Heterogeneous accumulation of fluorescent bile acids in primary rat hepatocytes does not correlate with their homogenous expression of ntcp

Abstract

Sodium taurocholate-cotransporting polypeptide (ntcp) is considered to be a major determinant of bile acid uptake into hepatocytes. However, the regulation of ntcp and the degree that it participates in the accumulation of specific substrates are not well understood. We utilized fluorescent bile acid derivatives and direct quantitation of fluorescent microscopy images to examine the regulation of ntcp and its role in the cell-to-cell variability of fluorescent bile acid accumulation. Primary-cultured rat hepatocytes rapidly accumulated the fluorescent bile acids, chenodeoxycholylglycylamidofluorescein (CDCGamF), 7-β- nitrobenzoxadiazole 3-α hydroxy 5-β cholan-24-oic acid (NBD-CA), and cholyl-glycylamido-fluorescein (CGamF). However, in stably transfected HeLa cells, ntcp preferred CDCGamF, whereas the organic anion transporter, organic anion transporting polypeptide 1 (oatp1a1), preferred NBD-CA, and neither ntcp nor oatp1a1 showed strong accumulation of CGamF by these methods. Ntcp-mediated transport of CDCGamF was inhibited by taurocholate, cyclosporin, actin depolymerization, and an inhibitor of atypical PKC-ζ. The latter two agents altered the cellular distribution of ntcp as visualized in ntcp-green fluorescent protein-transfected cells. Although fluorescent bile acid accumulation was reproducible by the imaging assays, individual cells showed variable accumulation that was not attributable to changes in membrane permeability or cell viability. In HeLa cells, this was accounted for by variable levels of ntcp, whereas, in hepatocytes, ntcp expression was uniform, and low accumulation was seen in a large portion of cells despite the presence of ntcp. These studies indicate that single-cell imaging can provide insight into previously unrecognized details of anion transport in the complex environment of polarized hepatocytes.

sodium taurocholate-cotransporting polypeptide (ntcp) is abundantly expressed on the sinusoidal plasma membrane of hepatocytes and is believed to play a major role in the enterohepatic circulation by accounting for the sodium-dependent uptake of bile acids and other anions (30). The activity of ntcp is generally inferred from the accumulation of substrates within cells. However, substrate accumulation is dependent on transport into and out of cells as well affinities for intracellular binding partners. The endogenous situation is complicated by the presence of multiple uptake transporters, such as sodium-independent organic anion transporting polypeptides (oatps) (33), by export transporters residing on the sinusoid and bile canaliculus and by the regulation of ntcp, which may be related to its intracellular distribution (2). In this report we utilized fluorescent bile acid derivatives to analyze organic anion transport by directly quantifying microscope images of hepatocytes and cultured cell lines captured with low-excitation intensity, low-light CCD cameras, and multi-channel semiautomated image acquisition. Fluorescent-labeled bile acids are distinct from their unlabeled counterparts and may have higher affinity and lower maximal transport, lower rates of biotransformation, and increased cell retention compared with unmodified bile acids (13, 28). However, they have been studied as substrates for ntcp and oatps as well as canalicular transporters, and their handling by hepatocytes and transporter-transfected cell lines can reveal important features of cellular organic anion processing (4, 21, 22). In the following studies, fluorescent bile acids were used to assess the substrate specificity of ntcp as well as the regulation of ntcp transporter activity and its relation to its intracellular distribution. Through imaging it has been found that individual isolated rat hepatocytes accumulate fluorescent bile acids to different levels (28). We find that accumulation does not correlate with the level of ntcp and that many hepatocytes have low accumulation despite the presence of ntcp, suggesting that ntcp may be downregulated in these cells or that other transporters or intracellular binding partners are responsible for the variable accumulation of fluorescent bile acids.

MATERIALS AND METHODS

Reagents, cell culture.

