KRT23 Acts as an Oncogene in Hepatocellular Carcinoma by Regulating P21 via PI3K/AKT/GSK3β Pathway

Background:Hepatocellular carcinoma (HCC) is a leading cancer worldwide for which diagnosis, treatment and progression are largely unknown. Keratin23 is a potential biomarker forHCC development; however, regulatory mechanisms underlying its expression remain unclear. Inthis research we explored the expression and effect of KRT23 underlying HCC development. Materials and methods:GEPIA was applied to analyze the expression of KRT23 in HCC samples and Kaplan-Merier survival analysis for patients’ prognosis. Next, IHC was further conducted for conrming its expression in HCC tissues. Meanwhile qRT-PCR and western blot analysis were applied to examine the expression of KRT23 on both mRNA and protein level in HCC cell lines compared with immortal hepatocyte LO2. Cell experiments including MTT assay, apoptosis analysis, cell cycle assay and clone formation assay were conducted for cell proliferation while transwell assay and scratch test for metastasis in vitro. Moreover, xenograft tumors in nude mice were further conducted for verication in vivo. As for mechanism in depth, immunouorescence and western blot were operated to explore the effect of KRT23 on EMT and PI3K/AKT/GSK3βsignaling pathway. Furthermore, Co-immunoprecipitation was applied for interaction between KRT23 and P21. Functional rescue experiments were conducted to further testify their mutual effect. Results:For this research, we discovered the high expression of KRT23 in HCC samples and cell lines. Functionally, KRT23 knockdown reduced cell proliferation and metastasis in vitro and vivo. Furthermore, KRT23 participated in EMT progression and interacted with P21 to mediate PI3K/AKT/GSK3βpathway in HCC development. Conclusion:To summarize, HCC by regulating P21 via PI3K/AKT/GSK3βpathway.


Background
Hepatocellular carcinoma (HCC) is still a noteworthy public health issue all over the world even in the coming decade in wake of its fast -rising incidence and high mortality 1,2 . As we all known, HCC development results from gradual accumulation of multiple genetic and factorial regulation, including hepatitis B and C viruses, alcohol consumption, a atoxin exposure, obesity and some with no speci c reason 3,4 . Detection and treatments for HCC has been advanced greatly over the past few years, but there is no promising improvement for patients, along with 5 -year survival rate lower than 18% caused by rather high diagnosis at advanced stage and then limited treatment options 5,6 . Considering the correlation between early detection and prognosis for patients, it's urgent to discover new biomarkers for HCC patients at risk and much more sensitive targets for current therapies.
Expressed in a tissue-speci c manner, keratins (KRTs) are accepted as structural proteins that has the intermediate laments of epithelial cells 7 . Keratin family possesses 50 members, in which KRT9-23 are classi ed to the acidic type while KRT1-8 are basic-neural 8 . Interacted with other proteins or other members in KRT family with various epigenetic modi cation, keratins take great effect on stress response, several cancers progression and signaling pathways 9 . As a member of acidic keratins, KRT23was rstly discovered in pancreatic cancer cells and was induced by histonehyperacetylation 10 .
Meanwhile, the high expression of KRT23 in microsatellite-stable colon cancer (MSS) was applied for distinction MSS from microsatellite-instable colon cancers (MSI) 8 . Moreover, KRT23 as a stress-inducible marker, correlated with the liver disease severity 7 , which predicted that KRT23 might participate in liver disease even cancer. All above reminds us the orientation of KRT23 on HCC development.
For this research, bioinformatic analyses were rstly applied and revealed the high expression of KRT23 in HCC patients and correlation of poor prognosis. Next functional experiments in vitro and xenograft tumors in nude mice were conducted to detect the effect of KRT23 in HCC progression. Furthermore, we found that KRT23 also mediated EMT and interacted with P21 to regulate PI3K/AKT/GSK3βpathway. Above all, KRT23 may has the potential as a new drug target and diagnostic biomarker for HCC patients.

