Differential expression of CLSPN between tumor and normal tissue samples

Firstly, we compared CLSPN expression levels among tumors and matched normal tissues from 33 cancers using the TIMER database (Fig. 1A). Elevated CLSPN expression was observed in BLCA (bladder urothelial carcinoma), BRCA (breast invasive carcinoma), KIRC (kidney renal clear cell carcinoma), KIRP (kidney renal papillary cell carcinoma), LIHC (liver hepatocellular carcinoma), LUAD (lung adenocarcinoma), et al. The CLSPN expression abundances of various tissues in males and females were displayed in Supplementary Figure S1A and B. Overall, no gender difference was observed in the mRNA expression levels of CLSPN (Supplementary Figure S1C). The data from Human Protein Atlas (https://www.proteinatlas.org/) [32, 33] also suggested that CLSPN was highly expressed in multiple cancer samples (Supplementary Figure S1D).

Fig. 1

Differential expression of CLSPN in pan-cancer. A Expression levels of CLSPN in different TCGA tumors from TIMER database, *P < 0.05; **P < 0.01; ***P < 0.001. B Human CLSPN expression levels in different tumor types derived from the GTEx database (*P < 0.05, **P < 0.01, ***P < 0.001). C Expression levels of CLSPN in 23 tumor cell lines based on the CCLE datasets (Kruskal–Wallis test: P = 2.6e − 24)

Subsequently, we combined TCGA and GTEx datasets to further validate the CLSPN expression differences in multiple normal tissues and tumor tissues (Fig. 1B, P < 0.05). CLSPN was significantly upregulated in tumor tissues when compared with normal tissues. Furthermore, the CLSPN expression in diverse tumor cell lines was demonstrated with significant differences based on CCLE datasets (Fig. 1C, Kruskal–Wallis test: P = 2.6e − 24).

The results from the TCGA database were used to explore the correlation between CLSPN expression and clinicopathological stages in various cancers, which revealed the stage-specific expressional changes of CLSPN in some tumor types, such as BRCA, KICH (Kidney Chromophobe), KIRC, LIHC, LUAD, et al. (Supplementary Figure S2A-T).

Prognostic value of CLSPN across cancers

We further conducted survival analysis in different cancer types to investigate the prognostic value of CLSPN. The results of Cox proportional hazards model demonstrated that CLSPN was significantly connected with OS in most cancers (Fig. 2A, P < 0.05). Kaplan–Meier survival curves revealed that high CLSPN expression was obviously related to poor OS in ACC (Adrenocortical carcinoma), KICH, KIRP, LUAD, MESO (mesothelioma), PAAD (pancreatic adenocarcinoma), SKCM (skin cutaneous melanoma), UVM (uveal melanoma) (Fig. 2B-I). The GEO (Gene Expression Omnibus) dataset further validated the influence of CLSPN on the prognosis of tumor patients in clinical cohort (Supplementary Figure S3A-I).

Fig. 2

Association between the CLSPN expression and the OS of cancer patients. A Forest plot displaying the effect of CLSPN expression on OS across 33 types of cancer using Cox regression model. B-I Kaplan–Meier survival curves of the correlations between the CLSPN expression and OS. A red line represents high CLSPN expression, and the blue lines represent low CLSPN expression (P < 0.05 indicated statistical significance)

Furthermore, the result of DSS analysis indicated CLSPN expression was correlated with patients prognosis (Supplementary Figure S4A). Kaplan–Meier survival analysis showed an association between CLSPN and poor prognosis in ACC, KICH, KIRP, LIHC, LUAD and PAAD patients (Supplementary Figure S4B-G, P < 0.05). Cox regression analysis of the DFI revealed that the increased CLSPN expression was a risk factor in KIRP, LIHC, LUAD, and PAAD (Supplementary Figure S4H, P < 0.05). A significant association were presented by KM survival analysis (Supplementary Figure S4I-N, P < 0.05). With regard to PFI, CLSPN expression level was related to the ACC, BLCA, KICH, KIRP, LIHC and LUAD patients' prognosis (Supplementary Figure S5A, P < 0.05). The KM survival analysis results were presented in Supplementary Figure S5B-I.

