To better understand the role of human equilibrative (hENTs) and concentrative (hCNTs) nucleoside transporters in physiology and pharmacology, we investigated the regional, cellular, and spatial distribution of two hCNTs (hCNT1 and hCNT2) and two hENTs (hENT1 and hENT2) in four human tissues. Using in situ hybridization and immunohistochemical techniques, we found that the duodenum expressed hCNT1 and hCNT2 mRNAs in enterocytes and hENT1 and hENT2 mRNAs in crypt cells. In these cells, the hCNT and hENT proteins were predominantly localized in the apical and lateral membrane, respectively. Hepatocytes expressed higher levels of mRNAs of hENT1, hCNT1, and hENT2 than of hCNT2 and expressed all these proteins at hepatocyte cell borders and in the cytoplasm. While the kidney expressed hCNT1 and hCNT2 mRNAs in the proximal tubules, hENT1 and hENT2 mRNAs were present in the distal tubules, glomeruli, endothelial cells, and vascular smooth muscle cells. Proximal tubules adjacent to corticomedullary junctions expressed hENT1, hCNT1, and hCNT2 mRNA. Immunolocalization studies revealed predominant localization of hCNTs in the brush-border membrane of the proximal tubular epithelial cells and hENTs in the basolateral membrane of the distal tubular epithelial cells. Chorionic villi sections of human term placenta expressed mRNAs and proteins for hENT1 and hENT2 but only mRNA for hCNT2. Immunolocalization studies showed presence of hENT1 in the brush-border membrane of the syncytiotrophoblasts. These data are critical for a better understanding of the role of nucleoside transporters in the physiological and pharmacological effects of nucleosides and nucleoside drugs, respectively.
- human tissue
human equilibrative (hENTs) and concentrative (hCNTs) nucleoside transporters are important in cellular homeostasis. They regulate neurotransmitter, circulatory, and renal functions by modulating extracellular adenosine available to interact with adenosine receptors (8, 29). For example, ablation of mouse (m)ENT1 has been shown to reduce hypnotic and ataxic responses to ethanol, increase propensity to consume ethanol (11), and diminish anxiety behaviors (10) in mice. In addition, nucleoside transporters are known to salvage nucleosides in the proximal tubules of the kidney (48), skeletal muscle (39), and the intestine (52). Furthermore, hENTs and hCNTs have been proposed to participate in regulation of cell growth (54), apoptosis (57), vasodilator response (42), and ischemic preconditioning responses in cardiac muscles (4). Recently, nucleoside transporters (ENT2 and CNT2) have been shown to mediate cellular uptake of cyclic ADP-ribose second messenger that is known to trigger Ca2+-mediated signaling processes (23, 24).
Nucleoside transporters also have the potential to be important in the efficacy and toxicity of various anticancer and antiviral nucleoside drugs. For nucleoside drugs to produce their pharmacological effect or toxicity, they need to be phosphorylated intracellularly. Since nucleoside drugs are hydrophilic, they need to be transported into cells and into cell organelles by the nucleoside transporters (33, 37, 45, 58). Thus nucleoside transporters can be the rate-limiting step in the efficacy and toxicity of nucleoside drugs. Indeed, this has been demonstrated in vitro for a number of drugs (12, 30, 34, 49). Recently, we have also shown (33) that mitochondrial expression of hENT1 facilitates the entry of fialuridine into the mitochondria and enhances the mitochondrial toxicity of this nucleoside drug. In addition, nucleoside transporters have the potential to play an important role in the absorption and disposition of nucleosides and nucleoside drugs. For example, expression of the second member of the concentrative nucleoside transporters (hCNT2) appears to be important in the intestinal absorption of ribavirin, a nucleoside drug used in the treatment of hepatitis C (47). The simultaneous expression of hENTs and hCNTs in kidney epithelial cells may explain why adenosine and deoxyadenosine are cleared differently by the kidneys (32).
To fully appreciate the role of nucleoside transporters in physiology and pharmacology of nucleosides and nucleoside drugs, the types of nucleoside transporters expressed in various tissues, their level of expression, and their regional and spatial distribution must be characterized. To do so, and as a continuation of our earlier studies (9, 33, 46), we report here the characterization of the types of nucleoside transporters expressed in the human intestine, liver, kidneys, and placenta, their level of expression, as well as their regional and spatial distribution in these tissues.
Tissue sources and preparation.
