Vitamin K2 inhibits the growth and invasiveness of hepatocellular carcinoma cells via protein kinase A activation



Heatocellular carcinoma (HCC) is a common human malignancy. Its high mortality rate is mainly a result of high intrahepatic recurrence and portal venous invasion (PVI). We previously reported that the development of PVI is related to levels of des-gamma-carboxy prothrombin (DCP), a serum protein that increases at a notably higher rate in patients with HCC. Because DCP is produced by a vitamin K shortage, we examined the biological effects of extrinsic supplementation of vitamin K2 in HCC cells in vitro and in vivo. Consequently, vitamin K2 inhibits the growth and invasion of HCC cells through the activation of protein kinase A, which modulates the activities of several transcriptional factors and inhibits the small GTPase Rho, independent of suppression of DCP. In addition, administration of vitamin K2 to nude mice inoculated with liver tumor cells reduced both tumor growth and body weight loss. In conclusion, similar to an acyclic retinoid—which was previously reported to prevent the recurrence of HCC—vitamin K2, another lipid-soluble vitamin, may be a promising therapeutic means for the management of HCC. Supplementary material for this article can be found on the HEPATOLOGY website ( (HEPATOLOGY 2004;40:243–251.)

Hepatocellular carcinoma (HCC) is one of the major malignancies in the world.1 Recent progress in treatment modalities has improved the prognosis of patients with HCC.2–4 However, the long-term prognosis remains disappointing because of the frequent recurrence of HCC5 and the development of portal venous invasion (PVI) of HCC in 16%–65% of patients,6, 7 resulting in the extensive spread of tumor cells throughout the liver.8

Prior to this study, we and others observed PVI more frequently in HCC patients with positive des-gamma-carboxy prothrombin (DCP), a serum protein induced by the absence of vitamin K, than in patients with negative DCP.9 Moreover, serum DCP levels at the time of the HCC diagnosis were found in a prospective study to be the most significant predisposing factor for the development of PVI.9

Prothrombin is a vitamin K–dependent plasma coagulation factor that is synthesized in the liver. The precursor of prothrombin contains 10 glutamic acid residues in its amino-terminal domain. These residues must undergo complete carboxylation to γ-carboxyglutamic acid in the presence of vitamin K for prothrombin to display coagulation activity. DCP is an abnormal prothrombin that is not completely carboxylated. Although the precise causes of DCP production are unknown,10 the concentration of vitamin K is normal in nontumorous parts of the liver but is decreased significantly in HCC tissues.11

Vitamin K is a fat-soluble essential vitamin; its only known physiological role is as a cofactor for γ-glutamylcarboxylase, which acts on several blood-clotting proteins. There are two naturally occurring forms of vitamin K: vitamin K1, which is synthesized by plants, and vitamin K2, which is synthesized by bacteria. Vitamin K2 is a collective reference to compounds composed of menaquinone-n (MK-n), where n stands for the number of five-carbon units. Among the MK-n, MK-4 has a unique synthetic pathway; it is not produced in significant amounts by bacteria, but is synthesized by humans.12 Because the administration of MK-4 decreases serum DCP levels in HCC patients,13 a prospective study of the administration of MK-4 to HCC patients with high DCP to decrease the occurrence of PVI is now in progress. Results to date are positive; the rates of PVI development in the vitamin K–treated group were 2% at 1 year and 13% at 2 years; the rates were 21% at 1 year and 55% at 2 years in the control group (P = .01).14

In this study, we examined the effects and mechanisms of the retardation of HCC cell growth induced by MK-4. The latter is a representation of vitamin K2, and it is widely used in Japan as a therapy for osteoporosis15, 16 with no obvious adverse effects.


HCC, hepatocellular carcinoma; DCP, des-gamma-carboxy prothrombin; PVI, portal venous invasion; MK, menaquinone; PKA, protein kinase A; MTT, 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide; AP, activating enhancer-binding protein; USF, upstream transcription factor; CREB, cyclic adenosine monophospate response element-binding protein; Cdc25A, cell division cycle 25A; cdk2, cyclin-dependent kinase 2.

