Somatic embryos of P. asperata were obtained from the embryogenic cell line 1931, which was initiated from an immature zygotic embryo in the National Spruce Germplasm Bank of China (2003DKA21003). According to Xia et al. (2016), the embryos were matured on semi-solid modified Litvay medium (Litvay et al., 1985) supplemented with 0.1% (w/v) casein acid hydrolysate (Sigma, St. Louis, MO, USA), 3% (w/v) sucrose (Beijing Chemical Works, Beijing, PR China), 5% (w/v) polyethylene glycol 4000 (Merck, Darmstadt, Germany), and 0.5% (w/v) gellan gum (Sigma). Prior to autoclaving (121°C, 20 min), the pH was adjusted to 5.8 ± 0.01. At the end of this procedure, filter-sterilized (0.2-μm filter, Pall Corp., Port Washington, NY, USA) L-glutamine (Sigma) and (±)-abscisic acid (Gibco, Grand Island, NY, USA) were added into the autoclaved and cooled medium to the final concentration of 0.05% and 60 μM, respectively. The embryo maturation cultures were maintained in darkness at 24 ± 1°C. After 2 months of culture, mature embryos with well-developed cotyledons were selected and transferred onto two layers dry sterile filter paper (Whatman™ 2, Kent, UK) to undergo PDT under a 16-h photoperiod with a low light density of 15 μmol m⁻² s⁻¹ (LED fluorescent tubes) at 24 ± 1°C for 0 (D0), 7 (D7), 14 (D14), or 21 (D21) days. Finally, the desiccated embryos were collected, and immediately frozen in liquid nitrogen, then stored at −80°C until use.

Somatic embryogenesis of conifers, which uses plant cell engineering to generate plantlets, is one of the most important biotechnologies. To date, somatic embryogenesis has been achieved in a variety of coniferous species (Klimaszewska et al., 2016). However, the low germination rate of mature somatic embryos is a major limitation that prevents the application of somatic embryogenesis for plant propagation on a large scale. In the present study, PDT was found to promote germination and enhanced the conversion of P. asperata somatic embryos (Figure 1). Furthermore, we investigated the global acetylome of the desiccated embryos, to provide a better understanding of its physiological functions and the stress tolerance mechanisms.

To acquire a general view of LysAc in the embryos during PDT (D0, D7, D14, and D21), multiple major proteins bands with the molecular weights similar to histones and non-histones were successfully examined through western blotting using an anti-acetyllysine antibody (Figure 2A). Figure 2A shows that the level of LysAc was higher at D14. For example, the signals of 70–250 KD bands were enhanced upon anti-acetyllsine antibody detection in D14 radicles compared with D0, D7, and D21 radicles. On this basis, to obtain global insight into the large-scale dataset of LysAc sites in the desiccated embryos, we used antibody-based affinity enrichment and high-resolution MS to identify and analyze acetylated proteins in the embryos from D14.

The lengths of most identified peptides were distributed between 8 and 17 amino acids, which was consistent with tryptic peptide properties and MS identification (Supplementary Figure S1). In total, we successfully identified 1360 unique acetylation sites in 717 acetylated proteins in the desiccated embryos. The MS proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaino et al., 2016) partner repository with the dataset identifier PXD005042 (Username: Reviewer31026@ebi.ac.uk; Password: V62RrjWZ). By searching these acetylated peptides identified against the previous proteome dataset of the embryos during PDT, we validated 1079 unique acetylation sites in 556 acetylated proteins (Figure 2B and Supplementary Table S1). Previously, plant lysine acetylome studies identified 74 acetylated proteins in A. thaliana (Finkemeier et al., 2011), 31 in S. tuberosum (Xing and Poirier, 2012), 97 in V. vinifera (Melo-Braga et al., 2012), 121 in G. max (Smith-Hammond et al., 2014b), 358 in P. sativum (Smith-Hammond et al., 2014a), 684 in F. ananassa (Fang et al., 2015), 227 in T. aestivum (Zhang et al., 2016), and 716 in O. sativa (Xiong et al., 2016). These different numbers are mainly due to the different intrinsic acetylation levels of the proteins among the various species. Here, our data extend the inventory of lysine-acetylated proteins in plants, providing information to reveal the role of LysAc in conifer embryos as well as their development.

