| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
REPORT
1Department of Cell Biology, Johns Hopkins University School of Medicine, Baltimore, Maryland; 2Children's Hospital Boston, Boston, Massachusetts; and 3Department of Neuroscience, University of Connecticut Health Center, Farmington, Connecticut
Submitted 12 October 2006 ; accepted in final form 5 December 2006
ABSTRACT
We report for the first time on the copper-dependent behavior of endogenous ATP7A in two types of polarized intestinal epithelia, rat enterocytes in vivo and filter-grown Caco-2 cells, an accepted in vitro model of human small intestine. We used high-resolution, confocal immunofluorescence combined with quantitative cell surface biotinylation and found that the vast majority of endogenous ATP7A was localized intracellularly under all copper conditions. In copper-depleted cells, virtually all of the ATP7A localized to a post-TGN compartment, with <3% of the total protein detectable at the basolateral cell surface. When copper levels were elevated, ATP7A dispersed to the cell periphery in punctae whose pattern did not overlap with the steady-state distributions of post-Golgi, endosomal, or basolateral membrane markers; only 
8–10% of the recovered ATP7A was detected at the basolateral cell surface. These results raise several questions regarding prevailing models of ATP7A dynamics and the mechanism of copper efflux.
intestine; biotinylation; Caco-2; trafficking
Among the proteins known to be important in the regulation of copper levels at the organismal and cellular levels are two copper-transporting membrane ATPases, ATP7A (Menkes protein) and ATP7B (Wilson protein). They belong to a large family of cation-transporting P-type ATPases, which translocate ions across cell membranes using the energy of ATP hydrolysis (22). ATP7A is expressed in most tissues, except liver, where ATP7B is expressed. ATP7B is also in the kidney and parts of the brain (7). Mutations in the genes encoding these proteins cause disease. Menkes disease is a fatal condition of copper deficiency, because of a failure of dietary copper accumulated in the small intestine to be delivered to the circulation for distribution to the rest of the body (17). Wilson disease is a treatable condition of copper excess, because of a failure to excrete copper that has accumulated in the liver (12). The presentations of these diseases indicate that both copper-ATPases function in the cellular efflux of copper in addition to their role in transporting copper into the secretory pathways of cells. Importantly, copper efflux by ATP7B in hepatocytes is into the apical space, whereas that of ATP7A in intestine is into the basolateral space.
Because of their importance in disease, the ATP7A and ATP7B genes and proteins have been studied extensively (see reviews, Refs. 30, 48, 59, 60). Many mutations causing Menkes and Wilson diseases have been identified and studies conducted to understand a specific mutation's effect on protein expression, stability, or function (10, 18, 19, 40, 61). In addition, studies of the cellular basis of the normal proteins' dual functions have revealed that both ATP7A and ATP7B proteins change their intracellular locations in response to changes in copper levels (30). Under low-copper conditions, the proteins are reported to reside in the trans-Golgi network (TGN), whereas in high-copper conditions their steady-state locations are shifted. The prevailing view has been that ATP7A moves to the plasma membrane (PM) (24, 45) and ATP7B to what has been identified as a multivesicular body (25). However, to date, their trafficking has been studied predominantly in nonpolarized cells engineered to overexpress the proteins. Furthermore, high-resolution analysis of the intracellular compartments through which the two copper-ATPases move and function has not been reported.
To begin to understand the mechanisms of copper efflux by the copper-ATPases, we have focused on endogenous ATP7A in intestinal epithelia. ATP7A plays an essential role in intestinal cells in the delivery of dietary copper to the blood for subsequent distribution to the rest of the body. Therefore, in this study, we asked whether the protein traffics to the basolateral PM of enterocytes in vivo and Caco-2 cells, an accepted model of human small intestine. We report that ATP7A remained overwhelmingly intracellular under all copper conditions. Although it relocated to a peripheral location and small puncta when copper levels were elevated, only a small fraction of the protein could be detected in the basolateral PM. Our results, together with those from several other recent reports, have led us to revise the current model of ATP7A-dependent copper efflux to include a dispersed vesicular ATP7A-positive compartment that actively sequesters excess cytoplasmic copper.
MATERIALS AND METHODS
Materials
The following chemicals were used: cupric chloride hydrate, cupric sulfate pentahydrate, bathocuproinedisulfonic acid (BCS), L-lysine monohydrochloride, sodium m-periodate, sodium orthovanadate (Na3VO4), DL-DTT, Tween-20, saponin, cycloheximide, sodium fluoride, and imidazole from Sigma Chemical (St. Louis, MO); glycine, SDS, Tris, and Triton-X 100 from Fisher Scientific (Fair Lawn, NJ); and 16% paraformaldehyde, 2-methyl-butane (isopentane), HEPES, and ultrapure sucrose from Electron Microscopy Sciences (Hatfield, PA), EM Science (Gibbstown, NJ), Invitrogen (Grand Island, NY), and ICN Biomedicals (Aurora, OH), respectively. All other chemicals were from J. T. Baker (Phillipsburg, PA).
