Study Of The Heterogeneity Of Tumor Endothelial Cells In Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is one the most prevalent cancers in China. Its incidence is 0.379%o in male and 0.142%o in female, and ranks the second cancer killer in China. Most patients with HCC are diagnosed at late stage, which is not amenable to curative treatment. However, systemic treatment (mostly chemotherapy) is generally not effective in HCC. Sorafenib is a multiple molecule inhibitor, which presents anti-angiogenesis effect through targeting VEGFRs, has been proved to be a standard of care for advanced HCC. Although VEGFR and fibroblast growth factor receptor (FGFR) are two major molecular targets for anti-angiogenesis treatments, other mechanism of these drugs, for example, the off target effect, is often questioned. Receptor kinase in tumor endothelial cell (TEC) is the major target of anti-angiogenesis treatment. However, the understanding of TEC remains largely unchanged during the last 10 years. High-throughput techniques, including genomic, gene expression profiling and proteomics, has not been widely used due to limited availability of TEC. On the other hand, many studies used human umbilical vein endothelial cell (HUVEC) and cultured microvessel endothelial cells (MEC), which are different from that from tumor vessels. We have established magnetic bead method to isolate TEC and NEC from human HCC specimens or orthotopically implanted human HCC in nude mice. Isolated EC can be passaged in vitro and we have proved that cultured EC within 3-4 passages remains similar cell surface marker, morphology and growth rate. We have successfully isolated CD105+TEC and CD31+TEC, and found they were different in proliferation, tube formation and cell surface markers (VEGFR1 and VEGFR2). Furthermore, we found they are also different in resistance to and anti-angiogenesis drug s(sorafenib,Sunitnib,Brivanib). We have analyzed the gene expression profiles between TEC and NEC, as well as remnant cells from tumor tissue (TCs) and remnant cells from nontumor liver tissue (NTCs) after isolation of CD31+ cells or CD105+cells.Combining with the results of gene expression profiles, we did our experiments in vitro and animal experiments to further confirm the source of the TEC, and to explore its molecular mechanism, so as to find new ways for anti-angiogenesis drugs and improve the efficacy of anti-angiogenesis. Our new findings include:1. We have established magnetic bead method to isolate TEC and NEC from human HCC specimens or orthotopically implanted human HCC in nude mice. Isolated EC can be passaged in vitro and we have proved that cultured EC within 3-4 passages remains similar cell surface marker, morphology and growth rat. The immune magnetic beads of the separation system (MACS) have no poisonous to the cells and can be degraded and does not affect to the cell characteristics, so it si suitable for sorting after subculture of subsequent experiments. We have successfully isolated CD105+TEC and CD31+TEC. These results suggested TEC is composed of different subgroups as our previous study suggested TEC from one single tumor may have several subgroups. We hypothesize that the composition of TEC may influence the net effect of anti-angiogenesis treatment; furthermore, These results make us hypothesize that continuous treatment by a anti-angiogenesis drug may induce change of composition of TEC subgroups (phenotype shift), which may result in a resistance to this treatment.2. After isolated by antibody conjugated magnetic beads, We compared the CD105+TEC and CD31+TEC in the difference of cell function and anti-angiogenesis drugs (sorafenib), we found they are also different in resistance to and anti-angiogenesis drugs (sorafenib), CD31+TEC has stronger resistance to the anti-angiogenesis drugs. Heterogeneity can be identified in growth rate, surface marker, tube formation, sprouting and resistance to anti-angiogenesis treatment between CD105+TEC and CD31+TEC, which may associated with difference in gene expression profiles.3. To determine expression of EC-related genes, we performed global transcriptomic profiles of CD31+ or paired CD105+ TEC and NEC isolates from 8 patients with HCC as well as matched CD31+ TEC, NEC, TC, and NTC isolates from another 8 patients with HCC. Unsupervised hierarchical clustering analysis revealed that TEC isolates, regardless of CD31 or CD105 status, were more closely related to TC isolates than to NEC and NTC isolates. NEC isolates were clustered largely separated from NTC isolates. Similar results were obtained with the use of principal component analysis. These results indicate that the transcriptomic profiles of TECs were much more similar to the profiles of tumor cells than to the profiles of matched NECs.Class comparison analyses with paired t-statistics revealed that a total of 443 genes were differentially expressed between CD31+ TECs and matched CD31+ NECs (p< 0.001). Similar analyses conducted by comparing NTCs to matched TCs or TECs to matched TCs identified 1151 or 201 differentially expressed genes, respectively. Venn diagram analysis revealed that a majority of the TEC-related genes (278 of 443 genes) overlapped with TC-related genes. In contrast, a majority of the EC-related genes (TECs vs TCs; 147 of 201 genes) did not overlap with the