A tumor endothelial cell-specific microRNA replacement therapy for hepatocellular carcinoma

Summary Current approved anti-angiogenic drugs (AAD) for hepatocellular carcinoma (HCC) inhibit tumor angiogenesis, but affect the hepatic vasculature resulting in adverse effects. Tumor endothelial cells (TECs) differ from normal endothelial cells. In this study, we aimed to detect TEC-specific miRNAs and develop an anti-angiogenic treatment specific for TECs. We established HCC orthotopic mouse models. TEC-specific miRNAs were detected using a microRNA array. Finally, we evaluated the therapeutic effects of the TEC-specific miRNA agonist cocktail. In total, 260 TEC-specific genes were detected. Among the top ten downregulated TEC-specific miRNAs, miR-139-3p and 214-3p were important for the TEC phenotype. The TEC-specific microRNA agonist cocktail showed significant anti-tumor effects by inhibiting tumor angiogenesis without affecting hepatic vasculatures in HCC orthotopic mouse models. Moreover, it significantly downregulated tip-cell sprouting-related genes. We identified two downregulated TEC-specific miRNAs; microRNA replacement therapy, which targets the downregulated TEC-specific miRNAs, is an effective and promising treatment for HCC.


INTRODUCTION
Systemic treatments for hepatocellular carcinoma (HCC) have developed remarkably. 1 Drugs used in HCC treatment mainly target cells composed of the tumor microenvironment, such as tumor endothelial cells (TECs) and immune-related cells. 2 Molecular targeted therapy targeting tumor angiogenesis has become the main systemic treatment for various cancer types, including HCC. 3 Current anti-angiogenic drugs (AADs) inhibit angiogenesis by inhibiting the vascular endothelial growth factor (VEGF) signaling pathway. 2 Although anti-angiogenic therapy has contributed to improving the survival of patients with cancer, many clinical unmet needs, such as the appearance of adverse events and acquired resistance, still exist.Current AADs have a high incidence of adverse events because of their non-specificity for TECs.Currently, AADs commonly inhibit VEGF signaling.VEGF is essential for tumor angiogenesis and maintaining vascular structures in healthy organs. 4We have previously reported that an anti-VEGF antibody or sunitinib, a tyrosine kinase inhibitor that targets the VEGF receptor, inhibited tumor angiogenesis and vascular structures of endocrine organs such as the thyroid, which was correlated with the appearance of hypothyroidism. 5VEGF blockade also affects the hepatic vasculature.We found that VEGF blockade induced a reduction in hepatic vasculatures and morphological changes in liver sinusoidal endothelial cells (LSEC), which resulted in the appearance of adverse events and the acceleration of metastasis in cancer cells. 6Although current AADs are effective drugs, the non-specificity of TEC should be improved to further advance cancer treatment.
Recently, studies of non-coding RNAs, including microRNAs, have attracted attention. 7Accumulating evidence indicates an important role of miRNAs in cancer progression. 8miRNAs regulate gene expression by promoting messenger RNA degradation or repressing messenger RNA translation.miRNAs are important regulators of various cellular processes, including development, differentiation, and signaling in cancer cells. 8ased on a deeper understanding of the role of miRNAs in cancer progression, miRNAs have been introduced as promising therapeutic targets for cancer treatment. 8Particularly, microRNA replacement therapy that restores downregulated cancer-specific microRNA using microRNA agonists/mimics is attracting attention as a next-generation molecular targeted therapy. 9Although studies regarding cancer cell microRNAs and the development of microRNA replacement therapy targeting cancer cells are ongoing, microRNAs associated with the phenotype of TECs have not been fully understood, and microRNA replacement therapy that targets TECs has not been studied yet.
Recent studies have shown that TECs have morphological, functional, and genetically different characteristics compared to vascular endothelial cells of healthy organs. 10In the study, we hypothesized that TEC-specific genes and microRNAs that are important to maintaining the phenotype of TECs exist, and regulating such genes specifically can inhibit tumor angiogenesis without affecting the vascular structure of healthy organs.This study aimed to establish a novel anti-angiogenic therapy targeting TEC using TEC-specific microRNAs.