Cell Culture
Caco-2-E3V cells were seeded on filters (polycarbonate membrane, 0.4-µm pore, 6.5- or 24-mm diameter) from Costar (Cambridge, MA), cultured in DMEM (with 0.12 µM Cu) +10% FBS in a humidified 5% CO2 incubator at 37°C as described (20) and used at 2–3 wk postconfluency when cells were fully differentiated and polarized. Skin fibroblasts from a Menkes disease patient, MNKY/–, and patient's mother, MNK+/–, were seeded on six-well plates from Corning (Corning, NY) and cultured in DMEM + 10% FBS as previously described (14).
Antibodies
Rabbit anti-ATP7A (CT77 and 78) was raised against a peptide consisting of the COOH-terminal 20 amino acids of mouse ATP7A coupled by glutaraldehyde to keyhole limpet hemocyanin (Covance, Denver, PA) (57). IgG fractions of CT 77 and 78 were prepared, as was affinity-purified CT78. A mouse monoclonal antibody to the 
-subunit of chicken Na-K-ATPase was obtained from D. Fambrough (Johns Hopkins University, Baltimore, MD). Rat monoclonal anti-zonula occludens (ZO)-1 hybridoma cells (R40.76) were from B. Stevenson (University of Alberta, Edmonton, Canada) (56). Guinea pig and rabbit polyclonal anti-dipeptidyl-peptidase IV (DPP-IV) were prepared as described (52). Mouse monoclonals against -tubulin (clone DM1A), GAPDH, mannosidase II, and transferrin receptor (TfR) were from Sigma, Research Diagnostics (Flanders, NJ), Covance Research Products (Berkeley, CA), and Harlan Sera-Lab (Loughborough, UK), respectively. E-cadherin, EEA1, 
-adaptin, TGN38, Golgin-84, and syntaxin 6 monoclonal antibodies were from BD-Transduction Laboratories (Lexington, KY). Alexa-secondary antibodies were from Molecular Probes (Eugene, OR), and Cy3- or Cy5-conjugated secondary antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA).
Copper Treatments
In vivo. Young male Sprague-Dawley rats weighing 50–65 g (Charles River Breeding Labs, Wilmington, MA) were starved overnight then administered 1 ml (per 50 g animal weight) of freshly prepared 2 mM CuSO4 or 800 µM BCS, diluted in 10 mM HCl, each hour by gastric intubation for 5 h. At the end of hour 5, the rats were anesthetized with ether and guillotined, and the upper half of the jejunum was excised for further processing. For morphology experiments, we used the duodenum and upper 5 cm of jejunum. Biochemical samples were obtained from untreated rats by excising specific segments along the small intestine and scraping their mucosae into homogenizing buffer (3 mM imidazole, pH 7.4, 0.25 M sucrose, protease inhibitor cocktail). Homogenates were boiled (5 min) in reducing sample buffer (see Cell Surface Biotinylation and Western Blot Analysis) and analyzed by SDS-PAGE and Western blot.
In vitro. Filter-grown Caco-2 cells were cultured in basal medium then switched to BCS or CuCl2, which was added to the medium in both the apical and basolateral chambers for all experiments. The conditions used were as follows: 1) 3–4 h in 200 µM BCS or 200 µM CuCl2 for extreme copper-depleted and copper-loaded conditions; or 2) 10 µM BCS for 16–18 h followed by 1, 10, or 100 µM CuCl2 for 1, 2, or 4 h, for physiological copper conditions. Identical concentrations were applied to the MNK fibroblasts. Cells were then processed for indirect immunofluorescence or cell surface biotinylation.
For copper washout experiments (see 
Fig. 8), 10 µg/ml cycloheximide was included in each step of the experiment.
|
|
|
|
|
|
|
|
Tissue.
The excised jejunum was flushed with cold 0.9% saline, and 5 mM pieces were cut and fixed by immersion (1 h) at room temperature in freshly prepared 2% paraformaldehyde-lysine-periodate (33). Tissue blocks were incubated in 0.5 M sucrose/0.1 M NaPO4, pH 7.4 (4°C, 4–24 h), embedded in Optimum Cutting Temperature compound (Triangle Biomedical Sciences, Durham, NC), 10 µm sections prepared (Leica cryostat, Wetzlar, Germany), mounted onto glass slides and stored at –20°C. Tissue sections on slides were further fixed in 95% methanol (10 min, –20°C), rehydrated in PBS (3 x 10 min, room temperature), blocked in 2% BSA-PBS-0.2% Tx100 (30 min), incubated (1 h) in primary antibody diluted in blocking solution, rinsed in 0.2% BSA-PBS-0.02% Tx100 (3 x 10 min), incubated in secondary antibody (30 min) diluted in blocking solution, rinsed 2 x 10 min in 0.2% BSA-PBS-0.02% Tx100 then 1 x 10 min in PBS.
