A cohort of 100 patients with clear cell RCC (ccRCC) diagnosed at the Saitama Medical Center between 2008 and 2019 were retrospectively analyzed. We evaluated the overall survival of the patients from the time of their first visit. Patient characteristics are shown in Table 1 . Patients who did not agree to participate in the study were excluded. The clinical parameters whose values were not available were excluded from the statistical analysis to compare the patient characteristics. Pathological findings were classified according to the Japanese General Rules for Clinical and Pathological Studies on Renal Cell Carcinoma (17). The Institutional Review Board of the Saitama Medical Center, Saitama Medical University, approved the clinical protocols (No. 117 and No. 2308).

The Cancer Genome Atlas (TCGA) RCC dataset retrieved from The Human Protein Atlas (https://www.proteinatlas.org/) was used for prognostic analysis for NRN1 and CXCR4 mRNA. For co-expression analysis, mRNA expression data in TCGA ccRCC RNA-sequencing (RNA-seq) dataset were retrieved from cBioPortal (https://www.cbioportal.org/). Gene ontology (GO) analysis was performed using The Database for Annotation, Visualization and Integrated Discovery (DAVID) (https://david.ncifcrf.gov/tools.jsp).

Patient-derived ccRCC spheroid cultures were established from primary tumors of patients with RCC after obtaining informed consent at the Saitama Medical Center. Tumor samples were processed and cultured as three-dimensional spheroids as described previously (13). Characteristics of the patient PDCs are summarized in Supplementary Table 1 . The Ethics Committee of the Saitama Medical Center, Saitama Medical University, approved all procedures (No. 1363-IV).

RNA extraction, cDNA synthesis, and qRT-PCR were performed as described previously (13). β-Actin (ACTB) was used as the internal control. All qRT-PCR primers used in this study are listed in Supplementary Table 2 .

siRNAs targeting NRN1 (siNRN1 #1 and #2) and control siRNA (siControl) were purchased from RNAi Inc. and transfected into PDCs with RNAiMAX reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. The siRNA sequences used are shown in Supplementary Table 3 . Flag-tagged NRN1 cDNA was subcloned into pcDNA3 (Invitrogen) and transfected into PDCs with Lipofectamine 3000 reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Cell viability was assessed using the CellTiter-Glo 3D Assay (Promega) on day 3 after plating and transfection.

RCC-PDC1 cells transfected with siRNAs or expression vectors were lysed in RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% SDS, 1% Triton X-100, 1% sodium deoxycholate, 1 mM PMSF, 1 μg/mL aprotinin) and the plasma membrane fraction was purified by centrifugation at 19,100g for 20 min at 4°C. Proteins were separated with SDS-PAGE and blotted on PVDF membrane, followed by reactions with anti-CXCR4 antibody (GTX22074, GeneTex) or anti-β-actin antibody (AC-74, SigmaAldrich) for loading control.

Formalin-fixed tissues and PDCs resuspended in iPGell (Genostaff) were embedded in paraffin and sectioned as previously described (13). A Histofine kit (Nichirei) based on the streptavidin-biotin amplification method was used for immunohistochemical analyses of NRN1 (anti-NRN1 antibody PRS4101, Sigma-Aldrich) and CXCR4 (anti-CXCR4 antibody GTX22074, GeneTex). Specialized pathologists evaluated the percentage of immune-positive tumor cells. Immunoreactivity of NRN1 and CXCR4 was detected in the cytoplasm of RCC cells and considered high when the cases had more than 10% of the positive carcinoma cells.

All animal experiments were approved by the Animal Care and Use Committee of Saitama Medical University, and carried out in accordance with the Guidelines and Regulations for the Care and Use of Experimental Animals of Saitama Medical University. For the xenograft model, 1.5 × 10⁶ RCC-PDCs were suspended in 150 μl of medium containing 50% Matrigel (BD Biosciences, San Diego, CA) and subcutaneously injected into the flank of male NOG mice (In-Vivo Science, Washington, DC) or male nude mice (BALB/cAJcl-nu/nu) as described previously (13). The RCC-PDC1-derived tumor generated in nude mice was cut into 2-mm cubes and inoculated into the flanks of 8-week-old nude mice. The tumors were measured in two-dimension using micrometer calipers, and the tumor volume was estimated according to the formula: 0.5 × ((smallest diameter)² × (longest diameter)). When the tumor volume exceeded 150 mm³, mice were randomly assigned to one of two groups: siNRN1 #1- and siControl-administered groups (n = 5, each group). siRNA duplexes (5 μg) were mixed with 4 μl GeneSilencer reagent (Gene Therapy System, San Diego, CA), dissolved in 50 μl Dulbecco’s modified Eagle’s medium, and then directly injected into the tumors twice weekly. The results were represented as mean ± SD. Student’s t-test was used for statistical analysis. At the end point of the experiment (3 weeks after siRNA administration), the tumors were dissected from the mice.