Genetic alteration analysis

We analyzed various tumor samples to explore the genetic alteration status of CLSPN. As shown in Fig. 3A, the highest alteration frequency of CLSPN (> 8%) was observed in patients with uterine endometrial tumors with “mutation” as the primary type. The types, sites and case numbers of the CLSPN genetic alteration were further presented in Fig. 3B. Moreover, we analyzed the potential correlation between genetic alteration of CLSPN and prognosis of cases with distinct cancer types. The result revealed that altered CLSPN had better prognosis in OS (P = 0.0122) and DSS (P = 0.0372) but not in PFS (P = 0.213) and DFS (P = 0.613), in comparison with patients without CLSPN alterations (Fig. 3C).

Fig. 3

Mutation feature of CLSPN in TCGA tumors obtained from the cBioPortal tool. A Alteration frequency with the mutation type of CLSPN in human pan-cancer. B Mutation sites of CLSPN are displayed. C The correlation between CLSPN mutation status and OS, PFS, DFS and DSS in pan-cancer (P < 0.05 indicated statistical significance)

Relationship between CLSPN expression and the tumor microenvironment

Numerous studies demonstrated that tumor immune microenvironment had an impact on the cancer therapeutic effectiveness. Accordingly, we further investigated the correlation between TME and CLSPN expressions using the ESTIMATE algorithm across 33 cancer types (Fig. 4A). CLSPN was significantly negatively associated with StromalScore, ImmuneScore and ESTIMATEScore in LUAD, LUSC, etc. (Fig. 4B - D). The top 4 tumors were most significantly related to CLSPN expression in StromalScore, ImmuneScore, and ESTIMATEScore were presented in Fig. 4E.

Fig. 4

Association of CLSPN expression with StromalScore, ImmuneScore and ESTIMATEScore in pan-cancer. A The heatmap of the relationship between CLSPN expression and StromalScore, ImmuneScore, ESTIMATEScore and TumorPurity. B Correlation of CLSPN expression with StromalScore. C Correlation of CLSPN expression with ImmuneScore. D Correlation of CLSPN expression with ESTIMATEScore. E Top 4 cancers significantly related to CLSPN expression by ImmuneScore, StromalScore, and ESTIMATEScore, respectively (P < 0.05 indicated statistical significance)

Algorithms of CIBERSORT and XCELL were applied to analyze the correlation of infiltrating immune cells and CLSPN expression in various cancers. In most cancer types, CLSPN expression and number of infiltrating CD8 + T cells showed a negative correlation, as depicted in Fig. 5A and B. Furthermore, CLSPN expression was associated with 47 immune checkpoint genes in LUAD, 37 in PRAD, and 42 in LIHC (Fig. 5C). These results suggested that CLSPN expression alteration may reflect tumor immunity level.

Fig. 5

The correlation between CLSPN expression and immunity, TMB, MSI in different cancer types. A The correlation between CLSPN expression and immune cell infiltration across all tumors in TCGA examined by the CIBERSORT database. B The correlation between CLSPN expression and diverse immune cells infiltration in pan-cancer based on X-Cell database. C The heatmap of the correlation between 47 immune checkpoint genes and CLSPN expression. D Radar map of the correlation between TMB and CLSPN expression. E Radar map of the correlation between MSI and CLSPN expression. *P < 0.05, **P < 0.01, ***P < 0.001

It has been reported that TMB and MSI are biomarkers of immune response of tumors. As shown in Fig. 5D, CLSPN notably correlated with TMB in several tumors, such as KICH, LUAD and READ. CLSPN was positively associated with the MSI in GBM, COAD, BRCA, SKCM and LUAD tissues (Fig. 5E).

Correlation of CLSPN expression with MMR gene and DNA methylation

To investigate whether CLSPN expression could predict tumor progression, we selected five typical MMR genes, and evaluated their association with CLSPN. DNA mismatch repair genes were highly associated with the CLSPN expression in almost all cancer types (Fig. 6A).

