Omics- and Pharmacogenomic Evidence for the Prognostic, Regulatory, and Immune-Related Roles of PBK in a Pan-Cancer Cohort

PDZ-binding kinase (PBK) is known to regulate tumor progression in some cancer types. However, its relationship to immune cell infiltration and prognosis in different cancers is unclear. This was investigated in the present study by analyzing data from TCGA, GEO, GETx, TIMER, CPTAC, GEPIA2, cBioPortal, GSCALite, PROGNOSCAN, PharmacoDB, STRING, and ENCORI databases. PBK was overexpressed in most tumors including adenocortical carcinoma (hazard ratio [HR] = 2.178, p < 0.001), kidney renal clear cell carcinoma (KIRC; HR = 1.907, p < 0.001), kidney renal papillary cell carcinoma (HR = 3.024, p < 0.001), and lung adenocarcinoma (HR = 1.255, p < 0.001), in which it was associated with poor overall survival and advanced pathologic stage. PBK methylation level was a prognostic marker in thyroid carcinoma (THCA). PBK expression was positively correlated with the levels of BIRC5, CCNB1, CDC20, CDK1, DLGAP5, MAD2L1, MELK, PLK1, TOP2A, and TTK in 32 tumor types; and with the levels of the transcription factors E2F1 and MYC, which regulate apoptosis, the cell cycle, cell proliferation and invasion, tumorigenesis, and metastasis. It was also negatively regulated by the microRNAs hsa-miR-101-5p, hsa-miR-145-5p, and hsa-miR-5694. PBK expression in KIRC, liver hepatocellular carcinoma, THCA, and thymoma was positively correlated with the infiltration of immune cells including B cells, CD4+T cells, CD8+ T cells, macrophages, monocytes, and neutrophils. The results of the functional enrichment analysis suggested that PBK and related genes contribute to tumor development via cell cycle regulation. We also identified 20 drugs that potentially inhibit PBK expression. Thus, PBK is associated with survival outcome in a variety of cancers and may promote tumor development and progression by increasing immune cell infiltration into the tumor microenvironment. These findings indicate that PBK is a potential therapeutic target and has prognostic value in cancer treatment.

Previous studies on PBK were limited to a single cancer type and lacked a comparative analysis of multiple tumors. The aim of the present study was to examine the role of PBK in different cancers and elucidate the molecular mechanisms and functions of PBK and its interacting molecules in carcinogenesis. We also analyzed the relationships between PBK expression, mutation and methylation status, clinical outcome, and tumor immune infiltration in cancers, and screened drugs that can potentially inhibit PBK. Our results demonstrate that PBK plays an important role in carcinogenesis and is a potential therapeutic target in cancer treatment.

Gene Expression Analysis
RNA expression and clinical data in The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) databases were downloaded from the University of California at Santa Cruz (UCSC) Xena database (https://xenabrowser.net/ datapages/). The R "limma" package was used to analyze the mRNA expression of PBK in the pan-cancer cohort.
The UALCAN portal (http://ualcan.path.uab.edu/analysisprot.html) was used to access the Clinical Proteome Oncology Analysis Alliance (CPTAC) dataset (Chen F. et al., 2019), which includes breast invasive carcinoma (BRCA), colon adenocarcinoma (COAD), OV, clear cell renal cell carcinoma, endometrial carcinoma (UCEC), lung adenocarcinoma (LUAD), and childhood brain cancer. We used these data to compare total protein expression between primary tumors and normal tissue by entering "PBK" into the search field.
We used the Gene Expression Profiling Interactive Analysis v2 (GEPIA2; http://gepia2.cancer-pku.cn/#index) "Pathological Stage Plot" module to obtain maps of PBK expression in different pathologic stages (stage I-IV) of TCGA tumors. Log2 (transcripts per million+1) expression data and an adjusted p value < 0.05 were used as cutoffs to generate the stage diagram.

