Gcn4p and novel upstream activating sequences regulate targets of the unfolded protein response

Abstract

Eukaryotic cells respond to accumulation of unfolded proteins in the endoplasmic reticulum (ER) by activating the unfolded protein response (UPR), a signal transduction pathway that communicates between the ER and the nucleus. In yeast, a large set of UPR target genes has been experimentally determined, but the previously characterized unfolded protein response element (UPRE), an upstream activating sequence (UAS) found in the promoter of the UPR target gene KAR2, cannot account for the transcriptional regulation of most genes in this set. To address this puzzle, we analyzed the promoters of UPR target genes computationally, identifying as candidate UASs short sequences that are statistically overrepresented. We tested the most promising of these candidate UASs for biological activity, and identified two novel UPREs, which are necessary and sufficient for UPR activation of promoters. A genetic screen for activators of the novel motifs revealed that the transcription factor Gcn4p plays an essential and previously unrecognized role in the UPR: Gcn4p and its activator Gcn2p are required for induction of a majority of UPR target genes during ER stress. Both Hac1p and Gcn4p bind target gene promoters to stimulate transcriptional induction. Regulation of Gcn4p levels in response to changing physiological conditions may function as an additional means to modulate the UPR. The discovery of a role for Gcn4p in the yeast UPR reveals an additional level of complexity and demonstrates a surprising conservation of the signaling circuit between yeast and metazoan cells.

Figure 1. Computational Selection of Candidate Regulatory Motifs

(A) Candidate regulatory motifs are overrepresented in UPR target promoters. Sequence motifs were ranked in order of overrepresentation, i.e., on the number of observed appearances in target promoters relative to the expectation from the total appearances in all promoters. −log₁₀ P, a metric of overrepresentation, is plotted against rank (circles). Eight motifs were chosen for experimental characterization (open circles). (B) Best words grouped into eight candidate motifs. The eight most overrepresented motifs from Fig. 1A, aligned to illustrate common core sequences. The example of each motif chosen for experimental characterization is underlined.

Figure 2. Identification of Two Novel Sequence Motifs Necessary and Sufficient for UPR Activation

(A) Motif 1 and Motif 8 are sufficient to confer UPR-responsive transcription on an artificial promoter. Single representative sequences of the KAR2-derived UPRE and candidate regulatory motifs Motif 1 and Motif 8 were cloned into a crippled promoter driving lacZ, transformed into yeast (WT, Δire1, and Δhac1), and β-galactosidase activity monitored in response to Tm treatment. (B) UPRE-2 (Motif 1) is necessary for UPR-dependent activation of the ERO1 promoter. lacZ was placed under the control of the WT ERO1 promoter (+ UPRE-2) or a mutant (− UPRE-2), and β-galactosidase activity monitored in response to DTT treatment. (C) UPRE-3 (Motif 8) is necessary for UPR-dependent activation of the DHH1 promoter. As in (B), except using the DHH1 promoter, in which UPRE-3 appears once. (D) Novel motifs explain a greater fraction of UPR target gene activation. Sets of genes whose promoters contain UPR-responsive UASs UPRE-1, UPRE-2, UPRE-3, or a combination, are here depicted in Venn diagram format as subsets of the 381-gene UPR target set.

Figure 3. GCN4 Encodes a Novel Transcription Factor in the UPR

(A) Overexpression of GCN4 is sufficient for activation of UPRE-2, but not UPRE-1 or UPRE-3. UPRE-driven transcriptional activity as a function of Gcn4p levels, elevated either as a result of overexpression (+ GCN4–2μ) or amino acid starvation (+ 3-AT), in the presence or absence of ER stress (Tm). (B) GCN4 and GCN2 are necessary for ER stress-dependent activation of UPRE-1 and UPRE-2. UPRE-driven transcriptional activity as a function of GCN4 pathway genes (WT, Δgcn4, and Δgcn2) in the presence or absence of ER stress (Tm). (C) GCN4 and GCN2 are required for UPR-dependent transcriptional activation of a subset of target genes. Fold changes in mRNA levels were determined by microarray for DTT-treated vs. -untreated WT, Δire1, Δgcn4, and Δgcn2 strains (columns). Histograms show distribution of log₂-fold changes for non-UPR target genes (light bars) and for UPR target genes (dark bars), which contain UPRE-1, UPRE-2, UPRE-3, or still unidentified UPREs (rows) in their promoters. (D) Target gene regulation differs significantly in WT and Δgcn4/Δgcn2 mutants. Means (μ) and standard deviations (σ) for log₂-fold change in gene expression for non-UPR target genes, and for genes that fall inside the UPR target gene set and contain UPRE-1, UPRE-2, or UPRE-3 in their promoters. Z statistic (z) and P value (P): higher z reflects a greater difference between the distribution for UPRE-containing target genes and nontarget genes; lower P indicates a more highly significant difference. For detailed calculations, see Materials and Methods.

