The materials ascorbic acid (Vit C), butylated hydroxytoluene (BHT), bovine lactoferrin (bLF) and EDTA used for antioxidant activity assays were purchased from Sigma (MO, USA).

A full-length cDNA fragment (2061 bp) of porcine lactoferrin (accession number NM 214362) driven by a rice actin promoter (1255 bp) [52] and Nicotiana tabacum RB7 MAR DNA fragment (accession number U67919) were first constructed in pUC18 and then inserted into the HindIII site of the vector pCAMBIA1300 (accession number AF234296) to form pCAM-Act-MLF (Fig. 1). This plasmid vector was then transferred to Agrobacterium tumefaciens strain EHA105 for rice transformation. The hygromycin phosphotransferase (hpt) gene driven by the 35S promoter was used as a selection marker.

One hundred grams of mature rice grains from transgenic rice were ground in liquid nitrogen in extraction buffer (20 mM Tris/HCl, 1 M NaCl, and 0.1% (v/v) Triton X-100, pH 7.4). The extracted proteins were partially purified by 4-55% ammonium sulfate precipitation; the precipitated proteins were collected by centrifugation and then resuspended and dialyzed against the initial buffer to remove excess salts. These proteins were further purified through a heparin-Sepharose column with an AKTA purifier 10 system (Amersham Pharmacia Biotech, Inc., UK). The amounts of purified protein were detected by the Bradford method using a protein assay kit (Bio-Rad, USA).

The antibody to be used against rpLF was raised in a rabbit, and an anti-LF antiserum was prepared and purified as described by Wu et al. [53], where its specificity against both the E.coli-expressed and rice-produced rpLFs had been evaluated (Fig. S2). Proteins from whole grains of TNG67 and rice bran as well as the whole grains and leaves of transgenic lines were extracted as described by Lee et al. [31]. The protein samples were then fractionated on a 10% SDS-PAGE gel and transferred onto a nitrocellulose membrane with a BioRad Trans-Blot system according to the manufacturer’s instructions. The membrane was first probed with primary anti-LF antibody and then treated with secondary anti-rabbit antibody conjugated with alkaline phosphatase (Sigma, MO, USA). After antibody recognition, the membrane was incubated with NBT (nitro blue tetrazolium chloride, 50 mg/mL) and BCIP (5-bromo-4-chloro-3-indolyphosphate, 50 mg/mL) in 10 mL of alkaline phosphatase buffer (100 mM NaCl, 5 mM MgCl₂, and 100 mM Tris-HCl, pH 9.5) for color development [54].

The microtiter plates were coated with 50 μl of protein overnight at 4°C. After blocking with 2% nonfat milk in PBS for 2 hours and rinsing three times with PBS, the rpLF antibody (50 μl) was added to each well and incubated for 2 hours. Then, 50 μl of alkaline phosphatase conjugated anti-rabbit IgG (Sigma-Aldrich, USA), the secondary antibody, was added. The reactions were then incubated with p-nitrophenyl phosphate substrate (Sigma, MO, USA) in diethanolamine buffer (pH 9.8) for 1 hour in the dark at room temperature. The absorbance at 405 nm was read in a Bio-Kinetic Reader EL 312e (Bio-Tek Instrument Inc., VT, USA).

A MALDI-TOF-MS instrument mode voyager ED PRO (Applied Biosystems, USA) equipped with a nitrogen laser radiating at 337 nm was used to obtain protein mass spectra. The MALDI matrix was prepared as a saturated solution of SA (sinapinic acid) in 50% acetonitrile containing 0.1% TFA. One microliter of the mixture was spotted on the sample plate and allowed to dry. The spectra were acquired in linear mode at an acceleration voltage of 25 kV, 90% grid voltage, 0.15% guide wire voltage, and 750 nsec delay with external calibration. The spectrum was obtained by accumulating an average of 200 shots in the positive ion mode. The N-terminal sequence of desalted protein samples was analyzed by SDS-PAGE following a reported protocol and sequenced using a Procise protein/peptide sequencer from Applied Biosystems.

Fig. (1). Schematic diagram of plasmid constructs pCAM-35S-LF, pCAM-Act-LF and pCAM-ActM-LF. Plasmid constructs in the vector pCAMBIA1300 (8958 bp, accession number AF234296) with a porcine lactoferrin (LF) cDNA fragment (2061 bp, accession number NM214362) driven by A) the cauliflower mosaic virus (CaMV) 35S promoter (850 bp, 35S-P) and B) a rice actin promoter (1255 bp, Act-P) and 3’-NOS terminator (300 bp) were created for rice transformation. C) In the pCAM-ActM-LF construct, a tobacco RB7 MAR DNA fragment (1100 bp, accession number U67919) was added upstream of the actin promoter.

The N-glycosylation site was analyzed using trypsin digestion followed by the addition of PNGase F (N-glycosidase F). Both deglycosylated and non-deglycosylated samples were subjected to LC-MS/MS analysis. LC-MS/MS was performed with Q Exactive using a Dionex Ultimate 3000 RSLC pump (Thermo Scientific, USA) coupled with an Acclaim PepMap RSLC C18 column (2 μm, 75 μm x 150 mm, Thermo Scientific, USA) at a flow rate of 0.25 μL/min using the gradient shown in Table S1. Xcalibur 2.2 was used to obtain the BPI chromatograms. Raw data were processed into peak lists by Thermo Proteome Discoverer 1.4 and searched in the Mascot database (Fig. S1A- S1C).