Human intestine (duodenum or jejunum), liver (adult or fetal, the former from donors or biopsies), kidney (donors or biopsies), and term placental tissue samples were obtained as fixed tissue sections from the Department of Pathology, University of Washington. The tissue slides were made from discarded tissues after resection of abnormal regions. Only grossly and histologically normal tissues were used. Additional human liver tissues were obtained from the Liver Bank (organ donors), Department of Pharmaceutics, University of Washington. Tissues were procured after approval from the Human Subjects Division, University of Washington. Tissues were fixed in 10% neutral buffered formalin for in situ hybridization and in methyl-Carnoy solution (30% chloroform, 60% methanol, and 10% glacial acetic acid) for immunohistochemical studies. All fixed tissues were processed and paraffin embedded according to standard protocols.
Synthesis of hybridization probes.
Full-length hENT2 and hCNT2 cDNAs (32, 38) were cloned from a human intestinal cDNA library with PCR-based strategies, and their sequences were verified with published sequences (NM_001532, NM_004212). Full-length hENT1 and hCNT1 cDNAs were obtained as described previously (32, 38). All of the cDNAs were subcloned into pCR4.1 (Invitrogen) for riboprobe preparation. pTri-β-actin-human antisense control template was obtained from Ambion (Austin, TX). All reagents for riboprobe preparation (except 35S-UTP) were obtained from Promega (San Luis Obispo, CA). 35S-UTP was obtained from New England Nuclear (Boston, MA). Riboprobes were synthesized as per the manufacturer's instructions (Promega). Briefly, 125–250 μCi of 35S-UTP was used for each reaction mixture that contained 2.5 mM of rNTP mix, RNA polymerase (T3, T7, or SP6 as appropriate), RNA polymerase buffer, RNAsin, and appropriate template DNA. The plasmids containing the various cDNAs were linearized by either NotI (antisense: hCNT2, hENT1, hENT2; sense: hCNT1) or SpeI digestion (antisense: hCNT1; sense: hCNT2, hENT1, hENT2) before use as templates for preparing the antisense and sense probes. The reaction mixture was incubated at 37°C for 75 min to synthesize RNA probes and subsequently incubated with RQ1 DNase to destroy template DNA. The mixture was separated on a Sephadex-G50 column preequilibrated with Tris-EDTA (TE) buffer, and fractions were collected by monitoring radioactive peaks with a Geiger counter. RNA was precipitated by adding 0.1 volume of 3 M sodium acetate and 2.5 volumes of ethanol. The precipitated pellets were washed with 70% ethanol, air dried, and reconstituted in nuclease-free water containing 10 mM dithiothreitol (DTT).
In situ hybridization of tissue sections.
Tissue sections were deparaffinized by washing three times in xylene (10 min each) and rehydrated in a graded series of ethanol washes. Sections were then rinsed three times with 0.5× standard saline citrate (SSC; 150 mM NaCl, 15 mM Na citrate, pH 7.0) and rinsed with Tris buffer (500 mM NaCl, 10 mM Tris, pH 8.0). Sections were then treated with Tris buffer containing freshly added proteinase K (25 μg/ml; Sigma, St. Louis, MO) and incubated at 37°C for 30 min before several washings. Tissue sections were later dehydrated with a graded series of ethanol washes and allowed to air dry. Subsequently, tissue sections were incubated (for 2 h at 50–55°C) with 50 μl of prehybridization mix (50% formamide, 0.3 M NaCl, 20 mM Tris, pH 8.0, 5 mM ethylenediaminetetraacetic acid, 1× Denhardt solution containing 0.01% Ficoll type 400, 0.01% polyvinylpyrrolidone, and 0.01% bovine serum albumin, 1× 10% dextran sulfate, 10 mM DTT, and 50 μg/ml yeast tRNA) in a tightly sealed incubation chamber. For hybridization, the radiolabeled probes were diluted by adding 500,000 cpm of 35S-labeled riboprobe in 50 μl of prehybridization mix and were directly added to the slides. The slides were then incubated at 50–55°C in a humidity chamber for 12–16 h. The sections were then briefly washed three times with 0.5× SSC and incubated for 30 min at 37°C with Tris buffer containing freshly added RNase (20 μg/ml). The sections were subsequently washed three times in 2× SSC (2 min each) and three times in 0.1× SSC containing 0.5% Tween 20 (Sigma) for 40 min at 50–60°C. The optimal wash temperature was standardized for each probe used. The sections were finally washed briefly with 2× SSC before being dehydrated for 2 min each in ethanol containing 0.3 M ammonium acetate and then air dried. The hybridized tissue sections were dipped in NTB2 nuclear emulsion (Kodak, Rochester, NY) diluted 1:1 with distilled water and allowed to dry in a dark room. To develop hybridization signals, the sections were incubated for 5 min at 12–15°C in Kodak D-12 developer (20.5 g/200 ml distilled water), rinsed in distilled water, and incubated in Kodak Rapid Fix (30.8 g/200 ml distilled water). After developing, the sections were washed in distilled water, stained with hematoxylin and eosin, dehydrated, and mounted with coverslips. Labeled sense cRNA probes were substituted for the antisense probes as a negative control. Also, additional control experiments were performed by pretreating slides with RNases before adding the labeled antisense probes or by completely omitting labeled antisense probes in the hybridization mixtures. In all cases, exposure was carried out for at least 4–6 wk.
