ARD1 was first described in 1993 as the deduced protein product of clones isolated from human and rat genomic cDNA libraries that encoded an about 18-kDa ADP-ribosylation factor (Arf) sequence at the C-terminus of a 64-kDa molecule (Mishima et al. 1993). Human ARD1 coding region cDNA hybridized with 3.7- and 4.1-kb mRNAs from all rat tissues examined. Both recombinant, full-length ARD1 and its Arf domain (M403-A574), activated cholera toxin ADP-ribosyltransferase activity, at that time a defining characteristic of Arf function, whereas the non-Arf (M1-K402) fragment of ARD1 did not. Using recombinant proteins on nitrocellulose membranes, only the Arf domain (and a recombinant ARF3), not the full-length ARD1, bound radio-labeled GTP (Mishima et al. 1993).
Further investigation of GTPase activity of the same recombinant proteins in solution provided evidence that the N-terminal non-Arf region of ARD1 acted as a GTPase-activating protein (GAP) domain for the Arf moiety (Vitale et al. 1996). Bound GDP dissociated from the Arf domain much more rapidly than it did from the full-length ARD1, but after addition of the non-Arf domain, GDP dissociation was slowed, essentially to that of GDP release from the intact ARD1. Thus, the ARD1 molecule comprises both GDI (GDP-dissociation inhibitor) and GAP domains that can influence biological activity of the Arf domain, although the extent to which these domains regulate its GTPase cycle in cells was not demonstrated.
Regulation of ARD1 Activity
The Arf domain of ARD1 was not a substrate for a partially purified Arf GAP from rat spleen that enhanced GTP hydrolysis by ARFs 1, 3, 5, and 6 as well as Arf-like ARL proteins. Nor did the non-Arf part of ARD1 display GAP activity toward potential substrates other than its own Arf domain (Ding et al. 1996). As often cautioned, the extent to which relationships of GAP (and GEF) activities assessed in vitro compare to those in living cells can be difficult to establish.
Structural requirements for functional interaction of the Arf and non-Arf domains of ARD1 were defined in some detail using site-specific mutagenesis and chimeric proteins (Vitale et al. 1997a). Replacement of seven Arf1 amino acids with those in the corresponding positions of the Arf domain in ARD1 enabled it to associate with the non-Arf domain that enhanced its GTPase activity. Further deletion of 15 amino acids from its N-terminus produced an Arf molecule structurally and functionally equivalent to the ARD1 Arf domain. Additional effects of numerous specific amino acid replacements in ARD1 and those of several phospholipids or detergents on interactions and/or activities of the GTP-binding Arf and regulatory non-Arf domains were also reported (Vitale et al. 1997a). Regions in the non-Arf domain responsible for acceleration of GTP hydrolysis by the intrinsic GTPase activity of the Arf domain, plus adjacent sequence necessary to establish the required physical interaction, were also identified. Evidence for interaction of acidic N427 and E428 (plus P432) in the Arf region with basic R249 and K250 in the non-Arf domain was also obtained as regulated release of bound GDP and control of GTP hydrolysis are equally important for temporal continuity of the Arf domain GTPase cycling.
Structural elements of the ARD1 GDP dissociation-inhibitor (GDI) region were similarly characterized (Vitale et al. 1997b). Although both this action and the enhancement of GTPase catalytic activity involve intramolecular conformational changes, these might well differ in energetic costs and/or the effects on them of environmental changes. Early studies had shown that bound GDPβS was released more slowly from the recombinant Arf domain than from intact ARD1 (Vitale et al. 1996). Experiments with recombinant proteins and/or their fragments more clearly established the discrimination between GDP and GTP ligands. It was further demonstrated that 15 amino acids immediately preceding the Arf domain were responsible for slowing dissociation of GDP, but not GTP, and site-specific mutagenesis revealed importance of the hydrophobic amino acids for stabilization of ARD1-bound GDP (Vitale et al. 1997b). Even as our understanding of structure-function relationships in these actions of the ARD1 molecule continues to improve, their relationships to and integration with ARD1 functions in critical reactions of the innate immune system to viral infection appear to present more new and intriguing questions.
Cellular Localization of ARD1
After multiple attempts to identify endogenous ARD1 at intracellular sites in several tissues, Vitale et al. (1998b) were able to show single protein bands of about 64 kDa using antibodies immunoreactive with different parts of the ARD1 molecule in affinity-purified membranes from human liver presumed representative of lysosome or Golgi structures (Vitale et al. 1998b). Systematic microscopic observations of subcellular distribution of overexpressed EGFP-tagged ARD1 revealed perinuclear fluorescence consistent with Golgi localization in <5% of cells after 4 h, but present in almost 20% by 6 h and approaching 60% after 10 h. At the same time, fluorescence in widely dispersed vesicular structures, apparently lysosomes, was recorded in only <5% at 6 h and 40% of cells at 10 h. Between 12 and 50 h, both vesicular and perinuclear fluorescence was seen in about 60% of cells, suggesting the possibility of ARD1 cycling between Golgi and lysosomal organelles (Vitale et al. 1998b).