According to the validated 1079 LysAc sites in the 556 acetylated proteins, the average degree of acetylation was 1.9 sites per protein, and the number of LysAc sites in each protein was ranged from 1 to 9 (Supplementary Figure S2). Among these proteins, 57% contained only one acetylation site, and the percentages of proteins with two, three, four, and five or more modification sites were 20, 10, 7, and 6%, respectively. For the acetylome, the mean number of modified sites per 100 amino acids was 0.78, with the highest value being 6.67 (Supplementary Table S2).

To determine the evolutionary conservation of the detected LysAc sites, we searched the 1079 acetylated peptides in P. asperata against the previously reported acetylomes of some other species: F. ananassa (Fang et al., 2015), O. sativa (He et al., 2016; Xiong et al., 2016), and A. thaliana (Finkemeier et al., 2011; Wu et al., 2011). The consistency analysis demonstrated that 65 peptides from 51 proteins had been identified in at least two plants (Supplementary Table S3). Among these, 33 conserved proteins were involved in metabolic processes, indicating key functions of LysAc in metabolic regulation. Notably, seven metabolism-related proteins were conservatively acetylated at two sites among various plants species, including chalcone synthase (CHS, MA_10426264g0020), large subunit ribosomal protein L3e (RP-L3e, MA_10432758g0010), adenosylhomocysteinase (ahcY, MA_10433454g0020), small subunit ribosomal protein SAe (RP-SAe, MA_112104g0010), large subunit ribosomal protein L19e (RP-L19e, MA_1839g0010), elongation factor 1-alpha (EEF1A, MA_434977g0010), and phosphoglycerate kinase (PGK, MA_92421g0010) (Supplementary Table S3).

The preferred amino acid residues surrounding LysAc sites have been identified in plants (Finkemeier et al., 2011; Melo-Braga et al., 2012; Nallamilli et al., 2014; Smith-Hammond et al., 2014a,b; Fang et al., 2015; He et al., 2016; Xiong et al., 2016; Zhang et al., 2016). To explore the conserved LysAc motifs of the desiccated somatic embryos of P. asperata, we used the Motif-X program to search for conserved sequence motifs in all of the validated 1079 acetylated-lysine sites. A total of 529 defined unique sites, accounting for 49% of the validated sites, were matched to seven conserved motifs (Figure 3 and Supplementary Table S1). Inspection of these seven motifs demonstrated that two types of amino acid residues surround acetylated sites: The first is phenylalanine (F), tyrosine (Y), lysine (K), or valine (V) upstream of the LysAc site, and the second is Y, F, or histidine (H) downstream of LysAc sites (Figure 3B). F and Y are two of the most conserved amino acid residues were detected both upstream and downstream of LysAc sites. F was located at the ±2 positions of the sites, while Y was detected at the ±1 positions. By contrast, the above two conserved residues (F and Y) in our dataset were the same as those in rice. In germinating rice seeds, F and Y are present at a high frequency around acetylated lysines (He et al., 2016). Xiong et al. (2016) found that F was an essential amino acid and was the most conserved residue surrounding the ±2 positions of LysAc sites in rice seedlings. Zhang et al. (2016) noted that H is preferred at the +1 position of LysAc sites, this is in accordance with our result. In addition, a residue with hydrophobic side chain group (V) was enriched at the −2 position. The above results suggested that these conserved motifs, which occur at the ±1 or ±2 positions of LysAc sites, are important for lysine acetylation in plants. Furthermore, a positively charged residue (K) was also found to be enriched at the −10 position in our study. These conserved amino acid residues surrounding LysAc sites suggest that residues with aromatic ring, positive charge or hydrophobic side chain might be functionally important for acetylation to occur.