Clinical data were analyzed using the log-rank test for the Kaplan-Meier method and Fisher’s exact test for patient characteristics. Experimental data were analyzed using a two-sided Student’s t-test for pairwise comparisons. For multiple comparisons, a two-way ANOVA test was used. Univariate and multivariate analyses were performed using the Cox proportional hazard model. Overall survival curves were generated using the Kaplan-Meier method, and statistical significance was determined using the log-rank test. Statistical computations were carried out using EZR software (version 1.54) (18).

To assess the clinical significance of NRN1 in RCC, we performed immunohistochemical (IHC) analysis of NRN1 in ccRCC tumor samples obtained from distinct patients. We determined that 21 of 100 tumor tissues exhibited high NRN1 IHC staining in the cytoplasm of cancer cells ( Figure 1A ). Kaplan-Meier analysis of the 100 cases showed that the patients with high NRN1 IHC staining showed significant shorter overall survival rate than those with low NRN1 IHC staining (P = 3.1e-4) ( Figure 1B ). In clinicopathological analysis, NRN1 immunoreactivity was significantly correlated with corrected serum calcium (P = 0.011) and C-reactive protein (CRP) (P = 0.024) ( Table 1 ). NRN1 immunoreactivity was also associated with the International Metastatic RCC Database Consortium (IMDC) risk group classification, which is used to stratify patients with metastatic RCC (P = 0.0027). Furthermore, univariate and multivariate analyses of overall survival showed that NRN1 immunoreactivity was an independent prognostic factor for RCC (P = 0.0012 and 0.036, respectively) ( Table 2 ). We also presented representative IHC images of NRN1 with disease stages (I-IV) in neoplastic and paired adjacent non-neoplastic tissues in Supplementary Figure 1 . These NRN1 IHC images and the association study of Table 1 may indicate that there is a positive tendency for the correlation between NRN1 immunoreactivity and stages. Grades were not basically associated with NRN1 immunoreactivity as shown in Table 1 . In addition, high NRN1 immunoreactivities were found in neoplastic areas at stages III and IV compared with their paired corresponding non-neoplastic areas. Of 100 paired analyses, high immunoreactivities were detected in 21 neoplastic areas whereas not in non-neoplastic areas (P < 1.0e-4 by McNemar’s test). In RNA-seq dataset of TCGA RCC cohort (n = 845), patients with high NRN1 mRNA levels had shorter overall survival rate than those with low NRN1 levels (expression level range: 0.1-204.2 FPKM, expression median level: 2.9, expression cut off level 6.6 FPKM, P = 3.4e-4) ( Figure 1C ). NRN1 mRNA levels in stages III and IV RCC tumors significantly higher in those in stages I and II RCC tumors (P = 5.4e-6) ( Supplementary Figure 2A ). Moreover, we examined RCC-subtype dependent NRN1 expressions and prognostic values based on the TCGA datasets for ccRCC (KIRC), papillary RCC (KIRP), and chromophobe RCC (KICH) through the Human Protein Atlas website. As shown in Supplementary Figure 3 , KIRC tissues expressed higher levels of NRN1 compared with KIRP and KICH tissues. In addition, high NRN1 expression was significantly associated with poor survival in patients with KIRC and KIRP. These results suggest that NRN1 may play an oncogenic role in ccRCC and renal papillary cell carcinoma at least.

We next explored genes co-expressed with NRN1 in ccRCC tumors. Because we previously identified that NRN1 was abundantly expressed in patient-derived testicular germ cell tumor spheroid cultures with the enrichment of cancer stemness characteristics, we particularly examined the co-expression of NRN1 with stemness-related genes in RCC. In TCGA PanCancer Atlas RNA-seq dataset of ccRCC cohort (n = 352), we found that C-X-C chemokine receptor type 4 (CXCR4) is a stemness-related gene whose expression is substantially correlated with NRN1 expression (Spearman’s correlation coefficient = 0.263, q-value 9.83e-6). In terms of other prototypic stemness-related genes such as CD44, OCT3/4, and SOX2, Spearman’s correlation coefficients for co-expression were <0.2. Based on the TCGA PanCancer Atlas RNA-seq dataset of ccRCC cohort, we extracted 1,974 and 3,265 genes co-expressed with NRN1 and CXCR4, respectively, with Spearmen’s correlation coefficient ≥0.2 and determined 941 common genes as the overlap of genes co-expressed with both NRN1 and CXCR4 ( Supplementary Figure 4 ). Among the top 10 identified Gene Ontology (GO) Terms using The Database for Annotation, Visualization and Integrated Discovery (DAVID), pathways related to such as extracellular matrix, collagen catabolic process, collagen fibril organization, angiogenesis, and cell adhesion were identified.

