The HAC1 Histone Acetyltransferase Promotes Leaf Senescence via Regulation of ERF022

Nutrient remobilization during leaf senescence nourishes the growing plant. Understanding the regulation of this process is essential for reducing our dependence on nitrogen fertilizers and increasing agricultural sustainability. Our lab is interested in chromatin changes that accompany the transition to leaf senescence. Previously, darker green leaves were reported for Arabidopsis thaliana hac1 mutants, defective in a gene encoding a histone acetyltransferase in the CREB-binding protein family. Here, we show that two Arabidopsis hac1 alleles display delayed age-related developmental senescence, but have normal dark-induced senescence. Using a combination of ChIP-seq for H3K9ac and RNA-seq for gene expression, we identified 44 potential HAC1 targets during age-related developmental senescence. Genetic analysis demonstrated that one of these potential targets, ERF022, is a positive regulator of leaf senescence. ERF022 is regulated additively by HAC1 and MED25, suggesting MED25 recruits HAC1 to the ERF022 promoter to increase its expression in older leaves.


Introduction 15
Plants continuously produce new organs. During vegetative growth, new leaves form from the shoot 16 apical meristem, and develop into protein-rich photosynthetic factories that export sugars. Eventually, 17 the older leaves enter senescence by catabolizing the photosynthetic apparatus and exporting nitrogen-18 rich amino acids to support continuing growth (Himelblau and Amasino, 2001). Understanding the 19 regulation of leaf senescence could maximize nitrogen recycling thus producing more nutrient-rich 20 seeds and reducing the need for fertilizers. 21 The transition into leaf senescence is preceded (Kim et al., 2018a) and accompanied by changes in gene 22 expression (Buchanan-Wollaston et al., 2005;van der Graaff et al., 2006;Breeze et al., 2011). Lists of 23 senescence-associated genes (SAG) have been generated from these transcriptome analyses. Enriched 24 biological processes from Gene Ontology (GO) analyses include response to the hormones salicylic acid 25 (SA), jasmonic acid (JA), abscisic acid (ABA) and ethylene. Also, enrichment of GO terms autophagy, 26 immune response, defense response, and response to reactive oxygen species demonstrates a 27 molecular relationship between defense and leaf senescence. Additional GO terms highly represented in 28 SAGs from age-related developmental senescence include response to chitin and glucosinolate 29 biosynthesis (Brusslan et al., 2015). The consistent enrichment of the phosphorylation term among SAG 30 lists is likely a result of high expression of receptor-like kinase gene-family members, which also are 31 known to regulate defense (Antolín-Llovera et al., 2014). 32 Changes in chromatin structure are hypothesized to promote and/or maintain leaf senescence 33 (Humbeck, 2013). We have previously shown a correlation between histone 3, lysine 4, trimethylation 34 (H3K4me3) and histone 3, lysine 9 acetylation (H3K9ac) histone modifications and increased expression 35 of senescence up-regulated genes (SURGs). A similar correlation was seen between histone 3, lysine 27 36 trimethylation (H3K27me3) marks and decreased expression of senescence down-regulated genes 37 (SDRGs) (Brusslan et al., 2012;Brusslan et al., 2015). Genetic analysis suggests histone deacetylases 38 regulate leaf senescence. HDA19 is a negative regulator of senescence (Tian and Chen, 2001) while 39 HDA6 is a positive regulator of leaf senescence (Wu et al., 2008). HDA9 works with POWERDRESS to 40 reduce the expression of four putative negative regulators of leaf senescence (NPX1,TMAC2,WRKY57 41 and APG9), thus promoting leaf senescence (Chen et al., 2016 HAC1 also plays a role in the response to jasmonoyl-isoleucine (JA-ile), the active form of JA. HAC1 67 acetylates histones associated with MYC2 target genes to promote their expression. The Mediator 68 Complex subunit, MED25 interacts with MYC2 and directly binds to and recruits HAC1 to target genes 69 (An et al., 2017). Transcriptome data showed that genes induced by JA-ile were less responsive in a hac1 70 mutant. In addition, genes co-regulated by JA-ile and HAC1 were enriched for many defense-related 71 biological process GO terms as well as leaf senescence. 