LMNB1 upregulation was observed in various cancer types and validated in kidney renal clear cell carcinoma
The expression and distribution of LMNB1 in different human tissues and cell types under physiological conditions was explored in the HPA website. LMNB1 was widely expressed in almost all the tissue types while especially enhanced in the lymphoid tissue, including thymus, appendix, lymph node, tonsil and bone marrow (Fig. S1A). Based on single cell RNAseq of human issues and blood cells, LMNB1 showed highest expression in granulocytes, monocytes, spermatids and extravillous trophoblasts (Fig. S1B), whereas low expression specificity in blood cell types (Fig. S1C). Based on the mass spectrometry of human plasma in the publicly available Peptide Atlas, the protein concentration of lamin B1 was estimated as 3.3 µg/L (Fig. S1D).
The TIMER2.0 webserver was applied to study the differential expression of LMNB1 between tumor and adjacent normal tissues across 33 human cancer types in TCGA database. As shown in Fig. 1.A, in the 21 distinct tumor types of which normal tissue data are available, LMNB1 expression in the tumor tissues is upregulated compared to the corresponding adjacent tissues in 19 kinds of tumors, including BLCA (Bladder Urothelial Carcinoma), BRCA (Breast invasive carcinoma), CESC (Cervical squamous cell carcinoma and endocervical adenocarcinoma), CHOL (Cholangio carcinoma), COAD (Colon adenocarcinoma), ESCA (Esophageal carcinoma), GBM (Glioblastoma multiforme), HNSC (Head and Neck squamous cell carcinoma), KIRC (Kidney renal clear cell carcinoma), KIRP (Kidney renal papillary cell carcinoma), LIHC (Liver hepatocellular carcinoma), LUAD (Lung adenocarcinoma), LUSC (Lung squamous cell carcinoma), PCPG (Pheochromocytoma and Paraganglioma), PRAD (Prostate adenocarcinoma), READ (Rectum adenocarcinoma), STAD (Stomach adenocarcinoma), THCA (Thyroid carcinoma), UCEC (Uterine Corpus Endometrial Carcinoma) (Fig. 1A). In contrast, only in KICH (Kidney Chromophobe), LMNB expression in the tumor tissues is downregulated compared to the corresponding control tissues. Besides, LMNB1 expression in the metastasis tissues of SKCM (Skin Cutaneous Melanoma) is much higher than the primary tumor tissues. In addition to transcriptome analysis, we further compared the protein level of lamin B1 in different tumor kinds through UALCAN resource which provides protein expression analysis using data from CPTAC. The total protein expression of lamin B1 was found significantly higher in the tumor tissues of all the 6 tumor kinds (BRCA, COAD, KIRC, LUAD, OV (Ovarian serous cystadenocarcinoma) and UCEC) whose normal tissue data were available in CPTAC, than the corresponding normal tissues (Fig. 1B). Furthermore, we applied GEPIA2 approach to explore the relationship between LMNB1 mRNA expression and the pathological stages across all TCGA tumors. The results indicated that when the pathological stage increased, LMNB1 expression showed a trend of gradual increase in ACC (Adrenocortical carcinoma), KIRC, LUAD, TGCT (Testicular Germ Cell Tumors), and a trend of gradual decrease in OV, while no obvious gradual trend but with stage-specific expression difference in KICH, KIRP, LIHC and SKCM (Fig. 1C).