Reagents were from Sigma-Aldrich (St. Louis, MO) unless noted. Dr. Alan Hofmann (University of San Diego, CA) kindly provided the fluorescent bile acid derivatives, chenodeoxycholylglycylamidofluorescein (CDCGamF), 7-β- nitrobenzoxadiazole 3-α hydroxy 5-β cholan-24-oic acid (NBD-CA), and cholyl-glycylamido-fluorescein (CGamF). Structures of these or similar compounds are presented elsewhere (22, 26). Coverslip-bottomed chambers were from MatTek (Ashland, MA) or Thermo Scientific (Lab-Tek, Nalge Nunc International, Rochester, NY). Rat hepatocytes were isolated by collagenase perfusion (3), attached to coverslip-bottomed chambers coated with 0.5 mg/ml Matrigel (BD Biosciences, San Jose, CA) or 1.5 mg/ml rat tail collagen, and cultured in Williams E media lacking phenol red and supplemented with 0.1 μM dexamethasone, 2 mM glutamax, 10 μg/ml insulin, 5.5 μg/ml transferrin, 6.7 ng/ml sodium selenite, and 100 U/ml penicillin-streptomycin (Invitrogen, Carlsbad, CA). All animal procedures were approved by the University Animal Use Committee. To create sandwich cultures, 1.5 mg/ml neutralized collagen was overlaid for 2 h in the cell-culture incubator, followed by readdition of media. Washes and incubations were in Williams media unless otherwise noted. Serum-free media contained 135 mM NaCl, 1.2 mM MgCl₂, 0.8 mM MgSO₄, 28 mM dextrose, 2.5 mM CaCl₂, and 25 mM Hepes, pH 7.4. Wide-field microscopy for FITC, rhodamine, and bright-field illumination was as described (24); confocal microscopy utilized a multichannel white light source with DAPI, FITC, Rhodamine, and Cy5 filter settings on a CARV II spinning-disk imager (Crisel Instruments, Rome, Italy) containing an iXon back-illuminated 897 EMCCD camera (Andor Technologies, Belfast Ireland), a PhotoFluor metal halide lamp (Chroma Technologies, Bellows Falls, VT), and a ×60 1.4 NA oil immersion or ×20 lens. This was supplemented with Perkin Elmer laser excitation, spinning-disk confocal imaging (Analytical Imaging Facility, Albert Einstein College of Medicine). Relative fluorescence units (RFU) equal the pixel intensity measured during a set of experiments. Controls were performed for each set of experiments, allowing RFU to be compared directly.

Accumulation assays and cell lines.

Fluorescent bile acid accumulation assays were performed as follows: HeLa cells that had been stably transfected with full-length rat oatp1a1 (Slco1a1) or rat ntcp (Slc10a1) under regulation of a metallothionein promoter were cultured to confluence on coverslip-bottomed chambers while being induced for protein expression with 100 μM ZnSO₄ (12). Cells were washed in Williams media E (lacking phenol red) and preincubated with Williams with or without reagents in the cell-culture incubator for 30 min (or other times as indicated), followed by addition of fluorescent bile acid to the mixture for 5 min, followed by imaging under wide-field fluorescence and bright-field microscopy with the ×60 lens. Accumulation assays for hepatocytes were done similarly except that the preincubation was for 20 min followed by a 10-min incubation with fluorescent bile acid followed by imaging under spinning-disk confocal imaging with the ×60 lens (unless noted). Cells expressing rat ntcp fused to the NH₂ terminus of green fluorescence protein (GFP) were generated with the pEGFP-N3 vector (Clontech, Mountain View, CA) in which full-length rat ntcp cDNA was inserted into the Xho1 and KpnI multicloning sites. Huh-7 cells were transfected using Polyfect (Qiagen, Valencia, CA) reagent, and cells were selected and maintained in RPMI containing 200 μg/ml G418. Fluorescent colonies were additionally selected and subcloned using fluorescence microscopy. For live cell imaging, ntcp-GFP-expressing cells were plated onto coverslip-bottomed dishes for 2 days, washed in Williams media, and placed in agents or equivalent concentration of solvent (e.g., DMSO) for 30 min at 37°C in the cell-culture incubator. The dishes were then imaged [or washed and then imaged for cyclosporin A (CsA) and myristolated PKC-ζ pseudosubstrate (Zps) treatments] by time-lapse fluorescence wide-field microscopy at 37°C with ×60 lens in bright-field and FITC fluorescence channels. For automated analysis, cells were treated with drugs as above, washed twice in PBS plus 5 μg/ml Hoechst nuclear stain, fixed in 4% formaldehyde in BRB80 buffer (80 mM dipotassium Pipes, 1 mM EGTA, 1 mM MgCl₂, pH 6.8) for 10 min, washed in PBS, and imaged by wide-field fluorescent microscopy with the ×20 lens.

Image analysis.

