvironment. In addition, phages through biding with its receptor LRP1 [23]. Correspond- forced midkine expression apparently attenuated the suppression ingly, we found that midkine-modulated secretome promoted of anti-PD-1 antibody on HCC tumor growth. These results + + - expansion of CD11b CD33 HLA-DR MDSCs from human PBMCs. suggested that the immunocompromised microenvironment In recent years, immune checkpoint inhibitors such as anti-PD-1 caused by VEGF inhibitors such as sorafenib might account for or anti-PD-L1 become an alternative therapeutic choice for HCC the low response rate of the combination therapy of anti-PD-1/ treatment, however the majority of patients do not respond to anti-PD-L1 antibodies and VEGF inhibitors. Meanwhile, breakdown them when use as single agent [11, 45]. In contrast, the of immune-tolerant state such as midkine inhibition in our study combination of anti-PD-1/anti-PD-L1 antibodies with vascular might increase the efﬁciency of the combination therapy. This is endothelial growth factor (VEGF) inhibitors has become a new also demonstrated by many other studies. For instance, inhibition Cell Death Discovery (2023) 9:92 L. Ding et al. of CXCR4 by AMD3100 signiﬁcantly increased the efﬁciency of 40 ng/mL IL-6 (Peprotech, Germany). Cells were cultured in a humidiﬁed atmosphere at 37 ˚Cwith 5% CO . Sorafenib (Selleck Chemicals, USA) was anti-PD-1 immunotherapy in sorfaenib-treated HCC tumors [37]. 2 dissolved in dimethylsulfoxide (DMSO), thus DMSO was used as vehicle MDSCs are essential components of the suppressive HCC tumor control. microenvironment. The frequency of inﬁltrating MDSCs is tightly correlated with the prognosis of HCC patients and efﬁciency of immune checkpoint inhibitors and tyrosine kinase inhibitors. For Plasmid constructs Human or mouse midkine expression lentivirus vector was constructed by example, Ly6G MDSCs are evidently increased in sorafenib- cloning the coding sequence of human or mouse midkine into the pCDH treated orthotopic liver tumors, and targeting Ly6G MDSCs by lentivirus vector (System Biosciences #CD510B). The empty pCDH lentivirus anti-Ly6G antibody apparently improves the inhibitory effects of vector was used as empty vector control (EV). To deplete midkine sorafenib via decreasing inﬁltrating MDSCs and enhancing expression, short hairpin RNAs targeting mouse midkine (Sh-MDK-1 and intratumoral cytotoxic T cells [46]. In HCC, CCRK promotes Sh-MDK-2) or human midkine (Sh-MDK-3 and Sh-MDK-4) were cloned into immunosuppressive MDSCs expansion and accumulation via NF- the pLKO.1 plasmid. The pLKO.1 plasmid inserted with a non-targeting кB/IL-6 signaling [47]. Depletion of CCRK enhances PD-L1 blockade sequences was used as non-targeting control (Sh-NC). The shRNA efﬁciency via reducing MDSC accumulation and increasing sequences were listed in Supplementary Table 2. intratumorous cytotoxic T cells. In our study, midkine over- expression signiﬁcantly increased the intratumoral accumulation Quantitative real-time polymerase chain reaction (qRT-PCR) of MDSCs, thus diminished the efﬁciency of PD-1 blockade in Total RNAs from tissue samples or cell lines were extracted by the TRIzol subcutaneous Hepa 1-6 tumors. On the contrary, midkine reagent (Takara, Japan). Complementary DNA strands were reverse- inhibition reduced the percentage of inﬁltrating MDSCs in transcribed by the ﬁrst-strand cDNA synthesis kit (Takara, Japan). SYBR sorafenib-treated Hepa 1-6 tumors, thus