Since analysis of TCGA and CPTAC datasets showed that in KIRC and LUAD tumor tissues, LMNB1 expression was upregulated both in mRNA and protein levels, as well as increased gradually with elevated pathological stages, we decided to validate the LMNB1 expression status in specimens of radical nephrectomy in our center. Firstly, we detected LMNB1 mRNA expression in 25 pairs of KIRC tumor tissues and adjacent normal tissues by qRT-PCR. Compared to the corresponding non-cancerous tissues, 21 of 25 (84.0%) cancerous tissues showed higher LMNB1 mRNA expression (Fig. 2A, B). Furthermore, western blotting and IHC staining confirmed that lamin B1 protein level in tumor specimens was much higher than the corresponding normal tissues (Fig. 2C-E). Consistent with TCGA analysis, IHC staining showed that KIRC with higher T stage had higher expression of lamin B1 (Fig. 2D, F), and KIRC with advanced Fuhrman nuclear grade (G3 and G4) also showed higher lamin B1 expression than low grade (G1 and G2) (Fig. 2G). Moreover, we explored the correlation between the clinicopathological features and lamin B1 expression in surgical resected specimens from a cohort of 130 consecutive KIRC patients by IHC staining. The results showed that high lamin B1 IHC staining in KIRC patients was positively associated with male gender (p = 0.001), pathological T stage(p < 0.001), distant metastasis (p = 0.008), Fuhrman grade (p = 0.002) and microvascular invasion (p = 0.001) (Table 1). These results indicated that LMNB1 upregulation was generally correlated with tumor initiation and progression.
Table 1
The correlation of lamin B1 expression with clinicopathologic features in 130 KIRC patients.
Clinicopathologic features
|
|
Lamin B1 IHC staining
|
p value
|
Negative or low
|
High
|
Gender
|
Male
|
21 (44.7%)
|
61 (73.5%)
|
0.001
|
Female
|
26 (55.3%)
|
22 (26.5%)
|
|
Age (year)
|
≤ 50
|
16 (34.0%)
|
22 (26.5%)
|
0.364
|
> 50
|
31 (66.0%)
|
61 (73.5%)
|
|
Pathological T stage
|
T1
|
37 (78.7%)
|
35 (42.2%)
|
< 0.001
|
T2
|
8 (17.0%)
|
16 (19.3%)
|
|
T3+T4
|
2 (4.3%)
|
32 (38.5%)
|
|
Lymph node involvement
|
N0
|
46 (97.9%)
|
76 (91.6%)
|
0.257
|
N1
|
1 (2.1%)
|
7 (8.4%)
|
|
Distant metastasis
|
M0
|
45 (95.7%)
|
65 (78.3%)
|
0.008
|
M1
|
2 (4.3%)
|
18 (21.7%)
|
|
Fuhrman grade
|
1
|
12 (25.5%)
|
10 (12.0%)
|
0.002
|
2
|
27 (57.4%)
|
34 (41.0%)
|
|
3+4
|
8 (17.1%)
|
39 (47.0%)
|
|
Recurrence
|
No
|
45 (95.7%)
|
76 (91.6%)
|
0.487
|
Yes
|
2 (4.3%)
|
7 (8.4%)
|
|
Microvascular invasion
|
No
|
44 (93.6%)
|
57 (68.7%)
|
0.001
|
Yes
|
3 (6.4%)
|
26 (31.3%)
|
|
Statistical significance was calculated using the Chi-square test or the Fisher exact test. |
LMNB1 upregulation was universally correlated with poor survival in human cancers
Since high expression of LMNB1 was found in the overwhelming majority of cancer types and correlated with cancer aggressiveness, we wonder whether it affects the prognosis of human cancers. The “Survival Analysis” module of GEPIA2 tool was applied to compare the OS and DFS based on the expression status of LMNB1 in TCGA datasets. As shown in Fig. 3A, LMNB1 expression higher than median was significantly associated with poor prognosis of OS for cancers of ACC (p < 0.0001), LGG (Brain Lower Grade Glioma) (p = 0.0005), PAAD (p = 0.0076), KIRP (p = 0.0021), LIHC (p = 0.0034), MESO (Mesothelioma) (p = 0.047), SARC (Sarcoma) (p = 0.01), while with favorable prognosis of OS for cancers of LUSC (p = 0.038), THYM (Thymoma) (p = 0.009) (Fig. 3A). On the other hand, DFS analysis revealed that highly expressed LMNB1 was notably correlated with poor prognosis for ACC (p < 0.0001), KIRP (p = 0.0001), LIHC (p < 0.0001), PRAD (p = 0.013), UVM (Uveal Melanoma) (p = 0.012), ESCA (p = 0.04), LGG (p = 0.025), PAAD (p = 0.035), SARC (p = 0.0004), while not correlated with favorable prognosis for any human cancer type (Fig, 3B).