Materials and Methods


Vitamin K2 (MK-4) was obtained from Eisai Co. (Tokyo, Japan) and Sigma (St. Louis, MO). N-2-[p-bromocinnamylamino]-5-isoquinoline sulfonamide (H89), a protein kinase A (PKA)-specific inhibitor, was purchased from Calbiochem (La Jolla, CA).

Cell Lines.

Human liver tumor cell lines (HepG2 and PRF/PLC/5) and murine embryonic fibroblasts (NIH3T3) were obtained from the RIKEN Cell Bank (Tsukuba Science City, Japan). A human primary hepatocyte cell culture was obtained from the Applied Cell Biology Research Institute (ACBRI, Kirkland, WA).

Cell Growth Assays.

Cells were seeded onto six-well plates at a density of 5 × 104 cells/well. The medium was changed the next day, and vitamin K2 was added at a concentration of 0, 30, 50, or 80 μM. After 3, 4, and 5 days, the numbers of viable cells were determined with the 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay (Sigma) according to the manufacturer's instructions. The mean ± SD mitochondrial dehydrogenase activity per well was determined from triplicate wells.

Flow Cytometric Assay.

The cell cycle distribution of the cells after 3 days of vitamin K2 treatment was analyzed by flow cytometry (Becton Dickinson, Franklin Lakes, NJ) using the fluorescent dye propidium ioide. The proportion of nuclei in each cell cycle phase was determined using MODFIT DNA analysis software (Becton Dickinson).

Matrigel Invasion Assay.

Cell invasion through a reconstituted basement membrane, Matrigel, was assayed by a previously reported method.17 Briefly, Transwell culture chambers with polycarbonate membranes (8.0-μm pore size) coated with 5% Matrigel in the upper compartments were obtained from BD Biocoat (Becton Dickinson). According to the method recommended in the invasion assay,17 we added conditioned medium prepared from NIH3T3 cells to the lower chambers as a chemotactic factor. HepG2 cells were added to the upper chamber and were allowed to migrate for 24 hours at 37°C. Cells that had migrated through the membrane to the lower surface were stained with Giemsa solution (International Reagents Co., Kobe, Japan) and counted in six different fields under a light microscope at ×40 or ×100 magnification. Each experiment was performed in duplicate wells and was repeated twice.

Microarray Procedures.

The complementary DNA microarray analysis was performed according to the manufacturer's instructions using Human Oligo Chip 30K Subset A (Hitachi Software Engineering Co., Yokohama-City, Japan), which contained about 10,000 open reading frame oligo probes, most with known functions. The detailed procedures and full microarray data are available from Gene Expression Omnibus at the National Center for Biotechnology Information Web site (; the sample ID is GSM10736) based on the MIAME guidelines issued by the Microarray Gene Expression Data group.18

Search for the Gene Regulatory Regions.

Searches for gene regulatory elements were performed using the Celera Discovery System and Celera's associated databases. Five-kilobase flanking sequences of the genes of interest were queried for the presence of transcriptional factor consensus elements.

Nuclear Extraction and Transcription Factor Array.

Nuclear extracts were prepared using mininuclear extraction methods.19 A simultaneous comparison of the activation of 54 transcriptional factors was performed using TranSignal Protein/DNA Arrays (Panomics, Redwood City, CA) according to the manufacturer's instructions.

Electrophoretic Mobility Shift Assay.

Nuclear extracts were prepared from HepG2 cells treated with 50 μM vitamin K2 for 0, 2, 6, 12, and 24 hours. The activating enhancer-binding protein (AP)-2 and upstream transcription factor (USF)-1 binding reactions were carried out with 5 μg of nuclear proteins and 1 μg AP-2 (GATCGAACTGACCGCCCGCGGCCCGT) or USF-1 (CACCCGGTCACGTGGCCTACACC) biotin-labeled probe according to the manufacturer's protocol (Panomics). To test for specificity, a 100-fold excess of unlabeled oligonucleotide was added to the binding reactions in a competition assay.

cAMP Response Element-Binding Protein Activation Assay.

