Genome-Wide Identification, Evolutionary Patterns, and Expression Analysis of bZIP Gene Family in Olive (Olea europaea L.)

Olive (Olea europaea.L) is an economically important oleaginous crop and its fruit cold-pressed oil is used for edible oil all over the world. The basic region-leucine zipper (bZIP) family is one of the largest transcription factors families among eukaryotic organisms; its members play vital roles in environmental signaling, stress response, plant growth, seed maturation, and fruit development. However, a comprehensive report on the bZIP gene family in olive is lacking. In this study, 103 OebZIP genes from the olive genome were identified and divided into 12 subfamilies according to their genetic relationship with 78 bZIPs of A. thaliana. Most OebZIP genes are clustered in the subgroup that has a similar gene structure and conserved motif distribution. According to the characteristics of the leucine zipper region, the dimerization characteristics of 103 OebZIP proteins were predicted. Gene duplication analyses revealed that 22 OebZIP genes were involved in the expansion of the bZIP family. To evaluate the expression patterns of OebZIP genes, RNA-seq data available in public databases were analyzed. The highly expressed OebZIP genes and several lipid synthesis genes (LPGs) in fruits of two varieties with different oil contents during the fast oil accumulation stage were examined via qRT-PCR. By comparing the dynamic changes of oil accumulation, OebZIP1, OebZIP7, OebZIP22, and OebZIP99 were shown to have a close relationship with fruit development and lipid synthesis. Additionally, some OebZIP had a significant positive correlation with various LPG genes. This study gives insights into the structural features, evolutionary patterns, and expression analysis, laying a foundation to further reveal the function of the 103 OebZIP genes in olive.


Introduction
The olive (Olea europaea L.) is a socioeconomically important oleaginous crop that is planted widely in the Mediterranean basin, symbolizing peace, health, and longevity. The olive is the only member of the Oleaceae family with edible fruit. Olive oil is rich in unsaturated fatty acids and other important secondary metabolites. It also contains more than 30 different phenolic compounds [1] that are strong antioxidants and free radical scavengers [2]. In fact, olive oil has great benefits for human health [3], especially for cardiovascular disease, obesity, and diabetes [4]. It has also been recognized as having potential value in reducing the incidence of Alzheimer's [5]. With the development of China's economy, the market demand for olive oil is increasing dramatically; nevertheless, the domestic planting area is small and the production is low; therefore, consumption depends on imports. As such, it is necessary for us to reduce the cost, increase the planting yield, and improve the quality of olive oil. accumulated mostly between August and September, and reaches a plateau in November. However, due to environmental conditions, different agricultural habits, and olive varieties, the manner of lipid accumulation may be different [40]. Therefore, we preliminarily evaluate the expressive relationship with lipid synthesis genes (LPGs) in two varieties. We analyzed the expressive relation between LPGs and OebZIP, combined with the dynamic diversification of oil content, predicted the upstream cis-element of LPGs that contains A-boxes, G-Boxes, and C-Boxes, and performed a correlation analysis. Our study was the first to perform genome-wide identification and an expression analysis of bZIP genes. The goal is to provide a foundation to further study the functions of OebZIP on lipid synthesis.

Identification and Filtration
To retrieve all members of the bZIP gene family, we used the HMMER v3.1b2 (http://hmmer. janelia.org/) Hidden Markov model (PF00170) as a probe and to screen all the candidate proteins with E values of less than 1e-10 in order to build a new, olive-specific Hidden Markov model for the secondary search. Then, the secondary search results of the protein sequences were further confirmed using SMART (http://smart.embl-heidelberg.de/) [48] and Pfam (http://pfam.xfam.org/searc) [49] to ensure their reliability. Finally, all putative bZIP gene models were utilized to analyze the amino acid length, PI, MV by ProtParam (https://web.expasy.org/protparam/) and named by their chromosomal location.