Alterations in Gene Expression
We investigated the inheritance of PBK gene modifications in different cancers using the cBioPortal portal (https://www. cbioportal.org/) (Gao et al., 2013) through "TCGA Pan-Cancer Atlas Studies" in the "Quick select" section with "PBK" as query. The results of genetic changes in TCGA tumor samples were displayed in the "Cancer Types Summary" module, including the type of mutation and copy number alterations. The comparison module also returned disease specificity and overall survival (OS), progression-free survival (PFS), and disease-free survival (DFS) in TCGA cancer samples with or without PBK gene mutations. A log-rank p value < 0.05 was used as the cutoff.
Gene Set Cancer Analysis (GSCALite; http://bioinfo.life. hust.edu.cn/web/GSCALite/) is a bioinformatics platform that allows methylation, drug sensitivity, genetic, gene variation, and survival analyses of cancer datasets (Liu et al., 2018). We used the GSCALite platform to analyze the differential methylation of PBK and TFs between tumors and paired normal tissues in different types of cancer in TCGA. After logging onto the GSCALite server, we used the "TCGA Cancer-methylation" module to select 32 TCGA cancer types. In the final map generated by GSCALite, only 14 cancer types showed methylation profiles of PBK and related TFs. Using the same platform, we analyzed the relationship between methylation and the expression of PBK and related TFs in different cancers.

Survival Analysis
The PrognoScan database (http://dna00.bio.kyutech.ac.jp/ PrognoScan/index.html) allows meta-analysis of the prognostic value of genes by comparing the relationship between gene expression and variables of interest in a large number of published reports (Mizuno et al., 2009). We used this platform to evaluate the relationship between PBK expression and patient prognosis in different cancer cohorts. Survival curves were generated with PrognoScan and the "survival" package in R from TCGA data. The survival results are shown as hazard ratio (HR) with 95% confidence interval (CI) and log-rank p value.

Correlation Analysis
Pearson correlation analysis was performed to evaluate the correlation between PBK and immune checkpoints, tumor mutation burden (TMB), microsatellite instability (MSI), DNA methylation, and mismatch repair (MMR). The "heatmap" package in R was used to display the results as a heatmap.

Immune Cell Infiltration Into the Tumor Microenvironment
The Tumor IMmune Estimation Resource (TIMER; http://timer. cistrome.org/)  web server is a comprehensive resource for systematic analysis of immune cell infiltration in different cancer types. We used the immune gene module of TIMER to assess the correlation between PBK expression in six immune cell types and their infiltration in 32 TCGA tumors. We also used the Tumor-Immune System Interactions database (TISIDB) (http://cis.hku.hk/TISIDB/) to evaluate whether there was a significant difference in PBK expression between responders and non-responders to immunotherapies (e.g., anti-programmed death 1 [PD1] or anti-programmed death ligand 1 [PD-L1]).

Enrichment Analysis of PBK-Related Genes
We used the Search Tool for the Retrieval of Interacting Genes/ Proteins (STRING) website (https://string-db.org/) with "PBK" and organism Homo sapiens to construct a network of the interaction between PBK and related genes. We then used the GSCALite platform to analyze the function of PBK-related genes by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses. Using the ggplot2 package in R v3.6.2, a point map was generated to visualize the results (p < 0.05).

Drugs Targeting PBK
We downloaded data on PBK-related drugs from PharmacoDB (https://www.pharmacodb.pmgenomics.ca/) (Smirnov et al., 2018), with p < 0.01 and a standardized coefficient of −1 to 1 as thresholds. The "heatmap" package in R was used to display the results as a heatmap.

PBK Mutation Profile in Different Tissues
We used cBioPortal to determine the mutation frequency of PBK in the TCGA database (10,967 samples in 32 studies). We found that PBK had a relatively high mutation frequency in PRAD, OV, and UCEC (>6%) (Figures 2A,B). A total of 40 mutations (including 31 missense, 7 truncation, 1 deletion, and 1 fusion) were detected between amino acid 0 and 322 ( Figure 2C). PBK gene mutations were unrelated to survival and prognosis of cancers in TCGA. Using cBioPortal, we determined that there was no significant difference in OS (p 0.163), DFS (p 0.883), disease-specific survival (DSS; p 0.173), or PFS (p 0.106) between samples with vs those without PBK mutation ( Figure 2D).

