Genome-wide identification of the Liriodendron chinense WRKY gene family and its diverse roles in response to multiple abiotic stress

Background Liriodendron chinense (Lchi) is a tree species within the Magnoliaceae family and is considered a basal angiosperm. The too low or high temperature or soil drought will restrict its growth as the adverse environmental conditions, thus improving L. chinense abiotic tolerance was the key issues to study. WRKYs are a major family of plant transcription factors known to often be involved in biotic and abiotic stress responses. So far, it is still largely unknown if and how the LchiWRKY gene family is tied to regulating L. chinense stress responses. Therefore, studying the involvement of the WRKY gene family in abiotic stress regulation in L. chinense could be very informative in showing how this tree deals with such stressful conditions. Results In this research, we performed a genome-wide analysis of the Liriodendron chinense (Lchi) WRKY gene family, studying their classification relationships, gene structure, chromosomal locations, gene duplication, cis-element, and response to abiotic stress. The 44 members of the LchiWRKY gene family contain a significant amount of sequence diversity, with their lengths ranging from 525 bp to 40,981 bp. Using classification analysis, we divided the 44 LchiWRKY genes into three phylogenetic groups (I, II, II), with group II then being further divided into five subgroups (IIa, IIb, IIc, IId, IIe). Comparative phylogenetic analysis including the WRKY families from 17 plant species suggested that LchiWRKYs are closely related to the Magnolia Cinnamomum kanehirae WRKY family, and has fewer family members than higher plants. We found the LchiWRKYs to be evenly distributed across 15 chromosomes, with their duplication events suggesting that tandem duplication may have played a major role in LchiWRKY gene expansion model. A Ka/Ks analysis indicated that they mainly underwent purifying selection and distributed in the group IId. Motif analysis showed that LchiWRKYs contained 20 motifs, and different phylogenetic groups contained conserved motif. Gene ontology (GO) analysis showed that LchiWRKYs were mainly enriched in two categories, i.e., biological process and molecular function. Two group IIc members (LchiWRKY10 and LchiWRKY37) contain unique WRKY element sequence variants (WRKYGKK and WRKYGKS). Gene structure analysis showed that most LchiWRKYs possess 3 exons and two different types of introns: the R- and V-type which are both contained within the WRKY domain (WD). Additional promoter cis-element analysis indicated that 12 cis-elements that play different functions in environmental adaptability occur across all LchiWRKY groups. Heat, cold, and drought stress mainly induced the expression of group II and I LchiWRKYs, some of which had undergone gene duplication during evolution, and more than half of which had three exons. LchiWRKY33 mainly responded to cold stress and LchiWRKY25 mainly responded to heat stress, and LchiWRKY18 mainly responded to drought stress, which was almost 4-fold highly expressed, while 5 LchiWRKYs (LchiWRKY5, LchiWRKY23, LchiWRKY14, LchiWRKY27, and LchiWRKY36) responded equally three stresses with more than 6-fold expression. Subcellular localization analysis showed that all LchiWRKYs were localized in the nucleus, and subcellular localization experiments of LchiWRKY18 and 36 also showed that these two transcription factors were expressed in the nucleus. Conclusions This study shows that in Liriodendron chinense, several WRKY genes like LchiWRKY33, LchiWRKY25, and LchiWRKY18, respond to cold or heat or drought stress, suggesting that they may indeed play a role in regulating the tree’s response to such conditions. This information will prove a pivotal role in directing further studies on the function of the LchiWRKY gene family in abiotic stress response and provides a theoretical basis for popularizing afforestation in different regions of China. Supplementary Information The online version contains supplementary material available at 10.1186/s12870-021-03371-1.


Background
Transcription factors (TFs) are important proteins that bind to specific DNA motifs to regulate gene expression and play crucial roles in plant growth, development, metabolism, and stress response [1,2]. WRKY transcription factors were first identified in plants and form one of their largest gene families [3]. After the first WRKY gene was cloned in sweet potato, the WRKY gene family function has been studied in several model plants, such as Arabidopsis thaliana (72), Oryza sativa (76), Amborella trichopoda (31), and Solanum lycopersicum (81) [4,5]. The main defining feature of WRKY proteins is their WRKY domain, which contains a conserved amino acid sequence WRKYGQK at the N-terminus and a C2H2 or C2HC zinc motif at the C-terminus of about 60 amino acids in length [1,6]. Both the heptapeptide sequence and zinc-finger motif are required for the high binding affinity of WRKY TFs to the consensus W-box (TTG ACC /T) cis-elements in their target genes, which regulates target gene expression [7,8]. Based on the number of WRKY domains and the structure of their zinc-finger motifs, WRKY proteins can be classified into three main groups: groups I, II, and III. Group II is specified by WRKY proteins containing a single WRKY domain with a C2H2 motif and can be further divided into five subgroups: IIa, IIb, IIc, IId, and IIe. Group I contains WRKY proteins with two WRKY domains including a C2H2 motif. The remaining WRKY proteins, containing a WRKY domain along with a C2HC motif, belong to group III [9,10].
The evolution of the WRKY gene family has been studied in a large number of reports, but a clear evolutionary history has so far not yet been defined. This has resulted in an active debate on the origin of individual WRKY genes and gene groups in different plant species [3,11]. According to a large phylogenetic tree using higher plants, the WRKY gene family can be classified into four clades, including groups I + IIc, groups IIa + IIb, group IId, and group IIe [3,12]. Previous reports have suggested two alternative hypotheses for the evolution of the WRKY family: The "Group I Hypothesis" and the "IIa + b Separate Hypothesis". The "Group I Hypothesis" postulates that the C-terminal WRKY domains of group I genes are the most ancient and that all other WRKY genes are derived from these, whereas the "IIa + b Separate Hypothesis" suggests that groups IIa and IIb evolved directly from a single domain algal gene, independent from the group I-derived lineage [3,13,14].
In addition, WRKY transcription factors are involved in a plant's stress response to herbivores, pathogens, and nematodes [15,16]. Substantial evidence has shown that the WRKY TFs play significant roles in the signaling and regulation of gene expression during biotic and abiotic stress responses [17,18]. For example, In Camelina (Camelina sativa (L.) Crantz), CsWRKY21 was highly expressed under cold stress and CsWRKY22 was highly expressed under drought stress [17]. In the Grape (Vitis vinifera L.), VvWRKY11 is involved in the response to dehydration stress, and VvWRKY24 expression is induced by cold stress [18]. In Cucumber (Cucumis sativus L.), five CsWRKYs responded strongly to salt stress and hightemperature stress [4]. In Wheat (Triticum aestivum), the expression of TaWRKY75-A was highly induced under salt stress [19]. In Ipomoea trifida (H.B.K.) G. Don, 11 ItfWRKYs were highly expressed under four abiotic stresses: cold, heat, salt, and drought [16]. In Arabidopsis, AtWRKY39 positively regulates the cooperation between the SA and JA activated signaling pathways that mediate the response to heat stress, while AtWRKY8 is highly expressed in plant roots and significantly upregulated under salt stress [20,21].
Liriodendron, a basal angiosperm genus, resides within the Magnoliaceae family and has diverged into two distinct species: the East Asian (L. chinense (Hemsley) Sargent) and the North-East American (L. tulipifera Linn) across all LchiWRKY groups. Heat, cold, and drought stress mainly induced the expression of group II and I LchiWRKYs, some of which had undergone gene duplication during evolution, and more than half of which had three exons. LchiWRKY33 mainly responded to cold stress and LchiWRKY25 mainly responded to heat stress, and LchiWRKY18 mainly responded to drought stress, which was almost 4-fold highly expressed, while 5 LchiWRKYs (LchiWRKY5, LchiWRKY23, LchiWRKY14, LchiWRKY27, and LchiWRKY36) responded equally three stresses with more than 6-fold expression. Subcellular localization analysis showed that all LchiWRKYs were localized in the nucleus, and subcellular localization experiments of LchiWRKY18 and 36 also showed that these two transcription factors were expressed in the nucleus. [22,23]. L. chinense is mainly distributed in the south of the Yangtze River, which has further diverged into two separate lineages, the eastern and western groups in China [23]. The flowering mutant gene super long blooming 1(slb1) of L. chinense was identified by transcriptome analysis [24]. There are few reports on abiotic stress research, such as auxin efflux carrier PIN-FORMED (PIN) [25], gibberellin oxidase (GAox) [26], myeloblastosis (MYB) [27], and C-regeneration binding factors (CBF) gene family [28]. Liriodendron hybrids are intraspecific hybrids crossed by L. chinense and L. tulipfera, which are valuable ornamental tree species in both private and public spaces because of their fast growth, drought tolerance, and strong resistance to diseases and insect pests [29,30]. Under aluminum (Al) stress, glutamate decarboxylase (GAD) was up-regulated in Liriodendron hybrids, and enhanced the Al tolerance [31]. Adverse temperature and soil conditions can inhibit the growth of L. chinense, such as cold, heat, and drought stress. Knowledge is still lacking on how L. chinense deals with abiotic stress at a molecular level. WRKY family is involved in plant growth and development and abiotic stress response and has been reported in many species with available genome sequences. The recently published L. chinense genome now provides ample resources with which to carry out bioinformatics-based identification and analysis of WRKY TFs.
In this study, we identified 44 LchiWRKY genes and classified them into three main groups. A comprehensive analysis, including determination of exon-intron structure, motif, gene ontology, gene duplication events, chromosome distribution, subcellular localization, and comparative phylogenetic analysis was performed. Furthermore, global transcriptome expression analysis was performed to identify the gene expression pattern of specific WRKY gene family members under different abiotic stress conditions in L. chinense. In addition, Quantitative real-time PCR (qRT-PCR) was used to verify the expression levels of LchiWRKY genes with different expression patterns. Subcellular localization experiment verified the expression position of LchiWRKY transcription factors. This study provides valuable clues for the functional characterization of the WRKY gene family in L. chinense.
This research reveals the functional roles of the L. chinenese WRKY genes under abiotic stress and provides a valuable theoretical basis for increasing forest yield and promoting afforestation of Liriodendron.