Generation of antibodies.
Antibody was generated from mice immunized according to a University of Maryland Institutional Animal Care and Use Committee-approved protocol (JH01). A monoclonal antibody against hCNT1 was generated by first making the mice tolerant to whole cell extract containing yellow fluorescent protein (YFP), then immunizing them with whole cell extracts prepared from Madin-Darby canine kidney (MDCK) cells expressing hCNT1-YFP and boosted with hCNT1 peptides. Polyclonal antibodies were custom generated by Covance (Denver, PA). The specificity of rabbit polyclonal hENT1 antibody, described previously (27), was confirmed by immunofluorescence studies and Western blotting (data not shown) using MDCK cells stably expressing hENT1-YFP fusion protein (32). Polyclonal rabbit antibodies were raised against peptides of the NH2 terminus tails of hCNT2 and the intracellular loop between transmembrane domains 6 and 7 of hENT1 (19, 43). A previously characterized rabbit polyclonal hENT2 antibody (19) was used in the study.
Immunohistochemistry of tissue sections.
Methyl-Carnoy-fixed, paraffin-embedded human kidney, intestine, and placental tissues were used for the study. Cryosections of human liver tissues (5-μm thickness) were prepared and fixed in methyl-Carnoy solution. All tissues were deparaffinized and rehydrated, and endogenous peroxidase activity was blocked by incubating the tissues (10 min at room temperature) in 0.3% H2O2 and 0.1% sodium azide in water. Slides containing tissue sections were then incubated with normal goat serum (10%) for 30 min to block nonspecific binding. After being washed once in phosphate-buffered saline (PBS) for 5 min, the tissue sections were incubated in a humidity chamber for 60 min (at room temperature) with the primary antibody. After being washed twice with PBS for 10 min each, the tissue sections were incubated for 30 min in a moist chamber with biotinylated secondary antibody (Vector Laboratories, Burlingame, CA) and avidin-biotin-horseradish peroxidase (HRP) complex (ABC; Vector). The tissue sections were finally washed twice with PBS for 10 min each and then visualized with 3,3′-diaminobenzidine (DAB; Sigma) with nickel chloride enhancement to give a blackish brown reaction product. Hematoxylin and eosin staining was performed, and the slides were dehydrated and sealed with coverslips. For all samples, negative controls for immunohistochemistry were performed by substituting the primary antibody with control serum from the same species.
Analyses of data and preparation of figures.
Multiple sections from several tissue donors were analyzed for each of the tissues examined (n = 3–5). Positive cellular labeling for in situ hybridization experiments was defined as five or more silver grains concentrated over a single cell. Positive staining in immunolocalization experiments was judged by visual comparison with control serum- or irrelevant IgG-stained sections. Various anatomic regions of tissues were identified from morphological structures seen in hematoxylin and eosin-stained sections. Images focusing on the hybridization signals (in situ hybridization) or immunostaining regions were captured with a ×40 dry or ×100 oil-immersion objective in a Zeiss inverted microscope and recorded with an Olympus DP-10 digital camera.
Specificity of in situ hybridization probes (controls) and integrity of mRNA in tissue sections.
Hybridization with antisense and sense control probes was performed for comparison of actual signal-to-noise ratio. Sense control probes for all four transporters (see methods) analyzed did not yield any clear hybridization signals and could easily be judged as background. An antisense β-actin probe was used in all tissue sections to judge the quality and integrity of mRNA in tissue sections. Only those parallel sections that showed intense hybridization signals for β-actin probe were used for hybridization experiments with various nucleoside transporter probes.
