Transcriptional and Translational Regulatory Mechanisms in Nutrient Control of Gene Expression

It has been reported that mammalian eIF2A promotes Met-tRNAi recruitment to the AUG codons of viral mRNAs and enhances initiation at near-cognate codons (NCCs) when TC assembly is impaired by eIF2α phosphorylation. We conducted ribosome profiling on a yeast eIF2AΔ mutant in the presence or absence of eIF2α phosphorylation induced by amino acid starvation, reasoning that mRNAs able to utilize eIF2A might show TE reductions only when eIF2 is impaired. Surprisingly, eIF2AΔ conferred no significant TE reductions for any mRNAs in non-starved cells, and affected only about 30 mRNAs in amino acid–starved cells. We were able to validate this conditional eIF2A dependence by luciferase (LUC) reporter analysis and polysome profiling of native mRNAs for only several genes. We found no evidence that eIF2AΔ alters the translation of mRNAs containing evolutionarily conserved, inhibitory uORFs. Our findings do not support an important role for yeast eIF2A in a back-up mechanism for Met-tRNAi recruitment when eIF2 function is reduced by stress.

A highly conserved response to amino acid starvation involves activation of the protein kinase Gcn2, which phosphorylates eukaryotic initiation factor 2, with attendant inhibition of global protein synthesis and increased translation of the yeast transcriptional activator GCN4. Gcn2 contains a domain related to histidyl-tRNA synthetase (HisRS–like domain), and a C-terminal ribosome-binding domain. Previous work indicated that Gcn2 is activated on translating ribosomes by uncharged tRNAs that accumulate in amino acid–starved cells and pair with the cognate codons in the empty ribosomal A, interacting directly with the HisRS–like domain to stimulate kinase activity. Gcn2 can also be activated by conditions that stall elongating ribosomes without reducing aminoacylation of tRNA, but it was unclear whether distinct molecular mechanisms operate in these two circumstances. We identified three regimes that activate Gcn2 in yeast by starvation-independent (SI) ribosome stalling, which leaves an empty A site on the stalled ribosome: (1) treatment with inhibitor tigecycline, which should stall ribosomes at all codons; (2) deleting the gene encoding tRNAArgUCC, which should stall ribosomes at AGG codons; and (3) depletion of translation-termination factor eRF1, which should stall ribosomes at stop codons. Subsequent genetic analysis demonstrated requirements for the HisRS–like and ribosome-binding domains of Gcn2, positive effectors Gcn1/Gcn20, and the tethering of at least one of two P1/P2 heterodimers of the 60S ribosomal P-stalk complex, for detectable activation by SI–ribosome stalling. Remarkably, no tethered P1/P2 proteins were required for strong Gcn2 activation by starvation for various amino acids, indicating that Gcn2 activation has different requirements for the P-stalk, depending on how ribosomes are stalled. We propose that accumulation of deacylated tRNAs in starved cells functionally substitutes for the P-stalk in binding to the HisRS–like domain for eIF2 kinase activation by ribosomes stalled with A sites.

Figure 1. Structural mimicry of the histidyl-tRNA charging enzyme (HisRS) by the HisRS–related sensor of amino acid starvation in Gcn2, the ancestral protein kinase of the Integrated Stress Response

A. Overall crystal structure of the C. thermophilum Gcn2 HisRS–related regulatory domain, containing structural mimics of both the catalytic domain (CD), where the 3′ acceptor end of tRNA is aminoacylated, and the anticodon binding domain (ABD), which binds the anticodon loop of tRNA to ensure its identify for aminoacylation.

B. Co-crystal structure of T. thermophilus authentic HisRS enzyme bound to histidyl tRNA (PDB:4RDX). We showed that substituting conserved ABD residues in the Gcn2 HisRS–related domain, well positioned to contact the tRNA anticodon loop, impairs both Gcn2 function in starved cells and eIF2 phosphorylation by purified Gcn2 in vitro. Mimicry in Gcn2 of two highly conserved structural domains for binding both ends of tRNA, each crucial for Gcn2 function, supports the model that deacylated tRNAs activate Gcn2 in amino acid–starved cells and exemplifies how a metabolic enzyme is repurposed to host new local structures and sequences that confer a novel regulatory function.

Click image to enlarge.

Click image to enlarge.

Figure 1. Structural mimicry of the histidyl-tRNA charging enzyme (HisRS) by the HisRS–related sensor of amino acid starvation in Gcn2, the ancestral protein kinase of the Integrated Stress Response

A. Overall crystal structure of the C. thermophilum Gcn2 HisRS–related regulatory domain, containing structural mimics of both the catalytic domain (CD), where the 3′ acceptor end of tRNA is aminoacylated, and the anticodon binding domain (ABD), which binds the anticodon loop of tRNA to ensure its identify for aminoacylation.

