Potential conflict of interest: Nothing to report.
Supported in part by the Junta de Castilla y Leon (Grants BIO39/SA27/10, SA023A11-2 and SA070A11-2), Spain; the Ministerio de Ciencia y Tecnologia (Grants SAF2009-08493 and SAF2010-15517), Spain; the Fundacion Investigacion Medica Mutua Madrileña (Call 2009), Spain, and Fundacion Samuel Solorzano Barruso (Grant FS/1-2011), Spain. E.L. was supported by the AP2008-0376 PhD grant from the Junta de Castilla y Leon/Fondo Social Europeo and J.V. by the AP2007-00105 PhD grant from the Ministerio de Educacion (AP2007-00105), Spain. J.M.B. and L.B. received grants from the Spanish Association Against Cancer (AECC), the Basque Department of Industry (Saiotek), and the Carlos III Health Institute, Spain (FIS grant PI12/00380).
Address reprint requests to: Jose J.G. Marin, University of Salamanca, Department of Physiology and Pharmacology, Campus Miguel de Unamuno, E.D. S-09, 37007 - Salamanca, Spain. E-mail: firstname.lastname@example.org; fax: +34-923-294669.
Reduced drug uptake is an important mechanism of chemoresistance. Down-regulation of SLC22A1 encoding the organic cation transporter-1 (OCT1) may affect the response of hepatocellular carcinoma (HCC) and cholangiocarcinoma (CGC) to sorafenib, a cationic drug. Here we investigated whether SLC22A1 variants may contribute to sorafenib chemoresistance. Complete sequencing and selective variant identification were carried out to detect single nucleotide polymorphisms (SNPs) in SLC22A1 complementary DNA (cDNA). In HCC and CGC biopsies, in addition to previously described variants, two novel alternative spliced variants and three SNPs were identified. To study their functional consequences, these variants were mimicked by directed mutagenesis and expressed in HCC (Alexander and SK-Hep-1) and CGC (TFK1) cells. The two novel described variants, R61S fs*10 and C88A fs*16, encoded truncated proteins unable to reach the plasma membrane. Both variants abolished OCT1-mediated uptake of tetraethylammonium, a typical OCT1 substrate, and were not able to induce sorafenib sensitivity. In cells expressing functional OCT1 variants, OCT1 inhibition with quinine prevented sorafenib-induced toxicity. Expression of OCT1 variants in Xenopus laevis oocytes and determination of quinine-sensitive sorafenib uptake by high-performance liquid chromatography-dual mass spectrometry confirmed that OCT1 is able to transport sorafenib and that R61S fs*10 and C88A fs*16 abolish this ability. Screening of these SNPs in 23 HCC and 15 CGC biopsies revealed that R61S fs*10 was present in both HCC (17%) and CGC (13%), whereas C88A fs*16 was only found in HCC (17%). Considering all SLC22A1 variants, at least one inactivating SNP was found in 48% HCC and 40% CGC. Conclusion: Development of HCC and CGC is accompanied by the appearance of aberrant OCT1 variants that, together with decreased OCT1 expression, may dramatically affect the ability of sorafenib to reach active intracellular concentrations in these tumors. (Hepatology 2013;53:1065–1073)
Hepatocellular carcinoma (HCC) and cholangiocarcinoma (CGC) are important causes of cancer-related death worldwide. Although surgery is potentially curative for patients with localized disease, these tumors are often in an advanced stage at the time of diagnosis, when surgery is no longer the recommended approach. Alternative treatments have low or very low efficacy in advanced liver cancer. Regarding chemotherapy, HCC and CGC are among the tumors with the highest refractoriness. Although some drugs, such as doxorubicin, can achieve a partial effect in some cases, no relevant survival benefit has been obtained. Chemoresistance is often present before the treatment, but it can be further enhanced in response to the pharmacological challenge. Mechanisms of chemoresistance (MOCs) have been classified based on their role in drug uptake (MOC-1a) or efflux (MOC-1b), intracellular drug metabolism (MOC-2), changes in the expression/function of molecular targets (MOC-3), changes in the DNA repair machinery (MOC-4), reduced activation of apoptosis (MOC-5a), and enhanced expression/activity of antiapoptotic proteins (MOC-5b). MOCs may involve changes in the expression levels of specific proteins and the presence of genetic variants affecting their function.
