Identification of p38 MAPK and JNK as new targets for correction of Wilson disease-causing ATP7B mutants

Antibodies, Reagents, and DNAs

Antibodies against phospho-p38, phospho-JNK, phospho-extracellular signal-regulated kinase (ERK), p38, JNK, and ERK were from Cell Signaling Technology (Beverly, MA). The DNAs of ATP7A and its mutants were already reported.13 DNAs of flag-tagged MKK3, MKK4, MKK6, and MKK7 were obtained from Addgene (Cambridge MA). To obtain the R778L, D765N, L776V, A874V, and L1083F mutants of ATP7B, the phosphorylated enhanced green fluorescent protein (GFP) C1-ATP7B construct was used as a template and site-directed mutagenesis was performed according to the manufacturer's instructions using the QuickChange kit (Stratagene, La Jolla, CA). Other reagents, antibodies, DNAs, and viruses have been reported.14

Cell Culture

HeLa and HepG2 cells and primary mouse hepatocytes were grown in Dulbecco's modified Eagle's medium supplemented with fetal calf serum 10% (decomplemented for HepG2 cells), l-glutamine, and penicillin/streptomycin.

Immunoprecipitation and Mass Spectrometric Analysis

To compare interactomes of ATP7B^H1069Q-GFP or ATP7B^WT-GFP, the tagged proteins were immunoprecipitated from cell lysates with the anti-GFP antibody. Pull-downs were fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and the protein bands were digested with trypsin. Peptide mixtures were analyzed by nano-chromatography tandem mass spectrometry (MS/MS) on a CHIP MS Ion Trap XCT Ultra (Agilent Technologies, Palo Alto, CA). Peptide analysis was performed using data-dependent acquisition of one MS scan (mass range 400-2000 m/z) followed by MS/MS scans of the three most abundant ions in each MS scan. Raw data from nano liquid chromatography-MS/MS analyses were employed to query a nonredundant protein database using in-house MASCOT software (Matrix Science, Boston, MA).

Microarray Analysis

To identify the effects of ATP7B^H1069Q expression on the transcriptome, total RNA was extracted from HepG2 cells (expressing either ATP7B^H1069Q-GFP or ATP7B^WT-GFP) and hybridized on Affymetrix GeneChips. All of the raw microarray data were formatted for the Gene Expression Omnibus to comply with Minimum Information About a Microarray Experiment and Microarray Gene Expression Database group standards (Gene Expression Omnibus number GSE51818). Microarray analyses were carried out with R, a free software environment. Genes that passed all filtering criteria were entered into the DAVID Gene Functional Annotation Tool. The gene ontology (GO) options GOTERM_BP_ALL and GOTERM_MF_ALL were selected, and a functional annotation chart was generated. A maximum P value of 0.05 was chosen to select only the significant categories.

Statistical Analysis

Data were expressed as mean values ± standard deviation. The Student t test was used to compare differences between two groups by GraphPad Prism 6 software (GraphPad Prism Inc., San Diego, CA). Statistically significant differences between them are indicated by the following: *P < 0.05, **P < 0.01, and ***P < 0.001.

Other Methods

Adenoviral and vesicular stomatitis viral infection, DNA transfection, RNA interference, quantitative polymerase chain reaction, fluorescent microscopy, immuno-electron microscopy (EM), inductively coupled plasma mass spectrometry (ICP-MS), and western blot analysis were done as reported.14 For details see the Supporting Information.

Expression of the ATP7B^H1069Q Mutant Induces Activation of p38 and JNK

We used unbiased approaches to understand which molecular pathway could be targeted to correct mutants causing WD. We first sought to determine how ATP7B^H1069Q mutant expression affects the transcriptome of hepatic cells. Comparison of microarrays from ATP7B^H1069Q-expressing and ATP7B^WT-expressing cells revealed 626 differently expressed genes (GSE51818). GO analysis of these hits indicated enrichment in genes that regulate cell death and response to stress (Fig. 1B; Supporting Table S1). Several of these genes (SQSTM1, TXNIP, IL6R, FOSL1, MAP2K6, and NET1) have been reported to operate in mitogen-activated protein kinase (MAPK) signaling pathways mediated by two proapoptotic stress kinases, p38 and JNK.15 This might be linked to WD pathogenesis because reactive oxygen species (ROS) accumulating in WD3 can stimulate the activity of p38 and JNK.15-18 Indeed, several ROS-sensitive genes (SOD2, MSRA, TP53I3) were up-regulated in ATP7B^H1069Q-expressing cells. This prompted us to investigate whether the activity of these kinases rises upon ATP7B^H1069Q expression. Western blot analysis revealed that phosphorylated p38 and JNK increased significantly in cells expressing the mutant (Fig. 1C), indicating that both kinases were activated. In the meantime the MAPK kinase/ERK pathway of the MAPK signaling cascade was not activated as the levels of phosphorylated ERK remained unaltered in mutant-expressing cells (Supporting Fig. S1A).

