• Open Access

Expression of SLCO1B3 is associated with intratumoral cholestasis and CTNNB1 mutations in hepatocellular carcinoma


To whom correspondence should be addressed. E-mail: ssekine@ncc.go.jp


Recent studies have shown that intratumoral cholestasis is a hallmark of CTNNB1 mutations in hepatocellular carcinomas (HCC). Here, we analyzed the expressions of genes involved in bile acid and bilirubin metabolism and their correlation with the mutational status of CTNNB1 in a series of HCC. The expressions of CYP7A1 and CYP27A1, which encode rate-limiting enzymes in bile acid synthesis, were unaltered or only marginally increased in CTNNB1-mutated HCC compared with those in HCC with wild-type CTNNB1. Among the genes involved in bile acid and bilirubin transport, the expression of SLCO1B3 was significantly elevated in HCC with CTNNB1 mutations, whereas the expression of ABCC4 was elevated in HCC with wild-type CTNNB1. Immunohistochemistry confirmed the frequent expression of SLCO1B3 in CTNNB1-mutated HCC at the protein level, but not in most HCC with wild-type CTNNB1. Immunohistochemistry for MRP4 (encoded by ABCC4) partly agreed with ABCC4 expression, but most cases did not express detectable levels of MRP4. Notably, all HCC with bile accumulation, including those without CTNNB1 mutations, expressed SLCO1B3, suggesting that SLCO1B3 expression, rather than CTNNB1 mutation, is the critical determinant of intratumoral cholestasis. As SLCO1B3 is involved in the uptake of a number of chemotherapeutic and diagnostic agents, SLCO1B3 expression and the status of CTNNB1 mutation might need to be considered in the drug delivery to HCC. (Cancer Sci 2011; 102: 1742–1747)

Bile acids are major components of bile and the liver plays a central role in their metabolism.(1–3) Bile acids are synthesized from cholesterol in the liver and secreted into bile. Bile plays essential roles in the absorption and excretion of lipid-soluble substances. In addition to widely recognized roles in lipid metabolism, bile acids act as a signaling molecule and modulate proliferation and energy metabolism in hepatocytes.(3,4) Mice deficient in the bile acid receptor FXR exhibit cholestasis and the spontaneous development of hepatocellular carcinomas (HCC), suggesting a potential linkage between bile acid signaling and hepatocarcinogenesis.(5,6)

Interestingly, recent reports have shown that HCC with activating CTNNB1 mutations frequently exhibit cholestasis.(7,8) This observation implies that β-catenin regulates bile metabolism in HCC. Activating mutations of CTNNB1, encoding β-catenin, are present in 30–40% of HCC.(9,10) Physiologically, β-catenin acts as a transducer of the Wnt signaling pathway,(11) and mutation of CTNNB1 leads to abnormal accumulation of β-catenin and constitutive activation of T-cell factor (TCF)-dependent transcription.(12) This results in the overexpression of β-catenin/TCF-regulated transcriptional targets in CTNNB1-mutated tumors and the promotion of tumorigenesis.

Based on these previous reports, we suspected that mutated β-catenin might coordinately induce genes critically involved in bile metabolism in HCC. To elucidate this issue, we examined the expressions of a list of genes that are involved in bile acid synthesis and transport in a series of HCC.

Materials and Methods

Cases.  We examined 44 cases of HCC obtained from 42 patients; all of these tumors had been previously analyzed for CTNNB1 mutations and the presence of intratumoral cholestasis.(8) Mutation analysis was done by direct sequencing of the N-terminal region of CTNNB1 using cDNA samples. Intratumoral cholestasis was histologically determined by the presence of bile pigments on hematoxylin–eosin and Hall’s bile acid staining. Eight non-tumoral liver tissues obtained during the resection of metastatic colorectal cancers were used as normal liver samples for comparison. All tissue samples were obtained at the National Cancer Center Hospital, Tokyo, Japan. The present study was approved by the Ethics Committee of the National Cancer Center, Tokyo, Japan.

