Keratin 23 is a PPARA-dependent, MYC-amplified oncogene that promotes hepatocyte proliferation

Abstract

Chronic activation of the nuclear receptor peroxisome proliferator-activated receptor alpha (PPARA) promotes MYC-linked hepatocellular carcinoma in mice. Recent studies have shown that MYC can function as an amplifier of transcription where MYC does not act as an ‘on-off’ switch for gene expression, but rather accelerates transcription rates at active promoters by stimulating transcript elongation. Considering the possibility that MYC may amplify the expression of PPARA target genes to potentiate cell proliferation and liver cancer, gene expression was analyzed from livers of wild-type and hepatocyte-specific Myc knockout mice (Myc^ΔHep) treated with the PPARA agonist Wy-14643. A subset of PPARA target genes was amplified in the presence of MYC, including keratin 23 (Krt23). The induction of Krt23 was significantly attenuated in Myc^ΔHep mice and completely abolished in Ppara-null mice. Reporter gene assays and chromatin immunoprecipitation confirmed direct binding of both PPARA and MYC to sites within the Krt23 promoter. Forced expression of KRT23 in primary hepatocytes induced cell cycle-related genes. These data indicate that PPARA activation elevates MYC expression which in turn potentiates the expression of select PPARA target genes involved in cell proliferation. Finally, KRT23 protein are highly elevated in human hepatocellular carcinomas. These results revealed that MYC-mediated transcriptional potentiation of select PPARA target genes, such as Krt23, may remove rate-limiting constraints on hepatocyte growth and proliferation leading to liver cancer.

Introduction

Peroxisome proliferator-activated receptor alpha (PPARA) is a ligand activated transcription factor and key mediator of lipid metabolism. PPARA agonists are considered ‘classic’ non-genotoxic carcinogens, and chronic PPARA activation causes hepatocyte proliferation, hepatomegaly, and eventually hepatocellular carcinoma (HCC) in rodents.^{(1, 2)} Studies have shown the Myc oncogene contributes to agonist-induced hepatocyte proliferation. MYC upregulation occurs indirectly via the PPARA-mediated downregulation of miRlet7c-1, which targets Myc mRNA for degradation.⁽³⁾ The contribution of MYC to hepatocyte proliferation was confirmed in vivo using liver-specific Myc knockout mice (Myc^ΔHep) treated with the potent PPARA agonist Wy-14643. Interestingly, agonist treated Myc^ΔHep mice were resistant to, but not completely protected from, hepatocyte proliferation and hepatomegaly.⁽⁴⁾ Additionally, Myc^ΔHep mice were only partially protected from N-nitrosodiethylamine (DEN)-induced HCC.⁽⁴⁾ DEN-treated Myc knockout mice had less than half the incidence of preneoplastic foci and adenomas. These data suggest that, in these models, MYC does not initiate cell proliferation but rather potentiates the proliferative effects initiated by alternate mechanisms.

A model of MYC action was recently proposed where MYC functions as a general amplifier of transcription, further elevating expression of transcriptionally active genes.^{(5, 6)} In this model, MYC binding accelerates transcription rates by facilitating the release of RNA polymerase from its ‘paused’ state, thereby promoting transcript elongation.^{(7, 8)} This model originated from the fact that MYC targets vary greatly between cell types and the lack of target gene overlap makes it difficult to ascribe a distinct gene signature associated with MYC activation.^(9–11) Studies supporting MYC amplification of gene expression have primarily relied on microarrays and next generation sequencing data derived from cell lines where MYC levels or activity is artificially modulated.⁽⁶⁾ Other studies have suggested that MYC does not universally amplify transcriptionally-active genes, and instead activates discrete sets of genes in a tumor-, tissue-, and cell type-specific fashion.^(10–12) A major obstacle for testing MYC amplifier activity in vivo is the need for coordinated control of both a defined set of target genes and MYC expression levels. Pharmacological activation of the transcription factor PPARA in combination with hepatocyte-specific Myc knockout mice affords a unique opportunity to test the MYC amplifier model in vivo. Myc ablation only provided partial protection against the proliferative effects of prolonged PPARA activation.^{(1, 2, 4)} This observation could be explained by an underlying mechanism involving the MYC-mediated amplification of PPARA target genes. Transcriptional amplification of select genes by MYC may therefore play a major role in agonist-induced HCC models.⁽¹³⁾