Current anti-angiogenic drugs affect not only tumor endothelial cells but also hepatic vasculatures
Our previous studies revealed that the current AADs that target VEGF signaling inhibited tumor angiogenesis and affected the vasculature of healthy organs. 6,11We investigated whether AADs approved for treating HCC, such as sorafenib and lenvatinib, also affect hepatic vasculature.Sorafenib significantly inhibited tumor angiogenesis and reduced perfusion in a mouse HCC orthotopic model (Figure 1A).Sorafenib also reduced microvessel density and blood perfusion in healthy and cirrhotic livers (Figure 1B).This phenomenon was also observed in livers treated with lenvatinib (Figure S1).To increase the clinical relevance, we evaluated tumor and hepatic blood perfusion changes in 14 patients with HCC treated with sorafenib using perfusion computed tomography (CT) (Figures 1C and 1D).Table S1 shows the baseline clinical characteristics of the patients, and Figure S2 shows the overall survival and progression-free survival of the registered patients.As with the results of the mouse experiments, sorafenib significantly reduced the total blood flow (TBF) not only in the tumor and but also the surrounding liver even in the clinical practice (Figure 1C).Taken together, these data indicate that current AADs affect the hepatic vasculature and reduce blood flow in the liver.

Tumor vessels are morphologically and functionally different from liver sinusoids
To understand the morphological and functional differences between TECs and hepatic endothelial cells, including LSEC, we created the HCC orthotopic mouse model (Figure 2A) and performed immunohistochemical staining.Although there was no significant difference in vessel intensity between TEC and LSEC, there was a significant difference in the distribution of vessel diameter between them, suggesting that TECs were significantly disorganized vessels (Figure 2B).In terms of function, tumor vessels showed leaky and less perfused vessels than the liver vasculature, followed by higher hypoxia in the tumor microenvironment (Figure 2B).Additionally, the number of proliferative endothelial cells was significantly higher in TEC than in LSEC (Figure 2B).

Genetic differences between tumor endothelial cells and liver sinusoidal endothelial cells
To shed light on the genetic differences between TEC and LSEC, we isolated the population of each vascular endothelial cell using magnetic beads.Fluorescence-activated cell sorting (FACS) analysis revealed that the TECs were isolated with over 80% purity (Figure S3).Compared to LSEC, 455 genes were significantly different between TECs and LSECs (Figure 3A).However, as cancer cell contamination is inevitable in the cell isolation method, we removed cancer cell-related genes from the 455 TEC-specific genes.Finally, we detected 260 TEC-specific genes that showed proliferative and pro-angiogenic features in the gene ontology analysis (Figures 3B and 3C).A microRNA array was used to detect TEC-specific miRNAs.We focused on downregulated miRNAs because the downregulation of TEC-specific miRNAs resulted in the upregulation of TEC-specific mRNAs (Figure 3D).The top 10 downregulated TEC-specific genes were detected using a microRNA array (Figure 3E).

Detection of important microRNAs that are associated with the TEC phenotype
Among the top 10 downregulated TEC-specific miRNAs, miR-139-3p, 214-3p, 199a-5p, 497-5p, and 455-3p were significantly downregulated in TECs (Figures 4A-4E and Figure S4).To assess whether these TEC-specific miRNAs are related to the TEC phenotype, tube formation assays using miRNA inhibitors were performed.In experiments involving miRNA inhibitors and mimics, we confirmed that the assay was appropriately conducted (Figure S5).Notably, the downregulation of miR-139-3p and 214-3p significantly increased tube formation (Figure 4F).The downregulation of these miRNAs simultaneously significantly increased tube formation compared with the downregulation of each micro-RNA (Figure 4G).These results suggest that miR-139-3p and miR-214-3p are important for achieving a TEC-specific phenotype in vascular endothelial cells.

Upregulation of miR-139-3p and 214-3p using miR mimics inhibited tube formation
Experiments using microRNA mimics were performed to determine whether miR-139-3p and -214-3p are promising therapeutic targets.Upregulation of miR-139-3p and 214-3p using microRNA mimics significantly reduced tube formation (Figure S6).MicroRNA mimics were also reduced tube formation under the exposure of microRNA inhibitors (Figure 5A).As both miR-139-3p and -214-3p were downregulated in the isolated TECs, we evaluated the effects of the miRNA mimics on the downregulation of both miRNAs (Figure 5B).The upregulation of miR-139-3p and 214-3p significantly reduced the number of tubes formed under these conditions.Additionally, upregulation of these miRNAs inhibited cell proliferation in vascular endothelial cells (Figure 5C) but not in cancer cells (Figure 5D).