To better understand the function of the lysine acetylome in desiccated embryos, the lysine-acetylated proteins were subjected to GO classification based on the corresponding biological processes, cellular components, and molecular functions (Figure 4 and Supplementary Table S4). In the functional classification of biological processes, the acetylated proteins of the desiccated embryos were involved in metabolic processes, cellular processes, and single-organism processes, accounting for 39, 27, and 21% of all lysine-acetylated proteins, respectively (Figure 4A). Similarly, previous studies have indicated that proteins involved in many metabolic and cellular processes can be acetylated at lysine sites. In addition, this reversible lysine acetylation is crucial for the regulation of biological processes in A. thaliana (Wu et al., 2011), F. ananassa (Fang et al., 2015), and O. sativa (He et al., 2016; Xiong et al., 2016). These lysine-acetylated proteins validated in the present study could be important for conifer metabolism and cellular regulation. Furthermore, the percentage of acetylated proteins in the desiccated embryos in the response to stimulus category was slightly higher than that in the non-desiccated embryos (Supplementary Figure S3A), implying an important regulatory role of lysine acetylation in the response of embryos to desiccation stress.

For the cellular component ontology, the acetylated proteins were mainly classified into the cell (38%), macromolecular complex (27%), organelle (25%), and membrane (10%) categories (Figure 4B). It has been reported that the main cellular components of rice are cells, organelles, membranes, and macromolecular complexes, accounting for 29, 23, 17, and 11% of all acetylated proteins, respectively (He et al., 2016). In strawberry leaves, the top three cellular components are organelles (55.66%), macromolecular complexes (24.53%), and membranes (18.92%) (Fang et al., 2015). Based on these data, we speculate that the main classifications of cellular components are essentially the same among acetylated plant proteins.

In the ontology of molecular functions, the analysis demonstrated that the acetylated proteins were associated with binding (42%), catalytic activity (41%), and structural molecular activity (11%) (Figure 4C). The percentage of acetylated proteins in the catalytic activity category was much higher in desiccated embryos than in non-desiccated embryos (Supplementary Figure S3C). It is well known that the catalytic activity of enzymes involves the binding of their substrates to form an enzyme-substrate complex (Cooper, 2000). Previous reports revealed that reversible lysine acetylation in the binding domain regulates metabolism to respond to different stressors in bacteria (Mo et al., 2015; Liu et al., 2016). In the present, we infer that the majority of the validated acetylated proteins, which are transcription factors or enzyme-related proteins, ensured that the embryos responded promptly to desiccation stress.

To reveal the subcellular localization of acetylation in the desiccated embryos, we conducted subcellular location prediction (Figure 4D and Supplementary Table S4). The prediction indicated that the majority of the validated acetylated proteins were distributed in diverse subcellular locations with 40% in cytosol, suggesting that LysAc in this compartment plays an important role in the embryos. Furthermore, a number of acetylated proteins were also localized to the chloroplasts (28%), nuclei (15%), and mitochondria (7%). Due to the different materials examined, our data differed slightly from other plant systems such as strawberry, rice, and Arabidopsis. Fang et al. (2015) demonstrated that the main subcellular components of strawberry leaves are chloroplasts, cytoplasmic components and nuclei, accounting for 48, 28, and 7% of all acetylated proteins, respectively. In whole rice plants, the main subcellular components are chloroplasts (42%), cytoplasm (29%), nuclei (12%), and mitochondria (8%) (Xiong et al., 2016). The subcellular components of Arabidopsis leaves mainly correspond to the intracellular (23.1%), cytoplasm (15.7%), and chloroplasts (14.4%) categories (Wu et al., 2011). Notably, the percentage of acetylated proteins in the chloroplasts was higher in the desiccated embryos than in the non-desiccated embryos (Supplementary Figure S3D), which was directly associated with the low-light conditions, indicating that the embryos carried out photosynthesis under light in response to nutrient-deficit stress. In addition, 1% of the acetylated proteins in the desiccated embryos were localized to the peroxisome, where fatty acids are activated to acyl-CoA and then converted to sucrose (Rylott et al., 2006).