Indirect immunofluorescence staining was performed, as previously described.28 HepG2 cells were seeded onto an eight-well slide-glass chamber the day before the experiments. Vitamin K2 (50 μM) was added to the medium for 0, 1, 2, or 3 hours in each well. Forskolin (10 μM) was used as a positive control.20 The cells were incubated with rabbit anti–cyclic adenosine monophospate response element-binding protein (CREB) phosphorylated at Ser133 (Cell Signaling Technology, Beverly, MA), and then with Alexa Fluor 488 anti–rabbit antibody (Molecular Probes, Eugene, OR). After incubation with 1 μg/mL propidium iodide (Sigma) for 5 minutes, the cells were observed using confocal laser scanning microscopy TCS SL (Leica Microsystems, Wetzlar, Germany).

In Vitro Kinase Assay for PKA Activity.

Cell lysates were prepared in cold extraction buffer (25 mM Tris-HCl, pH 7.4, 0.5 mM ethylenediaminetetraacetic acid, 0.5 mM egtazic acid, 10 mM β-mercaptoethanol). The supernatant was used in a nonradioactive assay for cAMP-dependent protein kinase activity (PKA Assay System, Promega, Madison, WI). Phosphorylated and unphosphorylated forms of the PKA peptide substrate were separated by 0.8% agarose gel electrophoresis.

Rho and Rac Activation Assays.

Cell extracts were prepared and normalized for the protein concentration by using 250 μg of protein for each assay with a Micro BCA Protein Assay Reagent Kit (Pierce, Rockford, IL). To assess the total Rho or Rac content, 25 μg (10%) of each extract was used. The remaining extract was reacted with the Rhotekin Rho-binding domain coupled to agarose, or with the p21-activated kinase 1–binding domain coupled to agarose. Immunoprecipitates were blotted with anti-Rho immunoglobulin G or anti-Rac immunoglobulin G according to the instructions in the kits from Cytoskeleton (Denver, CO).

In Vivo Analyses.

Because HepG2 cells were not able to be inoculated subcutaneously into the nude mice, we used PRF/PLC/5 cells in our in vivo analyses, which showed similar results to HepG2 cells in in vitro studies. Athymic Balb/c nude mice (6 weeks old) were injected subcutaneously with 1 × 107 growing PRF/PLC/5 cells in 0.1 mL Matrigel (Becton Dickinson). To evaluate the effect of vitamin K2 on tumor growth, administration of vitamin K2 or vehicle began after subcutaneous tumor formation, 14 days after inoculation with cancer cells. Because mice are estimated to drink approximately 30 mL water per day at this age, vitamin K2 was added to the water bottle to give a dose of approximately 20 mg/kg/d. The bottle was changed every 2 days. Four and five mice were enrolled in the vitamin K2 and vehicle groups, respectively. Mice were treated for 6 weeks. Part of each tumor was used for PKA activity assays and part for paraformaldehyde-fixed sections stained with hematoxylin-eosin. The percentage changes in tumor volume and body weight between the second and fourth weeks after the start of vitamin K2 administration were calculated. All experimental procedures were approved by the Committee on Animal Research of Tokyo University.

Immunoblot Analysis.

To examine the cell division cycle 25A (Cdc25A) phosphatase activity, the phosphorylation status of cyclin-dependent kinase 2 (cdk2) and cdc2 kinase at Tyr15 was determined using Western blotting with anti-Cdc25A (Santa Cruz Biotechnology, Santa Cruz, CA), anti-cdc2 (Cell Signaling Technology), or anti-cdc2-phospho Tyr15, which reacts with phospho-Tyr15 in both cyclin-dependent kinase 2 (cdk2) and cdc2.21

Statistical Analysis.

The results were analyzed using ANOVA with a post hoc Scheffe test (StatView J, Abacus Concepts Inc., Berkeley, CA).