Phylogenetic Analysis and Classification
In order to detect the classification and evolutionary profile of all OebZIP, the amino acid sequences of the Arabidopsis and olive were determined by ClustalW. The tree was constructed using MEGAX with the Maximum Likelihood (ML) method: A phylogeny test was performed with the bootstrap method of 1000 replicates, substitution with the Poisson model, and gaps Data treatment with pairwise deletion.

Gene Structure and Conserved Motif Analysis
We used the online tool Gene Structure Display Server (GSDS v2.0) (http://gsds.cbi.pku.edu.cn/ index.php) to analyze the structure of the OebZIP gene containing the constituents of the exons/introns. Next, we employed MEME v 5.0.2 (http://meme-suite.org/tools/meme) to explore the conservative motifs as well as the bZIP domain, with the following optimized parameters: the maximum number of motifs was set to 10; the width of each motif was 6-50 residues.

Chromosomal Location, Gene Duplication, and Evolutionary Analysis
MG2C v2.0 (http://mg2c.iask.in/mg2c_v2.0/) was used to locate the OebZIP gene on the chromosomes of Olea europaea.L. Firstly, according to multiple sequence alignments, we got a number of a pairwise genes that had similar lengths and sequences by Circos (http://circos.ca/). Afterward, the MCScanX was used to find duplication genes in the olive genome and to analyze collinearity between olive and six other species (Arabidopsis, rice, sesame, soybean, cacao, and grape) [50].

Dimerization Properties of OebZIP Proteins
The leucine zipper region of OebZIP proteins domain, which might be related to the dimerization stability and specificity, was partitioned into different boundaries of N and C terminal [51]. To describe and speculate on the dimerization foundation of 103 OebZIP members, we set seven special amino acid residues that covered four amino acid residues before the appearance of the first leucine in the bZIP domain and the following two amino acid residues, as an initial heptad named L 0 . Next, the amino acid sequences were manually arranged to the end or the tenth heptad, named L 9 . The nomenclature of the position within each heptad was g, a, b, c, d, e, and f.

Plant RNA Extraction and qRT-PCR
Total RNA was extracted using a kit (Tiangen, Beijing, China), the reverse-transcribed process was performed with a PrimeScript RT reagent kit (TaKaRa, Dalian, China), and primers were designed by Primer5. The 'Arbequina' and 'Grossa', due to the enormous difference in their oil contents, were chosen for the qRT-PCR experiment to explore the expression quantity of OebZIPs in leaves and fruits at fruit development stages. Then, LPG genes were texted. β-Actin was used as an internal reference gene, and the expression data were calculated by 2-(∆∆CT) method [52].

Identification and Naming of OebZIP
Based on a HMMER search analysis, we manually screened the candidates in two databases and got 103 bZIP genes in the olive genome; we named these genes according to their distribution on the chromosomes (OebZIP1~OebZIP103, Table S1). The coding sequence length, the isoelectric points (pIs), and molecular weights (MWs) of these genes were then analyzed, as shown in Table S1. The lengths of the all 103 OebZIP proteins were between 96 aa and 772 aa, with an average of 315 aa. The MWs of the proteins ranged from 11.22 (OebZIP99) to 84.31 (OebZIP6), with an average of 34.96. The isoelectric point (pIs) was between 4.85 (OebZIP97) and 10.99 (OebZIP96), with an average of 7.20.

Phylogenetic Analysis and Classification
To classify the subgroups of OebZIPs and explore the evolutionary relationship between the Arabidopsis and olive, we constructed an unroot phylogenetic tree based on the Maximum Likelihood (ML) method with protein sequences of 75 AtbZIP genes and the 103 OebZIP genes ( Figure 1). The 75 AtbZIP genes were divided into 12 subgroups, with 103 OebZIPs being distributed in each. Similar to Arabidopsis, most of the OebZIP genes belong to subgroups S and A, with each containing 25 and 19 OebZIP genes, respectively. Interestingly, this was also observed in rice, sesame, cacao, and olive, but not grape ( Figure S1). Subgroup J and K were the smallest classifications, with each including 1 OebZIP gene. The bZIP genes from olive and Arabidopsis are marked in red and black, respectively. Plant bZIP members were divided into 12 subgroups (A~K, S). Different color blocks represent different subgroups. The bootstrap was set to1000.