Furthermore, we carried out GO enrichment analysis to determine which functional terms were targeted for LysAc in the desiccated embryos (Figure 5A). In the analysis of cellular components, ribosome and ribonucleoprotein complexes were significantly enriched. Accordingly, the structural constituents of the ribosome were significantly enriched in the molecular function category. The detailed enrichment results are presented in Supplementary Table S5. Notably, oxidoreductase activities were significantly enriched in the desiccated embryos compared with those in the non-desiccated embryos (Supplementary Figure S4A), reflecting a crucial role of LysAc in the response to desiccation stress. Our data also demonstrated significant enrichment of validated proteins in various biological processes, including the oxoacid metabolic processes, organic acid metabolic processes, and carboxylic acid metabolic processes, indicating the pivotal role of LysAc in virtually all fundamental metabolic processes.

Furthermore, to search for enriched protein domains in the acetylated proteins of P. asperata, we applied protein domain enrichment analysis to the acetylome data. In total, 25 protein domains were significantly enriched (Figure 5B and Supplementary Table S5). The validated proteins with these enriched domains are involved in various metabolic pathways with important cellular functions. Specifically, the NAD(P)-binding domain was significantly enriched in the desiccated embryos compared with the non-desiccated embryos (Supplementary Figure S4B). A previous study suggested that enzymes containing an NAD(P)-binding domain are involved in catalyzing redox reactions (Hua et al., 2014). This further supports the hypothesis that LysAc plays a key role in the response to desiccation stress.

On the first day of PDT, the embryos were subjected to severe water and nutrient deficiency, leading to a series of physiological changes (Jing et al., 2016). The initial slow desiccation is thought to be the signal that triggers events that establish desiccation tolerance, embryo regeneration, and subsequent germination (Lipavská and Konrádová, 2004). In the subsequent days of PDT, with the gradually increased water content, the embryos exhibited shoot and root growth (Figure 1B). Additionally, somatic embryos after PDT germinated effectively and vigorously (Figure 1C).

Previous studies on plants have indicated that LysAc plays important roles in many cellular processes, including responses to changes in environmental conditions, shoot and root development, and hormone signaling (Chen and Tian, 2007; Servet et al., 2010). In the desiccated embryos, we detected three acetylated proteins that were located in the peroxisome (Supplementary Table S4), which is associated with the reduction of reactive oxygen species, such as hydrogen peroxide (Bonekamp et al., 2009) and the degradation of fatty acids (Rylott et al., 2006). Two pivotal enzymes, catalase (MA_10437148g0010 and MA_10427891g0010) and malate synthase (MA_120902g0010), were found to be modified at multiple sites by LysAc in the desiccated embryos (Supplementary Table S1) compared with the non-desiccated embryos. Additionally, the analysis of biological processes indicated that in the desiccated embryos, 25 acetylated proteins were enriched in response to stimulus-related GO terms (Supplementary Table S4). Some peroxidases, such as glutathione peroxidase (MA_125131g0010), cationic peroxidase SPC4 (MA_8813326g0010), and probable L-ascorbate peroxidase 8 (MA_119996g0010) (Supplementary Table S1), were not detected in non-desiccated embryos.

To reveal the preferential metabolic pathways associated with LysAc in the desiccated embryos, we performed KEGG pathway enrichment analysis. Cluster analysis showed that 251 acetylated candidate proteins revealed involvement in 22 enriched pathways (Figure 6, Supplementary Table S5 and Supplementary Package 1), among which carbon metabolism was significantly enriched. It is well known that glycolysis/gluconeogenesis and the tricarboxylic acid cycle are the main pathways of glycometabolism and denote the diverse biochemical processes of various carbohydrates in living cells (Plaxton, 1996; Mo et al., 2015; He et al., 2016). It has also been reported that acetylation of glycometabolism-related proteins (mainly enzymes) may be a crucial regulatory mechanism in processes such as plant development, growth, and stress tolerance (Fang et al., 2015; He et al., 2016). In our study, a large proportion of the enzymes in these two pathways were detected as being acetylated both in desiccated and non-desiccated embryos (Supplementary Figure S5), which facilitates the elucidation of the mechanism of LysAc-regulated glycometabolism. We infer that the lysine-acetylated residues of key enzymes are a prerequisite for regulating the glycometabolic pathway in living organisms.

