KPNA4 Serves as A Biomarker For Pancreatic Adnenocarcinoma KPNA4 Serves as A Promising Diagnostic and Prognostic Biomarker For Human Pancreatic Adenocarcinoma


 Objectives: Growing evidence suggests that aberrant expression of Karyopherin alpha (KPNA) has been involved in the tumor progression of various cancer types. However, the differential expression patterns and prognostic value of the seven KPNA subtypes remain to be investigated. Hence the transcriptional and survival data of KPNAs in patients with pancreatic adenocarcinoma (PAAD) were analyzed.Methods: Transcriptional and survival data related to KPNA expression in patients with PAAD were derived through ONCOMINE, Gene Expression Profiling Interactive Analysis 2, and Human protein atlas databases. The DNA alteration for KPNAs came from The Cancer Genome Atlas and c-BioPortal. The prognostic value analysis was performed with Kaplan–Meier Plotter. Gene functional enrichment analyses were conducted in database LinkedOmics and Metascape.Results: The mRNA expression levels of KPNA1-4, 6,7 were found significantly upregulated in pancreatic cancer tissues than in normal pancreas tissues, whereas the aberrant expression level of KPNA5 was no more significant in the former than in the latter. KPNA1, 4, and 7 were associated with tumor stage or grade of PAAD. Nevertheless, the obvious association with clinical outcomes was identified only in the aberrant expression of KPNA4. Further gene functional exploration about KPNA4 displayed KPNA4 strongly associated with adherens junction and tumor metastasis in PAAD. Conclusion: Our findings suggested that KPNA4 might be a potential diagnostic and a new biomarker of prognosis for patients with pancreatic adenocarcinoma.


Introduction
Pancreatic adenocarcinoma (PAAD) is one of the most lethal human malignancies for still unknown reasons. For the patients with PAAD, their ve-year survival rates have remained at <10% due to no improvement in early diagnosis and poor responsiveness to most standard therapy [1,2]. Though advances of several most common mutations and alterations as the signatures of PAAD, such as KRAS, CDKN2A, SMAD4, and TP53 mutation in PAAD development that has been detected [2], the utility of these research ndings in clinical practice still needs time. Hence, exploring the molecular mechanism underlying PAAD and identifying novel biomarkers will promote understanding this disease better and facilitate the development of targeted therapies effective.
Karyopherin protein includes both importins and exportins. Importin consists of two subunits, karyopherin alpha (importin alpha, also known as KPNA) and karyopherin beta (importin beta, also known as KPNB).
By binding the nuclear localization sequences (NLS) on cargo protein, importin transports the complex of proteins and KPNB into the nucleus. Seven subtypes of KPNA genes (KPNA1-KPNA7) and twenty-two subtypes of KPNB have been identi ed in humans. By binding the nuclear localization signal on the cargo protein, all KPNA acts as an adaptor protein between KPNB and cargo protein to translocate them into the nucleus, while only nine members of KPNB are involved in this process. Different KPNA isoforms present preferences for particular kinds of nuclear localization signals on cargo molecules, but the boundary may not always be absolute [3].
KPNAs perform the indispensable role of carrying proteins from the cytoplasm into the nucleus, which has been signi cant during the differentiation and development of animal cells [3]. Recently, their nontransport functions such as gene regulation and cancer pathogenesis in various cancers have been found [4,5]. These cancers include breast, brain, colon, gallbladder and prostate cancer, etc. [6][7][8][9][10][11]. KPNA7 displayed it can facilitate the growth of pancreatic cancer cells and inhibited the autophagy of these cells in vitro [12]. However, the relationship of the majority of KPNAs with the dysregulated mRNA expression, genetic alterations, prognostic value, and further molecular mechanisms in the patients with PAAD have not been completely investigated yet.
In our study, the expression patterns of different KPNAs family members in PAAD were analyzed from DNA, mRNA, and protein levels based on various databases. Furthermore, the prognostic values and potential functions of KPNAs were also explored. We found KPNA4 in patients with PAAD had potential diagnostic and prognostic value, and hence we investigated it further to elucidate the possible mechanism.

