Copper-dependent Interaction of Dynactin Subunit p62 with the N Terminus of ATP7B but Not ATP7A*

The P-type ATPase affected in Wilson disease, ATP7B, is a key liver protein required to regulate and maintain copper homeostasis. When hepatocytes are exposed to elevated copper levels, ATP7B traffics from the trans-Golgi network toward the biliary canalicular membrane to excrete excess copper into bile. The N-terminal region of ATP7B contains six metal-binding sites (MBS), each with the copper-binding motif MXCXXC. These sites are required for the activity and copper-regulated intracellular redistribution of ATP7B. Two proteins are known to interact with the ATP7B N-terminal region: the copper chaperone ATOX1 that delivers copper to ATP7B, and COMMD1 (MURR1) that is potentially involved in vesicular copper sequestration. To identify additional proteins that interact with ATP7B and hence are involved in copper homeostasis, a yeast two-hybrid approach was employed to screen a human liver cDNA library. The dynactin subunit p62 (dynactin 4; DCTN4) was identified as an interacting partner, and this interaction was confirmed by co-immunoprecipitation from mammalian cells. The dynactin complex binds cargo, such as vesicles and organelles, to cytoplasmic dynein for retrograde microtubule-mediated trafficking and could feasibly be involved in the copper-regulated trafficking of ATP7B. The ATP7B/p62 interaction required copper, the metal-binding CXXC motifs, and the region between MBS 4 and MBS 6 of ATP7B. The p62 subunit did not interact with the related copper ATPase, ATP7A. We propose that the ATP7B interaction with p62 is a key component of the copper-induced trafficking pathway that delivers ATP7B to subapical vesicles of hepatocytes for the removal of excess copper into bile.

In mammals, the overall copper status is mainly regulated by excretion of excess copper into the bile. This process is disrupted in Wilson disease, leading to massive accumulation of copper in the liver. Wilson disease is caused by mutations in ATP7B (reviewed in Ref. 5). Within the hepatocyte, ATP7B is located at the trans-Golgi network (TGN) 2 where it transports copper to apoceruloplasmin (6). When intracellular copper levels are in excess, it redistributes to a vesicular compartment located toward the biliary canalicular membrane where excess copper is eliminated from the cell into the bile either by exocytosis through fusion of copper-loaded vesicles with the membrane (7,8) or by effluxing copper directly across the apical membrane (9,10). When intracellular copper levels are returned to normal ATP7B cycles back to the TGN (9).
There are six metal binding domains within the N-terminal regions of ATP7B and the closely related copper ATPase defective in Menkes disease, ATP7A. Each domain is ϳ70 amino acids in length, and each contains one MXCXXC motif. The six repeating MXCXXC motifs bind 6 molar equivalents of copper via the conserved cysteines and are involved in regulating the copper transport activity and copper-regulated redistribution of the ATPases (10 -13). Several studies have revealed that a large portion of the N-terminal domain is not critical for the overall transport activity and the intracellular redistribution of the human copper-transporting ATPases (10, 14 -18). For example, the N-terminal region of ATP7B encompassing MBS 1-5 (amino acids 63-540) is not essential for copper-induced trafficking, with only one MBS close to the membrane channel necessary and sufficient to support trafficking (19). Hence the large N-terminal region of the ATPases that contain the six MBS may have additional roles and potentially contains other signals or motifs that could be important for the regulated trafficking and targeting of ATP7A and ATP7B (17, 20 -22).
Two proteins have been shown to interact with the N terminus of ATP7B. The copper chaperone ATOX1 interacts with the N termini of both ATP7A and ATP7B in a copper-dependent manner, and this interaction is essential for copper delivery to the secretory pathway (23)(24)(25)(26)(27). The recently discovered protein COMMD1 (MURR1), which is absent in Bedlington terriers suffering from copper toxicosis (28,29), interacts with the N terminus of ATP7B (29). The fact that COMMD1 interacts with ATP7B but not with ATOX1 or ATP7A (29) is consistent with its proposed role in copper excretion from the liver.
Despite their sequence and structural similarity and the presence of six MBS within the N termini of the copper ATPases, clearly there are sufficient differences in sequence and/or structure that confer specificity of interactions with proteins such as COMMD1 and differences in their trafficking pathways. The identification of a putative targeting signal in ATP7B that is not present or is different in ATP7A (20) suggests that additional specific protein interactions are likely to impact on their activity and/or trafficking.
