Abstract
The aim of this study was to investigate the gene and protein expression profiles of important drug-transporting proteins in human cell lines commonly used for studies of drug transport mechanisms. Human cell lines used to transiently or stably express single transporters [HeLa, human embryonic kidney (HEK) 293] and leukemia cell lines used to study drug resistance by ATP-binding cassette transporters (HL-60, K562) were investigated and compared with organotypic cell lines (HepG2, Saos-2, Caco-2, and Caco-2 TC7). For gene expression studies, real-time polymerase chain reaction was used, whereas monospecific polyclonal antibodies were generated and used to investigate protein expression by immunohistochemistry. Thirty-six transporters were studied for gene expression, and nine were studied for protein expression. The antibodies were validated using expression patterns in human tissues. Finally, the function of one ubiquitously expressed transporter, MCT1/SLC16A1, was investigated using [14C]lactic acid as a substrate. In general, the adherent cell lines (HeLa, HEK293) displayed low transporter expression, and the expression patterns were barely affected by transfection. The leukemia cell lines (K562, HL-60) and Saos-2 also had low endogenous transporter expression, whereas the organotypic cell lines (HepG2 and Caco-2) showed higher expression of some transporters. Comparison of gene and protein expression profiles gave poor correlations, but better agreement was obtained for antibodies with a good validation score, indicating that antibody quality was a significant variable. It is noteworthy that the monocarboxylic acid-transporting protein MCT1 was significantly expressed in all and was functional in most of the cell lines, indicating that MCT1 may be a confounding factor when the transport of small anionic drugs is investigated.
One problem with using cell lines in studies of a specific transporter function is that the results may be confounded by endogenous transporter activity (Goh et al., 2002). Only in rare cases has the background expression of endogenous transporters been investigated in such cell lines (Hilgendorf et al., 2007). Furthermore, because many traditionally used cell lines are of polyclonal origin, variations in gene expression and regulation may be a problem (Hayeshi et al., 2008). Another limitation is that most investigations of transporter expression have been conducted at the mRNA level. Although such studies may give a hint of protein expression, in particular for highly expressed transporter genes (Landowski et al., 2004), post-transcriptional regulatory mechanisms and variations in mRNA and protein stability will result in discrepancies between gene and protein expression (e.g., Tian et al., 2004). To the best of our knowledge, studies on protein expression patterns of transporters in cell lines have not yet been investigated, although techniques that eventually will allow such studies are becoming available (e.g., Kamiie et al., 2008). Western blotting will only mirror the functional transporter as long as it is correctly folded and sorted to the plasma membrane. For example, in some cases the ATP-binding cassette (ABC) transporter multidrug resistance protein (MRP) 2/ABCC2 will not be sorted to the cell membrane in human embryonic kidney (HEK) 293 cells in sufficient amounts (Keitel et al., 2003), which is a normal mechanism for short-term regulation of MRP2 (Sekine et al., 2008). The membrane constitution may also affect the function of the transporter; for example, breast cancer resistance protein (BCRP)/ABCG2 activity differs in insect Sf9 cells that lack cholesterol and eukaryotic HEK293 cells (Pál et al., 2007). Only in a few cases has the transporter gene or protein expression been correlated to the transporter function (e.g., Taipalensuu et al., 2004).
Transformed human cell lines are widely used tools in research on drug transport mechanisms. Common adherent cell lines that are easy to maintain, such as the HEK293 cell line generated through transformation with adenovirus type 5 (Graham et al., 1977) and the human cervix epithelioid carcinoma HeLa (Gey et al., 1952) cells, can easily be transiently or stably transfected with an uptake or efflux transporter of interest, as exemplified by selected references (Zhang et al., 1998; Ahlin et al., 2008). Suspension cell lines, often originating from various tumors of the immune system, such as K562 from chronic myeloid leukemia (Gahmberg and Andersson, 1981) and HL-60 from acute promyelocytic leukemia (Gallagher et al., 1979), are used to study drug resistance mechanisms mediated by ABC transporters (Puhlmann et al., 2005; Assef et al., 2009). Furthermore, the differentiated cell lines, such as the colorectal carcinoma cell line Caco-2 (Fogh et al., 1977), the hepatocellular carcinoma cell line HepG2 (Aden et al., 1979), and the human osteosarcoma cell line Saos-2 (Fogh et al., 1977), are used as organotypic cell models for the intestinal epithelium, hepatocytes, and osteoblasts, respectively (Rodan et al., 1987; Artursson and Karlsson, 1991; Maruyama et al., 2007).