TCGA andONCOMINE database Analysis
The gene expression array datasets of KPNA family members in PAAD in the Cancer Genome Atlas (TCGA) and of ONCOMINE (www.oncomine.org) database [13] were analyzed. The p-value was restricted at 0.05 and the fold change was set at 1.5. The gene rank was restricted to the top 10% and data type selected mRNA. For each KPNA gene, the comparison between cancer samples and normal control specimen was conducted.

GEPIA2 Analysis
Gene Expression Pro ling Interactive Analysis 2 (GEPIA2, http://gepia2.cancer-pku.cn/#index) [14] was used to analyze the transcription levels of KPNAs in patients with PAAD. The mRNA expressions of KPNAs in TCGA cancer specimens were compared with those that matched TCGA normal and GTEx pancreatic data. The cut-off of p-value and log2 of fold change was set at 0.01 and 1, respectively. The correlation analysis of KPNAs and other genes was calculated with the Pearson correlation coe cient.

The Human Protein Atlas (HPA) Analysis
The HPA (https://www.proteinatlas.org) can allow displaying immunohistochemistry (IHC) staining of consecutive sections of human normal and speci c cancer tissues by using two different antibodies: HPA and CAB. KPNA genes in normal tissues and pancreatic adenocarcinoma tissues are evaluated via this database [15]. Based on the immunohistochemical data of patients with or without pancreatic adenocarcinoma from HPA, we further evaluated the protein expression of KPNA genes.

c-BioPortal Analysis
The integrative analysis for cancer genomics of the KPNA family was provided by the c-BioPortal database (http://www.cbioportal.org/) [16,17]. The data of KPNAs came from two studies (TCGA Firehose Legacy and TCGA Pancancer Atlas), and the frequency of gene alterations (ampli cation, deep deletion, and missense mutations), copy number variance were assessed. Furthermore, the overall, disease-free, progression-free, and disease-speci c survival analysis by using c-BioPortal for the KPNA genes with or without alternation were also performed.
UALCAN Analysis UALCAN (http://ualcan.path.uab.edu), a website that includes transcriptional and clinical data from different cancer types in The Cancer Genome Atlas (TCGA) [18], was used to compare the mRNA expression with tumor grades in this study. The transcriptional expression panels for KPNA1-7 based on 176 PAAD patients' grades were compared with 4 normal individuals in the TCGA database. Students' ttest was the analyzed method and p 0.05 was considered statically signi cant.

The Kaplan-Meier Plotter Analysis
In this study, the prognostic value of the mRNA expression of KPNA family members in PAAD was tested using the Kaplan-Meier plotter (www.kmplot.com) [19]. Based on the median expression of mRNA (high vs. low expression), the patients with PAAD were divided into two groups. The overall survival (OS), and relapse-free survival (RFS) of these patients were assessed with the hazard ratio and log-rank p-value. The log P-value less than 0.01 was de ned as statistically signi cant.

LinkedOmics Analysis
To obtain a novel comprehension of the biological effects of the KPNAs, LinkedOmics (http://linkedomics.org), a platform to analyze and compare cancer multi-omics data within and across tumor types [19], was used to perform gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. The analysis was based on the mRNA level and histological type of the TCGA-PAAD dataset in this online database. Pearson correlation test and Gene set enrichment analysis (GSEA) was used as the statistical method and enrichment analysis respectively. The KEGG pathway analysis is used False Discovery Rate (FDR) as rank criteria (from LinkFinder Result).

Metascape Analysis
In our study, Metascape (https://metascape.org) [20], a database to provide a comprehensive gene annotation and analysis, was used to perform the functional enrichment and interactome analysis for the KPNA4 associated genes.