Vitamin K2 Reduces Growth of Human Liver Tumor Cells.

Because DCP increases at a higher rate in patients with HCC22 and is related to the development of PVI,9 we examined the biological effects of vitamin K2 on HCC in vitro. To first determine the effects of vitamin K2 on cell growth, we added vitamin K2 to the medium of HCC cell lines. As shown in Fig. 1a, vitamin K2 inhibited HepG2 cell growth in a dose-dependent manner. The inhibitory effect was apparent after only 3 days (data not shown). The cell growth rate for 5 days was lower in the vitamin K2–treated cells compared with untreated cells (the cell growth rate after a 5-day incubation was 485% ± 32% in control cells and 385% ± 45%, 237% ± 2%, and 175% ± 5% in cells treated with 30, 50, and 80 μM vitamin K2, respectively). From these results, the vitamin K2 concentration causing 50% cell growth inhibition was calculated to be 45 μM. This dose is much higher than the doses needed for the complete suppression of DCP production by DCP-producing HCC cells (0.1 μM; data not shown). Similar results were obtained using PRF/PLC/5 cells (data not shown). Within the range of vitamin K2 doses used in this assay, the viability of primary hepatocytes was not affected (data not shown). Growth inhibition depended primarily on the increase in the number of cells in G1 and G2/M phases of the cell cycle (Fig. 1b). These data suggest that high-dose vitamin K2 treatment functions as an inhibitor of HCC cell growth.

Figure 1.

Vitamin K2 reduces the growth of HCC cells. (a) HepG2 cells were plated at a concentration of 5 × 104 cells/well in a six-well plate. After 24 hours, cells were treated with 0, 30, 50, and 80 μM of vitamin K2. Cell growth was determined after 3 and 5 days by the MTT assay as described in Materials and Methods. The results are expressed as the percentage of viable cells compared with before vitamin K2 treatment. *P < .05 and **P < .01 versus cells without vitamin K2 determined at the fifth day. (b) HepG2 cells with or without treatment with 50 μM vitamin K2 for 5 days. Cell suspensions were analyzed by flow cytometry after cells had been stained with propidium iodide as described in Materials and Methods. The percentages of G1 phase cells were measured. Data show a representative result from two independent experiments that gave similar results.

Vitamin K2 Reduces Invasion of Human Liver Tumor Cells.

Because oral administration of vitamin K2 potentially reduces the development of portal venous tumor invasion in HCC patients,9 we evaluated the effect of vitamin K2 on liver tumor cell invasion in vitro. Invasion of HepG2 cells was inhibited by treatment with vitamin K2 in a dose-dependent manner (Fig. 2), indicating that vitamin K2 is a potent inhibitor of invasion—as well as growth—of liver tumor cells. This phenomenon did not depend on growth inhibition; the assay was performed 24 hours after the addition of vitamin K2, before growth inhibition was apparent.

Figure 2.

Vitamin K2 inhibits the invasion of HCC cells. Invasion of HepG2 cells in the absence or presence of 50 μM vitamin K2 in the upper compartment of the Transwell culture chambers was assessed after 24 hours of treatment. Data are expressed as the mean ± SD of three separate experiments with triplicate results. (Original magnification ×40.)

Identification of Vitamin K–Mediated Signal Transduction Pathways.

To clarify the mechanisms of growth and invasion inhibition by vitamin K2 in liver tumor cells, we used microarrays of approximately 10,000 genes to determine the changes in gene expression induced by vitamin K2. The expression levels of approximately 3.4% of the genes examined were increased approximately twofold after 3 days of vitamin K2 treatment (full microarray data, platform, and methods are available from Gene Expression Omnibus at the National Center for Biotechnology Web site:; the sample ID is GSM10736). These results suggested that vitamin K2 directly or indirectly modulates some types of transcription.