Fig. 6

Correlation analysis between CLSPN expression and five MMR genes and four DNA methyltransferases in pan-cancer. A The heatmap of association between CLSPN expression and five MMR genes (MLH1, MSH2, MSH6, EPCAM, PMS2). B The heatmap of correlation between CLSPN expression and the expression of four methyltransferases (DNMT1, DNMT2, DNMT3A, DNMT3B). C Spearman’s correlation analysis of CLSPN expression with four DNA methyltransferases across 33 cancers. *P < 0.05, **P < 0.01, ***P < 0.001

In addition, the relationships between CLSPN and four methyltransferases were also observed in the majority of cancer types (Fig. 6B and C). The correlation between CLSPN expression and CLSPN methylation was presented in Supplementary Figure S6A. We further evaluated the impact of single CpG on LUAD prognosis in MethSurv using TCGA data. As shown in Supplementary Figure S6B, the hyper methylation of CLSPN-body-Island-cg00463507 (HR = 1.433, P = 0.046), CLSPN − TSS200 − Island − cg04263115 (HR = 1.502, P = 0.02), CLSPN − TSS1500 − Island − cg10246273 (HR = 1.435, P = 0.025) indicated poorer OS in TCGA LUAD patients. However, the hyper methylation of CLSPN − 5'UTR;1stExon − Island − cg25109252 (HR = 0.662, P = 0.013) suggested a good OS. The heatmap suggested that cg02106385 of CLSPN displayed the highest level of DNA methylation in LUAD (Supplementary Figure S6C). Kaplan–Meier survival analysis was used to assess the relationship between promoter methylation of CLSPN and prognosis of patient (Supplementary Figure S7A—B).

Correlation between CLSPN expression and stemness score in pan-cancer

The stemness index correlated with tumor pathology and could be used to predict clinical prognosis. We explored whether CLSPN expression was related with stemness score in a variety of cancers by conducting a correlation analysis (Supplementary Figure S8A). The result indicated that CLSPN was positively associated with mRNAsi in ACC, BLCA, LUAD, etc. and mDNAsi in BRCA, CESC, LUAD, etc. (Supplementary Figure S8B). The top 6 tumors most positively correlated with mRNAsi and mDNAsi were presented in Supplementary Figure S8C.

The distribution of CLSPN in LUAD at single-cell level

We evaluated the expression of CLSPN in LUAD patients at the single-cell level using three datasets (NSCLC_EMTAB6149, NSCLC_GSE127465 and NSCLC_GSE143423) from TISCH database. The distribution of CLSPN expression in databases was presented in Supplementary Figure S9A. In NSCLC_EMTAB6149, 12 cell types were found. The result suggested that CLSPN was mainly expressed at the CD8Tex and malignant cells (Supplementary Figure S9B). In NSCLC_GSE127465, CLSPN was mainly distributed in CD8Tex, NK, DC, Mona/Macro, Mast, Neutrophils and malignant cells (Supplementary Figure S9C). In NSCLC_GSE143423, CLSPN was mainly concentrated in CD8T, Mona/Macro and malignant cells (Supplementary Figure S9D). These results indicated that CLSPN may function in tumor immune microenvironment.

CLSPN was upregulated in LUAD tissues and cell lines

Based on the bioinformatics analysis, we further evaluated the role of CLSPN in LUAD. The expression level of CLSPN mRNA and Claspin protein in LUAD tissues was significantly higher than that in adjacent normal tissues (Fig. 7A and B). The expression of CLSPN in Beas2B cell line (Normal pulmonary epithelial cell) and 6 human lung cancer cell lines was detected by RT-qPCR. CLSPN expression was significantly increased in lung cancer cell lines, especially in PC9 and A549 cell lines (Fig. 7C and D). We further explored the correlation between Claspin and immune infiltration. The immunofluorescence staining result suggested that Claspin was remarkably negatively associated with CD8 + T cell infiltration and immune checkpoints including PD-1 and PD-L1 (Fig. 7E), which confirmed the result of bioinformatics analysis above.

