3.1 NGF-NGFR communication inefficiency in tumor tissues of HCC patients
Although T cells are quantitatively exhausted in tumor tissues, the mechanism underlying this T cell exhaustion is poorly understood to date. In the present study, T cells and tumor cells were separated from the tumor tissue and adjacent tissue samples from the single-cell RNA-sequence dataset (Fig. 1A, GSE149614). It was observed that few T cells had infiltrated into to tumor tissues of HCC patients (Fig. 1B, Table S4). Subsequently, the CellChat R package was used for evaluating the results of the cell-cell communication assay. Importantly, the NGF secreted from tumor cells was received mainly by the NGFR expressed in T cells (Fig. 1C), and this NGF-NGFR interaction contributed to the primary NGF-NGFR communication signaling pathway network (Fig. 1D). Unfortunately, only two clusters of tumor cells (cluster 3 and cluster 7, which were from adjacent tumor tissues) successfully secreted NGF individually, while tumor cells in clusters 5, 8, 10, and 11 (tumor tissues) failed to secrete NGF (Fig. 1E). Furthermore, the double immunofluorescence assay was performed to evaluate the NGF-NGFR communication in vivo. NGF was observed to be co-localized with NGFR on the membrane of the infiltrated T cells in the tumor tissues of HCC patients (Fig. 1F). Together, these results suggested that the tumor cell-secreted NGF was received specifically by the NGFR expressed in T cells, and NGF-NGFR communication was inefficient in the tumor tissues of HCC patients.
3.2 The exhaustion of T cells in HCC patients was driven by the infiltration process
In order to further understand the molecular characteristics of the infiltrated T cells in HCC patients, T cells were separated from the tumor and adjacent normal tissue samples obtained from the single-cell RNA sequence (Fig. 2A). It was observed that few T cells existed in the tumor tissues (Fig. 2B). Subsequently, the monocle 2 R package was employed for the Pseudo-Time analysis of exhausted T cells [15]. Interestingly, in the process of T cell infiltration, the exhausted T cells that had infiltrated into the tumor tissues were those that had differentiated from the adjacent tissues (Fig. 2C), and the T cell-specific molecular markers of this sub-population were completely reprogrammed during this infiltration process (Fig. 2D). Importantly, the T cell exhaustion biomarkers (LAYN, CTLA4, and HAVCR2) were highly expressed in the tumor tissues (Fig. 2E). Collectively, these data demonstrated that quantitatively and functionally exhausted T cells that had infiltrated into the tumor tissues were those that had differentiated from the adjacent tissues and then infiltrated into the tumor tissue in the HCC patients.
3.3 NGF-NGFR communication inefficiency suppressed the mitotic spindle signaling pathway
The above-stated results demonstrated that the NGF-NGFR communication failure in HCC was driven by the differentiation process. However, the underlying mechanism of NGF-NGFR communication in T cell exhaustion remains largely unclear. In order to better understand the molecular regulation mechanism and biological functioning of NGF and NGFR in tumor tissues, GSEA was performed using the pan-cancer approach. The result revealed that overexpression of NGF and NGFR could activate apoptosis and the mitotic spindle pathway (Supplementary Fig. 1–2).
Furthermore, the TCGA-LIHC RNA-Seq data were divided into NGF/NGFR high-expression and NGF/NGFR low expression groups of patients. Subsequently, 3576 Differentially Expressed Genes (DEGs) were screened out successfully (Supplementary Fig. 3A) using the criteria |logFC| > 0.8 and P < 0.05 (Supplementary Fig. 3B). Next, the GO enrichment analysis was performed, which revealed immune receptor activity, growth factor activity, extracellular matrix organization, and extracellular structure organization, among others, as the main enriched processes (Supplementary Fig. 3C-E). Further, the GSEA assay results suggested that NGF-NGFR communication was involved in the mitotic spindle signaling pathway (Supplementary Fig. 3F). In summary, these results indicated that NGF-NGFR communication inefficiency might affect the immune response progression in HCC via the mitotic spindle signaling pathway.
3.4 NGF-NGFR communication inefficiency induces mitotic arrest via the microtubule network
The preliminary findings of the present study suggested that the mitotic spindle could be regulated by NGF-NGFR communication. Therefore, NGF-stable and low-expression Huh7 cells and NGFR-transient and low-expression Jurkat T cells were infected with NGF-shRNA-Lv and NGFR-shRNA-rAd, respectively (Table S5). The microtubule network in the Jurkat T cells, collected from the Huh7 and Jurkat T co-cultured system, was incubated with an anti-α-tubulin antibody for immunofluorescence staining. The microtubule network in the Jurkat T cells from the NGF and NGFR low expression co-culture system presented a typical arrangement in most of the cells, while cells in the NGF and NGFR low expression co-culture system or those in the group exposed to colchicine and paclitaxel exhibited an irregular bipolar array kind of arrangement. Consistently, in the NGF/NGFR low expression system or the cells exposed to colchicine and paclitaxel, a small nuclear envelope was observed following the anti-lamin B staining (Fig. 3A). Further, monomeric tubulin and polymeric tubulin were extracted to determine the organization of tubulin. As depicted in Fig. 3B, the polymeric tubulin levels decreased following the NGF/NGFR interference or upon colchicine exposure, and this phenomenon was reversed upon paclitaxel exposure. These results suggested that NGF-NGFR communication inefficiency inhibited mitotic spindle formation through the inhibition of microtubule aggregation.
