NOL3 is remarkably upregulated in ccRCC
Initially, we explored NOL3 mRNA levels in tumors and matched normal tissues in various types of cancer on the TIMER2.0 database. The expression of NOL3 mRNA is much more elevated compared with that in normal tissues in lung adenocarcinoma (LUAD), colon adenocarcinoma (COAD), squamous cell lung carcinoma (LUSC) and especially in Kidney renal clear cell carcinoma (KIRC) (Fig. 1A). Additionally, in the Oncomine database, five datasets indicated NOL3 mRNA expression is higher in tumors than in normal tissue (Fig. 1B). Consistent with the results of TIMER2.0 and the Oncomine database, Ualcan database also showed that the mRNA level of NOL3 is upregulated in ccRCC. Moreover, the protein expression of NOL3 in the CPTAC was different between normal and tumor tissues, and so was it among different stages (Fig. 1C).
Next, 40 paired samples derived from patients diagnosed with ccRCC in Xinhua Hospital were used to determine whether the expression of NOL3 is upregulated in tumor tissue, and the results revealed that NOL3 expression in tumor species is higher than in normal tissue (Fig. 2A). Besides, the results of immunohistochemistry also revealed that NOL3 expresses higher in tumor tissues than in paired adjacent normal tissue samples (Fig. 2B). The representative IHC images for different H-Scores are presented in Fig. 2C.
NOL3 is associated with malignant phenotype and unfavorable prognosis
The expression of NOL3 was evaluated in HK-2 and renal carcinoma cell lines (ANCH, Caki-1, A498, 786-O and 769-P) using Western Blot. The results revealed that NOL3 protein expression was higher in tumor cells (Fig. 3A). GFP-tagged recombinant plasmids were used to upregulate NOL3 expression in ACHN and786-O, and the efficiency of NOL3 overexpressing was confirmed using Western Blot and qPCR (Fig. 3B). Interestingly, the upregulation of NOL3 affected neither the migrating ability nor the invasive ability of 786-O and ACHN (Supplementary Figure S1A). However, the proliferation assays and colony formation assay clearly indicated that overexpression of NOL3 notably increased the proliferation of ccRCC cells (Fig. 3D and Supplementary Figure S1B). And the apoptosis rate of ccRCC cells demonstrated that the anti-apoptosis abilities of both ACHN and 786-O cells were enhanced (Fig. 3E). Furthermore, shRNA was used to deplete NOL3 expression in the same cell lines, and the efficiency of NOL3 downregulation was confirmed (Fig. 3C). As expected, no remarkable reduction in the number of ccRCC transwell migration or Matrigel invasion cells was noted in NOL3-silenced cells (Supplementary Figure S1 A), whereas NOL3 knockdown greatly impaired the proliferation ability and promoted apoptosis of both cell lines (Fig. 3D and F). Taken together, these findings supported the hypothesis that NOL3 could regulate the malignant phenotype of ccRCC cells.
Next, to validate the impact of NOL3 expression on clinical outcomes, the relationship between NOL3 RNA expression and prognosis in ccRCC was analyzed through the GEPIA website (Fig. 4A and B). The results suggested a survival advantage in both the overall survival (OS) and disease-free survival (DFS) for the low NOL3 expression group over the high NOL3 expression group, although no statistical significance was observed. 302 patients collected from Eastern Hepatobiliary Surgery Hospital and Xinhua Hospitals were recruited. Tissue microarrays derived from all patients were used to confirm the expression of NOL3, which is quantified by H-score (see Materials and Methods). ROC curve was employed to identify the proper cut-off value which classified patients into the NOL3-overexpression group and the NOL3-low expression group (Fig. 4C). Patients in the NOL3-low expression group have longer OS and PFS than those in the NOL3-overexpression group (Fig. 4D-I).