Images were quantified using ImageJ (ImageJ, National Institutes of Health, Bethesda, Maryland, http://rsb.info.nih.gov/ij/). Unless noted, fluorescence accumulation was quantified by scoring the percent of pixels per image that exceeded an intensity threshold (i.e., those that were positive for fluorescence signal) using an automated algorithm written as an ImageJ macro (supplemental materials, Macro S1; supplemental material for this article is available online at the American Journal of Physiology Gastrointestinal and Liver Physiology website). Scoring was normalized by adjusting the threshold so that control cells gave a score of 10%. More than 50 images were collected for each condition with two or more experimental chambers, and the entire experiment was repeated at least three times. Automated quantification of ntcp-GFP expressing cells was quantified using macros written for ImageJ (supplemental materials, Macro S2). Briefly, images of fluorescent cells were smoothed and segmented (digitally selected) with the same threshold for all images chosen by eye. Images of Hoechst-stained nuclei, used for cell counting and normalization of fluorescence intensity, were smoothed, and nuclei were segmented by threshold chosen automatically. Standard deviation of fluorescence intensity per image was calculated by ImageJ. Data are combined from three separate experiments with 25 images per condition and more than 1,200 cells for each experiment. Experiments performed on different days were normalized so that the average GFP fluorescence intensity at the location of the nuclei was equivalent.

Immunofluorescence.

Stably transfected HeLa cells were plated on coverslip-bottomed chambers, induced for transporter expression (12), and processed by being washed twice in BRB80 plus 5% glycerol, permeabilized with 0.2% digitonin and 10% DMSO for 1 min, and fixed in 1 mg/ml dithiobis(succinimidyl)propionate (or 4% formaldehyde) for 5 min, followed by washing in PBS, and blocking in PBS plus 5 mg/ml BSA. Primary and secondary antibodies were incubated for 40 min followed by washing. Immunofluorescence images were acquired the same day as the accumulation assays. Hepatocytes were plated on coverslip-bottomed chambers coated with collagen. Permeabilization and immunofluorescence was performed as described (25) using a freeze-thaw cycle, as detergents were found to decrease immunofluorescence intensity. Affinity-purified ntcp and oatp1a1 antibodies were generated as described (25, 32). HiLyte Fluor secondary antibodies were from Anaspec (San Jose, CA).

RESULTS

Accumulation of fluorescent bile acids in rat hepatocytes.

To assess whether real-time imaging could be used to measure fluorescent bile acid accumulation, we incubated 4-h primary-cultured rat hepatocytes with fluorescent bile acids under spinning-disk confocal microscopy. Cell-surface binding is evident by 20 s of exposure to 250 nM CDCGamF and reaches near maximal accumulation by 80 s (Fig. 1A). Accumulation was strongly decreased when cells were assayed after several days in culture, coincident with loss of expression of bile acid transporters (16). However, we were able to maintain accumulation using collagen sandwich culturing (7, 17). Figure 1, B–D, presents images from two-day sandwich-cultured primary rat hepatocytes incubated for 10 min with 250 nM fluorescent bile acids. CDCGamF strongly accumulated into the cells and was also excreted into the bile canaliculi (arrows) that develop during sandwich culturing. CGamF showed relatively low cytoplasmic fluorescence (Fig. 1C) but accumulated brightly within bile canaliculi, indicating good transport. NBD-CA strongly accumulated within cells but showed low fluorescence in bile canaliculi, which likely reflects the low fluorescence of the NBD dye in the aqueous environment rather than a failure to secrete NBD-CA into bile canaliculi (18, 27). All of the fluorescent bile acids accumulated within hepatocytes as visualized by spinning-disk confocal imaging, and, interestingly, all showed variable levels of accumulation between neighboring cells.

Fig. 1.

Imaging the accumulation of three fluorescent bile acid derivatives within individual primary rat hepatocytes. A: hepatocytes were cultured for 4 h and subject to the fluorescent bile acid accumulation assay with 250 nM henodeoxycholylglycylamidofluorescein (CDCGamF) under time-lapse spinning-disk confocal imaging. Time after addition is given in seconds. B–D: hepatocytes were cultured in a collagen sandwich for 48 h and subject to the accumulation assay with 250 nM CDCGamF, cholyl-glycylamido-fluorescein (CGamF), or 7-β- nitrobenzoxadiazole 3-α hydroxy 5-β cholan-24-oic acid (NBD-CA) as indicated, followed by spinning-disk confocal imaging. Note that bile acids accumulate in bile canaliculi (BC) that develop during culture but appear excluded from some unidentified organelles. Scale bars = 20 μm.

To determine whether this variability in accumulation is maintained during culture, CDCGamF accumulation was assayed over 6 days of sandwich culturing of hepatocytes (Fig. 2). Both accumulation and the cell-to-cell variable nature of the accumulation were maintained, indicating that the cells with low accumulation are not dead or dying. Staining of 6-day cultured hepatocytes with the membrane impermeant dye, propidium iodide (Fig. 3), revealed that the spread (attached) hepatocytes excluded this dye and did not show corresponding variability of this dye. Only the obviously apoptotic or necrotic cells, as seen in the bright-field images, had permeabilized membranes (Fig. 3). Cells with low accumulation of CDCGamF excluded propidium iodide and appeared normal in bright-field images. Again this indicates that the variable nature of the accumulation is not attributable to dead or dying cells and is intrinsic to normal-appearing rat hepatocyte primary cultures.