The present studies in hepatocyte-specific Myc-null mice and Ppara-null mice treated with the PPARA agonist Wy-14643 identified several PPARA target genes that were significantly amplified by MYC. Notably, Myc-dependent gene amplification was not universal as many ‘classic’ PPARA target genes involved in the control of lipid metabolism were unaffected by Myc status. Notably, keratin 23 (Krt23) was upregulated over forty-fold in response to PPARA activation and the response was markedly attenuated in liver-specific Myc-null mice. KRT23 is type I acidic cytokeratin protein that is highly expressed in several human cancers suggesting that it may contribute to PPARA-mediated hepatocellular proliferation.^(14–17) The current study examined the hypothesis that MYC directly acts as a transcriptional amplifier of select PPARA target genes, including Krt23, and that upregulation of Krt23 promotes hepatocyte proliferation and potentially HCC.

Materials and Methods

ANIMALS

Myc^fl/fl mice were provided by John M. Sedivy (Brown University) and described previously.⁽¹⁸⁾ For temporal hepatocyte-specific disruption, Myc^fl/fl mice were crossed with mice harboring the Cre-ER^T2 recombinase driven by the albumin promoter, designated Alb-Cre-ER^T2.⁽¹⁹⁾ obtained from Pierre Chambon and Daniel Metzger (University of Strasbourg), to yield the Myc^{fl/fl,ERT2-Cre} mouse line. The resulting hepatocyte-specific Myc knockout mice are subsequently referred to as Myc^ΔHep. For activation of the Alb-Cre-ER^T2 driven CRE recombinase, male mice were injected intraperitoneally with 2 mg tamoxifen per mouse while being fed a tamoxifen diet (1 mg/kg, Dyets Inc., Bethlehem, PA) for 3 days. For Wy-14643 treatment, mice were fed for three days with tamoxifen diet then switched to a grain-based diet containing 0.1% Wy-14643 (Bio-Serv, Frenchtown, NJ), while receiving 2 mg tamoxifen by intraperitoneal (IP) injection every other day for an additional 3 days. Ppara wild-type (Ppara^+/+) and conventional Ppara-null mice (Ppara^−/−) used in this study were described previously.⁽²⁰⁾ For Wy-14643 studies, mice were provided a grain-based control diet or matched diet containing 0.1% Wy-14643 for two days. For the MK-8745 study, wild-type mice were given 15 mg/kg/day MK-8745 by intraperitoneal injection (IP) and provided either a 0.1% Wy-14643 diet or matching control chow diet for 3 days. Male 6- to 8-weeks-old mice were used for all studies and all mouse strains were on the C57BL6N background. Mice were housed in light and temperature-controlled rooms and provided water and pelleted chow ad libitum. All animal experiments were performed in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care international guidelines and approved by the National Cancer Institute Animal Care and Use Committee.

SUPPLEMENTAL METHODS

Cell culture; Histological and immunohistochemical (IHC) analysis; Production of recombinant KRT23-expressing adenovirus; Preparation of adeno-associated virus (AAV) short hairpin (sh) RNA of KRT23; Primary hepatocyte isolation and adenoviral transduction; Western blot analysis; Quantitative reverse transcription PCR (qRT-PCR) assays; Luciferase reporter assays; Chromatin immunoprecipitation (ChIP) assays; Computational and Statistical analyses