Next, to determine the transcriptional pathways that vitamin K2 modulates, we analyzed protein/DNA arrays, examining 54 kinds of transcription factor activities. Among the transcriptional factors investigated, vitamin K2 up-regulates AP-2–, USF-1–, and CREB-related transcriptional activities (Fig. 3a). The activations of AP-2 and USF-1 were detectable at maximum activity 6 hours after stimulation with vitamin K2, as shown by conventional EMSA experiments (Fig. 3b and 3c). The activation of CREB was also confirmed by the increased nuclear staining of the phosphorylated CREB at Ser133, a key regulatory site controlling CREB activity, after stimulation with vitamin K2 (Fig. 3d).

Figure 3.

Changes in transcriptional activities induced by vitamin K2. (a) To evaluate the transcriptional factors related to vitamin K2 treatment, microarray analyses were performed using transcriptional factor arrays and nuclear extracts from HepG2 cells treated with 50 μM vitamin K2 for 0, 2, and 12 hours. Hybridization signals of transcriptional factors binding to the biotin-labeled probe mix are shown. Differences in spot intensities were observed for (1) AP-2, (2) CREB, and (3) USF-1. The intensities of EGR (4), which were not changed by vitamin K2 treatment, are shown as a control. (b) EMSA was performed using an AP-2 consensus binding nucleotide and HepG2 nuclear extracts treated with vitamin K2 for 0, 2, 6, 12, and 24 hours. The specificity of the DNA/protein band (arrow) was confirmed by adding excess cold probe. (c) EMSA was performed using a USF-1 consensus binding nucleotide and HepG2 nuclear extracts treated with vitamin K2. The specificity of the indicated DNA/protein band was confirmed by adding excess cold probe. (d) To determine CREB activity, the phosphorylated form of CREB was immunostained. HepG2 cells were preincubated with either 50 μM vitamin K2 for 1, 2, or 3 hours or with 20 μM forskolin for 1 hour as a positive control. The cells were fixed and probed against CREB phosphorylated at Ser133. Propidium iodide was used for nuclear staining. The phosphorylated CREB is green, and the nucleus appears red. (Original magnification ×400 using confocal laser scanning microscopy.) EGR, early growth response factor; EMSA, electrophoretic mobility shift assays.

To validate our microarray results, and to look for the mechanisms of regulation, we conducted a search for AP-2, USF-1, and CREB consensus sequences in genomic regions upstream of the top 20 genes that were induced by vitamin K2 treatment. As shown in Table 1, 12 of the 20 genes (60%) had at least one AP-2, USF, or CREB binding sequence in their 5′ flanking regions.

Table 1. Identification of Candidate AP-2, CREB, or USF Consensus Elements Within the 5 Kilo Base Upstream of Flanking Regions of the Top 20 Genes That Were Induced by Vitamin K2 Treatment
Signal RatioGeneAP-2CREBUSFLocation
128.6Cathepsin G   14q11.2
49.1Uskelin 1   7q32
36.3Indoleamine-pyrrole 2,3 dioxygenase   8p12–p11
32.6KIAA0710 + 12q13.2–q13.3
26.1Solute carrier family 12   15q13
20.1Antitrypsin, member 10   14q32.13
17.4Proprotein convertase subtilisin/kexin type 5 ++9q21.3
16.8PR domain containing 12+++9q33–q34
16.2HTGN29 protein   5q31.1
10.9Primase, polypeptide 2A  +6p12–p11.1
9.8Solute carrier family 27 + 5q31.1
8.5CCR4 carbon catabolite repression 4-lik   19
7.8Tudor and KH domain containing+ +1q12
7.4Zinc finger homeodomain 4   8q13.3
6.7Signal sequence receptor 4+ +Xq28
6.5Sirtuin+  19p13.3
6.5Engulfment and cell motility 2++ 20q13
6.4Dual specificity phosphatase 5+ +10q25
6.2Serum/glucocorticoid regulated kinase-like+  8q12.3–8q13.1

Mechanisms of Vitamin K2 Inhibition of Cell Growth and Invasion.

