Genome-Wide Identication of Histone Acetyltransferases and Deacetylases in Epiphytic Orchid Plant Dendrobium Catenatum

Background: Dendrobium catenatum is a kind of precious Traditional Chinese Medicine, and possesses unique developmental programs and epiphytic lifestyle. Histone acetyltransferases (HATs) and histone deacetylases (HDACs) are responsible for maintenance of histone acetylation homeostasis, and they are widely involved in developmental regulation and stress responses via remodeling chromatin structure, but their biological functions in orchid plants remain largely unknown. Results: Here we identied 8 HAT genes and 14 HDAC genes from D. catenatum genome. We carried out phylogenetic construction, gene structure and domain architecture analysis of these D. catenatum HAT/HDAC (DcHAT/DcHDAC) proteins using the well-dened homologs from the model plants Arabidopsis thaliana and Oryza sativa as references. DcHAT proteins can be classied into four families: GNAT family (3 members), MYST family (2), CBP family (2), and TAFII250 family (1), and DcHDAC proteins can be grouped into three families: RPD3/HDA1 family (10), SIR2 family (2), and HD2 family (2), in accordance with previously described classication. Cis-acting element analysis indicated that the promoter regions of DcHAT/DcHDAC genes contain diverse stress-responsive elements. Subcellular localization predictions suggested that DcHAT/DcHDAC proteins might be localized in nucleus or/and cytoplasm. Spatiotemporal expression proling showed that DcHAT/DcHDAC genes generally exhibit either universal or specic expression pattern in different tissues and organs. Finally, stress response assay suggested drought treatment signicantly represses the expression of DcHAG1 and DcHDA14, cold exposure evidently inuences the expression of DcHAG1 and DcHDT1, and heat shock has a broad impact on the expression of DcHAT/DcHDAC genes. Conclusions: In this study, we reveal the identication and expression proles of DcHATs and DcHDACs in epiphytic orchid plant D. catenatum, indicating their roles in the regulation of both long-term developmental programs and short-term stress responses. This study provides a Class-II members displayed diverse expression patterns but had uniformly low expression in pollinia; DcHDA8 harbored intermediated expression in almost all the organs; DcHDA14, DcHDA15, DcHDA5a seemingly had distinct expression levels in leaf, oral bud, and root/stem, respectively; DcHDA5b transcripts were absent in detected organs, indicating functional divergence of the DcHDA5a/5b genes. SIR2 members DcSRT1/2 were also weakly expressed in all the organs except for intermediated expression of DcSRT1 and DcSRT12 in pollinia and labellum, respectively. On the other hand, HD2 members DcHDT1/2 were strongly expressed in all the organs except for DcHDT1 not in pollinia. These results indicated that DcHATs/DcHDACs have undergone the functional conservation and divergence during evolution.

In plant, HATs were divided into four families: TATA-binding protein Associated Factor (TAFII250)/HAF, p300/ cAMP-responsive element Binding Protein (CBP)/HAC, GCN5-related N-terminal AcetylTransferase (GNAT)/HAG and MOZ-Ybf2/Sas3-Sas2-Tip60 (MYST)/HAM [5]. HDACs were classi ed into three families: Histone Deacetylase 2 (HD2), Silent Information Regulator 2 (SIR2), and Reduced Potassium Dependency 3/Histone DeAcetylase 1 (RPD3/HDA1) that was further divided into three classes: class I (RPD3-like), class II (HDA1-like) and class IV [18]. Both HATs and HDACs associated with dynamic histone acetylation levels have pleiotropic effects on regulation of plant development programs and response to environmental stresses and stimuli. In vegetative phase, histone acetylation regulates the cellular patterning of Arabidopsis root epidermis by relaying positional information, single mutants for HDAC members HDA6, HDA18 and HDA19 and HATs members GCN5 and HAF2 exhibited altered epidermal phenotypes [19,20]. Arabidopsis GCN5 is implicated in regulating PLETHORA (PLT) gradient-mediated root stem cell niche maintenance and transit amplifying cell proliferation [21]. AtHAG1/GCN5 is involved in establishment of competence for de novo shoot regeneration through acting as an epigenetic switch that renders somatic cells to acquire regeneration potential via catalyzing histone acetylation at key root-meristem gene loci in developing callus [22].
HDA6 and HDA19 function redundantly to repress somatic embryogenesis via suppressing the expression of master embryo regulators; simultaneous knockdown of HDA6/HDA19 causes arrested growth and the formation of embryo-like structures on the true leaves [23]. In reproductive phase, AtGCN5 is required to regulate the oral meristem activity through the WUSCHEL (WUS) / AGAMOUS (AG) pathway, and mutation of AtGCN5 leads to terminal ower, homeotic transformation and ectopic carpel [24]. Arabidopsis HDA9-mediated H3K27 deacetylation is required for Polycomb repressive complex 2 (PRC2) -mediated H3K27me3, leading to FLC repression and owering time regulation [25]. In some crops and vegetables, maize Rpd3/HDA1 member HDA108 is involved in regulation of plant height, shoot and leaf development, male and female in orescence patterning, and fertility [26]. Cotton GhHDA5 is involved in H3K9 deacetylation and ber initiation, and knockdown of GhHDA5 suppresses ber initiation and lint yield [16]. Tomato histones acetylation status affects ethylene-dependent fruit ripening and carotenoid accumulation. Knockdown of SlHDA1 or SlHDA3 results in accelerated fruit ripening process along with short shelf life characteristics, but knockdown of SlHDT3 results in delayed ripening along with prolonged shelf life [27][28][29]. In terms of environmental stimulus and stress response, histone acetyltransferase TAF1/HAF2 and GCN5 and histone deacetylase HD1 (HDA19) are involved in light regulation of growth and gene expression [30]. HAM1, HAM2 and HAG3 participate in UV-Binduced DNA damage repair and signaling, but HAF1 and HAC1 are only involved in UV-B signaling [31][32][33]. Populus AREB1 recruits the histone acetyltransferase unit ADA2b-GCN5 to elevate H3K9ac deposition and RNA polymerase II occupancy on drought-responsive genes (PtrNAC006/007/120), leading to their activation and increased drought tolerance. Accordingly, disruption of each member of the trimeric AREB1-ADA2b-GCN5 complex causes highly drought sensitivity [34]. Overexpression of a Populus RPD3/HDA1-type gene 84KHDA903 in tobacco enhanced drought tolerance [35]. In fact, Lys acetylation-mediated posttranslational regulation occur not only on histones, but also on nonhistone proteins. SRT2 can target the ATP synthase and the ATP/ADP carriers, ne-tuning mitochondrial energy metabolism [36]. HDA6 can deacetylate the K189 residue of BIN2 to inhibit its kinase activity and increase brassinosteroid signaling in Arabidopsis [37].
Tiepi-Shihu (Dendrobium catenatum, or D.o cinale, or D. candidum) is a rare traditional Chinese medicinal plant in Orchidaceae [38], for their traditional properties of nourishing the kidney, moisturizing the lung, bene ting the stomach, promoting the production of body uids and clearing heat [39]. The earliest records of Dendrobium utilization date back to"Shen Nong's Herbal Classic"( ) written nearly 2,300 years ago during the Eastern Han Dynasty. Especially, Dendrobium was listed as the head of the nine Chinese fairy grasses in "Collected Taoist Scriptures"( ) during the Tang Dynasty. At present, D. o cinale Kimura et Migo is individually listed in Pharmacopoeia of the People's Republic of China (2015 Edition) for rich polysaccharides, whereas D. nobile Lindl., D. chrysotoxum Lindl. and D. mbriatum Hook. are uniformly classi ed in the item of Dendrobium with dendrobine, erianin and dendrophenol as the active ingredients, respectively [40]. Modern pharmacological study suggested that D. catenatum possesses hepatoprotective, anticancer, hypoglycemic, antifatigue, antioxidant, anticonstipation, hypoglycemic, gastric ulcer protective, and antihypertensive effects, immunoenhancement, and so on [41].
The biological functions of HATs and HDACs in D. catenatum remain unknown, so a genome-wide analysis of HAT and HDAC superfamilies was performed to gain insight into their potential roles. In this study, we identi ed 8 HAT genes and 14 HDAC genes throughout the genome of D. catenatum. Evolutionary relationships of DcHATs and DcHDACs were assessed through phylogenetic reconstruction, gene structure and domain organization. Promoter cis-acting element and protein subcellular localization were investigated using online software. Gene expression pro ling in different tissues or organs as well as stress responses were determined via RNA-seq data and quantitative RT-PCR. Together, our results shed light on the involvement of HATs and HDACs in growth and development as well as environmental tness of D. catenatum, and contribute to further investigate the evolution and function of histone acetylation in orchid medical plants.