Results
The L. chinense genome contains a small number of WRKY genes that vary greatly in length Since WRKY genes have previously been demonstrated to be involved in abiotic stress response, we set out to characterize the L. chinense WRKY gene family. To gain an understanding of the gene diversity present within this family, we determined the basic physicochemical properties of the WRKY protein products, including sequence length, molecular weight (MW), isoelectric point (pI) and other indexes for all 44 identified LchiWRKY (Additional file 1: Table S1). The size of the LchiWRKY proteins varied dramatically, with LchiWRKY37 being the shortest protein at 143 amino acids (aa) in length, whereas LchiWRKY42, the longest, contains 787 aa. Their MW ranges from 16.274 to 90.094 kDa, and the pI from 4.95 (LchiWRKY41) to 9.89 (LchiWRKY36). These data suggest that different LchiWRKY proteins might operate in various microenvironments. The values of the grand average of hydropathicity were negative for all LchiWRKY proteins, indicating that they are all hydrophilic proteins. Almost all LchiWRKYs were defined as unstable proteins, and only 3 LchiWRKYs (LchiWRKY35, LchiWRKY37, LchiWRKY40) with an instability index of less than 40 were considered to be stable proteins.
The number of WRKY gene family in L. chinense is small, and their physical and chemical characteristics vary widely, so it is necessary to explore their taxonomic and phylogenetic characteristics.

The LchiWRKY gene family consists of three distinct phylogenetic groups
We constructed a phylogenetic tree that would indicate the LchiWRKY protein family's evolutionary interrelatedness. We based our phylogenetic tree on a multiple sequence alignment of the WRKY full length of each protein. We found that the 44 LchiWRKYs are divided into three separate phylogenetic groups: Group I, Group II, and Group III [13], out of which Group II contains five subgroups, being Group IIa, Group IIb, Group IIc, Group IId, and Group IIe ( Fig. 1 and Additional file 2: Table S2 [23]. A total of 1308 WRKY proteins were used to construct a Bayesian phylogenetic tree as described in the Methods section ( Fig. 2 and Additional file 3: Table S3).
All 1308 WRKYs clustered across major clades of the classification group. We frequently found WRKY members from all 17 species that belong to one specific group (I, II, III) to be clustering to the same clade. We found LchiWRKY protein sequences to be more closely related to their homologs from C. kanehirae than those from monocotyledons and dicotyledons. This finding is consistent with the evolutionary status of L. chinense being closer to A. trichopoda, another basal angiosperm, than model plants A. thaliana and O. sativa, two more recently evolved dicotyledonous and monocotyledons species [1].