Fig. 7

Identification of CLSPN expression in LUAD tissues and cell lines. A CLSPN mRNA expression in LUAD tissues and adjacent normal lung tissues (n = 28). B Claspin protein levels in paired tissues (n = 8). C CLSPN mRNA levels in 6 LUAD cell lines (H1299, Calu-3, SPCA1, HCC827, PC9 and A549) and human normal pulmonary epithelial cell (Beas2B). D Claspin protein levels in 6 LUAD cell lines and human normal pulmonary epithelial cell. E Representative immunofluorescence staining of Claspin, CD8, PD-1, and PD-L1 in high- and low-Claspin group (Scare bar = 50 μm). *P < 0.05, **P < 0.01, ***P < 0.001

Knockdown CLSPN suppressed the LUAD cells proliferation

To further investigate the function of CLSPN in LUAD, we selected A549 cells and PC9 cells which was with highest CLSPN expression for further functional study. Then, we constructed 3 shRNA to knockdown CLSPN in A549 and PC9 cells. RT-qPCR and Western blot analysis was performed to evaluate the knockdown efficiency of CLSPN (Fig. 8A and B). The results of CCK8 assays (Fig. 8C), EDU (Fig. 8D) and colony formation assays (Fig. 8E) indicated that the knockdown of CLSPN inhibited the proliferative activity of LUAD cells. Compared with negative control, knockdown CLSPN resulted in S and G2/M arrest in A549 and PC9 cells (Fig. 8F).

Fig. 8

Knockdown CLSPN significantly inhibited lung cancer cell proliferation and induced S and G2/M arrest in vitro. A RT-qPCR and western blot confirmed the knockdown efficiency of CLSPN in A549 cells. B The knockdown efficiency of CLSPN in PC9 cells. C The effect of CLSPN on cell viability was confirmed by CCK-8 assays. D EdU assays were used to evaluate cell proliferative ability after knockdown CLSPN (Scare bar = 100 μm). E Colony formation in untreated, NC and sh-CLSPN groups. F Cell cycle analyses were performed by flow cytometry. *P < 0.05, **P < 0.01, ***P < 0.001, ns: no significance

Knockdown CLSPN suppressed cell cycle signal both in vitro and in vivo

The functional enrichment analysis in Fig. 9A suggested that CLSPN was remarkably associated with cell cycle in LUAD. We further explored the correlation between CLSPN and Cyclin-dependent kinase (CDK) family and Cyclin family expression. The results demonstrated that CLSPN was positively correlated with the expression of CCNA2, CCNB1, CCNB2, CCNE2, CDK1, CDK2 (Fig. 9B). The expression of CCNA2, CCNB1, CDK1 and CDK2 were significantly decreased at the mRNA and protein level after knockdown CLSPN in A549 and PC9 cells (Fig. 9C and D).

Fig. 9

CLSPN associated with cell cycle signal. A Functional enrichment analysis of CLSPN through GSEA. B The correlation between CLSPN expression and CCNA2, CCNB1, CCNB2, CCNE2, CDK1, CDK2, CDK4 and CDK6 in the GEPIA2.0 database. C RT-qPCR validated the mRNA expression of CCNA2, CCNB1, CCNB2, CCNE2, CDK1, CDK2, CDK4 and CDK6 after knockdown CLSPN. D Knockdown CLSPN significantly inhibited the protein expression of CCNA2, CCNB1, CDK1 and CDK2 in A549 and PC9 cells. *P < 0.05, **P < 0.01, ***P < 0.001, ns: no significance

To explore the function of CLSPN on LUAD growth in vivo, we constructed LUAD xenograft mouse models by subcutaneously injecting A549 and PC9 cells with CLSPN—RNAi or vector lentivirus stably transduction in the right flank of BALB/c mice. The volume and weight of xenografts in sh-CLSPN group was lower than that in NC group (Fig. 10A and B). Knockdown CLSPN significantly reduced the tumor growth rate in A549 and PC9 cells (Fig. 10C). The immunohistochemistry result of subcutaneous tumor indicated that knockdown CLSPN significantly down regulated CCNA2, CCNB1, CDK1, CDK2 and Ki67 expression (Fig. 10D).

