The plant-specific B3 superfamily constitutes one of the largest transcription factor (TF) families in plants. All members of the B3 superfamily contain an approximately 110 amino acid region called the B3 domain (Swaminathan et al., 2008). This domain is initially named because it is the third basic domain in the maize protein VIVIPAROUS1(VP1) (McCarty et al., 1991). The B3 domain of VP1 has a sequence-specific DNA binding activity (Suzuki et al., 1997). Subsequently, the B3 domain has been found in 118 genes of Arabidopsis and in 91 genes of rice. B3 TF genes are also present in green algae, mosses, liverworts, ferns, and gymnosperms (Marella et al., 2006).

The B3 superfamily encompasses several distinct gene families, including LAV (LEAFY COTYLEDON2 [LEC2]-ABSCISIC ACID INSENSITIVE3 [ABI3]-VAL) (Wang et al., 2012), ARF (AUXIN RESPONSE FACTOR) (Ulmasov et al., 1997), RAV (RELATED TO ABI3 and VP1) (Kagaya et al., 1999) and REM (REPRODUCTIVE MERISTEM) (Romanel et al., 2009). B3 TFs are involved in a variety of biological processes, such as seed development, embryonic development, etc., generally through polycomb silencing in plants. In Arabidopsis, the VAL subfamily genes VAL1/HIGH-LEVEL EXPRESSION OF SUGAR-INDUCIBLE GENE 2 (HSI2), VAL2/HSI2-Like 1 (HSL1), and VAL3/HSI2-Like 2 (HSL2) are expressed in many organs throughout plant development. val1 val2 double mutant exhibits embryonic callus formation on both roots and shoots, suggesting a role of VAL1 and VAL2 in repressing the embryonic developmental program. Germination is delayed in this double mutant, and the analysis shows that HSI2-dependent silencing of DOG1 promotes the early release of seed dormancy (Chen et al., 2020). LAV family member LEAFY COTYLEDON 1 (LEC1), which is the Hap3 (HEMEACTIVATED PROTEIN 3) sub-unit of the CAAT-binding factor, controls various aspects of seed development from early embryogenesis to late seed maturation (Baumbusch et al., 2004). ABI3, LEC2, and FUSCA 3 (FUS3) also play key roles in the control of seed maturation in cooperation with LEC1 and abscisic acid (ABA) (Tsukagoshi et al., 2005). In Vitis Vinifera L., overexpression of VvFUS3 in tomatoes causes a reduction in total cell area and cell number, while the cell size of the fruit pericarp is increased. These results suggest that VvFUS3 has a role in seed development by influencing the ABA signaling (Ahmad et al., 2022). ARF family member AtARF1 binds the upstream promoter regions of many auxin response genes (Ulmasov et al., 1997). Mutation of ARF7 causes nearly an absence of lateral root formation, lacking of phototropism and auxin responses (Liscum & Briggs, 1995; Ruegger et al., 1997; Watahiki and Yamamoto, 1997). Both ARF4 and ARF3 regulate the abaxial-adaxial polarity in leaves and flowers in Arabidopsis (Pekker et al., 2005). RAV1 functions as a negative regulator of lateral root and rosette leaf development, and it is downregulated by the plant steroid hormone epibrassinolide (Hu et al., 2004). GmRAV negatively regulates short day (SD)-mediated flowering and hypocotyl elongation, and its overexpression in SD inhibits the growth of soybean leaves, roots, and stems (Zhao et al., 2008). The B3 gene also promotes flowering and functions in the maintenance of the vernalization response in Arabidopsis. For example, FLC (FLOWERING LOCUS C), a gene delays flowering, is transiently downregulated in the vrn1(VERNALIZATION1) mutant that has vernalization treatment, but levels of FLC RNA and protein are increased when the mutant is moved to normal temperature (Levy et al., 2002; Sung and Amasino, 2004). Brassica oleracea BoREM1, a putative ortholog of AtREM34, is specifically expressed in the cauliflower curd and involved in the determination of floral meristem identity (Franco-Zorrilla et al., 1999). In short, the B3 proteins of the LAV, RAV, ARF, and REM families are mainly involved in the signaling pathways of hormones including auxin, ABA, and brassinosteroid.