TFs and miRNAs Regulating PBK
Given the importance of PBK in cancer, we explored TFs and miRNAs that regulate PBK expression. We identified 153 transcriptional regulators of PBK in 4 databases including ChEA, NCODE, JASPAR, and TRANSFAC. After screening for those that were represented at least twice in these databases, 11 TFs were ultimately obtained ( Figure 4A). We analyzed the expression of these TFs in different cancers using the GSCALite platform. The 11 TFs were highly expressed in most cancers, with E2F1 showing the most significant upregulation ( Figure 4B). The heat map indicated that the 11 TFs in 32 cancers were associated with altered PBK expression ( Figure 4C): E2F1 in THYM (cor 0.916), tripartite motif-containing 28 (TRIM28) in THYM (cor 0.69), MYC in uveal melanoma (UVM) (cor 0.577), and TATA-binding protein (TBP) in LIHC (cor 0.497) were positively correlated with PBK level whereas signal transducer and activator of transcription 5A (STAT5A) and Spi-1 protooncogene (SPI1) showed negative correlations. E2F1 regulates the cell cycle, DNA repair, apoptosis, and cell proliferation (Fouad et al., 2020). TRIM28 is involved in the regulation of target gene transcription, response to DNA damage, downregulation of p53 activity, stimulation of epithelial-mesenchymal transition (EMT), autophagy induction, and regulation of reverse transcription-mediated transposition (Czerwińska et al., 2017). MYC promotes tumorigenesis by regulating cell proliferation, cell cycle, and genomic instability (Casey et al., 2017). Thus, PBK may contribute to tumorigenesis through interaction with these factors.

Changes in the Methylation State of PBK Influence Patient Prognosis
Alterations in DNA methylation state are related to gene expression changes in cancer (Masser et al., 2018). We used the GSCALite platform to analyze the methylation profiles of PBK and related TFs in various TCGA cancers. We first compared the methylation status of tumor and normal tissues in 14 cancer types and found that PBK methylation was decreased in THCA, LIHC, and LUAD ( Figure 5A). We also examined the correlation between PBK methylation and expression in 32 cancer types and found that the expression levels of PBK and its TFs were negatively correlated with PBK methylation, with positive correlations observed only in a few cases ( Figure 5B). Finally, we examined the association between PBK methylation and OS in 22 cancer types and found that in LGG, THCA, GBM, and PCPG, a higher methylation level was associated with a lower risk of death, whereas a positive association was observed between PBK methylation and mortality in UCEC ( Figure 5C).

High PBK Expression Is Correlated With Immune Cell Infiltration in Cancers
To determine whether PBK is involved in immune cell invasion into the tumor microenvironment, we analyzed the correlation between PBK expression and the infiltration of six immune cell types in 32 cancers in the TIMR database. PBK expression level was positively correlated with the degree of immune cell infiltration in KIRC, LIHC, THCA, and THYM ( Figures 6A-D). The correlations between PBK expression and immune cell infiltration in other cancers are shown in Supplementary  Figure 1. Our results suggest that in some cancers such as KIRC and LIHC, PBK overexpression can affect patient prognosis by increasing immune cell infiltration.

Relationship Between Immunotherapy, Immune Checkpoints, and PBK
Immune checkpoint molecules maintain immune system activation within a normal range; abnormal expression of these molecules is closely related to the occurrence and development of some tumors. Immune checkpoint inhibitors (ICIs) are a treatment option for certain malignancies. Here we analyzed the correlation between PBK expression and immune checkpoint molecules in different cancers. In CHOL, HNSC, KICH, KIRC, LIHC, and THCA, PBK expression was positively correlated with the expression of cluster of differentiation 86 (CD86), T cell immunoreceptor with Ig and ITIM domains (TIGIT), indoleamine 2,3-dioxygenase 1 (IDO1), cytotoxic T-lymphocyte-associated protein 4 (CTLA4), inducible T cell costimulator (ICOS), lymphocyte-activating 3 (LAG3), and tumor necrosis factor superfamily member 4 (TNFSF4). On the contrary, in GBM, LUSC, and THCA, PBK expression was negatively correlated with the expression of immune checkpoint molecules such as CD86, V-set immunoregulatory receptor (VSIR), hepatitis A virus cellular receptor 2 (HAVCR2), and neuropilin 1 (NRP1) ( Figure 7A). These results indicate that PBK plays an important role in tumor immune regulation and may thus impact patient survival. Although PBK expression was significant in urothelial cancer, there was no significant difference in expression level between immunotherapy responders and nonresponders (Supplementary Table 1).