In attempts to identify appropriate localization signals in the ARD1 molecule, intracellular distribution of >20 overexpressed ARD1 fragments was evaluated microscopically with particular attention to YXXL motifs in the Arf domain thought to be potentially responsible for Golgi localization (Vitale et al. 2000b). Selective mutation of one or both of the sequences indicated that each could contribute to ARD1 presence in Golgi and that to reach the lysosomes, passage of ARD1 through the Golgi was required. A pentapeptide sequence, KFERQ, had been implicated as a targeting signal for lysosomes. Mutagenesis in 344KTLQQ348 and 369KQQQQ373 sequences of the non-Arf region of ARD1 demonstrated that the latter was critical for lysosomal localization (Vitale et al. 2000a).
ARD1: E3 Ubiquitin Ligase Activity and Arf Domain Function
E3 ubiquitin ligase activity of GST-ARD1 or its RING finger domain (residues 1–110) was demonstrated in vitro using the recombinant proteins plus pure mammalian E1, E2 (usually UbcH6), ATP, and ubiquitin (Vichi et al. 2005). Activity was abolished by deletion or mutation of the ARD1 ring structure. Like activities of the Arf and GAP domains, that of the E3 ligase was apparently independent of the rest of the molecule and was unaffected by addition of GTPγS or GDP. Whether that is true also in the intracellular environment is not known. Although the ubiquitinated proteins failed to react with ARD1 antibodies, their increasing amounts and size during assays were paralleled by decreasing amounts of free ARD1 protein, consistent with auto-ubiquitination. Three bands of modified UbcH6 (E2) did react with UbcH6 antibodies as well as with those against ubiquitin (Vichi et al. 2005). Later studies of ARD1−/− mouse embryo fibroblasts stably expressing constructs for induced synthesis of ARD1 or a (C34A, H53A) mutant lacking E3 ligase activity supported the conclusion that auto-ubiquitination regulated ARD1 degradation and was responsible for maintaining its concentration at very low levels in all cells studied (Meza-Carmen et al. 2011). Mutant ARD1 without E3 ligase activity accumulated in cells incubated with proteasomal inhibitors to levels sevenfold those usually seen.
Cytohesin-1 had been shown earlier to be a specific guanine nucleotide-exchange factor (GEF) activator for ARD1 (Vitale et al. 2000b). Since cytohesin-1 (and other cytohesins) had been reported to modify receptor tyrosine kinase action, effects of ARD1 or its mutants on EFGR were explored (Meza-Carmen et al. 2011). There was no evidence of E3 ligase involvement, but Arf domain function was clearly important. Actions in vesicular trafficking are probably the most extensively studied of all Arf functions. These require continuity of Arf cycling between inactive GDP-bound and active GTP-bound states. Overexpression of mutants with single amino acid replacements that abolish GTP binding (T418N), thereby blocking activation, or prevent GTP hydrolysis (K458I) both interrupt cyclic activation/inactivation, causing accumulation, respectively, of GDP or GDP-liganded proteins (thus referred to as ARD1-GDP or ARD1-GTP mutants). Amounts of EGFR were higher in ARD1-GDP and lower in ARD1-GTP cells than in those expressing wild-type (WT) ARD1 with a functional Arf domain, and without significant differences in RNA levels (Meza-Carmen et al. 2011). All findings were consistent with the conclusion that relatively rapid EGFR turnover enables more facile control of receptor protein levels and functions than would alterations in gene expression and necessary subsequent protein adjustments. Regardless of total levels, about 80% of the EGFR was on the cell surface. It was notable also that in ARD1-GDP and –GTP cells differences in total amounts of TGFβR III and insulin receptor (IR) were similar to those of EGFR, although the limited IR data did not reach statistical significance (Meza-Carmen et al. 2011). Potential functions of ARD1 in the regulation of these (and perhaps other) growth factor receptors, via internalization, signaling, and/or degradation, should be of considerable interest and importance.
ARD1: Action/Function as TRIM23
Description of TRIM 23 as a member of the TRIM protein family provided new perspective on ARD1 action (Reymond et al. 2001). Activation of transcription with induction of type I interferon production is critical for initiation of innate immune responses. Association of TRIM 23 (ARD1) with HCMV (human cytomegalovirus)-encoded protein UL144, which was known to activate NF-kB, was first recognized in a yeast two-hybrid screen undertaken to identify UL144-interacting proteins in human cells (Poole et al. 2009). Involvement of tumor necrosis factor receptor (TNFR)-associated factor 6 ( TRAF6) and/or TAK1 in HCMV-induced NF-kB activation was known, but the screen revealed no evidence of their direct interaction with the viral UL144 protein. Then, interaction of UL144 with TRIM23 was found and confirmed, showing direct interaction of the UL144 C-terminal Zn-finger with the TRIM23 N-terminal Zn-finger region. To verify those findings, tagged UL144, TRIM23, and TRAF6 or fragments thereof were overexpressed in human fibroblasts. All experimental data supported the conclusion that TRIM23 was required for NF-kB activation by UL144 (but not for its activation by TNF or dsRNA). TRIM23 was necessary for ubiquitination also of TRAF6, but it did not catalyze the modification (using ubiquitin lysine 63, not lysine 48 for both) that resulted evidently from TRAF6 auto-ubiquitination (Poole et al. 2009).