In addition, we identified a significantly enriched pathway of pentose phosphate in the desiccated embryos (Figure 7A and Supplementary Table S5). The pentose phosphate pathway is considered to be a major source of cellular reducing power and plays an essential protective role against oxidative stress (Kletzien et al., 1994; Pandolfi et al., 1995). Glucose-6-phosphate dehydrogenase (MA_10267514g0010) catalyzes the first step in the pentose phosphate pathway and provides reductive potential in the form of NADPH (Slekar et al., 1996). The production of NADPH in turn supports antioxidant enzyme activity (Lehmann et al., 2009). In the pentose phosphate pathway, we infer that LysAc plays an important role in protecting the biomolecules of the desiccated embryos from oxidative damage.

Fatty acid metabolism was enriched in desiccated embryos compared with that in non-desiccated embryos (Figure 7B and Supplementary Table S5). Fatty acid degradation mainly proceeds through β-oxidation, which provides respiratory substrates under conditions of carbohydrate starvation (Dieuaide et al., 1992), produces hydrogen peroxide that may act as a signal in various stress responses in plants (Eastmond and Graham, 2000), and supplies metabolic energy and carbon skeletons for germination (Rylott et al., 2006). These functions are essential in the PDT process, not only in the response to water deficiency but also in supporting growth until photosynthetic competence is achieved.

According to the clustering analysis of subcellular locations in the desiccated embryos, we conducted functional enrichment to characterize the preferential target substrates of LysAc in different organelles (Figure 8). The response to stress was significantly enriched in mitochondria according to the biological processes category (Figure 8A). This enrichment indicated that during PDT, stress-related proteins were the preferential target substrates of LysAc in mitochondria. Accordingly, the molecular function category demonstrated that stress-related GO classifications, oxidoreductase, nutrient reservoir, and antioxidant activities, were significantly enriched (Figure 8C). In the nucleus, many chromatin organization-, protein complex-, protein-DNA complex-, and DNA conformation change-related classifications were significantly enriched (Figure 8A). Accordingly, many DNA, protein complex, protein-DNA complex, and chromatin classification in the cellular component category were significantly enriched in the nucleus (Figure 8B). The enrichment of nucleic acid, protein and DNA binding, and protein dimerization and heterodimerization activity was also significant in the nucleus in the molecular functions category (Figure 8C). In chloroplasts, the enzyme catalytic complex was significantly enriched (Figure 8B), suggesting that LysAc mainly occurs in catalytic metabolism-related proteins in this organelle. Consistent with this enrichment, several enzymes-related factors such as NADP, NAD, coenzymes, and cofactor binding, were significantly enriched in the chloroplasts, based on the classification of molecular functions (Figure 8C).

These differential clustering classifications in different organelles may contribute to stress and the intrinsic subcellular location of LysAc. First, during PDT, somatic embryos of P. asperata suffer from a water-deficit, nutrient-deficit, and oxidative stress. Thus, we looked for the acetylated proteins in response to stress, and the clustering analysis revealed that the stress-related GO classifications, including oxidoreductase activity, hydrolase activity, nutrient reservoir activity, and antioxidant activity, were significantly enriched. Furthermore, chromatin organization- and DNA conformation-related categories were significantly enriched in the nucleus. Previous studies have shown that the LysAc of histones is involved in chromatin structure, histone-DNA, and histone-histone interactions (Strahl and Allis, 2000; Chen and Tian, 2007; Hirschey et al., 2011).

To better understand the cross-linked pathways of these acetylated proteins, we established a PPI network among the acetylated proteins of the desiccated embryos based on information from the STRING database (Figure 9 and Supplementary Table S6). An overview network of 257 acetylated proteins was constructed to represent the diverse functions in the desiccated embryos. Among these networks, four highly-connected clusters, including the ribosome, proteasome, spliceosome, and metabolites of carbon metabolism clusters, were extracted using the MCODE algorithm (Supplementary Figure S6). Most of these interacting proteins were associated with the ribosome, proteasome, and spliceosome clusters, which is in agreement with the analysis of subcellular location that was associated with the synthesis and degradation of stress-related proteins. In addition, many interacting acetylated proteins were also classified as part of carbon metabolism, suggesting that LysAc on glycometabolism-related proteins plays an important role for embryo development.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer JK and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.