Fig. 2.

Fluorescent bile acid accumulation is maintained during collagen sandwich culture of hepatocytes. Hepatocytes were subject to the accumulation assay with 250 nM CDCGamF after 0 to 6 days in collagen sandwich culture as indicated. Scale bar = 20 μm.

Fig. 3.

Variable accumulation of fluorescent bile acids is not due to permeabilized plasma membranes as shown by exclusion of propidium iodide. 6-day sandwich-cultured hepatocytes were assayed with 250 nM CDCGamF as in Fig. 1B with the inclusion of 1 μM propidium iodide and rhodamine channel imaging. Large arrowheads (Perm) indicate occasional permeabilized apoptotic/necrotic cells that stain for propidium iodide; small arrows indicate cells with low (Lo) accumulation of CDCGamF. Bar = 20 μm.

Specific transport of fluorescent bile acids by ntcp and oatp1a1.

Several studies have indicated that fluorescent bile acids are substrates for ntcp and oatps (4, 22), and ntcp and oatp1a1 have been shown to have broad specificity for most common conjugated bile acids as well as sulfobromophthalein (12). To examine substrate specificity for the fluorescent bile acids, we utilized stably transfected HeLa cells that express either ntcp or oatp1a1 under control of a zinc-inducible metallothionein promoter. Cells were grown to confluence, induced for transporter expression, and exposed to the fluorescent bile acids. Statistical numbers of fluorescence microscopy images were acquired for each condition. As seen in Fig. 4A, ntcp allowed strong accumulation of CDCGamF but not NBD-CA, whereas oatp1a1 allowed strong accumulation of NBD-CA but not CDCGamF. Neither ntcp- nor oatp1a1-expressing cells showed significant accumulation of CGamF by these methods, and HeLa cells did not take up any of the dyes in the absence of transporter because accumulation was not seen in the absence of Zn induction (Fig. 4A, uninduced data). Ntcp-mediated accumulation of CDCGamF was also dependent on the presence of extracellular sodium (Fig. 4B), as expected because of the sodium-dependent nature of ntcp-mediated but not oatp-mediated transport (11, 33).

Fig. 4.

Specificity of fluorescent bile acid accumulation for cells transfected with sodium taurocholate-cotransporting polypeptide (ntcp) or organic anion transporting polypeptide 1 (oatp1a1). A: HeLa cells expressing or not expressing (i.e., uninduced, UnInd) ntcp or oatp1a1 were subject to fluorescent bile acid accumulation assays with 250 nM CDCGamF, CGamF, or NBD-CA as indicated. The y-axis is the average fluorescence intensity of individual cells, traced in ImageJ, and subtracted from image intensity of regions of the same chamber that lacked cells, expressed in relative fluorescence units (RFU). B: accumulation under low-sodium conditions was measured as above except that, 20 min before imaging, cells were washed in serum-free media or serum-free media with choline chloride substituted for sodium chloride (Low Na). Accumulation with or without Na⁺ is compared between cells cultured identically, measured on the same day with the same instrument settings, expressed in relative fluorescence units (RFU). In B data of ntcp- and oatp1a1-transfected cells were from different days, and fluorescence intensities between these 2 sets are not comparable. *P < 0.01, error bars are SD.

Inhibition of ntcp-mediated transport of CDCGamF by taurocholate.

A brief survey of agents that affect the activity of ntcp was performed to further evaluate the usefulness of the fluorescence assay and to monitor for agents that would also affect the cellular distribution of ntcp. Taurocholate (TC) is a classical substrate for ntcp, and we therefore assayed whether TC and the related bile acid and ntcp substrate, taurodeoxycholate (TDC) (12), would inhibit CDCGamF accumulation. HeLa cells stably transfected with ntcp were assayed for fluorescent bile acid accumulation. In these and subsequent experiments, accumulation of CDCGamF was quantified using a customized macro program written for ImageJ (supplemental materials, Macro S1). The macro scores the number of image pixels that exceed an intensity threshold, and the data are normalized to yield a score of 10% for the controls. Control experiments were performed alongside treated samples, and multiple plates were measured for each sample. Figure 5A demonstrates that 10 μM TC or TDC inhibited the accumulation of 100 nM CDCGamF by 98 and 95%, respectively. However, 1 μM TDC inhibited accumulation by only 40%, which is consistent with findings that show a higher affinity of fluorescent bile acids for cellular transporters compared with nonfluorescent analogs (13, 28, 34). Representative images from these experiments are seen in Fig. 5B. These demonstrate that CDCGamF accumulation was nonuniform across fields of confluent cells. Clusters or occasional solitary cells had high accumulation, whereas other cells were dark. This cell-to-cell variation in accumulation was seen for all experiments but did not prevent accurate quantitation when multiple fields of cells were averaged. Cells that grew above the plane of focus (as seen in bright-field images) also did not prevent accurate quantitation although dead or apoptotic cells frequently stained brightly for CDCGamF (Supplemental Fig. S1) and contributed to data noise. These characteristics presumably affect the measurement of bile acid accumulation by other methods although they may go undetected.