Enrichment of PBK-Binding Protein Networks
We explored the functional enrichment of PBK in cancer by mapping the protein-protein interaction network of PBK with the STRING database. Ten genes including baculoviral IAP repeat-containing 5 (BIRC5), cyclin B1 (CCNB1), cell division cycle protein 20 (CDC20), cyclindependent kinase 1 (CDK1), DLG-associated protein 5 (DLGAP5), mitotic arrest deficient 2-like 1 (MAD2L1), maternal embryonic leucine zipper kinase (MELK), pololike kinase 1 (PLK1), DNA topoisomerase II alpha (TOP2A), TTK protein kinase (TTK), and PDZ-binding kinase (PBK) formed an interaction network with PBK ( Figure 8A). We analyzed the association between these 10 genes and PBK in 32 TCGA cancers and found that the expression of all of the genes in all of the cancers was positively correlated with PBK expression ( Figure 8B). The results of the GO functional enrichment analysis showed that PBK is involved in the regulation of cell cycle, serine/ threonine kinase activity, and cell division ( Figure 8C). The KEGG pathway analysis results revealed enrichment in "cell cycle," "p53 signal pathway," and "cell senescence" ( Figure 8D). Thus, PBK has potential functions in cell cycle regulation.

Drug and Pathway Activity Analysis
We examined the relationship between the expression of PBK and its interaction partners and the activation or inhibition of cancer-related signaling pathways using the GSCALite platform. Consistent with results of the functional enrichment analysis, BIRC5, CCNB1, CDC20, CDK1, DLGAP5, MAD2L1, MELK, PLK1, TOP2A, TTK, and PBK were related to the activation of apoptosis, cell cycle, and DNA damage response; and BIRC5, CCNB1, CDC20, CDK1, DLGAP5, MAD2L1, MELK, PLK1, TOP2A, TTK, and PBK were related to the inhibition of hormone receptors and RAS/ MAPK pathways. Additionally, PBK was related to the induction of EMT as well as activation of TSC/mammalian target of rapamycin (mTOR) and inhibition of PI3K/AKT and receptor tyrosine kinase (RTK) signaling pathways ( Figures 9A,B).
We searched the PharmacoDB database for drugs related to PBK with a cutoff of p < 0.01. Drugs without targets molecules or pathways were excluded based on standard coefficients >0.1 and <−0.1. PBK was found to be resistant to 31 drugs and sensitive to 20 including those targeting the cell cycle such as paclitaxel, docetaxel, doxorubicin, and irinotecan; and apoptosis regulators such as brivanib, TW37, and elesclomol ( Figure 9C). The mTOR signaling pathway regulates cell metabolism, growth, proliferation, survival, and autophagy in many cancers; mTOR inhibitors suppress inflammation, proliferation, autophagy, and apoptosis and are used in cancer treatment (Chen and Zhou, 2020). PBK was found to be sensitive to rapamycin and temsirolimus, which target the mTOR signaling pathway.
The TCGA and GTEx data showed that PBK was highly expressed in ACC, BLCA, BRCA, CESC, CHOL, COAD, ESCA, GBM, HNSC, KICH, KIRC, KIRP, LGG, LIHC, LUAD, LUSC, OV, RAAD, PRAD, SKCM, STAD, THCA, UCEC, and UCS and downregulated in LAML compared to normal tissue. The Kaplan-Meier survival analysis revealed that high expression of PBK was a risk factor in ACC, DLBC, KIRC, KIRP, LGG, LIHC, LUAD, MESO, and PCPG and a protective factor in OV and THYM. Additionally, we found that PBK expression was positively correlated with cancer stage in ACC, KICH, KIRC, KIRP, and LUAD. PBK expression level was shown to be significantly associated with the progression of ACC and NSCLC (Lei et al., 2013;Kar et al., 2019). Thus, PBK is a prognostic biomarker in many cancers, especially in ACC and LUAD as well as in KIRC and KIRP, which has not been previously reported.
The expression of TFs related to PBK showed the highest positive correlations with UVM and THYM, while PBK-related miRNAs had the strongest inhibitory effects in UCS, TGCT, and MESO. These factors are involved in the regulation of the cell cycle, apoptosis, and cell proliferation and are linked to genome instability. For example, MYC is an oncogene in many cancers and plays an important role in regulating metabolic pathways, especially glutamine and glutamine breakdown (Tambay et al., 2021). TFs regulating PBK expression may thus influence its function in tumorigenesis.
PBK methylation was decreased in THCA, LIHC, and LUAD, which was negatively correlated with the expression level of PBK. Additionally, PBK methylation level was correlated with OS in LGG, THCA, GBM, PCPG, and UCEC cancer. The correlation was especially strong for THCA; therefore, while PBK expression cannot be used as a prognostic marker for THCA, we show here for the first time that a change in its methylation status may predict survival outcome.
