A dual role for SAGA-associated factor 29 (SGF29) in ER stress survival by coordination of both histone H3 acetylation and histone H3 lysine-4 trimethylation

Abstract

The SGF29 protein binds to tri-methylated lysine-4 of histone H3 (H3K4me3), which is a histone modification associated with active promoters. Human SGF29 is a subunit of the histone acetyltransferase module of the SAGA (Spt-Ada-Gcn5 acetyltransferase) and ATAC (Ada-Two-A-containing 2A) co-activator complexes. Previous work revealed that the SAGA complex is recruited to endoplasmic reticulum (ER) stress target genes and required for their induction. Here, we report the involvement of SGF29 in the survival of human cells from ER stress. SGF29 knockdown results in impaired transcription of the ER stress genes GRP78 and CHOP. Besides histone H3K14 acetylation, we find that SGF29 is also required for the maintenance of H3K4me3 at these genes, which is already present prior to ER stress. Reduced levels of H3K4me3 in the absence of SGF29 correlate with a decreased association of ASH2L, which is a core component of the SET1/MLL complexes, to GFP78 and CHOP. In conclusion, our results suggest that the H3K4me3-binding protein SGF29 plays a central and dual role in the ER stress response. Prior to ER stress, the protein coordinates H3K4me3 levels, thereby maintaining a 'poised' chromatin state on ER stress target gene promoters. Following ER stress induction, SGF29 is required for increased H3K14 acetylation on these genes, which then results in full transcriptional activation, thereby promoting cell survival.

Figure 1. SPT20 and SGF29 are required for cell survival after ER stress in U2OS cells.

U2OS cells were transduced with lentiviruses targeting SPT20 or SGF29 (two different shRNAs) or a non-target control shRNA. A. Different amounts of tunicamycin were tested for toxicity in U2OS WT cells. Schematic representation of FACS results of propidium (PI) positive cells. B. Recovery of unspliced XPB1 mRNA in U2OS wt cells after tunicamycin treatment (5ug/ul) for indicated times. Amount of XPB1 mRNA is measured by RT-PCR. C. and E. Analysis of mRNA expression levels of SPT20 and SGF29 by quantitative RT-PCR, corrected for β-ACTIN. Standard deviations represent technical triplicates. D. and F. Schematic representation of FACS results, increase in PI positive cells upon SPT20 knockdown (KD) and increase of PI- and annexin V-positive cells upon SGF29 KD. Samples were measured after 8 hours treatment with tunicamycin or DMSO followed by an o/n recovery. Standard deviations represent technical duplicates. Similar results were observed in two independent experiments.

Figure 2. The SGF29 protein is recruited to ER stress target genes and required for acetylation and subsequent transcription induction.

A. Analysis of mRNA expression levels of GRP78 and CHOP by quantitative RT-PCR. Levels were normalized to β-ACTIN and are presented as change compared to a control DMSO-treated sample. Samples were analyzed 4 and 8 h after tunicamycin treatment. B. Immunoblot analysis for doxycycline-inducible GFP-SGF29 and endogenous SGF29. C. Localization of the primer pairs used for ChIP. D. ChIP analysis of GFP-SGF29. On the x-axis are indicated amplicons for the GRP78 and CHOP genes or a non-coding control region. Cells were not treated (T0, black histograms) or treated for 4 h with tunicamycin (t=4, white bars). Standard deviations represent technical triplicates and similar results were observed in at least three independent experiments. E. H3K14ac ChIP (percentage of input relative to H3 ChIP) at the transcription start site of ER stress target genes and a non-coding control region for 0 and 4 hours treatment with tunicamycin. Standard deviations represent technical triplicates and similar results were observed in at least two independent experiments.

Figure 3. SGF29 regulates H3K4me3 levels on ER stress target genes and TAF3 is also involved in the ER stress response.

ChIP of H3K4me3 and U2OS cells transduced with lentiviruses carrying TAF3 or non-target control shRNAs. A. H3K4me3 ChIP (percentage of input relative to H3 ChIP) at the transcription start site of ER stress target genes and a non-coding control region for 0 and 4 hours treatment with tunicamycin. Standard deviations represent technical triplicates and similar results were observed in at least three independent experiments. B. Immunoblot analysis of proteins from SGF29 KD cells for global levels of H3K4me3. C. Analysis of mRNA expression of TAF3 by quantitative RT-PCR, corrected for β-ACTIN and standard deviations represent technical triplicates. D. Analysis of mRNA expression levels of GRP78 and CHOP by quantitative RT-PCR. Levels were normalized to β-ACTIN and are presented as change compared to a control DMSO-treated sample. Samples were analyzed 4 and 8 h after tunicamycin treatment. E. Schematic representation of FACS results, increase in propidium (PI) positive cells in TAF3 kd cells after tunicamycin treatment. Samples were measured after 8 hours treatment with tunicamycin or DMSO and o/n recovery. Standard deviations represent technical duplicates. Similar results were observed in two independent experiments. F. H3K4me3 ChIP (percentage of input relative to H3 ChIP) at the transcription start site of ER stress target genes and a non-coding control region for 0 and 4 hours treatment with tunicamycin. Standard deviations represent technical triplicates and similar results were observed in at least two independent experiments.

Figure 4. ASH2L and RBBP5 associate to GRP78 and CHOP promoters and ASH2L binding is SGF29 dependent.

ChIP of ASH2L and RBBP5 at the transcription start site of the GRP78 and CHOP genes. A. Immunoblot analysis of proteins from SGF29 KD cells for global levels of SET1/MLL subunits ASH2L, RBBP5 and WDR5. B. Immunoblot analysis of protein levels of inducible GFP-ASH2L and GFP-RBBP5 and endogenous RBBP5. C. ChIP analysis of GFP-ASH2L and GFP-RBBP5. Standard deviations represent technical triplicates and similar results were observed in at least three independent experiments. D. Analysis of mRNA expression of SGF29 by qPCR, corrected for β-ACTIN and standard deviations represent technical triplicates. E. Immunoblot analysis of GFP-ASH2L protein levels in GFP-ASH2L cell line infected with non target (lane 1 and 2) and SGF29 hairpin (lane 3 and 4). F. ChIP analysis of GFP-ASH2L in control and SGF29 KD lines. Standard deviations represent technical triplicates and similar results were observed in at least three independent experiments.