Characterization of hCNT1 monoclonal and hCNT2 polyclonal antibodies.
hCNT1 mouse monoclonal antibody was characterized by immunofluorescence staining of MDCK cells stably expressing hCNT1-YFP protein (32). This antibody recognized hCNT1 localized predominantly at the cell periphery of these cells (Fig. 1, A–C) but did not stain MDCK cells expressing YFP alone (Fig. 1, D–F). Also, this antibody did not stain wild-type (wt) MDCK cells (Fig. 1G). Similarly, hCNT2 rabbit polyclonal antibody identified predominant cell surface hCNT2 in MDCK cells stably expressing hCNT2 (Fig. 2A), which was greatly reduced after preincubation of antibody with hCNT2 peptides used to generate the antibody (Fig. 2B). Also, no detectable staining was observed when the primary antibody was omitted (Fig. 2C) or when wt MDCK cells were stained with hCNT2 antibody (Fig. 2D). In addition, both hCNT1 and hCNT2 antibodies did not cross-react with each other or with hENT1 or hENT2 (data not shown). Together, these results suggest that the antibodies generated against hCNT1 and hCNT2 show specific interaction with the respective proteins.
Expression and localization of nucleoside transporters in human intestine.
Several tissue sections from the duodenal regions of the small intestine (n = 5) were analyzed for mRNA expression of various nucleoside transporters. The distribution of in situ hybridization signals of hCNT1 mRNA showed a selective expression in the enterocytes (Fig. 3A) and the outer two-thirds portion of the villi (Fig. 3B). This expression progressively decreased in intensity as it reached the base of the villi. A moderate level of riboprobe signals for hCNT1 was also noted in the interstitial region of the villi (data not shown). A comparable hybridization pattern was also observed in sections obtained from the jejunum (n = 1; data not shown). Unlike the strong hybridization signals in the upper part of the villi, very low or no clear hCNT1 signal could be identified in the crypts or the glandular regions of the duodenum (Fig. 3C). In contrast, hCNT2 mRNA was expressed throughout the length of the villi (Fig. 3, E and F) at equal or higher intensity than that of hCNT1. Occasionally, mild to moderate hybridization signals were also present in some of the crypt regions (Fig. 3G, top left). No glandular regions or any other submucosal structures appeared positive for hCNT2 expression. Interestingly, unlike hCNT1 and hCNT2, both hENT1 (n = 3) and hENT2 (n = 3) mRNAs were expressed in the crypt regions of the small intestine (Fig. 3, I, K, M, O) with less or no clearly identifiable signal in the enterocytes (Fig. 3, J and N). Some hybridization signals were also observed at the base of the villi but at a lower intensity than in the crypt regions. Interstitial regions within the villi also showed some degree of positive signals for hENT1 mRNA. Intense expression of hENT1 and hENT2 in the crypt cells, but not in enterocytes, was also observed in the single jejunal section (data not shown).
Immunohistochemical analysis of duodenal sections showed presence of hCNT1 and hCNT2 protein staining in both enterocytes (Fig. 4, A and D) and crypt cells (Fig. 4, B and E). Consistent with the in situ hybridization results, the intensity of staining for both hCNT1 and hCNT2 was stronger in enterocytes (Fig. 4, A and D) than in crypt cells (Fig. 4, B and E). In the enterocytes, both hCNT1 and hCNT2 showed more intense staining at the apical side of the enterocytes (arrowheads, Fig. 4, A and D) than the basolateral side (arrows, Fig. 4, A and D). In the case of hENT1 and hENT2, strong immunoreactivity was observed in the crypt regions (Fig. 4, H and K). This intensity seemed higher than the respective immunoreactivity observed in the enterocytes (Fig. 4I) in the same histological section on the same slide. A certain degree of hENT1 and hENT2 staining was always observed in all the villi cells, with almost equal intensity at the apical and basolateral sides of the enterocytes (Fig. 4, G and J). For hENT1 and hENT2, immunohistochemical reactivity was observed predominantly in lateral membrane of the crypt cells (arrows, Fig. 4, H and K). Coimmunolocalization of hCNT1 and hENT1 clearly showed hCNT1 staining predominantly at the apical membrane and hENT1 staining predominantly at the basolateral membrane of the enterocytes (Fig. 5, A–C). However, higher magnification showed a low level of hCNT1 and hENT1 staining at the basolateral (Fig. 5D, arrow) and apical (Fig. 5E, arrowheads) surface of the enterocytes, respectively. Collectively, these results corroborate that hCNTs are predominantly concentrated in the apical membrane of the enterocytes and hENTs are predominantly concentrated in the lateral membrane of the crypt cells.