Only a limited number of proteins are known to be critical for copper homeostasis, and little is known of the protein/protein interactions required for this process. To better understand the function and regulation of ATP7B in copper homeostasis, we have employed yeast twohybrid technology to identify additional proteins that interact with ATP7B. The yeast two-hybrid system is a powerful and sensitive tool for investigating protein interaction networks (30). Here we describe the screening of a human liver cDNA library using the N-terminal region of ATP7B as a bait to search for interacting partners.
This screen identified the protein p62 (dynactin 4; DCTN4), a subunit of dynactin as potentially interacting with ATP7B. This interaction was confirmed using yeast two-hybrid assays and co-immunoprecipitation from mammalian cells. We showed that the p62 subunit of dynactin interacts with the N terminus of ATP7B but not ATP7A. This interaction was copper-dependent and required the CXXC motifs and the region between MBS 4 and MBS 6. We propose that the ATP7B interaction with p62 is a key component of the copper-regulated trafficking pathway that delivers ATP7B to subapical vesicles of hepatocytes for the removal of excess copper into bile.
Yeast Two-hybrid Screening-The screening procedure employed was a modification of the method described by Gietz and Wood (34). S. cerevisiae YGH1 cells were sequentially transformed with pAS2-1/ ATP7B-N and a human liver cDNA library constructed in the pACT2 vector (Clontech) using the lithium acetate/salmon sperm carrier DNA/polyethylene glycol method. Transformants were selected on SD medium lacking tryptophan, leucine, and histidine supplemented with 5 mM 3-amino-1,2,4-triazole (Sigma). To confirm the interactions, HIS ϩ transformants were tested for lacZ expression using ␤-galactosidase colony-lift filter assays (64). True positives were verified by curing the bait plasmid from the positive yeast clones, isolating the library plasmid(s) from Leu ϩ /Trp Ϫ /(LacZ ϩ ) auxotrophs, and re-transforming library plasmids back into S. cerevisiae YGH1 either with the pAS2-1 vector alone or with pAS2-1/ATP7B-N to confirm the interactions. Transformants were then re-assessed for ␤-galactosidase activity. Plasmids that were confirmed to encode interacting proteins were transformed into E. coli DH10B, and the cDNA inserts were subjected to sequence analysis to determine the nature and identity of the encoded protein.
To obtain a quantitative estimate of the extent of the interaction, liquid ␤-galactosidase assays that employed chlorophenol red-␤-D-galactopyranoside (CPRG) (Sigma) as the substrate were carried out following the Clontech protocol (64). For each culture within a single experiment, the ␤-galactosidase activity of triplicate samples was measured. The values obtained represent ␤-galactosidase activity calculated in Miller units using the average value obtained from at least three independent experiments Ϯ S.D.
DNA Sequence Analysis-Sequencing of plasmid DNA was carried out using ABI PRISM BigDye Terminator chemistry (Applied Biosystems) with reactions analyzed on an Applied Biosystems 3730S capillary sequencer apparatus at Micromon DNA Sequencing Facility, Monash University (Melbourne, Australia). Sequence analysis and basic sequence manipulation were performed using Sequencher (Gene Codes Corp.). Electronic data base searches were conducted using the BLASTN and BLASTP algorithms on the NCBI website (www.ncbi.nlm.nih.gov).
Immunoblot Analysis of S. cerevisiae YGH1 Transformants-Total protein extracts from S. cerevisiae YGH1 transformants were prepared using the trichloroacetic acid extraction method as described in the Yeast Protocols Handbook (64). Approximately 50 g of protein was fractionated by SDS-PAGE (7.5 and 12.5% (v/v) acrylamide). Proteins were transferred to Hybond-Cϩ nitrocellulose membranes (Amersham Biosciences) using a Trans-Blot semi-dry electrophoretic transfer appa-ratus (Bio-Rad) and detected with either sheep-derived, affinity-purified NC36 anti-ATP7B antibody (diluted 1:500 (19)) or rabbit-derived antibodies directed against the GAL4 DNA-BD (2 g/ml) (Sigma) or the GAL4 AD (1 g/ml) (Sigma). Secondary antibodies used for detection were a rabbit anti-goat IgG peroxidase conjugate (Sigma) and a goat anti-rabbit IgG peroxidase conjugate (Chemicon), respectively. Protein bands were detected after incubation in Lumi-Light Chemiluminescence blotting substrate (Roche Applied Science) and visualized using the Luminescent Image Analyzer LAS-3000 (Raytest Isotopenmessgeraete GmbH).