Results
Identi cation of HAT and HDAC genes in D. catenatum HATs and HDACs are responsible for catalyzing the deposition and removal of acetyl groups from histones, respectively. Both classes of enzymes netune genome-wide histone acetylation status, leading to modi cation of chromatin structure, uctuation of gene expression and reprogramming of developmental progresses. To obtain all the HAT and HDAC homologs in D. catenatum, the well-examined respective HAT and HDAC proteins in both Arabidopsis and rice were used as queries to BLASTp against D. catenatum genome; then the hits were further con rmed through reciprocal BLAST. Ultimately, 8 HAT genes and 14 HDAC genes were obtained in D. catenatum (Table 1 and Additional le 1), and they were named by reference to the corresponding Arabidopsis homologs with the pre x"Dc"for distinguishing. Of the 8 DcHATs, 3 members belong to the GNAT family, 2 belong to the MYST family, 2 belongs to the CBP family and 1 belongs to the TAFII250 family. Of the 14 DcHDACs, 10 belong to the RPD3/HDA1 family, 2 belong to the SIR2 family and 2 belong to the HD2 family.
HAT: p300/CBP/HAC family To determine the evolutionary relationship of DcHATs with those from rice, and Arabidopsis, total protein sequences from these organisms were used to construct a neighbour-joining phylogenetic tree (Fig. 1). We observed one subgroup related to CBP, TAFII250 and MYST, and three subgroups for GNAT. Eight DcHATs can be divided into four families, CBP (DcHAC1a/1b), GNAT (DcHAG1/2/3), TAFII250 (DcHAF1), and MYST (DcHAM1/2). HAT/HDAC members in D. catenatum usually have extremely longer genomic sequences than those in Arabidopsis and rice (Fig. 2), largely owing to existence of overlength introns.
The CBP family contains 5 members in Arabidopsis, 3 in rice and 2 in D. catenatum (Fig. 1). HAC-like proteins are usually characteristic of several successive domain combinations: an N-terminal ZnF-TAZ domain, a PHD-nger domain, a CBP-type HAT domain, a ZnF-ZZ domain and a C-terminal ZnF-TAZ domain, except AtHAC2 and OsHAC5 lacking the N-terminal ZnF-TAZ domain. Phylogenetic analysis (Fig. 1) revealed that DcHAC1a/1b paralog pairs were grouped together with AtHAC1/12 and OsHAC1a/1b pairs, consistently, DcHAC1a and DcHAC1b have 74.1% protein sequence identity, both display 51.6% and 50.5% identity with AtHAC1, respectively. No counterparts in D. catenatum were classi ed into AtHAC4/5, OsHAC5 or AtHAC2 branches, indicating CBP family in D. catenatum might have experienced less gene duplication events during evolution. In Arabidopsis, AtHAC1 participates in pleiotropic developmental programs, and its mutation causes delayed owering, decreased primary root length, and poor fertility [42]. AtHAC5 and AtHAC12 have redundant functions with AtHAC1 in regulate owering time by repressing the expression of a major oral repressor FLC [43]. AtHAC1/5 are required for salicylic acid (SA)triggered immunity and PR induction. AtHAC1/5 together with the master immune regulator NPR1 are recruited to PR chromatin through NPR1 interacting with transcription factor TGA upon SA signal, resulting in the formation of HAC-NPR1-TGA complex that activates PR transcription by histone acetylation-mediated epigenetic reprogramming [44]. So, DcHAC1a/1b might undertake most of HAC roles in development regulation and stress response of D. catenatum.
HAT: GNAT/HAG family HAG family was divided into three distinct clades (HAG1/GCN5, HAG2 and HAG3/ELP3), each of which contains a single member in examined species. All HAG family proteins possess a landmark GNAT-type HAT domain (Acetyltransf_1, PF00583), moreover, each clade has additional distinct domains, such as HAG1/GCN5 with a C-terminal bromodomain, HAG2 with Hat1_N domain (PF10394) and HAG3 with a central Elp3 domain (IPR006638). Bromodomain found in many chromatinassociated proteins can interact speci cally with acetylated lysine and mediate the interaction between histone acetyltransferases and histone proteins [45]. In HAG1 clade, Arabidopsis AtGCN5/HAG1 is a major HAT and usually acts as the catalytic subunit of several multicomponent HAT complexes that acetylate lysine residues of histone H3 [46]. AtGCN5/HAG1 plays an essential role in diverse plant development processes, such as meristem activity, cell differentiation, plant architecture, root, leaf and oral organogenesis, and de novo shoot regeneration [21,22,24,46,47]. AtGCN5/HAG1 is also involved in nutrient balance and fatty acid metabolism. AtGCN5 contributes to iron homeostasis by directly affecting the H3K9 and H3K14 acetylation level of the citrate e ux protein FRD3. Mutation of AtGCN5 impairs iron translocation from the root to the shoot [48]. AtGCN5 modulates fatty acid biosynthesis by affecting the H3K9/14 acetylation levels of FAD3; loss of AtGCN5 causes decreased ratio of α-linolenic acid to linoleic acid in seed oil [49]. AtGCN5 regulates stem cuticular wax biosynthesis by modulating the H3K9/14 acetylation of CER3; AtGCN5 disruption impairs cuticular wax accumulation [50]. In addition, AtGCN5/HAG1 is involved in response to environmental stresses. AtGCN5 is required for thermotolerance and salt tolerance through H3K9/14 acetylation and activation of the key stress-responsive genes [51,52]. DcHAG1 displays same domain organization with other plant GCN5 proteins, shares 62.5% and 63.4% sequence identity with AtHAG1/GCN5 and OsHAG1, respectively. But the GNAT-type HAT domain of DcHAG1 is 88.3% sequence identical to that of AtHAG1/GCN5. Considering the signi cance of HAG1 clade, it is requisite to investigate the regulatory function of DcHAG1 in D. catenatum. In HAG3 clade, Arabidopsis HAG3 participates in UV-B-induced DNA damage repair and signaling. Knockdown of HAG3 leads to reduction of DNA damage and growth inhibition by UV-B [31]. DcHAG3 has 88.4% and 90.4% sequence identity with AtHAG3 and OsHAG3, respectively, indicating DcHAG3 might have similar role in UV-proof.
HAT: TAFII250/HAF family Dendrobium genome contains a single TAFII250 homolog DcHAF1, like rice with one homolog OsHAF1, but different from Arabidopsis with a twin copy AtHAF1/2 ( Fig. 1). DcHAF1 gene consists of 20 exons, OsHAF1 contains 22 exons and AtHAF1/2 contain 21/17, respectively. However, the genomic sequence of DcHAF1 is nearly 90 kb long due to extremely large introns, much longer than that of OsHAF1 (14 kb) and AtHAF1/2 (10/8 kb). HAF1 encoding proteins in the examined species are about 2000 aa long, and DcHAF1 shares 46.7% and 53.3% amino acid sequence identity with AtHAF1 and OsHAF1, respectively. HAF proteins are usually characterized by coexistence of an N-terminal TBP-binding domain, a DUF3591 domain, a C2HC-ZnF domain, and a C-terminal bromodomain. DUF3591 domain is a unique HAF-type HAT domain, which encompasses a Nterminal UBQ motif, in uencing protein function in a nonproteolytic manner [53]. Bromodomain can speci cally recognize acetyl-lysine (Kac) and is functionally linked to the HAT activity of co-activators in regulating gene transcription [45]. Phylogenetic analysis revealed DcHAF1 clustered together with OsHAF1 to form monocot branch with high bootstrap value, whereas AtHAF1/2 cluster together to represent dicot branch, suggesting a duplication event might occur near the origin of dicots. In Arabidopsis, HAF2 (TAF1) gene is involved in integrating light signals and activating light-responsive gene transcription through histone acetylation; haf2 mutant displays reduced chlorophyll accumulation, pale cotyledon and yellowish young leaf [54]. HAF2 facilitates H3 acetylation deposition at the promoters of the core clock genes PRR5 and LUX to contribute to stable maintenance of circadian oscillation [55]. It is intriguing to unveil the biological mechanism underlying the divergence of HAF gene copy number between dicot and monocot during evolution.
HAT: MYST/HAM family D. catenatum genome has two MYST-like genes, DcHAM1/2, but so far both only have partial sequences available (Fig. 1). Based on known full-length MYST sequences in other species, MYST family members usually share the conserved domain combinations: an N-terminal chromodomain, a central C2H2-ZnF domain and a C-terminal MYST-type HAT domain MOZ-SOS.
Chromodomain is a methyl-speci c histone binding module [56], indicating that MYST-catalyzed histone acetylation might coordinate with histone methylation to perform the functions. In Arabidopsis, HAM1 and HAM2 act redundantly to control female and male gametophyte development as well as owering time. Knockout of HAM1 and HAM2 resulted in lethality; in ham sesquimutants ham1/+; ham2/ham2 and ham1/ham1;ham2/+, half of the ovules aborted due to an arrest of mitotic cell cycle at one-nucleate stage of megagametogenesis [57]. On the other hand, HAM1 and HAM2 regulate owering time by deposition of histone H4 lysine 5 acetylation (H4K5ac) within FLC and MAF3/4 chromatins; knockdown of HAM1/2 by amiRNA or T-DNA insertion results in early owering and reduced fertility, whereas overexpression of HAM1 caused late owering [58].