Results

HEPATIC KERATIN 23 (KRT23) AND OTHER PPARA TARGET GENES ARE AMPLIFIED BY MYC

Liver-specific Myc knockout mice are resistant to, but not completely protected from, PPARA-induced hepatocyte proliferation and DEN-induced HCC.⁽⁴⁾ Considering the possibility that MYC may act as an amplifier of certain PPARA target genes, microarray data from a previous study was re-analyzed (GSE43842).⁽⁴⁾ The analysis of the array data from livers of floxed wild-type (Myc^fl/fl) and Myc^ΔHep mice treated with Wy-14643 revealed that several PPARA target gene mRNAs were significantly amplified in the presence of MYC (Fig 1A; upper panel) while many others were not impacted by MYC status (Fig. 1A; lower panel). Myc^fl/fl and Myc^ΔHep mice were treated with Wy-14643 for 48 h to confirm Myc-dependent changes in Krt23 mRNA and protein expression. As expected, Myc mRNA was upregulated in Myc^fl/fl but ablated in Myc^ΔHep mice treated with Wy-14643 (Fig. 1B). Krt23 mRNA expression was increased in both wild-type and Myc^ΔHep mice in response to PPARA activation, while levels were significantly blunted in Myc^ΔHep mice (Fig. 1B). Western blotting revealed that KRT23 protein levels were also lower in agonist-treated Myc^ΔHep mice (Fig. 1C) and densitometry confirmed a statistically significant difference in protein expression (Fig. 1D). To test whether Krt23 expression can promotes hepatocyte proliferation independent of MYC, an adeno-associated virus (AAV) expressing Krt23 vector (AAV-Krt23) was generated. Myc^F/F and Myc^ΔHep mice were placed on a Wy-14643 diet for five days and then injected with AAV-Krt-23. Wy-14643 treatment and AAV-Krt23 induced more pronounced hepatomegaly and cell proliferation when compared to Wy-14643 or AAV-Krt23 alone (Supporting Fig. S1A and 1-H). However, hepatocyte-specific loss of MYC expression did not significantly attenuate hepatomegaly or Krt23 mRNA expression (Supporting Fig. S1B). KRT23 and MYC protein levels were significantly increased by the combination of Wy-14643 and AAV-Krt23 compared to Wy-14643 treatment or AAV-Krt23 overexpression alone. Surprisingly, KRT23 protein produced from the AAV-Krt23 was significantly attenuated by liver-specific Myc knockout through a yet to be determined mechanism (Supporting Fig. 1C-1E). These data support the view that hepatic Krt23 expression is PPARA-dependent and amplified by, but not dependent on MYC.

FIG. 1.

MYC expression amplifies select PPARA target genes. (A) Liver microarray data from Myc-floxed (Myc^fl/fl) and hepatocyte-specific Myc-null (Myc^ΔHep) mice treated with Wy-14643. Heatmap showing representative MYC-amplified genes from microarray data. Upper panel depicts PPARA target gene mRNAs amplified by MYC. Lower panel shows PPARA targets unaffected by MYC status. (B) Myc and Krt23 expression in wild-type and Myc^ΔHep mice treated with Wy-14643. (C) Western blotting analysis of KRT23 protein expression in liver from Myc^fl/fl and Myc^ΔHep mice treated with Wy-14643. (D) Densitometry analysis of KRT23 protein levels. (E) Quantification of Cxcr1, Fam40b, Fa2h, Krt23, Myc, and Otop1 mRNAs in wild-type, Myc^ΔHep, and Ppara-null mice treated with Wy-14643. (F) Quantification of PPARA-dependent MYC-independent Acox1, Cyp4a10, and Ehhadh mRNAs in wild-type, Myc^ΔHep, and Ppara-null mice treated with Wy-14643. Experiments were performed on at least five mice/group. Each data point represents the mean ± SD (ns, not significant; *, p < 0.05; ***, p < 0.001).

MYC AMPLIFICATION OF PPARA TARGET GENES IS NOT UNIVERSAL

To assess MYC-mediated amplification of select PPARA target genes, wild-type, Myc^ΔHep and Ppara-null mice were treated with Wy-14643 then livers harvested and mRNAs quantified. Fatty acid 2-hydroxylase (Fa2h), otopetrin 1 (Otop1), striatin interacting protein 2 (Strip2; a.k.a. Fam40b), C-X-C motif chemokine receptor 1 (Cxcr1), and Krt23 mRNAs were induced by Wy-14643 in wild-type mice, attenuated in Myc^ΔHep mice, and completely blocked in Ppara-null mice, confirming the expression patterns observed in the microarray data (Fig. 1E). In contrast, expression of the ‘classic’ PPARA target gene mRNAs acyl-coenzyme A oxidase 1 (Acox1), cytochrome P450 family 4 subfamily A polypeptide 10 (Cyp4a10), and enoyl-coenzyme A, hydratase/3-hydroxyacyl coenzyme A dehydrogenase (Ehhadh) was completely PPARA-dependent and unaffected by Myc status (Fig. 1F). These data in Wy-14643 treated Ppara-null and Myc^ΔHep mice substantiate that select PPARA target genes are further amplified by MYC.