In this study, we quantified the gene expression for 36 transporters [10 ABC, 25 solute carrier transporter (SLC), and the human peptide transporter 1/CDH17] in eight cell lines commonly used in research on transport mechanisms. We also investigated how environmental differences in, for example, culture conditions, and transient and stable overexpression of a transporter influenced the gene expression. In parallel, as a part of the Human Proteome Resource (HPR) Project (Uhlén et al., 2005), we generated monospecific polyclonal antibodies (msAb) for nine transporters, which we validated in human tissues and used with the intention of examining expression in six of the cell lines. Finally, we investigated the function of the highly ubiquitously expressed MCT1 in these cell lines.
Materials and Methods
Cell Lines.
Human cell lines were used throughout this study. The adherent cell lines, HeLa derived from an individual with cervical cancer (Gey et al., 1952) and HEK293 derived from human embryonic kidney cells (Graham et al., 1977), and the suspension grown cell lines, HL-60 derived from leukocytes of an individual with acute promyelocytic leukemia (Gallagher et al., 1979) and K562 derived from a chronic myeloid leukemia (Gahmberg and Andersson, 1981), were included in the study. Four differentiated cell lines were also studied. The Saos-2 cell line was derived from osteosarcoma in human bone (Fogh et al., 1977) and is used to study osteoblastic properties and for transfection purposes (Rodan et al., 1987; Matsson et al., 2007). The gene expression data used for Saos-2 empty vector cells have been published previously by our group (Matsson et al., 2007). Caco-2 and Caco-2 TC7 cell lines were derived from human colon cancer (Fogh et al., 1977) and are extensively used as in vitro models for the human intestine and to predict drug uptake over the intestinal barrier (Artursson and Karlsson, 1991). HepG2 cells were derived from hepatocellular carcinoma and are extensively investigated for their hepatocyte-like characteristics (e.g., Maruyama et al., 2007). The gene expression data used for HepG2 cells have been published previously by Hilgendorf et al. (2007).
Culture Conditions.
mRNA and function.
The trypsin solutions contained 0.05% trypsin (Invitrogen, Carlsbad, CA) and 0.002 to 0.02% EDTA (MP Biomedicals, Solon, OH). The adherent cell lines, unless otherwise stated, were harvested at 70 to 90% confluence on plastic support. The suspension cells (HL-60 and K562) were harvested directly from suspension after 72 h in culture. The cells used for real-time polymerase chain reaction (RT-PCR) and functional studies were cultured in normal cell culture flasks according to protocols from respective distributors.
Protein.
The cells used for immunohistochemistry (IHC) were cultured in normal cell culture flasks according to protocols from respective distributors. The cells were harvested, and agarose cell gels were prepared for IHC (Andersson et al., 2006).
Gene Expression Profiles.
The cell lines were harvested, and RNA was isolated from the samples using the RNeasy mini-kit (QIAGEN GmbH, Hilden, Germany) according to the protocol provided from the manufacturer, with the addition of an extra on-column DNase step (QIAGEN GmbH). RNA concentration was measured using a NanoDrop (Wilmington, DE) ND-1000 Spectrophotometer, and the RNA quality was examined using a bioanalyzer (Agilent Technologies, Santa Clara, CA). RNA was converted into cDNA using a high-capacity cDNA archive kit (Applied Biosystems, Foster City, CA). Five hundred nanograms of the total RNA samples was added to a master mixture containing 10 μl of 10× RT buffer, 4 μl of 25× deoxynucleoside-5′-triphosphate, 10 μl of 10× random primers, 5 μl of Multiscribe RTase (50 U/μl; Applied Biosystems), and 21 μl of nuclease-free water. The reverse transcriptase PCR mixture was incubated at 25°C for 10 min and at 37°C for 120 min in a Mastercycler gradient system (Eppendorf AG, Hamburg, Germany). The output cDNA (5 ng/μl) was stored at −70°C pending TaqMan analysis. Analyses of the cDNA samples were performed as described previously (Hilgendorf et al., 2007). The RT-PCR analysis was performed at least in duplicate with 1 to 2 ng cDNA/well.