enhanced the inhibitory Green PCR Master Mix (Applied Biosystems, USA) were used for qRT-PCR analysis on the ViiATM7/QuantStudio 7 Flex Real Time PCR System (Applied effects of anti-PD-1 antibody. MDSCs are generated and activated Biosystems, USA). GAPDH was used as internal control. Relative gene through a complicated process involving multiple signaling −ΔΔCq expression was calculated by the 2 method. The primers used for pathways, such as STAT3, NF-кB and ERK [48]. In our study, we qRT-PCR analysis were listed in Supplementary Table 3. found that MDSCs from midkine-overexpression HCC tumors showed increasing activation of NF-кB, Akt, ERK, and Western blot STAT3 signaling. Moreover, IL-10 production by MDSCs was Protein lysates from tissues samples or culture cells were prepared using evidently increased by midkine overexpression. IL-10 is one of RIPA lysis buffer (Beyotime, China) supplemented with protease inhibitors the several MDSCs-secreted immunosuppressive effectors [49, 50]. (Beyotime, China). Bradford reagent (Sigma, USA) was used to measure In our study, neutralizing antibodies against IL-10 largely protein concentration. A total of 10–40 μg protein lysates were separated abrogated the inhibitory effects of MDSCs on proliferation and by 8–12% SDS-PAGE gels and transferred to polyvinylidene ﬂuoride activation of cytotoxic T cells. These results partially explained the membranes (GE Healthcare, UK). The membranes were blocked by 5% potential inﬂuence of midkine on generation and activation of non-fat milk, then incubated with speciﬁc ﬁrst antibodies at 4 ˚C overnight MDSCs in HCC tumors. and corresponding second antibodies at room temperature for 1 h. The In summary, we found that sorafenib treatment increased western bands were detected by chemiluminescence imaging (Biorad, USA) using ECL kit (GE Healthcare, UK). Image J was used to quantify the protein intratumoral hypoxia and altered HCC microenvironment towards expression. Antibodies used in our study were: HIF-1α Rabbit mAb (CST an immune-resistant state in mouse models. Sorafenib treatment #36169, 1: 1000), GAPDH Rabbit mAb (CST #5174, 1: 5000), PD-L1 Rabbit facilitated midkine expression and secretion by HCC cells. Moreover, mAb (CST #13684, 1: 1000), mouse midkine antibody (Abcam #ab281534, 1: midkine overexpression stimulated immunosuppressive MDSC accu- 1000), human midkine antibody (Abcam #ab52637, 1: 1000), Erk1/2 Rabbit mulation in HCC tumor microenvironment, while midkine inhibition mAb (CST #4695, 1: 1000); Phospho-Erk1/2 (Thr202/Tyr204) Rabbit mAb exhibited opposite effects. In addition, midkine inhibition enhanced (CST #4370, 1: 1000); STAT3 Mouse mAb (CST #9139, 1: 1000), Phospho- anti-PD-1 immunotherapy in sorafenib-treated HCC tumors. Midkine STAT3 (Tyr705) Rabbit mAb(CST #73533, 1: 1000), Akt Antibody (CST #9272, overexpression promoted multiple pathways activation and IL-10 1: 1000), Phospho-Akt (Ser473) Rabbit mAb (CST #4060, 1: 1000). secretion by MDSCs. Our data provided a novel role of midkine in sorafenib-induced immunosuppressive microenvironment. Midkine Immunohistochemistry (IHC) staining might be a potential target for the combination of anti-PD-1 Tissue samples were ﬁxed by formalin and embedded by parafﬁn. Tissue immunotherapy in HCC patients. sections (5 μm) were deparafﬁnized, rehydrated, and rinsed in distilled water. Antigen retrieval was conducted using citrate buffer (pH 6.0). The endogenous peroxidase activity was quenched by 