Expression and localization of nucleoside transporters in adult and fetal human liver.
In situ hybridization of these tissues identified mRNA for all four of the nucleoside transporters examined (Fig. 6, A, B, D, E), albeit with differences in the level of expression, as judged by the riboprobe signal intensities (small, uniform, dense particles). Maximum expression was seen in the β-actin control (Fig. 6C). For each transporter, the signal intensity was constant throughout the hepatic parenchyma, with no appreciable differences in expression in areas of portal field around hepatic artery, portal vein, or portal interstitium. The signal intensity was maximal for hENT1 (Fig. 6D), hCNT1 (Fig. 6A), and hENT2 (Fig. 6E), followed by hCNT2 (Fig. 6B). However, in the case of hENT1, apart from expression in the hepatic parenchyma, prominent expression was also noted in other sinusoidal liver cells, such as endothelial cells and Kupffer cells (data not shown).
Consistent with the in situ hybridization results, immunostaining of the liver tissue also identified expression of all four nucleoside transporters (Fig. 6, G, H, J, K). Finer details of expression were not discernible within the hepatocytes in these tissue sections except for some concentrated staining of both hCNTs (Fig. 6, G and H) and hENTs (Fig. 6, J and K) at the hepatocyte cell borders. In addition to the sinusoidal pattern of expression in the hepatic parenchyma, smaller blood vessels located within the hepatic parenchyma also stained positive for hENT1 and hENT2 (data not shown).
Expression and localization of nucleoside transporters in human kidney.
Similar to previously reported expression profiles in rodent kidneys (3, 48), we identified expression of all four of the nucleoside transporters in adult human kidneys (Fig. 7). Interestingly, unique patterns of expression of these transporters were observed within the tubular regions of the nephron. The proximal tubular segments (Fig. 7, A and B), but not distal tubular segments (Fig. 7, A and C) showed moderate levels of concentration of hCNT1 antisense riboprobes. Endothelial cells lining the small and larger blood vessels and peritubular capillaries were negative for hCNT1 expression (data not shown). hCNT2 showed moderate in situ hybridization signals in both proximal tubular epithelial cells (Fig. 7, E and F) and distal tubular epithelial cells (Fig. 7G). Glomeruli showed no appreciable signals with hCNT1 (data not shown) and hCNT2 (Fig. 7E) riboprobes. In several kidney sections analyzed (n = 5), hENT1 mRNA signals were often found localized at the distal tubular segments (Fig. 7, I and K) of the nephron. The degree of expression was almost uniform in all of the distal tubule structures examined, but only low levels of expression were observed in some of the proximal tubules (Fig. 7, I and J). A moderate level of riboprobe signals of hENT1 was also observed in some of the proximal tubules adjacent to the corticomedullary junction (data not shown). Invariably, certain cell types in all of the glomeruli examined were labeled with hENT1 riboprobes (data not shown). Although this was a consistent event, it was not possible to identify the type of cells that express hENT1 mRNA, given the complexity of cell types that are present within the glomeruli (epithelial cells, mesangial cells, podocytes, etc.). Strong hybridization signals for hENT1 were also observed in the endothelial cells lining the larger blood vessels of the kidney and the vascular smooth muscle cells present in the kidney sections (data not shown). Distribution of hENT2 mRNA was similar to hENT1 expression in the proximal tubules (Fig. 7N), in the distal tubules (Fig. 7, M and O), and in endothelial cells lining large blood vessels and vascular smooth muscle cells.