One of the most promising strategies to overcome chemoresistance of primary liver cancer is the development of tyrosine kinase inhibitors (TKIs), such as sorafenib. This drug has been approved for the treatment of HCC, although the beneficial effect, regarding the inhibition of tumor progression and the enhancement of overall survival, is rather modest. Sorafenib has been reported to be effective in vitro against cells derived from CGC,[7, 8] although its efficacy in CGC patients is low. The mechanism of action of sorafenib depends on its access to the intracellular targets, which may be affected by changes in the expression and activity of transporters accounting for its uptake. The organic cation transporter-1 (OCT1, gene symbol SLC22A1), located at the basolateral membrane of healthy hepatocytes, is one of these transporters. OCT1 mediates the uptake of endogenous and exogenous organic cations, including drugs such as metformin, platinum derivatives, anthracyclines, and TKIs. The response to drugs whose hepatic uptake depends on this transporter, such as metformin, is affected by changes in OCT1 expression and by the appearance of less functional variants. In HCC and CGC, a decreased expression of OCT1 has been found.[3, 15] Moreover, a relationship between the presence of inactivating mutations in the SLC22A1 gene and a lower response to imatinib in patients with chronic myeloid leukemia has been reported. In the present study we investigated the expression of aberrant OCT1 variants in HCC and CGC and evaluated in vitro their potential impact on the sensitivity of these tumors to sorafenib.
Materials and Methods
Tumor samples from 23 HCC and 15 CGC (see patient and tumor information in Supporting Table 1) were obtained with written consent of patients from surgically removed tumors. None of these patients had received chemotherapy prior to the resection. The research protocol complied with the ethical guidelines of the 1975 Declaration of Helsinki and was reviewed and approved by the Human Subjects Committee of the University of Salamanca.
Table 1. Definition, Nomenclature, and Allelic Frequency of SLC22A1 Genetic Variants
Amino Acid Variation
Minor Allele Frequency (MAF)
NCBI SNP and
Ensembl databases. MAF refers to the frequency at which the less common allele occurs in a given population. Nucleotide positions refer to the open reading frame of the wildtype (Wt) NM_003057 sequence. Amino acid positions refer to the NP_003048 protein. Novel variants are underlined.
After retrotranscription (RT) of total RNA, the open reading frame (ORF) of SLC22A1 was amplified by polymerase chain reaction (PCR) with the high-fidelity AccuPrime-Pfx DNA polymerase (Life Technologies, Madrid, Spain) and gene-specific primers (Supporting Table 2). The amplicons were genotyped to detect OCT1 SNPs by gel-electrophoresis-based sequencing using gene-specific primers (Supporting Table 2) in an ABI PRISM-3100 Genetic Analyzer (Life Technologies).
Table 2. Screening of Inactivating Alternative Splice Variants and SNPs of OCT1 in Hepatocellular Carcinoma (HCC), Cholangiocarcinoma (CGC) and Adjacent Liver Tissue
Values are the proportion of samples of the same type but from different individuals that contained the genetic variant.
P < 0.05, on comparing mRNA OCT1 levels in HCC (n = 23) or CGC (n = 15) with that in adjacent liver tissue using paired t test.
Based on previous reports of alternatively spliced OCT1 variants, we designed primers annealing in exons 6 (Fw1) and 11 (Rv1) that are shared by all OCT1 isoforms (Supporting Table 2, Fig. 1). Analytical PCR was carried out with Platinum-Taq DNA polymerase (Life Technologies). The presence and size of the PCR products were determined by gel electrophoresis. Because sequencing of OCT1 ORF revealed the expression of novel spliced variants in HCC and CGC, additional Fw2 and Rv2 primers were used to confirm these findings (Supporting Table 2, Fig. 1). PCR carried out with Fw1 and Rv2 primers allowed us to detect an OCT1 variant lacking exon 10. The c.1276+1insGTAAGTTG mutation was detected using Fw2 and Rv1 primers.