HDAC: PRD3/HDA1 family
To determine the evolutionary relationship of DcHDACs with those from rice, and Arabidopsis, total protein sequences from these organisms were used to construct a neighbour-joining phylogenetic tree (Fig. 3). HDACs were grouped into three families, SIR2, HD2, and RPD3/HDA1, which is further divided into Class-I, II, and IV. For the members in each family or clade, the closer the genetic relationship, the more similar the gene structure (Fig. 4). All Dendrobium RPD3/HDA1 members contain the conserved histone deacetylase domain (Hist_deacetyl) (Fig. 3), but their length varies from 356 aa to 685 aa, except DcHDA2b and DcHDA5b with partial sequences. In addition, DcHDA9 has another transmembrane domain at N-terminus, whereas DcHDA15 possesses another ZnF_RBZ domain, which was found in a nucleoporin protein RanBP2 located on the cytoplasmic side of the nuclear pore complex functioning in nuclear protein import [59].
Class-I of RPD3/HDA1 family possesses 10 members in D. catenatum, 12 in Arabidopsis, and 14 in rice; and they can be divided into four subclasses: I-1 (HDA19), I-2 (HDA6/7), I-3 (HDA9/10/17), and rice-speci c I-4 (HDA701/709). Subclass I-1 compresses a single member each in D. catenatum and Arabidopsis, and 3 homologs in rice. In Arabidopsis, down-regulation of HDA19 (AtHD1) leads to pleiotropic defects including early senescence, serrated leaves, aerial rosettes formation, delayed owering, and defects in oral organ identity [60,61]. Deplete of HDA19 results in markable tolerance to abiotic stresses including salinity, drought, and heat [62,63]. In rice, knockdown of HDA703 (OsHDAC3) reduces rice peduncle elongation and fertility, knockdown of HDA710 (OsHDAC2) affects vegetative growth, and down-regulation of HDT702 (OsHDAC1) causes the production of narrowed leaves and stems [7]. overexpression of HDA702 enhances growth rate, alters plant architecture [64], and increases root growth through deacetylating OsNAC6 locus [65]. HDA702 implicates in formation of IDS1-TPR1-HDA1 transcriptional repression complex, contributing to negative regulation of salt tolerance [66]. These data suggest that HDA702/703/710 in rice may have divergent developmental functions compared with closely related homologs in Arabidopsis. DcHDA19 has 77.6% and 77.4% sequence identity with AtHDA19 and HDA702, respectively. It is interesting to explore whether the function of DcHDA19 is more near to AtHDA19 or HDA702. Subclass I-2 harbors a single copy each in D. catenatum (DcHDA6) and rice (HDA705), and two homologs in Arabidopsis (AtHDA6/7). Arabidopsis HDA6 is involved in leaf development [67], cellular patterning of root epidermis [68], owering time [69], repression of somatic embryogenesis [23], circadian rhythms [55], and light intensity adaption [70]. AtHDA6 is also required for repressing pathogen defense response; loss of HDA6 resulted in enhanced resistance to hemibiotrophic bacterial pathogen Pst DC3000 and constitutively activated expression of pathogen-responsive genes [71]. AtHDA7 is crucial for female gametophyte development and embryogenesis, and silencing of AtHDA7 causes reduced seed set [72]. In rice, overexpression of HDA705 in rice decreases the resistance to ABA and salt stress during seed germination, but enhances the resistance to osmotic stress during the seedling stage [73]. DcHDA6 has 67.2% and 66.6% sequence identity with AtHDA6 and HDA705, respectively, but only 49.9% with AtHDA7, indicating the function of DcHDA6 is more near to AtHDA6 than AtHDA7. Subclass I-3 consists of a single member each in D. catenatum and rice, and 3 members in Arabidopsis. Arabidopsis HDA9 is involved in preventing precocious owering under short day [74,75], suppressing seed germination [76], and promoting the onset of leaf senescence [77]. Moreover, the hda9 mutant is insensitive to salt and polyethylene glycol treatments [78]. DcHDA9 has 78.5% sequence identity with AtHDA9, indicating there might be functional similarity to a large extent.
Class-II of RPD3/HDA1 family contains 5 members each in D. catenatum, Arabidopsis and rice; and they can be divided into four subclasses: II-1 (HDA8), II-2 (HDA14), II-3 (HDA15), and II-4 (HDA5). Among them, subclass II-3 possesses a single member in D. catenatum, Arabidopsis, and rice. DcHDA15 contains an additional ZnF_RBZ domain (SMART, SM00547), and shares the sequence identity of 53.2% and 55% with AtHDA15 and OsHDA704, respectively. DcHDA15 gene, like rice HDA714, consists of 16 exons, but is around 112 kb in length, far larger than HDA714 (~ 20 kb). In Arabidopsis, HDA15 is a negative component of light-regulated seed germination. PHYTOCHROME INTERACTING FACTOR3 (PIF3) recruits HDA15 to repress chlorophyll biosynthesis and photosynthesis gene expression in etiolated seedlings [79]. HDA15-PIF1 acts as a key repression module directing the transcription network of seed germination [80]. NF-YC1/3/4/9 transcriptional co-repressors interact with HDA15 to inhibit hypocotyl elongation in the light partially via histone deacetylation [81]. Collected date suggested that DcHDA15 might have a role in photomorphogenesis and skotomorphogenesis. Subclass II-4 has a pair of paralogs in D. catenatum (DcHDA5a/b) or Arabidopsis (HDA5/18), and one member in rice (HDA713). DcHDA5a has the sequence identity of 53.8% and 59.4% with AtHDA5 and OsHDA713, respectively. In Arabidopsis, HDA5 regulates owering time through targeting the chromatin of owering repressor genes FLC and MAF1 as well as interacting with FVE, FLD and HDA6 to form a complex. The hda5 mutant is late-owering due to up-regulated expression of FLC and MAF1 associated with increased deposition of activation markers, histone H3 acetylation and H3K4 trimethylation (H3K4me3) [82]. HDA18-mediated histone acetylation affects cellular patterning in root epidermis, and loss or overexpression of HDA8 lead to cells at the N position having H fate [83].
Class-IV of RPD3/HDA1 family contains 2 copies in D. catenatum, but only a single member in Arabidopsis and rice, respectively. DcHDA2a displays protein sequence identity of 61.7% and 71.1% with AtHDA2 and OsHDA706, respectively. No related function was reported so far.
HDAC: HD2 family HD2 family is speci c to plant, and rstly appeared in Streptophyta green algae Charophyta, but not in Chlorophyta [84]. Here we identi ed two HD2 homologs in D. catenatum (DcHDT1/2), compared with 2 members in rice (HDT701/702) and 4 members in Arabidopsis (AtHD2A-D). HD2-like genes usually compress 7 ~ 9 exons and about 1 kb in length (Fig. 4) but they have signi cant sequence divergence during evolution. For instance, there are only 50.6% protein sequence identity between DcHDT1 and DcHDT2, which both also share 17.9% and 45.9% sequence identity with OsHDT701, 24.5% and 7.2% with OsHDT702, respectively. DcHDT1/2 are characterized of the conserved C2H2-type zinc nger domain at C-terminus, but this domain is loss in Arabidopsis AtHD2A/2C and rice HDT702. HD2 proteins lacking zinc nger domain (Gr2 HD2s) evolved from those containing zinc nger domain (Gr1 HD2s) during angiosperm diversi cation, and Gr2 HD2s might ful ll complementary functions with the corresponding Gr1 HD2s [84]. In Arabidopsis, HD2 histone deacetylases have been demonstrated to participate in repressing gene expression [85]. HDT1 (HD2A) and HDT2 (HD2B) paralogs are involved in regulating root meristem maintenance partially through ne-tuning gibberellin metabolism via deacetylation and repression of GA2ox2 [86]. HD2B was identi ed as a distinguished factor associated with seed dormancy by genome-wide association mapping of 113 wild-type accessions with natural variation; inactivation of HD2B contributes to maintenance of seed dormancy in dormant accessions [87]. In addition, HD2 family is involved in the response to biotic and abiotic stresses. HD2B is implicated in plant innate immune defense in the form of MPK3-HD2B regulatory module [88]. Loss of HD2C results in hypersensitivity to ABA, NaCl and salt stress [89]. Overexpression of HDT701 in rice reduces resistance to ABA, salt and osmotic stress during seed germination, but enhances resistance to salt and osmotic stress during the seedling stage [90]. Further work is required to exploit why Gr2 HD2s is loss in D. catenatum during evolution.
HDAC: SIR2 family SIR2 family contains 2 members in D. catenatum, Arabidopsis, and rice, respectively. All these proteins possess a single copy of SIR2-type HDAC domain, whose catalytic activity requires NAD as a cofactor [91]. SIR2 family was divided into two clades, SRT1 clade including DcSRT1/AtSRT1/OsSRT701 and SRT2 clade including DcSRT2/AtSRT2/OsSRT702, indicating SRT1/2 diverged before the split of dicot and monocot. Consistently, the members in SRT1 or SRT2 branch have similar exon-intron gene structure architecture, respectively. Protein alignment showed DcSRT1 shares sequence identity of 60.3% and 62% to AtSRT1 and OsSRT701, respectively. DcSRT2 shares 66.6% and 61.4% sequence identity with AtSRT2 and OsSRT702, respectively. In Arabidopsis, SRT2 but not SRT1 is localized at the inner mitochondrial membrane and interacts with some protein complexes, mainly involved in energy metabolism and metabolite transport. Mutation of SRT2 results in no growth phenotype but rather a metabolic disorder with altered levels in sugars, amino acids, and ADP contents [36]. AtSRT1 negatively regulates plant tolerance to stress and glycolysis but stimulates mitochondrial respiration through both epigenetic regulation and modulation of AtMBP-1 transcriptional activity [92]. The reduction and loss of AtSRT1 result in signi cant resistance to ABA, while overexpression of AtSRT1 leads to hypersensitivity to ABA, salt and osmotic stresses. SRT1/2 are involved in mediating transcriptional repression of ethylene signaling by decreasing the levels of H3K9 acetylation [93]. OsSRT1 participates in maintenance of genome stability and defense against DNA damage. Knockdown of OsSRT1 triggers increased H3K9ac and decreased H3K9me2, resulting in oxidative burst, DNA fragmentation, and cell death, whereas overexpression of OsSRT1 enhances tolerance to oxidative stress [94]. OsSRT1 represses glycolysis by both epigenetic regulation and deacetylating glycolytic GAPDH to inhibit its nuclear accumulation and moonlighting function as a transcriptional activator of glycolytic genes [95]. Taken together, D. catenatum SIR2 type HDACs would have a predominant role in energy metabolism.