Co-immunoprecipitation and Immunoblot Analysis-Co-immunoprecipitation of interacting proteins from mammalian cells was carried out using the ProFound TM Mammalian Co-immunoprecipitation Kit (Pierce), and experiments were carried out and optimized using the manufacturer's instructions. Specifically, 200 g of the affinity-purified anti-ATP7B antibody (NC36) (19), 200 g of the sheep-derived, affinity-purified anti-ATP7A antibody (R17-BX) (35), and as a control the sheep preimmune serum were covalently immobilized to the aminereactive antibody coupling gel. Menkes patient fibroblast cells that expressed either ATP7A (A12-H9) or ATP7B (WND-16) (33) were seeded in equivalent numbers and grown to confluence in 75-cm 2 flasks for 72 h either in the absence or presence of the copper chelators BCS (100 M) and D-penicillamine (100 M). Cells were lysed, and the lysates were applied to the gel according to the protocol provided. The protein complexes were eluted in the elution buffer provided, boiled in sample buffer with added dithiothreitol (20 mM final concentration), and fractionated by SDS-PAGE (7.5% (v/v) and 12.5% (v/v) acrylamide). Proteins were transferred to Hybond-Cϩ membrane (Amersham Biosciences) using a Trans-Blot semi-dry electrophoretic transfer apparatus (Bio-Rad) and analyzed by immunoblotting with either the NC36 anti-ATP7B antibody (diluted 1:500 (19)), the R17-BX anti-ATP7A antibody (diluted 1:1000) (35), or a goat-derived anti-human dynactin p62 polyclonal antibody N-17 (diluted 1:1000; Santa Cruz Biotechnology). The secondary antibody used for detection was a rabbit anti-goat IgG peroxidase conjugate The 260-amino acid region of p62 encoded by the library plasmid pACT2/p62 and that interacted with ATP7B-N is shaded in gray. B, ␤-galactosidase activity of S. cerevisiae YGH1 co-transformants. S. cerevisiae YGH1 was co-transformed with plasmids that encoded p62 and either the wild-type ATP7B N terminus (ATP7B-N) or the wild-type ATP7A N terminus (ATP7A-N) as indicated. A control was included that was co-transformed with pAS2-1/ATP7B-N and the pACT2 AD vector. ␤-Galactosidase activity of overnight cultures was measured using CPRG as the substrate. The values shown represent ␤-galactosidase activity calculated in Miller units and using the average value obtained from at least three independent experiments Ϯ S.D. C, expression of GAL4 DNA-BD and AD fusion proteins. Panels show the GAL4 DNA-BD and AD fusion proteins (indicated above each lane) that are expressed in each co-transformant as detected by immunoblot with affinity-purified GAL4 DNA-BD and -AD antibodies (Ab) (Sigma), respectively. The proteins in each lane were derived from the same co-transformants that were tested for ␤-galactosidase activity shown directly above. The position of protein molecular mass markers (Fermentas) are shown on the left in kilodaltons (kDa).
(Sigma). Protein bands were detected after incubation in Lumi-Light chemiluminescence blotting substrate (Roche Applied Science) and visualized using the Luminescent Image Analyzer LAS-3000 (Raytest Isotopenmessgeraete GmbH).

RESULTS
Screening of a Human Liver cDNA Library for Binding Partners of the ATP7B N-terminal Domain-To identify proteins that interact with the ATP7B N-terminal domain in vivo, a yeast two-hybrid screen of a human liver cDNA library was conducted using a bait protein that included the ATP7B N terminus from amino acids 1 to 657. This region includes all six metal binding domains to one amino acid before the first transmembrane domain. The cDNA encoding the entire ATP7B N terminus (ATP7B-N) was cloned into the vector pAS2-1 to create an inframe fusion with the GAL4 DNA-BD (pAS2-1/ATP7B-N). When transformed into S. cerevisiae YGH1 alone, or with the GAL4 AD vector (pACT2), this bait plasmid did not autoactivate expression of the HIS3 and lacZ reporter genes.
