Emerging Role of GCN1 in Disease and Homeostasis

Abstract

GCN1 is recognized as a factor that is essential for the activation of GCN2, which is a sensor of amino acid starvation. This function is evolutionarily conserved from yeast to higher eukaryotes. However, recent studies have revealed non-canonical functions of GCN1 that are independent of GCN2, such as its participation in cell proliferation, apoptosis, and the immune response, beyond the borders of species. Although it is known that GCN1 and GCN2 interact with ribosomes to accomplish amino acid starvation sensing, recent studies have reported that GCN1 binds to disomes (i.e., ribosomes that collide each other), thereby regulating both the co-translational quality control and stress response. We propose that GCN1 regulates ribosome-mediated signaling by dynamically changing its partners among RWD domain-possessing proteins via unknown mechanisms. We recently demonstrated that GCN1 is essential for cell proliferation and whole-body energy regulation in mice. However, the manner in which ribosome-initiated signaling via GCN1 is related to various physiological functions warrants clarification. GCN1-mediated mechanisms and its interaction with other quality control and stress response signals should be important for proteostasis during aging and neurodegenerative diseases, and may be targeted for drug development.

Keywords: GCN1; GCN2; RWD domain; amino acid starvation; disome; ribosomal stress surveillance; ribosome.

Figure 1

The GCN1–GCN2 pathway mediates the integrated stress response (ISR). In response to various stressors, eIF2α phosphorylation by stress-specific kinases attenuates global translation initiation, which is known as ISR and is conserved in eukaryotes (right part). GCN1 is a scaffold protein that interacts with ribosomes, GCN2, and other RWD-domain-containing proteins, and is essential for GCN2 activation under conditions of amino acid starvation and UV irradiation. Indirect GCN2 activation by these stressors was presented by dashed line. Yeast GCN1 was found in a colliding ribosome (disome) complex under unstressed conditions. The resultant GCN2 and subsequent eIF2α phosphorylation represses cap-dependent translation, but derepresses mRNAs carrying inhibitory upstream open reading frames, such as ATF4 and CHOP, which regulate the transcriptional activation of amino acid metabolism and apoptotic cell death, respectively.

Figure 2

Domain structure of the GCN1- and RWD-domain-containing proteins. GCN1 interacts with ribosomes via its N-terminal 3/4 region three-quarters and with RWD-domain-containing proteins via a C-terminal region named the RWD-binding domain (RWDBD). The GCN1 eEF-like domain interacts with the N-terminal portion of yeast GCN20, which is also conserved in animal ABCF3, with these interactions being essential for AAS-induced GCN2 activation. GCN2 can directly bind ribosomes via its C-terminal domain (CTD) in yeast. Yeast GCN1 and GCN20 form a complex (disome) that includes the RWD-containing protein Gir2 and ribosome-interacting GTPase 2 (Rbg2). RWD domain-containing protein 1 (DFRP2) and developmentally regulated GTP-binding protein 2 (DRG2) form a heterodimer via the DFRP domain and an unknown region of DRG2. IMPACT, which is a protein that is enriched in neuronal cells and is involved in neuritogenesis, and Gir2 can compete with GCN2 for GCN1 because their forced expression inhibits GCN2 activation. Ring finger protein 14 (RNF14), which is an E3 ubiquitin ligase that is involved in the ubiquitination of stalled ribosomes, possesses an RWD domain and two RING domains, RING1 (R1) and RING2 (R2), as well as an in-between RING fingers (IBR) domain.