Relative Expression Analysis.
The relative expression analyses were performed as described previously (Hilgendorf et al., 2007). To obtain a better overview of the gene expression and to simplify comparison with protein expression data, the relative expression levels were classified into four groups (negative, weak, moderate, strong), as described previously (Hilgendorf et al., 2007). Owing to the very low relative expression of many transporters, all the genes not reaching the set threshold level later than 35 cycles were said to be negative.
Immunohistological Screening of Human Drug Transporters.
The protein expression data for the transporters were generated as a part of the large HPR project, with the aim of mapping protein expression of the whole human genome in human tissues, tumors, and cell lines (Uhlén et al., 2005). In brief, first two protein epitope signature tags (PrEST), consisting of a 50 to 150 amino acid sequence unique to the specific protein, were identified and expressed as recombinant proteins as described previously (Nilsson et al., 2005; Pontén et al., 2008). Each PrEST was injected subcutaneously in New Zealand rabbits to produce an immune response. The resulting antibodies were affinity-purified from serum by depletion of tag-specific antibodies, followed by purification of monospecific antibodies using affinity columns loaded with the protein-specific PrESTs (Nilsson et al., 2005). Quality assurance was performed by 1) sequence verification of the PrEST clone, 2) analyzing the size of the resulting recombinant protein to ensure that the correct antigen has been produced and purified, and 3) checking the antibodies for cross-reactivity to PrESTs spotted on protein arrays (Pontén et al., 2008).
Internal validation of the antibodies was performed by 1) comparing IHC staining patterns of the msAb with the staining patterns of commercial antibodies and/or with experimental/bioinformatics data for a subset of 13 different normal tissues, 4 different tumor tissues, and 8 cell lines; 2) investigating the consistency of immunofluorescence images with previously published protein localization and/or bioinformatics data; and 3) investigating the presence, size, and strength of bands on Western blot gels (Berglund et al., 2008; Pontén et al., 2008). An external validation step was performed, in which tissue staining patterns were compared with data from the literature on transporter expression and localization.
To allow a high-throughput staining, the cell lines were subcultured and agarose cell gels were prepared, which were used to produce tissue microarrays containing approximately 450 cells each. The cell microarrays were IHC-stained in duplicate (Andersson et al., 2006), and the resulting images were annotated using an automated image analysis application (Strömberg et al., 2007). The staining patterns were divided into five groups (not representative, negative, weak, moderate, and strong expression) depending on the intensity of the staining and the number of cells stained. The localization of the transporters was assessed by ocular examination of stained images.
With this approach, staining patterns for nine msAbs were generated [MDR1, MRP1, MRP2, MRP4, ileal bile acid transporter (IBAT), PEPT1, MCT1, OCT2, and OATP1B3] (Uhlén et al., 2005). During the course of this study, three additional antibodies (BSEP, OAT4, and OATP2B1) were generated, but no staining of the cell lines has yet been performed.
Functional Studies.
The function of MCT1 was investigated to assess its potency as a confounding factor in transporter experiments. The functionality was investigated using a known MCT1 substrate, [14C]lactic acid (15.3 μM), and an MCT1 inhibitor, quercetin (100 μM) (Bröer et al., 1997; Shim et al., 2007).