The LchiWRKY family was at a relatively primitive evolutionary status and had fewer family members
Interestingly, the total number of WRKY families in the majority of monocotyledons and dicotyledons is more than that in L. chinense (44) and A. trichopoda (31) ( Table 1 and Additional file 4: Table S4), for example, among monocotyledons, S.polyrhiza has the least number of WRKY families (44), and among dicotyledons, V. vinifera has the least number of WRKY families (60), but most of them were more than that of L. chinense. For Fig. 1 Phylogenetic tree of LchiWRKYs. A total of 44 LchiWRKY proteins were aligned using MUSCLE (v3.8.31), and a phylogenetic tree was constructed using BEAST (v2.6.6) set to the Bayesian method with 10,000,000 MCMC chain length. Each color indicates an individual closely related group of WRKY proteins (I-III) within the phylogenetic tree, Posterior probability limit was 1.0 and the posterior value was shown on the branch. The scale bar indicates the branch length of the phylogenetic tree. The posterior value below 0.5 indicates a low homology among the different LchiWRKY protein sequences Magnolia plants, two whole-genome duplication (WGD) events occurred during the evolution of C. kanehirae [32], and only one WGD event occurred in L. chinense [23], which may lead to a greater number of WRKY genes in C. kanehirae (73) compared to L. chinense (44). As the three subgroups (I, II, II) of plants of different evolutionary statuses are clustered into the same branch, so it is speculated that the WRKY family is relatively conservative in the process of evolution. We infer that different plant subgroups have different numbers due to differential expansion speeds and finally lead to different total family numbers, as a result, the WRKY family number of some basal angiosperms and magnolias expanded slower than monocotyledons and dicotyledons [14].
Then we used the OrthoFinder (v2.2.3) to explore the WRKY evolutionary status of all the 17 species ( Fig. 3 and Additional file 4: Table S4), and the phylogenetic analysis showed that the evolutionary status of WRKY of L. chinense was closer to the Magnolia C. kanehirae, and the evolutionary status of both species is relatively lower than that of higher angiosperms, including monocotyledons and dicotyledons, while higher than basal angiosperms A. trichopoda and N.colorata (Fig. 3). These results suggested that L. chinense might experience a slower expansion model during evolution, resulting in a smaller number of family members than other higher plants. So it is necessary to explore the expansion pattern of the LchiWRKY gene family. To understand how the identified LchiWRKY genes are distributed over the L. chinense chromosomal landscape, we mapped their sequences onto its 19 chromosomes [23] (Fig. 4 and Additional files 5 and 6: Tables S5-S6). Considering the high diversity with the L. chinense WRKY gene family, we asked how then the different genes would be interrelated. Using L. chinense genome annotation data, we identified a total of 44 LchiWRKYs, which were renamed from LchiWRKY1 to LchiWRKY40 based on their chromosomal positioning. An additional four genes (LchiWRKY41, LchiWRKY42, LchiWRKY43, and LchiWRKY44) that were mapped to yet unassembled scaffolds were named LchiWRKY41 ~ LchiWRKY 44 respectively.
We found 40 LchiWRKYs to be unevenly distributed across 15 out of the 19 L. chinense chromosomes, with chromosomes 4, 7, 12, and 19 devoid of any WRKY gene. However, the remaining 4 genes mapped onto four separate scaffolds were not yet assembled into full chromosomes. Overall, the number of WRKY genes per chromosome varied from 0 to 5, with no apparent correlation between chromosomal length and the number of WRKY genes present (Fig. 4A).
Our previous study mapped several gene duplication events that occurred in the evolutionary past of L. chinense [23]. Such events could have also affected the number of WRKY genes in the L. chinense genome, as well as their distribution over the genome [3]. Duplication events are known to occur at a regional scale via tandem duplication, copying one gene's sequence in the direct vicinity of the original (< 200 kb), but may also occur at a larger scale via segmental duplication, copying whole sections of chromosomes [3,13]. The former is often caused by DNA replication errors, whereas the latter may result from polyploidy events followed by chromosomal rearrangements [3,33].
We could find a total of 4 WRKY gene pairs on the L. chinense genome. Out of these, no gene resulted from a segmental duplication event (Fig. 4A). The 4 events (LchiWRKY27-LchiWRKY15, LchiWRKY27-LchiWRKY14, LchiWRKY27-LchiWRKY20, Lchi-WRKY23-LchiWRKY35) have all likely resulted from tandem duplication, leading to 6 additional WRKY genes (LchiWRKY14, 15,20,23,27, and 35) in the L. chinense genome, which distributed on Chr8, 11,14,15, and 16. Two of them were in group I (LchiWRKY23) and IIa (LchiWRKY35), the others were in group IId (Fig. 4B). These results indicate that a few gene duplication events like group IId have indeed contributed to the expansion of the LchiWRKY gene family in the evolutionary past.
To gain an understanding of whether or not the duplicated LchiWRKY genes contribute to organism fitness, we measured the Ka/Ks nucleotide substitution ratios of 4 individual gene pairs to study the exerted selective pressure (Table 2).
Generally, a Ka/Ks ratio > 1 is consistent with positive selection, while Ka/Ks < 1 indicates a purifying or negative selection [34]. We found their Ka/Ks ratios all to be < 1, indicating that the duplicated LchiWRKYs have undergone a synonymous substitution or purifying selection with limited functional divergence during their evolutionary history. The analysis of gene replication events and selection pressure showed that the WRKY family expanded mainly through tandem duplications, but they were subjected to purifying selection in the process of evolution, so it is necessary to further explore the structural characteristics and functions of LchiWRKYs.
We found a single motif (motif 1, a WRKY motif ) to be widely distributed among all members of the LchiWRKY family, while all others were specific to one or a few phylogenetic groups we defined previously (Fig. 5). For example, motif 6 (a WRKY motif ) and motifs 7, 9, and 19 are only present in Group I members (LchiWRKY38, 25, 21, 22, 8, 7, 33, and 23), motif 10 and 11 are specific to Groups IIa (LchiWRKY34, 35, and 24) and IIb 5,28,1,11,and 2), motif 12 to Group IIb, motif 5 to Group IIc (LchiWRKY17, 19, 32, 9, 10, 6, 4, 44 and 43) and I and motif 16, 17 and 20 to Group IId 36,27,20,15 and 14). Remarkably, Groups IIe (LchiWRKY18, 29, 41 and 30) and III (LchiWRKY40, 12, 39, 13, 3, 31, 37 and 42) did not contain any conserved sequence motifs, suggesting an increased sequence The picture on the right panel was the picture corresponding to each species, and the name of the family of each species was annotated on the corresponding branch of the evolutionary tree diversification of these groups (Fig. 5). According to most reports, the functions of some motifs remain to be elucidated. Overall, the closely related LchiWRKYs in the phylogenetic clades shared similar motif compositions, suggesting that the LchiWRKYs within the same phylogenetic group may have similar functions.
To further understand the function of the identified 44 WRKY genes, we performed gene ontology (GO) annotation analysis (Additional file 8: Fig. S1). The results showed that a total of 34 genes were significantly enriched in different GO terms, mainly including 2 categories and 10 subcategories. In the molecular function category, 34 genes were significantly enriched in 3 subcategories, including 'Sequence−specific DNA binding' (GO: 0043565), 'DNA binding transcription factor activity' (GO: 0003700) and 'Transcription regulator activity' (GO: 0140110). In the biological process category, they were mainly enriched in 7 subcategories, including 'Regulation of transcription, DNA-templated (GO: 0006355), 'Regulation of nucleic acid−templated transcription' (GO: 1903506), 'Regulation of RNA biosynthetic process' (GO: 2001141), 'Regulation of RNA metabolic process' (GO: 0051252), 'Regulation of biosynthetic process' (GO: 0009889), 'Regulation of macromolecule biosynthetic process' (GO: 0010556) and 'Regulation of cellular biosynthetic process' (GO: 0031326). Molecular functional enrichment is more significant than others, including three terms: DNA binding transcription factor activity, sequence-specific DNA binding, and transcription regulator activity, indicating that these three terms may be the main functions of WRKY genes. The three molecular functional subcategories are more representative than the subcategories of biological processes.

Diversity of gene structure and two main promoters cis-element regulator analysis
We next examined gene structure by studying the exonintron organization and cis-element in all identified Lchi-WRKY genes, to gain more insight into the evolution of the WRKY family in L. chinense. Firstly, we could observe a large variation in gene size of different LchiWRKYs, ranging from 525 bp (LchiWRKY37) to 40,981 bp (Lchi-WRKY25) (Additional file 9: Table S8 and Fig. 6).  This suggested a large variation in intron length since we had seen previously that the different LchiWRKY CDS sequences were of a much more similar size. The exon number varied between 2 and 6 exons for all LchiWRKY genes, with the majority of genes (27 / 61.36%) having a total of 3 exons (Additional file 9: Table S8 and Fig. 6). Two types of introns exist in the conserved WDs. One of the introns is spliced exactly at the R residue, We classified this phase-2 intron as an R-type intron. The other is localized before the V residue, which is at the sixth amino acid after the second C residue in the C2H2 motif in the zinc finger region. This phase-0 intron is designated as a V-type intron. We could identify two major types of introns within the conserved WRKY domains of the LchiWRKY genes (Additional file 10: Fig. S2), which are similar to those conserved in Arabidopsis [13]. Interestingly, phylogenetic groups IIa and IIb possess a V-type, or phase-0 intron, while all other groups (I, IIc-e, and III) contain an R-type, or phase-2 intron (excepting the Lchi-WRKY42 gene).
The WRKY domain sequence within the L. chinense WRKY gene family shows a high degree of conservation (Additional file 11: Table S9), with only two WRKY sequence variants (WRKYGKK and WRKYGKS) present in the LchiWRKY10 and LchiWRKY37 proteins (both in subgroup II-c) respectively. Six proteins, classified in groups II-c and III, lack a C-terminal zinc-motif. Group I contains 8 LchiWRKYs, with each possessing two WDs and C2H2-type zinc finger motifs (C-X4-C-X22-H-X1-H) (Additional file 11: Table S9). Group II is comprised of 29 proteins, each containing a single WD and C2H2type zinc finger structure (C-X4-5-C-X23-H-X1-H) (Additional file 11: Table S9). We further divided Group II into five subgroups according to their clusters in the phylogenetic analyses, including IIa, IIb, IIc, IId, and IIe with 3, 6, 10, 6, and 4 members, respectively. Finally, 7 LchiWRKYs with a single WD were assigned to Group III, because of their unique C2HC zinc-finger structure (C-X7-C-X23-H-X1-C).
We then used the PlantCARE website to analyze 1.5 kb of the 5′ upstream promoter regions of the LchiWRKY genes to identify any conserved cis-acting elements [4]. Various types of cis-acting elements were found and all LchiWRKY genes contained several cis-acting elements in their promoter regions. The 12 most common elements are summarized in Fig. 7 (Additional file 12: Table S10). There are two types of cis-elements. One is related to growth and development, including WRKY TF binding element (W-box) and four light-responsive elements (Sp1, Box 4, G-box, and GT1 motif ). The other is related to stress response, including abscisic acid-responsiveness (ABRE), MeJA-responsiveness (CGTCA and TGACG motifs), anaerobic element (ARE), drought stressresponsive element (MBS), defense and stress-responsive Exon-intron structure of LchiWRKYs. An unrooted phylogenetic tree constructed using Beast (v2.6.6) set to the Bayesian method (left), green boxes indicate exons; blue boxes indicate untranslated 5′-and 3′-regions; black lines indicate introns, scale = bp indicates sequence length, each group was labeled with a differently colored branch element (TC-rich), low-temperature stress responsive element (LTR). Several of the cis-elements we identified showed a specific distribution across the different L. chinense WRKY gene groups. For example, group II-a lacks any LTR elements, suggesting that this group may not be sensitive to low-temperature stress. The Sp1 element can be found in all groups except IId and IIe, showing that the genes in these two groups might not be as responsive to light. Moreover, the TC-rich element could only be detected in groups IIa, IIb, and IIc, indicating that these three groups might be mainly involved in defense and stress response. Finally, the MBS element is not present in group III genes, implying that these genes may not be induced under drought stress. In summary, the L. chinense WRKY gene groups seem to possess their own sets of cis-responsive elements, indicating that the different groups may have specialized towards responding to specific sets of stimuli.
Analysis of gene structure, motif and cis-regulatory elements showed that there were two WRKY structural variants and intron insertion structures, 2 GO categories and 12 cis-elements in the WRKY family, which regulated growth and development and abiotic stress response. Therefore, the expression profile of WRKY in growth and abiotic stress needs to be further studied.