We also compared the interactomes of ATP7B^WT and ATP7B^H1069Q. Proteomics analysis revealed 31 ATP7B^WT-specific and 51 ATP7B^H1069Q-specific interactors (Supporting Tables S2 and S3). GO analysis indicated that specific binding partners of ATP7B^WT were enriched in “intracellular trafficking” proteins (Fig. 1D). In contrast, the ATP7B^H1069Q interactome contained numerous proteins belonging to the “unfolded protein binding,” “proteasome,” and “ubiquitination” GO categories (Fig. 1D). These differences are in line with an enhanced association of the mutant with the components of the ER-associated quality control machinery, resulting in ER retention of the mutant.9, 10 Notably, specific MAPK pathways, which include p38 and JNK, have recently been shown to affect proteostasis of proteins such as cystic fibrosis transmembrane conductance regulator (CFTR)19 and to regulate the expression of ER quality control/degradation genes.15 We thus examined whether activation of p38 and JNK might lead to the accelerated degradation of ATP7B^H1069Q. Indeed, overexpression of the upstream activators of p38 (MKK3 and MKK6) or JNK (MKK4 and MKK7) reduced ATP7B^H1069Q levels, indicating increased mutant degradation (Fig. 1E).

On the whole, the above observations indicate that induction of p38 and JNK may be triggered by ATP7B^H1069Q expression. This activation, in turn, may promote ER retention and degradation of the mutant, an increase in ROS levels (due to Cu accumulation), and further perturbation of ER proteostasis, generating a vicious circle (Fig. 1F). In this context, p38 and JNK represent regulatory centerpieces and potential druggable targets to prevent ER retention of the mutant and, hence, to rescue its localization and function.

Suppression of p38 and JNK Rescues Appropriate Localization and Trafficking of the ATP7B^H1069Q Mutant

To test whether deactivating the p38 and JNK pathways facilitates ATP7B^H1069Q export from the ER, we suppressed the activity of p38 and JNK first by using specific chemical inhibitors, SB202190 (SB90) and SP600125 (SP125), and then by RNA interference. We initially evaluated the impact of p38/JNK-specific inhibitors and small interfering RNAs in HeLa cells, given that silencing of the genes of interest can be easily achieved in this cell type. In cells treated with the Cu chelator bathocuproine disulfonate (BCS), ATP7B^WT was mainly detected over the Golgi membranes, while ATP7B^H1069Q remained in the ER (Fig. 2A, upper row). Neither p38 and JNK inhibitor (added for 24 hours) affected ATP7B^WT (Supporting Fig. S2A), but both reduced ATP7B^H1069Q within the ER and improved the delivery of the mutant to the Golgi (Fig. 2A[upper row],B). The correction of ATP7B^H1069Q localization was remarkably effective even at quite low concentrations (Supporting Fig. S2B). Interestingly, the combination of p38 and JNK inhibitors did not further increase mutant relocation from the ER to the Golgi (Supporting Fig. S3), suggesting that the two inhibitors target similar molecular pathways for correction of ATP7B^H1069Q. In control experiments the ERK inhibitor failed to correct the mutant from the ER to the Golgi (Supporting Fig. S1B). This indicates that only p38 and JNK branches of the MAPK signaling cascade are involved in mutant retention in the ER. Importantly, acute suppression of p38 and JNK did not impact ER export of ATP7B^H1069Q (Supporting Fig. S4). Therefore, inhibitors apparently act through the transcriptional regulation of quality control and/or trafficking genes rather than by reducing the direct phosphorylation of the mutant or components of its trafficking machinery.

Next, we investigated whether the inhibitors rescue Cu-responsive trafficking of ATP7B^H1069Q. CuSO₄ addition induced redistribution of ATP7B^WT from the TGN to the plasma membrane (PM) and peripheral vesicular structures but failed to induce ATP7B^H1069Q relocation from the ER (Fig. 2A, lower row). Instead, when corrected by p38/JNK antagonists, ATP7B^H1069Q efficiently moved from the Golgi to the PM and vesicles in response to elevated Cu (Fig. 2A,C).