The pAS2-1/ATP7B-N plasmid and the liver cDNA library constructed in the pACT2 vector were sequentially transformed into S. cerevisiae YGH1. Interaction of the bait protein with a library-encoded protein was necessary to bring the GAL4 DNA-BD and -AD into close proximity to activate transcription of the reporter genes. Approximately 3 ϫ 10 6 transformants were screened for histidine prototrophy. Among the 2,534 colonies isolated as HIS ϩ , 365 were also LacZ ϩ . The bait plasmid was cured from positive transformants, and the library plasmid DNA was isolated from each of these clones, transformed into E. coli, isolated, and then retransformed into S. cerevisiae YGH1 that was pretransformed with either pAS2-1/ATP7B-N or pAS2-1 vector only. Transformants were tested for ␤-galactosidase activity. True positives were those that were positive for ␤-galactosidase activity when both library and bait plasmids were present but negative for ␤-galactosidase activity when the library plasmid and pAS2-1 vector were present. Plasmid DNA from the true positive clones was isolated, sequenced, and shown to encode a number of different cDNAs. A library plasmid that encoded the dynactin subunit p62 was chosen for further detailed analysis based on the presence of many cysteine residues within its amino acid sequence and the involvement of dynactin in protein trafficking. This plasmid contained an insert of 3.1 kb, which included 778 bp of a 1380-bp open reading frame plus 2.3 kb of 3Ј-untranslated sequence. The open reading frame encoded amino acid 200 to the stop codon of dynactin subunit p62 (260 amino acids) (Fig. 1A).
Dynactin Subunit p62 Interacts with the N Terminus of ATP7B but Not ATP7A-To confirm the specificity of the ATP7B-N/p62 interaction, S. cerevisiae YGH1 was co-transformed with pACT2/p62 and bait plasmids constructed in pAS2-1 that encoded either ATP7B-N (pAS2-1/ATP7B-N) or ATP7A-N (pAS2-1/ATP7A-N) fused in-frame to the GAL4 DNA-BD. The negative control was a transformant that contained pAS2-1/ATP7B-N and the empty library plasmid pACT2. Co-transformants were analyzed for histidine prototrophy and ␤-galactosidase activity, and the latter was measured qualitatively by the colony-lift filter assay (data not shown) and semi-quantitatively using a liquid culture assay (CPRG assay) (Fig. 1B).
These experiments demonstrated a significant and specific interaction between p62 and ATP7B-N (Fig. 1B). The negative control showed that the positive ␤-galactosidase result with ATP7B-N and p62 did not arise from interaction with vector-derived proteins. There was no interaction between p62 and the copper ATPase ATP7A, which is closely related in structure and function to ATP7B (Fig. 1B). Immunoblot analysis (Fig. 1C) of the co-transformants confirmed that the appropriate fusion proteins were expressed so that the negative ␤-galactosidase results cannot be attributed to the lack of protein expression.
Co-immunoprecipitation experiments were carried out to verify the p62/ATP7B interaction in a mammalian cell environment. Using the ProFound TM Mammalian Co-immunoprecipitation Kit (Pierce), cell lysates derived from a human Menkes fibroblast cell line that expressed either ATP7A (A12-H9) or ATP7B (WND-16) (33) were applied separately to columns prepared with either immobilized ATP7A or ATP7B primary antibodies, respectively, or preimmune serum. SDS-PAGE analysis of the immune complexes eluted from the columns showed that the anti-ATP7B antibody precipitated both ATP7B and endogenous p62 from the lysate, whereas the preimmune serum did not precipitate either protein (Fig. 2A). In contrast, the anti-ATP7A antibody precipitated ATP7A but not p62 from the ATP7A-expressing cell line (Fig. 2B), although endogenous p62 could be detected in the cell lysate. We concluded that the interaction between ATP7B and dynactin subunit p62 represented a true and specific interaction in mammalian cells. Dynactin p62 Interacts with the N Terminus of ATP7B in a Copperdependent Manner-To determine whether copper was required for the interaction between p62 and the ATP7B N terminus, S. cerevisiae YGH1 co-transformed with pAS2-1/ATP7B-N, and pACT2/p62 was passaged for 48 h in copper-limited medium prepared by the addition of the copper chelator, BCS, and the liquid ␤-galactosidase assay was repeated (Fig. 3A, panel i). The extent of the interaction was reduced 4-fold by 100 M BCS and was almost completely disrupted by the addition of 500 M BCS. The addition of BCS did not affect the interactions between the unrelated interacting pairs Bim and Bcl2 and Fos and Jun (Fig. 3A, panel ii). This observation showed that the reduced interaction between ATP7B-N and p62 in the presence of BCS was not because of an effect of BCS on the activity of ␤-galactosidase. The addition of 50 M CuSO 4 to the BCS-supplemented medium restored the ATP7B/p62 interaction to normal levels (Fig. 3B). There was maximal interaction with the addition of 100 M CuSO 4 . Note that protein levels were not affected by the variations in the copper content of the medium, so changes in the extent of the interaction were not a reflection of copper-induced changes in protein expression (Fig. 3B).