Gene Structure and Motif Composition of Olive bZIP
As shown in Figure 2, the intron/exon structure was detected using the Gene Structure Display Server (http://gsds.cbi.pku.edu.cn/index.php/). The number of exons of OebZIP genes varied from 1 to 22, of which 22 (21.4%) of the 103 OebZIP genes had only one exon, and most of them were fastened to subgroup S. Subgroups C, D, and G had multiple exons: 5 genes had a number of exons ranging from 6 to 10, 14 had between 6 to 12 exons, and 7 had between 8 to 22. MEME 7.0 was used to detect the conserved motifs in 103 OebZIP proteins, and 10 motifs containing the bZIP domain were identified. As shown in Figure 2, most OebZIP proteins contained several motifs, while OebZIP72, OebZIP94, OebZIP66, OebZIP103, OebZIP19, and OebZIP95 only contain motif 1, which means that only the basic region has no obvious Leu zipper structure. OebZIP35 only contains motif 4, which only has a Leu zipper region and lacks the Basic region. Motif 8 is highly conserved and only exists in subgroup G, indicating that it is a conserved protein domain which may perform a special function. In subgroup A, except for OebZIP4, OebZIP24, and OebZIP99, all the others contained motifs 7 and 10. The distribution of the protein sequence was motif 7-motif 10-motif 7. OebZIP in Subgroup D contained the most motifs among the 13 subgroups, and was arranged in the order of motif 1-motif 9-motif 6-motif 2-motif 5-motif 3; among these, motifs 9, 2, 5 and 3 are unique in subgroup D. In addition, the evolutionary tree clustering results are consistent with the intron/exon distribution and motif distribution. This indicated that most of OebZIP genes in the same subgroup showed low structural diversity, close evolutionary relationships and high degrees of conservation.

Chromosomal Locations and Gene Collinearity Analysis of OebZIP
Sixty-seven OebZIP gene locations were mapped on the chromosome ( Figure 3). Thirty-six OebZIP locations were not distributed on the chromosome but rather, on scaffolds, and no OebZIP genes were distributed on chr6 and chr19. In some specific regions of chr3 and chr16, OebZIP genes were densely distributed. Gene replication is considered to be one of the main drivers of the evolution of the genome and genetic systems. Segment repeats and tandem repeats are the two main reasons for the expansion of plant gene families. In the olive genome, there were four pairs of tandem repeat events involved in 8 OebZIP genes; OebZIP7 and OebZIP8, and OebZIP52 and OebZIP53 were distributed on chr3 and chr16, respectively, while OebZIP74 and OebZIP75, and OebZIP92 and OebZIP93 were not distributed on the chromosome. As such, they were not represented graphically in Figure 4. Twenty-seven OebZIP genes were involved in 14 fragment repeats. A pair of genes that formed one gene duplication event was found to come from the same subgroup. As OebZIP7 and OebZIP8 are tandem genes, they were both assigned to subgroup F. Thus, 33.98% of the OebZIP genes were shown to be involved in gene duplication, which may be why so many OebZIP genes are caused by gene duplication. To explore the evolutionary relationship of bZIP genes in different species, we used McScanX to perform a gene collinearity analysis on five dicotyledonous plants: Arabidopsis thaliana, Sesamum indicum, Glycine max, Vitis vinifera L., and Theobroma cacao; and on a monocotyledon: Oryza sativa L. ( Figure 5). Many reference plant bZIP genes had a colinear relationship with some OebZIP genes, i.e., 14 (Arabidopsis), 9 (rice), 38 (grape), 35 (cacao), 81 (soybean) and 51 (sesame). Soybean had the most bZIP genes that were colinear with olive bZIP genes, followed by sesame; rice had the fewest. In addition, 5 OebZIP genes (OebZIP1, OebZIP29, OebZIP46, OebZIP33, OebZIP44) were found to be colinear with five dicotyledons, and except for OebZIP46, the rest were all in subgroup S, indicating that these genes are highly conserved during evolution and play an important role in the amplification of the bZIP gene family in olive.