We next examined the correlation of CXCR4 expression with RCC patient prognosis in the 100 ccRCC tumor specimens as described above. Prior to immunohistochemical study, the CXCR4 antibody was used in Western blotting where CXCR4 protein levels were decreased and increased by NRN1 silencing and overexpression in RCC-PDC1 cells, respectively ( Supplementary Figures 5A, B ). We determined 25 of 100 tumor tissues showed high CXCR4 IHC staining in the cytoplasm of cancer cells ( Supplementary Figure 5C ). Kaplan-Meier analysis of the 100 cases showed that the patients with high CXCR4 IHC staining showed significant shorter overall survival rate than those with low CXCR4 IHC staining (P = 3.0e-5) ( Supplementary Figure 5D ). In RNA-seq dataset of TCGA RCC cohort (n = 845), patients with high CXCR4 mRNA levels had shorter overall survival rate than those with low CXCR4 levels (expression level range: 2.1-372.0 FPKM, expression median level: 61.8, expression cut off level: 99.5 FPKM, P = 1.1e-5) ( Supplementary Figure 5E ). Similar to NRN1 expression, CXCR4 mRNA levels in stages III and IV RCC tumors significantly higher in those in stages I and II RCC tumors (P = 1.3e-6) ( Supplementary Figure 2B ).

To investigate the roles of NRN1 and CXCR4 in RCC, we established RCC-PDC spheroid cultures from 2 distinct patients with ccRCC, RCC-PDC1 and RCC-PDC2. Transplantation of RCC-PDC1 and RCC-PDC2 into the flanks of immunocompromised male NOG mice could successfully generate PDC-originated xenograft (PDCX) tumors. Hematoxylin and eosin (HE) staining and NRN1 immunostaining showed that RCC-PDC1 spheroid culture and its PDCX tumors recapitulated the morphological and immunohistological features of ccRCC ( Figure 2 ). In the case of RCC-PDC2, the original patient tumor contained a tissue portion of ccRCC cells with eosinophilic cytoplasm and high NRN1/CXCR4 IHC staining, which feature was recapitulated in the established PDC spheroid culture and its PDCX tumor ( Figure 2 ).

We next questioned whether NRN1 and CXCR4 play roles in RCC cell proliferation. We evaluated the effect of NRN1 knockdown on RCC-PDC viability by transfecting NRN1-specific siRNAs siNRN1 #1 and #2. These siRNAs significantly repressed NRN1 expression levels in PDCs compared with the control siRNA (siControl) ( Figures 3A, C ), and impaired the viabilities of RCC-PDC1 and -PDC2 spheroid cultures ( Figures 3B, D ). Conversely, NRN1 overexpression in PDCs significantly upregulated NRN1 mRNA levels compared with transfection with control vector ( Figures 3E, G ) and increased the viabilities of RCC-PDC spheroid cultures ( Figures 3F, H ).

We examined whether NRN1 expression modulates CXCR4 expression in RCC-PDC spheroid cultures. NRN1 silencing ( Figures 3I, J ) and overexpression ( Figures 3K, L ) significantly downregulated and upregulated CXCR4 expression, respectively. Moreover, CXCR4-specific siRNAs (siCXCR4 #1 and #2) significantly repressed CXCR4 expression levels compared with siControl ( Figures 3M, O ) and impaired the viabilities of RCC-PDC spheroid cultures ( Figures 3N, P ). Additionally, we evaluated the correlation between NRN1 and CXCR4 immunoreactivities in our 100 ccRCC cases. Of 25 high CXCR4 immunoreactivity cases, 12 (48%) and 13 (52%) had high and low NRN1 immunoreactivities, respectively, and of 75 low CXCR4 immunoreactivity cases, 9 (12%) and 66 (88%) had high and low NRN1 immunoreactivities, respectively (P = 1.0e-4 by Chi-square test).

As a potential NRN1-activated transcriptional factor, nuclear factor of activated T-cells (NFAT) c4 may activate CXCR4 transcription in ccRCC based on previous literature describing that NRN1 activates NFATc4 via binding to insulin receptor in neuronal cells (19), and that NFATc4-binding on the CXCR4 promoter increases CXCR4 expression in 3D spheroid culture of ovarian cancer cells (20). In this context, it is notable that NFATC4 mRNA expression is substantially correlated with NRN1 expression in TCGA PanCancer Atlas RNA-seq dataset of ccRCC cohort (Spearman’s correlation coefficient = 0.333, q-value 1.38e-10). Moreover, qRT-PCR experiment demonstrated that NRN1 knockdown decreased NFATC4 mRNA levels in both RCC-PDC1 and 2, supporting the notion of intermediate function of NFATC4 between NRN1 and CXCR4 ( Supplementary Figure 6 ).

We further questioned whether NRN1-specific siRNAs can repress in vivo RCC tumor growth. In RCC-PDC1-derived xenograft models in nude mice, siNRN1 #1 or siControl was directly injected into the subcutaneous tumors twice a week when the tumor volume exceeded 150 mm³. siNRN1 injection significantly repressed the growth of RCC-PDC1-derived tumors compared with siControl injection ( Figures 4A, B ), while body weights of mice were not substantially different between the 2 groups ( Figure 4C ). In the dissected tumors of siNRN1 #1-treated mice, NRN1 and CXCR4 immunoreactivities were negligible and NRN1 and CXCR4 mRNA levels were significantly decreased compared with those of siControl-treated mice ( Figures 4D–F ). In summary, NRN1 plays an oncogenic role in RCC in cooperation with CXCR4 ( Figure 4G ).