Cells. Caco-2 cells were rinsed in PBS, fixed in 3% PFA-PBS (4°C, 30 min), quenched in fresh 50 mM NH4Cl-PBS (4°C, 15 min), then permeabilized and blocked in 1% BSA-PBS plus 0.075% saponin (30 min). Cells were labeled with primary antibody diluted in blocking solution (1 h), rinsed (2 x 10 min, 0.1% BSA-PBS-0.0075% saponin), labeled with secondary antibody in blocking solution (30 min), and rinsed as before, and the membrane was cut out and mounted on a slide.
Mounting medium for both cell types was 25% glycerol in 100 mM Tris·HCl, 300 mM NaCl, pH 9.5–10.5, containing 2 mg/ml p-phenylenediamine (Sigma). Primary antibodies used were as follows: ATP7A (CT77, 1:400 of IgG), ZO-1 (1:200), DPP-IV (guinea pig, 1:1,000), Na/K-ATPase (1:500), TfR (1:50), -adaptin (1:100), TGN38 (1:5,000), mannosidase II (1:200), E-cadherin (1:200), EEA1 (1:50), and syntaxin 6 (1:50). Alexa-, Cy3-, or Cy5-conjugated secondary antibodies were used at 3–5 µg/ml.
Confocal images were acquired using an UltraView Confocal Imaging System (Perkin Elmer Life Sciences Division) or LSM Zeiss 510 Meta (Zeiss, Jena, Germany).
Quantitation of ATP7A Dynamics
Quantitative analysis of ATP7A in Caco-2 cells (Fig. 5) was performed using Volocity ver. 3.6.1 software (Improvision). Raw data (confocal stacks, 12-bit images), acquired under identical microscope settings, was entered and the "Classifier" tool was used to set thresholds described by signal intensity (of ATP7A) and structure size (µm3) that would exclusively select the larger supranuclear ATP7A-positive compartments at the TGN region. Thus ATP7A in these regions of interest was measured as the volume occupied by the ATP7A-positive TGN compartment, whose size and frequency decreased with increasing copper levels. Measurements were normalized to those of the 16 h BCS-treated cell sample in each experiment and expressed as "relative volume" of ATP7A in the TGN region. Data shown are the average of two to four stacks of images per condition, from three separate experiments.
Cell Surface Biotinylation and Western Blot Analysis
Filter-grown Caco-2 cells and MNK+/– fibroblasts on six-well plates were incubated overnight (37°C) in complete medium containing 10 µM BCS, then switched to 10 µM CuCl2 (2 h) for copper-induced ATP7A trafficking, before the cells were cooled down (1 h, 4°C). Cell surface biotinylation at 4°C was as follows: cells were rinsed in PBS then incubated 10 min in a borate buffer (154 mM NaCl, 10 mM boric acid, 7.2 mM KCl, and 1.8 mM CaCl2, pH 8), switched to borate buffer plus 0.5 mg/ml EZ-link-sulfo-NHS-LC-biotin (Pierce, Rockford, IL) for 15 min x 2, while the cells were gently shaken. For Caco-2 cells, the biotin was added to the apical chamber or the basolateral chamber. To quench the reaction, cells were washed 2 x 5 min with 100 mM glycine-PBS and subsequently rinsed 3 x 5 min with PBS. Cells were scraped into 600 µl of lysis buffer (1% Triton X-100, 20 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM Na3VO4, 50 mM NaF containing protease inhibitors: 2 µg/ml aprotinin, 1 mM benzamidine, 1 µg/ml pepstatin, 5 µg/ml antipain, 5 µg/ml leupeptin, 1 mM PMSF), sonicated, and sheared with a 26-gauge needle before centrifugation at 5,000 rpm for 15 min to remove insoluble cellular debris. The resulting supernatant (S) was used as the starting material, whereas the total pellet was analyzed to confirm solubilization of the proteins (Fig. 6A).
Biotinylated proteins were precipitated from 400–450 µl of the supernatant by incubation with neutravidin-agarose beads (Pierce) at 4°C for 2 h. After centrifugation, the supernatant was removed and saved, and protein bound to the beads was eluted in reducing sample buffer (final: 20 mM Tris·HCl, pH 8.8, 5% SDS, 15% sucrose, 2 mM EDTA, 5 M urea, 0.1 M DTT) by boiling for 5 min (eluate B1). This elution was repeated (eluate B2) to maximize recovery of biotinylated protein. The total eluate, aliquots of the starting material (S20, 20 µl and S40, 40 µl) and residual unbound material (Un, 32 µl) were analyzed by SDS-PAGE and Western blot (primary antibodies used: ATP7A, 1:2,000; E-Cadherin, 1:2,000; -tubulin, 1:5,000; Golgin-84, 1:500).