The LchiWRKY group II responds strongly to multiple abiotic stresses
Environmental stress can affect a plant's health and growth, as well as influence the regulation of crucial molecular processes. Under adverse conditions, a large number of stress-related genes are induced to help plants deal with stress. Therefore, to understand how L. chinense deals with abiotic stress and see whether WRKY genes play a role in stress response, as well as their regulatory patterns, we used leaf tissue transcriptome data from 7 subsequent time points (0 h, 1 h, 3 h, 6 h, 12 h, 1d, and 3d) to explore the expression of LchiWRKY genes in response to cold, heat and drought stress (Figs. 8, 9 and 10).
A total of 11 genes were induced by cold stress (Fig. 8 and Additional file 13: Table S11), their expression patterns were clustered into two groups, including persistent Cis-acting elements of LchiWRKYs. An unrooted phylogenetic tree is shown, constructed using BEAST (v2.6.6) set to the Bayesian method and 11 cis-acting elements on the 1.5 kb of LchiWRKYs promoter sequences. The differently colored boxes represent different cis-acting elements, each group was labeled with a differently colored branch low expression and high expression at different time points. LchiWRKY5, 23,14,27,36,19 showed peak expression at the 1d and 3d time points, while downregulated at the other time points. Only two genes showed long-lasting and continuous up-regulation after cold stress, with LchiWRKY5 being up-regulated from 3 h to 3d and LchiWRKY33 from 3 h to 1d, LchiWRKY14 was highly expressed at almost all time points. These findings may suggest that LchiWRKY5 and LchiWRKY14 mainly promote L. chinense in response to low-temperature stress. LchiWRKY family may play a major regulatory role at 1d and 3d response time points.
The response of LchiWRKY to high-temperature stress was more obvious than that to low-temperature stress, ( Fig. 9 and Additional file 14: Table S12). A total of 11 genes were induced by heat stress, and the expression patterns were clustered into two groups, including low expression patterns at almost all time points and high expression patterns at continuous time points (3 h ~ 3d).
LchiWRKY7 were lowly expressed in all-time points, most of them were differentially expressed in all-time points. LchiWRKY5, 14,23,27,36 were up-regulated significantly at the 1d and 3d time points, similar to how they responded to cold stress. LchiWRKY44, 18 reached its expression peak at 6 h and 12 h time points, while LchiWRKY33, 25 were up-regulated at the 1d and 3d time points, which were different from cold stress. Compared with cold stress, LchiWRKY18, 25 was specifically high expressed from 1d to 3d under heat stress. Under heat stress, we speculate that 1d and 3d may be important response time points.
Under drought stress ( Fig. 10 and Additional file 15: Table S13), a total of 13 LchiWRKY family members responded, and their expression patterns were divided into two groups, one group was a high expression from 6 h to 1 day, the other group was a low expression at all time points. LchiWRKY14 was highly expressed at all time points, It seems that duplication events such as tandem repeat events in the gene structure and promoter region of WRKY family members caused WRKYs to acquire diverse functions [36,37]. The expression of LchiWRKY27, 18,5,33,23,36 peaked at 12 h and 1d, and LchiWRKY16, 44,7,37,25,19 were low expressed at almost all time points. We speculate that 12 h and 1d may be the main response time points, which is different from low and high-temperature stress and may be caused by different biological pathways. LchiWRKY14, 27,18,5,33,23, 36 may be the main response genes to drought stress.
We found that some genes responded to low temperature, high temperature, and drought stress at the same time, for example, LchiWRKY5, LchiWRKY23, Lchi-WRKY14, LchiWRKY27, LchiWRKY36 were up-regulated at the 1d time points in three cases, suggesting that these five genes are the main responders to abiotic stress and play across regulatory role in abiotic stress resistance, and 1d may be important time points for LchiWRKY to respond to abiotic stress.

The qRT-PCR validation of LchiWRKY under multiple abiotic stresses
To verify the accuracy of abiotic stress transcriptome data, we performed qRT-PCR experiments of 10 genes with different expression levels, including high or low expression, LchiWRKY7 and 44 were low expression (0 < Log2FC < 3) in the three stresses and other genes (LchiWRKY5, 18, 19, 25, 27, 33, and 36) were high expression (Log2FC > 3) at different time points of the three stresses [38] (Figs. 8, 9 and 10).
Under cold stress ( Fig. 11 and Additional file 16: Table S14), LchiWRKY7, 36, 19 were up-regulated and reached the peak at 1d, while LchiWRKY23 was up-regulated and reached the peak at the 12 h time point, only LchiWRKY5 were up-regulated at all time points. And other genes' patterns like LchiWRKY25, 44,18,27,33 were down-regulated at all time points. As a result, these genes' expression patterns as the same as transcriptome data, and 1d was the main response time point. Under heat stress ( Fig. 12 and Additional file 17: Table S15), the response time was earlier than that of low-temperature stress. LchiWRKY19 was up-regulated at 3 h and reached the peak, then LchiWRKY7, 25 were up-regulated and reached the peak at 12 h. The peak expression of LchiWRKY36, 5, 33, 23 were at 1 d, while LchiWRKY44, 27,18 were low expressed at all time points. The above results showed that the expression pattern of high-temperature stress was different from that of low-temperature stress, but there was a common important response time point 1d, which indicated that this time point was the key time point for Liriodendron to respond to temperature stress.
Under drought stress ( Fig. 13 and Additional file 18: Table S16), the expression of most genes was relatively low than 0 h, except for LchiWRKY5,23, which had a significant peak value and were highly expressed at 12 h and 1 d, respectively. The response of LchiWRKY18, 19, 33 to drought stress was earlier, peaked at 1 h, and then began to decline. LchiWRKY25, 7,44,36,27 were all low expressed after 0 h, which may be due to the inhibition of drought stress. Compared with temperature stress, the expression levels of some genes were relatively low under drought stress. For instance, LchiWRKY7, 36 was highly expressed under both low and high-temperature stress, but low expressed under drought stress. Several genes had peak expression under three stresses, such as the expression of LchiWRKY5, 23 peaked at 12 h and 1d of three stresses, respectively.
The expression pattern of transcriptome and qRT-PCR experiment of LchiWRKY gene under abiotic stress fully showed that the WRKY family of L. chinense responded to abiotic stress, which provided the theoretical basis for the adaptability promotion of L. chinense in different regions.