3% hydrogen peroxide. For pimonidazole staining, the sections were incubated with monoclonal MATERIALS AND METHODS mouse antibodies against pimonidazole (1:50, Hypoxyprobe Inc., USA) at Patient samples 37 ˚C for 30 min. For HIF-1α staining, the sections were incubated with HIF- The collection and use of human samples were approved by the Ethics 1α Rabbit mAb (CST #36169, 1: 100) at 4 ˚C overnight. Next, the sections Committee of the First Hospital of Jilin University. The study was were stained with horseradish peroxidase anti-rabbit antibody and performed according to the guidelines with the Declaration of Helsinki detected by DAB kit (Thermo Fisher, USA). The nuclei were stained with and Ethics Committee of the First Hospital of Jilin University. Informed DAPI (Sigma, USA) at room temperature for 10 min. Images were obtained consents were obtained from all enrolled participants. Four pairs of HCC by Olympus FV1000 confocal microscopy. specimens and adjacent normal tissues were collected from the First Hospital of Jilin University between March 2019 and June 2019. Ten blood samples from healthy donors were collected, then peripheral blood Flow cytometry mononuclear cells (PBMCs) were isolated by Ficoll reagents (Sigma, USA) The orthotopic HCC tumor tissues or subcutaneous xenografts were according to the manufacturers’ instructions. dissected out and minced. Then tissues were digested with 0.8 mg/mL Collagenase IV (Sigma, USA) at 37 ˚C for 1 h. The cell suspensions were ﬁltered through 70 μm strainer and resuspended in 36% Percoll (GE Cell culture and reagents Healthcare, UK). PBMCs of anonymous human healthy donors were Murine HCC cell lines Hepa 1-6 and Hepa 1c1c7 and Human HCC cell lines isolated by Ficoll reagent (Sigma, USA) according to manufacturers’ HUH-7, SNU-449, HUH-6, and HepG2 were maintained in Dulbecco’smodiﬁed instructions. For cell surface staining, 1 × 10 cells were incubated with Eagle’s medium (DMEM, Gibco, USA) medium supplemented with 10% fetal anti-Fc receptor blocking antibody (2.4G2) at 4 ˚C for 15 min. Murine bovine serum (FBS, Gibco, USA), 100 units/mL penicillin and streptomycin samples were stained with anti-mouse CD45 APC, CD3 FITC, Gr-1 V450, (Gibco,USA). SortedMDSCs were cultured andexpandedby RPMImedium CD4 PE, CD8a V450, CD25 APC-CY7, Ly6C FITC, Ly6G PECY7, CD11 PE, and (Gibco, USA) supplemented with 10% FBS, 100 units/mL penicillin and F4/80 APC-CY7 from BD Bioscience (USA). Human samples were stained streptomycin (Gibco, USA), 40 ng/mL GM-CSF (Peprotech, Germany) and Cell Death Discovery (2023) 9:92 L. Ding et al. with anti-human CD45 APC, CD11 PE, CD33 FITC, and HLA-DR V450 from constructed by subcutaneously injected Hepa 1-6 cells (1 × 10 )intothe left BD Bioscience (USA). For intracellular staining of Arg1, Foxp3, and Ki67, ﬂank of 8-week-old male C57BL/6 mice. Two weeks after tumor injection, cells were ﬁxed and permeabilized by Fixation/Permeabilization solution mice were gave daily gavage of 50 mg/kg sorafenib (in PBS/1% Tween80), (BD Biosciences, USA) at 4 ˚C for 15 min. Then cells were washed and antibody against murine PD-1 (100 mg intraperitoneally every 4 days for ﬁve stained with anti-mouse Arg1, anti-mouse Foxp3, and anti-Ki67 from BD times, BioXCell, USA) or the IgG2b control isotype (100 mg intraperitoneally Bioscience (USA). Flow cytometry was performed on a B.D. Inﬂux cell sorter every 4 days for ﬁve times, BioXCell, USA). Tumor length was measures by (BD