Labeled proteins were visualized and quantitated with Versadoc model 5000 and Quantity One software (Bio-Rad). The amount of biotinylated protein at the cell surface (B1+B2) was analyzed in two ways: 1) as percent of material recovered in B1, B2 and Un, i.e., [(B1+B2)/(B1+B2+Un)] x 100; or 2) as percent of input, i.e., [(B1+B2)/total input] x 100, where S1 and S2 were used to estimate total input. Both methods yielded similar results within each experiment, and data in Fig. 6D are reported using method 1, averaged from three separate experiments per condition.
RESULTS
Anti-ATP7A Antibodies Are Specific
Two polyclonal antibodies to the COOH terminus of ATP7A (CT77 and 78) were used in this study. The specificity of only CT78 is demonstrated in Fig. 1, but CT77 IgG behaved similarly (data not shown). We first used them to detect endogenous ATP7A in fibroblasts from a Menkes disease patient (MNKY/–) and his mother (MNK+/–). CT78 recognized a single band migrating at 180 kDa in the MNK+/– (Fig. 1A) but not the MNK-null fibroblasts. This single band is consistent with the size of ATP7A, which is 1,500 aa. A band of similar size was found in extracts of Caco-2 cells. The antibody also immunoprecipitated endogenous ATP7A from lysates of ATP7A-positive cell types (data not shown).
Differential Levels of ATP7A Protein Along the Small Intestine
We focused first on ATP7A in intestinal epithelia, because this is the cellular site where ATP7A delivers dietary copper to the blood for subsequent distribution to the rest of the body. To date, the specific intestinal region(s) across which copper-absorption occurs has not been fully described. We found that ATP7A levels were high in the duodenum and upper third of the jejunum (Fig. 1B), with evidence of animal and/or sampling variability. Other regions of the intestine expressed much lower amounts.
Endogenous ATP7A Redistributes When Copper Levels Are Changed
To identify the intracellular structures containing ATP7A under low- and high-copper conditions, we first depleted copper levels in vivo and in vitro and found dramatic shifts of endogenous ATP7A localization in both polarized cell types (Fig. 2, A and D). In rats fed a chelator (BCS) that depletes intracellular copper, the majority of ATP7A in enterocytes was concentrated in the supranuclear region (Fig. 2A), whereas in Caco-2 cells, the protein was predominantly in large puncta around the nucleus (Fig. 2D). The distribution of ATP7A in cells exposed to high copper levels was much different. In enterocytes in vivo, the protein was concentrated in small puncta close to the basolateral surface, although some ATP7A remained in the perinuclear region (Fig. 2B). In Caco-2 cells ATP7A was found in small puncta dispersed throughout the cytoplasm (Fig. 2E). In untreated rats or Caco-2 cells cultured under basal conditions, endogenous ATP7A displayed a bimodal distribution (Fig. 2, C and F).
To further characterize the juxtanuclear compartment, which in studies of ATP7A in nonpolarized cells was identified as the trans-Golgi network (TGN), we double stained sections and cells with antibodies to proteins associated with the Golgi complex. Of the markers we examined, syntaxin 6, a t-SNARE that functions in post-TGN trafficking, showed the most extensive colocalization with ATP7A both in intestinal and Caco-2 cells (Fig. 3, A–C, and data not shown). We saw less overlap of ATP7A with three markers that localize to different regions of the Golgi (shown in the intestine, Fig. 3, D–L).
In Basal and High Copper Levels, the Majority of ATP7A Is in Vesicles in the Basolateral Region
ATP7A has been reported to move to the plasma membrane in nonpolarized cells that have been exposed to high copper levels (e.g., Refs. 6, 15, 44, 46). However, our immunofluorescence results of ATP7A's distribution after comparable treatment did not indicate a similar localization. Therefore, we double labeled sections and cells with antibodies to ATP7A and two basolateral membrane markers (Fig. 4), the integral membrane proteins NaK-ATPase in intestine and E-cadherin in Caco-2 cells. As can be seen in Fig. 4, both proteins displayed very clear basolateral staining. However, neither pattern corresponded to that of ATP7A in the same cell. In fact, the discontinuous, vesicular character of ATP7A was very evident in Caco-2 compared with the smooth labeling of E-cadherin in these cells (Fig. 4, inset).