From total RNA extracted from healthy human liver, the OCT1 ORF was amplified by RT-PCR and cloned into a pGEM-T-Easy vector using specific primers (Supporting Table 2), to which attB sites were added to obtain cDNA adapted for Gateway cloning (Life Technologies). The sequence of the wildtype OCT1 was confirmed and used to generate pGEM-T vectors containing the desired OCT1 variants (Table 1) by homemade site-directed mutagenesis. These plasmids were recombined with the pDONR221 vector to generate Entry plasmids, which were further recombined with a pcDNA3.1 destination vector to generate expression plasmids.
Cell Lines and Frog Oocytes
Human cell lines were obtained from ATCC (LGC Standards, Barcelona, Spain) (Alexander, SK-Hep-1, Caco-2, BeWo, Jar, and HEK-293), DSMZ (Braunschweig, Germany) (EGI-1, TFK1), and Health Protection Agency Culture Collections (Salisbury, UK) (COR-L23 and COR-L23/R). Partially chemoresistant cell lines LS 174T/R and WIF-B9/R were obtained as previously reported. Transient transfection was carried out with Lipofectamine LTX/PLUS reagent (Life Technologies). Transport studies were performed 2 days after transfection, as previously reported. [14C]-Tetraethylammonium bromide (TEA) (PerkinElmer, Barcelona, Spain) and quinine hydrochloride (Sigma-Aldrich, Madrid, Spain) were used as typical OCT1 substrate and inhibitor, respectively.
Mature female frogs (Xenopus laevis), purchased from Regine Olig (Hamburg, Germany), were used to obtain oocytes. The animals received humane care as outlined in the National Institutes of Health guidelines for the care and use of laboratory animals. Experimental protocols were approved by the Ethical Committee for Laboratory Animals of the University of Salamanca. Adaptations of previously published methods were used to determine by high-performance liquid chromatography-dual mass spectrometry (HPLC-MS/MS) sorafenib uptake by frog oocytes.
Total RNA was used as template to determine OCT1 expression by RT-QPCR using gene-specific primers spanning exon-exon junctions in the target mRNA (Supporting Table 2) and AmpliTaq Gold DNA polymerase in a 7500 Real-Time PCR System (Life Technologies). The screening of novel SNPs was carried out with primers specific for the mutated sequence (Supporting Table 2). Detection of amplicons was carried out using SYBR Green I. The abundance of OCT1 mRNA in each sample was normalized on the basis of its GAPDH content.
Immunostaining was carried out in cells fixed and permeabilized in ice-cold methanol using an antibody against V5 (Life Technologies) diluted 1:600 in 2% fetal calf serum in phosphate-buffered saline (PBS), and Alexa Fluor-488 antimouse IgG secondary antibody (1:1,000) (Life Technologies). Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI). Confocal laser-scanning microscopy was performed using a Zeiss LSM 510 confocal microscope.
Effect of Sorafenib on Cell Viability
Cells were seeded onto 96-well plates at subconfluence. After 24 hours the cells were transfected and 48 hours later exposed to sorafenib (Pharmacy Department, University Hospital, Salamanca, Spain) for the indicated time period. The formazan test from thiazolyl blue tetrazolium bromide (Sigma-Aldrich) was used to determine cell viability.
Protein Modeling and Statistical Methods
Tertiary structures were predicted on the web by Phyre2 server. Results are expressed as mean ± standard deviation (SD) from at least three different cultures carried out in triplicate. To calculate the statistical significance of the differences the paired t test was used.
Detection of OCT1 Variants in Liver Tumors
Complete sequencing of SLC22A1 cDNA obtained from 12 HCC (Supporting Table 3) and 9 CGC (Supporting Table 4) biopsies revealed the presence of several already described alternative spliced variants (Fig. 1) and SNPs (Fig. 2A).[17, 23-25] In addition, novel variants were identified. In some cases the presence of a single sequence suggested both homozygosity and homogeneity of the sample regarding the population of cells expressing OCT1. In contrast, both the wildtype and the variant sequence were frequently detected together. In paired nontumor tissue the presence of these SNPs was less common (Table 2).
Already known OCT1 variants that result in truncated proteins, through the loss of one or more exons and/or intron retention, have been reported to be nonfunctional.[17, 26] In contrast, SNPs may have different consequences on OCT1 function. To investigate this question plasmids containing wildtype or mutated OCT1 ORF were transfected into Alexander and SK-Hep-1 cells of hepatocellular origin, and TFK1 cells derived from CGC. These cells were selected from a panel of cells of different origin based on three criteria: (1) low endogenous OCT1 expression, (2) integrity of the mechanism accounting for sorafenib sensitivity, and (3) being derived from human liver tumors (Fig. 3).