The copper dependence of the ATP7B/p62 interaction was confirmed in mammalian cells by co-immunoprecipitation experiments.
To deplete cells of copper, equivalent numbers of WND-16 cells were seeded into flasks and grown in the presence of 100 M BCS and 100 M D-penicillamine for 72 h. Using the ProFound TM Mammalian Co-immunoprecipitation Kit (Pierce), cell lysates were prepared, and samples containing equal amounts of total protein were applied to a column prepared with immobilized NC36 ATP7B primary antibodies. SDS-PAGE analysis of the immune complexes eluted from the column showed that as before the NC36 antibody precipitated both ATP7B and endogenous p62 from the cells grown in normal medium (Fig. 3C, panel i). However, when cells were grown in copper-depleted medium, there was a substantial reduction in the amount of p62 that was co-precipitated with ATP7B (Fig. 3C, panel  ii). There was no difference in the amount of p62 or ATP7B protein in each lysate indicating that the BCS treatment did not affect the expression levels of these proteins. Hence, we concluded that the reduction in the amount of p62 that was co-precipitated with ATP7B was because of copper depletion, confirming that copper was required for the ATP7B/p62 interaction in mammalian cells. Note that concentrations of BCS between 50 and 500 M have been shown previously to reverse or disrupt the copper-induced trafficking of the ATPases (7, 8, 10, 36 -38).  I, and B). As a control, S. cerevisiae YGH1 cells were co-transformed with yeast expression plasmids encoding either GAL4 AD-Bim and GAL4 DNA-BD-Bcl2 fusions or GAL4 AD-Fos and GAL4 DNA-BD-Jun fusions (A, panel ii). Cultures of all co-transformants were passaged for 48 h in copper-limited media supplemented with the indicated concentrations of the copper chelator BCS (A), or in copper-limited media supplemented with 500 M BCS and the indicated concentrations of CuSO 4 (B). The extent of the interaction was then measured in three independent experiments using the CPRG liquid ␤-galactosidase assay. Values shown represent the mean Ϯ S.D. Panels below the graph in B show immunoblots of the GAL4 DNA-BD/ATP7B-N and GAL4 AD/p62 fusion proteins that are expressed in each culture represented in the graph. Total protein extracts from the untreated and treated cultures were subjected to SDS-PAGE and immunoblotting with affinity-purified GAL4 AD (1 g/ml) (Sigma) and NC36 (diluted 1:500 (19)). An image of a portion of the Ponceau Red-stained membrane is also shown to demonstrate approximately equivalent protein loading in each lane. C, the ATP7B/p62 interaction is dependent on copper in mammalian cells. WND-16 fibroblast cells (33) were grown for 72 h in the absence or presence of 100 M BCS and 100 M D-penicillamine. Co-immunoprecipitation of interacting proteins was carried out using the ProFound TM mammalian co-immunoprecipitation kit (Pierce). Cell lysates from treated and untreated cells were added separately to a column prepared with immobilized ATP7B affinity-purified primary antibodies (19). Samples (20 l) from the first two elutions (Co-IP elutions 1 and 2) were fractionated on SDS-polyacrylamide gels. As controls for protein size and expression, aliquots of cell lysates were prepared by adding 5 l of sample buffer to 20 l of lysate and then loaded onto each gel. Following electrophoresis, proteins were transferred to Hybond-C nitrocellulose membranes, and probed as follows: panel I, NC36 (diluted 1:500 (19)), or panel ii, N-17, an anti-human dynactin p62 polyclonal antibody (diluted 1:1000; Santa Cruz Biotechnology). The sizes of prestained protein molecular mass markers (Fermentas) are shown on the right in kilodaltons (kDa). The positions of ATP7B and p62 proteins are indicated by arrows on the left.