Bioscience, USA). Flowjo software was used to analyze the data. caliper every 3 days. At the end of drug treatment, the mice were anaesthetized by 3% isoﬂurane and sacriﬁced by broking the neck. Tumor xenografts were dissected out and weighed. Transcriptome RNA-sequencing and TCGA data analysis Total RNAs from sorafenib-treated or vehicle-treated orthotopic Hepa 1-6 tumors were extracted by TRIzol reagent (Thermo Fisher, USA). The Statistical analysis sequencing library was prepared using the Illumina’s TruSeq Stranded Statistical analysis was performed by GraphPad Prism 8 (GraphPad mRNA Sample Preparation kit (Illumina, USA) and sequenced on the Software, La Jolla, CA). Two-tailed Student’s t test and One-way ANOVA Illumina HiSeq 2500 platform. Fifty-base-pair sequenced reads were (Tukey’s post-hoc test) were used to compare difference between two or evaluated by the nextpresso pipeline (http://bioinfo.cnio.es/nextpresso/). more groups respectively. Data were shown as mean ± standard deviation The RNA sequencing data for the TCGA hepatocellular carcinoma samples (x ± s.d). P < 0.05 was considered as statistically signiﬁcant. were downloaded from the GDC Data Portal (https:// gdac.broadinstitute.org/) and analyzed as described. FastQC v0.11.0 was used to check sequencing quality. RNA-BisSeq method was performed to DATA AVAILABILITY ﬁlter the raw reads. The raw reads was mapped to mouse genome The data that support the ﬁndings of this study are available on request from the (GRCm38/mm10) or human genome (GRCh37/hg19) using the TopHat- corresponding author. 2.0.10 and evaluated by HTSeq. Differentially expressed genes were analyzed by DESeq2 and deﬁned as |Log Fold Change| ≥ 1.5 and adjusted p < 0.05. Samples were sequenced with three repeats. REFERENCES 1. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Tumor immune estimation resource (TIMER) and tumor- Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality world- immune system interactions database (TISIDB) analysis wide for 36 cancers in 185 countries. CA. 2021;71:209–49. TIMER was used to investigate the interactions between genes and tumor 2. Kulik L, El-Serag HB. Epidemiology and management of hepatocellular carcinoma. immune interactions (https://cistrome.shinyapps.io/timer/). The TISIDB was Gastroenterology. 2019;156:477–491.e471. used to analyze the tumor and immune system interaction (http:// 3. Bruix J, Reig M, Sherman M. Evidence-based diagnosis, staging, and treatment of cis.hku.hk/TISIDB/). In the present study, the correlation of midkine patients with hepatocellular carcinoma. Gastroenterology. 2016;150:835–53. expression with tumor immune inﬁltration was investigated by TIMER 4. Cheng AL, Kang YK, Chen Z, Tsao CJ, Qin S, Kim JS, et al. Efﬁcacy and safety of and TISIDB with the HCC database. The 22 tumor-inﬁltrating immune cells sorafenib in patients in the Asia-Paciﬁc region with advanced hepatocellular in HCC tissues were estimated by CIBERSORT algorithm. carcinoma: a phase III randomised, double-blind, placebo-controlled trial. Lancet Oncol. 2009;10:25–34. 