Prediction of Dimerization Properties
The structural characteristics of the Leu zipper region is L-X 6 -L-X 6 -L. Hydrophobic and electrostatic interactions mediate α-helical oligomerization to form homo-or heterologous leucine zipper structures which are involved in the bZIP protein dimerization before binding to DNA. Studies have demonstrated that every seven amino acid residues of the leucine zipper region are set as a heptad, and that the arrangement order is g, a, b, c, d, e, and f. The possibility, stability, and specificity of the leucine zipper dimerization structure are mainly determined by four sites: a, d, e, and g. The a and d positions are usually hydrophobic amino acids. The two monomer α-helical structures interact with each other through the hydrophobic amino acids at the a and d positions to form a hydrophobic inner core, enabling the dimer to exist stably. Asparagine at the a position can form a polar pocket on the hydrophobic interface. When a is asparagine, it can form a very stable N-N structure (a↔a ), so it tends not to interact with other amino acids, which limits the possibility of interactions between heterologous helices. The d position is generally a highly conserved leucine, which is one of the important factors affecting the dimerization structure. However, the e and g positions that flank the dimerization interface frequently contain charged amino acids including acidic amino acids glutamic acid (E) and aspartic acid (D), and the basic amino acids arginine (R) and lysine (K). As a result, salt bridges are formed to maintain stability between the spirals.
We performed a detailed dimerization analysis to characterize the amino acids present at positions a, d, e, and g of the OebZIP proteins ( Figure S3). The length of the Leu zipper in the OebZIP family varied from two to nine heptads. At the a position, asparagine (Asn/N) accounted for 19%; meanwhile, at position a, the asparagine appeared most frequently in the second heptad(L 2 ), followed by L 4 , accounting for 35.35% and 25.25%, respectively. In addition, at position a, the frequency of hydrophobic amino acids (I, V, L, M) was 35%, and charged amino acids (R, K, E, D) was 20%. At the d position, the frequency of hydrophobic amino acids was 81%, of which 50% was leucine and 31% was other hydrophobic amino acids (I, V, M).
Four types of g↔e interactions in each heptad were then analyzed, including attractive basic-acidic pairs (+/− attractive), attractive acidic-basic pairs (−/+ attractive), repulsive basic pairs, and repulsive acidic pairs ( Figure 6). As shown in Figure 6c, the L 1 heptad had a total of 32 of g↔e pairs, of which 87.88% were attractive pairs. The L 4 and L 5 heptad had 26 and 15 g↔e pairs respectively. The L 6 heptad only had attractive basic-acidic pairs (+/-). The L 7 , L 8 , and L 9 heptad had no complete g↔e pair. Based on the dimerization features, OebZIP proteins were divided into three categories: (I) Those that tend to form homodimerization. OebZIP proteins in this category contained the g↔e pairs and the asparagine in the a position. The members of subgroup A basically belong to this category; (II) Those with both homo-and heterodimerization. Most OebZIP belong to this category, such as those of subgroup C, which contained repulsive g↔e pairs and attractive g↔e pairs, or contain repulsive g↔e pairs. (III) Those that tend to form heterodimerization. OebZIP proteins in this category contained repulsive g↔e pairs only; for example, OebZIP30 contained three adjacent repulsive g↔e pairs, but there were fewer OebZIPs in this category.

