An endosperm-specific promoter from a glutelin gene Gt13a and its signal peptide were used in this study [14,17]. A human mature basic fibroblast growth factor (bFGF) gene was optimized with a rice codon bias. A total of 58 independent transformants was obtained using a two-strain Agrobacterium-mediated transformation [18] (Figure 1a). The positive transgenic plants were identified by polymerase chain reaction (PCR) analysis with specific primers (Figure 1b). Western blotting indicated that 35 (63.34%) transformants accumulated OsrbFGF in the rice endosperm (Figure 1c). The expression level of total OsrbFGF ranged from 99.11 to 185.66 mg/kg of brown rice from eight lines with high fertility (Figure 1d). Transgenic lines, 277–122 and 277–238, were the highest expressing lines and were used for further study.

The segregation for transgene expression was determined in the T1 seeds of the transgenic lines 277–238, 277–223, 277–122 and 277–179, genetic segregation in T1 seeds based on the expression: non-expression was analyzed. The results showed that 277–238, 277–223 and 277–179 showed a single locus and 277–122 two loci (Figure 2a). Southern blotting was carried out for further characterization of copy numbers in these transgenic lines. The results indicated that one fragment from transgenic lines 277–238 and 277–223 was detected by digestion of either EcoRI or HindIII, which is consistent with the results from genetic segregation. Digestion with EcoRI and HindIII of 277–122 yielded three fragments and two fragments, respectively. The results of genetic segregation and Southern blots of 277–122 indicated that it contained two loci. Although the transgenic line 277–179 showed two loci when digested with EcoRI or HindIII, the fragment obtained with HindIII was of smaller size to the entire expression cassette obtained with double digestion with HindIII/EcoRI. This suggested that it was a truncated locus. This further explained why 277–179 yielded two loci with Southern blotting, while genetic segregation of expression indicated a single locus (Figure 2 a,b). To further determine whether rice endosperm cells can properly cleave the signal peptide at the N-terminus of OsrbFGF, N-terminal analysis was carried out and revealed that two amino acids, Met and Ala, were missing at the N-terminus of OsrbFGF (Figure 2c).

After the highly expressing OsrbFGF transgenic lines were identified, we developed a purification protocol for large scale processing of OsrbFGF that could satisfy the demand of the markets. We firstly tried to separate the impurities using different ion exchange resins from rice seeds; however, all efforts failed (data not shown). Then we tried direct application of a heparin affinity column and washing with two different buffers that removed most of the impurities from the affinity column. A high purity OsrbFGF with a single band of 17 kDa was obtained (Figure 3). Endotoxin contamination was removed with one step of ultrafiltration using 5 kDa or 100 kDa filters. The resulting purity of OsrbFGF reached up to >95% and <1 EU/μg, respectively. The final yield of OsrbFGF was 8.33 mg/kg of brown rice with a recovery of 4.49%.

The NIH/3T3 cell line is generally used for evaluation of the biological activity of FGFs [19]. To evaluate the biological activity of OsrbFGF, stimulation of NIH/3T3 cells proliferation was tested. As shown in Figure 4a, OsrbFGF concentrations from 0.0007 to 12.5 IU/mL showed a dose-dependent cell proliferative effect on NIH/3T3 cells, reaching 1.04 × 10⁶ IU/mg. The result suggests that OsrbFGF has the same stimulation of cell proliferation as commercialized rbFGF.

bFGF is used for would healing therapy [20]. To evaluate its biological activities in vivo, OsrbFGF was applied to a mouse cut model for wound healing. Three dosages of OsrbFGF; i.e., 125 ng/treatment, 500 ng/treatment and 2000 ng/treatments were tested. Wound healing rates were observed on Day 0, Day 2, Day 4, Day 6, Day 8, Day 10, Day 12 and Day 14 after wound treatments. The results indicated that wound healing rates were dose-dependent. The efficacy of stimulating wound healing at a low-dose was more significant than that of 500 ng and 2000 ng within eight days after treatment (Figure 4b). In general, the recovery of wounded skin could promote epithelialization, increase expression of the CD68 and proliferation of cell nuclear antigen (PCNA) [21]. We therefore monitored the expression of PCNA and CD68 on the wound area on Day 3, Day 7 and Day 14 treatments from a dosage of 500 ng OsrbFGF. The results showed that CD68 and PCNA in the OsrbFGF treatments were increased significantly compared to that of the negative control on Day 3 and Day 7 of post-wound healing and was lower than that of post-wound healing in the control group on Day 14 (Figure 4 d,f). The results indicated that OsrbFGF could accelerate wound healing by stimulating cell proliferation and improving the expression of CD68 and PCNA in the early stage of treatment. It then decreased the expression to prevent the generation of scars and cell proliferation when the wound closed at the late stage of the treatment.