Periodic acid-Schiff (PAS) staining for glycoprotein detection was performed using the Thermo Scientific Pierce Glycoprotein Staining kit (Thermo Scientific, USA). In the assay, protein samples were first fractionated on a 10% SDS-PAGE gel and then either stained with Coomassie blue for the control or treated with a staining kit according to the manufacturer’s instructions.

Biotinylated Vicia Villosa Lectin (VVL) and Peanut Agglutinin (PNA) purchased from Vector Labs (Vector Labs, USA) were used for Tn-antigen or T-antigen (mucin-type O-linked glycoprotein) detection, respectively. For detection, the protein samples were fractionated on gels and transferred onto nitrocellulose membranes. Then, to block nonspecific binding, the membranes were incubated in a carbo-free blocking solution (SP-5040) for 30 minutes then incubated in PBS containing 2-20 µg/ml biotinylated VVL or PNA for 30 minutes at room temperature. VECTASTAIN ABC-AP (alkaline phosphatase, AK-5000) reagents and BCIP/NBT (SK-5400) were applied according to the manufacturer’s instructions, resulting in black bands with a colorless background.

The ferrous ion chelating activity of the rice-expressed rpLF was investigated according to the method of Dinis et al. [50]. Briefly, purified rice-expressed rpLF was added to a solution of 2 mM FeCl₂ and methanol. Then, the reaction was initiated by the addition of 5 mM ferrozine; the mixture was shaken vigorously then left standing at room temperature for 10 min. After the mixture had reached equilibrium, the absorbance of the solution was measured spectrophotometrically at 562 nm, where the Fe²⁺ chelating ability of rpLF was monitored by measuring the ferrous ion-ferrozine complex at 562 nm. The percentage of inhibition of ferrozine-Fe²⁺ complex formation was given by the following formula:

Ferrous ions chelating (%) = [(A₀-A₁)/A₀] x 100

where A₀ is the absorbance of the control, and A₁ is the absorbance in the presence of the rpLF sample. The control contains FeCl₂ and ferrozine complex formation molecules. In addition, EDTA was used as a positive control, and bLF was included for comparison.

Lecithin (300 mg) was sonicated in an ultrasonic cleaner in 30 mL of 50 mM phosphate buffer (pH 7.4) for 4 hours. FeCl₃ (0.5 mL, 1 mM), ascorbic acid (0.5 mL, 400 mM) and rpLF were then added to the sonicated solution (0.5 mL, 10 mg lecithin/mL). The mixture was incubated for 1 hour at 37°C, and oxidation was measured by the thiobarbituric acid (TBA) method [55]. The absorbance of the sample was read at 532 nm against a negative control, which contained all reagents except lecithin. The percent inhibition formula is given below:

Inhibition of lipid peroxidation (%)= [1- (A₁/A₀)] x 100

where A₀ is the absorbance of the negative control reaction and A₁ is the absorbance in the presence of rpLF. In addition, the synthetic antioxidant butylated hydroxytoluene (BHT) was used as a positive control, and bLF was included for comparison.

The reductive capability of pure rpLF was quantified by the method of Oyaizu [56]. The pure rpLF was mixed with phosphate buffer (0.2 M, pH 6.6) and potassium ferricyanide (1%). The mixture was then incubated at 50°C for 20 minutes, and the reaction was terminated by adding a portion of trichloroacetic acid (10%). The upper layer of solution was mixed with FeCl₃ (0.1%), and the absorbance was measured at 700 nm in a spectrophotometer against ascorbic acid (Vit C), which was used as a control. The bLF was included for comparison.

The free radical scavenging activity of pure rpLF was measured by DPPH using the method of Blois [57]. A 0.1 mM solution of DPPH in ethanol was prepared, and pure rpLF at different concentrations was added. After 30 minutes, the absorbance of the reaction mixture was measured at 517 nm against BHT, which was used as a control. A low absorbance of the reaction mixture indicated high free radical scavenging activity and was calculated using the following equation:

DPPH scavenging effects (%)=100- [(A₀-A₁/A₀) x100]

where A₀ is the absorbance of the control reaction, and A₁ is the absorbance in the presence of pure rpLF. The synthetic antioxidant butylated hydroxytoluene (BHT) was used as a positive control, and bLF was included for comparison.

The results were evaluated as mean values of three times of the same sample, and data are represented as the mean ± standard deviation (mean ± SD).

Fig. (2). The protein expression levels of rpLF in various transgenic rice lines. A) The relative amount of rpLF from the transgenic rice lines pCAM-Act-LF and pCAM-ActM-LF was compared to that of pCAM-35S-LF (set as 1). Data were measured from the Western blot results shown below. B) Western blot analysis of rpLF from the transgenic rice lines pCAM-35S-LF, pCAM-Act-LF and pCAM-ActM-LF and non-transgenic (NT) rice. The same amount of total extractable protein (45 μg) was analyzed.

