General amino acid control in fission yeast is regulated by a nonconserved transcription factor, with functions analogous to Gcn4/Atf4

Abstract

Eukaryotes respond to amino acid starvation by enhancing the translation of mRNAs encoding b-ZIP family transcription factors (GCN4 in Saccharomyces cerevisiae and ATF4 in mammals), which launch transcriptional programs to counter this stress. This pathway involves phosphorylation of the eIF2 translation factor by Gcn2-protein kinases and is regulated by upstream ORFs (uORFs) in the GCN4/ATF4 5' leaders. Here, we present evidence that the transcription factors that mediate this response are not evolutionarily conserved. Although cells of the fission yeast Schizosaccharomyces pombe respond transcriptionally to amino acid starvation, they lack clear Gcn4 and Atf4 orthologs. We used ribosome profiling to identify mediators of this response in S. pombe, looking for transcription factors that behave like GCN4 We discovered a transcription factor (Fil1) translationally induced by amino acid starvation in a 5' leader and Gcn2-dependent manner. Like Gcn4, Fil1 is required for the transcriptional response to amino acid starvation, and Gcn4 and Fil1 regulate similar genes. Despite their similarities in regulation, function, and targets, Fil1 and Gcn4 belong to different transcription factor families (GATA and b-ZIP, respectively). Thus, the same functions are performed by nonorthologous proteins under similar regulation. These results highlight the plasticity of transcriptional networks, which maintain conserved principles with nonconserved regulators.

Keywords: ATF4; GCN4; fission yeast; ribosome profiling; translational control.

Fig. 1.

Transcriptomic response to amino acid starvation. (A) Scatter plot comparing mRNA levels of wild-type cells before and after 3-AT treatment (RNA-seq). All cells were grown in EMM2 without amino acids, and 3-AT was added as indicated. All data have been normalized to reads per kilobase per million mapped reads (RPKMs). The dashed lines correspond to twofold differences. The results of a single experiment are shown. Genes in green have been selected as significantly up-regulated by 3-AT over multiple independent biological replicates (see Methods for details). (B) As in A, comparison between wild-type cells and gcn2Δ mutants in the absence of 3-AT. (C) As in A, comparison of gcn2Δ mutants before and after 3-AT treatment. (D) As in A, comparison of wild-type and gcn2Δ cells after exposure to 3-AT.

Fig. 2.

Translational responses to amino acid starvation. Scatter plots comparing log₂ changes in mRNA levels and translation efficiencies between 3-AT-treated and untreated cells are shown. All cells were grown in EMM2 without amino acids, and 3-AT was added as indicated. The fil1 gene is plotted in black and highlighted by the arrows. (A) Wild-type cells. Genes whose TE is reproducibly induced upon 3-AT treatment in wild-type cells are shown in red (at least 1.5-fold induction in seven of seven experiments, including both CHX-treated and untreated cells). Cells from this experiment were pretreated with CHX. (B) As in A, data for gcn2Δ mutants. (C) Wild-type cells incubated with CHX. Genes encoding ribosomal proteins are displayed in yellow. (D) As in C, wild-type cells not incubated with CHX before collection.

Fig. 3.

Characterization of fil1Δ mutants. (A) Relative growth rates of wild-type cells (WT), fil1Δ, and fil1Δ expressing the fil1 gene from the leu1 locus. Cells were grown in rich medium (Rich), minimal medium with the addition of 20 amino acids (EMM+aa), and standard minimal medium (EMM). Data are normalized to the rich medium samples. Each dot corresponds to an independent biological replicate (n = 3 or 6 as shown), and horizontal lines indicate the mean. Significance was determined by using two-sample two-sided Student’s t tests. **P < 0.01. Only significant comparisons are shown. (B) Venn diagram showing the overlap between genes induced by 3-AT treatment in wild-type cells and those expressed at low levels in fil1Δ mutants (no 3-AT). All cells were grown in EMM2 without amino acids, and 3-AT was added as indicated. The P value of the observed overlap is shown. (C) Scatter plot comparing mRNA levels of fil1Δ cells before and after 3-AT treatment. All data have been normalized to RPKMs. The dashed lines correspond to twofold differences. Genes in green are significantly up-regulated by 3-AT in wild-type cells (Methods). (D) As in C, but comparison of wild-type and fil1Δ cells in the absence of 3-AT treatment. (E) As in C, cells expressing the fil1 coding sequence from the nmt1 promoter.