5. Llovet JM, Ricci S, Mazzaferro V, Hilgard P, Gane E, Blanc JF, et al. Sorafenib in ELISA assay advanced hepatocellular carcinoma. N. Engl J Med. 2008;359:378–90. The levels of midkine, IL-10, and IFN-γ in serum samples or conditional 6. Tang W, Chen Z, Zhang W, Cheng Y, Zhang B, Wu F, et al. The mechanisms of medium were evaluated by ELISA kit (Abcam # ab193761, # ab279416, # sorafenib resistance in hepatocellular carcinoma: theoretical basis and ther- ab185986, #ab255729 and #ab174443, USA) according to manufacturer’s apeutic aspects. Signal Transduct Target Ther. 2020;5:87. instructions. Brieﬂy, samples were incubated with antibody-coated plates 7. Kudo M, Finn RS, Qin S, Han KH, Ikeda K, Piscaglia F, et al. Lenvatinib versus for 1 h at room temperature and washed for ﬁve times. Then plates were sorafenib in ﬁrst-line treatment of patients with unresectable hepatocellular incubated with biotinylated antibody, streptavidin antibody, substrate carcinoma: a randomised phase 3 non-inferiority trial. Lancet. 2018;391:1163–73. solution, and stop solution consecutively. The absorbance at 450 nm was 8. Yang YM, Kim SY, Seki E. Inﬂammation and liver cancer: molecular mechanisms measured by a microplate reader. All samples were done in triplicates. and therapeutic targets. Semin Liver Dis. 2019;39:26–42. 9. Fujiwara M, Garo LP, Murugaiyan G. PD1 blockade in cancer: impact on myeloid cells. Trends Cancer. 2020;6:443–4. T cell suppression assay 5 + + 10. Kudo M. Scientiﬁc rationale for combined immunotherapy with PD-1/PD-L1 5×10 CD3 CD8 cytotoxic T cells were stimulated with CD3/CD28 antibodies and VEGF inhibitors in advanced hepatocellular carcinoma. Cancers. dynabeads (Invitrogen, USA) in 24-well plates. At the same time, 5 × 10 2020;12:1089. sorted MDSCs were added into each well in the presence of human 11. Zhu AX, Finn RS, Edeline J, Cattan S, Ogasawara S, Palmer D, et al. Pembrolizumab recombinant IL-2 (R&D, USA) for 3 days. In separated experiment, cells in patients with advanced hepatocellular carcinoma previously treated with were supplemented with IL-10 antibody (Abcam #ab133575 or #ab189392, sorafenib (KEYNOTE-224): a non-randomised, open-label phase 2 trial. Lancet USA) or IgG isotype control (Abcam, USA). The cells were then stained with Oncol. 2018;19:940–52. surface markers of CD3/CD4/CD8 and intracellular protein of Ki67 12. El-Khoueiry AB, Sangro B, Yau T, Crocenzi TS, Kudo M, Hsu C, et al. Nivolumab in (eBioscience, USA) and Granzyme B (eBioscience, USA) for ﬂow cytometry. patients with advanced hepatocellular carcinoma (CheckMate 040): an open- All samples were done with three repeats. label, non-comparative, phase 1/2 dose escalation and expansion trial. Lancet. 2017;389:2492–502. Murine HCC models 13. Finn RS, Qin S, Ikeda M, Galle PR, Ducreux M, Kim TY, et al. Atezolizumab plus Hepa 1-6 or Hepa 1c1c7 cells were stably transduced with midkine, EV, sh- bevacizumab in unresectable hepatocellular carcinoma. N Engl J Med. MDK-1, sh-MDK-2, or sh-NC lentiviral particles as indicated for animal 2020;382:1894–905. experiments. Orthotopic HCC models were constructed as previously 14. Jain RK. Antiangiogenesis strategies revisited: from starving tumors to alleviating described [37]. 1 × 10 Hepa 1-6 or Hepa 1c1c7 cells were orthotopically hypoxia. Cancer Cell. 2014;26:605–22. implanted in the liver of 8-week-old male C57BL/6 mice. Brieﬂy, mice were 15. Mo Z, Liu D, Rong D, Zhang S. Hypoxic characteristic in the immunosuppressive anaesthetized by ketamine/xylazine, then median liver lobe was exposed by microenvironment of hepatocellular carcinoma. Front Immunol. 2021;12:611058. median laparotomy. 1 × 10 cells (suspended in 100 μL 1: 1 PBS and matrigel 16. Filippou PS, Karagiannis GS, Constantinidou A. Midkine (MDK) growth factor: a mixture) were orthotopically injected in the subcapsular region of liver by a key player in cancer progression and a promising therapeutic target. Oncogene. 