Endogenous ATP7A Traffics from the Post-TGN Region Into Vesicles in a Temporal and Copper-Dependent Manner
For the experiments described so far, we chose copper-depletion and loading conditions that were designed to obtain maximal responses of the copper-ATPases and had been used by others (e.g., Refs. 34, 58). However, we wanted to determine the extent to which ATP7A protein redistributed under conditions reported to be in the range of dietary copper (10–30 µM) (2, 50). For example, 80% of a 1 µM 64CuCl2 dose applied to the apical side of filter-grown Caco-2 cells was found in the basolateral medium 30–60 min later (29). Therefore, we first incubated filter-grown Caco-2 cells in 10 µM BCS for 16 h, to deplete intracellular copper levels and stage the protein in the TGN (Fig. 5A). After thorough rinsing, we then incubated cells in 1, 10, or 100 µM CuCl2 for 1 and 4 h and fixed and labeled them with anti-ATP7A antibody. By immunofluorescence, the protein showed significant redistribution after only 1 h at 1 µM Cu (Fig. 5B) and almost complete dispersal by 4 h (Fig. 5C). Quantitation of signal loss from large perinuclear punctae, which were labeled by the post-TGN marker, syntaxin 6 (not shown), indicated that 40% of ATP7A protein had moved out of these structures after only 1 h in 1 µM Cu (Fig. 5E). Interestingly, exposure to higher copper levels resulted in greater loss within the same time frame.
The Small Amount of Endogenous ATP7A at the Basolateral Surface in Copper-Depleted Cells Increases When Copper Levels Are Raised
We next used surface biotinylation to determine whether ATP7A was at the plasma membrane of either nonpolarized MNK+/– fibroblasts or polarized Caco-2 cells. We chose copper levels that were within the physiological range and would localize ATP7A at either of the two extremes, the TGN region or vesicles in the basolateral region. Cells were cultured overnight in 10 µM BCS to deplete copper, then transferred to medium containing 10 µM CuCl2 the next day for 2 h at 37°C. The cells were biotinylated at 4°C thereafter.
We first evaluated the solubility of ATP7A under low-copper and high-copper conditions and confirmed that >98% of total ATP7A was present in the supernatant fraction, in both MNK+/– fibroblasts (Fig. 6A), and Caco-2 cells (data not shown). After cell surface biotinylation, the supernatant fraction was incubated with neutravidin beads to precipitate any biotinylated material, which was then separated by SDS-PAGE and immunoblotted with various antibodies. In nonbiotinylated controls of MNK+/– cells (Fig. 6B) and Caco-2 cells (data not shown), none of the proteins examined were precipitated by the neutravidin beads, verifying the specificity of this assay.
Analysis of the precipitated biotinylated material from BCS-treated cells indicated that very low amounts of ATP7A were at the cell surface of MNK+/– fibroblasts (Fig. 6B) or the basolateral surface of Caco-2 cells (Fig. 6C). In copper-depleted fibroblasts, only 2.4 ± 0.5% of recovered ATP7A was detected at the cell surface (Fig. 6D). Upon a subsequent 2-h exposure to 10 µM CuCl2, this amount rose to 8.8 ± 0.3% in the fibroblasts. Similarly, in Caco-2 cells, 2.9 ± 0.5% and 8.9 ± 1.2%, of recovered ATP7A were detected at the basolateral surface (Fig. 6C) after copper-depletion and subsequent high-copper treatment, respectively. Thus fibroblasts exhibited a 3.8-fold increase (±0.75) in ATP7A at the cell surface when copper levels were raised and Caco-2 cells a 3.1-fold increase, ± 0.4.
Each biotinylation experiment contained several important controls. Two cytosolic proteins, Golgin84, a TGN resident, and -tubulin, a cytoskeletal protein, were used as negative controls for the cell surface biotinylation. Both proteins were absent in the biotinylated fraction of the assay (Fig. 6, B and C). E-cadherin was used as a positive control to assess the extent of biotinylation at the basolateral surface. We consistently obtained 91 ± 2.6% biotinylation of this marker regardless of the copper conditions used. Finally, 0.7–0.9% of recovered ATP7A and 1.5–1.6% of E-cadherin were detected at the apical surface of Caco-2 cells in low or high copper (Fig. 6C), confirming the domain-specific labeling of our assay.
Peripherally Located ATP7A Is in a Unique Vesicular Compartment
We next focused on identifying the vesicular compartment that contained the majority of ATP7A under elevated copper conditions. In nonpolarized cells ATP7A is reported to cycle between the plasma membrane and Golgi, leading us to reason that the protein might transiently reside in an endocytic or recycling compartment. Therefore, we double labeled cells and sections from animals exposed to high copper with antibodies to accepted markers of early and recycling endosomes: EEA1 and TfR, respectively. As can be seen in Fig. 7, the two markers were present in discrete puncta, but there was little to no overlap with ATP7A. The dispersed ATP7A did not overlap with the post-TGN marker syntaxin 6 (Fig. 7), nor with an endoplasmic reticulum marker (protein disulfide isomerase, results not shown).