The Interaction between ATP7B and p62 Is Dependent Upon the CXXC Motifs-To investigate the requirement for the CXXC motifs and the involvement of the different MBS within the ATP7B N terminus in the interaction with p62, bait proteins comprising the GAL4 DNA-BD fused to mutant ATP7B N-terminal regions were generated (Fig. 4A). A fusion construct was generated in which the cysteines in all six motifs were altered to serines (SXXS) (pAS2-1/ MBS1-6c/s). In addition, fusion constructs were generated that contained only metal binding domain 6 (deleting amino acids 63-540) (pAS2-1/MBS1-5del) or metal binding domains 1-3 (deleting amino acids 300 -599) (pAS2-1/MBS4 -6del). The fusion constructs were co-transformed into S. cerevisiae YGH1 with pACT2/ p62 and ␤-galactosidase activity measured (Fig. 4B). The interaction was negligible when all six CXXC motifs were mutated to SXXS, clearly demonstrating a requirement for these motifs in the interaction. There was no interaction between p62 and MBS4 -6del and a reduced but significant interaction between p62 and MBS1-5del.
These results indicated that the region between MBS 4 and 6 is essential for the p62/ATP7B interaction. Again, immunoblot analysis of co-transformants demonstrated expression of all fusion proteins (Fig. 4C).

DISCUSSION
ATP7A and ATP7B are key proteins in copper homeostasis, and a detailed understanding of the cellular constituents that regulate their activity and trafficking is required to complete the picture of how these molecules function. Under basal copper levels, ATP7B has a steady state distribution at the TGN (7,14). In elevated copper, ATP7B traffics to large multivesicular compartments within the cytoplasm of nonpolarized cells (7,14,19,39), and in polarized hepatocytes it redistributes to a subapical vesicular compartment associated with apical vacuoles that are reminiscent of bile canaliculi (7)(8)(9)(10). Recent studies from our group have provided support for a mechanism whereby the copper-loaded vesicles fuse with the apical membrane to expel the copper into the bile (8). B, GAL4 DNA-BD fusion constructs incorporating these mutations were co-transformed with pACT2/ p62 into S. cerevisiae YGH1 cells. The extent of interaction was then measured in three independent experiments using the CPRG liquid ␤-galactosidase assay. Values shown represent the mean Ϯ S.D. Panels below the graph show immunoblots of the GAL4 DNA-BD/ATP7B N-terminal mutation/deletion and GAL4 AD/p62 fusion proteins that are expressed in each culture represented in the graph. In the top right panel the expected protein bands are indicated by an asterisk. Protein extracts were subjected to SDS-PAGE and immunoblotting with affinity-purified GAL4 AD (1 g/ml) (Sigma) and NC36 (diluted 1:500 (19)). The position of protein molecular mass markers (Fermentas) are shown on the left in kilodaltons (kDa).
Little is known of the precise route or the associated proteins required for the trafficking of ATP7B through the cell. The N-terminal copper-binding motifs are present in all of the copper-transporting P-type ATPases, albeit in varying numbers, and play a critical but incompletely defined role in the transport and trafficking functions of both ATP7B and ATP7A (17,19,40,41). Each metal binding domain binds copper (11)(12)(13)22), but the presence of all six domains is not necessary for the catalytic activity or the copper-induced redistribution of the proteins (10,14,15,17,19). In particular, metal binding domain 6 appears to have an important role in copper transport and trafficking (17,19). The six metal binding domains are between 20 and 60% identical to each other, suggesting that they have functions that are distinct from just binding copper and possibly mediated by protein/protein interactions. In ATP7B, a putative vesicle targeting signal exists within metal binding domain 6 and may exert its effect through interactions with the cell sorting machinery (20). This signal is not present or is different in ATP7A supporting the notion that different proteins and interactions are likely to be involved in the different trafficking pathways of the ATPases in the cell. So far, two proteins are known to interact with the ATP7B N terminus. ATOX1 interacts with the N termini of both ATP7A and ATP7B in a copper-dependent manner to allow the delivery of copper to the ATPases (23)(24)(25)(26)(27). COMMD1, possibly involved in vesicular copper sequestration, also interacts with the N-terminal region of ATP7B but not ATP7A (29). Given the importance of the N-terminal region in both trafficking and copper transport, we undertook a yeast two-hybrid screening approach to identify additional interacting partners.