Subcellular localization prediction
To investigate the functional compartmentation of DcHATs/DcHDACs, we predicted their subcellular localization through different software (Table 2). SLP-Local predictor showed the potential subcellular localization sites for most of DcHATs/DcHDACs with low reliability index (RI < 3), except for DcHAF1, which might be localized in the nucleus or cytosol with higher reliability (RI = 6). However, Target P detected potential localization for nearly one half of these proteins with higher reliability (RC < 2), and all of them are localized in nucleus or cytosol. Uniprot program generated prediction results for only half of DcHATs and DcHDACs, including a SIR2-type HDAC, DcSRT2 with mitochondria localization as well as DcHAC1a/1b, DcHAG2, and DcHDA5a/5b/6/8/9/15/19 with nucleus localization. NES predictor (NetNES) detected the potential leucine-rich nuclear export signal (NES) for DcHATs and DcHDACs with a threshold value of 0.5. NES was found in most of DcHAT members except for DcHAG1, but only found in some DcHDAC members, such as DcHDA2b/6/9 and DcSRT2. The presence of the NES in 6 DcHATs and 4 DcHDACs indicates that they might perform the functions in both nucleus and cytosol. WoLF PSORT displayed low frequency values for DcHATs and DcHDACs; most of them were predicted with a major localization in the nucleus or cytosol, however, DcHAF1 generated the relatively close localization scores for nucleus (7) and cytosol (6), consistent with the dual localization predicted by other software (Table 2).