Quantitative RT-PCR.  RNA extraction and reverse-transcription reactions were performed using standard protocols. Quantitative RT-PCR reactions were performed using SYBR Green PCR master mix (Applied Biosystems, Foster City, CA, USA). The expression level of each gene was determined using GUSB as a standard, as previously described.(13) The primer sequences are shown in Table 1.

Table 1.   Primers used in the RT-PCR analysis
 Forward primerReverse primer

Immunohistochemistry.  Among the tumors subjected to RT-PCR analysis, 41 lesions were available for histological analysis. Liver tissue samples were fixed in 10% buffered formalin, embedded in paraffin and cut into 4-μm-thick sections. Antigen retrieval was performed by autoclaving in 10 mmol/L of citrate buffer (pH 6.0) for 10 min. Anti-SLCO1B3 (1:250 dilution; Sigma, St Louis, MO, USA) and anti-ABCC4 antibodies (1:500 dilution; Abnova, Taipei, Taiwan) were used as the primary antibodies and the signals were detected using peroxidase-labeled anti-rabbit and anti-goat polymers (Histofine simple stain; Nichirei, Tokyo, Japan). 3-3′-Diaminobenzidine tetrahydrochloride was used as a chromogen. Normal liver tissue served as a positive control for SLCO1B3 and normal prostatic tissue was used as a positive control for ABCC4. The staining results were evaluated as follows: ++, diffuse (>50%) expression; +, focal (10–50%) expression; and −, no (<10%) expression.

Statistical analysis.  For quantitative PCR analysis, statistical significance was confirmed using a two-tailed Mann–Whitney U-test. The Fisher–Freeman–Halton exact test was used to analyze each 2 × 3 table. < 0.05 was considered statistically significant.


We first determined the expressions of genes encoding enzymes critical for bile acid synthesis, postulating that CTNNB1-mutated HCC show increased bile acid production. CYP7A1 and CYP27A1 are the rate-limiting enzymes of the classical and alternative pathways of bile acid synthesis, respectively.(3) HSD3B7 is another critical enzyme in bile acid synthesis and its mutation has been linked to a defect in bile acid synthesis.(14) The results showed that CYP7A1 expression was increased in the HCC regardless of the CTNNB1 mutation status (Fig. 1). CYP27A1 expression was significantly but only marginally elevated in HCC with CTNNB1 mutations compared with those without mutation. The HSD3B7 expression level was unaltered between normal liver and the HCC. These results indicate that CTNNB1 mutations do not induce genes for bile acid synthesis in HCC.

Figure 1.

 Expression of genes related to bile acid metabolism. Box plots of the expression of genes related to bile acid metabolism in hepatocellular carcinomas (HCC). The expression of each gene was determined using quantitative RT-PCR with GUSB used as a reference. N, normal liver; Mt-, HCC with wild-type CTNNB1; Mt+, CTNNB1-mutated HCC.

Next we examined four genes involved in bile acid and bilirubin uptake from portal blood. The expressions of three of the genes that were examined, SLC10A1, SLCO1A2 and SLCO1B1, tended to be reduced in HCC, and no significant associations with the status of CTNNB1 mutation were seen (Fig. 1). However, SLCO1B3 expression was closely correlated with the presence of CTNNB1 mutations. While HCC without CTNNB1 mutations showed remarkably reduced SLCO1B3 expression levels, CTNNB1-mutated HCC retained expression levels comparable with that observed in normal liver.

Two genes for canalicular transporters, ABCC2 (encoding MRP2) and ABCB11 (encoding BSEP), showed a modest increase in CTNNB1-mutated tumors. Between two basolateral efflux transporters, ABCC3 (encoding MRP3) and ABCC4 (encoding MRP4), ABCC4 was significantly elevated in HCC with wild-type CTNNB1. Of note, expression of NR0B2, which mediates the feedback regulation of bile acid signaling,(15,16) did not differ between HCC with or without CTNNB1 mutations.

The expressions of SLCO1B3 and MRP4 were further determined at the protein level using immunohistochemistry. SLCO1B3 expression in the HCC was membranous and consistent with the results of the RT-PCR analysis; the expression of SLCO1B3 was significantly correlated with the presence of CTNNB1 mutations (Fig. 2, Table 2). SLCO1B3 was expressed in pericentral hepatocytes with a membranous pattern in normal liver (Fig. 3).