Page 5/17
Transcription Levels of KPNAs in Patients with PAAD.
The transcriptional levels of seven KPNA family members were compared between cancers and those in normal tissue specimen by using ONCOMINE databases (Signi cant correlations are shown in Figure   1, Table 1).
The analysis revealed that the mRNA expression of KPNA1, KPNA2, KPNA3, KPNA4, and KPNA6 was upregulated in patients with PAAD, KPNA5 was downregulated. No data is available for KPNA7. In Ishikawa's dataset [21], KPNA1 was upregulated in PAAD compared with that in the normal samples, with a fold change of 1.957 and p-value of 0.029 (Table 1). The transcription levels of KPNA2 were signi cant elevation in patients with PAAD in three datasets (Table 1). In Segara's dataset, KPNA2 was upregulated in PAAD with a fold change of 4.194 and a p-value of 7.03E-6 [22]. In Pei's and Badea's dataset, KPNA2 was upregulated in PAAD compared with that in the normal samples, with a fold change of 2.485 (p-value=7.30E-5) and 2.644 (p-value=2.16E-8), respectively [23,24]. The transcription levels of KPNA3 was signi cantly higher in patients with PAAD in Segara's datasets [22]. The fold change of mRNA expression of KPNA3 was 2.194 and a p-value of 1.20E-4 ( Table 1). The mRNA levels of KPNA4 (fold change =1.504 and p-value= 0.034) and KPNA5 (fold change = 1.536 and p-value= 5.54E-4) in patients with PAAD were signi cantly higher than those in the normal specimen in Ishikawa's and Buchholz's datasets, respectively (Table 1) [21,25]. The transcriptional levels of KPNA6 in PAAD (fold change = 1.883 and p-value=0.017) were signi cantly different from those in the normal (Table 1) [21].  Collectively, KPNA4 demonstrated the most obviously prognostic value for the patients than the other KPNAs members (Figure 7h). Then receiver operating characteristic curve (ROC) of KPNA4 is analyzed by R, the area under the curve (AUC) 0.978 implies that high expression of KPNA4 may be a diagnostic biomarker for patients with PAAD ( Figure 7p).
The clinical parameters of KPNA4 in patients with PAAD were also analyzed statistically ( Table 2). The differential mRNA expression of KPNA4 is signi cantly associated with tumor T stage and histologic grade. All these results suggest that KPNA4 might be used to predict prognosis in PAAD. To nd the biological signi cance of KPNA4 in PAAD, KPNA4 co-expression mode in the PAAD cohort was predicted by the function module of LinkedOmics (Figure 8). Total 19696 genes were revealed a signi cant correlation with KPNA4. 8941 genes (dark red dots) were displayed signi cant positive correlations with KPNA4, whereas 10754 genes (dark green dots) were found signi cant negative correlations (false discovery rate, FDR < 0.01) (Figure 8a). The top 50 signi cant genes positively and negatively correlated with KPNA4 were shown in the heat map (Figure 8b). The biological process (BP), cellular components (CPs), and molecular functions (MFs) were predicted with Gene Ontology (GO) analysis. The results showed that these correlated genes participate primarily in biological regulation and metabolic process (BP), membrane and nucleus (CPs), protein binding, and ion binding (MFs) ( Figure   8c).
Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis with GSEA demonstrated enrichment of these genes is correlated primarily with adherens junction, and focal adhesion. (Figure 8d) ( Figure 9).
Adherens junction is one part of cell adhesion junction, the other two parts are desmosomes and tight junction. Each part contains classes of proteins composed of groups of genes or gene families. Three primary molecular families: cadherins, armadillo proteins, and plakins constitute the adherens junction [26]. To a certain extent, the metastasis and epithelial-mesenchymal transition (EMT) of cancer cells are associated with loss of cell adhesion and increase of cellular mobility [26,27]. Therefore, we explored whether KPNA4 may impact the metastasis of PAAD or not. 19696 genes correlated with KPNA4 and 64 genes involved in cell adhesion junction and EMT were compared and evaluated ( Table 3). The genes associated with EMT mainly include SNAIL, SLUG, TWIST, Zeb1-3, and TGFB1. 49 overlap genes were found in these two gene sets ( Figure 10). The Venn diagram was created by website tools (http://bioinformatics.psb.ugent.be/webtools/Venn/).