Wood formation is a specific biological process in trees. The major components of wood are lignin, cellulose, and hemicelluloses, which are deposited in the secondary cell walls (SCWs) of fiber cells and vessels in the xylem. These three polymers maintain the overall structure and strength of plant SCWs (Wessels et al., 2019). After years of efforts, the mechanism of lignification and SCW deposition has been preliminarily analyzed, and regulatory networks that regulate wood formation have been proposed in some review articles (Zheng-Hua and Zhong, 2015; Borthakur et al., 2022). The regulatory network for SCW thickening includes many NAC and MYB genes, such as NST1 to NST3/SND1 (Mitsuda et al., 2005; Zhong and Ye, 2006; Zhong et al., 2007), VND1 to VND7 (Kubo et al., 2005; Yamaguchi et al., 2008; Zhou et al., 2014), and PtrWND2B/6B (Zhong et al., 2011), are master regulators of xylem differentiation and SCW thickening. However, there are still more upstream regulatory genes for these master regulators, which need to be revealed. VNI2 (VND-INTERACTING2) and VND7 were co-expressed in the elongated ductal precursors of the root. Overexpression of VNI2 significantly inhibited the differentiation of normal vessel cells. Transient transformation of the reporter gene showed that VNI2 is a transcriptional repressor, which can inhibit the expression of VND7-regulated vessel-specific genes. Sumoylation of LBD30 in poplar affected the expression of SND1 and NST1 in the transient transformation of protoplast systems. As LBD30 is sumoylated by SIZ1, this protein modification activates the regulatory network of SND1/NST1-mediated secondary cell wall formation in fibers. In addition, lignin is the most direct material for the study of wood formation. Lignin accounts for ~20% of SCWs, and it is polymerized by three hydroxy cinnamyl alcohols, including sinapyl alcohol, coniferyl alcohol, and p-coumaryl alcohol (Sarkanen and Hergert, 1971; Watahiki and Yamamoto, 1997). These alcohol precursors are termed the syringyl (S), guaiacyl (G), and hydroxyphenyl (H) lignin subunits, respectively. In angiosperms, lignin is polymerized primarily from S and G monolignols and trace amounts of the H monolignol (Wang et al., 2021). The biosynthesis of monolignols lignin occurs in several consecutive reactions involving 11 enzyme families and 24 metabolites, in a branched grid-like pathway (Vanholme et al., 2013; Wang et al., 2014; Sulis and Wang, 2020). These families include phenylalanine ammonia-lyase (PAL), cinnamate 4-hydroxylase (C4H), p-coumaroyl-CoA 3-hydroxylase (C3H), p-coumarate CoA ligase (4CL), hydroxycinnamoyl transferase (HCT), caffeoyl shikimate esterase (CSE), caffeoyl-CoA O-methyltransferase (CCoAOMT), cinnamoyl-CoA reductase (CCR), conifer aldehyde 5-hydroxylase (CAld5H, first named F5H, ferulate 5-hydroxylase), caffeic acid 3-O-methyltransferase (COMT) and cinnamyl alcohol dehydrogenase (CAD) (Zhao and Dixon, 2011). In this pathway, C3H catalyzes the hydroxylation of coumaric acid to form caffeic acid through the 3-carbon hydroxylation of aromatic rings of various phenol intermediates (Palafox-Carlos et al., 2014). The reduction of feruloyl-CoA to conifer aldehyde is mediated by CCR (Li et al., 2005). Coniferaldehyde is catalyzed by CAD to produce confieryl alcohol, G monolignols (Baucher et al., 1996). Coniferaldehyde can also be converted to sinapyl alcohol by CAld5H/F5H, COMT, and CAD, to produce S monolignols (Lapierre et al., 1999). Overexpression of CAld5H have a 64% increase in the S/G ratio in poplar, and co-transformation of F5H and COMT results in 2- to 3- times higher S/G ratio than F5H alone (Li et al., 2003). The lignin S/G ratio varies among different tree species, and it is a more important factor affecting the pulp yield than the lignin content (Del Río et al., 2005; Studer et al., 2011). High pulp yield is correlated with a high S/G ratio in wood biomass, so genes controlling S/G ratio in tree have great promising to breed pulp-specific trees.