Fig. 5.

Inhibition of Ntcp-mediated accumulation of CDCGamF by taurocholate (TC) and taurodeoxycholate (TDC). A: HeLa cells expressing ntcp were subject to fluorescent bile acid accumulation assays. 10 min before imaging, cells were incubated with TC or TDC for 5 min followed by addition of 100 nM CDCGamF for 5 min. Accumulation was quantified by automated image analysis (see materials and methods) as the percent of image pixels that exceeded a fluorescence-intensity threshold using macro programs written for ImageJ software. B: representative fluorescence images (top), scaled identically, are presented along with their bright-field images (bottom). These and all subsequent error bars are SE of the mean. *P < 0.01 vs. control. Scale bar = 20 μm.

Inhibition of ntcp-mediated transport by actin depolymerization.

Previous studies showed that actin and microtubule depolymerization can inhibit the increase in ntcp-mediated transport that is seen when dibutyryl cAMP is added to ntcp-transfected cells (6, 23). To determine whether cytoskeletal depolymerization could affect baseline ntcp-mediated transport, we incubated ntcp-expressing HeLa cells with either 10 μM nocodazole, to depolymerize microtubules, or 10 μM latrunculin, to depolymerize actin, and measured accumulation as in Fig. 5. Actin depolymerization reduced CDCGamF accumulation by 45%, whereas microtubule depolymerization actually increased accumulation by 13% compared with controls although the latter did not meet statistical significance (Fig. 6). Images from the experiments (Fig. 6B) indicate that latrunculin-treated cells have poor, but not zero, accumulation, and this is reflected in the scoring. Some bright-staining debris of an unknown nature was seen in the latrunculin-treated cells but not in control or nocodazole-treated cells. The nocodazole-treated cells frequently had a “wavy” cellular periphery (Fig. 6B, inset within the nocodazole bright-field image), suggestive of cell spreading.

Fig. 6.

Inhibition of ntcp-mediated accumulation of CDCGamF by latrunculin (Latr) but not nocodazole (Noco). A: ntcp-expressing HeLa cells were subject to bile acid accumulation assays with 100 nM CDCGamF; the 30-min preincubations and imaging contained 0.5% vol/vol DMSO and 0.5% vol/vol ethanol (control) or solvents plus 10 μM latrunculin B or 10 μM nocodazole. B: representative fluorescence images (top) and bright-field images (bottom) are presented as in Fig. 5. Inset: magnification of the boxed region of the nocodazole-treated cells showing deformation of the plasma membrane. C: stably transfected ntcp-green fluorescence protein (GFP) cells were treated with latrunculin or nocodazole as above, followed by live-cell imaging to examine for changes in ntcp distribution. In control cells, ntcp concentrated into bands (band) that frequently formed between cells, as well as into intracellular puncta. Ntcp was present at the plasma membrane as seen in the spread lamella (lam.). Latrunculin treatment resulted in retracted lamella and aggregation of ntcp near the nucleus, whereas nocodazole treatment resulted in spreading of the plasma membrane but retention of ntcp in bands and puncta. *P < 0.01 vs. control. Scale bar = 20 μm.

To determine whether nocodazole and latrunculin affect the subcellular distribution of ntcp, we prepared stable cell lines expressing ntcp-GFP and treated these with the depolymerizing agents as above. Although ntcp-GFP was able to transport CDCGamF (Supplemental Fig. S2), we did not quantify GFP and CDCGamF within the same cells because they both exhibit green fluorescence. As seen in Fig. 6C, in untreated cells ntcp-GFP is found in the plasma membrane in thin lamellae as well as concentrated to “bands” that frequently form at cell-to-cell junctions. Alternately, ntcp is found in puncta and hazy structures that accumulate near the center of the cell. Treatment with latrunculin resulted in cellular retraction and aggregation of ntcp around the nucleus. A thin retraction lamella remained attached to the coverslip as is typically seen when actin is depolymerized in cultured cells (5). In contrast, nocodazole treatment resulted in spreading of ntcp-containing plasma membrane, but the ntcp-containing bands and puncta remained intact. We speculate that the plasma membrane spreading caused the small increase in transport that we observe with nocodazole treatment because more ntcp may be present at the cell surface in the spread cells.