Figure 2.

 SLCO1B3 expression, gross morphology and histology in hepatocellular carcinomas (HCC). Immunohistochemistry for SLCO1B3 (A–D), gross morphology (E,F) and histology (G,H) in a SLCO1B3-expressing HCC (A,C,E,G) and a SLCO1B3-negative HCC (B,D,F,H). Low-power views of SLCO1B3 staining (A,B). Areas of the tumor are indicated by arrowheads. Extensive SLCO1B3 expression in a case of cholestatic HCC (A) and almost completely negative staining in a case of non-cholestatic HCC (B). Focal SLCO1B3 expression is observed in non-neoplastic cirrhotic liver on the backgrounds of HCC (A,B). Magnified views showed membranous SLCO1B3 expression (C), and no SLCO1B3 staining (D). A SLCO1B3-expressing HCC has a greenish cholestatic appearance (E), whereas a SLCO1B3-negative HCC has a homogenous whitish appearance (F). A HE-stained section shows bile pigments in a SLCO1B3-expressing HCC (G) but not in a SLCO1B3-negative HCC (H).

Table 2.   Correlations among SLCO1B3 expression, CTNNB1 mutation and cholestasis
 TotalSLCO1B3 immunohistochemistryP-value
  1. ++, diffuse expression; +, focal expression; −, no expression.

CTNNB1 mutation
 Present1811436.8 × 10−4
 Present1512306.4 × 10−8
Figure 3.

 SLCO1B3 expression in normal liver tissue. Normal liver tissue shows pericentral SLCO1B3 expression (A). A magnified view showing membranous expression of SLCO1B3 in normal hepatocytes (B). CV, central vein; P, portal tract.

Immunohistochemistry for MRP4 showed diffuse expression in one case and focal staining in three cases (Figs 4,5). All four cases positive for MRP4 also showed high levels of ABCC4 expression, indicating concordance between the mRNA and protein expression levels. However, negative or faint expression of MRP4 was observed in the other HCC. Non-neoplastic liver tissue did not express immunohistochemically detectable levels of MRP4. Prostatic tissues used for positive controls exhibited diffuse and strong membranous expression.

Figure 4.

 MRP4 expression in hepatocellular carcinomas (HCC). Most tumor cells show membranous expression of MRP4 with some heterogeneity in this tumor (A), whereas the majority of HCC did not express immunohistochemically detectable levels of MRP4 (B).

Figure 5.

 Expressions of SLCO1B3, ABCC4 and AXIN2 mRNA, SLCO1B3 and MRP4 protein, CTNNB1 mutation status and intratumoral cholestasis in each tumor sample. The expressions of SLCO1B3, ABCC4 and AXIN2 were determined using quantitative RT-PCR. SLCO1B3 and MRP4 expressions were determined using immunohistochemistry (IHC). The results of the CTNNB1 mutation analysis and for intratumoral cholestasis were obtained in our previous study.(8) The case numbers are identical to those in our previous study.

Next, we sought to determine the correlation between SLCO1B3 expression and intratumoral cholestasis. The results showed that HCC with SLCO1B3 expression frequently showed bile accumulation (Figs 2,5, Table 2). Remarkably, all three CTNNB1 mutation-negative, cholestatic HCC expressed SLCO1B3, implying that the presence of bile accumulation is more closely correlated with SLCO1B3 expression than the mutational status of CTNNB1.

While CTNNB1-mutations affecting N-terminal regions of β-catenin is the common cause of activation of β-catenin signaling in HCC, β-catenin signaling could potentially be activated by uncommon genetic alterations such as atypical CTNNB1 mutations or APC mutations.(17,18) To exclude this possibility, we examined the expression of AXIN2, a ubiquitous target of β-catenin/TCF.(19,20) As expected, the expression of AXIN2 was upregulated in CTNNB1-mutated HCC, but the levels of AXIN2 expression were not significantly elevated in any of the CTNNB1 mutation-negative HCC with SLCO1B3 expression (Fig. 5). This finding indicates that a minor subset of HCC express SLCO1B3 even in the absence of active β-catenin signaling.