72 Here we show that hac1 mutants have delayed age-related developmental leaf senescence. Potential 73 HAC1 targets are identified by RNA-seq and ChIP-seq utilizing WT and two hac1 alleles. T-DNA insertion 74 mutants in three potential HAC1 targets were tested for leaf senescence phenotypes, and an erf022 75 mutant disrupting the expression of ERF022 showed delayed senescence. These findings implicate this 76 AP2/ERF transcription factor as a novel positive effector of leaf senescence regulated by histone 77 acetylation co-mediated by HAC1 and MED25. 78 79 Total Protein: One leaf hole-punch was ground in liquid nitrogen in a 1.5 ml microfuge tube using a blue 103 plastic pestle. 100 µL 0.1 M NaOH was added and the sample was ground for another 30 sec (Jones et  104 al., 1989). Samples were incubated at room temperature for 30 min, centrifuged at 14000 rpm for 5 min. 105

Materials and Methods
The Bradford protein assay (Bio-Rad Protein Assay Dye Reagent) was used to determine protein 106 concentration in each supernatant using a bovine serum albumin standard. For each 107 genotype/condition, n = 6. 108 Percent Nitrogen: Elemental analysis for % nitrogen was done by Midwest Microlab, Indianapolis, IN. 109 100 dried seeds from one individual plant were in each sample (n = 8 for each genotype). 110 Gene Expression: Total RNA was isolated from the Indicated leaves using Trizol reagent. 1000 ng of 111 extracted RNA was used as a template for cDNA synthesis using MMLV-reverse transcriptase (New 112 England Biolabs) and random hexamers to prime cDNA synthesis. The cDNA was diluted 16-fold and 113 used as a template for real-time qPCR using either ABsolute QPCR Mix, SYBR Green, ROX (Thermo 114 Scientific) or qPCRBIO SyGreen Blue Mix Hi-Rox (PCR Biosystems), in Step One Plus or Quant Studio 6 115 Flex qPCR machines. All real-time qPCR reactions used a 61 o C annealing temperature. 116 For chlorophyll, total protein, percent nitrogen and gene expression, significant differences were 117 determined using a t-test. 118 RNA-seq: Indicated leaves were harvested and stored in liquid nitrogen. RNA was extracted and RNA-seq 119 library production was performed using the breath adapter directional sequencing (BrAD-seq) method 120 (Townsley et al., 2015). Real-time qPCR using ACT2 primers was the initial quality test. Libraries were 121 sequenced at the Genome High-Throughput Facility (GHTF) at University of California, Irvine (UCI). 122 ChIP-seq; Nuclei preparation and ChIP was performed as described previously (Brusslan et al., 2012). 123 Libraries were produced and sequenced at the GHTF at UCI. 124 Bioinformatics: RNA-seq raw data reads were aligned to the Arabidopsis TAIR 10 genome using 125 Rsubread (Liao et al., 2013), and subject to quality control of count data and differential expression 126 using NOISeq (Tarazona et al., 2015). The values were FPKM normalized using Tmisc and HTSFilter 127 removed genes with low expression levels ( Two Arabidopsis hac1 alleles [hac1-1 (SALK_080380) and hac1-2 (SALK_136314), Supplemental Figure 1] 136 displayed darker green leaves when compared to WT. Age-related chlorophyll loss is shown in Figure 1A. 137 At 28 days, total chlorophyll levels in leaf 7 were equal, but as the leaves aged, chlorophyll levels 138 decreased faster in WT than the two hac1 alleles. A significant difference in chlorophyll levels was 139 detected between WT and both hac1 alleles at day 48. The retention of chlorophyll was accompanied by 140 reduced mRNA levels for genes associated with leaf senescence ( Figure 1B). AtNAP encodes a positive 141 regulator of leaf senescence associated with ABA synthesis (Liang et al., 2014;Yang et al., 2014). NIT2 142 encodes a nitrilase that is highly expressed in leaf senescence, and contributes to auxin synthesis 143 (Normanly et al., 2007) and glucosinolate catabolism (Vorwerk et al., 2001). NYC1 encodes a chlorophyll 144 b reductase required for light harvesting complex disassembly (Kusaba et al., 2007). The chlorophyll and 145 gene expression data show that hac1 alleles display delayed leaf senescence. 