Gene Expression Profiles of OebZIPs on Transcription Data
We analyzed the expression profiles based on transcriptome data including different periods and tissues. In July, 92, 95, 97, and 94 OebZIP genes were expressed, respectively, in fruits, pedicels, stems, and leaves; In November, 89, 96, 95, and 93 OebZIP genes were expressed, respectively, in fruits, pedicels, stems, and leaves. Among them, only one OebZIP gene (OebZIP33) was not expressed in four tissues in two periods, while the rest were expressed in at least one tissue. However, the gene expression abundance was significantly different. As shown in Figure S4, expression of most OebZIP genes was tissue-specific. For example, the OebZIP84 gene was expressed in fruits only, while the OebZIP62 gene was expressed in all tissues except for fruits. Moreover, most OebZIP genes were significantly up-or down-regulated in different periods. The highly expressed OebZIP genes in fruits may be involved in fruit oil accumulation, and there were 17 and 16 bZIP genes with higher expression in fruit in July (OebZIP1, 7, 8 ) were found to be highly expressed in both periods; 12 OebZIP genes were upregulated and 15 OebZIP genes were downregulated. Therefore, 15 candidate OebZIP genes that possess higher expression in fruits were selected for the next analysis.

Expression Patterns of OebZIP Genes in Different Tissues
To give an insight into the functional role of the OebZIP gene, we used qRT-PCR to detect the expression profiles of candidate OebZIP genes throughout the development of fruit, and the expression of 15 candidate OebZIP genes in the leaf and fruit of 'Arbequina' were analyzed ( Figure S5). All candidate genes were expressed in both fruits and leaves, but the expression levels were significantly different between different tissues at various development periods. Similar to the result of the transcriptome data analysis, most of the OebZIP genes had high expression in fruits. The OebZIP22, OebZIP53, OebZIP79, OebZIP84, OebZIP91, and OebZIP99 had higher expression abundance in fruits than leaves in all periods (July to November).

Expression of OebZIP Genes in Different Varieties During Fruit Development
We found that different varieties of olive displayed a big difference in the speed of oil accumulation during the fast oil accumulation stages. The fruit development and oil accumulation period of some varieties were significantly earlier than those of others. Therefore, 'Arbequina' ('A') and 'Grossa' ('G') were selected for the real-time PCR, and to explore the candidate OebZIP genes among two varieties. As shown in Figure 7, the expression of candidate genes between the two varieties ('A' and 'G') was quite different. We found that some OebZIP genes (OebZIP1, OebZIP22, OebZIP37, OebZIP52, OebZIP53, OebZIP8, OebZIP85, and OebZIP89) in 'A' were concentrated in the early stages of oil accumulation (July to August). Seven OebZIP genes (OebZIP1, OebZIP7, OebZIP50, OebZIP53, OebZIP59, OebZIP91, and OebZIP99) in variety 'G' were concentrated in the middle stages (September to October) of oil accumulation. Furthermore, the expression levels of five OebZIP genes (OebZIP1, OebZIP7, OebZIP59, OebZIP84, OebZIP85, and OebZIP85) in variety 'G' increased dramatically from July to October, reached a peak in September or October, and then decreased rapidly. This observation is similar to the rate of oil accumulation in 'G', varieties indicating that these OebZIP genes have tight functions in olive fruit development and lipid synthesis. Interestingly, the expression level of the OebZIP22 gene in 'A' in July was the same as that of the OebZIP22 gene in 'G' in August, the expression levels and dynamic changes of the two varieties were again similar, indicating that OebZIP22 gene may play an important role in fruit development, given the different fruit development periods of the two varieties. In addition, although two pairs of tandem genes (OebZIP7 & OebZIP8, OebZIP52, and OebZIP53) and one pair of fragment genes (OebZIP1 and OebZIP37) had sequence similarities, their expression patterns and abundances were disparate, indicating that the repeated genes in olive achieved functional differentiation during their evolution.

Predicting Interaction with Lipid Synthesis Genes
Since the expression of some OebZIP genes was found to be significantly different between the two varieties, these genes might correlate with oil accumulation. To preliminarily investigate this interaction, we referred to previous studies and scanned 44 lipid synthesis genes (LPGs) named according to their chromosomal location from the olive genome (Table S2).