Univariate and multivariate Cox regression models demonstrated that NOL3 expression was a potential independent risk factor in predicting the clinical outcome of ccRCC patients (Table 1–3). Additionally, the C-index of combined NOL3 expression and TNM stage or SSIGN score for OS or PFS of ccRCC patients was found greater than that of NOL3 or any other markers alone. This implied that combining NOL3 expression with the TNM stage or SSIGN score might provide a better accuracy in predicting ccRCC patient’s prognosis.
Hypoxia stimulated NOL3 expression and promoted IL13 secretion which induced macrophage M2 polarization in ccRCC
Hypoxia is a common characteristic of the microenvironment of solid tumors, especially in ccRCC. It is also known as a poor prognosis factor in ccRCC patients. We further investigated the relationship between hypoxia and NOL3 expression using GSEA analysis (Fig. 5A) From real-time quantitative PCR analysis, Western Blot of 786-O and ACHN exposed to 21% O2, 10% O2, 5% O2, 2% O2 and 1% O2, respectively, it was observed that NOL3 expression was induced by low oxygen and achieved the greatest upregulation in the condition of 5% or 2% O2 (Fig. 5B and C). Additionally, according to GSEA enrichment analysis, NOL3 is associated with interleukin-13 production (Fig. 5D). The results of ELISA showed that NOL3 promoted the production of intracellular IL13 and increased the secretion of IL13 into extracellular supernatant (Fig. 5E and F, Supplementary Figure S2 A).
The growth of ccRCC is also associated with a marked immune suppressive microenvironment, partially characterized by an accumulation of macrophage of the M2 phenotype. Moreover, previous research reported that IL13 plays a key role in the dynamic process of transition between M1 and M2 phenotypes24. THP-1, a man-derived leukemic monocyte lymphoma cell line was used to investigate the relationship between IL13 and M2 polarization. Q-PCR was employed to explore the percent of M2 or M1 subtype markers expressed in THP-1. The results showed that after cocultured with NOL3-overexpression 786-O and ACHN cells, the mRNA expression level of M2 subtype markers, such as NOS2, CD163, and ARG1, were remarkably enhanced. However, the mRNA expression level of M1 subtype markers, such as CD86, CD80, and iNOS, was decreased after coculturing with ccRCC cells (Fig. 5G and Supplementary Figure S2 B). Meanwhile, knocking down of NOL3 in 786-O and ACHN impaired its stimulation of THP-1 towards M2 polarization (Fig. 5H and Supplementary Figure S2 C).
Besides, the migration assays were employed to measure the activation of THP-1, and the results showed that THP-1 cocultured with ccRCC cells tend to be more active, which was regulated by NOL3 (Fig. 5I and J, Figure S2 D). Next, ELISA assays were used to determine the secreting function of THP-1, and the secretion of IL-10, TGF-β, and TNF-α was enhanced after cocultured with ccRCC cells. Likewise, the secretion of IL10, TGF-β, and TNF-α was regulated by NOL3 (Fig. 5K-M, Supplementary Figure S2 E-G).
Taken all together, NOL3 was associated with the production and secretion of IL13 and hypoxia promoted the expression of NOL3 and facilitated the M2 polarization of macrophages in ccRCC.
M2 polarized macrophages induce VEGF secretion and promote TKI resistance in ccRCC
Since ccRCC exhibited the highest expression of VEGFA among all epithelial cancers, VEGFA together with its receptors (in particular, VEGFR2) have long been the main therapeutic targets in this disease25. VEGF promotes ccRCC progression via directly inducing angiogenesis and indirectly facilitating immunosuppressive tumor microenvironment (TME). On the other hand, M2 polarized macrophages in the TME, also known as tumor-associated macrophages (TAMs), represent the main cluster of the TME, which contribute to tumor progression through different effects such as provision of a niche for stem cells, genetic instability, epithelial to mesenchymal transition (EMT), invasion and metastasis26. Thus, we tend to investigate whether TAMs promote VEGF secretion in the ccRCC.