Inhibition of ntcp-mediated transport by PKC-ζ and cyclosporin.

We also examined the effects of an inhibitor of PKC-ζ (Zps) as well as a known ntcp inhibitor, CsA (8, 15), on the ntcp-mediated transport of CDCGamF. PKC-ζ is a kinase that associates with cell membranes and participates in vesicle trafficking and establishment of cell polarity (14). Previous studies have found that inhibition of PKC-ζ reduces cAMP-mediated increases in ntcp-mediated transport (19) as well as intracellular trafficking of ntcp (25). As seen in Fig. 7, inclusion of 1 μM Zps decreased accumulation of CDCGamF by 34% compared with control, whereas 25 μM CsA decreased accumulation by 82%. Surprisingly, inclusion of both Zps and CsA reduced accumulation by only 34% and was less effective at transport inhibition than CsA alone. This suggests that the inhibitory activity of CsA may require PKC-ζ activity or that there is some cross talk in these pathways. When ntcp-GFP-expressing cells were similarly treated, the PKC-ζ inhibitor caused aggregation of ntcp-containing membrane and formation of wispy or spidery (Fig. 7C) strings of membrane. Using time-lapse imaging, we observed that these regions frequently showed lower intracellular ntcp-GFP motility, and we speculate that this results in the altered subcellular distribution of ntcp-GFP, consistent with findings that this drug blocks trafficking of ntcp-containing vesicles in vivo and in vitro (25). In contrast cyclosporin did not produce gross alterations in the distribution or motility of ntcp-GFP that could be detected at this level of resolution.

Fig. 7.

Inhibition of ntcp-mediated accumulation of CDCGamF by PKC-ζ pseudosubstrate (Zps) and cyclosporin A (CsA). A: ntcp-expressing HeLa cells were subject to bile acid accumulation assays with 100 nM CDCGamF and a 28-min preincubation with solvent alone (control) or 1 μM Zps, 25 μM CsA, or both drugs followed by washing before CDCGamF incubation and imaging. B: representative fluorescence images (top) and bright-field images (bottom) are presented as in Fig. 5. C: ntcp-GFP-expressing cells were treated with Zps or CsA as above followed by imaging. Zps induced a “spidery” appearance of ntcp-GFP and stalling of ntcp-GFP-containing membrane, whereas CsA did not grossly affect ntcp-GFP distribution. *P < 0.01 vs. control. Scale bar = 20 μm.

To gain a quantitative perspective, we captured statistical numbers of low-magnification images of ntcp-GFP-expressing cells after their exposure to the drugs described above. The images were subject to automated image analysis to examine for differences in ntcp-GFP distribution. A macro was written for ImageJ to allow automatic quantitation of ntcp-GFP fluorescence (supplemental materials, Macro S2). As can be seen in Fig. 8, treatment with either PKC-ζ inhibitor or latrunculin altered the variation in ntcp-GFP fluorescence (i.e., the standard deviation of the ntcp-GFP pixel intensity), whereas the other treatments were not different from control. This supported our observations of the high-magnification images, indicating that actin depolymerization and PKC-ζ inhibition caused aggregation of ntcp. These treatments also decreased ntcp-mediated accumulation of CDCGamF, and these studies support the notion that the organization of ntcp within cells can affect its transport activity. They also show that quantification of GFP fluorescence can provide a method to distinguish different drugs that affect ntcp transport activity.

Fig. 8.

Automated image analysis of the effects of drugs on the distribution of ntcp-GFP. Ntcp-GFP-expressing cells were mock treated (control) or treated with 1 μM Zps, 10 μM latrunculin B, 10 μM nocodazole, 25 μM CsA, or 10 μM TC for 30 min as indicated. Cells were then fixed, and statistical numbers of images were acquired and quantified using macros written for ImageJ (supplemental material). Variation in Ntcp fluorescence is the average SD of the pixel intensities of each field of fluorescent cells. *P < 0.05 vs. control.

Variation in the accumulation of CDCGamF within individual hepatocytes does not correlate with the level of ntcp or oatp1a1.