Immunohistochemical analysis of hCNT1 (Fig. 8, A–C) and hCNT2 (Fig. 8, E–G) in human kidneys showed mild to moderate levels of immunoreactivity in the proximal tubular segments (Fig. 8, A and E). Higher magnification showed a concentration of hCNT1 staining in the apical (brush border) membrane of proximal tubular epithelial cells (arrows, Fig. 8A). Human CNT2 immunoreactivity was also observed in the apical membrane of some, but not all, proximal tubular epithelial cells (arrows, Fig. 8E), indicating heterogeneity in the expression of hCNT2. A few distal tubular segments (Fig. 8F) and glomeruli (arrows, Fig. 8G) were also immunoreactive for hCNT2, but the peritubular capillaries were negative (data not shown). Unlike hCNTs, hENT1 immunoreactivity was significant in almost all of the distal tubular segments (Fig. 8J) of the kidney sections examined (n = 5). Analysis at higher magnifications of distal tubular epithelial cells showed intense hENT1 staining in the basolateral regions (arrows, Fig. 8J) and to a lesser extent in the intracellular regions. Presence of hENT1 staining in the distal tubules was confirmed by colocalization with a distal tubule protein, epithelial-membrane antigen (Fig. 9, G–I). hENT1 staining was also identified in the proximal tubular epithelial cells located at corticomedullary junctions (Fig. 8I), and, interestingly, hENT1 staining in these regions was found in both the apical (arrows, Fig. 8I) and the basolateral (arrowheads, Fig. 8I) regions. Double immunolocalization of hCNT1 and hENT1 in the human kidney showed colocalization of hCNT1 with hENT1 at the apical surface (Fig. 9C, inset, arrows) of the proximal tubular epithelial cells, but only hENT1 was found at the basolateral regions (Fig. 9B, inset, arrows). P-glycoprotein was used as a positive control to label the apical surface of the proximal tubular epithelial cells (Fig. 9D). Several cell types also stained strongly positive for hENT1 within the glomeruli regions (arrows, Fig. 8K). Some staining of hENT2 staining was also noted in the distal tubular epithelial cells (Fig. 8N) but not in the proximal tubular epithelial cells (Fig. 8M) or the glomerulus (Fig. 8O).
Expression and localization of nucleoside transporters in human placenta.
Our results showed intense hybridization signals with the hENT1 probe in chorionic villi regions (Fig. 10A). In addition, positive hybridization signals were also identified for hENT2 (Fig. 10B) and hCNT2 (Fig. 10C) at the epithelial layers, but these signals were less intense than that of hENT1. The muscular layers in the placenta showed only a background level of hybridization, and no hybridization signals were seen with hCNT1 probes both in epithelial and muscular layers (data not shown).
Immunohistochemical analysis of placental tissue sections also showed expression of hENT1 staining in the outer trophoblastic layer (Fig. 10G). Careful examination of cross sections of chorionic villi showed concentration of hENT1 expression in the brush-border membrane (arrows, Fig. 10J) of the syncytiotrophoblasts. Endothelial cells of blood vessels present in those regions were also positive for hENT1 (data not shown). Like hENT1, hENT2 signals were also seen in the outer trophoblastic layer (Fig. 10H) but with lesser intensity than hENT1, and this staining was predominantly cytoplasmic. No significant hCNT2 protein staining could be identified (Fig. 10I) even though moderate levels of its corresponding mRNA signals were identified by in situ hybridization. Also, no appreciable hCNT1 protein staining was detected in the trophoblastic layers (Fig. 10K), consistent with a lack of in situ hybridization signal.
Our previous studies, using brush-border membrane vesicles prepared from human intestine, showed that hCNT1 and hCNT2, but not hENT, transport activity was present in the apical membrane of the enterocytes (46). However, when mRNA isolated from these intestines was microinjected into Xenopus laevis oocytes hENT1/2 activity was also observed (9, 46). On the basis of these data, we proposed (9) that hCNTs and hENTs were expressed on the apical and basolateral membranes, respectively, of the enterocytes, allowing vectorial transport of nucleosides and nucleoside drugs from the intestinal lumen to the blood. Consistent with these observations, in the present study we observed significant expression of hCNT1 and hCNT2 in the apical membrane of the enterocytes. However, it is possible that the hENT activity we observed in Xenopus oocytes came from the expression of these transporters in crypt cells. The significance of hENT localization in crypt epithelial cells is not clear at present, and further studies are needed to clarify their role. One interesting observation is that in situ hybridization exhibited abundant hENT1 mRNA in crypt cells but not in enterocytes. This raises the possibility that the low levels of hENT1 protein (but not mRNA) in enterocytes could be remnants of this protein expressed (in preenterocytes) in crypt cells. This is likely if hENT proteins have a long half-life. Nevertheless, at any given time point, our data unambiguously identified hCNT1 and hCNT2 as being predominantly localized in the well-differentiated enterocytes, and hENT1 and hENT2 as being predominantly localized in the rapidly proliferating crypt cells. Interestingly, an earlier study showed in rat intestinal IEC-6 cells that agents known to promote proliferation (EGF, transforming growth factor-α, and cell wounding) upregulate ENT-type transporters, whereas agents inducing differentiation (glucocorticoids) induce CNT2 expression (2). Together, our present observations suggest that both concentrative and equilibrative transporters are likely to be important in the absorption of nucleoside drugs.