Expression Patterns of Lipid Synthesis Genes
Based on the abundance of genes expression ( Figure S7) and the prediction of upstream cis-acting elements (Figure 8a) OeLPPAT2, OeSACPD1, OeSACPD2, and OeSADCP4) were selected. These basically covered most of the genes involved in the lipid synthesis pathway ( Figure 8b); therefore, these genes were selected for detailed qRT-PCR analysis. The expression patterns between LPGs were significantly different (Figure 9). OeKASII.6 (β-ketoacyl -ACP synthetase II, C16~C18), OeKASIII.1 (β-ketoacyl-ACP synthetase III, C2~C4), OeLPAAT2 (lysophosphatidate acyltransferase) and OeDGAT1 (diacylglycerol acyltransferase) had higher expression abundance than the other LPG genes in both varieties. Moreover, the expression patterns of different members in the gene family were also significantly different. For example, the OeKASII and OeSACPD families had different expression patterns. Some genes (OeFAD2.5, OeSACPD4, and OeKASII.4) maintained an extremely low level of expression in two varieties, indicating that they may be less correlated with the regulation of fruit development and lipid synthesis. In addition, several LPGs had different expression patterns between the two varieties, which might be caused by the difference in the period of oil accumulation, as described above. There were 7 genes in 'A' (OeACC1, OeBCCP1, OeDGAT1, OeFAD2.4, OeFAD2.5, OeLPPAT2, and OeSACPD1) that were significantly expressed in the early stage (July to August). In contrast, in 'G' there were 6 genes (OeACC1, OeBCCP1, OeDGAT1, OeGPAT5, OeKASII.6, OeLPPAT2) that were upregulated progressively throughout the oil accumulation until they reached a peak in the middle stages of oil accumulation (September to October), and then downregulated rapidly. Compared with the expression patterns of OebZIPs and LPGs, the result indicates that some OebZIPs (OebZIP1, OebZIP7, OebZIP91, and OebZIP99) and LPGs (OeACC1, OeDGAT1, OeLPPAT2, and OeKASII.6) showed similar expression patterns in different varieties.

Evolutionary Analysis of the OebZIP Gene Family
Olive is a very important oil crop around the world. With the completion of olive genome sequencing and the rapid development of bioinformatics, the mining and identification of gene families based on genome-wide searches have become more and more effective. Therefore, it is necessary to explore the structural and evolutionary characteristics of OebZIP genes from the level of gene families. Although the bZIP gene family has been studied in many plants, this is not the case for olive. In this study, a total of 103 OebZIP genes were identified and screened based on the olive genome. The average length of OebZIP proteins is 315 aa. With respect to numbers and aa length, compared to other plants, Arabidopsis (78, 321 aa) [38], rice (89, 311 aa) [29], sesame (63, 339 aa) [30], soybean (160, 343 aa) [31], grape (47, 336 aa) [32], and apple (114, 411 aa) [35], the number of bZIP family members in olive is larger and similar to apple, while the coding sequence length of the genes is clearly different. Several studies [53,54] have found that there is no positive correlation between the numbers of bZIP genes and the genome size. The differences in the number of bZIP members in plants might be due to whole-genome duplication (WGD) events [37]. Phylogenetic tree