LchiWRKY18 and 36 were expressed in the nucleus
To understand the biological function of WRKY proteins, we performed subcellular localization analysis (Additional file 1: Table S1). The results showed that 44 LchiWRKY genes were expressed in the nucleus [4,39,40], while only LchiWRKY42 was also expressed in the cell membrane [39,41]. These results suggest that most LchiWRKY transcription factors were expressed in the nucleus, but may also function in other organelles, such as cell membrane. Therefore, we speculate that transcription factors mainly perform biological functions in the nucleus and regulate the expression of target genes in combination with the promoter sequence of target genes. There may also be some other situations, for example, some transcription factors (LchiWRKY42) may also have biological functions in the cell membrane.
To verify the results of subcellular localization analysis, we selected two genes LchiWRKY18 and Lchi-WRKY36 for the subcellular localization experiment.
The fusion vectors pJIT166-LchiWRKY18-GFP and pJIT166-LchiWRKY36-GFP were constructed and transformed into callus protoplasts of Liriodendron hybrids by PEG-6000 mediated. The subcellular localization expression vector pJIT166-GFP was constructed as a positive control. The fluorescence of LchiWRKY18-GFP (35S::LchiWRKY18::GFP) fusion protein was preferentially detected in the nucleus, while the spontaneous fluorescence of the positive control pJIT166-GFP (35S::GFP) fusion protein could be detected in the whole-cell space (Fig. 14). Similarly, the fluorescence of LchiWRKY36-GFP (35S::LchiWRKY18::GFP) fusion protein was also preferentially detected in the nucleus. The results of the subcellular localization experiment show that LchiWRKY18 and 36 are expressed in the nucleus, which is consistent with the results of subcellular localization analysis, indicating that the main biological functional position of transcription factors is in the nucleus to regulate the transcription of target genes.

Lchi has fewer WRKY members, diversified biological functions and physicochemical properties, and conserved WRKY domain and motif structure
WRKY genes comprise a large family of transcription factors that are conserved across the whole plant kingdom. Genome-wide analysis of WRKY gene families has been widely carried out in a large number of species whose genomes have been sequenced [9,42]. In the current study, a search for WRKY genes in the L. chinense genome resulted in the identification of 44 members, which we designated as LchiWRKY1 through LchiWRKY44, based on their chromosomal location. Compared with some higher angiosperms, the LchiWRKY family has a smaller number. For example, there are 125 ZmWRKY genes in the maize (Zea mays) genome [43], 108 HbWRKY genes in the Rubber Tree (Hevea brasiliensis) genome [44], 102 LuWRKY genes in Flax (Linum usitatissimum L.) genome [45], 86 HvWRKY genes in Barley (Hordeum vulgare L.) genome [46], and 64 IiWRKY genes were identified in the whole genome of Isatis indigotica [47]. A total of 61 CsWRKY genes in the Cucumber (Cucumis sativus L.) genome [4], 57 CmWRKY genes were identified in the melon (Cucumis Melo L.) genome [48], and 54 AcWRKYs in Pineapple (Ananas comosus (L.) Merr.) genome [49]. Although the number of WRKY members of these plants with different evolutionary statuses is different, they are more than L. chinense. The results showed that the number of WRKY family members of L. chinense might be related to its evolutionary status and family expansion model, L. chinense was close to the basal angiosperm [23], so speculate the number of LchiWRKY family members might be less than that of higher angiosperm plants.
Gene ontology (GO) showed that most WRKY genes were enriched in molecular function, biological process, and cellular component. For instance, GO analysis of Tartary buckwheat WRKYs showed that most FtWRKYs are enriched in molecular function and biological processes, and some genes are enriched in a cellular component [41]. In Orchardgrass, 93 DgWRKYs were enriched in biological regulation and cellular process of biological process category, and binding and nucleic acid binding TF activity of molecular function category, while some of them were in membrane and membrane part of cellular component category [50]. In Rice, novel candidate OsWRKY genes were significantly enriched in sequencespecific, DNA, heterocyclic and regulatory region binding activities of molecular function [51]. In Taraxacum antungense Kitag, 44 TaWRKYs were significantly enriched in 23 terms of biological process, and cellular processes were the most representative [52]. In L. chinense, A total of 34 LchiWRKYs were significantly enriched in the two GO categories of biological processes and molecular functions, including 10 subcategories, which do not contain cellular components. This is different from the above species. It is speculated that it may be the difference in the regulatory function of WRKY transcription factors among different species. Therefore, biological processes and molecular function may be the main biological functions and processes involved in LchiWRKYs.
The protein length (MW) and isoelectric point (pI) of LchiWRKY proteins showed a large degree of variation across the gene family, indicating that LchiWRKYs possess different protein structures and may have adapted to different microenvironments. These features suggest that LchiWRKYs may have biological regulatory functions in different environments/conditions, especially during abiotic stress. The WRKY protein length, molecular weight, and isoelectric point are different in different species, for example, in Pineapple, the protein length of AcWRKYs varied from 122 to 1320 amino acid (aa), and MW varied from 13.7 to 144.5 kDa, and pI ranged from 5.11 to 10.08 [49]. In Maize, the length of ZmWRKYs varied from 99 to 729 aa, and MW varied from 11.27 to 78.7 kDa, and pI ranged from 4.55 to 10.78 [43]. Therefore, the biological functions of WRKY genes in different species are different, which may be related to the physical and chemical properties and spatial structure of proteins.
The conserved WRKY domains of the LchiWRKY proteins were assessed in this study. Multiple sequence alignments revealed that two LchiWRKY proteins (Lchi-WRKY10 and LchiWRKY37) in group IIc had sequence variation in their WRKY domain. Most characterized WRKY proteins exhibit a binding preference to their cognate cis-acting W-box element, with the help of the WRKY domain. According to previous studies, variations in the WRKYGQK motif within the WRKY domain might influence normal interactions of WRKY genes with their downstream target genes, and it might therefore be worthwhile to further investigate the binding specificity and functionality of these two WRKY proteins [53,54]. R-type introns (PR intron) and V-type introns (VQR intron) are two major introns within the WRKY family in different plants; In L. chinense, the groups I, IIc, IId, III we identified contain R-type introns, and groups IIa and IIb contain V-type introns [4,55]. These two types of introns may play important roles in biological processes, such as heat stress, cold stress, salt stress, and drought stress. For example, LchiWRKY5 and LchiWRKY23 were highly expressed under cold, heat, and drought stress [4,11,55]. These WRKY protein variants and two intron insertions have different forms in different species, which indirectly reflects the conservation and diversity of WRKY gene in the process of plant evolution, so that the functions of WRKY gene among different species are not only similar but also different [4,11,54,55].
The conserved motifs reveal the grouping specificity of motif distribution of L. chinense, a total of 20 motifs were distributed in 7 subgroups, each of which had a specific motif, except for groups IIe (LchiWRKY30) and III (LchiWRKY40, LchiWRKY42). In Orchardgrass, 10 motifs were determined and the same phylogenetic group has the same conserved motif, only group III didn't have a C2H2 zinc finger motif [50]. In Common walnut, 15 motifs were detected in JrWRKYs, only two genes contain the CCCC domain, while other phylogenetic subgroups contain group-specific motifs [40]. The above results show that although the number of WRKY is different in different species, the motif has specificity in different phylogenetic groups, and the same subgroup has the same motif.