Fig. 4.

ChIP-seq analysis of Fil1 binding. (A) Enrichment of Fil1-TAP in the asn1 locus. Cells were grown in EMM2 without amino acids, and 3-AT was added as indicated. The arrow corresponds to the asn1 gene and the box to the coding sequence. Enrichment is shown for two independent biological replicates in the absence of 3-AT exposure (red and black lines). The x axis shows the chromosomal coordinates (chromosome 2). (B) Changes in mRNA levels of the asn1 gene upon 3-AT treatment of wild-type cells (Left), in fil1 mutants compared with wild-type cells without 3-AT exposure (Center), or in fil1 mutants compared with wild-type cells after 3-AT-treatment (Right). Data are from RNA-seq experiments. Each point corresponds to an independent biological replicate, and the horizontal lines show the mean (n = 3). (C) Venn diagram showing the overlap between genes bound by Fil1 (without 3-AT treatment) and those expressed at low levels in fil1Δ mutants. The P value was calculated as described in Methods. (D) As in C, showing the overlap between genes bound by Fil1 (no 3-AT) and those expressed at increased levels in fil1Δ mutants. (E) As in C, displaying the overlap between genes bound by Fil1 (upon 3-AT exposure) and those induced by 3-AT treatment of wild-type cells. (F) As in C, comparing genes bound by Fil1 (with 3-AT) and those repressed by 3-AT treatment of wild-type cells. (G) As in C, comparing genes bound by Fil1 in cells untreated or treated with 3-AT. (H) As in C, comparing genes bound by S. pombe Fil1 and S. cerevisiae Gcn4.

Fig. 5.

Translational control of the fil1 mRNA. (A, Top) Structure of the fil1 transcript and distribution of RPFs. Lines represent the location of the UTRs and the box that of the coding sequence (CDS). The position of six uORFs (five AUG and one CUG) are indicated. Cells were grown in EMM2 without amino acids and incubated with 3-AT for 1 h. A, Middle and Bottom display the density of RPFs along the transcript for untreated (A, Middle) and 3-AT–treated (A, Bottom) cells. (B) Quantification of RPF read density for the fil1 coding sequence (CDS) and 5′-leader sequences. Data are presented for control and 3-AT–treated cells. Each dot corresponds to an independent biological replicate (n = 4), and the horizontal lines indicate the mean. (C, Upper) Western blots to measure Fil1–TAP protein levels. Cells were incubated in EMM2 without amino acids and containing 3-AT for the indicated times. One sample (His) was incubated with both histidine and 3-AT for 180 min (histidine is expected to prevent the effects of 3-AT). The experiment was performed with wild-type cells (right) and gcn2Δ mutants (left). Histone H3 was used as a loading control. C, Lower shows quantification of three independent biological replicates of the experiment above, with data normalized to the levels of untreated cells of the corresponding genotypes. (D) Expression of a mCherry fluorescent reporter in wild-type cells containing the 5′ leader of adh1, or fil1 with the six mutated uORFs (fil1-6x). Data are presented for fluorescence (protein) or RNA levels and normalized to the levels of the wild-type fil1 reporter. Each dot corresponds to an independent biological replicate (n = 3), and horizontal lines indicate the mean. (E) Expression of an mCherry fluorescent reporter in wild-type cells containing the 5′-leader sequences of adh1, wild-type fil1 (fil1-wt), or fil1 with the six mutated uORFs (fil1-6x), or in wild type with fil1 5′-leader incubated in the presence of histidine (His), or with the fil1 5′-leader in gcn2Δ cells. Cell treatment is as in C, except that the cells were incubated with 3-AT for 5 h. All data are normalized to the levels of the corresponding reporter in untreated cells. Each dot corresponds to an independent biological replicate (n = 3), and horizontal lines indicate the mean. Significance was determined by using one-sample one-sided Student’s t tests. Only significant comparisons are shown. **P < 0.01. The data are shown for fluorescence levels estimated by flow cytometry (E, Left) and for mRNA abundance quantified by qPCR (E, Right).