Figure 3

Structure prediction of GCN1 RWDBD and docking simulation using the RWD domain. (A) The three-dimensional structure of the RWDBD domain of human GCN1 was constructed according to its amino acid sequence (2260–2408) using the AlphaFold2 program. The confidence of the predicted model is colored in blue, cyan, and yellow for very-high (pLDDT > 90), confident (90 > pLDDT > 70), and low (70 > pLDDT > 50) values, respectively. The model did not include very-low confidence values (pLDDT < 50). (B) A docking simulation of human GCN1 RWDBD using the RWD domain of GCN2 was performed using the AlphaFold2 program. (C) Interaction analysis of human GCN1 RWDBD with the RWD domain of GCN2. Hydrogen bonds (3.8 Å), salt bridges (6 Å), and π-cation bonds (6 Å) resulting from the interaction analysis using iCn3D (https://www.ncbi.nlm.nih.gov/Structure/icn3d/full.html accessed on 28 October 2022) are denoted by gray, green, and red dashed lines, respectively. The side chains of Arg-2312 of GCN1 and Asp-37 of GCN2, which were involved in a salt bridge, are depicted using stick models. (D) RWDBD sequence of human GCN1 and binding sites to RWD domains predicted by docking simulations and interaction analysis. The amino acids in the GCN1 RWDBD that were predicted to be involved in binding to the each RWD domain protein are indicated by asterisks. The α-helix structures H1 through H8 in the RWDBD model are presented at the top of the sequence. (E) Multiple alignments of RWD domains. Amino acid sequences of human RWD-containing proteins, GCN2/EIF2AK4 (amino acids 25–137), IMPACT (14–116), RNF14 (11–137), E3 ubiquitin-protein ligase RNF25 (18–128), DFRP2/RWDD1 (10–114), RWD domain-containing protein 2A (RWDD2A) (14–134), RWD domain-containing protein 2B (RWDD2B) (41–165), RWD domain-containing protein 3 (RWDD3) (7–114), RWD domain-containing protein 4 (RWDD4) (9–111), and WDR59 (393–494) were used. The binding sites that were predicted to be involved in binding to GCN1 RWDBD are highlighted in yellow.

Figure 4

Quality control of aberrant mRNA translation. (A) The nonsense-mediated mRNA decay (NMD) pathway recognizes and eliminates mRNAs carrying a termination codon at an aberrant position. NMD proceeds by recruiting the NMD machinery, including up-frame shift proteins, via the exon junction complex, which is located upstream of the exon–exon junction. A premature translation termination codon leads to the dynamic assembly of up-frame shift proteins. (B) Nonstop mRNAs lacking a termination codon because of the incorrect attachment of a poly(A) tail to an open reading frame are eliminated by nonstop decay (NSD). The yeast translation factor complex Dom34–elongation factor 1 alpha-like protein Hbs1 (Pelota–Hbs1 in mammals) promotes the dissociation of the translation elongation complex, mRNA, and peptidyl-tRNA. The RNase L inhibitor 1 (RLI1, ATP-binding cassette sub-family E member 1, ABCE1 in mammals) is a ribosomal recirculation factor that cooperates with Dom34–Hbs1 to dissociate ribosomes. (C) The disome formed at a difficult codon triggers the recruitment of the zinc finger protein 598 (ZNF598) in mammals (E3 ubiquitin-protein ligase Hel2 in yeast) and the subsequent ubiquitination of 40S ribosomal subunit proteins. (D) The readthrough of the stop codon results in the incorporation of amino acids into the nascent polypeptide chain without proper termination. Because most mRNAs have an additional termination codon in the 3′ UTR, the ribosome cannot reach the poly(A) tail and stalls, resulting in the formation of a disome. GCN1 binds to the disome and recruits the CCR4-Not complex 3’-5’-exoribonuclease subunit Ccr4 (CCR4/NOT) complex to degrade the stalled mRNA.

Figure 5

Role of GCN1 in ribosomal stress surveillance. (A) Mammalian GCN1 detects ribosome collision by direct binding, as observed in yeast, and is recruited by ZAKα to the disome fraction. GCN1 can recruit the downstream effectors RNF14 and CCR4/NOT to resolve stalled ribosomes by ribosome-associated quality control (RQC), as well as GCN2, to activate ISR and attenuate the global translation initiation depending on stress severity. (B) The inhibition of translation elongation can trigger both the GCN2–ISR and the ZAKα–ribotoxic stress response (RSR) pathways. GCN2 is activated in response to UVB irradiation and elongation inhibitors, such as tigecycline, which stall ribosomes with a vacant A-site, although intermediate anisomycin can also activate GCN2 in human MCF10A cells but not in yeast cells (dashed line). ZAKα is activated by a broad spectrum of elongation inhibitors, including cycloheximide, and is conserved in animals but absent in yeast.

Figure 6

Altered hepatic metabolism in tamoxifen-induced Gcn1 CKO mice. A combination of tamoxifen administration and Gcn1 CKO in mice resulted in a decrease in liver and white adipose tissue weight, reduced hepatic lipid storage, as well as reduced glucose and very-low-density lipoprotein levels in the blood. The alteration of the liver proteome in CKO mice implicates a decrease in glycogen deposition, leucine degradation, and fatty acid oxidation (indicated in blue), and an inverse increase in the peroxisomal pathway, resulting in inefficient energy production (indicated in red) [127]. Solid arrows indicate the probable increment in the metabolic flux and gray dashed arrows indicate the probable decrease in the metabolic flux, respectively. Triple arrows (→→→) represent the multiple steps of the metabolisms.