The expansion of the LchiWRKY family results from tandem repeat events and functional and structural diversity of replication genes
Tandem and segmental duplication events have played a critical role in the expansion of the WRKY gene family [6]. In Rubber tree (Hevea brasiliensis), among the 25 pairs of duplicated gene pairs, 2 pairs of genes were derived from tandem duplication events and 23 pairs of genes were segmental duplication events [44]. In Quinoa (Chenopodium quinoa Wild.), a total of 39 pairs of replication gene pairs, 2 pairs of genes are tandem replication, 37 pairs of genes are segmental replication [14]. In Cucumber (Cucumis sativus L.), 14 segmental replication events occurred in 25 genes [4]. In Pineapple (Ananas comosus), 7 pairs of genes were tandem duplication genes and 17 segmental replication events occurred in 27 genes [49]. In Maize (Zea mays), 52 segmental duplication events occurred in 78 genes, and there was no tandem replication event [43]. In Barley (Hordeum vulgare L.), 5 tandem duplication events and 13 pairs of genes were segmental duplications [46]. In Chinese woad (Isatis indigotica), 5 gene pairs were tandem duplication and 7 gene pairs were segmental duplication [47]. In addition, whole-genome duplication events may also affect the number of gene families. For basal angiosperms, for example, there are no ancestor whole-genome duplication events in A. trichopoda [32], while there is an ancestor whole-genome duplication events in Water lily [56], which may lead to the total number of WRKY in water lily being more than that in A. trichopoda (Table 1). For Magnolia plants, two whole-genome duplication events occurred in C. kanehirae and one whole-genome duplication event occurred in L. chinense [23,32], which may also lead to more WRKY in C. kanehirae than L. chinense (Table 1).
A comparison of the number of WRKY genes in L. chinense with other sequenced higher angiosperm genomes shows that L. chinense possesses a smaller number of WRKY genes. This could indicate a reduced number of duplication events in L. chinense, only six genes had tandem replication events. Among these genes, four (LchiWRKY14, 15,20,27) are from group II-d, and the other two are from group I (LchiWRKY14) and II-a (Lchi-WRKY35), This suggests that group IId may play a major role in the expansion of LchiWRKY family, and tandem repetition is the main expansion model. Therefore, the total number of gene families may be related to wholegenome duplication events, tandem duplication, and segmental duplication events, which ultimately determines the difference in the number of WRKY families in different species.
We found that the expression pattern of duplicated paralogous LchiWRKY genes was always different. Divergence of gene expression plays an important role in the preservation of duplicated genes [57,58]. Several paralogous gene pairs show a differing response to abiotic stress, suggesting that they may play diverse roles in L. chinense in response to abiotic stress (Figs. 8, 9 and 10). For instance, LchiWRKY23 was highly expressed in response to cold and drought stress at 12 h ~ 1d, while its paralogous gene LchiWRKY35 was not induced upon cold and drought stress. Still, some WRKY genes and their paralogues, like LchiWRKY14 and LchiWRKY27, shared similar high transcript abundance profiles under heat stress at 1d, 3d, suggesting that they may have redundant functions. Further analysis showed that the duplicated genes contained the same cis-regulatory elements (Fig. 7), such as LTR elements, which mainly responded to lowtemperature stress, so they could all respond to lowtemperature stress, such as LchiWRKY27,14. Combined with the results of gene structure and motif, these replicated genes also have diversity in structure and motif. For instance, LchiWRKY23 has 6 motifs and LchiWRKY35 has 7 motifs, only motif 1 is their common motif (Fig. 5). LchiWRKY27 lacks motif 18, and LchiWRKY14, 15 and 20 have it (Fig. 5). LchiWRKY35 has 3 exons and Lchi-WRKY23 has 2 exons and 4 UTRs, but no UTR is found in LchiWRKY35 (Fig. 6).

LchiWRKY gene has a cross-regulation function in response to multiple abiotic stress
According to previous studies, WRKY transcription factors are involved in abiotic stress regulation, including low temperature, high temperature, salt, and drought [4,16,44]. For instance, in Ipomoea trifida, 11 Itfwrkys (Itfwrky8, 15.1, 22.1, 34, 41, 48, 66, 69, 77, 79, and 80) were highly expressed under salt, drought, low temperature, and high temperature [16]. In Cucumber (Cucumis sativus L.), 21 CsWRKY genes were induced to express under heat stress, and the expression trend was the same from 3 h to 6 h, five CsWRKY genes (CsWRKY27 / 41 / 50 / 52 / 57) were induced to express under salt and heat stress [4]. In Camelina (Camelina sativa (L.) Crantz), the expression of CsWRKY21 under cold stress was higher than that under drought stress, while the expression level of CsWRKY22 under drought stress was higher than that under cold stress [17]. In Quinoa (Chenopodium quinoa Wild.), nearly 46 CqWRKY genes are induced to express under salt or drought stress. The expression level of CqWRKY52a-1 / 2 was similar to that of group II-d, which was highly expressed under salt stress [14]. In Chickpea (Cicer arietinum L.), 15 genes were down-regulated in roots and 10 genes were down-regulated in stems under abiotic stress [59]. The OsWRKY55 negatively regulates drought tolerance in transgenic rice lines [60], The transcription level of HvWRKY42 increased in Barley (Hordeum vulgare L.) under drought stress, which indicated that HvWRKY42 was regulating the drought adaptability of barley [61]. The HvWRKY3 showed 13 fold high expression and 4 genes showed 2-fold high expression under drought stress [46]. In Maize, under drought stress, 10 ZmWRKYs were upregulated in roots and 4 ZmWRKYs showed 10-folds expression in leaves at 8 h [43]. In Pineapple, AcWRKY35 was induced by drought stress, and 7 AcWRKYs were expressed by cold stress [49]. In Camellia japonica, 13 CjWRKYs were highly expressed under salicylic acid (SA) stress for 12 h, indicating that the CjWRKY gene can regulate C. japonica to adapt to SA environmental conditions [62].
In the study, we explored the regulatory function of LchiWRKY under abiotic stress by treating somatic embryogenesis of Liriodendron hybrids seedlings with low, high-temperature, and drought stress, and the expression patterns of low temperature, high-temperature, and drought stress in the WRKY family were obtained. Low temperature is one of the major environmental stresses that affect L. chinense growth and development, and high temperature and soil drought cause the death of trees and reduce the yield of forests. According to the cold stress expression pattern (Fig. 8), the expression of 11 LchiWRKY genes was significantly induced. The overall expression pattern is divided into two clusters, including high expression at different time points and persistent low expression at all time points. Group I and II responded to cold stress, but group III did not respond, indicating that the response of group III to regulating cold stress was not obvious, while the regulation of groups I and II was more obvious, which also reflected the functional specificity of the group. LchiWRKY5 and 14 may be the main members of the WRKY family in response to low-temperature stress, and 1d and 3d are the main response time points of the WRKY family. It shows that when L. chinense is stimulated by cold stress, it first adapts to the environment through self-protection, and then regulates its adaptability through cold stressrelated resistance genes expressed at 12 h or 1d. Group II is the main cold stress regulatory gene, which promotes L. chinense to adapt to low-temperature environment stress.
The expression patterns of high-temperature stress were also divided into two groups (Fig. 9), including 1d and 3d high expression patterns and persistent low expression patterns. Similar to cold stress, group III was not significantly induced, and there were more response genes in group II than in group I. LchiWRKY14,18 are the key genes of the WRKY family to respond to hightemperature stress. 1d and 3d are also the main response time points of high-temperature stress. This also shows that the adaptation of L. chinense to a high-temperature environment is similar to that to a low temperature. First, it carries out self-protection and then regulates its adaptability through high-temperature response genes expressed at 12 h or 1d. Group II is also the main regulatory gene of L. chinense in response to high-temperature stress, which indicates that the function of group II is very important for temperature stress.
The expression patterns of drought stress can also be divided into two groups, containing a high expression pattern of 12 h, 1d, and a persistent low expression pattern (Fig. 10). Different from low temperature and hightemperature stress, group III genes were significantly induced, indicating that group III may be drought-specific response genes. The number of genes in groups I and II are still more than that in group III, indicating that these two groups are also important genes in response to drought stress. LchiWRKY14,18 may be the main response gene and 12 h, 1d may be the main response time point to drought stress. This shows that L. chinense adapts to drought stress as well as temperature stress. It first protects itself and then adapts to the environment by specifically regulating gene expression at 12 h or 1d. LchiWRKY family can regulate L. chinense to adapt to drought environment, but the regulation function of groups I and II are more obvious than that of group III.
Interestingly, some WRKY genes have high expression patterns under different stresses, such as LchiWRKY5, 23,14,27,36, which are highly expressed at 1d time points of three stresses, indicating that these genes may have a cross-regulation function under different stresses. These five LchiWRKYs were also the key regulatory genes for the growth and development of L. chinense to promote L. chinense to adapt to different adverse environmental conditions. The results of qRT-PCR under low, hightemperature, and drought stress also showed that 1d was the main response time point of the WRKY family, which was consistent with the results of the transcriptome, and reflected the accuracy of transcriptional data.