Bioinformatic analysis
Gene Expression Pro ling Interactive Analysis(GEPIA)was applied for detecting the expression of KRT23 in HCC and normal patients. Then Kaplan-Merier survival analysis was used for the prediction of patients' prognosis. Meanwhile the interaction between KRT23 and otherproteins was explored in STRING: functional protein association networks.
Cell culture and tumor tissues collection All cells involved in this research were purchased fromthe Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). Cells were Cultured with DMEM of high glucose (Hyclone, America), which contained 10% fetal bovine serum (FBS, Gibico, America) and 100 U/ml penicillinstreptomycin (Geneview, America). Meanwhile all cells were maintained in sterile and humid cell incubator with 5% CO 2 at 37℃and stored at liquid nitrogen. On the other hand, clinical tumor samples and tumor adjacent samples around tumors 5 cm were collected from pathology department ofFirst A liated Hospital of Xi'an Jiaotong Universityafter approved and authorized by the institute of medical ethics.

Cell transfection assay
Speci c interfering RNAs against the sequence of KRT23 or P21 were all designed and synthesized by Genepharma (Shanghai, China) and concrete sequences has been listed in Table 1. Lipofectamine2000 Transfection Reagent (Geneview) was applied as its instruction for importing siRNAs into cells when the cell density attached around 50%-70%. In view of transfection e ciency, siKRT23#1 was selected for further experiment and construction of lentivirus which inserting the sequence of siKRT23#3 into the vector GV248. Lentivirus-KRT23-RNAi was bought from GeneChem (Shanghai, China) and cells transfected with lentivirus were cultured with DMEM/high glucose containing 5 μg/ml puromycin.
RNA isolation and quantitative real-time PCR (qRT-PCR) RNA extraction was conducted as the protocol of Trizol Reagent (Invitrogen, USA). The sequences of primers applied for this research had been displayed in Table 2. cDNAs were synthesized as the instruction of EasyQuick RT MasterMix (CWBIO, China) and qRT-PCR was examined using UltraSYBR Mixture ((CWBIO, China).β-actin was regarded as the control reference when the expression of genes was calculated as the formula 2 −ΔΔCt .

MTT assay
MTT assay was used to analyze the cell proliferationafter transfection.Cells were seeded into 96-well plates at a density of 5000 cells/well. In the following days, 10 μL of 5 mg/ml MTT was added into every well and after incubating for 4 h in cell incubator at 37℃ discardedthe suspension and 150 μL DMSO were applied for purple crystal solution at the bottom of 96-well plates. At the end, EnSpire Multimode Plate Reader (PerkinElmer) was used for measuring the optical density (OD) at 490nm.

Cloning formation assay
Stable transfected cells were seeded into 6-well plates for 500 cells/well. When one separate cell cloned and formed clusters along with no less than 50 cells after 14 days, the clone formation was xed by methanol for 15 min and then colored with 0.1% crystal violet for 20 min. Images were collected under microscope when the plates were dry.

Flow cytometry
Cell apoptosis was conducted as the instruction of Annexin V-APC/7-AAD apoptosis kit (MULTI SCIENCES, China). After transfection for 48 h, cells were gathered and washed by PBS for two times. Then 500 μL 1×Binding buffer was applied to resuspend cells and following 5 μL Annexin V-APC and 10 μL 7-AAD were added for incubation away from light at room temperature for 5 min. Finally, samples were examined by FALS CALIBAR (USA). As for cell cycle, as the protocol of Cell Cycle Detection Kit (KeyGEN BioTECH, China), 1×10 6 cells were collected and washed by PBS after transfection for 48 h.
Then 70% cold ethyl alcohol was supplied for xing cells overnight at 4℃. The next day, PBS was used to wipe off the xing liquid and then 500 μL PI/RNase A solution prepared in advance at adequate ratio of 9:1 were added and incubated at room temperature for 60 min. identically, samples were detected on FALS CALIBAR (USA).