A human liver cDNA library was screened with the N terminus of ATP7B. One of the proteins confirmed to interact with ATP7B-N in yeast, dynactin subunit p62, was chosen for further analysis. Based on the proposed role of dynactin in tethering cargo to the retrograde microtubule motor protein dynein, the subunit p62 represented a candidate likely to be implicated in the copper-induced trafficking of ATP7B from the TGN to the subapical vesicular compartments. Yeast two-hybrid experiments showed that p62 interacted specifically with the N terminus of ATP7B but not with that of the related ATPase ATP7A. The lack of interaction between p62 and the ATP7A N terminus and also between p62 and ATOX1 (data not shown) confirmed that the interaction with ATP7B was specific. Co-immunoprecipitation confirmed the interaction in mammalian cells, while the inability of the ATP7A antibody to co-immunoprecipitate ATP7A and p62 further confirmed the specificity of the interaction.
Dynactin is a multisubunit complex that binds both to microtubules (42) and to the retrograde microtubule motor protein dynein (43,44). Cytoplasmic dynein is targeted to a variety of subcellular structures and requires the activity of dynactin for all of its known cellular functions that include organelle and vesicle transport (reviewed in Ref. 45). Although the specific function of the interaction between dynein and dynactin is unknown, there is evidence that dynactin enhances the microtubule binding capabilities and motility of cytoplasmic dynein (46) and links dynein to its cellular cargo (47). The 11 different dynactin subunits range in size from 22 to 150 kDa, and for many, their individual roles within the dynactin complex have been well characterized (reviewed in Ref. 48).
Dynactin subunit p62 is a ubiquitously expressed 53-kDa protein, with a punctate cytoplasmic and centrosomal subcellular localization, similar to other dynactin subunits (49). The function of the p62 subunit is not fully understood, but its absence does not significantly impact dynactin stability (50). The most prominent feature of p62 is its highly conserved cysteine-rich domain (amino acids 30 -114) in the N-terminal region in which 8 of the 11 cysteines in this region form a LIM domain (49,51,52). The LIM domain is a zinc binding protein/protein interaction domain (reviewed in Ref. 53), and it has been proposed that p62 uses this region to bind to other dynactin subunits or to membraneassociated phospholipids (48,49). There are five additional cysteine residues in the C-terminal half of p62, two of which form a CXXC motif that is highly conserved among homologs from Drosophila melanogaster, Caenorhabditis elegans, mouse, and Neurospora and hence is likely to be functionally important. The clone isolated from the liver cDNA library in this study encoded the C-terminal region of p62 (amino acids 200-stop (460)) that does not include the LIM domain but contains the five additional cysteine residues that may also be implicated in the copper dependence of the interaction between p62 and ATP7B.
Hepatocytes establish a horizontal polarity axis with a horizontal microtubule arrangement (54). The microtubules within these polarized cells are organized from sites near the apical surface and are oriented with their minus ends emanating at the apical plasma membrane and their plus ends attached to or near the basolateral surface (55,56). Hepatocytes use predominantly an indirect biosynthetic route to target their biliary canalicular membrane proteins initially to the basolateral membrane and associated endosomes, followed by transcytosis to the apical lumen (57). Transport to the apical domain by both direct and indirect routes involves microtubule-based motility. Previous studies have shown that the microtubule inhibitor nocodazole prevented copper-induced trafficking of ATP7B (9), and biliary copper excretion was inhibited by treating rats with colchicine, an inhibitor of microtubule polymerization (58). These observations suggest that microtubules are involved in the copper-induced trafficking of ATP7B to the apical surface of hepatocytes. Because dynein is a minus-end directed microtubule motor protein (59) and because ATP7B traffics toward the apical surface of polarized hepatocytes in elevated copper (9, 10), we propose that dynein/dynactin, through interaction between p62 and ATP7B, mediates microtubule-based copper-induced trafficking of ATP7B to the apical membrane via the indirect transcytotic route. Microtubules are not required for the delivery of proteins to the basal membrane (60,61). This observation is consistent with our finding that ATP7A, which in elevated copper traffics to the basolateral membrane of polarized epithelial cells (62,63), did not interact with p62.