As described, both transfected HeLa cells and cultured primary rat hepatocytes exhibit dramatic cell-to-cell variability in their accumulation of fluorescent bile acids. In both cases, neighboring cells accumulate different levels of fluorescent bile acid, but, when multiple fields are averaged, accumulation is consistent and quantifiable. To examine whether this is caused by variable levels of transporter, we employed correlative-immunofluorescence technology, wherein fields of cells are imaged in fluorescent bile acid assays, processed for immunofluorescence, and then reimaged for the presence of the transporter protein. A typical field of ntcp-transfected HeLa cells that have accumulated CDCGamF is shown in Fig. 9A along with its corresponding bright-field image, which shows that healthy cells are present throughout the field. Immunofluorescence staining for ntcp revealed that the level of ntcp itself was strikingly variable (red channel), and those cells with high bile acid accumulation (green channel) had high ntcp expression. Arrows indicate adjacent cells with high and low fluorescence for both CDCGamF and ntcp. This was not due to fluorescence bleed-through, as the green channel, which previously contained CDCGamF fluorescence, was blank during imaging of ntcp. These experiments explain the variability in CDCGamF accumulation, and they further validate that ntcp is responsible for this accumulation. It is also apparent that CDCGamF accumulation is restricted to the cell periphery in HeLa cells. For quantification, the images were aligned, cell outlines were traced, and the mean fluorescence intensity for CDCGamF and ntcp was measured per cell and subtracted from background (the pixel mode value of the image). These values were plotted as a correlation, seen at the right of the images (correlation coefficient of 0.69, n = 277 cells). Many of the cells clustered around the origin, with fluorescence equal to or sometimes lower than background for both ntcp and CDCGamF, confirming what was seen in the images.

Fig. 9.

Cell-to-cell variability of CDCGamF accumulation is explained by variable ntcp protein expression in stably transfected HeLa cells but not in rat hepatocytes. A: ntcp-expressing HeLa cells were subject to bile acid accumulation assays for 100 nM CDCGamF. Cells were then processed for immunofluorescence, and identical fields were reimaged for ntcp protein, as indicated. Arrows point to a pair of cells with high and low accumulation of CDCGamF and high and low expression of ntcp. Right: quantification of the data by correlation plot. Images were aligned, cell outlines traced, and the mean fluorescence intensity for CDCGamF and ntcp was measured and subtracted from background (the mode pixel value of the image) using ImageJ. The correlation coefficient was 0.69, n = 277. B: hepatocytes (Heps) were cultured for 4 h and subject to the fluorescent bile acid accumulation assay with 250 nM CDCGamF, imaged with a ×20 lens, and then processed and reimaged for ntcp immunofluorescence. Arrows point to a pair of cells with high and low accumulation of CDCGamF but equivalent expression of ntcp. Right: correlation plot of ntcp vs. CDCGamF fluorescence (●), and no primary antibody control vs. CDCGamF fluorescence (○), performed as above; correlation coefficient was 0.42, n = 162 (ntcp) and 0.18, n = 282 (no primary). C: hepatocytes were imaged for CDCGamF accumulation and oatp1a1 immunofluorescence as in the previous panels. Arrows point to a pair of cells with high and low accumulation of CDCGamF but equivalent expression of oatp1a1. Correlation coefficient was 0.51, n = 197 (no primary is identical to above). Scale bars = 20 μm. D: hepatocytes were cultured for 8 h and subject to the fluorescent bile acid accumulation assay with 250 nM CDCGamF. 25 min before imaging, cells were incubated in the presence (left bar) and absence (right bar) of sodium for 15 min and then exposed to 100 nM CDCGamF for 10 min followed by imaging with a ×20 lens. CDCGamF accumulation is the average fluorescence intensity of individual cells, traced in ImageJ, and subtracted from background intensity for regions of the same chamber lacking cells. *P < 0.05.

We predicted that a similar result would be seen with primary rat hepatocytes; i.e., the amount of ntcp would vary from cell to cell. Figure 9B presents correlative immunofluorescence for CDCGamF and ntcp in hepatocytes. Ntcp localized to the plasma membrane as well as the cell interior and tended to concentrate at cell junctions as was seen in HeLa cells (Fig. 9, A and B) and ntcp-GFP cells (Fig. 6C). However, the amount of ntcp was very similar from cell to cell despite the variable accumulation of CDCGamF. Arrows in Fig. 9B indicate adjacent cells with high and low CDCGamF accumulation but similar levels of ntcp. Although ntcp can transport CDCGamF and is responsible for the cell-to-cell variation of accumulation in ntcp-transfected HeLa cells, in primary hepatocytes the presence of ntcp did not guarantee high accumulation of CDCGamF. The correlation plot of individual cell fluorescence indicates weak correspondence of the two signals (correlation coefficient of 0.42, n = 162 cells). As a control, correlative-immunofluorescence experiments were performed for CDCGamF vs. immunofluorescence procedure in the absence of primary antibody (Fig. 9B), and this gave low fluorescence signal and weak correlation between the two signals (correlation coefficient of 0.18, n = 282 cells). The images have inherent positive correlation attributable to features (e.g., cell structures) that are present in both channels.