Poplar is an ideal species to study the formation of wood, the important biomass energy material in nature. Although the pathway enzyme genes of cellulose, hemicellulose, and lignin biosynthesis and many upstream TF regulators have been identified, these studied regulators are only a small proportion of TFs. Among B3 TF family, PagVAL2-B1 is identified as an candidate regulator involved in wood formation, because yeast one hybridization shows that it can bind to PagCAld5H2 promoter (Wang et al., 2021). Many remaining genes that are specifically expressed during wood formation may have potential functions in regulating specific wood properties. The involvement of B3 TF family genes in wood formation and their specific regulation on wood property formation needs investigation. In this study, we focused on the characterization of the B3 family in P. alba × P. glandulosa, and identified potential key genes involved in wood formation.

The plant-specific superfamily of B3 transcription factors is distinguished by the existence of one or more B3 domains, or a fusion of B3 domains with supplementary domains like AP2 (APETALA2), AUX_IAA, and zf-CW (Peng and Weselake, 2013). The B3 transcription factor superfamily, comprises four families, among which the LAV(LEC2/VP1/VAL) and ARF families are well-studied in Arabidopsis, characterized by diverse functions in plant growth and development, and extensively studied in Arabidopsis (Yamasaki et al., 2004; Swaminathan et al., 2008; Yamasaki et al., 2008; Agarwal et al., 2011). In contrast, the identification and characterization of B3 genes in woody plants is limited. New insight into this superfamily can be obtained through genome analysis. In this study, we performed a comprehensive characterization of B3 genes in P. alba × P. glandulosa and investigated their expression profiles in leaf, phloem, and xylem. We identified 160 putative B3 genes in the genome of P. alba × P. glandulosa, showing a discrepancy of gene number in each family with Arabidopsis. Different species are expected to have varying numbers of B3 genes in each family, and the number of B3 genes identified in different studies may also be influenced by the use of different database sources, methods, and parameters (Romanel et al., 2009; Peng and Weselake, 2013). For example, previous studies have classified B3 proteins lacking typical AP2 domains as members of the RAV family (Magnani et al., 2004; Kim et al., 2006). Similarly, in the LAV family, we identified only 10 VAL subgroup members, while LEC2-ABI3 was not identified according to our analysis; this could be due to the protein sequence variation from species to species. In addition to the B3 domain, the VAL members contain a zf-CW domain, with the one exception of PagVAL3-A1. For the RAV family, our results are consistent with previous results showing that some proteins contain the B3 and AP2 domains (Alvarez et al., 2009). Based on the evolutionary relationships between RAV proteins in Arabidopsis and P. alba × P. glandulosa, it was found that 8 out of 20 RAV proteins had AP2 domains. In the ARF family, 24 out of 56 members contained three domains (B3, Auxin_resp, and AUX_IAA), while the remaining 32 contained only two domains (B3 and Auxin_resp). The same observations have been reported in other plants, such as Arabidopsis, apple, and papaya (Li et al., 2015; Peng et al., 2020). We speculated that loss of the zf-CW, AP2, and AUX_IAA domains occurs during evolution. In a gene family, the emergence of new members can occur through domain duplication or loss. For example, in the MYB family, the plant-specific R2R3 organization is thought to have evolved from an R1R2R3-type ancestral gene from which the first repeat domain R1 motif was lost (Braun and Grotewold, 1999; Riechmann et al., 2000; Dias et al., 2016).

The domain arrangement observed in REM proteins is more intricate compared to other families. Thirty-one members of the REM family contain multiple B3 domains: for example, PagREM6-A/B contained the highest number of five ( Supplementary Figure S4 ). The complexity of domain architecture in REM proteins suggests that they may have evolved from one or more domain duplication events, which can lead to functional diversification over the course of evolution. However, the study of genes with domain duplications is often challenging as the functional redundancy of many proteins can make it difficult to ascertain their specific roles (Waltner et al., 2005). Previously, domain duplication has been reported in the RAV family as well as other families of transcription factors. For example, in the liverwort Marchantia polymorpha, a RAV protein has been identified to contain two B3 domains, indicating the occurrence of domain duplication in this protein family (Swaminathan et al., 2008). Among the 167 basic helix-loop-helix (bHLH) proteins that have been identified in the rice genome, it has been discovered that one particular protein (OC_Os01g09930) contains two bHLH domains that are duplicated (Li et al., 2006).