Our studies indicate that vitamin K2 activates the AP-2, USF-1, and CREB signal transduction pathways. Because PKA is an only known common activator of these three signaling pathways as far as we investigated,23–25 we evaluated PKA activation after vitamin K2 treatment. Vitamin K2 rapidly stimulated PKA activity in liver tumor cells, producing maximum activity 2 hours after stimulation (Supplementary Fig. 1a). This stimulation was dose-dependent within the range of 20–100 μM (Supplementary Fig. 1b and 1c). However, the PKA activity was saturated at over 100 μM vitamin K2 (Supplementary Fig. 1b and 1c).

The selective PKA inhibitor H89 blocked the activation of PKA by vitamin K2. To determine whether vitamin K2 inhibited cell growth via signals transduced through PKA, we evaluated the effects of vitamin K2 on cell growth in the presence and absence of H89. This inhibitor blocked the anti–cell growth properties of vitamin K2 (Fig. 4a). Furthermore, H89 also blocked the inhibitory effect of vitamin K2 on liver tumor cell invasion (Fig. 4b). These results indicate that vitamin K2 inhibition of cell growth and invasion is PKA-dependent and that activation of PKA blocks liver tumor cell growth and invasion.

Figure 4.

The antiproliferative effects of vitamin K2 involve PKA activation. (a) HepG2 cells treated with 50 μM vitamin K2 alone or in the presence of 5 μM H89. After treatment for 5 days, cell numbers were estimated using MTT assay. Results are expressed as percentages of cell growth compared with untreated cells and are the mean ± SD of quadruplicate determinations from three separate experiments. *P < .05 versus control; **P < .05 versus cells treated with vitamin K2 alone. (b) An invasion assay was performed using HepG2 cells treated with 50 μM vitamin K2 alone, or in the presence of 5 μM H89, in the upper compartment of the Transwell culture chambers. Data are expressed as the mean ± SD of three separate experiments with triplicate results. *P < .05. (Original magnification ×100.)

Rho Activation Is Inhibited by Vitamin K2.

Recent studies have highlighted the important roles of Rho family proteins, including RhoA and Rac1, in the invasion of carcinoma cells and the inhibition of RhoA activity by PKA-mediated signaling. Therefore, we investigated the effects of vitamin K2 on RhoA and Rac activities. Although vitamin K2 did not affect the total RhoA protein expression, it inhibited the activation of RhoA in a dose-dependent manner, and H89 blocked this inhibition (Fig. 5). Therefore, vitamin K2 was believed to activate PKA, which in turn inhibits RhoA activation. Rac activities were not determined because the amount of activated Rac in HepG2 cells was too small to assess the effects (data not shown).

Figure 5.

Vitamin K2 induces the PKA-dependent inhibition of RhoA. HepG2 cells were untreated or treated with 50 or 100 μM vitamin K2 alone or in the presence of 5 μM H89 for 6 hours. Cells were harvested and cell extracts were assayed for Rho activity as described in Materials and Methods. Representative immunoblots for total and active RhoA are shown. Quantified data on the RhoA activity and total RhoA expression are presented as a percentage compared with the control from three independent experiments. *P < .05.

Effects of Administration of Vitamin K2 on HCC Cells in Nude Mice.