There have been few studies on the function of PBK in the tumor immune microenvironment. We observed a negative correlation between PBK expression and the infiltration of immune cells including B cells, CD4 + T cells, CD8 + T cells, and macrophages in most tumors, but this association was especially strong in KIRC, LIHC, THCA, and THYM. Based on this observation combined with the survival data, we speculate that PBK affects the prognosis of cancers-particularly KIRC and LIHC-by enhancing immune cell infiltration.
ICIs are a type of immunotherapy with demonstrated efficacy and safety in multiple cancers (He and Xu, 2020). Patients with high TMB or MSI are more likely to experience long-term survival benefits from immunotherapy (Samstein et al., 2019;Zhu et al., 2021). To date there have been no studies on the relationship between TMB and MSI and PBK expression in cancers. Our results showed that PBK expression in various types of tumor was related to TMB and MSI, although additional studies are needed in order to determine the significance of this observation in terms of response to immunotherapy.
Functional enrichment analysis showed that PBK and related genes were functionally enriched in cell cycle regulation. Previous studies have shown that PBK binds to the CDK1/cyclin B1 complex on the mitotic spindle and induces its phosphorylation, leading to cell proliferation and cell cycle progression (Côté et al., 2002;Park et al., 2010). Dysregulation of the cell cycle results in abnormal cell proliferation, inhibition of apoptosis, and increased invasion and metastasis (Zaretsky et al., 2016). Thus, PBK may play an oncogenic role by interfering with cell cycle regulation. This finding can guide the development of targeted drugs for cancer treatment.
BIRC5, CCNB1, CDC20, CDK1, DLGAP5, MAD2L1, MELK, PLK1, TOP2A, TTK, and PBK formed an interaction network that was associated with activation of cell apoptosis and the DNA damage response and inhibition of RAS/MAPK signaling. PBK was also related to the induction of EMT, activation of TSC/ mTOR signaling, and inhibition of the PI3K/AKT and RTK pathways. Thus, PBK and PBK-related genes promote tumorigenesis through a variety of mechanisms.
The low solubility and toxicity of the PBK inhibitors HI-TOPK-032 and OTS964TOPK limit their clinical application (Matsuo et al., 2014;Ikeda et al., 2016). As such, new PBK inhibitors for cancer treatment are needed. In this study, we screened 31 drugs that potentially target PBK and identified several candidates including linifanib, which is used in the treatment of LIHC, COAD, STAD, NSCLC, and HNSC (Hsu et al., 2014;Chen et al., 2016;Horinouchi, 2016;Chan et al., 2017;Han et al., 2021); neratinib, which is used in BRCA (Saura et al., 2021); erlotinib and gefitinib, which are used in NSCLC (Yang et al., 2017); and BMS-536924, which is used in GBM and ESCA (Adachi et al., 2014;Zhou, 2015). The anticancer effect of these drugs may be related to PBK inhibition.

CONCLUSION
We carried out a comprehensive evaluation of PBK to investigate its potential roles in promoting cancer and as a prognostic indicator through bioinformatics analyses. We identified 4 TFs and 3 miRNAs that were related to PBK upregulation and downregulation, respectively, in cancer. We also found that PBK expression is related to cancer immunity: in some cancers, PBK may affect tumor progression by increasing immune cell infiltration into the tumor microenvironment, making it a potential target for cancer immunotherapy. Functional enrichment analysis showed that PBK and its related genes regulate tumor-related signaling pathways. Finally, we found that PBK is associated with sensitivity and resistance to various anticancer drugs and is related to drug-targeted genes in tumor cells. These findings indicate that evaluating PBK expression levels may have diagnostic utility in cancer and can guide treatment decisions, although experimental studies are needed to confirm this possibility.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

ETHICS STATEMENT
Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

AUTHOR CONTRIBUTIONS
YL designed the experiment and collected and analyzed the data. YL and XJ wrote the manuscript. GP provided technical guidance. CS and GP revised the manuscript. All authors have read and approved the final manuscript.

FUNDING
This study was supported by the National Natural Science Foundation of China Youth Fund Project (no. 81801908) and Hunan Provincial Natural Science Committee Project (no. 2021JJ70074).