KRT23 RESPONSE IS PPARA-DEPENDENT AND LIVER-SPECIFIC

Since PPARA agonist-induced cell proliferation is hepatocyte-specific and not observed in other tissues, genes that contribute to hepatocyte proliferation would be expected to increase only within the liver. Myc, Krt23, and Cyp4a14 mRNA levels were examined in tissues from wild-type and Ppara-null mice treated with the PPARA agonist Wy-14643. Cyp4a14, a well-characterized PPARA target gene, was used as a positive control to assess PPARA activation. PPARA was activated to varying degrees in all tissues as evaluated by Cyp4a14 mRNA expression (Fig. 2A). Induction of Myc and Krt23 mRNAs by Wy-14643 was liver-specific and not observed in Ppara-null mice indicating that their expression is PPARA-dependent (Fig 2B and C). Basal Krt23 levels were also significantly decreased in Ppara-null mice compared to wild-type mice (Fig. 2C). Induction of hepatic KRT23 protein expression by Wy-14643 administration was further confirmed by Western blotting (Fig. 2D). Histological examination of liver sections by H&E staining and immunohistochemistry (IHC) revealed pronounced hepatocyte swelling in Wy-14643 treated animals. KRT23 staining was primarily cytosolic and was increased significantly in response to Wy-14643 treatment when compared to controls; staining was markedly decreased in Myc^ΔHep mice (Fig. 2E and F). Knockdown of Krt23 with AAV-shRNA-Krt23 inhibited Wy-14643 induced hepatocyte proliferation and hepatomegaly (Supporting Fig. S2). Combined treatment of mice with Wy-14643 and AAV-KRT23 significantly increased the number of BrdU-positive hepatocytes (Supporting Fig. S1G and 1H). These data show that Myc and Krt23 expression in response to PPARA activation was liver-specific, supporting a role for these proteins in hepatocyte proliferation.

FIG. 2.

Liver-specific KRT23 response supports a role in PPARA-dependent hepatocyte proliferation. Wild-type (Ppara^+/+) and Ppara-null (Ppara^−/−) mice were treated with Wy-14643 for 48 hours. (A) qRT-PCR analysis of the PPARA target gene mRNA Cyp4a14 in different tissues. (B and C) Expression of Myc and Krt23 mRNA in Ppara^+/+ and Ppara^−/− mice with and without Wy-14643. (D) Western blotting of KRT23 in Wy-14643 treated Ppara^+/+ and Ppara^−/− mice. (E) Hematoxylin and eosin (H&E) staining of liver sections from control, Wy-14643-treated and Wy-14643 treated in Myc^ΔHep mice. (F) Immunohistochemical staining of KRT23 in liver sections from control, Wy-14643 treated and Wy-14643 treated in Myc^ΔHep mice. Representative images are shown. Scale bars represent 100 nm (200X). At least five mice were analyzed/genotype and treatment group. Each data point represents the mean ± SD (***, p < 0.001).

KRT23 IS THE PRINCIPAL KERATIN UPREGULATED DURING PPARA-INDUCED HEPATOCYTE PROLIFERATION

Historically keratins are considered structural filament forming proteins but more recently studies have shown these proteins facilitate cell signaling pathways by acting as scaffolds.⁽²¹⁾ A dendrogram of KRT protein sequences depicts sequence homology of keratins expressed in liver (Fig. 3A). Krt7, Krt8, Krt10, Krt18, Krt19, and Krt20 are commonly used as prognostic and diagnostic markers for liver disease and cancer.^{(21, 22)} Notably, Krt10 and Krt19 are used as diagnostic markers for liver tumor pathology.⁽²³⁾ As expected, Krt23 mRNA expression increased sixty-fold in wild-type mice and levels were unchanged in Ppara-null mice treated with Wy-14643. Krt4 and Krt28 mRNA levels also increased but to much lower extent, three-fold and two-fold, respectively, in response to PPARA activation. (Fig. 3B). Ductular reactions are a characteristic response to liver injury and correlate with liver disease severity. KRT7, KRT8, KRT18, KRT19, and KRT23 are used as markers for ductular reaction. To determine whether Krt23 affects ductular reactions, Krt7 and Krt19 mRNA levels were measured in primary hepatocytes infected with adenovirus expressing KRT23. Results revealed that the overexpression of KRT23 does not impact the expression of Krt7, Krt8, Krt18, or Krt19 mRNA (Fig. 3C) suggesting that KRT23 has a minimal role in the development of ductular reactions.

FIG. 3.

KRT23 is the principal keratin upregulated during PPARA-induced hepatocyte proliferation. (A) Dendrogram depicting homology between mouse KRT protein sequences. Diagnostic and prognostic markers denoted by ‘#’ and ‘*’, respectively. (B) Ppara^+/+ and Ppara-null mice were treated with Wy-14643 for 48 hours then livers were collected and expression of Krt mRNA measured by qRT-PCR. (C) Analysis of Krt7, Krt8, Krt18, Krt19 and Krt23 mRNA in primary hepatocytes treated with AAV-KRT23. Experiments were performed on at least five mice/group. Each data point represents the mean ± SD (nd, not detected; *, p < 0.05; **, p < 0.01; ***, p < 0.001).