28 G needle. One week after implantation, these mice were randomly divided 2020;39:2040–54. into sorafenib treatment group and vehicle group. Sorafenib treatment 17. Muramatsu T. Structure and function of midkine as the basis of its pharmaco- group received daily gavage of 50 mg/kg sorafenib (in PBS/1% Tween80), logical effects. Br J Pharm. 2014;171:814–26. while vehicle group received equal volume of PBS/1% Tween80. Mice were 18. Takemoto Y, Horiba M, Harada M, Sakamoto K, Takeshita K, Murohara T, et al. sacriﬁced 4 weeks post tumor implantation. Orthotopic HCC tumors were Midkine promotes atherosclerotic plaque formation through its pro-inﬂamma- dissected out, and inﬁltrating immune cells were isolated and evaluated by tory, angiogenic and anti-apoptotic functions in apolipoprotein E-knockout mice. ﬂow cytometry. In separated experiments, HCC tumor xenograft model were Circ J. 2017;82:19–27. Cell Death Discovery (2023) 9:92 L. Ding et al. 19. Weckbach LT, Gola A, Winkelmann M, Jakob SM, Groesser L, Borgolte J, et al. The 44. Sato W, Kadomatsu K, Yuzawa Y, Muramatsu H, Hotta N, Matsuo S, et al. Midkine cytokine midkine supports neutrophil trafﬁcking during acute inﬂammation by is involved in neutrophil inﬁltration into the tubulointerstitium in ischemic renal promoting adhesion via beta2 integrins (CD11/CD18). Blood. 2014;123:1887–96. injury. J Immunol. 2001;167:3463–9. 20. Cohen S, Shoshana OY, Zelman-Toister E, Maharshak N, Binsky-Ehrenreich I, 45. Finn RS, Ryoo BY, Merle P, Kudo M, Bouattour M, Lim HY, et al. Pembrolizumab as Gordin M, et al. The cytokine midkine and its receptor RPTPzeta regulate B cell second-line therapy in patients with advanced hepatocellular carcinoma in survival in a pathway induced by CD74. J Immunol. 2012;188:259–69. KEYNOTE-240: a randomized, double-blind, phase III trial. J Clin Oncol. 21. Guo X, Pan Y, Xiong M, Sanapala S, Anastasaki C, Cobb O, et al. Midkine activation 2020;38:193–202. of CD8(+) T cells establishes a neuron-immune-cancer axis responsible for low- 46. Chang CJ, Yang YH, Chiu CJ, Lu LC, Liao CC, Liang CW, et al. Targeting tumor- grade glioma growth. Nat Commun. 2020;11:2177. inﬁltrating Ly6G(+) myeloid cells improves sorafenib efﬁcacy in mouse ortho- 22. Cerezo-Wallis D, Contreras-Alcalde M, Troule K, Catena X, Mucientes C, Calvo TG, topic hepatocellular carcinoma. Int J Cancer. 2018;142:1878–89. et al. Midkine rewires the melanoma microenvironment toward a tolerogenic and 47. Zhou J, Liu M, Sun H, Feng Y, Xu L, Chan AWH, et al. Hepatoma-intrinsic CCRK immune-resistant state. Nat Med. 2020;26:1865–77. inhibition diminishes myeloid-derived suppressor cell immunosuppression and 23. Zhang Y, Zuo C, Liu L, Hu Y, Yang B, Qiu S, et al. Single-cell RNA-sequencing atlas enhances immune-checkpoint blockade efﬁcacy. Gut. 2018;67:931–44. reveals an MDK-dependent immunosuppressive environment in ErbB pathway- 48. Wang Y, Zhang T, Sun M, Ji X, Xie M, Huang W, et al. Therapeutic values of mutated gallbladder cancer. J Hepatol. 2021;75:1128–41. myeloid-derived suppressor cells in hepatocellular carcinoma: facts and hopes. 24. Zhu WW, Guo JJ, Guo L, Jia HL, Zhu M, Zhang JB, et al. Evaluation of midkine as a Cancers. 