LchiWRKY transcription factor is mainly expressed in the nucleus
As transcription factors, they are mainly expressed in the nucleus [40,[49][50][51]63], but some are also expressed in other organelles, such as chloroplast, mitochondria, cytoplasm, and cell membrane [43,51,63]. Subcellular localization analysis of maize showed that most ZmWRKY transcription factors were expressed in the nucleus, and only a few ZmWRKYs were expressed in other organelles [43]. In pineapple, 51 AcWRKYs were located in the nucleus and only 3 in the chloroplast [49]. Similarly, in Orchardgrass, most of DgWRKYs (88.17%) were located in the nucleus, and only a few in chloroplasts (7), cytoplasm (2), mitochondria (1), and peroxisome (1) [50]. The difference is that in walnut, all JrWRKYs were expressed in the nucleus [40]. The Subcellular localization analysis of L. chinense showed that all LchiWRKYs were expressed in the nucleus, except the LchiWRKY42 was also expressed in the cell membrane. Combined with phylogenetic tree (Fig. 1) analysis, LchiWRKY42 is in a separate branch, so it is speculated that its function is different from other genes, which may be due to functional differentiation in the process of evolution.
The subcellular localization experiments of Lchi-WRKY18 and LchiWRKY36 in L. chinense showed that these two transcription factors were located in the nucleus (Fig. 14), indicating that they were expressed in the nucleus and regulated the expression of target genes by binding the promoter sequence of target genes. In other species, the subcellular localization experiments of WRKY transcription factor are mainly in the nucleus, For example, in Sugarcane (Saccharum spp.), the subcellular localization experiment of ScWRKY5 showed that it was expressed in the nucleus [64]. In sandalwood (Santalum album L.), the SaWRKY1 protein was also expressed in the nucleus [39]. In Liriodendron hybrids, the LhWOX1 protein was expressed in the nucleus [65].
Based on the above analysis, the expression position of WRKY transcription factor is different in different species, but most members are mainly expressed in the nucleus and a few in other organelles, indicating that the main function of WRKY transcription factors is to regulate gene expression, but also participate in other biological processes.
Overall, the above findings provide insight into the potential functional role of L. chinense WRKY genes. The LchiWRKY family has a cross-regulation function, which makes L. chinense adapt to different stress environments. Our comprehensive analyses helped select candidate WRKY genes for further functional verification and genetic improvement of L. chinense agronomic traits and environmental resistance. Additionally, to provide a theoretical basis for promoting afforestation of Liriodendron in different areas.

Conclusions
A comprehensive analysis of the WRKY gene family in L. chinense was carried out in the present study. Forty-four full-length WRKY genes were characterized and further classified into three main phylogenetic groups, phylogenetic comparison of WRKY genes from 17 different plant species provided valuable clues about the evolutionary characteristics of LchiWRKY genes, and L. chinense has fewer WRKY members than other higher angiosperms. LchiWRKY family has a highly similar exon-intron structure and motif compositions within the same subgroup and the regulatory functions of different subgroups are specific. Subcellular localization analysis and experiments showed that LchiWRKYs were mainly expressed in the nucleus, and GO analysis focused on biological processes and molecular functions. The analysis of gene replication events and selection pressure showed that the expansion of the WRKY family was mainly through tandem repeat events, and suffered from purification selection during evolution, which may lead to a small number of WRKY members. More importantly, LchiWRKY is also involved in the regulation of abiotic stress, 23,14,27,and 36 cross-regulate the response to low, high-temperature, and drought stress. These results provide a valuable resource for a better understanding of the biological roles of WRKY genes in L. chinense. At the same time, it also provides a theoretical basis for the growth and development and afforestation in different areas of Liriodendron.

Datasets and sequence retrieval
The complete genome, transcript/protein sequences, and genome feature file of L. chinense (Lchi)

Phylogenetic analysis
Multiple sequence alignment (MSA) of Amborella trichopoda, Liriodendron chinense, Brachypodium distachyon, Cinnamomum kanehirae, Gossypium raimondii, Oryza sativa, Arabidopsis thaliana, Capsella grandiflora, Cucumis sativus, Eucalyptus grandis, Malus domestica, Nymphaea colorata, Spirodela polyrhiza, Panicum hallii, Populus trichocarpa, Theobroma cacao, Vitis vinifera WRKY domain-containing full-length proteins with FASTA format was done using MUSCLE (v3.8.31) set the following parameters: the maximum number of iterations was 1 with '-maxiters 1' , and find diagonals with '-diags -sv -distance1 kbit20_3' [74]. The trimmed MSA file was generated with trimAl (v1.4) [75] set to 'auto-mated1' mode and then used to construct the WRKY phylogenetic tree. The Bayesian phylogenetic tree was constructed using the BEAST software (v2.6.6), by inputting the trimmed file in FASTA format with BEAUti 2 program under the following settings: the Site model was Gamma Site model, substitution rate was 1.0, and substitution model was Dayhoff of site model, clock.rate of clock model was 1.0, Priors of the tree was set to the Yule Model, the Birth Rate was 1.0, the MCMC chain length was 10,000,000, and then output the file in XML format. After the BEAST program is completed and the TreeAnnotator program set posterior probability limit was 1.0 with a Burnin percentage of 90, target tree type was maximum clade credibility tree, node heights were common ancestor heights to build the Bayesian phylogenetic tree (the Bayesian tree file was shown in Additional files 2 and 3: Tables S2-S3) [76]. The phylogenic tree was then visualized using the R package ggtree [77,78] (http:// www. bioco nduct or. org/ packa ges/ relea se/ bioc/ html/ ggtree. html). OrthoFinder (v2.2.3) was used to explore the WRKY evolutionary status of all the 17 species by using a folder of their protein full-length sequences in the FASTA format with the parameter '-f ' , and used the DIA-MOND sequence search and MSA gene tree inference, then MAFFT was used for multiple sequence alignment and the IQ-TREE program was used to construct species tree with maximum likelihood species tree inference, and the 17 rooted species tree file was shown in (Additional file 4: Table S4) [79][80][81], and then displayed the rooted species tree by using the R package ggtree [77].