Figure 4. Gcn4p Protein Levels Are Upregulated during the UPR

(A) Gcn4p levels, but not eIF-2α phosphorylation, rise under ER stress in a UPR-dependent manner. Cells bearing a C-terminally myc-tagged allele of GCN4 were treated with Tm for 0, 15, 30, 60, or 120 min. Western blots probed with anti-myc recognizing Gcn4p-myc (top panels) or phospho-specific anti-eIF-2α antibody (bottom panel). Gcn4p blot for the Δgcn2 mutant is 5x overexposed so that the bands are visible. (B) Quantitation of the Gcn4p-myc protein levels in Figure 4A. Data reflect an average of four experiments, normalized against Gcn4p levels in the WT t = 0 samples.

Figure 5. GCN4 Acts with or Downstream of HAC1

UPRE reporter activity as a function of Hac1p expression and UPR pathway genes. To express Hac1p in the absence of ER stress, we used an intron-less allele of HAC1, which is constitutively translated.

Figure 6. Hac1p and Gcn4p Directly Interact with UPRE-1 and UPRE-2

³²P-labeled oligos bearing either UPRE-1 or UPRE-2 promoter were incubated with crude cell extracts, and subjected to nondenaturing polyacrylamide gel electrophoresis. (A) Extract: Samples were of the WT, or bore deletions in IRE1Δire1), GCN4 (Δgcn4), or GCN2 (Δgcn2), and were treated with Tm (+) or mock treated (−). Labeled oligos contained either UPRE-1 (1) or UPRE-2 (2). Binding reactions were incubated with no unlabeled competitor (−) or with 100x excess of unlabeled WT UPRE-1 (1), an inactive mutant version of UPRE-1 (1*), UPRE-2 (2), or an inactive mutant version of UPRE-2 (2*). (B) Extract: Samples from a strain overexpressing GCN4 (2μ-GCN4; lanes 1 and 2) or from a strain expressing myc-tagged Gcn4p and HA-tagged Hac1p (GCN4-myc and HA-HAC1). Binding reactions were incubated with no antibody (−), anti-myc recognizing Gcn4p-myc (myc), anti-HA recognizing HA-Hac1p (HA), or both antibodies simultaneously (myc/HA). Bands represent the following: a, Gcn4p + Hac1p + anti-myc + anti-HA; b, Gcn4p + Hac1p + anti-HA; c, Gcn4p + Hac1p + anti-myc; d, Gcn4p + Hac1p; e, Gcn4p. *, an unidentified band that appears only when extracts include both Gcn4-myc and HA-Hac1p and when both antibodies are included in the binding reaction.

Figure 7. Multiple Alignment of UPRE-1 and UPRE-2 from Three Budding Yeasts

Alignment of partial promoter sequences from S. cerevisiae and homologous sequences in related yeasts. Numerical coordinates reflect the distance from the first nucleotide of the initiation codon in the S. cerevisiae promoter. (A) A segment of the KAR2/YJL034W promoter and homologs. The core sequence of UPRE-1 is indicated. (B) A segment of the ERO1/YML130C promoter and homologs. The core sequence of UPRE-2 is indicated (above). The consensus binding site of Gcn4p is aligned (below).

Figure 8. Model of Gcn4p/Hac1p Action in the UPR

(A) The expanded circuitry of the UPR. The classical UPR (red), the S-UPR (blue), and regulated Gcn4p levels (green) are integrated at target promoters. Transcriptional regulation of HAC1 mRNA levels, providing one level of gain control, is depicted as a rheostat under supervision of a logical AND gate informed by multiple inputs from the ER. Splicing of HAC1 mRNA by Ire1p, providing a binary on/off control, is depicted by a switch. Regulation of Gcn4p levels by Gcn2p under changing cellular conditions adds an additional layer of gain control. Together, activity levels of Hac1p, Gcn4p, and the proposed UPR modulatory factor (Leber et al. 2004) collaborate to determine the magnitude of the transcriptional output signal. (B) Mechanism of Gcn4p/Hac1p action at target promoters. In the absence of Hac1p, Gcn4p is present in the cell as a consequence of baseline activity of Gcn2p. At normal concentrations, Gcn4p is unable to bind or activate a target UPRE, but it may bind when Gcn4p levels are elevated. Upon induction of the UPR, Ire1p is activated and Hac1 is synthesized. Hac1p can bind, but not activate, target UPREs. Binding of target DNA by a Gcn4p/Hac1p heterodimer results in a transcriptionally active complex. Gcn4p levels are upregulated under UPR induction, perhaps as a consequence of stabilization by interaction with Hac1p.