TEC-specific microRNA agonist cocktail inhibited tumor growth through anti-angiogenesis without affecting hepatic vasculatures
To examine the effects of the detected TEC-specific miRNAs in treating HCC, the miRNA agonist cocktail (miR cocktail), including miR-139-3p and 214-3p mimics, was administered to the HCC orthotopic mouse models.A pilot study was performed using miR-negative and miR-positive controls (PC) to establish the assay.After injection, the administered miR-PC was detected in the tumor and TEC populations (Figures S7A and S7B).Moreover, the known target genes of miR-PC were significantly downregulated in the whole tumor and TEC populations (Figures S7C and S7D).We administered the miR cocktail to two orthotopic HCC mouse models based on the pilot study.Interestingly, the miR cocktail significantly inhibited tumor growth in both orthotopic HCC models (Figures 6A and 6B).
Changes in the tumor microenvironment due to the miR cocktail were evaluated using immunohistochemical staining (Figures 6C-6F).Tumor angiogenesis was significantly inhibited in tumors treated with the miR cocktail.Importantly, the miR cocktail did not decrease the vessel density in the hepatic vasculature.Significant hypoxia and necrosis in the tumor were detected following treatment with the miR cocktail.In addition, the miR cocktail treatment induced a significant reduction in cell proliferation, whereas an increase in cell apoptosis was observed only in the tumor, but not in the liver.These results suggest that the miR cocktail inhibits tumor growth by suppressing tumor angiogenesis without affecting the hepatic vasculature.

TEC-specific microRNAs specifically inhibited vessel sprouting via the downregulation of vascular tip cells-related genes
These vascular endothelial cells were isolated to confirm whether the miR cocktail was transfected into TEC and LSEC.Notably, the miR-139-3p and 214-3p levels were significantly increased in both TEC and LSEC populations (Figures 7A and 7B).The expression gene array was performed, and gene ontology analysis was conducted to determine what happened in the tumor treated with the miR cocktail (Figure 7C).We found that angiogenesis-related genes were significantly downregulated by the miRNA cocktail.Particularly, various genes related to the dynamics of vascular endothelial cells, such as development, angiogenesis, vasculogenesis, cell movement, migration, and proliferation, were significantly reduced by the miR cocktail administration (Figure 7D).These results suggested that the miR cocktail significantly inhibited tumor angiogenesis via the downregulation of various angiogenesis-related genes.However, one remaining question was raised: there was no impact on the liver vasculature, regardless of transfection into LSEC (Figures 7A and 7B).To shed light on the reason why the miR cocktail only affects TECs, but not LSEC, we focused on vascular tip-cell-related genes.Goveia et al. reported approximately 43 vascular tip-cellrelated genes in cancer, which were correlated with the vessel sprouting of TEC. 12 Interestingly, among the 43 vascular tip-cell-related genes, 78% were downregulated by the miR cocktail treatment (Figure 7E).An aortic ring assay was performed to confirm whether the miR cocktail regulated vessel sprouting.The miR cocktail significantly reduced vessel sprouting (Figure 7F).Finally, the regulation of TEC-specific micro-RNAs detected in this study downregulated the genes related to vessel sprouting.Therefore, they could only affect tumor angiogenesis, but not the hepatic vasculature.