The four adherent cells lines (HeLa, HEK293, Caco-2, and HepG2) were seeded into 96-well plates (at 50,000–70,000 cells/well) 2 days before the experiment. Before the incubation, the cell monolayers were washed twice with 200 μl of Hanks' balanced salt solution (HBSS). The uptake of lactic acid was studied for 5 min, and the transport process was terminated by washing the cell monolayers five times with 200 μl of ice-cold HBSS. Thereafter, 50 μl of trypsin solution was added to detach the cells from the surface, and 200 μl of 1 M NaOH was added for at least 1 h to lyse the cells. This process was followed by neutralizing the solution with 1 M HCl, and 100 μl of the solutions was added to scintillation vials and supplemented with 3 ml of Ultima Gold scintillation mixture (PerkinElmer Life and Analytical Sciences, Waltham, MA). The radioactivity was measured in a Beckman LS6000IC liquid scintillation counter (Beckman Coulter, Fullerton, CA). The two suspension cell lines (HL-60 and K562) were incubated in 96-well plates (at 100,000 cells/well) using the same solutions and incubation times as for adherent cells. The cell solutions were added to a filter plate (Millipore Corporation, Billerica, MA), and the incubation solution was removed. The transport processes were terminated by washing the cells five times with ice-cold HBSS. The filters were removed and placed in scintillation vials and allowed to dry for 30 min at 37°C. Three milliliters of Ultima Gold scintillation mixture (PerkinElmer Life and Analytical Sciences) was added to the vials, and the samples were analyzed in a Beckman LS6000IC liquid scintillation counter (Beckman Coulter). The scintillation data were normalized to the protein content of each well using a bicinchoninic acid protein assay reagent kit (Thermo Fisher Scientific Inc., Rockford, IL).
The function of the MCT1 was investigated by comparing the transport in the presence (Uptake+inhibitor) and absence (Uptake−inhibitor) of inhibitor. MCT1 was classed as functional if the lactic acid uptake was significantly inhibited by quercetin (using eq. 1 and two-sided paired t test). Cell lines showing nonsignificant inhibition or no substrate uptake were defined as nonfunctional.
Results
Gene Expression Patterns.
The relative gene expression patterns for the cell lines are summarized in Figs. 1 and 2; see also Supplemental Data Fig. S1. HeLa cells expressed 21 of the 36 investigated transporters, 20 of which displayed only weak expression, with the only highly expressed transporter being MCT1. Likewise, HEK293 cells expressed 21 transporters, of which 16 had a low expression, and again MCT1 showed a higher expression. HL-60 cells expressed 16 (6 ABC and 10 SLC) transporters with MRP1, MCT1, OATP4A1, and OATP4C1 at significant levels, whereas K562 cells expressed 20 transporters with only MCT1 at a pronounced level. The osteosarcoma cell line Saos-2 expressed 14 transporters, 12 of which had weak expression. Of the efflux transporters, MRP4 expression was significant, whereas MCT1 was the uptake transporter with the highest expression. The gene expression of Caco-2 was comparable with previous publications (Seithel et al., 2006; Hilgendorf et al., 2007). The expression of MCT5 was higher than that of MCT1, which is in agreement with the small intestinal-like properties of Caco-2 cells that also express other SLC transporters present in the small intestine, such as PEPT1 and IBAT (Fig. 2) (Hilgendorf et al., 2007). In HepG2, a high level of expression of MRP2 was observed. In summary, with a few exceptions, the endogenous gene expression in the cell lines was low, the general exception being MCT1 (Fig. 1).
Relative gene expression of ABC transporters (a) and SLC transporters + human peptide transporter (HPT) 1 (b) in HeLa, HEK293, HL-60, K562, Saos-2, Caco-2, and HepG2. The bars represent the mean relative expression levels for each transporter. Caco-2 TC7 data are presented in Fig. 2c.
The effect of the transfection process and culture conditions on the relative gene expression profile. a, the effect of stable transfection on the transfected protein and endogenous transporters. b, gene expression of HEK293 cells in suspension (dark gray bars) and adherent (light gray bars) culture. c, gene expression of Caco-2 TC7 (dark gray bars) and Caco-2 (light gray bars) cells cultured on plastic support for 4 days; see under Materials and Methods. d, gene expression of Caco-2 cultured on a plastic support for 4 days (dark gray bars) and Caco-2 cells cultured on a filter support for 21 days (light gray bars).
Ubiquitously Expressed Transporters.
Apart from MCT1, 8 of 36 transporters investigated [three ABC (MRP1, 4, and 5) and five SLC (PEPT2, MCT5, OCT1, OCTN1, and OCTN2)] were expressed in all the investigated cell lines (Supplemental Data Fig. S1). Most of these were expressed at low levels; however, MCT1 showed higher expression in all the cell lines except for Caco-2, where MCT5 substituted for MCT1 as the most highly expressed monocarboxylate transporter.
Effect of Culturing Conditions and the Transfection Process.