PPARA DIRECTLY REGULATES KRT23 BY BINDING TO PROMOTER RESPONSE ELEMENTS

Genomatix MatInspector (Genomatix, Munchen, Germany) was used to bioinformatically identify potential peroxisome proliferator response elements (PPREs) within the Krt23 promoter. Five PPRE sites were identified within 1 kb upstream of the Krt23 transcriptional start site (TSS) (Fig. 4A). Predicted MYC binding sequences, known as an enhancer box (Ebox), are shown for reference. Inserts (A-C) containing the predicted PPRE sites were synthesized and cloned into the pGL4.11 reporter and luciferase assays performed to confirm direct PPARA binding within the Krt23 promoter. Another PPRE-luc construct containing a Acox1 PPRE repeat was used as a positive control and empty pGL4.11 plasmid was used as a negative control. Luciferase signal was significantly elevated in primary hepatocytes transfected with the pGL4.11-A construct which contained a single PPRE site, referred to as PPRE(1), indicating direct regulation by PPARA (Fig. 4B). pGL4.11-B and -C reporter signal was similar to empty vector. PPARA binding was also assessed by ChIP assays on chromatin isolated from livers of wild-type mice on control diet or treated with Wy-14643-treated using a PPARA antibody. The results revealed an enrichment using primer sets ‘A’ and ‘B’ (Fig. 4C). Enrichment was stronger with primer set ‘A’ which targets PPRE(1) in agreement with the luciferase data. Further enrichment was also observed at PPRE(1) after treatment with Wy-14643. Samples from Ppara-null mice on control diet were used as a negative control. Together these data indicate that PPARA regulates Krt23 transcription primarily through direct binding of a PPRE site found within 100bp of the Krt23 TSS.

FIG. 4.

PPARA directly regulates Krt23 through promoter PPRE sites. (A) Schematic representation of predicted PPRE and Ebox sites found within the Krt23 promoter (−1KB). ChIP primer binding sites and reporter construct inserts are shown below. (B) Luciferase-based reporter assays confirmed functional PPREs sites found within the Krt23 promoter. (C) PPARA ChIP assays assessed PPRE binding in liver samples from Ppara^+/+ and Ppara-null mice treated with Wy-14643. Experiments were performed with at least four replicates. Each data point represents the mean ± SD (ns, not significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001).

MYC DIRECTLY REGULATES KRT23 BY BINDING TO EBOX SITES WITHIN ITS PROMOTER

Genomatix MatInspector also identified potential MYC binding Ebox sites within the Krt23 promoter. Luciferase assays were performed to confirm the presence of functional MYC-binding sites. Three Ebox sites (1–3) were identified within 1 kb upstream of Krt23 exon 1 and MYC-binding element reporter constructs were generated for each (Fig 5A). The PPRE(1) site was also included in all constructs. Constructs were transiently co-transfected with either empty vector (pBABE-empty) or a construct that expresses constitutively-activated MYC (pBABE-MYC) into primary hepatocytes.⁽³⁾ Ebox 1, 2, and 3 contain functional MYC response elements (Fig. 5B). ChIP assays were then performed and the results quantitated showing that MYC can bind to all Ebox sites when compared to the negative (NEG) control primer sets that targeted a region around −2 kb upstream of the Krt23 TSS. However, only Ebox 1 showed enrichment after Wy-14643 treatment (Fig. 5C and D).

FIG. 5.

MYC directly regulates Krt23 through promoter Ebox sites. (A) Schematic representation of KRT23 promoter and predicted Ebox binding sites. Reporter construct inserts are detailed below. (B) Luciferase-based reporter assays confirmed MYC transactivation activities in primary hepatocytes transfected with empty plasmid (Vector) or a MYC expression vector (MYC). (C) ChIP primer site locations within the KRT23 promoter used to confirm MYC Ebox binding. (D) MYC ChIP assays for determining promoter occupancy was examined using liver tissue from control and Wy-14643 treated mice. Experiments were performed with at least four replicates. Each data point represents the mean ± SD (***, p < 0.001).