analysis with the Arabidopsis bZIP protein, which was divided into 12 subgroups, showed that the AtbZIP genes have a close kinship with the OebZIP genes. The number of identified OebZIP genes in the subgroups were similar to other plants (Arabidopsis, rice, sesame, soybean, and cocoa) except grape. Additionally, we found that subgroup H contains 2 OebZIP genes, similar to Arabidopsis bZIP. This indicates strong homology, and that subgroup H (OebZIP73, OebZIP88) may play an important role in photomorphogenesis.
Gene replication is considered to be one of the main drivers of genetic system evolution. The two main reasons for the expansion of plant gene families are tandem repeats and segment repeats. Twenty-two olive, 119 soybean, 8 tartary buckwheat, and 28 grape bZIP genes are involved in gene duplication events (Figure 4), accounting for around 21.4% (olive), 86.2% (soybean) [31], 8.3% (tartary buckwheat) [54], 59.6% (grape) [32]. The contribution of gene duplication events to the amplification of gene families was significantly different among different species. In addition, by comparing the quantity of the tandem repeats, i.e., 4 pairs of olive, 1 pair of soybean [31], 2 pairs of tartary buckwheat [54], 0 pairs of grape, we concluded that the contribution of tandem repeats during the evolution of OebZIP genes was limited, which is consistent with reports on grape, soybean, and tartary buckwheat. Therefore, segment gene duplication plays an important role in bZIP gene family expansion in olive. Soybean had the most bZIP genes that were colinear with olive bZIP genes ( Figure 5) and tartary buckwheat bZIP genes [54], which may be due to the large genome of soybean and the large number of GmbZIP genes.
Most OebZIP genes in the same subgroup possess the same features (Figure 2), as also reported for apple [35], melon [55], and cassava [33], i.e., the OebZIP genes in each S subgroup have one exon and the motif distribution is simple, just containing motif 1 (basic area) and motif 4 (Leu area). Some specific motifs (motif7, motif10) only exist in a specific subgroup (subgroup A). Meanwhile, the order of motif arrangement in the same subgroup is consistent. Furthermore, motif 8 was found to be quite conserved and focused on subgroup G, indicating this motif might have some specific function.
The function of bZIP proteins requires dimerization. We investigated the dimerization patterns of OebZIP proteins, which are characterized mainly by the four amino acids presenting at the a, d, g, and e positions (Figure 6). At the d position, the frequency of Leu aa accounts for 50%, which is close to AtbZIPs (56%), but significantly less than in OsbZIPs (71%) [29], ZmbZIPs (70%) [56], and GmbZIPs (68%) [31]. The frequency of Asn at position a was 19%, which was lower than that found in AtbZIP (40%) [51] and GmbZIP (25%) [31]. Another difference is that in the olive, the frequency of Asn at the a position was the highest in the L 2 heptad, followed by the L 4 heptad, while in [31] Arabidopsis [51] and strawberry [34], the frequency of Asn in the a position was highest in the L 2 heptad, followed by the L 5 heptad. Therefore, it was shown that the bZIPs in olive are more prone to heterodimerization, and the length of Leu zipper is shorter compared to soybean [31], Arabidopsis [51], and strawberry [34].