B. Co-crystal structure of T. thermophilus authentic HisRS enzyme bound to histidyl tRNA (PDB:4RDX). We showed that substituting conserved ABD residues in the Gcn2 HisRS–related domain, well positioned to contact the tRNA anticodon loop, impairs both Gcn2 function in starved cells and eIF2 phosphorylation by purified Gcn2 in vitro. Mimicry in Gcn2 of two highly conserved structural domains for binding both ends of tRNA, each crucial for Gcn2 function, supports the model that deacylated tRNAs activate Gcn2 in amino acid–starved cells and exemplifies how a metabolic enzyme is repurposed to host new local structures and sequences that confer a novel regulatory function.

We collaborated with Jinwei Zhang to determine the crystal structure of the HisRS–like domain of Gcn2 from the fungus Chaetomium thermophilum. The structure is dramatically similar to authentic HisRS, including an α-helical insertion domain harboring the catalytic center and a separate anticodon binding domain (ABD) that, in authentic HisRS, bind to the acceptor stem and anticodon loop of tRNAHis, respectively. The structural elements for forming histidyl adenylate and aminoacylation of tRNAHis are lacking, consistent with repurposing of the HisRS–like domain as a sensor of amino acid starvation. This model is supported by the location of numerous substitutions in the conserved catalytic core of the HisRS–like domain that hyperactivate or impair Gcn2 function. We further showed that residues in the ABD predicted by the structure to contact the tRNA anticodon loop or connect the ABD to the catalytic core are also crucial for Gcn2 activation in vivo and high-level kinase activity by purified Gcn2 in vitro. The presence of two conserved structural domains for binding different ends of tRNA in the HisRS–like domain, both essential for kinase activation, supports the model that uncharged tRNA directly activates Gcn2 on stalled ribosomes in amino acid–starved cells.

Evidence suggests that the mammalian heterodimer MCT-1/DENR promotes reinitiation (REI) following translation of short uORFs whose penultimate codons depend on the heterodimer for releasing the cognate deacylated tRNA from post-termination 40S complexes (postTCs), including uORF1 of ATF4 mRNA, which mediates the same delayed REI mechanism governing GCN4 translation. Substituting GCN4 uORF1 with penultimate codons that vary in dependence on the corresponding translation machinery–associated protein heterodimer Tma20/Tma22 for 40S recycling, we found little evidence that Tma20/Tma22 enhances REI at GCN4 uORF1 variants equipped with Tma–dependent penultimate codons. Making similar substitutions at uORF4, which differs from uORF1 in being permissive for REI, we implicated Tma20/Tma22 in blocking, not stimulating REI, by releasing empty postTCs, in a manner only marginally more pronounced with Tma20/Tma22–dependent penultimate codons. Thus, dissociation of tRNA from postTCs can proceed without Tma20/Tma22 at both uORF1 and uORF4, even at Tma–dependent penultimate codons, and Tma20/Tma22 enhances dissociation of postTCs at uORF4 variants with either class of penultimate codons. The latter function is impeded at uORF1 by its specialized REI–promoting elements. We conclude that Tma20/Tma22 and MCT-1/DENR play distinct roles in the delayed REI mechanism.

Previously, we showed that elimination of Gcn5, the histone acetyltransferase (HAT) subunit in co-factor SAGA (Spt-Ada-Gcn5 acetyltransferase), did not fully impair nucleosome eviction at many starvation-induced genes, suggesting that Gcn5 might cooperate with other HATs in this process, similar to the functional cooperation we had identified for different CRs (chromatin remodelers). The role of the HAT complex NuA4, responsible for most H4 and H2A acetylation in yeast, was of particular interest, as it was shown to be recruited to the Gcn4 target genes ARG1 and ARG4. We examined the effects of disrupting the NuA4 complex, by eliminating its nonessential scaffold subunit Eaf1, on promoter nucleosome eviction and transcriptional activation at both starvation-induced and constitutively expressed genes. We also examined whether depleting Eaf1 from the nucleus (by anchor-away technology) confers defects in nucleosome eviction or transcription in cells lacking Gcn5 in order to evaluate whether NuA4 and Gcn5 make independent, additive contributions to these processes at particular genes in vivo. Our results revealed that NuA4 acts on par with Gcn5, and functions additively, in evicting and repositioning promoter nucleosomes, and in stimulating transcription, at starvation-induced genes. However, NuA4 is generally more important than Gcn5 in promoter nucleosome eviction, recruitment of the TATA-binding protein (TBP), and transcription at most other genes expressed constitutively in yeast. NuA4 also predominates over Gcn5 in stimulating TBP recruitment and transcription of genes categorized as principally dependent on the cofactor TFIID versus SAGA, except for the highly expressed subset encoding ribosomal proteins (RPs), where Gcn5 contributes strongly to PIC assembly and transcription. We found that both SAGA and NuA4 are recruited to promoter regions of starvation-induced genes in a manner that appears to be controlled by their HAT activities, and thus most likely act directly to promote transcription of these genes. Our findings reveal an intricate interplay between these two HATs in nucleosome eviction, PIC assembly, and transcription that differs between the starvation-induced and basal transcriptomes.