Functional Evaluation of OCT1 Variants
Expression of wildtype OCT1 induced quinine-sensitive TEA uptake by HCC and CGC cells (Fig. 4A-C). This ability was also observed in S14F, L160F, G401S, and P197S variants, whereas it was partly or completely lost in the rest of detected variants. To validate this transport assay, wildtype OCT1 was also expressed in frog oocytes. This maneuver markedly enhanced their ability to take up, in a quinine-sensitive manner, both TEA (Fig. 4D) and sorafenib (Fig. 4E). Moreover, the expression of the novel variants in this model also confirmed the lack of ability of R61S fs*10 and C88A fs*16 to transport sorafenib, which was maintained in P197S (Fig. 4E).
The effect of SNPs on the targeting to the plasma membrane was investigated by immunodetection of the V5-tag placed in the constructs. In this set of experiments, we also included C88R and S189L, whose effects on protein targeting were not known, and G465R, whose functional consequences are controversial. Although G465R has been described as a loss-of-function variant, our results indicate that when expressed in HCC and CGC cells this variant has a reduced, but not abolished, OCT1-mediated transport (Fig. 4A-C). When G465R was investigated in Alexander cells, similarly to wildtype OCT1, it was targeted to the plasma membrane (Fig. 5). In contrast, both C88R and S189L were mainly localized intracellularly (Fig. 5). This was consistent with the abolished ability of the latter two variants to mediate TEA uptake (Fig. 4).
Regarding the novel OCT1 variants, both R61S fs*10 and C88A fs*16, which encode truncated proteins (Fig. 2B), resulted in impaired targeting to the plasma membrane (Fig. 5) and lack of the ability to mediate sorafenib uptake by oocytes (Fig. 4E) and TEA uptake by transfected cells (Fig. 4A-C). In contrast, the functional variant P197S resulted in an entire OCT1 protein targeted to the plasma membrane (Fig. 5).
Relationship Between OCT1 Transport Ability and Sorafenib Sensitivity
Based on studies addressing the dose- (Fig. 3) and time- (Supporting Fig. 1) dependent sensitivity of Alexander cells to sorafenib, short-term (6 hours) exposure of HCC and CGC cells to sorafenib was carried out (Fig. 6). Under these conditions only OCT1 variants with a relatively well-preserved ability to mediate TEA transport (Fig. 4) were able to induce sensitivity to sorafenib in all cells assayed (Fig. 6). Regarding the SNPs identified here, P197S, but not R61S fs*10 or C88A fs*16, enhanced the sensitivity to sorafenib in cells expressing these variants (Fig. 6). Interestingly, OCT1 inhibition with quinine reduced, in a dose-dependent manner, the sensitivity to sorafenib due to the expression of functional variants of this transporter.
Screening of OCT1 Inactivating SNPs in Liver Tumors
Selective identification of loss-of-function SNPs was performed by RT-QPCR in a larger series of HCC and CGC biopsies (Table 2). The abundance of each variant was normalized by the abundance of total OCT1 mRNA. We confirmed previous findings regarding the reduced expression of OCT1 in HCC and CGC as compared with adjacent paired tissue. The frequency of R61S fs*10 in this limited series of HCC and CGC was 17% and 13%, respectively, whereas C88A fs*16 was only found in HCC (17%). Previously described inactivating SNPs, whose minor allele frequency has been calculated in larger populations (Table 1), appeared with different frequency in HCC and CGC (Table 2). When all OCT1 variants were considered together, the result was that at least one inactivating SNP was present in 48% HCC and 40% CGC.
