Abstract

   Hepatocellular carcinoma (HCC) is an aggressive and chemoresistant
   cancer type. The development of novel therapeutic strategies is still
   urgently needed. Personalized or precision medicine is a new trend in
   cancer therapy, which treats cancer patients with specific genetic
   alterations. In this study, a gene signature was identified from the
   transcriptome of HCC patients, which was correlated with the patients’
   poorer prognoses. This gene signature is functionally related to
   mitotic cell cycle regulation, and its higher or lower expression is
   linked to the mutation in tumor protein p53 (TP53) or catenin beta 1
   (CTNNB1), respectively. Gene–drug association analysis indicated that
   the taxanes, such as the clinically approved anticancer drug
   paclitaxel, are potential drugs targeting this mitotic gene signature.
   Accordingly, HCC cell lines harboring mutant TP53 or wild-type CTNNB1
   genes are more sensitive to paclitaxel treatment. Therefore, our
   results imply that HCC patients with mutant TP53 or wild-type CTNNB1
   genes may benefit from the paclitaxel therapy.

   Keywords: bioinformatics, hepatocellular carcinoma, precision medicine,
   taxanes

1. Introduction

   Primary liver cancer is still the sixth most common cancer type and the
   third leading cause of cancer-related death in the world [[34]1]. Most
   (75–85%) of primary liver cancer cases involve hepatocellular carcinoma
   (HCC), whose major risk factors include chronic hepatitis B or C virus
   (HBV or HCV) infection, aflatoxin contamination in foods, excess body
   weight, heavy alcohol intake, smoking, and type 2 diabetes [[35]1]. HCC
   development involves a complex, multi-step histological process of
   normal hepatocyte malignant transformation involving various genetic
   and epigenetic alterations [[36]2]. The most frequent genetic
   alterations include mutations in telomerase reverse transcriptase
   (TERT) promoter, tumor protein p53 (TP53), and catenin beta 1 (CTNNB1)
   genes, as well as copy number variations and aberrations in DNA
   methylation [[37]3]. TERT promoter mutations occur in dysplastic
   nodules and early HCC, and this gene is viewed as a gatekeeper for
   malignant transformation. TP53 and CTNNB1 mutations function as drivers
   during HCC development. Additional molecular alterations include focal
   DNA amplification (for example, vascular endothelial growth factor
   A/VEGFA, MYC proto-oncogene, bHLH transcription factor/MYC) and
   deletions (for example, cyclin-dependent kinase inhibitor 2A/CDKN2A,
   axin 1/AXIN1), and DNA methylation of promoter regions [[38]3].

   The current options for HCC management include surgical resection,
   liver transplantation, percutaneous local ablations (such as ethanol
   injection and radiofrequency thermal ablation), transarterial
   chemoembolization, transarterial radioembolization, and systemic
   pharmacological therapies [[39]4,[40]5]. The main curative treatments
   for HCC are surgical resection and liver transplantation, which are,
   however, only suitable for 15% to 25% of patients [[41]6]. Furthermore,
   HCC is a chemoresistant and extremely refractory tumor type, and no
   reliable and effective treatments are available for those with advanced
   or metastatic disease [[42]6]. Molecular targeted agents and
   immunotherapy have been regarded as treatment options in recent years.
   Although several multi-kinase inhibitors, such as sorafenib,
   regorafenib, lenvatinib, and cabozantinib, have been approved for
   treating advanced HCC [[43]7,[44]8,[45]9,[46]10], they only provide a
   short increase in median overall survival
   [[47]7,[48]8,[49]10,[50]11,[51]12]. Immune checkpoint inhibitors, such
   as human anti-PD-1 monoclonal antibodies (nivolumab and pembrolizumab)
   and human anti-CTLA4 monoclonal antibody (ipilimumab), were approved
   for advanced HCC from 2017 to 2020, which greatly extend the patients’
   overall survival [[52]13,[53]14,[54]15].

   Personalized or precision medicine has become a new trend in cancer
   treatment, which helps doctors select treatment for patients based on
   their genetic alterations [[55]16]. Because HCC is a highly
   heterogeneous disease, grouping HCC patients into relatively
   homogeneous molecular subtypes may offer a significant clinical benefit
   through precision treatment [[56]17]. HCC is usually classified based
   on tumor burden [[57]18]. Recent advances in multi-omics technologies
   provide an opportunity for developing personalized treatment against
   HCC. For example, next-generation sequencing analyses of HCC identified
   new mutational signatures and defined new tumor subtypes that may
   benefit from targeted treatments in the future [[58]19,[59]20].

   In this study, an integrated bioinformatics analysis determined that a
   prognostic mitotic gene signature is associated with TP53/CTNNB1
   mutation statuses in HCC. This gene signature could be targeted by
   paclitaxel, a clinically approved anticancer drug. Our results support
   that HCC patients with mutant TP53 or wild-type CTNNB1 genes may
   benefit from the paclitaxel anticancer therapy.