Transwell assay
Chambers of 8μm pole size coated with Matrigel (BD Biosciences) were applied for cell invasion. After transfection for 24 h, cells were resuspended using medium with no FBS and seeded on the top of chambers which had been put into 600 μL medium with 10% FBS in 24-well plates. Chambers were taken out after placed in cell incubator for 24 h and washed for 3 times with PBS. Wiped gently with swabs internally, chambers were put into 95% alcohol for xation for 15 min and then stained by 0.1% crystal violet for 20 min. Images were collected after removing super uous crystal violet by PBS under the microscope at a magni cation of 200×. Considering migration, the protocol was similar with that for invasion, just with no Matrigel for chambers.

Wound healing experiment
Stable transfected cells were seeded into 6-well plates to attach a con uence of 90%-95% and after stable adhesion, 10 μL sterile tips were applied for scratching. In case of interference from cell growth, medium with no FBS was used during the period. Pictures were collected respectively at 0 h, 24 h, and 48 h under the microscope at an ampli cation of 200×.

The oncogenicity in nude mice
Randomly divided into 2 groups, male athymic 4-week-old nude mice were purchased from the Centre Animal laboratory of Xi'an Jiaotong university and their good care was strictly according to the institution guidelines. Stable-transfected cells were used for xenograft tumors. 200 μL stably transfected cells HepG2 containing at least 5×10 6 cells in PBS were injected subcutaneously for detection of cell growth in vivo. Tumor volumes were measured every four days for drawing the growth curve over time. Mice were were narcotized by using ketamine for 100mg/kg and sacri ced usingcervical dislocation method. Xenograft tumors were collected after 32 days. As for metastasis, 100 μL stably transfected cells HepG2 were injected slowly from tail vein for 2 months and then H&E staining was conducted to assess pulmonary metastasis. The animal experiment was thoroughly operated according to the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and was authorized by the medicalethics committee of the First A liated Hospital of Xi'an JiaotongUniversity.

H&E staining and Immunohistochemistry (IHC)
Xenograft tumors were stored at 10% formalin overnight and then embedded by para n. All conduction was operated as the protocol of Kit of Streptomyces Habilis Ovalbumin -Biotin (ZSGB-BIO, China).Thickness for 4mm and put on slides, those tissues was dewaxed using xylene and then hydrated by grade-reduced concentration of alcohol. Then stained with H&E and observed using microscope. As for IHC, primary antibodies were incubated overnight at 4℃afterdewaxing and hydration. The next day secondary antibodies were incubated and then detected using DAB solution. Stained with hematoxylin for cell nucleus, next slides were dehydrated by grade-ascending concentration of alcohol and then transparentized by xylene. Finally sealed by coverslips with neural gum, the slides were observed by microimaging scanning system (Leica). Antibodies applied in detail has been listed in Table 3.
CO-immunoprecipitation (CO-IP) and western blot RIPA buffer with proteinase and phosphatase inhibitors was applied for cell protein isolation on ice for 10 min and then primary antibodies or IgG were added for incubation at room temperature for 4 h. Next, activated beads were mixed with protein lysates above and kept shocking at 4℃overnight. The next day, beads were washed for 3 times using RIPA buffer and resuspended by 1×loading buffer. Conduction remaining was similar with western blot. Proteins were uploaded into 5% gel for concentration and then 10% for dividing during electrophoresis process. Following proteins were transferred onto PVDF membrane with 0.45 μm or0.22 μm pole size, which were blocked with 5% milk without fat and then primary antibodies were incubated respectively at 4℃ overnight. Secondary antibodies were incubated for 1 h at room temperature the next day. All bands were observed by FD TM FDbio-Dura Ecl kit (Fdbio science, China). Application for antibodies in detail was shown in Table 3.

Immuno uorescence
Sterile slides were put into 6-well plates before Stable transfected cells were seeded into plates. Slides with adherent cells were xed with 4% paraformaldehyde for 15 min and then blocked with goat serum for 30 min. Following primary antibodies were incubated overnight at 4℃ in a wet black box. The next day, after washed for 3 times by PBS, slides were incubated with secondary antibodies for 1 h at room temperature. After stained with DAPI for nucleus and covered with anti-fading buffer, images were collected under inverted immuno uorescent microscope (Leica). The details for antibodies showed in Table 3.