The interaction between ATP7B-N and p62 was dependent on copper as is the interaction between the ATPases and ATOX1 (23)(24)(25)(26)(27). This copper dependence further implicates p62 and dynactin as key components in the copper-responsive trafficking pathway of ATP7B. The interaction between the ATP7B N terminus and p62 also required the conserved CXXC motifs of ATP7B, because mutation of the cysteines in all six metal binding domains disrupted the interaction. In the full-length ATP7B, this mutation abolished the copper-regulated redistribution of ATP7B from the TGN to subapical vesicular compartments (19) and the copper transport activity of ATP7B (17). Together these results suggest that the ATP7B/p62 interaction requires the copperbound MBS. Importantly, although the CXXC motifs were critical for the ATP7B/p62 interaction, because ATP7A and ATOX1 did not interact with p62, it is likely that the ATP7B/p62 interaction involves sequence and/or structural elements in addition to the CXXC motifs.
Experiments using the ATP7B-N deletion mutants localized the region important for ATP7B/p62 interaction to between MBS 4 and 6. There was significant but reduced interaction between p62 and the ATP7B-N MBS1-5del mutant. The full-length protein with this mutation could still traffic as wild-type ATP7B (19) and transport copper (17). In contrast, the MBS4 -6del protein did not interact with p62, and the full-length protein with this mutation could not traffic in response to copper (19) nor transport copper (17). These data link the transport and trafficking function of ATP7B to its ability to interact with p62. A recent study by Guo et al. (10) showed that the N-terminal 63 amino acids of ATP7B play a role in the copper-regulated trafficking of ATP7B and could potentially contain a targeting signal that directs the protein from the basolateral membrane to the apical surface. Although all of the wild-type and mutant constructs used in this study contained these N-terminal 63 amino acids (19), the relative contributions of this region and the ATP7B/p62 interaction in the copper-regulated trafficking of ATP7B in hepatocytes needs to be further investigated.
Our data clearly demonstrate a requirement for copper, functional CXXC motifs, and the region between MBS 4 and 6 for the interaction between ATP7B and p62 to occur. Based on our current understanding of the role of dynactin in the cell, we propose that this interaction is important for the copper-regulated trafficking of ATP7B to the apical surface of polarized hepatocytes.
It was proposed previously that as copper is bound to the N-terminal MBS of the ATPases, there is a conformational change, and a specific targeting signal within the N terminus is exposed, which enables the protein to traffic through interaction with components of the cell sorting machinery (17, 20 -22). Using our data in combination with that of Guo et al. (10), we propose a model for copper-induced conformational changes and subsequent microtubule-mediated apical trafficking of ATP7B in response to elevated copper concentrations. When intracellular copper levels are low, the MBS of ATP7B are not entirely saturated with copper, and the protein is predominantly located at the TGN. When copper levels become elevated, more copper is delivered to ATP7B and all available MBS become loaded with copper. The N-terminal region, replete with copper, undergoes a conformational change, and the protein traffics via a default pathway to the basolateral membrane. Here p62 interacts with the N terminus of ATP7B, in a mechanism that may involve the N-terminal 63 amino acids, to target the copper-bound protein for microtubule-mediated trafficking toward the biliary canalicular membrane. This model assumes indirect transcytosis of ATP7B to the apical membrane via the basolateral membrane. This model would also explain why p62 did not interact with the similar copper transporter ATP7A. The N termini of ATP7A and ATP7B differ significantly in the region between MBS 4 and 6 and within the region prior to MBS 1, with this latter region in ATP7A comprising only 10 amino acids, whereas ATP7B has 63 (10). Copper-loaded ATP7A would traffic to the basolateral membrane but would not be recognized by p62 for microtubule-mediated apical trafficking.
The identification of dynactin subunit p62 as an interacting partner of ATP7B represents an important discovery of a key cellular component implicated in the copper-regulated trafficking pathway of ATP7B. The discovery of this interaction and its dependence on copper will not only be instrumental in deciphering other signals and interactions that are integral to this pathway but is already providing insight into the molecular basis for the mechanistic differences between the trafficking pathways of ATP7A and ATP7B. This study is significant in furthering our understanding of the molecular mechanisms and pathways that lead to the maintenance of copper homeostasis and that will provide new targets for diagnosis and therapy of copper-related diseases.