The Expression Pattern of OebZIP was Related to Fruit Development and Lipid Synthesis
Plants have a large number of bZIP genes and diverse functions, but investigations of bZIP functions have been mainly focused on some model plants [38]. Many study already found that the bZIP gene family plays a role in all stages of plant growth and response to biotic/abiotic stresses [18][19][20][24][25][26]28,[57][58][59][60][61][62][63]. However, very few bZIP genes involved in fruit development and lipid synthesis have been identified in plants, especially in olive. We explored the possibility of the involvement of candidate OebZIP genes in regulating lipid synthesis and gene expression levels of LPG genes during fruit development. In a previous study, the oil content of different olive varieties was significantly differentiated in the fast oil accumulation stage ( Figure S6). Several LPGs have different expression patterns between the two varieties; this result is similar to that of a study on the gene expression levels of OeFAD2 [64], OeFAD3 [65], and OeSACPD [66] during fruit development between two olive varieties ('Picual' and 'Arbequina'). The expression levels of some LPGs, such as OeACC1, OeDGAT1, OeKASII.6, and OeLPAAT2, were consistent with the oil accumulation trend. The oil in 'A' accumulated mainly in early stage fruit development, while the oil 'G' accumulated mainly in the late stage. The gene expression pattern of OeDGAT1 in 'G' was consistent with a previous report of OeDGAT1 in 'Koroneiki' [67], presenting a typical normal curve, i.e., following maximal mRNA levels in September and then declining substantially. Although most TAGs in olive accumulated before they matured, it has been demonstrated that some oil cultivars, like 'A' in our study, have much shorter oil-filling periods than 'Koroneiki' and 'G' [68][69][70] (Figure S6). Olive oil is known for its high oleic acid content. Several researchers have reported that the composition of olive oil fatty acids, particularly oleic acid, fluctuated according to the variety [68,71,72]. The variation of OeFAD2 [64,73], OeFAD3 [65], and OeSACPD [66,74] gene expression is the main factor that affects the composition of fatty acids [37]. The gene expression levels of OeSACPD1 and -2 in 'A' in our study were similar to those in a previous report in the same variety, peaking in the early stages (July to August) and then decreasing [66]. Moreover, the expression level of OeFAD2.2, which is considered to be mainly responsible for the linoleic acid content in olive oil, as reported by Hernandez [64] and Bruno [73], was higher than that of OeFAD2.5 in our study, indicating that OeFAD2.5 may not be a significant factor affecting the oil content of mesocarp.
The goal of this study was to obtain more insight into the expression patterns and functions of OebZIP genes during fruit development and lipid synthesis. An expression profile analysis indicated that OebZIP genes expression showed significant variation among different tissues, which implied that they might have diverse functions. It is therefore hypothesized that highly expressed genes in fruits are closely related to fruit development and lipid synthesis. In previous studies, some candidate OebZIPs expressed in fruits showed different expressive patterns in the two varieties. In addition, paired gene duplication events showed different expression patterns, suggesting that functional differences had occurred during evolution, as similarly demonstrated in previous reports on rice and grape.
Only four genes, i.e., OebZIP1, OebZIP7, OebZIP91, and OebZIP99, were significantly upregulated in the fast oil accumulation stage between the two varieties. In Arabidopsis, AtbZIP53 was reported to be involved in regulating seed development, and seed maturation was affected when AtbZIP53 [28] was overexpressed or knocked out. In olive, OebZIP1 was homologous to AtbZIP genes, and OebZIP1 was clustered into the same group with AtbZIP53 in the phylogenetic tree ( Figure 1). Therefore, we speculate that the OebZIP1 may also regulate olive fruit development and oil accumulation. In addition, OebZIP22 was found to have a similar expression pattern between the two varieties, but the expression of OebZIP22 in 'A' took precedence over that of 'G'. OebZIP22 is a homolog of TGA3 (AtbZIP22). In Arabidopsis, TGA3 was involved in mediating hormonal cross-talk between salicylic acid and cytokinin [75]. Therefore, OebZIP22 may be involved in hormone regulation in fruit development and lipid synthesis. In addition, through a correlation analysis, we found that the expression patterns of some OebZIP genes in fruits of different varieties were significantly correlated with oil synthesis genes ( Figure 10). For example, there was a significant positive correlation between the expression of OebZIP1, OebZIP37, OebZIP 89, and OeFAD2.4. In addition, there were two a-boxes upstream of the OeFAD2.4 promoter region, and the possible interactions between OeFAD2.4 and OebZIP1, 37, and 89 transcription factors might be the focus of future research. The functions of these OebZIP genes need to be verified by experiments in the future.

Conclusions
In this study, for the first time, we identified 103 bZIP genes based on the olive genome, which were divided into 12 subgroups. We analyzed the genetic structure, conserved domain distribution, the dimerization feature, the evolutionary relationship and the gene replication events. Meanwhile, some important candidates of OebZIP genes for olive fruit development and lipid synthesis were identified. This study laid the foundation for the study of the structure and function of the bZIP gene family in olive. It provides important clues for increasing the oil content of olive from the direction of bZIP TF regulation of lipid synthesis.
Supplementary Materials: Supplementary Materials can be found at http://www.mdpi.com/2073-4425/11/5/510/s1, Figure S1: Classification of bZIP genes in six plants, Figure S2: A detailed motif introduction for all OebZIP proteins, Figure S3: Amino acid sequences of the Leu zipper region of OebZIP proteins, Figure S4: Heatmap of 103 OebZIP genes expressed among four tissues based on transcription data, Figure S5: Relative expression of OebZIP genes in fruit and leaf, Figure S6: Dried fruit oil content during the fast oil accumulation stages (July to November), Figure S7: Heatmap of 44 LPG genes expressed among four tissues based on transcription data. Table  S1: The basic characteristics of olive bZIP gene family, Table S2: The chromosomal location of olive LPG genes.