Statistical analysis
The software SPSS 23.0 was applied for statistical analysis. Difference between two paired groups were analyzed by Student's t-test and χ2 testwas applied for analysis of clinical data. p value <0.05 was considered as signi cant difference and p<0.01 very signi cant.

Overexpression of KRT23 in HCC samples and HCC cell lines
As is shown in Fig. 1A and B, the results of bioinformatic analysis from GEPIA revealed that the expression of KRT23 was higher when compared with that in normal patients (*p<0.01). Meanwhile, the overexpression of KRT23 indicated poor prognosis that the overall survival for HCC patients with higher expression of KRT23 was shorter than those with lower expression (p=0.038). Next, the results of immunochemistry (IHC) showed that the expression of KRT23 in HCC tumor samples was signi cantly higher than tumor adjacent tissues (Fig. 1C), accompanied with correlation between the expression of KRT23 and some clinical factors, as shown in Table 4, including tumor size (p=0.006), HBV infection (p=0.030), AFP (p=0.001), pathological grade (p=0.027), TNM stage (p=0.040) and cancer embolus (p=0.034). Then we detected the expression of KRT23 in HCC cell lines. The result of qRT-PCR displayed that KRT23 was overexpressed on mRNA level when compared with that in immortal hepatocyte LO2 (Fig.  1D, *p<0.05, **p<0.01), which was familiar with the result of western blot that detecting the expression of KRT23 on protein level (Fig. 1E).Moreover, immuno uorescence was applied to acknowledge the location of KRT23 in cytoplasm (Fig. 1F). In view of that, three special interfering RNAs against the sequence of KRT23 was designed and transfected effect was examined by both qRT-PCR and western blot to choose the siKRT23#1 for the research in depth ( Fig. 1G and H, **p<0.01). In all, the expression of KRT23 was higher in both HCC patient samples and HCC cell lines. relating to poor prognosis.
KRT23 regulated HCC cell proliferation, including apoptosis and cell cycle Clone formation assay and MTT assay were conducted to assess the in uence for proliferation. In clony formation assay, the number of clony formation in KRT23 knockdown group was obviously less than that in control and NC group ( Fig. 2A, *p<0.05, **p<0.01). As for MTT assay, when the expression of KRT23 was knocked down, the HCC cell growth overtimewas retarded when compared with both control and NC group (Fig. 2B, *p<0.05, **p<0.01). Furthermore, ow cytometry suggested that the number of apoptotic cells in KRT23 knockdown group was increased in contrast to control and NC group ( Fig. 2D and E, *p<0.05). In addition, KRT23 knockdown decreased the cell number in S phase when compared with that in control and NC group ( Fig. 2F and G, *p<0.05). Following the classical biomarkers for apoptosis and cell cycle were detected by western blot (Fig. 2H and I). Just as the results shown, the expression of PARP/cleaved PARP and Bax were augmented when KRT23 was knocked down while Bcl2 was alleviated; the relative markers for cell cycle involving cyclinD1, cyclinE, CDK4, CDK6, C-MYC were all weakened in siKRT23 group. Above all, KRT23 knockdown inhibited HCC cell proliferation and accelerated apoptosis.

KRT23 mediated HCC cell migration and invasion
For this part, transwell assay and wound healing assay were used for HCC cell metastasis. In siKRT23 group, the HCC cell number for migration and invasion was apparently decreased compared with that in control and NC group ( Fig. 3A and B, **p<0.01). similarly, when KRT23 was knocked down, the motility of HCC cells over time was attenuated (Fig. 3C, D and E, *p<0.05, **p<0.01). Meanwhile, the expression of MMP9 was detected using western blot and KRT23 knockdown strongly abrogated its expression on protein level (Fig. 3F). To sum up, KRT23 knockdown suppressed the metastasis of HCC cells.