DISCUSSION
In this study, we identified two downregulated TEC-specific microRNAs, miR-139-3p and 214-3p, in orthotopic HCC mouse models.We further demonstrated that the upregulation of these miRNAs due to the miRNA agonist cocktail administration specifically inhibited tumor angiogenesis without affecting hepatic vasculature.These miRNAs regulate the genes related to vascular tip cell sprouting in tumor angiogenesis, which could separate the anti-angiogenic action between tumor vessels and healthy vessels.
Nowadays, the treatment for HCC has remarkably developed and many molecular targeted agents (MTAs) have been approved for treating HCC. 13 The approved MTAs for HCC commonly inhibit tumor angiogenesis by blocking the VEGF signaling pathway. 2 Additionally, anti-VEGF antibodies are also used in combination therapy with immune checkpoint inhibitors. 14Although current MTAs have proven survival benefits for patients with HCC, the incidence of adverse events and acquired resistance are becoming issues that must be solved.The incidence rate of adverse events of approved MTAs is >90% at any grade. 11,15,16The dose of MTAs must be reduced by the appearance of adverse events, which results in deterioration of the anti-tumor effect and induction of acquired resistance.Maintaining liver function is essential for treating HCC. 17 However, as shown in this study, currently approved MTAs, such as sorafenib or lenvatinib, affect hepatic vasculature via VEGF blockade because VEGF is important for maintaining hepatic vasculature.This study showed that the TEC-specific microRNA cocktail did not affect the hepatic vasculature, which could be an ideal treatment in anti-angiogenic therapy for HCC.
Recent studies have shown that TEC has morphologically, functionally and genetically different characteristics compared to the vascular endothelial cells of healthy organs. 18Nanda et al. reported approximately 46 tumor endothelial markers (TEM), such as TEEM, TEM5, and TEM8. 19Hide et al. also reported that TECs are sensitive to growth factors and exhibit chromosomal abnormalities. 20These data indicated There was no significant difference in microvessel intensity between the two groups.There was a significant difference in the distribution of the vessel diameters (p < 0.001, n = 6 random fields per group).Whole-mount staining of 70 kD dextran and CD31 showed more leaked dextran (p < 0.05, n = 5 random fields per group, scale bar: 100 mm).Whole-mount staining of 2,000 kD dextran and CD31 showed fewer perfused vessels (p < 0.05, n = 5 random fields per group, scale bar: 100 mm).Immunohistochemical staining for CAIX and CD31 showed that the tumor had more hypoxia (p < 0.05, n = 6 random fields per group, scale bar: 20 mm).Immunohistochemical staining of PCNA and CD31 showed that proliferative endothelial cells were detected more in TEC (p < 0.05, n = 5 random fields per group, scale bar: 300 mm).TEC, tumor endothelial cell; LSEC, liver sinusoidal endothelial cell; HCC, hepatocellular carcinoma.
that TECs exhibit abnormalities and different phenotypes.Therapies targeting the TEC-specific gene have also been reported and have shown promising efficacy. 20However, studies on TEC-specific miRNAs have not yet been reported.This study is the first to report effective TEC-specific miRNAs in treating HCC, which has novelty and clinical significance from the point of view.In this study, tumor angiogenesis was specifically inhibited by restoring/replacing the TEC-specific downregulated miRNAs.6 This study was the first to demonstrate an effective microRNA replacement therapy that targets TEC.miRNAs significantly correlate with vascular development. 27This study revealed that microRNA replacement therapy targeting TEC-specific microRNAs could be a promising treatment for future molecular-targeted therapy.
This study found two important miRNAs that are essential for achieving the TEC phenotype.The first is miR-139-3p.Chi et al. reported that miR-139-3p downregulation was correlated with angiogenesis by regulating the expression of the S100A14 pathway. 28miR-139-3p is also associated with cancer cell proliferation. 29The second factor was miR-214-3p expression.Xiao et al. reported that miR-214-3p is correlated with endothelial migration and angiogenesis by regulating TrkB/ShcB expression. 30In this study, we found that these TEC-specific miRNAs significantly regulate the expression of vascular tip cell-related genes.Generally, vascular sprouting occurs during the fetal period or cancer progression. 31s the phenomenon does not generally occur in quiescent vascular cells in healthy organs, the treatment using the TEC-specific microRNAs detected in the study can only inhibit tumor angiogenesis with specificity, without affecting quiescent vascular cells in healthy organs.
In conclusion, we identified two downregulated TEC-specific miRNAs in orthotopic HCC mouse models.Furthermore, we demonstrated that the TEC-specific miRNA cocktail significantly inhibited tumor growth by inhibiting tumor angiogenesis without affecting hepatic vasculature.TEC-specific miRNA replacement therapy that targets only tumor-sprouting vessels can be a novel therapeutic concept and a promising treatment for patients with HCC.

Limitations of the study
This study found promising TEC-specific miRNAs that could be therapeutic targets in treating HCC.However, this study had some limitations.Although we found that these miRNAs significantly inhibited tumor angiogenesis by downregulating vascular tip cell-related genes, we could not identify the genes specifically binding to the detected TEC-specific microRNAs.Further studies are needed to identify the target genes.And we detected TEC-specific miRNAs only from downregulated genes isolated from TEC because the concept of the miRNA replacement therapy is to compensate the lacking miRNA.However, important miRNAs could also exist on the upregulated miRNAs.Therefore, further experiments are needed in the study.Moreover, its clinical relevance should be studied in future research.Whether TEC-specific microRNAs are detected in TECs isolated from human HCC samples and whether the regulation of these miRNAs is functional even for patients with HCC should be studied.However, the concepts identified in this study are important.The issue regarding the non-specificity of the targeted molecules for TEC in current AADs should be solved for further development in the field of anti-angiogenic treatment.The findings of the study showed promising outcomes from the point of view.(E) Waterfall plot of 46 tip-cell-related genes in TECs treated with the miR cocktail.Among them, the miR cocktail treatment downregulated 78% of the tip-cellrelated genes.(F) Aortic ring assay in miR negative control (NC) and miR cocktail (n = 3).*p < 0.05, **p < 0.01, ***p < 0.001, n.s.; not significant TEC, tumor endothelial cell; miR, microRNA; LSEC, liver sinusoidal endothelial cell.