Gene duplication is the primary source of the evolution of genes and gene families (Cannon et al., 2004). Previous research has demonstrated that the expansion of the B3 superfamily in Brassica rapa is primarily driven by tandem duplication, as evidenced by the presence of both two-copy and multi-copy tandem-arrayed B3 genes in its genome (Peng and Weselake, 2013). Duplicated B3 genes and other transcription-factor genes have also been found in the Arabidopsis genome (Riechmann et al., 2000; Romanel et al., 2009). Similar findings have been reported in Cicer arietinum, where CaARF4/CaARF5 and CaARF21/CaARF22 are duplicated in tandem but exhibit distinct structural differences (Die et al., 2018). Likewise, the REM family we identified is distributed on many chromosomes (Chr04A/B, Chr09A/B, Chr14A/B, and Chr15A/B). Tandem duplications appear to have occurred in 40 genes among 74 members of the REM family. In addition, only 44 of all the 160 B3 genes are products of tandem duplication events. Consistent with this, we found that the rest of the gene belongs to segmental duplication. This analysis suggests that segmental duplication may happen more frequently than tandem duplication in P. alba × P. glandulosa, and that numerous duplicated chromosomal blocks are retained in their genome after duplications (Cannon et al., 2004).

The B3 domain is not specific for angiosperms. It is also present in genes from gymnosperms, ferns, mosses, liverworts, and green algae (Marella et al., 2006). Evolution and genomic synteny can provide clues to gene function, and can be an effective tools for gene family analysis in the species with available whole-genome sequences (Lyons et al., 2008; Zhang et al., 2012). There is only a single B3 gene in Chlamydomonas reinhartii, a single-celled green algae with available complete genome sequence, and an ortholog of this gene has been identified in green algaes Mesostigma viride and S. muscicola, suggesting that the B3 domain arose before the development of multicellularity in the plant lineage (Merchant et al., 2007). This single B3 gene in C. reinhartii is similar in structure to the existing VAL subgroup in the most recent common ancestor of green algae and higher plants. Thus it is hypothesized that the C. reinhartii B3 gene is the ortholog of the VAL subgroup (Swaminathan et al., 2008). In the metaphyte lineage, a series of gene duplication events occurred before the speciation event between mosses and vascular plants. The first led to the LEC2-ABI3 subgroup, and subsequent gene duplications led to the RAV, ARF, and REM families (Swaminathan et al., 2008). Because of the evolutionary diversity and differences in genome polyploidy between different species from green algae to vascular plants, we selected seven species for collinearity analysis using orthologous genes. Eighteen B3 TF genes that were highly expressed in the xylem were selected for collinearity analysis. These genes can be found to be homologous in seven species and have collinearity on chromosomes, suggesting that these B3 genes were already present in early subaerial/terrestrial plants (Wang et al., 2020). For example, SM000256S08685 gene in S. muscicola, PAC:32923104 gene in P. patens, Aspi01Gene22980.t1 gene in A. spinulosa, and AT4G32010.1 gene in A. thaliana are orthologs of Pag.B16G000101 (PagVAL2-B1) gene in P. alba × P. glandulosa, belonging to the VAL subgroup. These findings suggest that the VAL genes have undergone a complex evolutionary history.