Dispersed ATP7A Relocates to the Golgi Region When Copper Levels in the Medium Are Decreased
A key feature of ATP7A's dynamics in nonpolarized cells is its reversibility when copper levels are changed (e.g., Refs. 39, 44). Therefore, we asked whether endogenous ATP7A returned to the Golgi region when cells were sequentially exposed to basal copper levels, then high copper and finally BCS, to deplete copper. As shown in Fig. 8, Caco-2 cells initially displayed the bimodal distribution of ATP7A. When incubated with excess copper for 3 h, relatively more ATP7A was dispersed throughout the cytoplasm in small puncta than in large puncta marked by syntaxin 6. After washout of the copper and incubation of cells in BCS for an additional 3 h, ATP7A showed the opposite pattern, with relatively more in the Golgi and less at the periphery. Cycloheximide was present throughout this experiment, yet the ATP7A signal did not diminish, indicating that preexisting ATP7A was redistributing in response to copper levels in the medium.
DISCUSSION
We report for the first time on the copper-dependent behavior of endogenous ATP7A in two types of polarized intestinal epithelia, rat enterocytes in vivo and filter-grown Caco-2 cells, an accepted in vitro model of human small intestine. We used high-resolution, confocal immunofluorescence combined with quantitative cell surface biotinylation and found that the vast majority of endogenous ATP7A was localized intracellularly under all copper conditions. In copper-depleted cells, virtually all of the ATP7A localized to a post-TGN compartment, with <3% of the total protein detectable at the basolateral cell surface. When copper levels were elevated, ATP7A dispersed to the cell periphery in punctae whose pattern did not overlap with the steady-state distributions of post-Golgi, endosomal, or basolateral membrane markers; only 8–10% of the recovered ATP7A was detected at the basolateral cell surface. These results raise several questions regarding prevailing models of ATP7A dynamics and the mechanism of copper efflux: 1) Does the relatively small amount of ATP7A at the basolateral surface account for all of the copper efflux from Caco-2 cells? 2) What role do the dispersed ATP7A vesicles play in copper homeostasis? 3) Does ATP7A traffic via a unique exoendocytic pathway?
Copper Absorption in the Small Intestine
The intestinal sites and mechanisms for uptake of dietary copper from the gut lumen have not been fully defined. The recent generation of an intestine-specific knockout of the small copper transporter, Ctr1, yielded mice exhibiting neonatal defects in peripheral accumulation of copper. The result provides direct evidence that this transporter, which is expressed throughout the small and large intestine (66), plays an essential role in copper absorption from the diet (37). However, Ctr1-null mouse embryonic fibroblasts exhibit residual copper uptake, suggesting that additional copper transport activities must exist (27). These may include the divalent metal ion transporter, DMT1, which is expressed in the apical membrane of duodenal enterocytes and responsible for nonheme iron uptake (31). DMT1 also reportedly transports both ionic forms of copper (1, 11).
Our studies indicate that the majority of intestinal ATP7A in the rat intestine is expressed in the duodenum and upper jejunum. A similar expression pattern may exist in adult humans, given clinical reports of patients suffering hypocupremia years after gastric bypasses that divert stomach contents directly into the lower end of the small intestine (23). In contrast, studies in hamsters demonstrated that 64Cu absorption from the mucosal to serosal side occurred predominantly in the lower part of the small intestine, suggesting that ATP7A is distributed differently in this species (8). Thus the specific pattern of ATP7A expression along the small intestine most likely determines the sites at which copper is delivered into the bloodstream.
Recently, endogenous ATP7A was reported to be in the duodenal brush border and basolateral membranes of rats maintained for 6 wk on an iron-deficient diet (49). Both indirect immunofluorescence and subcellular fractionation were used. Furthermore, these same animals expressed increased amounts of ATP7A protein. We were unable to reproduce the reported findings. That is, depending on the copper treatment, endogenous ATP7A was either in the TGN region or dispersed toward the basolateral region in both duodenal and jejunal enterocytes. We never detected the protein in the apical membrane of cells as assessed by indirect immunofluorescence (data not shown). The fact that others (35) did not detect ATP7A in the apical surfaces of intestinal or Caco-2 cells using several approaches reinforces the view that intestinal ATP7A conveys copper vectorially to the basolateral environment.
Implications of ATP7A's Post-TGN Location
Studies in nonpolarized cells have shown that ATP7A resides in the Golgi region in basal or copper-depleted conditions. We confirmed and extended this observation in polarized cells by showing that intestinal ATP7A resides in a TGN subcompartment overlapping almost completely with the steady-state localization of syntaxin 6, a t-SNARE implicated in a variety of endosome-TGN trafficking steps (64). Interestingly, we found that the Golgi adaptor protein, AP1, overlapped much less than syntaxin 6 with ATP7A, even though the two markers have been colocalized in other cell types (e.g., Ref. 53). Syntaxin 6 forms a complex with other t-SNAREs, syntaxin 16 and vti1a (21), and a v-SNARE, VAMP4 (32), all of which are found on small synaptic vesicles and clathrin-coated vesicles in isolated brain synaptosomes. Biochemical studies have reported that syntaxin 6 accompanies glucose transporter 4 to the adipocyte plasma membrane in exocytotic vesicles formed in response to insulin stimulation (41, 53). Finally, ultrastructural studies in nonpolarized cells indicate that syntaxin 6 resides in tubulovesicular structures. Although the functional consequences of the apparent colocalization of ATP7A and syntaxin 6 in the post-TGN region are not presently clear, the fact that ATP7A leaves this region when copper levels are elevated but syntaxin 6 does not suggests that copper-dependent sorting occurs here. Further study, including immunoelectron microscopy, will be required to identify the ATP7A-positive structure(s) in the TGN and to clarify the intriguing relationship with syntaxin 6.