It is wildly accepted that mitotic spindle organization is necessary for the progression of the cell cycle [16]. Therefore, the flow cytometry analysis of the cell cycle, the mitotic index (H3-Ser28 staining) evaluation, and EDU staining were performed simultaneously to investigate the NGF-NGFR communication failure in cell proliferation regulation. As visible in Fig. 3C, a G0/G1 phase cell cycle arrest occurred in the NGF/NGFR low-expression system (58–67%). Consistently, decreased H3-Ser28- and EDU-positive cells were observed in the NGF/NGFR low-expression system (Fig. 3D-E). These results indicated that NGF-NGFR communication inefficiency suppressed cell proliferation through the inhibition of mitotic spindle formation.
3.5 NGF-NGFR communication inefficiency suppressed mitotic spindle formation through the HDAC1 unclear trans-localization inhibited PREX1 expression
Previous results of the present study suggested that NGF-NGFR communication inefficiency suppressed T cell proliferation by affecting the organization of the mitotic spindle. However, the detailed molecular mechanism underlying this effect remains to be elucidated so far. Therefore, the gene expression of the mitotic spindle pathway regulated by the NGF-NGFR communication (Table S5) was analyzed, and it was observed that NGF and NGFR were positively correlated with PREX1 in TCGA, LICH, GTEx, and the liver (Fig. 4A-D). Consistently, PREX1 was down regulated in the NGF/NGFR low-expression tumor tissues of HCC patients (Fig. 4E) and also in the NGF and NGFR low-expression co-culture system (Fig. 4F). However, PREX1 was suppressed through the deacetylation of histones at the promoter region with HDAC1 [17]. Therefore, the nucleus and cytoplasm were isolated from the cells of the NGF-NGFR low-expression co-culture system using a commercial nucleus extraction kit. The results revealed that HDAC1 was non-distinctly localized (Fig. 4G), and a similar phenomenon was observed in the confocal immunofluorescence staining analysis of HDAC1 (Fig. 4H). These results indicated that NGF-NGFR communication inefficiency suppressed mitotic spindle formation through HDAC1 unclear trans-localization-inhibited PREX1 expression.
3.6 Overexpressed PREX1 reversed the inhibition of mitotic spindle organization due to NGF-NGFR communication inefficiency
The above results indicated that NGF-NGFR communication inefficiency inhibited the organization of the mitotic spindle and the PREX1 expression. However, this does not necessarily suggest that the inhibition of the mitotic spindle formation due to NGF-NGFR communication inefficiency was induced by the suppression of PREX1. Therefore, three PREX1-siRNAs were constructed and transfected into Jurkat T cells. The western blotting analysis was performed, and the results revealed that the PREX1 effectively interfered with PREX1-siRNA1 (Fig. 5A and Fig. 5Ba). Subsequently, monomeric tubulin and polymeric tubulin were extracted to determine the organization of tubulin. It was revealed that the interference of PREX1 expression effectively inhibited the organization of tubulin (Fig. 5Bb-c), which was also confirmed in the immunofluorescence staining analysis (Fig. 5C). Consistently, the results of EDU staining and the mitotic index (H3-Ser28 staining) revealed that the interference of PREX1 expression suppressed cell proliferation (Supplementary Fig. 4A and Supplementary Fig. 5A).
In order to further investigate the role of PREX1 expression inhibited organization of mitotic spindle in NGF-NGFR communication inefficiency, the PREX1 expression in the NGF/NGFR low-expression co-culture system was reversed through the transfection of the pcDNA-PREX1 plasmid. Western blotting and immunofluorescence staining were performed to determine the organization of α-tubulin, and the results revealed that PREX1 expression effectively restored the polymerization of tubulin that was inhibited due to NGF-NGFR communication inefficiency (Fig. 5D-E). Similarly, cell proliferation suppressed due to NGF-NGFR communication inefficiency was also restored upon PREX1 expression (Supplementary Fig. 4B and Supplementary Fig. 5B). These results indicated that the NGF-NGFR communication inefficiency inhibited the organization of tubulin through the inhibition of PREX1 expression and, thus, induced the inhibition of cell proliferation.