The B3 domain is highly conserved in B3 TFs, and it consists of 2 α-helices and 7 β-strands. Differences in amino acid residues may lead to different DNA binding ability or binding sequences of B3 proteins (Yamasaki et al., 2004; Waltner et al., 2005). The B3 domain of Arabidopsis VAL protein can recognize and bind to Sph/RY elements (CATGCA), where histone H3 lysine 27 trimethylation (H3K27me3) is marked and suppresses the transcription of downstream genes. When the B3 domain in VAL protein is destroyed, the expression of a downstream gene DOG1 is significantly upregulated, indicating that VAL proteins function through B3 domain-mediated binding (Chen et al., 2020). In addition, The AP2 and B3 domains in RAV1 protein can bind to the CAACA and CACCTG motifs, respectively, promoting the high-affinity activity of RAV1 in binding DNA sequences (Kagaya et al., 1999). In the ARF protein, the DNA-binding domain (B3 domain and Auxin-resp domain) is in the N-terminal region, while the protein-protein interaction domain (AUX_IAAs) is in the C-terminal portion (Wang et al., 2007). The promoter regions of B3 genes contain hormone-responsive elements such as ABA response elements (ABREs) and GA response elements (TCA motif and GARE sequences), and consistent some B3 TF genes are induced by hormone. The study on the ABA-dependent activation of the CRC (CRUCIFERIN C) promoter by the LEC1-NF-YC2 trimer indicates that ABRE motif is a necessary cis-element. Moreover, a seed-specific ABRE-binding protein, bZIP67, can functionally interact with LEC1-NF-YC2 to enhance the activation, while bZIP67 alone cannot activate the promoter. These results indicate that ABRE motif plays an important role in B3 protein-mediated regulation. (Curaba et al., 2004; Gazzarrini et al., 2004; Yamamoto et al., 2009).

The patterns of gene expression can serve as a critical indicator of the function of genes. In this study, some B3 genes exhibit tissue-specific expression patterns in P. alba × P. glandulosa: for example, PagARF10-A2, PagARF16-A2, PagARF17-A1, and PagARF11-A2 were found preferentially expressed in leaves, and PagVAL2-A1/A2/B1/B2 genes were preferentially expressed in xylems. In addition, 31 ARF genes were found to be expressed in all three tissues, among which nine were preferentially expressed in the differentiating xylem. Moreover, it seems that many ARFs function redundantly, thus single loss-of-function mutants do not exhibit growth and developmental phenotypes. In Arabidopsis, the arf7 arf19 double mutant has a strong auxin-related phenotype, including severely impaired lateral root formation and abnormal gravitropism in both hypocotyls and roots, which are not observed in the arf7 and arf19 single mutants (Okushima et al., 2005).

The VAL family genes PagVAL2-B1 and PagVAL2-A1 were highly expressed in the differentiating xylem of two age trees. Its specific expression in the xylem is consistent with our previous identification that it is a putative upstream of PagCAld5H2. Although studies have shown the essential role of VAL genes in regulating the transition from seed maturation to seedling growth, few of them have been studied for their functions in the process of tree development. As one of the most environmentally cost-effective and renewable sources, poplar wood has been widely used for timbering, paper making, and many other commercial applications (Wang et al., 2013). In trees, the secondary growth of stem and the lignin biosynthesis gives rise to wood formation (Zheng et al., 2021; Luo and Li, 2022). To understand their function, we investigated the co-expression gene network analysis, focusing on the one key module related to wood formation we identified, followed by pathways enrichment analysis. Our results from pathways analyses indicated that the differentially expressed genes (DEGs) across tissues and their corresponding network genes are enriched in many different pathways. Among which, PagVAL2-B1 and that corresponding network genes are enriched in regulation of SCWs biosynthesis and phenylpropanoid biosynthesis. This result suggests that PagVAL2-B1 can positively regulate secondary growth and lignin biosynthesis in poplar.The members belonging to the plant-specific B3-domain transcription factor family exhibit significant and diverse roles, particularly concerning vegetative and reproductive growth. We performed a detailed genome-wide analysis of 160 B3 family members based on the HMMER search and BLASTP method in P. alba × P. glandulosa. Motif, domain, promoter, and gene structure analyses were conducted for further identification. Gene duplication events and evolutionary analyses of B3 members were performed to illustrate the relationship within the P. alba × P. glandulosa and seven species genomes. Among the four B3 genes that were highly expressed in mature xylem, one (PagVAL2-B1), was studied for its co-expression relationships with other wood formation-related genes by WGCNA. Our analysis suggests that PagVAL2 family genes are involved in wood formation. Further using genetic transformation to study their specific roles in wood formation could help in the design to improve wood properties. Among the 160 B3 TF genes, 15% had incomplete coding region due to the annotation quality, and this may affect the gene structure analysis. The annotation needs further improvement to help the design in genetic transformation and CRISPR-based gene editing.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

MW, HL, and QL conceived the project. MW, HL, and QW performed the experiments. MW, HL, RL, and LY ran the program and visualized the data. MW wrote the draft, HL and QL edited the paper. All authors contributed to the article and approved the submitted version.