Figure 1 .
Figure1.Evaluation of anti-angiogenic action for hepatic vasculatures in current anti-angiogenic drugs sorafenib (A) Whole-mount staining of tumor vessels and perfusion in HCC orthotopic mice treated with sorafenib (n = 6 random fields per group).Sorafenib significantly reduced vessel intensity (p < 0.01) and blood perfusion (p < 0.05).The scale bar represents 300 mm.(B) Whole-mount staining of tumor vessels and perfusion in healthy and cirrhotic livers treated with sorafenib (n = 5 random fields per group).Sorafenib significantly reduced blood perfusion in healthy (p < 0.01) and cirrhotic livers (p < 0.001).The scale bar represents 100 mm.(C) Perfusion CT in sorafenib-treated patients (n = 14).The scale bar represents 3 cm.(D) Quantification of tumor and hepatic blood flow in patients treated with sorafenib.Sorafenib significantly reduced blood flow in the tumor (p < 0.01) and liver (p < 0.001).HCC; hepatocellular carcinoma, CT; computed tomography.

Figure 2 .
Figure 2. Evaluation of morphological and functional differences between TECs and LSECs (A) The macro-finding of the HCC orthotopic mouse model.(B) Whole-mount staining of tumor vessels and hepatic vasculature (scale bar: 300 mm).There was no significant difference in microvessel intensity between the two groups.There was a significant difference in the distribution of the vessel diameters (p < 0.001, n = 6 random fields per group).Whole-mount staining of 70 kD dextran and CD31 showed more leaked dextran (p < 0.05, n = 5 random fields per group, scale bar: 100 mm).Whole-mount staining of 2,000 kD dextran and CD31 showed fewer perfused vessels (p < 0.05, n = 5 random fields per group, scale bar: 100 mm).Immunohistochemical staining for CAIX and CD31 showed that the tumor had more hypoxia (p < 0.05, n = 6 random fields per group, scale bar: 20 mm).Immunohistochemical staining of PCNA and CD31 showed that proliferative endothelial cells were detected more in TEC (p < 0.05, n = 5 random fields per group, scale bar: 300 mm).TEC, tumor endothelial cell; LSEC, liver sinusoidal endothelial cell; HCC, hepatocellular carcinoma.
miRNA replacement therapy aims to restore tumor suppressor miRNA function in tumor cells using synthetic miRNA mimics.Serej et al. and Orellana et al. demonstrated the efficacy of replacement therapy with miR Let-7 in treating cancer.

Figure 5 .
Figure 5.In vitro tube formation assay using the detected TEC-specific microRNAs mimics (A) Tube formation assay using miRNA mimics.After exposure to each or both miRNA inhibitors, each or both miRNA mimics were added to the culture media.The number of tubes formed was then counted (n = 3).(B) Tube formation assay using miRNA mimics.After exposure to both microRNA inhibitors, miRNA mimics were added to the culture media.(C) Cell proliferation assay of vascular endothelial cells (n = 6).(D) Cell proliferation assay of cancer cells (Hepa1-6, n = 6).*p < 0.05, **p < 0.01, ***p < 0.001, n.s.; not significant TEC; tumor endothelial cell.

Figure 7 .
Figure 7. Evaluation of genetic change in TECs isolated from tumors treated with the TEC-specific microRNA agonists cocktail (A) Expression of TEC-specific microRNAs in TEC isolated from Hepa1-6 orthotopic mouse model treated with a TEC-specific miRNA agonist cocktail (miR cocktail).(B) Expression of TEC-specific microRNAs in LSEC isolated from Hepa1-6 orthotopic mouse model treated with the miR cocktail.(C) Gene ontology analysis of TECs treated with the miR cocktail (n = 1 in each group).(D) Angiogenesis-related gene ontology analysis of TECs treated with miR cocktail.(E) Waterfall plot of 46 tip-cell-related genes in TECs treated with the miR cocktail.Among them, the miR cocktail treatment downregulated 78% of the tip-cellrelated genes.(F) Aortic ring assay in miR negative control (NC) and miR cocktail (n = 3).*p < 0.05, **p < 0.01, ***p < 0.001, n.s.; not significant TEC, tumor endothelial cell; miR, microRNA; LSEC, liver sinusoidal endothelial cell.