Because vitamin K2 exhibited antiproliferative effects in vitro, the effect of vitamin K2 treatment on HCC growth was assessed in vivo. Athymic nude mice were inoculated subcutaneously with 1 × 107 PRF/PLC/5 cells, and tumors were allowed to develop for 2 weeks. Inoculated animals were then randomized to receive either oral vitamin K2 (20 mg/kg/d) or vehicle (ethanol). Six weeks after grouping, four of the five vehicle-treated mice developed large visible tumors. Although three of the four vitamin K2–treated mice also developed visible tumors, the tumors were smaller than those in vehicle-treated mice. Consistent with the final tumor sizes, the ratio of tumor size changes during the 2 weeks between the second and fourth weeks after grouping was lower in vitamin K2–treated mice (vehicle-treated mice, 135.0 ± 38.2%; vitamin K2–treated mice, 63.0 ± 24.2%; P = .05). Moreover, the ratio of body weight loss over these 2 weeks was significantly lower in vitamin K2–treated mice (vehicle-treated mice, 10.3 ± 5.2%; vitamin K2–treated mice, −7.6 ± 1.5%; P = .009) (Supplementary Fig. 2a and 2b). Consistent with the in vitro studies, there was a tendency toward higher PKA activity in the tumors of vitamin K2–treated mice (Supplementary Fig. 2c). An examination of cross-sections of resected tumors stained with hematoxylin-eosin revealed lower cell densities and noticeable vesicles in tumors of vitamin K2–treated mice (Supplementary Fig. 2d).

Effects of Vitamin K2 on Cdc25A Phosphatase Activity.

A newly synthesized vitamin K analogue, called compound 5, acts as a specific Cdc25A inhibitor.26 In addition, down-regulation of Cdc25A has been implicated in delaying cell cycle progression at the G1/S and G2 checkpoints.27 Therefore, we investigated whether or not vitamin K2 alters the expression or activity of Cdc25A, which works as a phosphatase in Tyr15 phosphorylation of cdk2 and cdc2.27, 28 Because our study revealed that vitamin K2 stimulates PKA after a 2-hour treatment, and the experiments using compound 5 were usually performed 6 hours after treatment,29 we used the cell lysate 2 and 6 hours after vitamin K2 addition. As shown in Supplementary Fig. 3, vitamin K2 did not significantly alter the expression of Cdc25A and cdc2; moreover, vitamin K2 did not change the phosphorylation states of cdk2 and cdc2 at Tyr15. These results suggest that the growth inhibition effect of vitamin K2 is independent of Cdc25A activity.


In this study, we showed that vitamin K2 inhibits the invasiveness and growth of HCC cells through PKA activation. Because DCP is induced by a vitamin K shortage, we added vitamin K2 to in vitro HCC cell cultures and subsequently observed inhibitory effects of vitamin K2 on cell growth and invasion. Extrinsically administered vitamin K2 completely suppresses DCP production in HCC cells in culture (data not shown). Thus, at first we thought that the inhibition of cell growth and invasion produced by vitamin K2 might depend on DCP suppression; that is, DCP might have growth-promoting effects. However, we now believe—for several reasons—that the inhibition of growth and invasion by vitamin K2 is not via DCP. First, vitamin K2 inhibits HCC cell growth only at high doses of approximately 50–100 μM; these doses are much higher than those needed for the complete suppression of DCP production by HCC cell lines. Second, vitamin K2 is reported to inhibit the growth of Hep3B cells— a liver tumor cell line that does not produce DCP—and other cancer cell lines derived from nonliver tissues,30–32 supporting the idea that DCP is not involved in this growth inhibition. Third, an adequate amount of recombinant DCP did not promote the growth of Hep3B cells (data not shown). Thus, the growth inhibition produced by vitamin K2 came to be considered not dependent upon DCP suppression.

Although several studies have already revealed that vitamin K2 has a transcriptional regulatory function for some candidate genes,33 comprehensive gene expression changes have not been examined. Our study showed that vitamin K2 might modulate gene expression and activate AP-2–, USF-1–, and CREB-related transcriptional abilities. Although changes in gene expression might depend on vitamin K2's role as a γ-glutamylcarboxylase cofactor, certain changes in expression could be related to modifications of transcriptional factor activities induced by vitamin K2. In fact, more than half of the top 20 vitamin K2–up-regulated genes contain consensus sequences for these transcriptional factors in their flanking regions (see Table 1).