146 The reduction of total chlorophyll was also evaluated in detached leaves floated in water in the dark 147 (dark-induced senescence), and no difference was noted between WT and the two hac1 alleles ( Figure  148 1C). There are molecular differences in the signaling pathways between dark-induced and 149 developmental senescence; most prominent is the role of SA in developmental, but not in dark-induced 150 senescence (Buchanan-Wollaston et al., 2005;van der Graaff et al., 2006;Guo and Gan, 2012). Thus, it is 151 possible that alterations in the signaling of developmental senescence do not necessarily accompany 152 changes in dark-induced senescence. These results support a role for HAC1 as a promoter of age-153 related, developmental leaf senescence. 154 A trending increase in total leaf protein concentration accompanied the significant increase in 155 chlorophyll levels in both hac1 alleles (Figures 2A-B). However, the delayed senescence in the hac1 156 alleles did not result in greater percentage of seed nitrogen ( Figure 2C). Delayed senescence in wheat 157 was reported to increase grain nitrogen concentration (Zhao et al., 2015), however the relationship 158 between percentage of seed nitrogen and leaf senescence is complex (Chardon et al., 2014;Havé et al., 159 2017). 160

hac1 Mutants Display Altered Levels of Histone Modifications and Changes in Gene Expression During 162
Leaf Senescence 163 ChIP-seq was performed on the same tissue shown in Figure 2 to identify genes associated with a loss of 164 H3K9ac and/or H3K4me3 histone modifications in both hac1 alleles. HAC1 catalyzes H3K9 acetylation, 165 and both H3K9ac and H3K4me3 are associated with active gene expression (Berr et al., 2011). As 166 expected, H3K9ac significantly decreased at 968 loci and increased at only 555 loci in both hac1 alleles. 167 H3K4me3 modifications were similarly affected, with 548 loci showing a loss and only 33 loci showing a 168 gain of H3K4me3 marks. RNA-seq was used to identify differentially expressed genes (DEGs) between 169 WT and both hac1 alleles. Accordingly, the number of up-regulated DEGs (12) was much smaller than 170 the number of down-regulated DEGs (143) in both hac1 alleles. These 143 down-regulated DEGs were 171 subject to pathway enrichment analysis, and significant enrichment of glucosinolate biosynthesis, plant-172 pathogen interaction, as well as glutathione and ascorbic acid metabolism were revealed. These 173 pathways are stress-related and their down-regulation in hac1 likely slows the rate of leaf senescence. 174 One GO term enriched in the up-regulated DEGs in both hac1 alleles is ribosome biogenesis, which 175 occurs during rapid protein synthesis, and would be important for anabolic growth, not catabolic 176 senescence. Cytokinin action delays dark-induced senescence, in part, by maintaining the expression of 177 genes associated with ribosome GO terms (Kim et al., 2018b). 178 The Venn diagram in Figure 3 shows the overlap of genes with reductions in H3K9ac and H3K4me3 179 marks, as well as decreased expression in both hac1 alleles. Our analysis identified 44 genes 180 (Supplemental Table 2) with reductions in H3K9ac marks and gene expression. These potential HAC1 181 targets have enriched GO terms including response to chitin and response to abiotic stimulus. These GO 182 biological process terms have previously been associated with SAGs (Brusslan et al., 2015). Two of the 183 potential HAC1 targets, IGMT1 and CYP81F2 (green highlight in Supplemental Table 2), encode indole 184 glucosinolate biosynthetic enzymes, providing evidence that these secondary compounds are important 185 during leaf senescence and potentially regulated via histone acetylation. We also observed significant 186 reductions in H3K4me3 marks for these two genes in both hac1 alleles, further bolstering the presence 187 of chromatin changes. 188 189