pOsPMP02 was used to construct an endosperm-specific expression cassette [14]. The coding sequence of the human bFGF gene (Genbank accession No. NM 002006) was synthesized by Heron Blue Biotechnology Inc. (Bothell, WA, USA) using rice-preferred genetic codons. The synthesized gene was cleaved by SchI and XhoI and then cloned into pOsPMP02 to produce pOsPMP276. A binary vector for Agrobacterium-mediated transformation was produced by digesting pOsPMP276 with HindIII and EcoRI and an entire expression cassette with a 1983-bp fragment containing the Gt13a promoter and the signal peptide, codon optimized bFGF gene and Nos terminator was inserted into a binary vector JH2600[25]. The resulting binary plasmid was designated as pOsPMP277. The binary plasmid was transferred into the Agrobacterium tumefaciens strain EHA105 harboring plasmid pOsPMP05 containing a selective marker gene [14], and pOsPMP277 were co-transformed into the calli regenerated from a rice variety TP309 by the Agrobacterium-mediated transformation as described previously [26].

To screen the positive transgenic lines, a forward primer F1 (5′-GAGGGTGTGGAGGCTCTTGT-3′) from the Gt13a signal peptide and a reverse primer R1 (5′-GCCAGTGAATTCCCGATCTAGTAAC-3′) from the Nos terminator were used for PCR amplification.

To obtain the expression levels of transgenic lines, Western blot was used. Briefly, 100 mg of T1 seeds from a PCR positive transgenic plant were ground with 1 mL of extraction buffer (66 mM Tris-HCl, pH 6.8, 2% SDS, 1 mM dithiothreitol (DTT)), and then centrifuged at 10,620× g for 10 min. Fifteen micrograms of the crude protein extracts and 350 ng of recombinant bFGF derived from E. coli were separated by a 15% SDS-PAGE gel and then transferred to a nitrocellulose membrane. A rabbit polyclonal to bFGF (Abcam, Cambridge, UK) in a dilution of 1:2000 and alkaline phosphatase goat anti-rabbit IgG (ZSGB-BIO, Beijing, China) in a dilution of 1:10000 were used. Detection of target protein was carried out using the substrates p-Nitro-Blue Tetrazolium Chloride and 5-Bromo-4-Chloro-3-Indolyl Phosphate (BIOSHARP, Hefei, China). Immunoblotting procedures were performed as described previously [17]. The expression level was determined from Western blotting by the Image-pro Plus software. Briefly, four concentrations of 50 ng, 100 ng, 200 ng and 400 ng of rbFGF derived from E. coli were used for quantitation of OsrbFGF level. The density of each known concentrations of rbFGF in Western blot was obtained by Image-pro Plus software. The expression level of each transgenic line was calculated based on the equation y = 2.0018* − 118.05 (R² = 0.98722).

About 200 milligrams of young leaf were ground with liquid nitrogen, and genomic DNA extraction was carried out using a TIANGEN DNA Quick Plant System (TIANGEN Biotech, Beijing, China). Genomic DNA was digested with EcoRI, HindIII, or EcoRI/HindIII (New England Biolabs, Ipswich, MA, USA), respectively, at 37 °C overnight and then was separated by 0.8% agarose gel. After transfer to a MILLIPORE NY + membrane, hybridization was carried out following the instructions of Roche DIG High Prime DNA Labeling and Detection Starter Kit I. A probe with an 824 bp fragment derived from the bFGF encoding region was prepared by amplifying with the primers F1 and R1.