TC-related genes. There appear to be more differences in gene expression between tumor cells and nontumor cells than there are differences between tumor cells and TECs. Taken together, these results suggest that the activities of TEC genes are largely contributed by tumor cells and that these genes are related to HCC prognosis. Hierarchical clustering of 443 TEC-related genes in TEC, NEC, TC and NTC isolates from 16 patients with HCC revealed that these genes can be separated into 7 main clusters (C1-C7) on the basis of expression similarities. The genes in C5, C6, and C7 were more abundantly expressed in TECs and TCs than in NECs and NTCs, such as osteopontin (SPP1), matrix metalloproteinase 9 (MMP9), and matrix metalloproteinase 12 (MMP12) genes. Note that the top rank genes highlighted in these clusters were previously reported to be HCC related.4. The above-described results suggest that TECs are not ECs embedded inside tumors and that instead TECS originate from tumor cells with the characteristics of ECs. We did our experiments in vitro and animal experiments to further confirm the source of the TEC:1) We choose the AFP positive HCC tissue, so the AFP as a sign of tumor cells. We found that both TECs and TCs also had elevated AFP expression compared to that in NECs both in the results of global transcriptomic profiles and qRT-PCR. Because one hallmark of tumor cells is aneuploidy, we tested this hypothesis by randomly selecting 7 CD31+ TEC isolates from viable cell cultures, and we found that all 7 TEC isolates exhibited aneuploidy, as indicated by the fact that they showed a gain in the number of chromosomes per metaphase spread. In contrast, none of the 3 NEC isolates used as controls showed chromosome gain.2) To determine whether TECs were tumor-initiating cells, we tested the tumorigenicity of CD31+ TEC isolates along with that of matched CD31+ NEC, TC, and NTC isolates from 8 patients with HCC. A total of 104-106 primary cells were subcutaneously injected into NOD/SCID mice. We found that freshly isolated primary TECs and TCs from 6 out of 8 patients induced tumors 7-17 weeks after implantation, and the latency of tumorigenesis for the TCs was much shorter than that for the TECs. Histological evaluation revealed that xenograft tumors derived from some of the TEC isolates had histological features similar to those of matched primary HCC specimens and expressed CD31.3) To examine whether HCC cells express CD31, CD31 mRNA expression could be detected in HCC-LM3 cell lines by qRT-PCR. There were few hCD31 expression detectable in the HCC-LM3cell lines, HUVECs used as a control expressed CD31. We also evaluated the capillary tube formation capacity of cultured HCC cell lines. We found that not like CD31+ TEC isolates, HCC-LM3 cells had no ability to induce capillary tube formation. To determine whether HCC cells could be incorporated into tumor vasculature in vivo, we subcutaneously injected HCC-LM3 cells into nude mice to induce tumor formation. We then performed immunofluorescence staining of tumor sections using human-specific anti-CD31 or murine-specific anti-CD31 antibodies. We found that human CD31+ cells were often located adjacent to mouse CD31+ cells and that together they formed vessel-like structures, suggesting that HCC cells were incorporated into tumor vasculature together with mouse ECs.4) To distinguish HCC cells from ECs, we also used HCC tissues and adjacent nontumor tissues to perform immunostaining with AFP-specific and CD31 antibodies. We found that CD31+ AFP+ cells could frequently be found in HCC tumor tissues but not in nontumor liver tissues, and most of the dual positive cells could be found on or near blood vessels.5. We also did the SNP data analysis, the analysis of the data found that HCC cells have the biggest changes on chromosome copy number, but also found TEC and HCC cells have many similarities in the copy number changes, and copy number changes on chromosome 8 are more similar. The concrete results are in further bioinformatics analysis.Conclusions.There have different subgroups of TEC in HCC and they have different in resistance to anti-angiogenesis drugs. The global transcriptomic profiles analysis revealed that TEC isolates were more closely related to TC isolates than to NEC and NTC isolates. These evidences that a subset of TECs may originate from HCC cells.The potential application of this work1. The heterogeneity of TEC in HCC may be one of the reasons to resistint anti-angiogenesis drugs The different subgroups of TEC are different in resistance to and anti-angiogenesis drugs (sorafenib,Sunitnib,Brivanib) Theses results provides a new research direction for further research.2. A subset of TECs may originate from HCC cells. These results can be expected to guide the development of new antiangiogenic therapies.Originalities of this work1. This is the first time to sepearate two different subtypes of TEC(CD31+and CD105+TEC) in human HCC tissues by antibody conjugated magnetic beads that we have established magnetic bead method, and this heterogeneity was identified in growth rate, surface marker, tube formation, sprouting and resistance to anti-angiogenesis treatment, which may associated with difference in gene expression profiles.2. This is the first report verifying that a subset of TECs may originate from HCC cells.