Our localization of ATP7A to a very late compartment of the secretory pathway of intestinal epithelial cells has implications for the posttranslational processing of the multicopper oxidase, hephaestin. A single-pass transmembrane protein with functional and structural homology to plasma ceruloplasmin, apo-hephaestin is synthesized and glycosylated in the ER and early Golgi, loaded with copper via the action of ATP7A in a post-TGN compartment as we have shown, then targeted to the basolateral plasma membrane, where it collaborates with the intestinal iron exporter, ferroportin, to effect Fe3+ release into the circulation (reviewed in Ref. 65). Interestingly, copper chelation of cultured T84 intestinal cells results in the retrieval of apo-hephaestin from the Golgi to the ER, where it retrotranslocates and is degraded by cytosolic proteasomes (36). Therefore, we would argue that the post-TGN compartment containing ATP7A and syntaxin 6 must be capable of sensing the copper status of this compartment, perhaps through some step in ATP7A's catalytic cycle, and redirecting excess apo-hephaestin back to the ER.
Revision of the Long-Standing Model of ATP7A's Copper-Dependent Trafficking
The results of this and recent studies (2, 35, 39) are at variance with a widely held view that in high copper ATP7A traffics from the TGN to the plasma membrane, where it effluxes copper out of the cell (see INTRODUCTION). We found that, after elevation of copper levels, endogenous ATP7A redistributed predominantly to novel vesicles localized in the basolateral region, not in the plasma membrane of intestinal or Caco-2 cells. The results of our quantitative biotinylation experiments show very clearly that the amount of ATP7A in the plasma membrane represents a minor fraction of the total cellular pool. Camakaris and colleagues (39) came to a similar conclusion using biotinylation to study the dynamics of ATP7A in their copper-resistant CUR3 cells, which express 70-fold more ATP7A than do their CHO-K1 parents.
We propose a revised model (Fig. 9), which incorporates results from this and other studies (e.g., Refs. 4, 35, 39, 62). In the model, elevation of intracellular copper (Fig. 9, left) leads to sorting of membrane-bound ATP7A in a post-TGN compartment into small vesicles that disperse toward the basolateral membrane. Excess (toxic) Cu1+ is temporarily stored in these novel ATP7A-positive vesicles. The copper-loaded vesicles periodically release their contents into the basolateral milieu by exocytosis; in doing so, the vesicle membrane fuses with the plasma membrane, increasing ATP7A levels there. Given the small percentage of total cellular ATP7A in the plasma membrane when copper levels are elevated, the protein must be rapidly and continuously retrieved from that site. Alternatively, before its fusion with the plasma membrane, the copper-loaded vesicle might become depleted of ATP7A through a maturation-type process, similar to that in the endosomal system (54). As long as excess copper is present in the cytoplasm, ATP7A continues to cycle between at least two membrane compartments, pumping Cu1+ across the membrane bilayer at these sites, but with different consequences. At the plasma membrane, Cu1+ is transported directly into the basal environment, whereas in intracellular vesicles it is trapped in a closed environment. When cytoplasmic copper levels fall (Fig. 9, right), ATP7A is retrieved from both compartments and delivered to the post-TGN compartment.
|
At present, we have few definitive answers to these questions, but there are intriguing clues from the literature. For example, Voskoboinik et al. (62) reported that isolated membrane vesicles enriched in ATP7A from CUR3 cells expressing the exogenous protein accumulated luminal 64Cu. At the time, they ascribed this activity to inside-out plasma membranes containing ATP7A, although it is now likely that the majority were intracellular in origin. Assuming some of the vesicles contain active ATP7A, there must then be a shunt to balance the luminal Cu1+ charge. Could ClC4, the intracellular chloride channel reported to be required for metallation of the secretory protein ceruloplasmin (63), accompany ATP7A in the dispersed vesicles? The isolation and full characterization of the dispersed vesicles could answer some of these questions.