2. Materials and Methods

2.1. Cancer Genomics Analysis

   Four microarray data sets from HCC patients, including [60]GSE14520
   [[61]21], [62]GSE45267 [[63]22], [64]GSE50579 [[65]23], and
   [66]GSE62232 [[67]19], were obtained from the public Gene Expression
   Omnibus (GEO) depository database at the National Center for
   Biotechnology Information (NCBI). The differentially expressed genes
   (DEGs) between HCC tumor and adjacent normal tissues were obtained
   using the GEO2R online tool [[68]24], and the criteria used to define
   DEGs were as follows: an adjusted p-value < 0.01 and |Log[2]
   fold-change| > 1. The full DEG list is shown in [69]File S1. The
   overlapped genes among four microarray data sets were visualized by a
   heat map generated using MORPHEUS software
   ([70]https://software.broadinstitute.org/morpheus/; accessed on 21
   January 2021). The prognostic values of these overlapped genes in HCC
   patients were further explored by GEPIA2
   ([71]http://gepia2.cancer-pku.cn/; accessed on 21 January 2021)
   [[72]25] and/or cBioPortal ([73]https://www.cbioportal.org/; accessed
   on 21 January 2021) [[74]26,[75]27], using The Cancer Genome Atlas
   (TCGA) hepatocellular liver carcinoma (LIHC) data set.

2.2. Pathway Enrichment Analysis

   For pathway enrichment in Gene Ontology (GO) biological process, Kyoto
   Encyclopedia of Genes and Genomes (KEGG), and Reactome, selected genes
   were analyzed by STRING ([76]https://string-db.org/; accessed on 21
   January 2021), a database of known and predicted protein–protein
   interactions based on computational prediction, knowledge transfer
   between organisms, and interactions aggregated from other primary
   databases [[77]28]. The following parameters were used: organism = Homo
   sapiens; network type = full network; network edges = evidence; active
   interaction sources = experiments and databases; minimum required
   interaction score = 0.4; max number of interactors to show = queried
   proteins only.

2.3. Gene–Drug Association Analysis

   Gene–drug association was analyzed using GLAD4U
   ([78]http://dlad4u.zhang-lab.org/; accessed on 26 January 2021)
   [[79]29] via the WebGestalt website ([80]http://www.webgestalt.org/;
   accessed on 26 January 2021) [[81]30]. GLAD4U is a gene retrieval and
   prioritization tool based on existing biomedical literature [[82]29].
   WebGestalt is a functional enrichment analysis web tool that integrates
   various primary databases, including GLAD4U [[83]30]. The gene–drug
   interaction network was further constructed using STITCH
   ([84]http://stitch.embl.de/; accessed on 26 January 2021), a database
   of known and predicted interactions between chemicals and proteins
   based on computational prediction, knowledge transfer between
   organisms, and interactions aggregated from other primary databases
   [[85]31]. The following parameters were used: organism = Homo sapiens;
   network edges = evidence; active interaction sources = experiments and
   databases; minimum required interaction score = 0.4; max number of
   interactors to show = queried proteins only.

2.4. Cancer Cell Drug Sensitivity Analysis

   The drug sensitivity data, gene mutation status, and gene expression
   levels in HCC cell lines ([86]Table S1) were downloaded from
   CellMinerCDB ([87]https://discover.nci.nih.gov/cellminercdb/; accessed
   on 25 July 2021) [[88]32] using the data in the Cancer Therapeutics
   Response Portal (CTRP;
   [89]https://portals.broadinstitute.org/ctrp.v2.1/; accessed on 25 July
   2021) [[90]33,[91]34,[92]35]. The CTRP database links cancer cells’
   genetic features to drug sensitivity [[93]33,[94]34,[95]35], and
   CellMinerCDB is an interactive web-based tool allowing integration and
   analysis of genetic and pharmacological data in cancer cell lines
   across various data sets, including the CTRP [[96]32].

3. Results

3.1. Identification of a Prognostic Mitotic Gene Signature in Hepatocellular
Carcinoma

   To identify a common gene signature in HCC patients, four microarray
   data sets ([97]Table 1) were employed, and the common DEGs and their
   fold-change values are visualized in [98]Figure 1 (left). To
   characterize the most essential genes in HCC patients, the prognostic
   roles of these DEGs in patients’ overall survival were analyzed using
   the TCGA-LIHC data set. As shown in [99]Figure 1 (right), 30
   upregulated and 2 downregulated genes were significantly linked to the
   patients’ overall survival. The interactions of these 32 genes were
   constructed by STRING [[100]28]. Pathway enrichment results found that
   most of the upregulated genes are related to the cell cycle procession,
   especially mitosis ([101]Figure 2). For example, mitosis is initiated
   by activating cyclin-dependent kinase 1 (CDK1) with its binding
   partner, cyclin B1 (CCNB1). Aurora kinase A (AURKA) and NIMA (never in
   mitosis gene A)-related kinase 2 (NEK2) are also mitotic kinases. Once
   activated, these mitotic kinases simultaneously or sequentially
   phosphorylate more than 1000 mitotic substrates (such as cell division
   cycle 20/CDC20 and pituitary tumor transforming gene 1/PTTG1) to
   regulate mitotic progression [[102]36]. These results indicate that the
   commonly upregulated genes contributing to poor prognoses of HCC
   patients are correlated with the aberrant regulation of mitosis.

Table 1.

   Microarray data sets for hepatocellular carcinoma patients.
   Access Number                Platform                Normal Tumor References