It is possible that supraphysiological doses of vitamin K2 might alter the effects of key signaling molecules. Our current studies demonstrate that vitamin K2 works as an activator of PKA, which is a common regulator of AP-2, USF-1, and CREB transcriptional factors.23–25 AP-2 and USF-1 activities are reportedly related to growth or metastasis inhibition.34, 35 PKA itself also induces cell cycle arrest,36 not only at G1 phase, but also at G2/M phase,37, 38 which is consistent with our study. A synthesized vitamin K derivative inhibits Cdc25A, and inhibition of Cdc25A leads to G1/S and G2/M arrest,27 but we could not detect any effect of vitamin K2 on Cdc25A activity. Therefore, we speculate that vitamin K2 activates PKA, consequently inhibiting cell growth and invasion via the activation of downstream transcriptional factors and some other mechanisms, independent of Cdc25A activity.

We demonstrated that the inhibition by vitamin K2 of the growth, invasion, and Rho activity of HCC cells was almost completely reversed by the selective PKA inhibitor H89, suggesting that vitamin K2 exerts its antitumor effects on HCC cells in a PKA-dependent manner. Consistent with our results, it has been reported that PKA inhibits RhoA activation.39 The modulation of PKA activity by vitamin K2 was apparent only at high doses, which is consistent with the finding that high doses of vitamin K2 were needed to inhibit HCC cell activities. The activation of PKA seems to be another physiological role for vitamin K2, in addition to its function as an enzyme cofactor.

Coumarin and its derivatives have been reported to inhibit cancer cell growth and spread.40 Coumarin inhibits vitamin K epoxide reductase, impairing vitamin K recycling. Therefore, these reports seem to contradict our results. However, their mode of action appears to be independent of the anticoagulant activity,41 which is in turn anti–vitamin K activity, and their mechanisms have not been examined fully. By contrast, these results might be consistent with our finding that the inhibitory effect of vitamin K2 on growth and invasion seems to be dependent on its effect other than as a γ-glutamylcarboxylase cofactor, which is important in coagulation.

Although our results showed that high-dose vitamin K2 activates PKA, the mechanism is unknown. Neither PKA expression nor the production of cyclic adenosine monophosphate, a known PKA activator, was modulated by vitamin K2 treatment in our study (data not shown). Although several studies have reported cyclic adenosine monophosphate–independent PKA activation,42, 43 the precise mechanisms are still unknown. To learn more about PKA activation by vitamin K2, we believe that it is important to identify the vitamin K2–binding protein. In addition—and consistent with our results—it has been reported that, using neuron cells, vitamin K enhances differentiation via PKA activation.44

Most of the present studies were performed using relatively high doses of vitamin K2. In Japan, vitamin K2 is usually administered to osteoporotic patients at a dose of 45 mg/d, resulting in a maximum serum level of approximately 3 μM.45 Several reports have showed that extrinsically administered vitamin K2 is concentrated in the liver, and the concentration in liver tissues is at least 10 times greater than in plasma46; thus in vivo liver tumors might be exposed to higher concentrations of vitamin K2 because of more prolonged exposure and greater uptake by active transport.47

We previously reported that DCP-producing HCC correlates highly with impending portal tumor invasion. Our data indicate that vitamin K2 supplementation actually reduces the growth and invasion of HCC cells. Although vitamin K and its analogues are reported to inhibit the growth of HCC and other carcinoma cells, the mechanisms are unclear.26, 48 Compound 5, a newly synthesized vitamin K analogue that acts as a specific Cdc25A inhibitor, is promising because of its strong inhibitory effects on growth.26, 29, 48 However, unexpected side effects should be carefully determined in clinical trials, because compound 5 has never been used as a drug in humans. Vitamin K2 is used clinically in Japan to treat osteoporosis either alone or in conjunction with 1α,25-(OH)2 vitamin D315, 16 and has essentially no known adult human toxicity, making orally administered vitamin K2 a very promising option for chemoprevention or tumor dormancy. Clinical trials are in progress in our institution to evaluate vitamin K2 in patients with HCC. Future trials may determine the efficacy of vitamin K2 for other tumor types.


We would like to thank Mitsuko Tsubouchi for her excellent technical assistance. We are also grateful to many of our colleagues for their helpful discussion during the course of this work.