The above results were confirmed by biochemistry and EM. Western blot exhibited reduced amounts of ATP7B^H1069Q in the PM subcellular fractions in comparison to ATP7B^WT, while both p38 and JNK inhibitors significantly increased quantities of the mutant delivered to the cell surface (Supporting Fig. S5). In turn, EM revealed ATP7B^H1069Q mainly over the ER membrane profiles, while cells incubated with p38 antagonists contained the mutant in TGN membranes (Fig. 2D). In response to CuSO₄ treatment inhibitor-treated cells delivered ATP7B^H1069Q from the TGN to the cell surface and endosome/lysosome-like “vesicles” (Fig. 2D), which normally receive ATP7B^WT upon Cu overload.14 Thus, it turns out that, once corrected to the Golgi by p38/JNK inhibitors, the mutant reacts to increasing Cu by trafficking to Cu excretion sites, just as ATP7B^WT does.

To further verify the above conclusions, we silenced different isoforms of p38 (MAPK11-MAPK14) and JNK (MAPK8-MAPK10) in cells expressing ATP7B^H1069Q (Fig. 3A). Both reduction in ER retention and recovery of Golgi and vesicle targeting of the mutant were detected after depletion of MAPK8, MAPK11, and MAPK14 (Fig. 3B,C). In addition, we sought to suppress MAP3K11, the upstream activator of both p38 and JNK,20 to evaluate whether this treatment improved ATP7B^H1069Q recovery from the ER. MAP3K11 has been reported to effectively correct Δ508-CFTR.19 Indeed, depletion of MAP3K11 resulted in strong reduction of ATP7B^H1069Q in the ER (Fig. 3B,C).

We next investigated whether inhibitors of p38 or JNK overcome the ER retention of other ATP7B mutants with residual activity.7, 11 Correction was achieved for D765N, L776V, and R778L mutants but not for A874V and L1083F variants (Fig. 4). Meanwhile, p38 and JNK antagonists failed to correct ATP7A mutants with post-Golgi trafficking defects (Fig. 5), indicating that inhibitors act mainly at the ER level. In summary, our observations suggest that suppressing p38 or JNK allows various ATP7B mutants to escape ER retention in a manner that depends on the type of mutation.

p38 and JNK Inhibitors Rescue ATP7B^H1069Q Trafficking and Targeting in Hepatic Cells

As a further step, we tested whether suppressing p38 and JNK is efficient at rescuing the mutant in WD-relevant cellular systems such as HepG2 cells and primary hepatocytes.21 In HepG2 cells, ATP7B^WT moved from the Golgi to the vesicles and cell membrane in response to Cu (Fig. 6A), while most of the ATP7B^H1069Q was arrested in the ER, regardless of changes in Cu concentrations (Fig. 6A). However, incubation with either p38 inhibitor or JNK inhibitor allowed significant amounts of ATP7B^H1069Q to reach the Golgi and to be transported toward the PM and associated vesicles in response to Cu overload (Fig. 6A,B). Moreover, the ATP7B^H1069Q mutant was targeted to the appropriate canalicular surface domain in polarized HepG2 cells (Supporting Fig. S6). Importantly, primary hepatocytes exhibited a similar response to p38 inhibitor and JNK inhibitor, which rescued ATP7B^H1069Q from the ER and allowed it to traffic toward Golgi and post-Golgi Cu excretion sites (Fig. 6C,D). Therefore, hepatic cells tend to deliver the ATP7B mutant to the right compartments upon suppression of p38 and JNK signaling pathways.

Correction of the ATP7B^H1069Q Mutant With p38 or JNK Inhibitors Facilitates Cu Excretion

We reasoned that correcting ATP7B^H1069Q from the ER to the appropriate Cu excretion sites should reduce the total amount of intracellular Cu. To test this hypothesis we used ICP-MS. ICP-MS measurements revealed higher concentrations of the metal in cells expressing ATP7B^H1069Q, while treatment with p38 or JNK inhibitor significantly reduced intracellular Cu levels (Fig. 7A).