LMNB1 was identified as a biomarker of CD4+ Type 2 T helper cell infiltration in human cancers
For the past few years, immune cells infiltrated in the tumor microenvironment (TME) has been found playing key roles in the tumorigenesis and tumor progression (29, 30). Previous study has demonstrated that a gene hub including LMNB1 in KIRC was positively correlated with multiple kinds of tumor infiltrating lymphocytes, such as activated CD4+ T cells, CD8+ T cells, regulatory T cells and follicular helper T cells, but negatively correlated with resting mast cells, resting NK cells and activated NK cells (31). Herein, we aimed to provide a comprehensive research on the potential relationship between LMNB1 expression and immune infiltration levels across all TCGA cancers. Firstly, Sangerbox tools helped us identified that activated CD4+ T cells and type 2 T helper cells were the main immune pathways which were significantly correlated with high expression of LMNB1 in most of human cancer types, especially ACC, KIRP, KIRC, LIHC, LGG, PRAD, THCA and UVM (Fig. 4A). Then we focus our attention on the details of CD4+ T cell infiltration levels. With the aid of TIMER2.0 resource, we employed the EPIC, TIMER, QUANTISEQ, XCELL, CIBERSORT, CIBERSORT-ABS methods to analyze the correlation between the infiltration levels of CD4+ T cell subtypes and the mRNA expression of LMNB1 across different TCGA tumor kinds. We found that all the TCGA cancers excluding UCS (Uterine Carcinosarcoma) showed statistical positive correlation between high LMNB1 expression and the immune infiltration of CD4+ Th2 cells based on XCELL algorithm (Fig. 4B). In contrast, the infiltration level of CD4+ central memory T cells and CD4+ effector memory T cells was found negatively correlated high LMNB1 expression in more than half of the TCGA cancer kinds (Fig. 4B). The scatter plots of the cancer types which occupied the top 10 purity-adjusted Spearman’s rho coefficient were showed (Fig. 4C, Fig. S2). Since CD4+ Th2 cells and Th2 cytokines such as interleukin-4 (IL-4), IL-5, IL-6, IL10 and IL13 were thought to be associated with immunosuppressive contexture and tumor-promoting effects (32–37), we believed that LMNB1 could be considered as a biomarker of immunosuppressive microenvironment.
Pathways of cell cycle and nuclear division were involved in the effects of LMNB1 on tumor pathogenesis
In order to further illustrate the molecular mechanism of LMNB1 in tumor initiation and progression, we screened out LMNB1-related genes for subsequent pathway enrichment analysis. By using GEPIA2 tool, we acquired the top 100 genes which had similar expression patterns with LMNB1 in the combined data of all TCGA cancer types, and the most 6 correlated genes are TMPO (thymopoietin) (R = 0.81), KIF11 (kinesin family member 11) (R = 0.79), NUSAP1 (nucleolar and spindle associated protein 1) (R = 0.79), KIF15 (kinesin family member 11) (R = 0.78), MCM6 (minichromosome maintenance complex component 6) (R = 0.77), PLK4 (R = 0.77) (Fig. 5A). The corresponding heatmap also verified that the above top 6 related genes showed statistical positive correlation with LMNB1 in almost every cancer type (Fig. S3). On the other hand, with STRING website we obtained another top 50 genes which encode the proteins that had been experimentally proved to physically bind to lamin B1 (Fig. 5B), such as RPA1 (replication protein A1) (interaction score = 0.873) and RPA3 (replication protein A3) (interaction score = 0.835). The venn diagram of the above two gene sets showed an intersection consisting of two genes, namely, TMPO and ZWINT (ZW10 interacting kinetochore protein) (Fig. 5C). Then, we carried out KEGG and GO enrichment analysis with the union of the above two gene sets. As shown in Fig. 5D, KEGG analysis revealed that “cell cycle” and “DNA replication” pathways appeared to play important roles in the influence of LMNB1 on the tumorigenesis and development (Fig. 5D). GO term analysis showed that “nuclear division”, “organelle fission”, “mitotic nuclear division” and “chromosome segregation” in biological process (Fig. 5E, F), “chromosomal region” and “spindle” in cellular component (Fig. S4A, C), “tubulin binding” and “microtubule binding” in molecular function (Fig. S4B, D) were most involved in the effects of LMNB1 on human tumors. Overall, we hold the opinion that LMNB1 played an important role in tumor growth and cell mitosis.