KRT23 INDUCED G2/M-RELATED GENE EXPRESSION IN MOUSE PRIMARY HEPATOCYTES

To test the impact of KRT23 on cell cycle control, adenovirus expressing KRT23 was used to infect primary mouse hepatocytes to determine whether KRT23 might act as a liver-specific, MYC-amplified oncogene. KRT23 expression upregulated mRNAs encoded by G2/M-related genes such as cell division cycle 25C (Cdc25c), cyclin B1 (Ccnb1), cyclin B2 (Ccnb2), and cyclin dependent kinase 1 (Cdk1) (Fig. 6A). Interestingly, these mRNAs are unchanged in Hepa-1 cells infected with KRT23-expressing adenovirus (Supporting Fig. S3). The ‘classic’ proliferation markers including Ki67 and Pcna mRNAs were also significantly increased after forced KRT23 expression (Fig. 6A). Forced KRT23 expression in primary hepatocytes led to the upregulation of aurora kinase A (Aurka) and B (Aurkb), as well as many AURK-associated co-factors including microtubule nucleation factor (Tpx2), mitotic checkpoint serine/threonine kinase (Bub1), transforming, acidic coiled-coil containing protein 3 (Tacc3), and centromere protein A (Cenpa). AURKA/B target gene mRNAs such as centromere protein A (Cdca8), baculoviral IAP repeat-containing 5 (Birc5), and inner centromere protein (Incenp) were also significantly increased in KRT23-expressing primary hepatocytes (Fig. 6B). KRT23, CDC25C, and CCNB1 proteins were significantly elevated after infection (Fig. 6C). Upregulation of AURKA protein expression was also confirmed in KRT23-expressing cells (Fig. 6D). KRT23 is known to interact with and sequester the scaffold protein 14–3-3 epsilon (14–3-3ε), also known as YWHAE, which regulates many cellular processes including signal transduction and the cell cycle.⁽¹⁴⁾ To verify the interaction between KRT23 and 14–3-3ε, liver lysates from Wy-14643-treated mice were used in KRT23 immunoprecipitation (IP) assays. Pull-downs were probed by Western blotting using a 14–3-3ε antibody and results support a strong interaction between KRT23 and 14–3-3ε (Fig. 6E).

FIG. 6.

Adenovirus expressing KRT23 induces G2/M-related gene expression in primary hepatocytes. (A) qRT-PCR analysis of G2/M-related genes Cdc25c, Cdk1, Ccnb1, Ccnb2, Ki67, and Pcna mRNA in primary hepatocytes infected with KRT23 expressing adenovirus (Ad-KRT23). Uninfected cells (Control) and cells infected with GFP expressing adenovirus (Ad-GFP) were used as negative controls. (B) Aurora kinase (Aurk) and Aurk-associated gene expression in KRT23 adenovirus infected primary hepatocytes. (C) Western blotting of KRT23, CDC25C, CCNB, and (D) AURKA protein expression in KRT23 adenovirus infected cells. ACTB expression was used as a loading control. (E) Immunoprecipitation with anti-KRT23 antibody followed by western blotting of 14–3-3ε in liver tissue from control or Wy-14643 treated mice. Each data point represents the mean ± SD (***, p < 0.001).

WY-14643-INDUCED KRT23 EXPRESSION IS PARTIALLY REPRESSED BY AURORA KINASE INHIBITION

Previous studies have shown AURKA upregulates MYC and promotes cancer progression in the liver.⁽²⁴⁾ A regulatory loop may exist between MYC, KRT23, and AURKA, where MYC amplifies KRT23 expression which then promotes AURKA subsequently reinforcing MYC dysregulation. To test whether inhibition of AURKA signaling influences MYC and KRT23 expression, MK8745, a specific AURKA inhibitor, was used in the Wy-14643-induced hepatocyte proliferation model. Liver/bodyweight ratios and the number of BrdU-positive hepatocytes were increased by Wy-14643, and but not significantly inhibited by MK-8745 treatment (Supporting Fig. S4). Myc and Krt23 mRNA expression were partially but significantly inhibited by MK-8745 treatment (Fig. 7A). Western blotting and quantitative densitometry indicated that the protein levels were also significantly decreased in response to treatment (Fig. 7B and C). In agreement with MYC downregulation, PPARA-dependent MYC-amplified genes were suppressed by AURK inhibition (Fig. 7D). A previous study found that the initial increase in MYC expression involved the PPARA-independent downregulation of miRlet7c-1 which targets Myc mRNA.⁽³⁾ These data suggest that MYC expression may be reinforced through an alternate mechanism involving KRT23. MYC amplifies Krt23 expression then KRT23 increases AURKA signaling possibly through an interaction with 14–3-3ε. Studies have shown AURK signaling bolsters MYC expression.⁽²⁴⁾ By inhibiting AURK while PPARA is activated, MYC levels decreased and MYC-amplified gene expression was attenuated. Together these data provide evidence to support the MYC-mediated amplification of PPARA target genes, in particular Krt23, might contribute to agonist-induced hepatocyte proliferation possibly through a mechanism involving AURKA (Fig. 7E).