3.7 NGF-NGFR communication inefficiency impaired the anti-tumor immunotherapy with PD-1 mAb in both mouse models and patients
The NGF-NGFR communication inefficiency was demonstrated to suppress T cell proliferation in the tumors of HCC patients. However, which particular kind of immune cells is suppressed due to NGF-NGFR communication inefficiency was unclear. The landscape of immune cells revealed that the memory CD4 + T cells, CD8 + T cells and activated NK cells had infiltrated into the NGF expression LIHC patients (Supplementary Fig. 6), while only memory CD4 + T cells had infiltrated into the NGFR expression LIHC patients (Supplementary Fig. 7). Furthermore, immune cells that had infiltrated the tumor tissues, including the CD56dim natural killer cells, central memory CD4+ T cells, natural killer T cells, activated CD8+ T cells, natural killer cells, and effector memory CD8+ T cells, were down regulated in the NGF/NGFR low-expression group of patients (Fig. 6). These results indicated that NGF-NGFR communication inefficiency inhibited the immune cell infiltration in the tumor tissues of patients.
In patients, the PD-1/PDL-1 interaction was demonstrated to induce the functional exhaustion of T cells by inhibiting cell proliferation and activation[18], while the NGF- NGFR communication was observed to be positively correlated with the expressions of PD-L1 and PD-1 in TCGA, LIHC, and GTEx (Supplementary Fig. 8). Therefore, to understand the role of NGF-NGFR communication in the anti-tumor immunotherapy with PD-1 mAb, the BALB/c nude mice were injected subcutaneously with the NGF-sh-Lv-infected Huh7 cells. After the successful formation of the subcutaneous tumor, the immune system was reconstructed via an injection of CD3+ T cells (1*106), which had been separated from PBMCs and transfected with NGFR-sh-rAd, and PD-1 mAb (Fig. 7A). Decreased levels of CD3+ cells and CD8+ cells (Supplementary Fig. 9) and increased tumor growth (Fig. 7B) were observed in the IgG mAb-exposed and NGF-sh-Lv/NGF-sh-rAd-injected mice. These results suggested that NGF-NGFR communication inefficiency impaired the PD-1 mAb therapy for tumor inhibition in the mouse model.
In order to further understand the role of NGF-NGFR communication in PD-1 mAb anti-tumor immunotherapy in HCC patients (Table S2), NGF–NGFR communication was studied in twenty-nine PD-1-positive patients exposed to PD-1 mAb, twelve non-responders, and seventeen responders. Two typical cases of NGF and NGFR expression and change in the tumor diameter (red line) upon PD-1 mAb anti-tumor immunotherapy are presented in Fig. 7C-D. Importantly, the NGF and NGFR low-expression patients were from the PD-1 mAb non-responders (Fig. 7E), and the changed diameter was positively correlated with NGF and NGFR expressions (Fig. 7F). These results suggested that the tumor inhibition ability of PD-1 mAb anti-tumor immunotherapy was weakened due to the NGF-NGFR communication inefficiency in the tumor tissues of HCC patients.
3.8 NGF-NGFR communication inefficiency resulted in incursive clinicopathological characteristics and disappointing prognosis in HCC patients
In order to unravel further details of the NGF-NGFR communication in HCC patients, thirty-three cancer types and nine HCC RNA-seq datasets were collected and evaluated for NGF and NGFR expressions. A down regulation of NGF was observed in the tumor tissues of BLCA, BRCA, CESC, COAD, KICH, KIRP, LIHC, PRAD, and UCEC, while NGFR was downregulated in BLCA, BRCA, CHOL, COAD, KIRP, LIHC, LUAD, PRAD, READ, STAD, and UCEC (Fig. 8A). In the HCC RNA-seq datasets, NGF was significantly downregulated in three RNA-seq datasets, while NGFR expression was low in all RNA-seq datasets (Fig. 8B). Consistently, the downregulation of NGF and NGFR in the tumor tissue of HCC patients was observed in the qRT-PCR analysis (Table S3, Fig. 8C) and western blotting (Fig. 8D) as well. These results suggested an impaired NGF-NGFR communication in the tumor tissues of these patients.
In order to further elucidate the contribution of NGF-NGFR communication inefficiency, the correlation of the down-regulated NGF and NGFR with the survival of HCC patients was evaluated. The low expression NGF/NGFR group patients presented poor overall survival (31 vs. 22 months, P = 0.023), reduced time to disease-free survival (31 vs. 22 months, P = 0.033), reduced disease-free interval (24 vs. 17 months, P = 0.01), and worse progression-free survival (21 vs. 15 months, P = 0.0048) (Supplementary Fig. 10A-D). A similar result was obtained for the HCC patients in the low expression NGF or NGFR patients (Supplementary Fig. 5). In addition, the multivariate analysis revealed NGF and NGFR levels, BMI scores, virus infection, and tumor stage as the independent risk factors for OS, DSS, DFI, and PFI (Supplementary Fig. 10E-H). Collectively, these results suggested that the frequently low-expressed NGF and NGFR were correlated with incursive clinicopathological characteristics and disappointing prognosis in HCC patients.