17 patients with recurrent ccRCC treated with Pazopanib were recruited from Xinhua Hospital and Eastern Hepatobiliary Surgery Hospital, and separated into a longer survival group and a shorter survival group according to the median survival (9.7 months). IHC staining quantified with H-Scores was performed on the tumor tissues of these patients. The results revealed a negative correlation between NOL3 expression and survival time (Fig. 6A). Thus, we deduced that NOL3 expression negatively affects pazopanib efficacy.
M2 polarized macrophages (TAMs) in the tumor tissue of all the 302 patients recruited were also detected using IHC staining of CD163 (Fig. 6B). ROC was employed to generate the cut-off value and patients were divided into two groups, TAMs high expression group and a TAMs low expression group. We discovered a positive correlation between NOL3 expression and TAMs infiltration in ccRCC samples (Fig. 6C).
Previous research reported that TAMs induces angiogenesis via VEGF secretion27. To validate this phenomenon in ccRCC, 786-O cells were co-cultured with THP-1 cells to simulate the TME of renal cancer, where the interaction between the renal cancer cells and the TAMs is maintained. To validate the effect of TAMs on ccRCC cells and eliminate the effect of TAMs on VEGF, after co-cultured with TAMs, 786-O or ACHN cells were isolated from the co-culture system and cultured separately for another 24h. Then, the culture medium was collected for further experiments. We detected the concentration of VEGF in the medium using the ELISA kit and found that VEGF is increased in the co-culture system of 786-O and THP-1 compared to that in the medium of either single 786-O or THP-1. Meanwhile, upregulation of NOL3 in 786-O cells further enhanced the generation of VEGF in the co-culture system while knocking down of NOL3 impaired it (Fig. 7A and B). Moreover, hypoxia further facilitated the secretion of VEGF positively regulated by NOL3, while this enhancement was counteracted by the downregulation of NOL3 (Fig. 7C and D). This indicated that the presence of TAMs is necessary for the VEGF regulatory efficacy of NOL3.
To further validate the previous conclusions of the co-culture assays in vivo, subcutaneous tumor formation was performed using nude mice. In line with the previous results, overexpression of NOL3 enhanced the tumorigenic capacity of ACHN and promoted TKI resistance (in particular, pazopanib) in mice with normal viability of macrophages. Interestingly, the depletion of macrophages with Clodronate Liposomes remarkably restored the anti-tumor effects of pazopanib. (Fig. 7E and F)
Taken together, overexpression of NOL3 in ccRCC modulates the mutual interaction between ccRCC and TAMs, resulting in the upregulation of VEGF, which promotes TKI resistance.
The combination of NOL3 expression and TAMs biomarkers better predicts the prognosis of ccRCC patients
The remarkable upregulation of NOL3 expression and its synergy with TAMs in ccRCC inspired us that the combination of NOL3 expression and TAMs markers may serve better in the prediction of ccRCC prognosis. Therefore, a total of 308 patients with the postoperative pathological diagnosis of ccRCC from Xinhua hospital and Eastern Hepatobiliary Surgery Hospital were recruited and respectively assigned to a training or validation cohort. IHC was applied to the tissue samples of these patients and the staining results were quantified by H-score generated by two experienced pathologists independently (see Materials and methods for details). Moreover, M2-polarized macrophages in the tumor tissue of these patients were also detected using IHC. Also, a ROC was employed to generate the cut-off value and patients were divided into two groups (Fig. 8A). Kaplan-Meier analysis was employed to generate survival cures which suggested that ccRCC patients in the NOL3lowCD163low group tend to have the best prognosis (in particular, longer overall survival and progression-free survival) among all the four groups, while patients with high expression of NOL3 and CD163 at the same time have the most unfavorable clinical outcomes (Fig. 8B-G). A Cox regression model was built based on the clinicopathological features and the clinical outcomes of the patients, which demonstrated that NOL3 expression and CD163 expression are independent risk factors for prognosis in ccRCC patients. (Table 1–3) C-index Consistency analysis validated that the combination of NOL3 expression and CD163 positive TAMs infiltration is a better index of ccRCC patients’ prognosis. (Table 4)