Promoter cis-element analysis
To detect the cis acting regulatory elements in DcHAT and DcHDAC gene promoter regions that determine the gene expression pro les, 2 kb sequences upstream of the ATG initiation codon were assessed via PlantCARE database [96]. Some phytohormone and stress-responsive cis-acting elements were selected to display (Fig. 5). According to function classi cation, these elements were mainly divided into two groups: phytohormone response elements including MeJA-responsive (82), abscisic acid responsive (27), gibberellin-responsive (16), auxin-responsive (14) and salicylic acid-responsive (10); stress response elements including anaerobic induction (28), low-temperature responsive (21), drought-inducibility (15), defense and stress responsive (10), anoxic speci c inducibility (2) and wound-responsive (1). Overall, the four elements with the highest number were MeJA-responsive (82), anaerobic induction (28), abscisic acid responsive elements (27), and low-temperature responsive (21). Among these genes, DcHDA2b promoter only had a single predicted element, while DcHAG2 had maximal 18 elements. MeJA-responsive element (16/21) and anaerobic induction element (16/21) were found in most of DcHAT and DcHDAC promoter regions. Although experimental evidence is required to further check the correlation between cis-elements and gene expression response to different conditions, collected data revealed the involvement of DcHAT and DcHDAC genes in stress responses.