KRT23 controlled HCC proliferation and metastasis in vivo
The function of KRT23 were further veri ed in vivo. For assurance, the e ciency for stably transfected cells Hep3B and HepG2 were evaluated using both qRT-PCR and western blot to be available as shown in Fig. A(**p<0.01). 200μlHepG2 stably transfected with sh-NCor sh-KRT23 in PBS, which contained 5×10 6 cells was subcutaneously injected to assess tumor growth. As the image shown, the tumor volume in sh-KRT23 group was evidently less than sh-NC group (Fig. 4B), and the growth rate over time was sharply retarded when KRT23 was knocked down (**p<0.01). Considering lung metastasis models, 100μl suspension of HepG2 stably transfected with sh-NC or sh-KRT23 was injected from tail vein. There were much more pulmonary metastatic nodules in sh-NC group when compared with that in sh-KRT23 group (Fig. 4D). Moreover, the expression of Ki67 was detected and as shown in Fig. 4E and F (**p<0.01), the number of Ki67 positive cells was predominantly richer than that in sh-KRT23 group. In a word, KRT23 regulated HCC growth and pulmonary metastasis in vivo.
KRT23 regulated Epithelial-mesenchymal transition and PI3K/AKT/GSK3β pathway As shown in Fig. 5,We inferred that KRT23 may affect EMT progression. No matter the results for immuno uorescence showed or the images in western blotpresented, the expression of E-cadherin, the epithelial relative marker was invigorated when KRT23 was knocked down while the expression of Ncadherin, vimentin, ZEB1 and β-catenin which all represented the characteristics of mesenchyme, were signi cantly weakened when in contrast to that in sh-NC or control group (Fig. 5A and B). Furthermore, KRT23 also regulated PI3K/AKT/GSK3β pathway. As is exhibited in Fig. 5C, when KRT23 was knocked down, the expression of PI3K, P-AKT and GSK3β were all decreased on protein level, in spite of no signi cant difference for AKT.
KRT23 interacted with P21 to mediate HCC cell proliferation and metastasis via PI3K/AKT/GSK3β pathway Following we revealed that there was some interaction between KRT23 and P21 (CDKN1A)to mediate HCC progression, which was reminded by the result from STRING (Fig. 6A). Then qRT-PCR and western blot were applied to preliminarily explore their mutual effect. As is shown in Fig. 6B (*p<0.05, **p<0.01), the expression of P21 was usually elevated when KRT23 was knocked down whether on mRNA or protein level. Subsequently, the interaction between KRT23 and P21 was further con rmed by CO-IP assay (Fig.  6C). Meanwhile, the e ciency of small interfering RNA against the sequence of P21 had been detected to be satis ed (Fig. 6D, **p<0.01).
Functionally, P21 knockdown accelerated HCC cell growth over time when compared with NC group, which can be rescued when co-transfected siP21 and siKRT23 ( Fig. 7A and B, *p<0.05, **p<0.01). In clony formation assay, the number of clone formation in P21 knockdown group were more than NC group, but the increase was abolished if KRT23 knockdown was added ( Fig. 7C and D, *p<0.05, **p<0.01). As for metastasis, P21 knockdown accelerated HCC cell migration and invasion and the number for migrated and invaded cells was obviously reduced in co-transfected with siP21 and siKRT23 group in contrast to that in siP21 group (Fig. 7Eand F, *p<0.05, **p<0.01). similarly, the mobility of HCC cells was strengthened when P21 was knocked down. However, KRT23 knockdown distinctly impeded the activation of P21 knockdown on cell moving ( Fig. 7G and H, *p<0.05, **p<0.01). In viewof mechanism, the expression of PI3K, P-AKT and GSK3β were all improved in P21 knockdown group, which was reversed by KRT23 knockdown (Fig. 7I).

