ApoB mRNA editing requires the coordinated assembly of a multi-protein editosome [14
] that governs substrate and site specificity [39
], the proportion of substrate edited [10
] and possibly export of the edited product from the nucleus. Thus, regulation of editosome assembly and trafficking of editing complexes is necessary for efficient editing. ACF is an obligate component of the editing machinery and is the site-specific RNA-binding protein [39
] that docks APOBEC-1 to cytidine 6666. Recently, phosphorylation of ACF [26
] was identified as a mechanism whereby apoB mRNA editing can be regulated. Hyper-phosphorylation of serine residues, in active nuclear 27S editosomes, was implicated in ethanol-stimulated editing [26
To better understand the role of ACF phosphorylation in basal and metabolically stimulated editing, the sites of phosphorylation and the enzymes responsible were characterized. The expression of all mutants was validated to be at similar levels through the selection of McArdle cell lines stably expressing each variant. While it is likely that overexpressed wtACF or mutant ACF competed with endogenous ACF leading to the observed changes in editing activity, the overall abundance and subcellular distribution of endogenous ACF was unchanged in response to ectopic expression of phosphorylation site mutants. We have demonstrated in this report that mutation of S154 and S368 to alanine inhibited the ability of McArdle cells to regulate (increase) editing activity in response to ethanol. Mutation of these residues to aspartic acid fully stimulated editing activity even in the absence of ethanol treatment.
S154 and S368 were predicted as phosphorylation sites, potentially targeted by either PKA or PKC. Protein kinase activator studies demonstrated that only PKC activators stimulated editing activity and enhanced ACF phosphorylation in hepatocytes. The role of PKC in ACF phosphorylation was further supported by in vitro protein kinase assays in which liver expressed PKC isoforms phosphorylated recombinant ACF, whereas PKA had to be used at 10-fold higher concentrations to affect ACF phosphorylation. Thus our data support the role of PKC phosphorylation of ACF64, particularly PKC isoforms α β II and ζ, in ethanol regulated hepatic editing activity.
Immunological data indicated that rat liver ACF possesses both serine and threonine phosphorylation sites although only serine phosphorylation has been observed by radiolabeling during metabolic stimulation [26
]. We believe this discrepancy to be due to inherent differences in the experimental systems used. Detection of phosphorylated residues by antibodies does not require phosphate turnover (unlike metabolic labeling with 32
P). These data suggest that threonine phosphorylation is subject to slower turnover than serine phosphorylation in rat primary hepatocytes and/or that phosphothreonine is significantly less abundant. Our mutagenesis studies were designed to identify sites of phosphorylation that affect ACF64 function during metabolically stimulated editing. Given that threonine residues were not identified in this mutagenic screen, our collective data only support the role of regulated ACF64 serine phosphorylation in the metabolically modulation of apoB mRNA editing. Phosphorylated threonines are likely constitutive or subject to slow phosphate turnover. ACF threonine phosphorylation however could play an inhibitory role, by maintaining editing at low levels under basal metabolic conditions. Consistent with this possibility are data showing that inhibition of protein synthesis stimulated editing in primary hepatocytes [25
]. This could be due to reduced protein synthesis of a negative regulator of editing, such as a protein kinase active on ACF threonines. If constitutively expressed, a protein kinase with activity on ACF T49/T50 and T176 for example could inhibit editing activity. The phenotype of S145D and S368D suggested that ethanol stimulated phosphorylation of S154 and/or S368 will override the inhibitory effects of threonine phosphorylation and activate editing activity without necessarily requiring dephosphorylation of phosphothreonine sites. We therefore predict that at steady state hepatic ACF64 will be phosphorylated at different sites and to varying degrees depending on the metabolic demand. ACF containing multiple phosphorylations was predicted by two-dimensional gel analysis [26
Hepatic ethanol-stimulated editing [41
] coincides with hyper-phosphorylation of ACF [26
]. Our results demonstrate that liver expressed PKC isozymes α β II and ζ have the greatest ability to phosphorylate recombinant ACF. PKC θ over expression has been reported to modulate editing in rat hepatoma cells, whereas PKC isozymes α β II, ε and ζ were unable to modulate editing [43
]. However, the authors did not provide direct evidence that individual isoforms phosphorylated proteins involved in editing. Moreover, PKC θ is not a physiologically relevant isozyme since it is not expressed in liver [44
]. Further, the relative expression level of each PKC isoform was not indicated and therefore, non-selective phosphorylation of proteins may have been induced experimentally by overexpression of protein kinase. In this regard, although PKA is predicted to phosphorylate ACF and is regulated by ethanol [45
], PKA had very low activity in vitro on recombinant ACF64 and PKA-specific activators did not modulate editing in primary hepatocytes (). Neither direct activation of PKA with cAMP analogs nor indirect activation via adenylate cyclase stimulated editing. Conversely, activation of PKC with indolactam V significantly enhanced apoB editing () and PKC isozymes phosphorylated purified ACF in vitro. Our data is in agreement with results previously published demonstrating that activation of PKC induces editing in intestinally derived Caco-2 cells [43
Although phosphorylated ACF has only been recovered in the nucleus of hepatocytes, the cellular domain in which ACF is phosphorylated remains to be determined. In vivo, PKC substrate specificity is believed to be imparted by the subcellular distribution of the enzyme and substrate [46
]. Therefore, PKC interaction with and phosphorylation of ACF in the perinuclear environment, at some point during nuclear import or at some point in time subsequent to nuclear import remain formal possibilities.