Correlation Analysis between KPNA4 and genes associated with cell adhesion junction in Patients With PAAD
The overlap 49 genes associated with cell adhesion and EMT were explored in GEPIA2 further (Table 3 and Figure 11). Pearson correlation coe cient more than 0.2 and p-value less than 0.01 were set as the threshold, respectively. 43 genes were demonstrated as a linear association with KPNA4. CTNNA1, CTNNB1, TJP1, and MUPP1 were showed an extremely strong correlation with KPNA4 (Table 3 and Figure  11a). CTNNA1and CTNNB1 belong to the adherens junction cluster. The plot is based on TCGA PAAD tumor, TCGA PAAD normal and GTX pancreas dataset.
Further network investigation was conducted between KPNA4 and those linear correlated genes in Metascape. The process and pathway enrichment analysis, protein interaction network of associated genes with KPNA4 were displayed in gure 11b and 11c. These genes are primarily related to the cell-cell junction organization. The pathway enrichment demonstrated that these genes signi cantly enriched in pathway neoplastic cell transformation and neoplasm invasiveness (Figure 11d). These results suggested that abnormal expression of KPNA4 may have a major effect on the metastasis of PAAD.

Discussion
KPNAs are not only passive transporter for proteins, their correlations with different types of cancers are observed recently. Many studies have demonstrated that KPNAs appears to be important in cancer cell growth, migration, and invasion [9,10]. More specially, compared with corresponding normal tissues, KPNAs have relatively higher expression levels in various tumors [10,11,[28][29][30]. However, the relationship between the elevation of distinct KPNA family members and the exact roles in PAAD is not yet clear. Our study is the rst time to perform a deeper analysis of KPNAs in PAAD. We hope our ndings will extend the knowledge of pathogenesis and progression for patients with PAAD.
The diseases associated with KPNA1 are mainly virus infection, neuronal damage, and in ammation [31]. Recent studies found it can help nucleocytoplasmic translocation of acetyl-CoA synthetase 2 (ACSS2) to promote glioma tumorigenesis [6,32]. Also, phosphorylation of E47 accumulating in the nucleus in a KPNA1 dependent manner can facilitate epithelial-mesenchymal transition and metastasis of colon cancer [9]. However, KPNA1 is only an assistant and no direct effect on these cancers. In our report, the mRNA expression of KPNA1 was higher in human PAAD than in normal tissues. However, this elevation was not coincident between transcription and translation level. Compared with normal controls, KPNA1 protein in tumor tissue showed no obvious elevation with immunohistochemical staining. Although the upregulation of KPNA1 correlates with histological grades of the patients with PAAD, high KPNA1 is no correlation with the clinical stages. An increased expression of KPNA1 was signi cantly associated with poor OS in all of the patients followed up for 80 months, but the RFS was of no signi cance. All these ndings suggest that KPNA1 is not an appropriate prognostic biomarker for PAAD.
As a member of the most research in the KPNA family, KPNA2 was found overexpression in various cancers, which is strongly associated with poor prognosis. The elevation of KPNA2 can promote the development of gallbladder and lung cancer [10,28]. USP1-mediated deubiquitination and stabilization of KPNA2 are indispensable for metastasis in breast cancer [8]. KPNA2 is also involved in the survival and metastasis of prostate cancer [33], its upregulation promotes metabolic abnormalities in glioblastomas by controlling c-myc [34]. In our study, KPNA2 was highly expressed on both transcription and translation levels. However, this aberrant upregulation is of no signi cance for tumor stages and grades in patients with PAAD. The prognostic value of KPNA2 is not obvious either.
Like KPNA1, KPNA3 has also a strong association with virus infection and brain damage in the previous studies [35]. However, increased expression of KPNA3 was revealed in colorectal cancer and hepatocellular cancer, and that could promote tumors to a high proliferation rate, invasion, and chemotherapy resistance [29,30] KPNA7 is the only studied member of the KPNA family in pancreatic cancer presently. Overexpressed KPNA7 promotes proliferation of pancreatic cancer cells in vitro [12,36], depletion of KPNA7 can lead to a dramatic reduction in cell growth of pancreatic and breast cancer [37]. In our study, although KPNA7 displayed a marked correlation with tumor stage and grade in patients with PAAD, its elevation is not signi cantly different with both poor OS and RFS.
KPNA4 plays a crucial role in infection and in ammation diseases too. However, its upregulation contributed to the development and progression of prostate cancer. Inhibition of KPNA4 expression of prostate cancer can suppress this process both in vivo and in vitro [11]. The elevation of KPNA4 mRNA expression is also found in the head and neck of squamous cell carcinoma (HNSCC). After the knockdown of KPNA4, the nuclear localization signal-dependent transport activity was attenuated and therefore the malignant phenotypes of HNSCC were inhibited [38]. The elevation of KPNA4 in other cancers such as papillary thyroid cancer and lung cancer is mediated by microRNA, targeting this miRNA/KPNA4 axis can help to inhibit malignant phenotypes of these cancers [39,40].
In our study, the elevation of KPNA4 appears in many cases of PAAD. Therefore, as a potential diagnostic and prognostic marker, it might be worthwhile to investigate further clinically. Our results suggest the high expression of KPNA4 in tumor tissues occurred on both mRNA and protein levels, and a signi cant correlation was found between this expression with tumor stage and grade in the patients of PAAD.
Besides, we clari ed that upregulation of KPNA4 was signi cantly correlated with not only poor OS, but also poor RFS in all of these patients. Furthermore, analysis with ROC supports KPNA4 might be a biomarker for patients with PAAD.
To explore the molecular mechanisms in regulating abnormal KPNA4 expression, we analyzed the KPNA4 co-expression network. We found that the functional consequence of KPNA4 is primarily involved in adherens junction and focal adhesion. The constitution of the cell adhesive junctions includes adherens junction, desmosomes, and tight junction. The function of cell adhesive junctions includes maintenance of the integrity of epithelial cells and regulation of signal communication between cells [26]. The downregulation of cell adhesion molecules is necessary for epithelial-mesenchymal transition (EMT), a process that facilitates the invasion and metastasis of cancer [26,27]. For mining regulators potentially responsible for KPNA4 upregulation, following protein-protein interaction analysis was performed with these genes associated with adherens junction and EMT. The results demonstrated CTNNA1(E-Cadherin) and CTNNB1(beta-catenin) was an extremely strong correlation with KPNA4 in both GEPIA2 and Metascape databases. Many studies have revealed the pivotal impact of E-Cadherin and beta-catenin on the progression and invasion of tumors. Future studies worth investigating the detailed mechanism between distinct KPNA4 and these molecules in PAAD.