Expression pro lings of DcHATs and DcHDACs in tissues and organs
To explore the tissue and organ-speci city of DcHATs and DcHDACs, their transcript levels were monitored in different plant tissues and organs, including leaf, root, green root tip, white part of root, stem, ower bud, sepal, labellum, pollinia, and gynostemium by RNA-seq data [97]. In DcHAT superfamily ( Fig. 6A and Additional le 2), CBP members DcHAC1a/1b displayed similar expression pattern in vegetative organs and oral buds, but showed different patterns in mature oral organs, such as high expression of DcHAC1a and DcHAC1b in pollinia and root, respectively. Both MYST members DcHAM1/2 had distinct expression pro les, such as DcHAM1 with intermediated expression but DcHAM2 lacking expression in detected tissues and organs. GNAT members DcHAG1/2/3 and TAFII250 member DcHAF1 displayed high expression in almost all the tissues and organs. In DcHDAC superfamily ( Fig. 6B and Additional le 2), Class-I RPD3/HDA1 members had similar and intermediated expression patterns, except for DcHDA6 with the highest expression in pollinia where DcHDA19/9 were absent. Class-IV members DcHDA2a/2b had similar but weak expression pro les. Class-II members displayed diverse expression patterns but had uniformly low expression in pollinia; DcHDA8 harbored intermediated expression in almost all the organs; DcHDA14, DcHDA15, DcHDA5a seemingly had distinct expression levels in leaf, oral bud, and root/stem, respectively; DcHDA5b transcripts were absent in detected organs, indicating functional divergence of the DcHDA5a/5b genes. SIR2 members DcSRT1/2 were also weakly expressed in all the organs except for intermediated expression of DcSRT1 and DcSRT12 in pollinia and labellum, respectively. On the other hand, HD2 members DcHDT1/2 were strongly expressed in all the organs except for DcHDT1 not in pollinia. These results indicated that DcHATs/DcHDACs have undergone the functional conservation and divergence during evolution.

Expression of DcHATs and DcHDACs in response to environmental stresses
To examine the responses of DcHATs and DcHDACs to cold stress, the expression levels of DcHATs and DcHDACs were evaluated through analyzing the RNA-seq data from the leaves of D. catenatum seedlings grown at 20 °C (control) and 0 °C for 20 h, respectively [98] (Fig. 7 and Additional le 3). In DcHAT superfamily, the expression DcHAG1 transcription was markedly induced, but that of DcHAM1 was repressed by cold exposure. In DcHDAC superfamily, the transcription level of DcHDA14 was increased, but that of DcHDT1/2 was obviously decreased in responding to cold exposure.
To examine the responses of DcHATs and DcHDACs to drought stress, the transcriptional dynamics of DcHATs and DcHDACs were evaluated by analyzing the RNA-seq data from the leaves of D. catenatum seedlings under drought-recovery treatment [99] (Fig. 8 and Additional le 4). Brie y, the plants were irrigated on the 1st day, and kept dry for next 6 days, and rehydrated on the 8th day. Leaves were sampled at 18:30 on the 2nd (DR8), 7th (DR10), 8th (DR11), and 9th (DR15) days, respectively. The results suggested that 5-day drought treatment decreased the expressions of DcHAG1, DcHDA14, and DcHDA9, but increased the expression of DcHDA5a, and then rehydration restored the transcription of DcHAG1, DcHDA14, and DcHDA9 to the normal level.
To explore the response of DcHATs and DcHDACs to heat shock, the expression pro les of DcHATs and DcHDACs in the leaves of D. catenatum seedlings treated at 35 °C for different times were monitored by real-time quantitative RT-PCR (RT-qPCR) (Fig. 9). The results show that the expression levels of 4 DcHAT genes and 10 DcHDAC genes were signi cantly in uenced by heat treatment, including DcHAM2 in MYST family, DcHAG1/2/3 in GNAT family, DcHDA6 /19/9 (Class I), DcHDA2a/2b (Class IV), DcHDA5a/5b (Class II) in RPD3/HDA1 family, and DcHDT1/2 in HD2 family. At 3 h after treatment (HAT), 8 genes were involved in early response to heat shock, including 5 upregulated genes and 3 downregulated genes. At 6 HAT, 7 genes were involved (6 upregulated genes and 1 downregulated genes). At 12 HAT, 5 genes were evidently induced, but no downregulated genes were detected. These indicate that the expression of DcHATs and DcHDACs are generally in positive response to heat shock. Notably, DcHAM2 and DcHDA5b transcripts were highly induced upon exposure to heat shock, but not detected in the expression pro les of different tissues and organs examined, indicating these two genes are mainly involved in heat shock response.

Discussion
Evolution and functional conservation of HAT and HDAC gene family HATs/HDACs catalyzed histone acetylation-deacetylation interconversions realize the switch between permissive (open) and repressive (close) states of chromatin, leading to induction/silencing and upregulation/downregulation of gene expression in response to developmental cues and environmental stimuli. Genome-wide identi cation of HATs/HDACs have been documented in some representative plant species, such as Arabidopsis, sweet orange, apple, tomato, rice, maize, and liverwort. Unexpectedly, HAG family in apple, orange, and tomato was reported to contain 50, 45, and 26 members, respectively [9,10,12], far more than that in other reported species. So, we further checked and screened the true HAG paralogs through more strict criteria satisfying the rst hit for each other in reciprocal blast and inclusion of Acetyltransf_1/Elp3 domain combination. eventually, we only detected 5, 4, and 3 HAGs in apple, orange, and tomato, respectively. Intriguingly, an ideal grain-weight QTL in rice has been identi ed and veri ed as a novel HAT member, OsglHAT1 (Os06g44100) with Acetyltransf_1 domain but lack of Elp3 domain, which further broads the scope of HATs and also provides a single target for directional yield improvement breeding [100]. Thus, it is logical to speculate that any protein containing Acetyltransf_1 domain might harbor HAT activity.
The epiphytic orchid plant D. catenatum has highly specialized organogenesis and unique living mode, but no HATs/HDACs homologs were reported so far. In present study, we identi ed 8 HATs and 14 HDACs in D. catenatum, as well as 7 HATs and 13 HDACs in its closely relative Phalaenopsis equstris (Fig. 10). Overall, the gene number in each HAT/HDAC family remains relatively stable throughout the green plant lineage during evolution, resulting in gene copy number that is not positively correlated with genome size. This result indicated that HATs/HDACs associated with histone acetylation/deacetylation have functional conservation and dosage stability during evolution, and genome rearrangement and gene deletion might happen to offset gene dosage effect induced by the polyploidization events during evolution.
Examination on acetylated proteins in rice suspension cells shows lysine acetylation on nuclear proteins only occupies 17.2%, and acetylation on nonhistone proteins accounts for the majority [101]. In fact, lysine acetylation on nonhistone proteins as the ancestral function rst happened in prokaryote. In bacteria, protein acetylation is widely involved in the regulation of central and secondary metabolism, virulence, transcription, and translation [102]. More than 70% of the acetylated proteins are metabolic enzymes and translation regulators in Escherichia coli [103]. Lysine acetylation can control central metabolic metabolism by directly switching off enzyme activity [104]. Acetylation is involved in tight regulation of acetyl-CoA synthetase activity necessary for maintenance of energy homeostasis in bacteria [102]. So, protein acetylation is an ancient and widespread strategy for protein activity regulation, especially on the aspect of energy metabolism [104], which is directly associated with survival of organisms and rapid environmental adaption. In eukaryote, DNA wraps around the histone octamer to assemble into nucleosome as the basic unit of chromatin, and lysine acetylation acquires novel role to occur at histone tails, leading to modi cation of chromatin structure and regulation of gene expression on epigenetic level. In fact, acetylation on nonhistone and histone can be linked together by acetyl-CoA, which is a hub metabolite and can supply acetyl group for both. Cells sense their metabolic state through monitoring the levels of the key indicator, acetyl-CoA. Disruption of cytosolic acetyl-CoA carboxylase ACC1 leads to elevated level of acetyl-CoA and increased deposition of histone acetylation predominantly at H3K27ac; comprehensive analysis suggests that H3K27ac is an essential link between cytosolic acetyl-CoA level and gene expression in response to the dynamic metabolic environments in plants [105,106].