In their detailed biotinylation study of ATP7A in CUR3 cells, Pase et al. (39) measured several kinetic parameters that relate to the model described above. The authors postulated the existence of an intracellular ATP7A pool, which was in rapid, dynamic equilibrium with ATP7A in the plasma membrane. They estimated that the two pools constituted 40% of the total cellular ATP7A, each being 20% at 500 µM copper, leaving 60% of cellular ATP7A elsewhere. Furthermore, they suggested that a distinct pool of ATP7A remained in the TGN to carry out the copper-ATPase's biosynthetic function; this conclusion was based on the finding that surface-biotinylated ATP7A molecules retrieved to the TGN in low copper were preferentially recruited to the cell surface when copper levels were raised again. Although the size of the TGN pool was not estimated in the Pase study, we would argue that it must be much smaller than 60%, since we found no detectable protein in the post-TGN compartment of Caco-2 cells after 1 h in 100 µM copper (Fig. 5, D and E). Pase et al. reported a similar finding when they exposed CUR2 cells to 500 µM copper; these cells express 10-fold more ATP7A than the parent CHO-K1 line. Thus, if we assume 10% of ATP7A is in the TGN in high copper, a large fraction (50%) of the dispersed ATP7A vesicles remains functionally unaccounted for. Could they be sequestering Cu1+?
Intriguing evidence from mutagenesis studies suggests that ATP7A and ATP7B traffic as acyl-phosphate intermediates (4, 47). Does dephosphorylation, which completes the catalytic cycle, put the copper-ATPases into a "retrieval" conformation? That is, is unloaded ATP7A retrieved to an earlier compartment and only moved forward by loading with copper and becoming phosphorylated? This mechanism might account for the acceleration of ATP7A's exit rate from the post-TGN compartment as copper levels rise (Fig. 5 and Ref. 39). That is, as long as the protein is copper loaded it would move forward, ultimately to the plasma membrane. Thus vesicles progressively closer to the plasma membrane could be loaded with more Cu1+.
What structural signals and cellular machinery direct ATP7A's itinerary in polarized epithelial cells? A di-L motif in ATP7A's COOH terminus was shown to target the protein to the basolateral region of polarized MDCK cells (13) and retrieve it from the plasma membrane region of unpolarized cells (43). Furthermore, a PDZ-binding motif at ATP7A's COOH terminus appears to play a role in targeting and maintaining the protein in the basolateral region of MDCK cells (13). La Fontaine and colleagues (55) used yeast two hybrid to identify a binding protein (PISP/AIPP1) with a single PDZ domain, which also binds to a plasma membrane Ca-ATPase. Its role in ATP7A targeting/efflux is not yet known.
Does ATP7A Traffic via a Unique Exo-Endocytic Pathway?
An important issue is the mechanism by which dispersed ATP7A is retrieved when copper levels are lowered. Use of dominant-negative mutants by several groups suggested that neither a caveolin- nor clathrin-mediated process was used (5, 26). These authors assumed that most of the ATP7A was in the plasma membrane, which the quantitative results discussed above show not to be the case. Thus the mechanism of ATP7A's endocytosis needs to be revisited.
In apparent contradiction to reports by others, our results indicate that the ATP7A vesicles present in copper-loaded intestinal cells are not part of the known endosomal system, as defined by the steady-state localizations of EEA1 and transferrin receptor, accepted markers of early and recycling endosomes, respectively (42). Petris and Mercer (44) reported in CHO cells that endocytosed transferrin and exogenous myc-tagged ATP7A briefly (5 min) colocalized when copper levels (200 µM) were lowered, suggesting that ATP7A moves through an early endosomal compartment. Endosomal rabs 5 and 7 were also reported to colocalize with overexpressed ATP7A in unpolarized cells exposed to high (189 µM) copper levels (38). Because we were unable to localize either rab using antibodies currently available, we cannot rule out this latter possibility. However, we would argue that the abundance of distinct tubulovesicular organelles in the peripheral and perinuclear regions of most cells makes decisions about possible overlap of fluorescence signals difficult. Additionally, the use of nonphysiological copper levels could have influenced the behavior of ATP7A in these studies.
In conclusion, the results we have presented, together with other data in the literature, suggest that 1) the ATP7A-positive vesicles seen when copper levels rise represent a novel compartment through which ATP7A traffics, 2) some of these vesicles may store copper temporarily, and 3) the vesicles periodically fuse with, but do not accumulate at, the basolateral surface to facilitate copper efflux by ATP7A.
GRANTS
This research was supported by National Institutes of Health Grants R01-DK-32949 (B. Eipper), P01-DK-072084, and R01-GM-064645 (A. Hubbard).
ACKNOWLEDGMENTS
We thank Mark Donowitz and Sandra Guggino for reading a draft of the manuscript.
Address for correspondence: A. Hubbard, Dept. of Cell Biology, Johns Hopkins Univ. School of Medicine, 725 N. Wolfe St., Baltimore, MD 21210 (e-mail: alh{at}jhmi.edu)
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.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  L. T. Braiterman, S. Heffernan, L. Nyasae, D. Johns, A. P. See, R. Yutzy, A. McNickle, M. Herman, A. Sharma, U. P. Naik, et al. JAM-A is both essential and inhibitory to development of hepatic polarity in WIF-B cells Am J Physiol Gastrointest Liver Physiol, February 1, 2008; 294(2): G576 - G588. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Linz, N. |