Most of the earlier studies on hepatic nucleoside transporters were performed on rat primary hepatocytes or rodent hepatic cell lines (3, 44). Liver parenchymal cells isolated from adult rats show activity corresponding to pyrimidine-preferring rat (r)CNT1, purine-preferring rCNT2, and nitrobenzylmercaptopurine riboside (NBMPR)-insensitive rENTs (15). Interestingly, in these studies, Northern blot analysis identified significant presence of rCNT2 mRNA but not (or only negligible levels of) rCNT1, rENT1, or rENT2 mRNA (15). In contrast, in human liver we observed significant amounts of mRNA as well as protein signals for all of the nucleoside transporters (except hCNT3, which was not analyzed in this study). Recently, Fernandez-Veledo et al. (21) also showed by real-time PCR analysis that human liver expresses hENT1, hENT2, hCNT1, and hCNT2 mRNA but not hCNT3. Since these transporters are also expressed at protein levels in human hepatocytes (our present data), these data suggest that the steady-state levels of nucleoside transporters could be different between humans and rodents. We also found that all of the nucleoside transporters were present at the periphery of the hepatocytes, suggesting their presence at the cell surface. In rat hepatocytes, rCNT1 is localized at the canalicular membrane and rCNT2 is localized at the sinusoidal membrane (16). Also, the trafficking of intracellular rCNT2 increases and the protein accrues at the sinusoidal plasma membrane in the presence of physiological agents such as bile acids (20). In a separate study, we demonstrated (22) that hENT1 and hCNTs proteins are targeted to both sinusoidal and canalicular membranes in human hepatocytes cultured in a sandwich configuration. Significant expression of various hENTs and hCNTs in hepatocytes suggests that these transporters are important in the efficacy and/or toxicity of nucleoside drugs (e.g., ribavirin) used in the treatment of hepatic disorders (e.g., hepatitis C).
In situ hybridization and immunohistochemical analysis revealed distinct segmental expression of various transporters within the human nephron. Abundance of hCNT1 and hCNT2 was observed at the brush border (apical) surface of the proximal tubular regions of the nephron. The apical localization of hCNTs in the proximal tubules is not completely surprising because earlier studies with brush-border membrane vesicles prepared from rabbit kidney cortex identified cit (rCNT1) and cif (rCNT2) transport activities (55). In addition, proximal tubule brush-border membrane vesicles of rat kidneys showed only rCNT activity, while only rENT activity was present in the basolateral membrane (36). Concentrative nucleoside transport activity corresponding to the cit system (except guanosine is also a substrate) was also demonstrated in the brush-border membrane vesicles prepared from human proximal tubular epithelial cells (25). The present data provide direct evidence for localization of both hCNT1 and hCNT2 in the apical membrane of the proximal tubules in the human kidney and suggest involvement of hCNTs in active reabsorption of nucleosides from the tubular lumen. Indeed, consistent with our findings and those of others, the renal clearance of many of the natural and pharmacologically active nucleoside analogs used in the clinic is lower than their glomerular filtration clearance (41, 48, 50). Apart from the proximal tubules and a low level of expression of hCNT2 in the distal tubules and the glomeruli, no significant expression of hCNTs was identified in other regions of human kidneys. Expression of CNT1 in proximal tubules and CNT2 in glomeruli was previously demonstrated in the rat kidney (48). We predict that detailed transport studies on proximal tubule membrane vesicles or primary tubular epithelial cells prepared from human donor kidneys will confirm the role of hCNT1 and hCNT2 in the reabsorption of nucleosides and nucleoside drugs. Because we did not have access to an hCNT3 antibody we could not confirm the presence of hCNT3 in the human kidney. However, as described below, others have now reported expression of CNT3 in human and rat kidney (14, 18).