Subcellular localization of HAT and HDAC proteins
HAT and HDAC proteins not only modify the histone tails to change chromatin status on genome level but also catalyze nonhistone proteins to directly regulate the activities of speci c targets, and the diversity of substrate type results in the divergence of subcellular localization of HAT and HDAC proteins. A proportion of HAT/HDAC proteins are exclusively localized in the nucleus, such as Arabidopsis GCN5, HAM2, HDA19 and rice HDT701 [90,107]. However, distinct proteins have different domain preference even in nucleus. AtHAG1/2 and AtHAM1/2 are rich at the periphery of the nucleolus [108]. AtHD1 is localized in the euchromatic regions and excluded from the nucleolus, different from AtHD2 [109]. Microbial-associated molecular pattern (MAMP)-activated MAP kinase MPK3 directly interacts with and phosphorylates HD2B, leading to relocalization of HD2B from the nucleolus to the nucleoplasm and global genome-wide shifts in the H3K9 acetylation landscape [57]. In present study, the nucleus localization of about a half of DcHATs (3/8) and DcHDACs (9/14) was successfully predicted by Uniprot or Wolf PSORT. Some members have cytoplasm-predominant localization or cytoplasm/nucleus dual localization. Arabidopsis RPD3 type Class II HDACs, HDA5, HDA8, and HDA18 are localized in the cytoplasm [107]. AtHDA15 shuttles from the cytoplasm to the nucleus in response to light [110]. Transient expression in Arabidopsis protoplasts showed the dual localization of OsHAC701, OsHAG702, and OsHAG704 [6]. Here, subcellular localization prediction by Wolf PSORT indicated that most of DcHATs might localize in both the nucleus and cytoplasm, suggesting a potential shuttling process between these compartments. Some members localize in mitochondrial or chloroplast. Arabidopsis SRT2 resides at mitochondrial and participates in ne-tuning energy metabolism [36]. Similarly, DcSRT2 was predicted to have mitochondrial localization. So, HAT and HDAC proteins ful ll diverse roles to realize broad and ne regulation on both genome-wide and speci c-locus levels partially by functional compartmentation.

Roles of HATs and HDACs in developmental regulation and stress response
HATs and HDACs usually function with other cofactors and crosstalk with different pathways. HAT/HDAC protein can act as a subunit of multiprotein complex. Arabidopsis HD2C deacetylase interacts with SWI/SNF chromatin remodeling complex to mediate heat stress response [111]. HAT/HDAC protein together with transcription factor functions as the most general regulatory combination. AP2 recruits the co-repressor TOPLESS and the histone deacetylase HDA19 to regulate ower development by controlling the expression territories of numerous oral organ identity genes [112]. HDA19 interact with HSL1 to directly repress seed maturation gene expression during germination [113]. HDA15 together with PIF3 and PIF1 is involved in repressing chlorophyll biosynthesis and seed germination, respectively [79,80]. HATs/HDACs associated with histone acetylation status crosstalk with phytohormone signal pathways. Ethylene induces histone acetylation in both genome-wide level and a group of ethylene regulated genes [114][115][116]. The key jasmonate (JA) signaling repressor JAZ1 interacts with two ethylene-stabilized factors EIN3/EIL1, recruits HDA6 to co-repress transcriptional activity of EIN3/EIL1 [117]. HATs/HDACs interplay with each other or other epigenetic modi cation. HD2C can interact with HDA6 to regulate gene expression; ABAresponsive genes ABI1/2 with enhanced expression in hda6 or/and hd2c single and double mutants are associated with increased histone H3K9/K14 acetylation and decreased H3K9 dimethylation (H3K9me2) [89]. FVE/MSI4 interacts with HDA6 and the histone demethylase FLD to regulate owering time by repressing FLC through decreasing histone H3K4me3 and H3 acetylation [69]. Histone demethylases LDL1 and LDL2 interact with HDA6 and CCA1/LHY co-repress TOC1 expression, involved in regulating circadian rhythms [55].
HATs/HDACs are involved in plant responses to various abiotic and biotic stresses. Class-I HDAC (HDA19) and class-II HDACs (HDA5/14/15/18) execute different mechanisms underlying salinity stress responses; hda19 single mutant and hda5/14/15/18 plant exhibit more tolerance and sensitivity to salt stress, respectively [63]. Drought stress signi cantly increases the expression of a subset of HATs (OsHAC703, OsHAG703, OsHAF701 and OsHAM701) and global acetylation level in rice [118]. Here we observed that 5-day drought treatment decreased the expression of HAT member DcHAG1 and HDAC members DcHDA14/DcHDA9, indicating that histone acetylation and deacetylation in distinct loci may contribute to the similar stress response. In rice, cold treatment leads to decreased expression of OsHDA704/OsHDA712/OsSRT701 but increased expression of OsHDA702 [119]. In D. catenatum, we found that cold exposure induced the expression of HAT members DcHAG1 and HDAC members DcHDA14, but repressed the expression of HAT members DcHAM1 and HDAC members DcHDT1/2, indicating that different DcHAT/DcHDAC families might take charge of distinct responses to cold stress for netuning regulation. Furthermore, the promoter region of DcHAG1 was found to contain four low-temperature responsive elements, supporting its signi cance in cold response. It has been demonstrated that heat stress at 38 °C for 3 h increases the expression of AtHDA6/7/5/8/14 in Arabidopsis [18]. The hda19-1 mutant is sensitive to high temperature, and displays abnormal cotyledon morphogenesis at 29 °C, compared with normal development at 25 °C [120]. In present study, high temperature treatment caused the dynamic expression changes of 50% of DcHATs and 71% of DcHDACs, which were upregulated at least at one time point. This result suggested that histone acetylation/deacetylation are widely involved in heat stress response for genome-wide programing of downstream gene expression. Defense hormone JA can induce the expression of HDA6 and HDA19 in Arabidopsis [121]. But in D. catenatum, exogenous application of JA repressed the expression of DcHAG2, DcHDT1/2 and DcHDA15 (Additional le 5), indicating JA has different effects on histone acetylation in different species. Additionally, HDA19 participates in the repression of SA-mediated basal defense responses, and loss of HDA19 enhances SA content and the expression of SA accumulation-related genes and PR genes, thereby causing increased resistance to Pseudomonas syringae [122]. In rice, HDT701 negatively regulates innate immunity by modulating histone H4 acetylation of defense-related genes, and overexpression and silencing of HDT701 leads to more susceptibility and resistance to the rice pathogens Magnaporthe oryzae and Xanthomonas oryzae pv oryzae, respectively [123]. Southern Blight caused by necrotrophic pathogen Sclerotium delphinii is a disaster disease of D. catenatum [124], but its infection had no obvious in uence at the transcriptional level of DcHATs/DcHDACs (Additional le 5). It is worth investigating their changes on translational and posttranslational levels. On the other hand, the pathogen may promote disease by manipulating host histone acetylation during infection. The soybean pathogen Phytophthora sojae can produce the cytoplasmic effector PsAvh23, which binds to ADA2 subunit of the HAT complex SAGA to disrupt ADA2-GCN5 module formation and subsequently suppresses the expression of defense genes by decreasing GCN5-mediated H3K9ac levels, and increases plant susceptibility [125]. The sugar beet cyst nematode (Heterodera schachtii) 32E03 effector interacts with histone deacetylase HDT1 and inhibits its activities, regulates rRNA gene dosage, promote cyst nematode parasitism [126]. Therefore, identi cation of HAT/HDAC candidates directly involving plant-pathogen interactions will provide key targets for directed genetic improvement and accelerate crop resistance breeding.