Discussion
HCC withdismal prognosis has brought great burdens for public health worldwide although the various prevalence of underlying risk factors in different areas 11,12 . Once discovered, a great deal of HCC patients have missed the best opportunity for surgical resection, following that 70% patients undertake recurrence within 5 years 11 . Furthermore, HCC is an aggressive solid tumor easily non-sensitive to chemotherapeutics applied at present 13 . Above all, it's imperative to improve the e cacy for identi cation of novel biomarkers at preliminary stage of HCC and then make therapies for HCC much more costeffective.Located on chromosome 17q21.2 and with a molecular weight of 48 kDa, KRT23 expresses a 1.65 kb mRNA and has been reported to have some function in several cancers.Protein KRT23 was highly upregulated in primary rectal cancer, which was analyzed by immunochemistry 14 . Meanwhile, KRT23 activated hTERT expression and then accelerated colorectal cancer growth 15 . In prostate basal cell carcinoma, KRT23 was recognized to be overexpressed 16 . As for liver disease, the expression of KRT23 in serum samples was signi cantly upregulated in HCC, chronic hepatitis compared to normal samples 17 .
Moreover, KRT23 was investigated to be a host factor induced by HCV in liver 18 .KimD et al. 19 detected that KRT23 was one of PPARA targeted, MYC-ampli ed oncogene in HCC.Taken together, all research above highlighted that KRT23 had the potential to be a diagnostic biomarker for liver cancer 19 . Our research is the rst to systematically investigate the effect of KRT23 in HCC development and discover thatKRT23 was overexpressed in HCC tissues and cell lines. Functionally, KRT23 knockdown suppressed HCC cell proliferation and metastasis in vitro and vivo. Above all, KRT23 might act as an oncogene in HCC.
P21, also known as CDKN1A which resided in 6p21.2 20,21 , has been investigated involvement in cell cycle regulation, differentiation, cell migration, cytoskeletal dynamic, apoptosis, transcription, DNA repair, induction of totipotent stem cells, autophagy and senescence 22,23 . Commonly accepted is that P21 as a member of cyclin-dependent kinase inhibitors (CDKI) interacts with various cell cycle proteins to mediate G1/S growth phase arrest in response to different stimuli 24,25 . P21 was regulated in P53 dependent or independent way at transcriptional level and also controlled by phosphorylation, ubiquitination and localization at post-transcriptional level 26 . Likewise, the function of P21 in cancer progression is rather sophisticated considering cellular environment and its subcellular location 27,28 . In our study, analysis of STRING indicated that KRT23 might interact with P21 and KRT23 knockdown elevated the expressionof P21 whether on mRNA level or protein. Co-immunoprecipitation was conducted to verify the interaction between KRT23 and P21. Furthermore, the promotion for cell proliferation, migration and invasion caused by P21 knockdown was partly reversed by KRT23 knockdown.
Epithelial-mesenchymal transition (EMT) mediated by transcriptional factors is a well-known accelerating process for neoplasia metastasis, along with mesenchymal-like morphology for easier separation from primary sites to distant organs and regulation of related biomarkers, E-cadherin for epithelial reduced while N-cadherin, vimentin and others for mesenchymal elevated 29,30 . Rather multiple researches haveelucidated that there are various signaling cascades participated in EMT, including PI3K/AKT/GSK3β pathway 31 , which is reported to activated in preliminary stage of tumorigenesis and take great effect on cell aerobic glycolysis, tumor cell growth, metastasis, angiogenesis, autophagy and chemoresistance [32][33][34] . As the downstream of phosphatidylinositol 3-kinase (PI3K), protein kinase B (AKT) is the crucial molecule in response to different stimuli and often activated by phosphorylation at different sites, most commonly at Ser 473 35 . Then activated AKT inhibits the activity of glycogen synthasekinase 3 Beta (GSK3β) via its phosphorylation at Ser 9, which generating abundant cellular glycogen for cell survival and subsequent acceleration of important gene transcription in cancer development 36,37 . Our research unraveled that KRT23 knockdown controlled EMT process and restrained the activation of PI3K/AKT/GSK3β pathway. Moreover, activation of PI3K/AKT/GSK3β pathway caused by P21 knockdown was abrogated by KRT23 knockdown. In all, KRT23 mediated PI3K/AKT/GSK3β pathway via regulating P21.