ACF is a single polypeptide that comprises three tandem RRMs [6
]. The protein itself has been implicated in apoB mRNA-binding, APOBEC-1 binding and protection of edited apoB mRNA from NMD [6
]. In general, each RRM is composed of four antiparallel β-sheets flanked by two α-helices (reviewed in [47
]). RNA recognition occurs through the β– strands and their connecting loops. Other complex embellishments to the RRM fold have been documented that allow protein-protein interactions at the α-helices [36
], or β-strands [50
]-, although the nature of the ACF-APOBEC-1 interaction is unknown at present. In HuD, ssRNA makes contact primarily to the β-sheets of RRMs 1 and 2 (, red residues), whereas the outward facing αhelices are exposed to solvent () [35
]. Based on ACF sequence homology to HuD and the conserved structure of the RRM fold, S154 of ACF is predicted to localize within the second helix of RRM 2, where it would be accessible to protein kinases, but would not be expected to contact RNA or affect RNA-binding when phosphorylated. As such, it is plausible that this residue could be involved in protein-protein interactions (). The three-dimensional model for the position and orientation of S154 within RRM 2 is consistent with previous studies demonstrating that treatment of extracts with calf-intestinal alkaline phosphatase reduced co-immunopurification of APOBEC-1 with ACF, but did not affect ACF’s ability to bind to apoB RNA [26
]. Regulation of protein-protein interactions by changes in the phosphorylation state of one or more proteins has been well documented [52
]. Additionally, phosphorylation may alter tertiary fold promoting protein-protein interactions as is the case with CREB and CREB binding protein [54
Another residue of interest in ACF is S171, which is a glutamine in HuD. This residue is located in the loop betweenβ–strands 2 and 3 of RRM 2. Loop 3 is the most genetically divergent region among RNP proteins and is important for substrate specificity [47
]. Q171 of HuD is 4 Å from the non-bridging phosphate oxygen of U5, suggesting that the introduction of negative charge in this area could affect substrate specificity irrespective of the phosphorylation state. Editing activities for the S171D mutant were 15% and 28% greater than untreated McArdle cells in the absence and presence of ethanol, respectively. The introduction of negative charge by aspartic acid mutation in S171D may have contributed to the slight elevation in editing activity, the data suggest that it is unlikely that S171 is an ethanol regulated phosphorylation site.
Our mutagenesis analyses predicted that some serine/threonine residues in and of themselves are important for ACF structure and also editing. The introduction or ablation of negative charge at these residues could affect protein folding, protein-protein interactions, protein-RNA interaction and thus function independent of protein phosphorylation. In this case, mutagenesis of these sites to alanine and aspartic acid would disrupt editing. Case in point, S132 is conserved in rat and human ACF as well as in HuD and is located in between RRMs 1 and 2. The data suggested that mutation to alanine at this site was tolerated since editing was comparable to wild type ACF64. However, mutation to aspartic acid was detrimental. Although serine residues are polar, the R group is uncharged at pH 7.0. Mutation to alanine, a nonpolar amino acid, with only a methyl moiety as the R group would be less likely to interfere with protein folding. To the contrary, substitution with negatively charged aspartic acid could alter protein folding. Therefore, S132 is unlikely to be a phosphorylation site.
Limited structural information precludes modeling outside of ACF RRMs 1 and 2. Consequently, mutations amino terminal to amino acid 55 (T49/T50) and those carboxy terminal to amino acid 223 (i.e. S368) cannot be evaluated in the context of this model. However, S368 is located within ACF’s nuclear localization sequence (amino acids 360–401) [33
]. ACF is a nucleocytoplasmic shuttling protein related to hnRNP proteins [33
] and its subcellular distribution is sensitive to metabolic perturbations [33
]. Significantly, phospho-ACF accumulates in the nucleus of hepatocytes treated with ethanol or insulin [14
]. The S368 to alanine mutation was refractory to ethanol, whereas the aspartic acid mutation stimulated editing in untreated McArdle cells to levels comparable to cells treated with ethanol. These data suggest that phosphorylation of S368 may modulate ACF nucleocytoplasmic shuttling. In fact, several examples exist in the literature where phosphorylation mediated nuclear localization ([53
] reviewed in [58
]). The coordinated action of PP1 [26
] and PKC could regulate that subcellular distribution of ACF, and therefore the proportion of apoB mRNA that is edited.
In conclusion, the data suggest a model of editing regulation in McArdle cells whereby ACF is phosphorylated minimally on serine and threonine residues in basal McArdle cells. Metabolic stimulation of editing by ethanol may be accompanied by phosphorylation of S154 and S368 by PKC. Protein phosphatase 1 activity has been implicated in removing phosphate from metabolically regulated sites of ACF phosphorylation, enhancing export of ACF from the nucleus [26
] and reducing editing activity. Our findings suggest that phosphorylation of ACF by PKC in response to ethanol is part of the mechanism for nuclear import of ACF and activation of editing activity. In this regard, PKC and PP1 activities on ACF are predicted to act in concert to modulate the overall phosphorylation status of ACF, regulating its protein interactions and subcellular distribution.