FIG. 7.

Aurora kinase inhibition attenuates MYC and KRT23 expression in the PPARA agonist-induced liver proliferation model. (A) qRT-PCR analysis of Krt23 and Myc mRNA in livers from Wy-14643 treated mice for 3 days or mice treated with Wy-14643 and MK-8745 (Wy + MK-8745) for 3 days. Vehicle treated mice were used as a negative control (Vehicle). (B) Western blot analysis of KRT23 and MYC in liver from Wy-14643 treated mice or treated with MK-8745 in mice. (C) Densitometric analysis of MYC and KRT23 protein expression. (D) Analysis of PPARA-dependent, MYC-amplified gene expression in response to AURK inhibition. (E) Proposed mechanism of action for KRT23 during PPARA-mediated hepatocyte proliferation. Experiments were performed on at least five mice/group. Each data point represents the mean ± SD (ns, not significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001).

KRT23 EXPRESSION CORRELATES WITH HUMAN HCC PROGRESSION

Previous studies revealed that KRT23 levels were significantly increased in patients with steatohepatitis when compared to patients with steatosis or healthy liver⁽²⁵⁾, and elevated in HCC patient sera when compared to control sera.⁽²⁶⁾ To investigate whether KRT23 may contribute to human liver cancer, protein levels were measured in human liver tumor biopsies by western blot (Fig. 8A). MYC, CCNB1, and AURKA expression was also assessed. Expression of KRT23, MYC, CCNB1, and AURKA was elevated in tumor samples when compared to normal liver tissue. Western blot data was further supported by immunohistochemical analysis which showed elevated KRT23, MYC, and AURKA expression in human HCC sections (Fig. 8D). These results confirm KRT23 expression is also significantly elevated in human HCC. Moreover, expression levels roughly correlated with tumor progression indicating KRT23 may act as a useful diagnostic biomarker for human tumor progression.

FIG. 8.

KRT23 expression correlates with human HCC progression. (A) Western blot analysis of KRT23, MYC, CCNB1, and AURKA protein expression in human liver biopsies. (B) Densitometric analysis of KRT23 and MYC protein expression. (C) Densitometric analysis of CCNB1 and AURKA protein expression. (D) Immunohistochemical staining of KRT23, MYC, and AURKA in from normal or cancerous human liver sections. Representative images are shown. Scale bars represent 20 µm (400X).

Discussion

Keratins are conventionally thought of as structural proteins, but more recently several studies have revealed that they also play active roles in cell signaling pathways.^{(22, 23)} KRTs also commonly serve as diagnostic and prognostic markers for cancer progression and clinical outcomes associated with a wide range of cancers.⁽²³⁾ To date, there is little information related to the functional consequences of KRT dysregulation.^{(21, 23)} The current study identified KRT23 as a potentially novel PPARA-dependent and MYC-amplified liver-specific oncogene. Prolonged PPARA activation causes hepatocyte proliferation and HCC in rodents. The PPARA agonist Wy-14643 significantly induced Krt23 mRNA and KRT23 protein expression in livers of wild-type mice. Krt23 expression was unchanged in Ppara-null mice treated with a PPARA agonist indicating that Krt23 induction is PPARA-dependent. Although Krt23 expression was also upregulated in Myc^ΔHep mice, gene response was much lower than in wild-type animals supporting the role of MYC as a possible amplifier for Krt23. These results indicate that Krt23 expression is influenced by, but not dependent on, MYC expression. Considering the high degree of sequence homology, KRT23 and possibly other KRT family members may promote proliferation by similar mechanisms.

The underlying PPARA-MYC regulatory interactions controlling Krt23 expression was further assessed by analyzing the Krt23 promoter region. PPARA and MYC binding sites were bioinformatically predicted in the 1 KB region upstream of the Krt23 TSS. The Krt23 promoter contains five predicted PPARA binding sites, one of which was found to functional bind PPARA by reporter gene and ChIP assays. Notably, three MYC-binding Ebox sites were identified within 1 KB of the Krt23 TSS. Luciferase reporter assays indicated transactivation by MYC at all three sites; these sites were also shown by ChIP to directly bind MYC. The identification of multiple functional MYC binding sites within a PPARA target gene is supports the possibility that some of these MYC-amplified PPARA target genes might play a role in agonist-induced hepatocellular proliferation leading to HCC. Several studies recently provided convincing support for a novel model of MYC action that represents a paradigm shift in our understanding of this important oncogene.^{(5, 6)} In this model, MYC does not act as an ‘on-off’ switch for gene expression, but rather accelerates transcription rates at active promoters by stimulating transcript elongation.⁽⁸⁾ While the ‘MYC amplifier model’ was defined by cell culture studies, it was not certain that this hypothesis is physiologically-relevant, or whether amplification of an existing gene expression program might drive tissue-specific tumorigenesis. The present data indicate that the Wy-14643-mediated upregulation of Krt23 is both PPARA-dependent and amplified by MYC.