2021;13:5127. diagnostic serum biomarker in hepatocellular carcinoma. Clin Cancer Res. 49. Zheng W, Song H, Luo Z, Wu H, Chen L, Wang Y, et al. Acetylcholine ameliorates 2013;19:3944–54. colitis by promoting IL-10 secretion of monocytic myeloid-derived suppressor 25. Batlle E, Massague J. Transforming growth factor-beta signaling in immunity and cells through the nAChR/ERK pathway. Proc Natl Acad Sci USA. cancer. Immunity. 2019;50:924–40. 2021;118:e2017762118. 26. Grivennikov SI, Greten FR, Karin M. Immunity, inﬂammation, and cancer. Cell. 50. Yaseen MM, Abuharfeil NM, Darmani H, Daoud A. Mechanisms of immune sup- 2010;140:883–99. pression by myeloid-derived suppressor cells: the role of interleukin-10 as a key 27. Reynolds PR, Mucenski ML, Le Cras TD, Nichols WC, Whitsett JA. Midkine is immunoregulatory cytokine. Open Biol. 2020;10:200111. regulated by hypoxia and causes pulmonary vascular remodeling. J Biol Chem. 2004;279:37124–32. 28. Wu H, Zhen Y, Ma Z, Li H, Yu J, Xu ZG, et al. Arginase-1-dependent promotion of AUTHOR CONTRIBUTIONS TH17 differentiation and disease progression by MDSCs in systemic lupus ery- All authors guaranteed the integrity of the entire study. The experiments were thematosus. Sci Transl Med. 2016;8:331ra340. conducted by LD, NW, and QW Clinical studies were conducted by XF. Data were 29. Lopez-Valero I, Davila D, Gonzalez-Martinez J, Salvador-Tormo N, Lorente M, Saiz- analyzed by YX. The manuscript was prepared and reviewed by SW. All authors have Ladera C, et al. Midkine signaling maintains the self-renewal and tumorigenic read and approved the manuscript. capacity of glioma initiating cells. Theranostics. 2020;10:5120–36. 30. Saraiva M, Vieira P, O’Garra A. Biology and therapeutic potential of interleukin-10. J Exp Med. 2020;217:e20190418. ETHICAL APPROVAL AND CONSENT TO PARTICIPATE 31. Cai YQ, Lv Y, Mo ZC, Lei J, Zhu JL, Zhong QQ. Multiple pathophysiological roles of The experiments involved human subjects were followed with the ethical standards midkine in human disease. Cytokine 2020;135:155242. of the institutional committee and with the 1964 Helsinki declaration and its later 32. Zheng L, Li H, Huang J, Shin J, Luo S, Guo C, et al. Serum midkine levels for the amendments or comparable ethical standards. The animal studies were followed diagnosis and assessment of response to interventional therapy in patients with with the ethical standards of the ethics committee of the First Hospital of Jilin hepatocellular carcinoma. J Inter Med. 2021;4:39–45. University. 33. Omran MM, Farid K, Omar MA, Emran TM, El-Taweel FM, Tabll AA. A combination of alpha-fetoprotein, midkine, thioredoxin and a metabolite for predicting hepatocellular carcinoma. Ann Hepatol. 2020;19:179–85. 34. Yazihan N, Ataoglu H, Akcil E, Yener B, Salman B, Aydin C. Midkine secretion COMPETING INTERESTS protects Hep3B cells from cadmium induced cellular damage. World J Gastro- The authors declare no competing interests. enterol. 2008;14:76–80. 35. Ohuchida T, Okamoto K, Akahane K, Higure A, Todoroki H, Abe Y, et al. Midkine protects hepatocellular carcinoma cells against TRAIL-mediated apoptosis ADDITIONAL INFORMATION through down-regulation of caspase-3 activity. Cancer. 2004;100:2430–6. Supplementary information The online version contains supplementary material 36. Sun B, Hu C, Yang Z, Zhang X, Zhao L, Xiong J, et al. Midkine promotes hepa- available at https://doi.org/10.1038/s41420-023-01392-3. tocellular carcinoma metastasis by elevating anoikis resistance of circulating tumor cells. Oncotarget. 