Next, we compared how the cell culture conditions affect the gene expression of transporters (Fig. 2). The cell membranes are often reversibly modified by a transfection agent to facilitate the uptake of the transgene to obtain a high transient or stable expression of a gene in a cell line. Moreover, an excess of foreign genetic material is introduced into the cell. Thus, the transfection process creates a cellular stress reaction that may alter its gene expression profile compared with that of untransfected “control” cells. Therefore, empty vector-transfected cells are usually used as controls. However, to our knowledge, the effect of the transfection procedure on the background expression of drug-transporting proteins has not been investigated. Therefore, we studied whether stable transfection, using the methods previously described by Lohmann et al. (2007) with each of four different transporters (BCRP, OAT3, OCT1, OATP1B1) (Fig. 2a), and transient transfection using polyethyleneimine as transfection agent of HEK293 cells with OCT1 (data not shown) affected the endogenous gene expression of transporters. As expected, the expression levels of each of the introduced transporter proteins were well above those of the endogenous transporters (Fig. 2a). The endogenous expression of some transporters was slightly increased by the transfection process. However, because these transporters were expressed at very low levels, these changes were insignificant.
HEK293 cells were cultured both adherently and in suspension for transport studies, and we wondered whether the different conditions significantly alter transporter expression. As can be seen in Fig. 2b, HEK293 cells grown in suspension and adherent cultures exhibited similar expression patterns, with the exception of some ABC transporters, and MCT1 had a slightly higher expression in the suspension cells (Fig. 2b).
Recent studies showed large interlaboratory differences in transporter expression when different Caco-2 clones were cultivated under various conditions according to local traditions (Hayeshi et al., 2008). Here, we investigated the variability in transporter expression for two commonly used Caco-2 clones, Caco-2 (obtained from American Type Culture Collection, Manassas, VA) and its subclone Caco-2 TC7, using the same cell culture conditions. Caco-2 TC7 cells had a slightly higher gene expression of the SLC transporters, but the differences between the two cell lines was marginal, suggesting that comparable culture conditions may reduce variability in Caco-2 cells (Fig. 2c). When used as an intestinal in vitro model, the Caco-2 cells are grown on filters for 21 days, and during this period the gene expression of many drug transporters gradually increases (Seithel et al., 2006). In analogy, we assumed that Caco-2 cells grown on plastic for 4 days only would have a lower expression of transporters than cells grown on filters for 21 days. The results in Fig. 2d confirm this assumption, but as for the two Caco-2 clones in Fig. 2c, the differences were not very large.
In conclusion, in our hands, changes in culture conditions only resulted in small differences in the gene expression of transporters, indicating that experiments performed under such different conditions in the same laboratory ought not to result in large differences in transporter expression. Cellular stress from the introduction and expression of a transgene resulted in increased expression levels of some endogenous transporters. However, these transporters had very low expression levels; therefore, the differences are considered to be negligible compared with the high expression level of the transgene (Fig. 2a).
Antibody Validation in Human Tissues.
We were able to generate antibodies that met the release criteria from the HPR for nine of the transporters, which is a normal success rate (Berglund et al., 2008). Forty-eight human organs from three individuals were stained with each of the antibodies, and we compared the results from IHC with data from the literature to learn about the specificity of each of the nine transporter-specific antibodies (Fig. 3). The references used for this external validation are given in Table 1. This external validation, in combination with the internal HPR validation, indicated that four of the antibodies displayed an unspecific and partly atypical staining pattern compared with what is presented in the literature. It is noteworthy that some of these antibodies gave the postulated staining pattern in tissues where their expression is expected but also stained tissues where no staining had been observed previously. Thus, as anticipated, the peptide transporter PEPT1 stained the apical side of the intestinal epithelium (Fig. 3o) and the renal tubuli (data not shown), but it also stained many other tissues (e.g., the cerebral cortex), for which negative results have been published in the literature (Table 1; Fig. 3p). Likewise, according to literature, IBAT and the cation transporter OCT2 are mainly restricted to the intestine and kidney, respectively, and although our antibodies stained these tissues (Fig. 3, m and q), they also exhibited broader staining patterns than expected and stained many tissues that had been expected to give negative results (e.g., Figs. 3n and 4r; Table 1). Finally, MRP4 showed poor correlation between the antibody staining and the wide tissue distribution reported in the literature (Fig. 3, k and l; Table 1).