Analysis of Leaf Senescence Phenotypes in Potential HAC1 Targets 190
We measured leaf senescence in T-DNA insertion lines disrupting three regulatory genes from the list of 191 44 potential HAC1 targets (yellow highlights in Supplemental Table 2). These include ERF022, MYB15 192 and TMAC2. Two of these genes: ERF022 and TMAC2 also show a reduction in H3K4me3 marks. ERF022 193 and MYB15 encode transcription factors while TMAC2 plays a negative role in ABA response (Huang and 194 Wu, 2007). Flowering time, NIT2 gene expression, and chlorophyll levels were quantified in these 195 mutants ( Figure 4A-C). We also showed that full-length mRNAs spanning the T-DNA insertion were not 196 produced in each mutant allele ( Figure 4D). was observed for med25 and hac1-1/med25 ( Figure 5A), but an additive effect in flowering phenotype 223 was not present. Chlorophyll levels were measured in leaf 7 in 45 day old plants, and higher chlorophyll 224 levels were observed in hac1-1, med25 and the hac1-1/med25 double mutants, and although all lines 225 were significantly greater than WT, none were significantly different from each other ( Figure 5B). These 226 data suggest that HAC1 and MED25 do not have an additive effect, as loss of one or both show similar 227 delays in flowering and chlorophyll loss. The erf022 mutant was also included in this experiment; it 228 bolted later and had more chlorophyll than WT, but it did not differ from the hac1-1, med25 or hac1-229 1/med25 mutant lines. 230 Gene expression was also evaluated in these mutant lines. As expected, ERF022 expression was 231 minimally detected in the erf022 mutant. A strong additive effect was seen between hac1-1 and med25 232 with much lower ERF022 expression in the hac1-1/med25 double mutant than in either single mutant 233 ( Figure 5C). These data suggest that MED25 guides HAC1 to histones at the ERF022 locus to direct 234 histone acetylation for increased chromatin accessibility. With respect to two other SAGs, NIT2 and 235 Lhcb2.4, the erf022 mutant showed the largest effect: minimal up-regulation of NIT2 ( Figure 5D) and 236 minimal down-regulation of Lhcb2.4 ( Figure 5E) as compared to hac1-1, med25 and hac1-1/med25. 237 These data suggest that loss of ERF022 has a more profound effect on the leaf senescence phenotype 238 than its down-regulation through loss of both HAC1 and MED25. Although the ERF022 transcript levels 239 were similar to the hac1-1/med25 double mutant ( Figure 5C), it is probable that the mRNA produced in 240 the erf022 mutant is inefficiently translated due to the T-DNA insertion in the 3'-UTR and led to a 241 stronger phenotype in erf022. In addition, there are likely more genes mis-regulated in hac1-1/med25 242 and these may have compensating effects on leaf senescence. 243

Conclusion 245
hac1 mutant alleles display a delay in leaf senescence implicating histone acetylation as a contributor to 246 the regulation of leaf senescence. A combined approach using ChIP-seq, RNA-seq and genetic analysis, 247 identified ERF022 as a novel positive effector of leaf senescence regulated by H3K9ac and H3K4me3 248 marks. ERF022 is possibly a direct target of HAC1, which operates in concert with MED25 to allow full 249 expression of ERF022 in older leaves. senescence, total chlorophyll was measured from leaf 7 from plants that had grown 28, 38 or 48 days. 379 Significant differences from WT are indicated