Treatment of mice with carcinogens such as diethylnitrosamine causes HCC through activation of G2/M-related gene expression.⁽²⁷⁾ Thus, cell cycle-related gene expression was measured in mouse primary hepatocytes. Adenovirus-expressing KRT23 induced G2/M-related gene expression including Cdc25c, Ccnb1, Ccnb2, and Cdk1. Since Hepa-1 is a mouse hepatoma cell line, perhaps forced exogenous KRT23 expression is unable to further enhance cell cycle progression and proliferation.⁽²⁸⁾ Previous studies revealed that KRT23 interacts with 14–3-3 epsilon (14–3-3ε) in human colon cancer cells and that this interaction influences many cellular processes including cell cycle and apoptosis.⁽¹⁴⁾ KRT23 immunoprecipitation using Wy-14643 treated mouse liver lysates confirmed the direct association of KRT23 with 14–3-3 in vivo.

Aurora kinases (AURKA) are known to control G2/M-related genes and mediate the oncogenic effects in HCC at least partially through the upregulation of MYC.⁽²⁴⁾ Aurk and Aurk-related gene expression was assessed in mouse primary hepatocytes infected with KRT23-expressing adenovirus. KRT23 induced both Aurka and Aurkb mRNA expression in primary hepatocytes. AURK-associated co-factors and downstream target gene mRNAs including Tpx2, Bub1, Tacc3, Cenpa, Cdca8, Birc5, and Incenp were significantly increased by KRT23 expression. Upregulation of AURK by KRT23 would then be expected to form a feed forward loop where KRT23 upregulates AURKA/B expression thereby further promoting Myc expression and the subsequent amplification of Krt23 expression. Conversely, AURKA/B inhibition would be expected to decrease KRT23 expression through downregulation of MYC. To verify whether the inhibition of AURK suppresses KRT23, mice were treated with Wy-14643 and the AURKA inhibitor MK-8745. Wy-14643-induced Krt23 and Myc mRNA expression were partially inhibited by MK-8745 treatment. Upregulation of PPARA-dependent, MYC-amplified target genes were also significantly attenuated by AURKA inhibition. KRT23 expression was also highly elevated in human hepatocellular carcinomas suggesting dysregulation may have clinical significance. These data support the view that the PPARA-dependent MYC-mediated amplification of KRT23 may involve a regulatory loop that includes aurora kinase signaling. Interestingly, 14–3-3 proteins act as a phosphodocking scaffold and mediate many phosphoregulatory pathways including those associated with aurora kinases.^{(29, 30)} Moreover, 14–3-3ε is an AURK target and phosphorylation regulates 14–3-3 function.^{(29, 30)} The interaction between KRT23, 14–3-3, and AURKA/B represents another potential regulatory mechanism that warrants further exploration.

The present studies demonstrate that MYC can act as a gene amplifier in vivo using a controlled system where a well-defined set of genes can be turned on and off by PPARA in the presence or absence of MYC. Furthermore, demonstrating that MYC-mediated amplification of KRT23 contributes to cell proliferation underscores the pleiotropic mechanisms driving MYC-induced carcinogenesis in different cell and tissue types. As MYC is commonly thought of as ‘undruggable’ in a clinical setting, results from these studies would support that targeting MYC-amplified, tissue-specific, and seemingly unrelated transcription factors may represent therapeutically efficacious targets for MYC-associated tumors. These studies also highlight KRT23 as a potentially novel drug target and diagnostic biomarker for human liver tumors.

LIST OF ABBREVIATIONS:

AURKA

aurora kinase A

ChIP

chromatin immunoprecipitation

Ebox

enhancer box

HCC

hepatocellular carcinoma

IP

immunoprecipitation

KRT23

keratin 23

PPARA

peroxisome proliferator-activated receptor alpha

PPRE

peroxisome proliferator-activated receptor response element

qRT-PCR

quantitative real-time polymerase chain reaction

TSS

transcription start site