Examples of human tissues stained with msAb of human transporters (a-r). Brown-black staining is transporter-specific antibody staining, and the tissue sections are counterstained with hematoxylin (blue staining) to enable visualization of microscopic features. See under Results for details.
Antibody validation from tissue staining patterns
In contrast, the five remaining antibodies gave typical staining patterns, with specific staining in the tissues where it had been expected. Thus, MDR1 stained many tissues as anticipated, including the apical surface of intestinal epithelial cells, bile canaliculi, renal tubuli, and placenta (e.g., in Fig. 3a). Furthermore, tissues lacking MDR1, such as skeletal muscle, did not generally show any antibody staining (Table 1; Fig. 3b). MRP1 was expressed in many tissues, which is consistent with the literature, with staining being detected in, for example, kidney (Fig. 3c) and testes, and as anticipated no staining in liver (Fig. 3d; Table 1). As expected, the MRP2 antibody stained the bile canaliculi in hepatocytes (Fig. 3e), duodenum, and kidney, and the tissue staining pattern corresponded well with data published earlier, for example, no staining of heart tissue (Fig. 3f; Table 1). MCT1 displayed ubiquitous expression, as anticipated from the literature (e.g., Fig. 3, g and h), and the liver-specific OATP1B3 stained the hepatocytes but was absent in other organs (Fig. 3, i and j; Table 1).
Protein Expression and Comparison with Gene Expression.
As a part of the HPR project, the protein expression is analyzed by IHC in a panel of 47 cell lines (Strömberg et al., 2007), including six of the eight cell lines included in the present study (Table 2, Protein expression; Fig. 4). In general, the results confirmed the unspecific reactivity of antibodies with poor validation. For example, all the cell lines were stained for IBAT, whereas the corresponding gene expression was found only in the cell line of colonic origin, which is known to express IBAT (Hilgendorf et al., 2007) (Table 2, Protein expression). Likewise, all six cell lines were moderately or strongly stained by the small intestinal peptide transporter PEPT1, which has shown weak expression levels in earlier studies (e.g., Bleasby et al., 2006). The five msAbs that displayed the expected tissue expression patterns were assumed to give more reliable results (Table 2, Protein expression). Examples of IHC for the cell lines are shown in Fig. 4. Similar to the gene expression data, the MDR1 antibody stained Caco-2 and HepG2 cells at moderate levels and HeLa and HEK293 at weak levels but did not stain HL-60 and K562 cells, which were expected to express this protein (Table 2). A clear membrane localization of MDR1 was observed in Caco-2, HepG2, and HeLa cells. The msAbs for MRP1 stained HL-60 and Caco-2 weakly, whereas the other four cell lines were moderately stained. This is generally consistent with the gene expression, with the exception of HeLa and HL-60 (Table 2, Overlap). For MRP2, HeLa, HEK293, and HL-60 displayed weak, Caco-2 and HepG2 moderate, and K562 strong staining. This was comparable with gene expression, except for K562, in which no gene expression for MRP2 was observed at the selected time point (Table 2, Gene expression). Both MRP1 and especially MRP2 showed intracellular staining patterns in most of the cell lines, which is in agreement with earlier findings (Fig. 4) (Keitel et al., 2003). Consistent with the gene expression data, MCT1 showed pronounced typical membrane staining in all six cell lines, whereas the protein expression for OATP1B3 was negative in all the cell lines investigated (Table 2; Fig. 4). The overlap between gene and protein expression in the cell lines was investigated for all nine transporters using four expression levels (negative, weak, moderate, and strong) (Table 2, Overlap). The expression levels were in agreement in 43% of the cases. When the five transporters with good validation scores were considered, the overlap increased to 67%, and in most of the remaining cases, the gene and protein expression were comparable. For the four antibodies with poorer validation scores, the agreement was 13%, i.e., much lower than for the transporters with better validation scores (Table 2, Overlap).