Conclusion
In conclusion, this study revealed that KRT23 was overexpressed in HCC tissues and cell lines and mediated HCC cell growth and metastasis in vitro and vivo. Furthermore, KRT23 participated in EMT and interacted with P21 via regulating PI3K/AKT/GSK3β pathway. Our research gave the orientation for KRT23 as a novel biomarker for HCC detection and even potential target molecule for clinical treatment. The study was conducted in accordance with the National Institutes of Health "Guide for the Care and Use of Laboratory Animal" and approved by the medical ethics committee of Xi'an Jiao Tong University.

Consent for publication
All authors involved have read this manuscript and approve to submit to your journal for publication.

Availability of data and materials
The whole data and materials supporting the conclusions were included in this manuscript.

Con icts of interest
The authors declare that they have no con ict of interest.          KRT23 regulated Epithelial-mesenchymal transition and PI3K/AKT/GSK3β pathway. A. The results of immuno uorescence revealed that the expression of epithelial marker E-cadherin was increased while mesenchymal relative biomarkers N-cadherin and vimentin were decreased when KRT23 was knocked down. B. The western blot showed that on protein level KRT23 knockdown suppressed the expression of mesenchymal markers, including ZEB1, N-cadherin, β-catenin and vimentin while epithelial marker Ecadherin was elevated.C.The expression of PI3K, P-AKT and p-GSK3β were all alleviated when KRT23 was knocked down. However, there was no signi cant difference for the expression of AKT and GSK3β. Figure 5 KRT23 regulated Epithelial-mesenchymal transition and PI3K/AKT/GSK3β pathway. A. The results of immuno uorescence revealed that the expression of epithelial marker E-cadherin was increased while mesenchymal relative biomarkers N-cadherin and vimentin were decreased when KRT23 was knocked down. B. The western blot showed that on protein level KRT23 knockdown suppressed the expression of mesenchymal markers, including ZEB1, N-cadherin, β-catenin and vimentin while epithelial marker Ecadherin was elevated.C.The expression of PI3K, P-AKT and p-GSK3β were all alleviated when KRT23 was knocked down. However, there was no signi cant difference for the expression of AKT and GSK3β.   KRT23 interacted with P21 to mediate HCC cell proliferation and metastasis via PI3K/AKT/GSK3βpathway. A-B. P21 knockdown accelerated the cell growth overtime but when cotransfected with siKRT23, the increase was abolished. *p<0.05, **p<0.01.C-D.The number of clone formation was evidently increased in siP21 group when compared with NC group, which was reversed by KRT23 knockdown. *p<0.05, **p<0.01.E-F. P21 knockdown promoted cell migration and invasion while in co-transfected with siP21 and siKRT23 group, the migrated and invaded cells were reduced rapidly. *p<0.05, **p<0.01.G-H. After P21 was knocked down, the cell motility was reinforced, which was partly abrogated when co-transfected with siKRT23.*p<0.05, **p<0.01.I.P21 knockdown strengthened the expression of PI3K, P-AKT and P-GSK3β but in co-transfected with siP21 and siKRT23 group, the elevation was weakened to some extent. Similarly, the expression of AKT and GSK3βhad no obvious change. KRT23 interacted with P21 to mediate HCC cell proliferation and metastasis via PI3K/AKT/GSK3βpathway. A-B. P21 knockdown accelerated the cell growth overtime but when cotransfected with siKRT23, the increase was abolished. *p<0.05, **p<0.01.C-D.The number of clone formation was evidently increased in siP21 group when compared with NC group, which was reversed by KRT23 knockdown. *p<0.05, **p<0.01.E-F. P21 knockdown promoted cell migration and invasion while in co-transfected with siP21 and siKRT23 group, the migrated and invaded cells were reduced rapidly. *p<0.05, **p<0.01.G-H. After P21 was knocked down, the cell motility was reinforced, which was partly abrogated when co-transfected with siKRT23.*p<0.05, **p<0.01.I.P21 knockdown strengthened the expression of PI3K, P-AKT and P-GSK3β but in co-transfected with siP21 and siKRT23 group, the elevation was weakened to some extent. Similarly, the expression of AKT and GSK3βhad no obvious change.   Schematic diagram for the function of KRT23 on HCC development for this research.