Rapid induction of the type I interferon response by viral infection is critical for an innate immune reaction. Involvement of several E3 ubiquitin ligases and their substrates provides complex feedback controls for the potentially harmful inflammatory reactions that result. To clarify some of these regulatory relationships, Arimoto et al. (2010) looked for effects of overexpression of one of the ligases, RNF-125, on gene expression and found a dramatic increase of >200% in TRIM23 mRNA. Exploration of TRIM 23 function revealed that it increased expression of an NF-kB-driven reporter gene in a NF-kB essential modulator (NEMO)-dependent manner. Direct interaction of the TRIM23 Arf domain with CC1 and LZ domains, which contain the sites of NEMO ubiquitination (via lysine 27), was demonstrated (Arimoto et al. 2010). Information regarding the role of UbcH5 as an E2 enzyme for TRIM23 ubiquitination of NEMO was notable, particularly perhaps the enhanced association of UbcH5 with NEMO after virus infection and the decreased NEMO ubiquitination by TRIM23 in UbcH5-depleted cells. Behavior of the alternatively spliced TRIM23 molecules (Venkateswarlu and Wilson 2011) in these interactions will surely be of interest, as an Arf domain contribution to intracellular trafficking might facilitate innate immune responses.
ARD1 was cloned because of its C-terminal Arf sequence (which lacks the first nine Arf amino acids). Initial investigation of the predicted 64-kDa protein was directed toward understanding how the Arf domain interacted with the non-Arf remainder of the molecule to accelerate hydrolysis of bound GTP using separately synthesized recombinant protein fragments. The region of the ARD1 molecule that serves as a GAP to terminate activation of the Arf moiety, including adjacent residues required for its functional interaction with Arf structure, appears to overlap B-box and C-C sequences of the TRIM23 molecule (Fig. 2). GDI action of 17 amino acids immediately N-terminal to the Arf domain mimics behavior of the N-terminal amphipathic α-helix of an Arf molecule that would precede the sequence designated Arf domain at ARD1 position 403 (Fig. 2). Both ARD1 and TRIM23 sequences assign E3 ubiquitin ligase actions to the same novel RING structure near the N-terminus. Generation of ARD−/− mouse embryo fibroblasts stably transfected with constructs for inducible expression of wild type or specifically mutated ARD1 molecules enabled demonstration that degradation of endogenous EGF receptor (EGFR) required continuity of the ARD1 GDP/GTP cycle. Cell content of TRIM23 was apparently maintained at low levels by auto-ubiquitination, with no TRIM23 contribution to EGFR ubiquitination. For auto-polyubiquitination of ARD1 in vitro with purified proteins, UbcH5a, 5b, 5c, and UbcH6 had been effective as E2 enzymes. UbcH5 was also involved in TRIM23 ubiquitination of NEMO. These and related findings may be a preview of the importance of only recently recognized antiviral actions of TRIM23.
- Arimoto K, Funami K, Saeki Y, Tanaka K, Okawa K, Takeuchi O, Akira S, Murakami Y, Shimotohno K. Polyubiquitin conjugation to NEMO by tripartite motif protein 23 (TRIM23) is critical in antiviral defense. Proc Natl Acad Sci U S A. 2010;107(36).Google Scholar
- Ding M, Vitale N, Tsai S, Adamik R, Moss J, Vaughan M. Characterization of a GTPase-activating protein that stimulates GTP Hydrolysis by both ADP-ribosylation factor (ARF) and ARF-like proteins. J Biol Chem. 1996;271:2400–9.Google Scholar
- Meza-Carmen V, Pacheco-Rodriguez G, Soo Kang G, Kato J, Donati C, Zhang C, Vichi A, Payne DM, El-Chemaly S, Stylianou M, Moss J, Vaughan M. Regulation of growth factor receptor degradation by ADP-ribosylation factor domain protein (ARD)1. Proc Natl Acad Sci U S A. 2011;108:10454–9.PubMedCentralCrossRefPubMedGoogle Scholar
- Mishima K, Tsuchiya M, Nightingale MS, Moss J, Vaughan M. ARD1, a 64-kDa guanine nucleotide-binding protein with a carboxyl-terminal ADP-ribosylation factor domain. J Biol Chem 1993;268:8801–8807.Google Scholar
- Venkateswarlu K, Wilson LM. ADP-ribosylation factor domain protein 1. UCSD Molecule Pages. 2011. doi: 10.1038/m.p.a003292.01.