For additional comparison we performed the same experiments for oatp1a1 (Fig. 9C). Oatp1a1 was present at cell junctions and at the plasma membrane and appeared to show modest variability in intensity between neighboring cells (correlation coefficient of 0.51, n = 197 cells). However, many cells had low accumulation of CDCGamF but high levels of oatp1a1. Arrows of Fig. 9C indicate neighboring cells with high and low accumulation of CDCGamF but identical intensities of oatp1a1. Overall these studies indicate that the determinants of fluorescent bile acid accumulation within hepatocytes are more complex than within transporter-transfected cells. We additionally found that sodium depletion reduced the accumulation of CDCGamF in rat hepatocytes by 18% (Fig. 9D), which is similar to previous reports showing a 20–50% reduction in fluorescent bile acid accumulation in hepatocytes upon sodium depletion (10, 18, 28, 35). This compares to 90% and 50% reduction that can be found for natural bile acids, TC, and chenodeoxycholylglycine, respectively, upon sodium depletion (31). The lower level of sodium dependence for the accumulation of fluorescent bile acids may reflect their lower dependence on ntcp.

DISCUSSION

This study demonstrates the utilization of high-resolution, low-light detection microscopy in providing detailed measurements of the temporal and spatial dynamics of fluorescent bile acid accumulation in hepatocytes and transfected cell lines. In transfected HeLa cells, ntcp allowed accumulation of CDCGamF, whereas oatp1a1 allowed accumulation of NBD-CA, whereas neither allowed strong accumulation of CGamF by these methods (Fig. 4). Previous reports indicate that CGamF, which is sensitive to pH and hydrophobicity (18), is weakly transported by ntcp (22) but is a good substrate for OATP1B3 (1), while CDCGamF can be transported by ntcp (22), and unconjugated 7-β NBD bile acids, similar to NBD-CA, can transit through the liver into bile (27). In our studies, TC and TDC inhibited ntcp-mediated accumulation of CDCGamF by 98% and 95%, respectively, and depolymerization of actin reduced the accumulation of CDCGamF by 45% and caused retraction of ntcp-GFP into a region surrounding the nucleus (Fig. 6). In contrast, depolymerization of microtubules did not lead to large-scale changes in the organization of ntcp and did not reduce ntcp-mediated CDCGamF accumulation. An inhibitor of PKC-ζ, which has been shown to inhibit the vesicular traffic of ntcp (25), resulted in a 34% reduction in the accumulation of CDCGamF as well as aggregation of ntcp-GFP, whereas the immunosuppressant, cyclosporin, decreased accumulation of CDCGamF by 82% but did not alter ntcp distribution within the resolution limit of these assays (Figs. 7 and 8). We also found that inclusion of the PKC-ζ inhibitor to cyclosporin inhibition assays reduced the degree of inhibition from cyclosporin (Fig. 7), suggesting that there may be cross talk in the mechanism of inhibition of these two drugs. Overall, the data support observations that ntcp transporter activity can be affected by changes in its cellular distribution and that drugs may affect the intracellular trafficking and distribution of ntcp (6, 23).

In contrast to results in transfected HeLa cells, CDCGamF, NBD-CA, and CGamF were all well transported into hepatocytes as seen by microscopy (Fig. 1). A striking feature for both HeLa cells and hepatocytes was the variable accumulation of fluorescent bile acids. In hepatocytes the variability was maintained through 6 days of primary culture and did not reflect damaged or permeabilized membrane as assessed by exclusion of propidium iodide. Correlative immunofluorescence revealed that, in HeLa cells, strong accumulation of CDCGamF corresponded with strong staining for ntcp. However, in hepatocytes, ntcp fluorescence was uniform and did not correspond with the accumulation of CDCGamF. Low accumulation of CDCGamF was seen in a portion of the cells despite their expression of ntcp. This suggests either that ntcp activity is downregulated in these cells or that other transporters or intracellular binding factors contribute more predominantly to the accumulation, or in this case lack of accumulation, of the fluorescent bile acids. Potential explanations for this cellular variation include the hepatocyte-isolation procedure itself or differences in the hepatic acinar zone from which the cells are derived (9). Studies that have looked at accumulation of fluorescent bile acids in vivo indicate that lithocholyllysyl fluorescein accumulates to a greater and somewhat variable extent in zone 3 compared with zone 1 of rat liver although cholyllysyl fluorescein and CGamF appear to accumulate uniformly (20, 29). The mechanism and specificity of this variation remains to be elucidated.

GRANTS

The work was supported by NIH Grants DK023026, DK041918, and DK041296.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.