Even though the basolateral membrane of the rat proximal tubules showed rENT activity (55) and recombinant hENTs were localized in the basolateral membrane of the canine kidney cell line MDCK (32, 40), the present study did not identify significant expression of hENT1 or hENT2 proteins in the proximal tubules of the human kidney except in the proximal tubular segments of the corticomedullary junctions. Here, hCNT1, hCNT2, and hENT1 were expressed, with the latter present in both apical and basolateral membranes. In the distal tubular cells, hENT1 was found mainly in the basolateral membrane. These data suggest that complex regulatory signals could govern the localization of hENT1 in various tubular segments of the nephron. Higher levels of hENT1 and moderate levels of hENT2 were also identified in the distal tubules, glomeruli, smaller blood vessels, and vascular muscle cells present in the kidney sections. While the physiological significance of these segmental patterns of hCNT and hENT expression in the human kidney is not clear at present, hENTs are likely involved in renal homeostasis. One possibility is their participation in adenosine receptor signaling, regulating glomerular filtration rate by looping a signal sensed at the distal tubules to the afferent arterioles in the glomerulus, a phenomenon known as tubuloglomerular feedback (51, 53). Direct association of a concentrative nucleoside transporter (rCNT2) and an A1 adenosine receptor was previously demonstrated in rodents (17).
When this manuscript was being prepared, Damaraju et al. (14) reported expression of hENT1, hENT2, hCNT1, hCNT2, and hCNT3 mRNAs in the human kidney. In addition, hENT1 protein was shown to be localized in the apical membrane of the proximal tubules and in both the apical and basolateral membranes of the loop of Henle and collecting ducts. Our present identification of apical localization of hENT1 in the proximal tubule epithelial cells and basolateral localization in the distal tubule epithelial cells is consistent with these findings. We also observed the presence of hENT1 in the apical and basolateral membrane of the proximal tubule epithelial cells adjacent to the corticomedullary junctions. However, in contrast to their finding of an absence of hCNT proteins in the kidney, we observed moderate levels of hCNT1 protein in the apical membrane of the proximal tubule epithelial cells along with heterogeneous presence of hCNT2 protein in the proximal tubules, distal tubules, and glomeruli. Differences in antibody detection limits could contribute to these differences in observations. Interestingly, Damaraju et al. (14) and Errasti-Murugarren et al. (18) found abundant expression of human and rat CNT3, respectively, in the apical membrane of proximal tubule epithelial cells. Collectively, these data support our earlier hypothesis (32) that the simultaneous expression of hCNTs and hENTs in the apical and basolateral membranes of kidney tubule epithelial cells can lead to either active reabsorption or secretion of nucleosides and nucleoside drugs.
We observed significant expression of hENT1, hENT2, and hCNT2 mRNAs but only hENT1 and hENT2 proteins in human placental trophoblasts. The mRNA results are consistent with a previous study that identified expression of hENT1, hENT2, and hCNT2 in the human placenta with real-time PCR analysis (1). Higher expression of rENT1 and rENT2 mRNAs than rCNT1 and rCNT2 mRNAs has also been reported in the rat placenta (35). Our results on hENT1 protein expression in the syncytiotrophoblasts of human term placenta are consistent with an earlier study that used antibodies raised against erythrocyte nucleoside transporter (5, 6). The abundance of hENT1 and hENT2 in the human placenta could allow entry of nucleoside drugs, such as ribavirin, into the fetal compartment where they can produce their toxicity (7, 13, 26, 28, 31, 56). Why hCNT2 protein is not expressed in the trophoblasts is not clear. Nor is it clear whether the expression of various nucleoside transporters might change with gestational age.
Collectively, these results have found complex cell-dependent spatial and regional expression of the various nucleoside transporters in the human intestine, liver, kidney, and placenta. Where functional studies are lacking (e.g., for the placenta), such studies should be conducted to determine whether the proteins expressed in these tissues are functional. Nevertheless, the patterns of expression observed here clearly indicate that nucleoside transporters expressed in each tissue have the potential to significantly modulate, both locally and systemically, the activity of natural nucleosides (e.g., adenosine) and the pharmacological efficacy and toxicity of nucleoside drugs.
This study was supported by National Institutes of Health Grant RO1-GM-054447.
We thank Drew Brostrom and Dr. Andrei Mikheev for their support in characterization of the hCNT1 antibody.
Present address of Y. Lai: Pfizer, St. Louis, MO.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society