Based on the association between CTNNB1 mutations and intratumoral cholestasis,(7,8) we postulated that active β-catenin signaling regulates bile acid metabolism in HCC. While previous analysis in a mouse model suggested that β-catenin induces bile acid synthesis genes under physiological conditions,(21) they were not upregulated in HCC with CTNNB1 mutations. In contrast, the expression of SLCO1B3, a solute carrier organic anion transporter protein, was associated with the presence of CTNNB1 mutations and more closely with intratumoral cholestasis. SLCO1B3 is physiologically involved in the uptake of bile acids;(22–26) however, CTNNB1-mutated HCC did not exhibit elevated NR0B2 levels, a hallmark of active bile acid signaling.(15,16) These findings suggest that the cholestatic appearance of HCC is not linked to an increase in bile acid synthesis or uptake.

The exact mechanism by which mutated β-catenin induces SLCO1B3 remains elusive. While we performed in vitro studies using several HCC cell lines, the activation of β-catenin signaling did not induce SLCO1B3 expression in any of the cell lines (data not shown). Furthermore, some of the HCC expressed high levels of SLCO1B3 in the absence of CTNNB1 mutations. These observations imply the presence of β-catenin-independent regulation of SLCO1B3 in some HCC.

MRP4 is a basolateral transporter involved in the efflux of bile acids, steroids and a range of xenobiotic substances.(27)ABCC4, encoding MRP4, was significantly upregulated in HCC with wild-type CTNNB1. However, the MRP4 protein was expressed at low levels in most of the HCC compared with prostatic tissue, where MRP4 is physiologically expressed. While a significant correlation was observed between ABCC4 expression and CTNNB1 mutation in HCC, the functional significance remains to be elucidated.

Bilirubin, the main bile pigment, is another important substrate of SLCO1B3. Previous in vitro experiments have shown that the introduction of SLCO1B3 into human cells or xenopus oocytes induced bilirubin uptake.(24,25) Furthermore, two genome-wide association studies identified genetic variations within SLCO1B3 as being associated with serum bilirubin levels,(28,29) suggesting a physiological role in bilirubin clearance in vivo. As the green color of bile is caused by its bilirubin content, it is reasonable to assume that the cholestatic appearance of HCC mainly reflects their ability to uptake bilirubin. While some previous studies reported conflicting results on the correlation between intratumoral cholestasis and SLCO1B3 expression,(30–32) our data suggest that expression of SLCO1B3 is the major determinant of intratumoral cholestasis in HCC.

Eight cases of SLCO1B3-positive HCC without cholestasis were observed. In fact, SLCO1B3 is a bidirectional carrier, and the efflux of bilirubin is reduced by binding to glutathione S-transferase.(33) Furthermore, some transporters, such as MRP3, can export bilirubin. Thus, SLCO1B3 expression is a critical, but not the sole, determinant of bilirubin accumulation in cells. It is expected that some molecules involved in bilirubin transport, other than SLCO1B3, are expressed differently in SLCO1B3-positive HCC without cholestasis.

A number of chemotherapeutic and diagnostic agents, in addition to bile acids and bilirubin, are also known as substrates of SLCO1B3(26,34–36) For example, gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA), an increasingly used magnetic resonance imaging contrast agent, is also a substrate of SLCO1B3.(35) The majority of HCC are depicted as hypointense areas during the hepatobiliary phase of Gd-EOB-DTPA-enhanced magnetic resonance imaging as HCC generally have a decreased capacity to take up this contrast agent. However, a subset of HCC that express high levels of SLCO1B3 can be detected as iso- or hyperintense masses.(30–32) Considering these previous and current observations, a significant proportion of Gd-EOB-DTPA-accumulating HCC might harbor CTNNB1 mutations. The present observations suggest that SLCO1B3 expression and the status of CTNNB1 mutation might need to be considered in drug delivery to HCC.


The authors thank Mr Shigeru Tamura for photographic assistance. This work is supported by a grant from the Takeda Science Foundation and a grant for Research on Publicly Essential Drugs and Medical Devices from the Japan Health Science Foundation.

Disclosure Statement

The authors have no conflict of interest.