Qualitative gene and protein expression, as well as overlap, data for the nine transporters investigated using IHC
Localization of the five transporters, showing good validation scores, in HEK293 and Caco-2 cells. Arrows indicate intracellular staining in MRP1 and MRP2. Brown-black staining is antibody-specific, and the tissue section is counterstained with hematoxylin (blue staining) to enable visualization of microscopic features. Images were annotated using an automated image analysis application. Staining levels: red, strong; orange, moderate; yellow, weak; and white, absent. See under Results for details.
MCT1 Function.
Because both gene and protein expression of MCT1 was significant, we reasoned that this ubiquitously expressed transporter could, if functional, be a confounding factor in transport experiments, in particular if small drugs with monocarboxylic acid moieties are studied (Meredith and Christian, 2008). Using the endogenous MCT1-substrate [14C]lactic acid, a significant uptake that could be completely or partly inhibited by quercetin was observed in five of six cell lines (Fig. 5).
MCT1 function in six cell lines. The cells were incubated with the MCT1 substrate [14C]lactic acid only (dark gray bars) and with [14C]lactic acid together with the MCT1 inhibitor quercetin (light gray bars). *, P < 0.05 and **, P < 0.01. ns, not significant.
Discussion
In this study, we used RT-PCR to assess the endogenous gene expression levels of 36 transporters in human cell lines commonly used in drug transport studies and investigated the protein expression of nine of these transporters using a new type of msAb, produced as a part of the HPR (Uhlén et al., 2005).
The low endogenous transporter expression in HEK293 and HeLa cells (Fig. 1; Supplemental Data Fig. S1) supports their suitability as systems for the overexpression of human transport proteins (Zhang et al., 1998; Ahlin et al., 2008). IHC showed that some ABC transporters, especially MRP2, are partially located intracellularly in all the cell lines (Fig. 4). These results are in agreement with studies indicating that MRP2 overexpressed in HEK293 cells may remain intracellular and not be functional at the plasma membrane (Sekine et al., 2008), but whether MRP2 is functional, as well as a potential confounding factor, remains to be investigated. On the other hand, clear membrane localization was observed for MCT1 in all the investigated cell lines (Fig. 4). This result suggested that MCT1 might be functional in these cell lines and that, based on subcellular localization, HEK293 and HeLa cells are a more attractive alternative for overexpression of SLC transporters than ABC transporters.
The two leukemia cell lines K562 and HL-60 have been extensively used in studies of multidrug resistance mediated by ABC transporters (Baran et al., 2007; Assef et al., 2009). In many studies, K562 displayed resistance to MDR1 substrates, whereas MRP1 is known to be a prominent contributor to drug resistance in HL-60 cells (e.g., Baran et al., 2007; Assef et al., 2009). Surprisingly, in our hands, these cell lines exhibited low levels of or no gene and protein expression of MDR1 (Table 2). Because similar results have been observed by others (Baran et al., 2007; Assef et al., 2009), it is possible that our K562 and HL-60 cells have not been exposed to selection pressure from cytostatics for a long time and therefore lost the expression of MDR1. However, the two cell lines expressed several other multidrug resistance proteins of the MRP family (Fig. 1; Supplemental Data Fig. S1), which is consistent with the literature (Gillet et al., 2004; Johnsson et al., 2005). Because the K562 and HL-60 cell lines are used to overexpress individual ABC transporters (Baran et al., 2007; Assef et al., 2009), our results propose that the endogenous expression of ABC transporters should be monitored to avoid biased conclusions regarding drug efflux mechanisms. An alternative for expression of ABC transporters is Saos-2 cells, which also have a low background expression in general and of ABC transporters in particular (Fig. 1; Supplemental Data Fig. S1). In previous studies, Saos-2 cells have been used successfully to overexpress single ABC transporters (Wierdl et al., 2003). The gene expression data obtained with the previously studied cell line Caco-2 (Fig. 1; Supplemental Data Fig. S1) were in agreement with the literature, indicating that our results are reproducible over longer periods (Hilgendorf et al., 2007; Hayeshi et al., 2008).