Fig. 10

Knockdown CLSPN suppressed LUAD cells growth in vivo. A A549 cells or PC9 cells with different CLSPN expression levels were subcutaneously inoculated in the right flank of BALB/c mice (5 × 106 cells/mouse; n = 5 in each group), and then the tumor volume was calculated using the following formula: V (mm3) = (L × W.2) × 0.5 (L: tumor length, W: width). The mice were sacrificed at day 21 after subcutaneous implantation. The formation of tumor masses was presented. B The histogram displayed the tumor weight in different groups. C Tumor size was measured every 3 days until day 21 (Data were represented as the mean ± SD). D Immunohistochemistry staining for Claspin, CCNA2, CCNB1, CDK1, CDK2 and Ki67 in subcutaneous tumors (Scare bar = 50 μm). ***: P < 0.001

Drug sensitivity analysis of CLSPN

Next, we analyzed the data from CellMiner™ to investigate the IC50 values of anti-cancer drugs and CLSPN expression. We discovered that CLSPN expression was positively associated with the drug sensitivity of PF − 06463922, salinomycin, KU − 55933, Olaparib, AZD − 3463, etc. and negatively associated with sensitivity of Birinapant, 6 − Thioguanine, 6 − THIOGUANIN and Nelarabine (Supplementary Figure S10).

Validation of the affinity of the candidate drugs by molecular docking analysis and CMap analysis

Claspin was discovered as an adaptor or scaffold protein necessary for Chk1 activation in response to DNA replication blocks and stalled replication forks [2]. Previous researches reported that the repetitive phosphopeptide motif in human Claspin was important for Claspin-Chk1 interaction to mediate Claspin function [34, 35]. Considering that Claspin acted as a scaffold protein facilitating the recruitment of CHK1 into larger complexes, we used the structure of CHK1 kinase domain and Claspin phosphopeptide complex (PDB ID:7ako) for molecular docking to screen related FDA-approved drugs. The results indicated that the drugs bound to the complex of Claspin phosphopeptide and CHK1 mainly through strong electrostatic and hydrogen-bonding interactions (Fig. 11A-E). The first five drugs with the lowest binding energy were Darifenacin, Dihydroergotamine, Netupitan, Fosaprepitant and Eltrombopag. The binding energy were − 8.9 kcal/mol, − 8.8 kcal/mol, − 8.8 kcal/mol, − 8.4 kcal/mol, − 8.4 kcal/mol indicating a highly stable binding. We performed CMap analysis to validate the drugs predicted by molecular docking, which indicated that eltrombopag might serve as a potential therapeutic drug in SKCM patients with highly expressed CLSPN, and darifenacin might act as potential therapeutic drug in PRAD, CESC, BRCA, COAD, LUAD, PAAD patients with CLSPN high expression (Fig. 11F).

Fig. 11

Validation of the affinity of the candidate drugs by molecular docking analysis and CMap analysis. A Binding mode of the protein complex and Darifenacin. B Binding mode of the protein complex and Dihydroergotamine. C Binding mode of the protein complex and Netupitant. D Binding mode of the protein complex and Fosaprepitant. E Binding mode of the protein complex and Eltrombopag. (i) The binding sites of drugs in the 3D structure of the protein complex were displayed by PyMOL software. (ii) AutoDockTools showed the interaction between the protein complex and drugs. (iii) 2D interactions of compounds and their targets. The directional bonds between the protein complex and ligands were drawn as dashed lines, and the interacting protein complex residues and ligands were visualized as structural diagrams. Hydrophobic contact was represented by the spline part, highlighting the interaction between the hydrophobic part of the ligand and the label of contacting amino acid. F CMap analysis to validate the drugs predicted by molecular docking in diverse cancers