The transgenic rice seeds were ground and extracted (1:5, w/v) with the extraction buffer (50 mM phosphate buffer, pH 7.5, 1 mM EDTA, 1 mM reduced L-Glutathione, 250 mM NaCl) for 1 h at room temperature. Then, the crude extract was clarified by the following 3 μm and 0.22 μm filtration through a positive pressure filter, respectively. Subsequently, the filtrate was applied onto the heparin 6 fast flow (GE Healthcare, www.gelifesciences.com) column equilibrated with 50 mM phosphate buffer (pH 7.5) containing 250 mM NaCl and 1 mM reduced l-glutathione. Two wash steps were used to remove the impurity proteins and increase the OsrbFGF recovery. Wash buffer 1 (50 mM phosphate buffer (pH 7.5), 600 mM NaCl, 1 mM reduced l-glutathione) and Wash buffer 2 (50 mM phosphate buffer (pH 7.5), 900 mM NaCl, 1 mM reduced l-glutathione) were used to remove impurities. After washing, the OsrbFGF was eluted by elution buffer (10 mM Tris-HCl (pH 7.5), 1.7 mM NaCl, 1 mM reduced l-glutathione). Before lyophilization, the OsrbFGF elution was concentrated and partially desalted using a 5 kD ultrafiltration filter with Pellicon XL (Millipore, Billerica, MA, USA), and then endotoxins were removed by a 100 kD ultrafiltration filter with a Viva Flow 50 (Sartorius, Goettingen, Germany). Finally, OsrbFGF was lyophilized with the protection of OsrHSA (1:50, w/w) [15]. The final protein concentration was determined using the Coomassie Plus (Bradford) Assay Kit (Thermo Fisher Scientific Inc., Rockford, IL, USA).

The purified OsrbFGF was transferred to a polyvinylidene difluoride (PVDF) membrane and then put into a protein sequencer PPSQ-31A (Shimadzu Cooperation, Kyoto, Japan) reactor to run 15 cycles.

The mitogenic activity of OsrbFGF was assessed through NIH/3T3 cell proliferation using the methylthiazoletetrazolium (MTT) method. NIH/3T3 cells were seeded in flat-bottom, 96-well plates at an initial density of 2 × 10⁵ cells per mL (200 μL per well) and cultured in Dulbecco’s modified eagle minimum essential medium (DMEM) supplemented with 10% FBS for 24 h and then cultured in fresh maintenance DMEM containing 0.75% FBS for 24 h. After the medium was removed, 200 μL maintenance medium containing a four-fold serial dilution of OsrbFGF or the recombinant bFGF standard with the initial concentration of 50 IU/mL was added into the wells and the cells were incubated for 48 h. The number of viable cells was determined by adding 20 μL methylthiazoletetrazolium (MTT 5 mg/mL) to each well. The culture was incubated for 4 h. After removal of the media, 150 μL dimethyl sulfoxide (DMSO) was added to each well. The plate was kept at room temperature for 30 min and was measured at 490 nm in a VersaMax 400 nm (Molecular Devices, Sunnyvale, CA, USA). All the data were analyzed with the GraphPad software. All experiments were repeated three times.

Female Sprague-Dawley rats (SD) with 200–250 g of body weight were obtained from Shanghai Slac Laboratory Animal Co. Ltd. The SD rats were randomly divided into four groups, and the excisional wound splinting model was generated as described previously [27]. Three dosages of OsrbFGF of 2000 ng/mL (n = 15), 500 ng/mL, (n = 15), 125 ng/mL, (n = 15) and negative control with normal saline (n = 15) were designed for testing the healing rates. The wound healing rate = (Area of original wound − Area of actual wound)/Area of original wound × 100.

For the immunohistochemistry assay, a dosage of 500 ng OsrbFGF was chosen. Fixation, sectioning and immuno-electronic microscopic observations followed the procedure as previously described [28]. A mouse antibody against CD68 or PCNA in a dilution of 1:100 and a goat anti-mouse IgG-HRP in a dilution of 1:200 were used. Finally, each slide was stained with the diaminobenzidine (DAB) substrate. After immunofluorescence staining, CD68 or PCNA positive cells were counted from three fields per section at the wound site between the edges in six successive sections.