by asterisks (t-test, p < 0.05) and were observed for both 380 hac1 alleles at 48 days. B) RNA was extracted from WT and the hac1-1 allele at 48 days from leaf 6 of the 381 same plants shown in panel A, and gene expression for three SAGs was measured by real-time qPCR. C) 382 Leaf 5 was removed from plants grown for 21 days, and floated on water in the dark for the indicated 383 number of days to observe dark-induced senescence. One leaf disc was removed from each leaf and 384 chlorophyll was measured. No significant differences were observed. All error bars show the 95% 385 confidence interval. 386 and chlorophyll (A) and total protein (B) were measured in hole-punch disks from leaves 12-14, n = 8. 388 Significant differences between WT were observed for chlorophyll, but not total protein (t-test, p < 389 0.05). Seeds were harvested from individual plants and batches of 100 dried seeds were subject to 390 elemental analysis (C). No significant differences in percent nitrogen were observed, n = 8. All error bars 391 show the 95% confidence interval. 392 modifications. WT, hac1-1 and hac1-2 (49 days, leaf 12-14) were subject to RNA-seq and ChIP-seq using 394 H3K9ac and H3K3me3 antibodies. Genes that showed a significant reduction in both hac1 alleles in 395 comparison to WT were considered to have lower expression (RNA-seq) or reduced histone marks (ChIP-396 seq). 397   Figure 1. Delayed age-related senescence in hac1 alleles. A) To observe age-related developmental senescence, total chlorophyll was measured from leaf 7 from plants that had grown 28, 38 or 48 days. Significant differences from WT are indicated by asterisks (t-test, p < 0.05) and were observed for both hac1 alleles at 48 days. B) RNA was extracted from WT and the hac1-1 allele at 48 days from leaf 6 of the same plants shown in panel A, and gene expression for three SAGs was measured by real-�me qPCR. C) Leaf 5 was removed from plants grown for 21 days, and floated on water in the dark for the indicated number of days to observe dark-induced senescence. One leaf disc was removed from each leaf and chlorophyll was measured. No significant differences were observed. All error bars show the 95% confidence interval. Chlorophyll, protein and seed nitrogen content in hac1 alleles. Plants were grown for 49 days and chlorophyll (A) and total protein (B) were measured in hole-punch disks from leaves 12-14, n = 8. Significant differences between WT were observed for chlorophyll, but not total protein (t-test, p < 0.05). Seeds were harvested from individual plants and batches of 100 dried seeds were subject to elemental analysis (C). No significant differences in percent nitrogen were observed, n = 8. All error bars show the 95% confidence interval.  gene expression and panel C shows total chlorophyll (n =6 for all genotypes). One biological replicate is shown, however similar results were obtained in a second biological replicate. Error bars for panels B and C show the 95% confidence interval. A t-test was used to evaluate significant differences: * = p<0.05, ** = p<0.01, *** = p<0.001. Gene expression is measured as 40 -ΔCt. The ΔCt value is the Ct value of ACT2 -the Ct value of the gene of interest. Panel D shows that full-length mRNAs were not produced in T-DNA inser�on alleles. The cDNAs templates are shown above the PCR products and the primers are shown below. Primer sequences are available in Supplemental  Figure 5: Senescence phenotypes in hac1-1/med25 double mutants. All lines were evaluated for flowering �me (panel A). At 45 days of growth, chlorophyll was measured in leaf 7 and RNA was extracted from leaf 6. Total chlorophyll levels (µg per leaf disk) are shown in panel B. ERF022 (panel C), NIT2 (panel D) and Lhcb2.4 (panel E) mRNA levels are shown. A t-test was used to evaluate significant differences: * = p<0.05, ** = p<0.01, *** = p<0.001. All error bars show the 95% confidence interval, n=6 for all genotypes.)