The endogenously expressed transport protein of potential concern in all these cell lines was MCT1, which displayed a comparably high gene and protein expression and also was found to be functional in five of six investigated cell lines (Table 2; Fig. 5). MCT1 is one of several ubiquitous transporters found in the expression analysis but was the only one that reached what could be considered as high expression levels. It transports lactic and pyruvic acid and hence is of importance in glycolysis and gluconeogenesis. In general, MCT1 accepts short-chain monocarboxylates and small drugs with carboxylate groups such as salicylate as substrates, suggesting that care should be exercised to avoid biased results, at least when investigating small monocarboxylate compounds. It is fortunate that very few drug substrates have been identified for MCT1 (Meredith and Christian, 2008).
Other complicating factors in cell culture studies are the variability introduced by laboratory-to-laboratory differences in cell culture procedures, cellular stress (as exemplified here by the introduction of a transgene), and clonal variation. In a recent study, a large laboratory-to-laboratory variation in the mRNA expression and function of transport proteins was observed for Caco-2 cells (Hayeshi et al., 2008). Because, in our experience, the time-dependent variability inside our laboratory—where the same culture procedures are used over time—is much smaller than that observed between laboratories, we investigated how transporter expression varied between suspension and adherent cultures, plastic and filter grown cells, transfection procedure, and cell clones with somewhat different properties. Our results indicated that, although differences could be observed, they were generally smaller than expected and in most cases probably of limited significance (Fig. 2). We conclude that maintenance of good, reproducible cell handling and cell culture conditions will reduce the variability in transporter expression.
The unspecific reactivity observed, in particular for the msAb with poor validation scores, was a point of concern but may also be an inherent property of these antibodies (Berglund et al., 2008; Pontén et al., 2008). In contrast to monoclonal antibodies that react with a single epitope on the antigen, msAbs have the potential to react with several epitopes and hence obtain a broader reactivity (Nilsson et al., 2005). Thus, in theory, these antibodies may maintain reactivity to their antigen in tissues of poorer quality, where the antigen may be partly denatured, such as human brain tissues obtained post mortem. Therefore, it cannot be excluded that, at least partially, the atypical reactivity observed with the poorly validated antibodies is real but has not been observed previously (Table 2, Protein expression). According to our own and the internal HPR validation approach (Berglund et al., 2008; Pontén et al., 2008), the msAb, with good correlation between the literature and the tissue staining patterns observed in this study, also gave a good correlation between gene and protein data for the cell lines (Table 2, Overlap). This suggests that a thorough antibody validation process is essential to get protein expressional data of high quality. We emphasize that the extensive validation of the msAbs in this study is unique and has barely been achieved for the commercial monoclonal antibodies used in many published studies. The validation showed that a tissue-specific staining pattern in a single or just a few tissues does not ensure the absence of nonspecific staining in a multitude of other tissues.
In conclusion, this is the first study that compares the gene and protein expression in human cell lines for a broad range of transporters. The results indicate that, in general, the gene and protein expressions were both low in the investigated cell lines. Independently of the antibody quality, specific staining in the expected tissues was obtained, but for the poorly validated antibodies, additional unspecific staining of those tissues that were expected to be negative was observed, underscoring the importance of thorough validation of the antibodies. We also showed, for the first time, that the monocarboxylic acid transporter MCT1 is ubiquitously expressed and functional in human cell lines, and we suggest that this should be considered when transport of small monocarboxylic drugs is investigated.
Footnotes
-
This work was supported in part by AstraZeneca; the Swedish Research Council [Grant 9478]; the Knut and Alice Wallenberg Foundation; the Swedish Fund for Research Without Animal Experiments; and the Swedish Board of Agriculture.
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.109.028654
-
↵
The online version of this article (available at http://dmd.aspetjournals.org) contains supplemental material. -
- ABC
- ATP-binding cassette
- MRP
- multidrug resistance protein
- HEK
- human embryonic kidney
- BCRP
- breast cancer resistance protein
- SLC
- solute carrier transporter
- HPR
- human proteome resource
- msAb
- monospecific polyclonal antibody
- RT-PCR
- real-time polymerase chain reaction
- IHC
- immunohistochemistry
- PrEST
- protein epitope signature tag
- IBAT
- ileal bile acid transporter
- HBSS
- Hanks' balanced salt solution.
- Received June 6, 2009.
- Accepted September 2, 2009.
- Copyright © 2009 by The American Society for Pharmacology and Experimental Therapeutics
