Upland cotton (Gossypium hirsutum L.) cultivar CRI24 that is relatively amendable for regeneration was used in this study. Cotton seedlings derived from the germination of surface-sterilized cottonseeds on MS medium were maintained in a growth chamber with a controlled environment for 7 days, with a 14/10 h photoperiod at 28°C. As previously described (Ge et al., 2015), the hypocotyls excised from the aseptic seedlings were inoculated with Agrobacterium tumefaciens strain LBA4404 cells harboring the 35S::GhLBDs or pER8:GhLBD18-1 plasmids. Upon the induction of calli, somatic embryos were generated and developed into plantlets. All the explants, calli, somatic embryos, and regenerated seedlings were maintained in a tissue culture room at 25–28°C under a 16 h light/8 h dark photoperiod.

Total RNA was isolated from 7-day-old sterile seedlings of G. hirsutum CRI24 plants using the Plant Total RNA Extraction Kit (TIANGEN Biotech, Beijing, China). The cDNA was synthesized using a PrimeScript RT reagent kit with gDNA eraser (Takara, Dalian, China). Primers (Supplementary Table 1) for GhLBDs were designed using Premier Primer 5 and used to amplify the full-length coding sequences of GhLBDs. The amplified products were each linked to the pBI121 expression vector under the transcriptional control of 35S CaMV promoter using the proper restriction endonuclease. The constructed vectors were introduced into LBA4404 strain, and positive clones were selected to transform the cotton hypocotyls with kanamycin as the selectable agent.

For expression analysis of GhLBDs, LRs, primary roots (PRs), and explants at different time points in the early stage of cotton callus formation were collected, from which total RNA and cDNA were prepared as described above. For quantitative real-time reverse transcription polymerase (RT-PCR), endogenous gene GhHistone 3 (Gh_D03G0370) was used as an internal reference gene. The quantitative RT-PCR was performed on the Applied Biosystems 7900HT system (Thermo Fisher Scientific, Waltham, MA, United States) using ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech, Nanjing, China). CT-method (2^–ΔΔCt) was used to quantify the relative expression level of the target genes. All the primer sequences are given in Supplementary Table 1.

Callus induction medium was supplemented with 2,3,5-triiodobenzoic acid (TIBA) or Indole-3-acetic acid (IAA)(2,4-D). To test the dose effects of auxin on the induction of LBD expression, a range of four different concentrations of TIBA, 0, 10, 20, and 30 μM, and a range of five different concentrations of 2,4-D, 0, 0.1, 0.3, 0.4, and 0.5 mM were used.

The hypocotyls of 7-day-old sterile seedlings of overexpressing GhLBDs, together with those of the negative control lines, were cut into segments of 1 cm in length and placed on CIM for 0–120 h prior to fixation in 50% FAA (50% ethanol, 10% formalin, 5% acetic acid, and 35% water) overnight. The fixed tissues were then embedded in paraffin prior to sectioning and microscopy observation previously described (Alexander et al., 2018). Selected calli were harvested following their culture on CIM medium for a period of 15, 25, and 45 days and prepared for the analysis of cell morphology under a HITACHI SU3500 scanning electron microscope (Hitachi, Tokyo, Japan).

LBD protein sequences from A. thaliana were used as queries for searching the homologs in G. hirsutum database using BlastP program, and the hits with e-values of 1e-5 were considered as significant (Zhang et al., 2015). As a result, 132 candidate LBD genes in G. hirsutum were identified. MEGA 7.0 was used to construct phylogenetic tree. PROSITE¹ and InterProScan² were used for domain search.

Sterile seedling hypocotyls were used as explants to induce callus initiation on CIM. The observation of callus initiation process indicated that there were no obvious morphological and cytological variations during the initial 48 h on CIM (Figure 1), suggesting that this could be a preparation stage for callus induction. The two ends of a hypocotyl became slightly enlarged, and the cell mass from the procambium and cambium began to form on the ends of hypocotyl following 72 h on culture, which marked the initiation of callus. This is congruent with previous studies that the calli of different explants mainly originated from the pericycle cells in the tissue (Ping and Howell, 2007; Atta et al., 2008). The callus cells were then quickly proliferated and became distinctively visible at 96 h (Figures 1A,B). Notably, one end of the hypocotyl produced healthy and large calli, in contrast to the other end with slowly growing and smaller calli in the first 144 h. Such an observation could be attributed to the variable auxin levels at the two ends of a hypocotyl, in which the auxin gradient polarity stemming from the differential distributions of the endogenous auxin had affected the callus initiation and development.

The knockout mutation of the WOX11-LBD16 pathway resulted in a complete loss of pluripotency in the callus of Arabidopsis (Liu et al., 2018), suggesting that some specifically expressed LBD genes during the cotton callus induction stage may also be crucial for callus proliferation and pluripotency acquisition, in addition to their known role in callus induction. In light of the transcriptome data, we have identified a number of annotated LBD genes that presented relatively a higher level of expression during callus initiation stage (Figure 2B), supporting the notion that LBDs play a key role in regulating callus initiation and growth.

It is generally recognized that phylogenetic analyses may enable the revelation of functional redundancy of gene families as the genes belonging to the same subgroup normally have overlapping functions. For further investigation into their potentially diverse or overlapping functions, all the annotated 131 LBDs were used to perform phylogenetic analyses (Supplementary Figure 1). LBDs were divided into eight subgroups, among which the aforementioned LBDs that were highly expressed during callus initiation stage were grouped into subgroup IV (Supplementary Figures 1, 2). By virtue of domain motif analysis, most members the subgroup IV were found to contain motif 8 that is unique to the subgroup, suggestive of functional divergence. This is well in line with a previous study that diversification of regulatory regions led to subfunctionalization or neofunctionalization (Li et al., 2005).

To further determine the evolutionary and homologous relationship, these above highly expressed GhLBD genes were used to perform phylogenetic analysis with Arabidopsis AtLBDs. As shown in Figure 2C, AtLBD16, AtLBD18, and AtLBD29 were clustered together with the aforementioned LBDs that were highly expressed during callus initiation stage. Based on the phylogenetic analysis, Gh_A11G1187/Gh_D11G1344, Gh_D08G1484/Gh_A08G1200, Gh_A08G0557/Gh_D8G0651, Gh_A11G1186/Gh_D11G1343, and Gh_A12G2597/Gh_D12G1018 were named as GhLBD29, GhLBD18-1, GhLBD16-1, GhLBD16-2, and GhLBD18-2, respectively (Figure 2C). PROSITE (see text footnote 1) and InterProScan (see text footnote 2) were used to search for the LOB domain in the obtained sequences and found that the LOB domain was highly conserved in LBD16, LBD18, and LBD29 between G. hirsutum and Arabidopsis (Figures 2A,C).

The regulatory network of controlling LR formation is highly conserved with that of controlling callus formation (Fan et al., 2012; Xu et al., 2018). The comparative expression of GhLBD16/18/29 homologous between the PR and LR was analyzed using qRT-PCR. As shown in Figure 2D, four out of the five candidates, including GhLBD16-1/2, GhLBD18-1/2, and GhLBD29, exhibited higher expression levels in LR relative to PR. Considering the similar regulation mechanisms between LR formation and callus development, expression patterns of these GhLBDs in the callus initiation process were detected by qRT-PCR (Figure 2D). Significant variation was observed in their expression patterns, including the expression abundance and peak point. The expressions of GhLBD16-1, GhLBD16-2, and GhLBD29 were similar, with a rapid response to the induction during the early initiation stage within 24 h. It is envisaged that GhLBD16 and GhLBD29 may function in concert by forming homodimers or heterodimers to induce cell proliferation (Lee et al., 2017). In contrast, GhLBD18-1 and GhLBD-2 showed relatively low expressions during the callus initiation process of 48 h. Notably, all the five GhLBDs showed considerably higher expressions at the late initiation stage from 72 to 144 h in comparison to the early initiation stages. Data shown in Figure 1 also suggested that 72 h might be the key time point for callus initiation and proliferation. The variation in peak time points and expression abundance among these GhLBDs are the manifestations of their functional redundancy and potential divergency. Such a premise is supported by the analysis of paraffin sections (Figure 1B) that exhibited conspicuous callus phenotype at 72 h stage.

To investigate GhLBDs’ response to exogenous auxin, five different concentrations of 2,4-D and three different concentrations of auxin transport inhibitor TIBA were added in the callus-induced medium. After 72 h induction, the expression of GhLBD16/18/29 was analyzed using qRT-PCR. Five GhLBDs genes showed very low expression without exogenous auxin treatment. A significant increase in GhLBDs’ expression was observed after exogenous auxin treatment (Figure 3B). The paraffin sections derived from the callus samples supplemented with different auxin concentrations indicated that 0.4 mM 2,4-D was the minimal amount to be effective in promoting callus initiation (Figure 3A), significantly enhancing GhLBDs’ expression at 72 h (Figure 3B). Lower concentrations of auxin with less than 0.4 mM were able to induce slightly, but not sufficient enough in inducing callus initiation at 72 h. TIBA treatment disrupted the localization and movement of auxin through impacting on auxin polar transport protein PINs (Cristian et al., 2010), resulting in the reduction in auxin level at the ends of hypocotyl explants. The low auxin level in turn cannot be sufficient enough in activating the expression of GhLBDs and delayed callus initiation at the ends of hypocotyl explants. As is evident in Figures 3C,D, TIBA treatment significantly inhibited the expression of GhLBDs and callus proliferation in cotton hypocotyl explants, in agreement with the notion that the auxin polar location is positively associated with GhLBDs’ expression and callus induction.

To confirm the functions of GhLBD16/18/29 homologs, transgenic cotton lines overexpressing GhLBD16/18/29 under the transcriptional control of 35S CaMV promoter were generated. To abrogate the confounding effects of auxin induction on GhLBD16/18/29 expression, the exogenous auxin was omitted from CIM to determine whether the GhLBD-induced callus growth is dependent on exogenous auxin. After subculturing on CIM without exogenous auxin for a period of 7 days, there was no callus formation discernible by naked eyes in both overexpression lines and controls that are transformed with pBI121 empty vector. Cytological examination on paraffin sections found small yet distinctive callus cell masses at the ends of hypocotyls of GhLBD18-1 transgenic lines, which was in contrast to the control where only the proliferated cells in the procambium and cambium were spotted (Figures 4A,B). This is clearly suggestive of the role of GhLBD18-1 in promoting callus initiation without the need for exogenous auxin. Similar observations have also been made with the transgenic lines overexpressing GhLBD16-1, GhLBD16-2, GhLBD18-2, or GhLBD29 (Supplementary Figure 3B). After being subcultured for 15 days on CIM without exogenous auxin, calli of considerable sizes were formed at the ends of hypocotyl explants derived from all the GhLBDs’ overexpressing lines, but no callus was formed in the control (Figure 4C and Supplementary Figure 4). Notably, both the GhLBDs overexpressing lines cultured on CIM without exogenous auxin and the control lines cultured on CIM containing exogenous auxin produced calli in similar sizes after 15 days in culture. At 25 days after culture, abundant calli were displayed in the transgenic lines overexpressing GhLBD18-1, in sharp contrast to the controls that only a few tiny calli were discernible. Consistent observations were made at 45 days in culture (Figures 4C,D). The morphology of the callus cells was analyzed using SEM. Despite the size difference (Figures 4C,D), the callus cells were all in globular shape in both GhLBD18-1 overexpression lines and control plants at 15-day subculture (Figures 4E,F). Rod-shaped cells were observed in GhLBD18-1 overexpression lines, whereas the control cells remained as globular or became oval shape at 25 and 45 days in subcultures (Figures 4E,F). Given both GhLBDs’ expression and exogenous auxin could induce callus initiation, the question arises as to whether these two different inducers could operate in a synergistic manner. To address this question, a comparison was made for the calli overexpressing GhLBD18-1 on cultures with and without exogenous auxin. Following 7 days in culture, the callus mass was substantially larger when exogenous auxin was added to the CIM medium for both transgenic and WT control (Figure 5). Corroborating results were also obtained in the transgenic calli expressing GhLBD16/18-2/29 (Supplementary Figure 3). Taken together, our results demonstrated that all the tested GhLBD genes, including GhLBD16/18/29, could not only promote the initiation and growth of callus cells independent of exogenous auxin, but also act synergistically with exogenous auxin in regulating callus proliferation.

Transgenic cotton lines overexpressing GhLBD18-1 were divided into two groups, i.e., high (H-type) and low (L-type), depending on whether the expression of GhLBD18-1 reached 10-fold of the WT level. As GhLBD18-1 was fused in frame with the β-glucuronidase (GUS) gene being controlled by 35S promoter in pBI121, the transgenic calli were stained blue in color (Figure 6A). According to the growth rate and morphology size of callus, we divided the callus into three types, namely, strong (S, enlarged size), intermediate (I, moderate size), and weak (W, small size) (Figure 6B). The calli derived from the H-type GhLBD18-1-overexpressing lines showed active growth and robustness on the CIM without exogenous auxin, in contrast to those of L-type transgenic lines. In addition, the numbers of S and I types of calli derived from the H-type transgenic lines were significantly increased, whereas the number of the W-type calli was obviously reduced (Figure 6C). For the lines transformed with an empty vector, callus initiation and growth were substantially inhibited, whereby most hypocotyl segments failed to produce discernible callus tissues at 30 days on CIM medium without exogenous auxin (Supplementary Figure 5).

Antibiotic and herbicide resistance genes are commonly used for positive cell selection in plant genetic transformation, but it is highly desired to develop an alternative selection approach in order to abrogate the needs for potentially harmful chemicals. Moreover, antibiotics that are used as selective agents could inhibit cell growth vigor and impose negative effects on cell proliferation and differentiation (Padilla and Burgos, 2010). The overexpression of GhLBD18-1, therefore, may represent an alternative selectable marker as it could give the transformed cells an advantage of cell growth over the untransformed cells. The overexpression of GhLBD18-1 driven by a β-estradiol-inducible promoter pER8 (Zuo et al., 2000) fostered positive callus cell growth as evidenced by the larger calli than the control calli at 30 days on CIM containing 10 μM β-estradiol (Figures 7A,B). Further examination of the large calli by qRT-PCR showed that more than 95% of them were genuine transgenic (Figure 7C). In essence, these results indicated that the overexpression of GhLBD18-1, coupled with β-estradiol as an induction agent, may serve as an effective and novel selection system for gene transformation in cotton.

The cotton GhLBD gene family is composed of 131 members that could be classified into eight subgroups via phylogenetic analysis (Supplementary Figure 1). As cotton is an allotetraploid plant species, many of these genes could be in homeologous pairs (Yang et al., 2019). By virtue of the expression analysis in callus induction stage, five preferentially expressed GhLBD genes, including GhLBD16-1, GhLBD16-2, GhLBD18-1, GhLBD18-2, and GhLBD29, were selected as candidates for study, and their potential functional role was explored in relation to callus induction and callus growth vigor. The five selected GhLBD genes were all significantly induced by exogenous auxin or inhibited by auxin transport inhibitor TIBA (Figure 3), suggesting that these GhLBDs may act downstream of auxin regulation and play a similar role in regulating callus formation and development. Differential expression of these GhLBD genes was illustrated, whereby GhLBD16-1, GhLBD16-2, and GhLBD29 presented similar expression patterns that are significantly induced in the early initiation stage (1–144 h), whereas GhLBD18-1 and GhLBD18-2 expressed rather late, with a relatively high level of expression at 72–144 h following the culture initiation. A similar study also proved that LEAFY is a pioneer transcription factor in promoting floral fate (Jin et al., 2021). fgfbp genes family showed temporal and spatial expression patterns to maintain zebrafish embryo development (Li et al., 2018). Our results are congruent with previous studies that Arabidopsis orthologs of GhLBD16, GhLBD17, GhLBD18, and GhLBD29 functioned redundantly in triggering callus formation (Fan et al., 2012). Because these GhLBD genes were grouped in the same clade in phylogenetic analysis (Supplementary Figures 1, 2), functional overlapping or divergency to some extent could be anticipated, as elucidated in previous studies (Mangeon et al., 2012). Overexpression of each of the five GhLBD genes in cotton was sufficient to trigger callus formation without exogenous auxin (Figure 4 and Supplementary Figures 3–5), suggestive of functional redundancy, but detailed examination on their individual role in modulating callus initiation and proliferation requires further exploration.

Gene expression is often closely related to its function and is commonly used to infer functionality, although it may not be directly proportional. Appropriate upregulation of BOC1 expression could reduce callus browning in rice, both high and low expressions of BOC1 led to serious callus browning and increased ratio of browning callus (Zhang et al., 2020). In transgenic cotton plants overexpressing GhLBD18-1, callus size and growth vigor were found to have a direct positive correlation with transgene expression level, whereby about 60 vs. 29% large callus, and about 35 vs. 26% moderate-sized callus in highly expressed GhLBD18-1 lines and moderately expressed GhLBD18-1 line, respectively (Figure 6). Altogether, these results indicated that GhLBD18-1 expression promotes callus growth in a dose-dependent manner. However, excessive callus proliferation could be unfavorable for cell pluripotency acquisition, and appropriate level of GhLBD18-1 expression may be required to ensure the optimal callus proliferation and cell fate transition to embryogenic callus.

The promotion of callus initiation and proliferation independent of exogenous auxin by overexpressing GhLBD in cotton suggested their functional role downstream of auxin regulation for callus growth. To avoid the potential pleiotropic effects and precise modulation of callus growth, a β-estradiol-inducible promoter pER8 was employed to upregulate GhLBD18-1 expression when it is required. It is also known to have moderate strength relative to 35S CaMV promoter (Schlücking et al., 2013). Furthermore, we have also examined the feasibility of using GhLBD18-1 as a selectable marker to substitute for the currently used antibiotics or herbicide-resistance marker gene. The abrogation of antibiotic and herbicide screening would enable calli to thrive on CIM without chemical stress. Moreover, the use of antibiotics or herbicide resistance in screening for transgenic plants is accompanied by the frequent escape of non-transgenic cells that may form a chimeric tissue with transgenic cells, causing undesirable complications in the selection and lowering transgenic efficacy. Furthermore, there are ever-increasing public concerns for the unintended gene flow of chemical resistance into the environment, which also results in the heavy regulatory burden and prolonged delays in releasing transgenic crops. Therefore, the employment of GhLBD18-1 coupled with pER8 has clear advantages as a promising transgenic selection system in plant genetic transformation relative to the classic antibiotic/herbicide-based selection systems.