Conclusions
In present study, we identi  Alignments and phylogenetic analyses HAT and HDAC proteins were aligned using Clustalw [127], and phylogenetic analysis was performed using MEGA7 program [128]. The full-length amino acid sequences of HATs and HDACs were used for constructing neighbor-joining (NJ) trees with the following settings: pairwise deletion option for gaps/missing data treatment; p-distance method for Substitution model; and bootstrap test of 1000 replicates for internal branch reliability.

Gene Structure and domain organization analysis
Exon-intron gene structure schematics of HAT and HDAC genes were analyzed by online program Gene Structure Display Server GSDS 2.0 ( http://gsds.cbi.pku.edu.cn/) [129]. The domain organization of HAT and HDAC proteins were analyzed using SMART and Pfam databases.
In silico analysis of transcriptome data For spatio-temporal expression of DcHAT and DcHDAC genes, the raw RNA-seq reads of different tissue and organs in D. catenatum were downloaded from NCBI, including leaf (SRR4431601), root (SRR5722140), Green root tip (SRR4431599), white part of root (SRR4431598), stem (SRR4431600), ower bud (SRR4431603), sepal (SRR4431597), labellum (SRR4431602), pollinia (SRR5722145) and gynostemium (SRR4431596) [97]. For drought stress and recovery experiment in 8-month-old D. catenatum plants [99], irrigation was performed on the 1st day, omitted from the 2nd to the 7th day, and resumed on the 8th day, watering every 2 days at 15:30. Consequently, the raw RNA-seq reads were obtained from the leaves that were harvested  [98]. Reads of all the samples were aligned to Dendrobium reference genome via HISAT package [134]. The mapped reads of each sample were assembled and gene expression levels were calculated by the FPKM method via StringTie [135].

Consent to publish
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this published article and its Additional les. The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests. Phylogenetic analysis and domain organization of HATs in D. catenatum, compared with homologs in Arabidopsis and rice. Full-length HAT amino acid sequences from D. catenatum, Arabidopsis and rice were aligned via ClustalW, and the unrooted neighbor-joining (NJ) trees were constructed using MEGA7 software with the following settings: pairwise deletion option for gaps/missing data treatment; p-distance method for Substitution model; and Bootstrap test of 1000 replicates for internal branch reliability. Four families (CBP, GNAT, TAFII250, and MYST) are clustered here. The HAT proteins from D. catenatum, Arabidopsis and rice are indicated by green triangle, red circle and green square, respectively. Different conserved protein domains are colored as indicated.  Phylogenetic analysis and domain organization of HDACs in D. catenatum, compared with homologs in Arabidopsis and rice. Full-length HDAC amino acid sequences from D. catenatum, Arabidopsis and rice were aligned via ClustalW, and the unrooted neighbor-joining (NJ) trees were constructed using MEGA7 software with the following settings: pairwise deletion option for gaps/missing data treatment; p-distance method for Substitution model; and Bootstrap test of 1000 replicates for internal branch reliability. HDAC family is divided into three families (RPD3/HDA1, SIR2, and HD2) here. The HDAC proteins from D.
catenatum, Arabidopsis and rice are indicated by green triangle, red circle and green square, respectively. Different conserved protein domains are colored as indicated.     Expression of DcHATs and DcHDACs in response to drought stress. Heat map showing expression pattern of DcHATs and DcHDACs in leaves under different drought treatments. The seedlings were watered on the 1st day, dried from the 2nd to the 7th day, and re-watered on the 8th day. Leaves were harvested at different times; DR6, DR8, DR10, and DR15 indicate sampling at 18:30 on the 2nd, 7th, 8th, and 9th days, respectively. Color scale from green end to red end indicating gene expression from low level to high level.

Figure 9
Expression of DcHATs and DcHDACs in response to heat shock. Expression pro les of DcHAT/DcHDAC genes in leaves at 20°C (control) and 35 °C (heat shock) were detected by RT-qPCR. DcACTIN was used as an internal reference. Experiments were repeated three times. The error bars indicate SD and the asterisks indicate statistical signi cance between the heat shock (35℃) and the control (20℃) based on Student's t test (*, P<0.05; **, P<0.01).

Figure 10
Distributions of HATs and HDACs in some representative species. Red digit in HAG column indicating the gene number have been modi ed based on more stringent screening criteria in this study. The photos on right showing D. catenatum grown on the cliff, arti cial rock wall, and pine trunk, respectively.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.