Sorafenib is a very active antitumor drug in most cancer cell lines, which include those derived from CGC and HCC.[7, 8] Unfortunately, the efficacy of this drug in clinical oncology is very different. Indeed, regimens that have incorporated this drug are far from optimal because a marked refractoriness to sorafenib is an initial characteristic of liver tumors. Moreover, cancer cells often activate MOCs during treatment. Regarding the refractoriness to sorafenib, the identified MOCs include: (1) up-regulation of ABC proteins, such as MDR1 and ABCG2, which reduce intracellular drug content (MOC-1b); (2) enhanced drug inactivation by uridine glucuronosyl transferase 1A (MOC-2); (3) the appearance of genetic variants in the intracellular targets of sorafenib (MOC-3); and (4) since sorafenib uptake is an essential requirement to be effective against tumor cells, changes in the expression/activity of the transporters involved in sorafenib uptake can also lead to drug resistance (MOC-1a). In this regard, OCT1 has been reported to be involved in sorafenib uptake by hepatocytes. This and other carrier systems may account for sorafenib uptake by tumor cells. Thus, the present study indicates that sorafenib has a strong effect even on cell lines with very poor expression of OCT1.
In agreement with previous studies, we observed here a marked reduction in OCT1 expression in both HCC and CGC. In the case of HCC, this event may be at least partially due to an enhanced methylation and reduced activity of the SLC22A1 promoter. OCT1 down-regulation has already been associated with chemoresistance in certain types of cancer, for instance, to cisplatin. Moreover, OCT1 expression levels have been suggested to be a useful biomarker to predict the success of imatinib-based therapy for chronic myeloid leukemia. The present study suggests that reduced OCT1-mediated sorafenib uptake may be involved in a poorer response to this drug.
The functional consequences of some OCT1 SNPs found in HCC and CGC have already been studied. M408V and L160F variants, with relative high frequency in HCC and CGC, have been reported to maintain transport ability. Although a trend to lower OCT1 expression has been reported in the livers from patients harboring the M408V variant, its impact on the clinical efficiency of metformin is minor. Patients with chronic myeloid leukemia harboring the wildtype genotype GG of the c.480G>C (L160F) variant have a poorer response than patients with the mutation. The genetic variants R61C, C88R, S189L, G220V, and R287G are known to reduce the transport of typical substrates, such as metformin and methylpyridinium. An important role of some highly conserved glycine residues has been suggested. Indeed, G220V and G465R induced in transfected cells a lower ability to take up TEA. Surprisingly, although the G220V variant induced in the three cell lines a poorer sensitivity to sorafenib, this was almost unaffected by G465R. This supports the concept that pharmacokinetic consequences of OCT1 genetic variants may differ depending on the substrate. Thus, P283L and P341L, two variants not found here in liver tumors, do not affect metformin uptake, but do reduce that of methylpyridinium[23, 24] and lamivudine. Moreover, S14F variant has an impaired ability to take up metformin, whereas the transport of methylpyridinium by this variant is enhanced. In the present study, TEA transport and OCT1-induced sorafenib sensitivity were not impaired by S14F.
G401S and G465R have been described as loss-of-function variants when transfected in MDCK cells. In our experimental settings G465R reduced, but did not abolish, and G401S did not affect both TEA uptake and OCT1-induced sorafenib sensitivity. This apparent discrepancy may be due to differences in protein targeting. In the present study V5-tagged G465R was clearly targeted to the plasma membrane of Alexander cells, whereas GFP-tagged G465R was poorly incorporated to the plasma membrane of MDCK cells.
Among the novel SNPs, P197S maintains the transport ability, probably because this change induces a conservative substitution between two neutral amino acids. In contrast, R61S fs*10 and C88A fs*16 induce frameshifts resulting in truncated nonfunctional proteins.
Altered exon skipping and intron retention mechanisms account for novel alternative spliced variants found in HCC and CGC. Short and nonfunctional OCT1 isoforms resulting from alternative splicing have also been found in glioma cells. Some of these truncated variants have been associated with altered pharmacokinetics of OCT1 substrates, such as metformin.
In conclusion, HCC and CGC development is accompanied by OCT1 down-regulation together with the appearance of genetic variants that may affect the ability of these tumors to take up and hence respond to sorafenib. Moreover, it should be considered that OCT1 is also involved in the pharmacokinetic of other antitumor drugs such as cisplatin, irinotecan, mitoxantrone and paclitaxel. These findings suggest the potential impact of an appropriate selection of HCC and CGC patients suitable for treatment with sorafenib.
The authors thank Nicholas Skinner for revision of the English spelling, grammar, and style of the article.