2017;8:32523–35. Correspondence and requests for materials should be addressed to Shudong Wang. 37. Chen Y, Ramjiawan RR, Reiberger T, Ng MR, Hato T, Huang Y, et al. CXCR4 inhibition in tumor microenvironment facilitates anti-programmed death Reprints and permission information is available at http://www.nature.com/ receptor-1 immunotherapy in sorafenib-treated hepatocellular carcinoma in reprints mice. Hepatology. 2015;61:1591–602. 38. Liang Y, Zheng T, Song R, Wang J, Yin D, Wang L, et al. Hypoxia-mediated Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims sorafenib resistance can be overcome by EF24 through Von Hippel-Lindau tumor in published maps and institutional afﬁliations. suppressor-dependent HIF-1alpha inhibition in hepatocellular carcinoma. Hepa- tology. 2013;57:1847–57. 39. Choudhuri R, Zhang HT, Donnini S, Ziche M, Bicknell R. An angiogenic role for the neurokines midkine and pleiotrophin in tumorigenesis. Cancer Res. 1997;57:1814–9. Open Access This article is licensed under a Creative Commons 40. Muramaki M, Miyake H, Hara I, Kamidono S. Introduction of midkine gene into Attribution 4.0 International License, which permits use, sharing, human bladder cancer cells enhances their malignant phenotype but increases adaptation, distribution and reproduction in any medium or format, as long as you give their sensitivity to antiangiogenic therapy. Clin Cancer Res. 2003;9:5152–60. appropriate credit to the original author(s) and the source, provide a link to the Creative 41. Zhou SL, Zhou ZJ, Hu ZQ, Huang XW, Wang Z, Chen EB, et al. Tumor-associated Commons license, and indicate if changes were made. The images or other third party neutrophils recruit macrophages and T-Regulatory cells to promote progression material in this article are included in the article’s Creative Commons license, unless of hepatocellular carcinoma and resistance to Sorafenib. Gastroenterology. indicated otherwise in a credit line to the material. If material is not included in the 2016;150:1646–1658.e1617. article’s Creative Commons license and your intended use is not permitted by statutory 42. Dong N, Shi X, Wang S, Gao Y, Kuang Z, Xie Q, et al. M2 macrophages mediate regulation or exceeds the permitted use, you will need to obtain permission directly sorafenib resistance by secreting HGF in a feed-forward manner in hepatocellular from the copyright holder. To view a copy of this license, visit http:// carcinoma. Br J Cancer. 2019;121:22–33. creativecommons.org/licenses/by/4.0/. 43. Li C, Wei S, Xu X, Jiang Y, Xue L, Jiang P, et al. Sorafenib attenuated the function of natural killer cells inﬁltrated in HCC through inhibiting ERK1/2. Int Immuno- pharmacol. 2019;76:105855. © The Author(s) 2023 Cell Death Discovery (2023) 9:92

Journal

Cell Death Discovery – Springer Journals

Published: Mar 11, 2023

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Midkine inhibition enhances anti-PD-1 immunotherapy in sorafenib-treated hepatocellular carcinoma via preventing immunosuppressive MDSCs infiltration

Midkine inhibition enhances anti-PD-1 immunotherapy in sorafenib-treated hepatocellular carcinoma via preventing immunosuppressive MDSCs infiltration

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Midkine inhibition enhances anti-PD-1 immunotherapy in sorafenib-treated hepatocellular carcinoma via preventing immunosuppressive MDSCs infiltration

Midkine inhibition enhances anti-PD-1 immunotherapy in sorafenib-treated hepatocellular carcinoma via preventing immunosuppressive MDSCs infiltration

References (52)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies