The allotetraploid cotton cultivar Gossypium hirsutum (accession TM-1) was used in this study. Cotton plants were cultivated in a glasshouse with a photoperiod of 14 h/10 h (light/dark) and a temperature of 28°C of day and 20°C of night. Cotton boll age was determined by tagging each pedicel on the day of flowering. Cotton bolls were collected on selected days post-anthesis (DPA). Fresh cotton fibers attached to seeds were used for Agrobacterium infiltration.

To test whether genes can be expressed in cotton fibers via Agrobacterium-mediated transient transformation, binary overexpressing vectors were constructed. Coding sequences (CDS) of GhMYB212 (Gh_D11G3078) (Sun et al., 2019), GhCFE1 (Gh_A05G1404) (Lv et al., 2015), GhSusC (Gh_D06G0825) (Brill et al., 2011), and GhFSN1 (Gh_A12G1049) (Zhang et al., 2018) were cloned from G. hirsutum TM-1 and inserted downstream of the constitutive CaMV35S promoter in pBI121 vector, respectively. Additionally, the 1,944 bp promoter sequences of GhMYB212 and 2,035 bp promoter sequences of GhFSN1 were replaced by the CaMV35S promoter in the pBI121 vector, respectively, to construct the recombinant plasmids of pGhMYB212::GUS and pGhFSN1::GUS.

All the plasmids were introduced into the Agrobacterium cells as described previously (Chen et al., 1994). The primers used for vector construction are listed in Supplementary Table 1.

Transformed Agrobacterium was cultured at 28°C in liquid Luria-Bertani (LB) medium (supplemented with 50 μg/mL rifampicin and 50 μg/mL kanamycin) until the OD₆₀₀ reached 1.5. Then, the Agrobacterium cells were collected by centrifugation (4,000 rpm, 10 min) and resuspended with transformation solution [1/2 Murashige and Skoog (MS) medium supplemented with 165 μM acetosyringone and 3% sucrose (w/v), pH = 5.8] to an OD₆₀₀ between 0.8 and 0.9. Before the resuspension procedure, the transformation solution was degassed with a vacuum pump (Sanplatec Inc., Satoshi Kato, Japan) under 0.07 MPa for 3 min.

Cotton bolls at selected DPA were collected, sterilized with 70% ethanol, and carefully stripped off to avoid damaging to the fibers. Then, fibers attached to the seeds were placed into a conical flask containing the resuspended Agrobacterium solution. After being treated with a vacuum pump at a pressure of 0.07 MPa for 2 min, 0.01% (v/v) Tween 20 was added into the transformation solution. The cotton fibers were infected by Agrobacterium cells under dark (25°C and 50 rpm/min) conditions for a determined period of time. Later, cotton fibers were taken out and rinsed with sterile water more than five times until the water is clear. Then, the fibers were transferred to 1/2 MS medium [supplemented with 0.3% (w/v) plant gel, 50 μg/mL rifampicin, and 50 μg/mL kanamycin, pH = 5.8] and placed in an incubator with a temperature of 26°C/20°C (day/night) and a photoperiod of 16 h of light and 8 h of dark. After co-incubation, fibers were used for gene expression analysis, enzyme activity determination, and fluorescence signal observation. Figure 1 summarizes the procedure to generate transient transgenic cotton fibers.

Cotton fibers transformed with Agrobacterium containing p35S::GUS, pGhMYB212::GUS, or pGhFSN1::GUS constructs were subjected to GUS histochemical staining and enzyme activity determination, with fibers transformed with empty Agrobacterium as the negative control. The GUS staining and activity assay were performed according to Jefferson et al. (1987). Fibers on each seed were collected and used as one biological replicate. Three biological replicates were used for each reaction.

Coding sequences of enhanced GFP gene (eGFP) under the control of CaMV35S promoter were transiently transformed into cotton fibers. Fluorescence of eGFP was observed using a confocal laser scanning microscope (TCS SP8, Leica, Germany) with the 488 nm excitation and 495–530 nm emission wavelengths. Fibers transformed with empty Agrobacterium were used as the negative control.

Total RNA was isolated from transformed fibers using the Biospin Plant Total RNA Extraction Kit (Hangzhou Bioer Technology Co., Ltd., Hangzhou, China). The cDNA was synthesized from RNA using the HiScript III RT SuperMix (+gDNA wiper) (Vazyme, Inc., China) according to the manufacturer’s instructions. Gene-specific primers for quantitative reverse transcript PCR (qRT-PCR) analysis were designed using Primer Premier 5.0 software. The qRT-PCR reaction was performed on a 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, United States) using AceQ SYBR Green Master (Vazyme, Inc., China). The cotton ubiquitin 7 (GhUBQ7, accession number: DQ116441) was used as a reference for normalization (Li et al., 2018). Relative expression levels were calculated according to the method described by Livak and Schmittgen (2001). Three biological replicates were used for each sample. Detailed primer information is shown in Supplementary Table 1.

To test the application of subcellular localization in cotton fiber cells via the developed Agrobacterium-mediated transient transformation, a nuclear-localized protein histone B2, AtHTB2 (Zhang et al., 2020), and a plasma membrane-localized aquaporin, AtPIP2A (Qiu et al., 2020), were selected. Open reading frames of these two proteins without stop codon were fused with mCherry to generate 35S::AtH2B-mCherry and 35S::AtPIP2A-mCherry constructs, respectively. The recombinant vectors were transiently expressed in fiber cells by using the Agrobacterium-mediated method as described above. Subcellular localization was observed using a confocal laser scanning microscope at the 587 nm excitation and 610 nm emission wavelengths (TCS SP8, Leica, Germany).

As shown in Figure 1, procedures of transient expression of the protein in cotton fiber cells are involved in the screening of the target gene, Agrobacterium strain, fiber developmental stage, infection time, and co-incubation time. In the beginning, we selected eGFP as the target gene due to its ready observation. The plasmid containing 35S::eGFP was introduced into two commonly used Agrobacterium tumefaciens strains GV3101 and LBA4404, respectively. The cotton fibers at 20 DPA were selected to perform Agrobacterium infection. Fiber cells were infected with Agrobacterium for 3 h, followed by 3-day incubation. Strong eGFP fluorescence signals were observed in cotton fiber cells that were infected with GV3101 or LBA4404 (Figures 2A,B), indicating that the eGFP protein was successfully expressed in fiber cells. However, no fluorescence signals were observed in the control fibers, which were not infected with Agrobacterium (Figure 2C). At 10× magnification, the eGFP signals were similar between fiber cells infected with GV3101 and LBA4404; however, at 40× magnification, strong fluorescence signals were observed in the cytoplasm of fibers infected with LBA4404, while many weak signals in fibers with GV3101 (Figures 2A,B). Then, we used the GUS report system to further verify the transient expression of proteins in cotton fiber cells. GUS staining showed that the control fibers that were not infected with Agrobacterium exhibited weak signals (Figure 2D), which is consistent with previous reports that the enzyme β-glucuronidase is present in developing cotton fibers (Sudan et al., 2006). Further analysis demonstrated that fiber cells infected with LBA4404 or GV3101 harbors 35S::GUS construct showed much stronger GUS expression than that in the control fibers (Figure 2D). Additionally, LBA4404-inoculated fiber cells exhibited deeper staining than that in GV3101-inoculated fiber cells. GUS activity showed that fibers transformed with LBA4404 or GV3101 had higher GUS activity than the control (Figure 2D). Furthermore, LBA4404-inoculated fiber cells showed higher GUS activities than those in GV3101-inoculated fiber cells. These results suggest that the target gene could be transiently expressed in cotton fiber cells via the Agrobacterium-mediated transient transformation method, and the LBA4404 strain is more efficient to perform the fiber cell infection process. Thus, the LBA4404 strain was used in the following experiments.

To determine the optimal parameters for transient expression, the infection time was investigated using the GUS report system. Cotton fibers were infected with Agrobacterium carrying the 35S::GUS plasmid for 0, 2, 3, and 4 h, respectively. After incubation for 1 day, GUS staining was performed on these fibers. Compared with the weak GUS staining signal in cotton fibers without Agrobacterium infection (0 h), 2, 3, and 4 h infection resulted in a strong GUS signal in fiber cells (Figure 3A). GUS activity further showed that the activities in 3 and 4 h infected fibers were significantly higher than that in 2 h fibers, and there was no significant difference between 3 and 4 h infected fibers (Figure 3B). Therefore, we chose 3 h infection as the optimal parameter for our transient transformation system.

We also optimized the incubation time using the GUS report system. Cotton fibers infected with Agrobacterium for 3 h were harvested immediately (0 day) or were further incubated on medium for 1, 2, or 3 days. GUS staining and activity determination showed that the GUS enzyme activities in cotton fibers without incubation were very weak, whereas they were much higher in 1-, 2-, or 3-day incubated fibers (Figures 3C,D). GUS activities in 2 and 3-day incubated fibers were significantly higher than that in 1-d fibers, and there was no significant difference between 2 and 3 days incubated fibers (Figure 3D). Thus, 2-day incubation was selected in our transient transformation system.

Taken together, three parameters, including LBA4404 Agrobacterium strain, 3 h infection time, and 2 days incubation time, were determined to study the transient expression in cotton fibers.

We selected four previously characterized cotton genes that were preferentially expressed in fibers, including GhMYB212 (Sun et al., 2019), GhCFE1 (Lv et al., 2015), GhSusC (Brill et al., 2011), and GhFSN1 (Zhang et al., 2018) to study the expression levels by using the transient transformation method. Expression analyses in G. hirsutum TM-1 showed that GhCFE1 and GhMYB212 were preferentially expressed in fibers undergoing rapid elongation (5–15 DPA), while GhSusC and GhFSN1 were mainly expressed in fibers at the secondary cell wall thickening stage (15–25 DPA) (Figure 4A), similar to previous reports. The coding regions of these four genes were cloned and inserted downstream of the 35S promoter in the pBI121 binary expression vector to obtain 35S::GhCFE1, 35S::GhMYB212, 35S::GhSusC, and 35S::GhFSN1 constructs, respectively. Furthermore, the four constructs were individually transformed into Agrobacterium and used for subsequent transient transformation. Fibers at 20 DPA were subjected to transformation and were collected for gene expression analysis. Quantitative RT-PCR assay showed that the transcript levels of GhMYB212, GhCFE1, GhSusC, and GhFSN1 increased by 44.19, 1,535.33, 11.59, and 205.74 times in transiently transformed fibers in comparison to the control that transformed with empty pBI121 binary expression vector, respectively (Figure 4B). These results confirmed the success of our method in transient overexpressing target genes in cotton fibers.

To further test the efficiency of the method, we cloned the promoter regions (∼2 kb) of GhFSN1 and GhMYB212 into pBI121 vector upstream of GUS to make the recombinant plasmids, pGhFSN1::GUS and pGhMYB212::GUS, respectively. Cotton fibers at 10 and 20 DPA were transiently transformed with Agrobacterium and then were subjected to GUS staining and activity determination. The staining results revealed that weak signals were observed both in 10 and 20 DPA fibers without Agrobacterium infection (Figure 5A). The signal of pGhFSN1::GUS was higher in 20 DPA fibers than that in 10 DPA fibers; on the contrary, the signal of pGhMYB212::GUS was higher in 10 DPA fibers than that in 20 DPA fibers. GUS activity assays confirmed that the control fibers had low β-glucuronidase activities (Figure 5B). Again, GUS activities in 20 DPA fibers were significantly higher than that in 10 DPA fibers transformed with pGhFSN1::GUS, while GUS activities in 10 DPA fibers were significantly higher than that in 20 DPA fibers transformed with pGhMYB212::GUS. These results are consistent with the GhFSN1 and GhMYB212 endogenous transcript levels quantified by qRT-PCR in cotton fibers (Figure 4A), indicating that the Agrobacterium-mediated transient transformation method was suitable for promoter characterization in fiber cells.

Having validated that proteins could be transiently expressed in cotton fibers, we further investigated whether the expressed proteins were correctly subcellular localized in fiber cells. A nuclear-localized protein, histone B2 (AtHTB2) (Zhang et al., 2020), and a plasma membrane-localized aquaporin, AtPIP2A (Qiu et al., 2020), were ligated with mCherry protein at the N-terminal, respectively, to form 35S::AtHTB2-mCherry and 35S::AtPIP2A-mCherry constructs. Agrobacterium-mediated transient transformation with the 35S::AtHTB2-mCherry construct produced a strong fluorescence signal confined to nuclear (Figure 6A), while fiber cell expressing AtPIP2A-mCherry showed strong fluorescence at the plasma membrane (Figure 6B). These results suggested that the transient transformation method was suitable for subcellular localization studies in cotton fiber cells.