To further support these findings, we used copper sensor 3 (CS3), which becomes fluorescent in the presence of bioavailable Cu.22 We found low CS3 signal in cells expressing ATP7B^WT, whereas CS3 fluorescence was higher in mutant-expressing cells (Fig. 7B). p38 and JNK inhibitors significantly decreased CS3 signal in cells expressing ATP7B^H1069Q (Fig. 7B,C). More importantly, reduction in CS3 fluorescence (and hence in Cu levels) correlated with the amount of ATP7B^H1069Q expressed in inhibitor-treated cells (Supporting Fig. S7). Collectively, the above findings indicate that correcting the mutant to the appropriate compartments with p38/JNK antagonists allows the cells to eliminate excess Cu.

p38 and JNK Inhibitors Reduce Degradation of ATP7B^H1069Q by Improving Mutant Sorting Into the Secretory Pathway

In the ER the failure of misfolded protein to pass the quality control check directs such a protein to degradation.23 Therefore, we analyzed whether p38 or JNK inhibitors counteract ATP7B^H1069Q degradation. We found that the ATP7B^H1069Q protein product increased in response to p38 and MAPK inhibition (Fig. 8A), while the levels of its messenger RNA remained the same in treated cells (not shown). This indicated that suppressing p38 and JNK attenuates degradation of the mutant. To support this conclusion we treated cells with cycloheximide and found that p38 and JNK inhibitors significantly slowed down the kinetics of mutant turnover (Fig. 8B,C). Thus, our results suggest that suppression of the p38 and JNK pathways allows ATP7B^H1069Q to circumvent ERAD.

How do p38 and JNK antagonists suppress ATP7B^H1069Q degradation? The first possibility is that they directly inhibit the ERAD components that comprise the proteasome.23 Being protected from ERAD, aberrant ATP7B molecules would reside in the ER longer and, therefore, may have a higher chance to be incorporated into carriers that transport them from the ER to the Golgi. To test whether this is the case, we treated ATP7B^H1069Q-expressing cells with the proteasome inhibitor MG132. MG132 increased ATP7B^H1069Q levels (Fig. 8D) but failed to overcome mutant retention in the ER (Fig. 8E). This indicates that inhibitor-mediated recovery of ATP7B^H1069Q from the ER does not involve ERAD components.

Therefore, ATP7B^H1069Q may be protected from degradation as a result of accelerated trafficking from the ER. To verify this we first observed whether p38 and JNK inhibitors accelerated overall trafficking rates in the secretory pathway. This was not the case as the trafficking kinetics of the generic cargo protein VSVG was not altered in SB90-treated or SP125-treated cells (Supporting Fig. S8).

We further hypothesized that the suppression of p38 and JNK might specifically accelerate trafficking rates of the ATP7B^H1069Q mutant from the ER to the Golgi. To test this hypothesis we used selective photo-bleaching, which allowed us to analyze protein exchange between different compartments in live cells.24 We bleached ATP7B^WT-GFP or ATP7B^H1069Q-GFP in the Golgi region and observed how the material from the ER recovered GFP signal in the bleached area (Fig. 8F). Recovery of ATP7B^H1069Q-GFP fluorescence in the Golgi reached the plateau with the same dynamics (t_1/2 ≈ 1 min) in both control and inhibitor-treated cells (Fig. 8G). However, the amount of ATP7B^H1069Q-GFP that reappeared within the Golgi over the same time interval was markedly higher in cells incubated with p38 or JNK inhibitor (Fig. 8F,G). These observations indicate that suppressing p38 and JNK does not shorten the time needed for a given ATP7B^H1069Q molecule to be transported from the ER to the Golgi. Instead, p38 and JNK antagonists allow a higher number of ATP7B^H1069Q molecules to be sorted from the ER and transported to the Golgi during this interval.

To further support this point we evaluated the distribution of ATP7B^H1069Q at the ER export sites (ERES) using immuno-EM. ERES contain coated buds (Fig. 8H-J, arrowheads) emerging from the ER cisternae and associated vesicular-tubular membranes (Fig. 8H-J, empty arrows) that carry newly synthetized proteins to the Golgi. Normally, membrane proteins directed into the secretory pathway (like ATP7B^WT) undergo enrichment in the ERES membranes (Fig. 8H, empty arrows) over regular ER cisternae.25 Such concentration in ERES was not detected for ATP7B^H1069Q. Indeed, ATP7B^H1069Q-associated gold particles decorated mainly ER cisternae (Fig. 8I, arrows) and were scarce in the coated buds (Fig. 8I, arrowhead) and neighboring vesicles (Fig. 8I, empty arrows). Incubation with either p38 or JNK inhibitor increased the density of ATP7B^H1069Q in the ERES buds (Fig. 8J, arrowhead) and vesicular profiles (Fig. 8J, empty arrows) and reduced ATP7B^H1069Q signal in ER cisternae (Fig. 8J, arrows; quantification in Fig. 8H). This trend indicates that inhibiting p38 and JNK facilitates sorting of ATP7B^H1069Q from the ER into nascent Golgi-directed transport intermediates.