LMNB1 was required for DNA homologous recombination repair and contributed to PARPi resistance
Interestingly, we found that several genes in the gene aggregate which encodes lamin B1-binding proteins or shows similar expression pattern with LMNB1 were also associated with DNA homologous recombination repair (HRR), such as BRCA1, CHEK1, RAD51, RAD54L and BRIP1 (Fig. 5F, Fig. S4C, D). To further explore the detailed role of LMNB1 on the DNA HRR, we used GEPIA2 to analyze the correlation between the expression of LMNB1 and a self-defined gene set named as PROfound signature, which consists of 15 HRR-associated genes prespecified by Johann de Bono, et al. in their phase 3 PROfound clinical trials (38). The heatmap showed that LMNB1 expression had a strong positive correlation with the PROfound signature in the vast majority of TCGA cancer types, and the corresponding scatter plots of the cancer types whose Pearson correlation coefficient was not less than 0.80 were displayed (Fig. 6A). In addition, TIMER2.0 assisted us to show the detail of the correlation between LMNB1 and every single gene in the PROfound signature across all TCGA tumor kinds (Fig. 6B).
PROfound trials revealed that metastatic castration-resistant prostate cancer (CRPC) patients with at least one alteration in the PROfound signature genes would benefit for progression-free survival from olaparib (a selective PARP inhibitor) treatment (38). Since LMNB1 was revealed to be closely related to the PROfound signature, we speculated that LMNB1 might influence the treatment effects of olaparib. Firstly, in the PRAD cohort of TCGA database, LMNB1 appeared to be strongly correlated with the PROfound signature (R = 0.75, p = 5.6E-90) (Fig. 6C). Then we stably overexpressed LMNB1 gene in a CRPC cell line PC3, and qRT-PCR analysis showed that forced expression of LMNB1 increased the mRNA level of BRCA1, BRCA2, CHEK1, CHEK2 and ATM (Fig. 6D). To further validate the relationship between the expression of LMNB1 and HRR genes in PRAD, lamin B1 and BRCA1, which was selected as a representative of HRR proteins were examined in 81 paraffin-embedded prostate cancer tissues by IHC staining (Fig. 6E). As expected, we found that the tumor epithelial areas in the PRAD specimens displaying strong staining of lamin B1 also had heavy signals of BRCA1 (Fig. 6E). Statistically, a positive relationship between lamin B1 and BRCA1 was observed (Spearman r = 0.5407, p < 0.0001) (Fig. 6F). In addition, Pearson correlation analysis also indicated that LMNB1 mRNA level was significantly associated with BRCA1 mRNA expression in a panel of prostate cell lines (Pearson r = 0.8977, p = 0.0025) (Fig. 6G).
In order to investigate the effects of LMNB1 on the PARPi therapy, we stably knocked know LMNB1 in another PRAD cell line 22Rv1, by infecting with two lentiviruses containing specific short hairpin RNAs (shRNAs) targeting different regions of LMNB1 gene (Fig. 7A, Table S2). The cell viability assay showed that silencing of LMNB1 in 22Rv1 cells resulted in a prominent decreased IC50 of olaparib (97.1 µM for shLMNB1#1 and 108.8 µM for shLMNB1#2 versus 177.6 µM for shNC) (Fig. 7B). Western blotting also demonstrated that LMNB1 knockdown caused more serious cell apoptosis characterized by cleaved PARP and cleaved caspase3 in the condition of olaparib incubation (Fig. 7C). Finally, GSEA of PRAD cohort in TCGA database showed that the pathways of “DNA repair” in hallmark gene set (Fig. 7D, E, Table S3), “homologous recombination”, “mismatch repair”, “nucleotide excision repair”, “base excision repair” in KEGG gene set (Fig. S5A, B, Table S4), and “double strand break repair” in GO biological process gene set (Fig. S5C, D, Table S5) were positively associated with high LMNB1 expression, indicating that LMNB1 did play a crucial role on DNA repair and PARPi therapy.