Plant materials and stress treatment
Liriodendron hybrids seedlings generated through somatic embryogenesis were used as the starting material throughout this study. Before any experiments were performed, plantlets were taken out of the culture medium vessel and acclimatized in a greenhouse for 2 weeks (22 °C, long day photoperiod of 16 h light/8 h dark and 75% relative humidity). For various abiotic stress treatments, plants were transferred to a growth chamber (long time photoperiod of 16 h light/8 h dark and 75% relative humidity): to simulate cold or heat or drought stress, plantlets were subjected to a 4 °C or 40 °C or 15% PEG6000 (the 15% concentration was determined according to the reports of references [47,49,82,83] and the preliminary experimental results (Additional file 19: Fig. S3), the control group was added the same amount of water to the substrate and the temperature was 22 °C) treatment respectively for 1 h, 3 h, 6 h, 12 h, 1d, and 3d in the growth chamber. Each treatment consisted of five biological replicates (five plantlets) for each time point. All treated leaf tissue samples of each plantlet were immediately frozen in liquid nitrogen and stored at − 80 °C for further RNA extraction.

RNA extraction, library preparation, and sequencing
RNA degradation and contamination were monitored on 1% agarose gels. RNA purity was checked using the NanoPhotometer ® spectrophotometer (IMPLEN, CA, USA).RNA integrity was assessed using the RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 system (Agilent Technologies, CA, USA), cDNA was synthesized using a HiScript ® III 1st Strand cDNA Synthesis Kit (Vazyme, Nanjing, China) with the extracted RNA as a template.
A total amount of 1 μg RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were generated using NEBNext ® UltraTM RNA Library Prep Kit for Illumina ® (NEB, USA) following the manufacturer's recommendations and index codes were added to attribute sequences to each sample. Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature in NEBNext First Strand Synthesis Reaction Buffer (5X). First-strand cDNA was synthesized using random hexamer primer and M-MuLV Reverse Transcriptase (RNase H-). Second strand cDNA synthesis was subsequently performed using DNA Polymerase I and RNase H. Remaining overhangs were converted into blunt ends via exonuclease/polymerase activities. After adenylation of 3'ends of DNA fragments, NEB-Next Adaptor with hairpin loop structure was ligated to prepare for hybridization. To select cDNA fragments of preferentially 250 ~ 300 bp in length, the library fragments were purified with AM Pure XP system (Beckman Coulter, Beverly, USA). Then 3 μl USER Enzyme (NEB, USA) was used with size-selected, adaptor-ligated cDNA at 37 °C for 15 min followed by 5 min at 95 °C before PCR. Then PCR was performed with Phusion High -Fidelity DNA polymerase, Universal PCR primers, and Index (X) Primer. At last, PCR products were purified (AMPure XP system) and library quality was assessed on the Agilent Bioanalyzer 2100 system.
The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumia) according to the manufacturer's instructions. After cluster generation, the library preparations were sequenced on an Illumina Novaseq 6000 platform and 150 bp paired-end reads were generated.

Transcriptome (RNA-seq) analysis
The L. chinense 7 time points leaf transcriptome with biological triplicates (three plantlets) of heat, cold, and drought stress was analyzed using RNA-seq pipeline, Raw data (raw reads) of FastQ format files were firstly carried out quality control through FastQC (v0.11.9) to obtain the overview of Raw data, including Basic Statistics, Per base sequence quality, Per sequence quality scores, Per sequence GC content, Per base N content, Sequence Length Distribution, and Adapter Content. Then the Raw data were filtered according to the quality control results by using Trimmomatic (v0.39) with the following parameters: remove adapters (ILLUMINACLIP: TruSeq3-PE. fa:2:30:10), remove leading low quality or N bases (below quality 10) (LEADING:10), remove trailing low quality or N bases (below quality 10) (TRAILING:10), scan the read with a 4-base wide sliding window, cutting when the average quality per base drops below 20 (SLIDINGWIN-DOW:4:20), drop reads below the 90 bases long (MIN-LEN:90) [84]. In this step, clean data (clean reads) were obtained by removing reads containing adapter, reads containing ploy-N and low-quality reads from Raw data. At the same time, Q20, Q30, and GC content of the clean data were calculated. All the downstream analyses were based on clean data with high quality.
Reference genome and gene model annotation files L. chinense, were downloaded from the NCBI website (https:// www. ncbi. nlm. nih. gov/ assem bly/ GCA_ 00301 3855.2), index of the reference genome was built using Hisat2 (v2.2.1) and paired-end clean reads were aligned to the reference genome using Hisat2 (v2.2.1) [85]. We selected Hisat2 as the mapping tool for that Hisat2 can generate a database of splice junctions based on the gene model annotation file and thus a better mapping result than other non-splice mapping tools. Then we used the Kallisto (v0.46.1) to quantify gene expression by calculating all clean reads mapped to this gene with "kallisto quant" command, and count Transcripts Per Million (TPM) values of individual genes, considering the effect of sequencing depth and gene length for the reads count at the same time and is currently the most commonly used method for estimating gene expression levels [86]. We used DESeq2 (v1.16.1) with the (ANOVA method: p.adjust< 0.05, |Log2FC| > 1) to determine differentially expressed genes (DEGs), and TPM values were used to calculate the transcript abundance of LchiWRKYs. All L. chinense genes had been annotated previously in the Pfam database (http:// pfam. xfam. org/) and Blast2GO website (http:// www. blast 2go. com). The data quality, mapping rate, gene annotation, and normalization (TPM values) of heat, cold, and drought stress RNAseq are shown in (Additional files 20, 21, 22, and 23: Tables S17-S20). Expression heatmaps were created using the pheatmap (v1.0.12) R package, based on the transformed data of log2 (TPM + 1) values. The transcriptome data used in this study has been archived and can also be obtained on the NCBI website, cold and heat stress accession numbers were PRJNA679089 (https:// www. ncbi. nlm. nih. gov/ biopr oject/ PRJNA 679089/), and drought stress accession number was PRJNA679101 (https:// www. ncbi. nlm. nih. gov/ biopr oject/ PRJNA 679101/).

qRT-PCR analysis
Primers used for quantifying the expression of LchiWRKYs were designed by using the primer3 website (https:// www. yeast genome. org/ prime r3) (Additional file 24: Table S21). qRT-PCR was carried out with a Roche Lightcyler ® 480II instrument using 2x AceQ ® qPCR SYBR ® Green Master Mix (Without ROX) (Vazyme, Nanjing, China). The composition of the PCR mix was as follows: 10 μl 2x AceQ ® qPCR SYBR ® Green Master Mix (Without ROX), 0.4 μl of each primer and 1 μl of cDNA template (10 ng/μl), adding 8.2 μl ddH 2 O in a final volume of 20 μl. The housekeeping L. chinense 18S gene was used as an internal control. The melt and standard curve of WRKY qRT-PCR primers were shown in Additional file 25: Fig.  S4. The reaction was carried out as follows: 95 °C for 10 min, followed by 45 cycles of 95 °C for 10 s, then 60 °C for 30 s. All reactions were run in 96-well plates. Each reaction was performed in biological triplicates, as well as three technical replicates. All data generated from real-time PCR amplification was analyzed using a 2 − △△CT method.
The callus of Liriodendron hybrids cultured for 20 days was used to prepare protoplasts, and protoplasts were slowly and gently dissolved into a solution (10 mL) containing 0.5Mmannitol, 20 mM MES, pH 5.7, 20 mM KCl, 0.1% (w/v) bovine serum albumin, 10 mM CaCl 2. and digested at 28 °C under dark conditions for 3 h [65], The protoplasts were transformed by PEG-6000, pipetted into a 6-well cell culture plate, and cultured at 23 °C under dark for 16 ~ 48 h [65], then the fluorescence effect of protoplasts was observed by ZEISS LSM 800 fluorescence microscope (Carl Zeiss, Germany).