Conclusions
In this study, our results implied that the increased expression of KPNA4 in pancreatic adenocarcinoma tissues might serve as a promising diagnostic and prognostic biomarker for PAAD. Although our study has some contributions, there are still several limitations. Firstly, all the analyzed experimental data were taken from online databases. Second, the patient numbers of pathologic stage III and IV are less than 10, which might lead to bias in statistics. To verify our nding, additional larger clinical samples are needed.

Declarations
Ethics Approval and consent to participate This study was approved by the Academic Committee of the 2 nd XiangYa Hospital. All the data were obtained from the online websites and published literatures, it was con rmed that all written informed consent was obtained. The study was performed according to the principles expressed in the Declaration of Helsinki.

Consent for publication
Publication consent was obtained from all authors.

AUTHOR CONTRIBUTIONS
Yuqian Zhou and Jirong Huo conducted and designed the research; Hongyi Zhu wrote the manuscript; Yi Chu, Min Luo, Wenyu Li, and Liang Lyu analyzed data and prepared gures.

Availability of data and material
All data in our study derived from online databases and is publicly available.

CONFLICTS OF INTEREST
The authors declare no competing interest.

FUNDING
This study is funding by Hunan Provincial Natural Science Foundation to Hong Yi Zhu (Grant No. 2019JJ50868), and Hunan Provincial Natural Science Foundation to Yi Chu (Grant No. 2018JJ3720).