In the present study, the expression levels of recombinant porcine lactoferrin (rpLF) in transgenic rice lines driven by either the 35S promoter (35S-LF), actin promoter (Act-LF) or actin promoter in combination with tobacco nuclear matrix attach region RB7 MAR [58] sequences (ActM-LF) were evaluated (Fig. 1).

Through agrobacterium-mediated transformation, more than 20 independent transgenic rice lines were obtained for each construct, and at least 3 homozygous lines of each construct were selected for characterization for more than 10 generations. Typically, the homozygous lines with one T-DNA insertion event can be obtained within 2 or 3 generations of hygromycin selection, of which selection can be confirmed when approximately 20-40 progeny seeds from each generation are all germinated normally under hygromycin selection. The representative Western blot results of transgenic lines with three different constructs and their relative amounts of rpLF were measured and are presented (Fig. 2A, B). The expression levels of rpLF in ActM-LF transgenic rice were approximately 2 and 6 times higher than those in Act-LF and 35S-LF, respectively (Fig. 2B). Thus, the ActM-LF transgenic line was chosen for rpLF protein purification and further characterization.

Purification of rpLF from rice grains yielded 815 mg of total extractable protein and 987 μg of rpLF, representing approximately 0.12% purity in this protein fraction (Table 1). These extracted proteins were further purified through 4-55% ammonium sulfate precipitation, and the rpLF purity reached 0.78% (6.5-fold improvement) with an approximately 25% loss of rpLF (Table 1). After rpLF was passed through the heparin-Sepharose column with an AKTA purifier, 326 μg of protein was eluted in the 0.34 M NaCl fraction, indicating approximately 100% purity and representing 33% recovery (Table 1). The purity of rpLF in all purification steps was analyzed by Coomassie blue-stained SDS-PAGE and Western blot assays (Fig. 3). In the Coomassie blue-stained gel, many protein signals appeared in the first three purification steps with 20 μg protein samples (Fig. 3A, lanes 1-3), and only one rpLF protein band was observed in the column-purified sample loaded with 1 μg of protein (Fig. 3A, lane 4). The column-purified rpLF was collected from eluted fractions which contains a higher concentration of the target protein, e.g. the fractions E4 to E9 (Fig. S3). These high concentrated protein fractions were then pooled and desalted and concentrated further through Amicon 30K UFC903008 (Merck Millipore, USA). In the Western blot assay, rpLF bands were observed for all protein samples loaded with either 20 μg for the first three steps (Fig. 3B, lanes 1-3) or 0.01 μg for the column-purified step (Fig. 3B, lane 4). In summary, column-purified proteins had an rpLF purity of approximately 100% based on the purification data and protein gel analysis results.

To determine if the rice-expressed rpLF could be correctly processed as it is in porcine, the purified rpLF was analyzed by the Edman degradation assay. The results of the Edman degradation assay revealed that the N-terminal sequence of rpLF starts with APKK (Fig. S1-A), demonstrating that the 19 amino acid signal peptide of rpLF could be removed correctly in transgenic rice.

The same amounts of purified rpLF, horseradish peroxidase (a glycoprotein, used as a positive control), soybean trypsin inhibitor (negative control) and bovine lactoferrin (bLF) were fractionated on a 10% SDS-PAGE gel and stained with Coomassie blue (Fig. 4A) or treated with PAS (Fig. 4B). The results showed that rpLF is a glycoprotein (Fig. 4B, lane 3) as it was stained with PAS; however, rpLF contained fewer polysaccharides than bovine lactoferrin (Fig. 4B, compare to lane 4). Furthermore, the MS/MS assay demonstrated that two potential N-glycosylation sites located at N365 and N472 were glycosylated in the purified rpLF (Fig. S1B-S1C).

Fig. (3). SDS-PAGE and rpLF identification in various purification steps. A) Coomassie blue-stained SDS–PAGE for crude extraction (20 μg, lane 1), supernatants of 4% ammonium sulfate treatment (20 μg, lane 2), precipitates of 55% ammonium sulfate treatment (20 μg, lane 3), and column-purified rpLF (1 μg, lane 4). The molecular weight markers are indicated in kDa. B) Western blot analysis of rpLF with the same amount of protein (20 μg) as above, except 0.01 μg of column-purified rpLF (lane 4) was used.

Fig. (4). Glycosylation stain polyacrylamide gel electrophoresis. A) Polyacrylamide gel stained with Coomassie blue; the same amount of horseradish peroxidase (lane 1, positive control), soybean trypsin inhibitor (lane 2, negative control), rpLF (lane 3, indicated by arrow) and bLF (lane 4) was subjected to SDS-PAGE and stained with Coomassie blue. Lane M, molecular mass markers. B) The same amount of protein samples from A) stained with periodic acid-Schiff. Lane M, molecular mass markers.

To determine the accurate molecular mass of rpLF and to check if rpLF contains multiple forms in transgenic rice, the MALDI-TOF MS assay was performed. At least two major peaks with molecular masses of 76652.5 and 77269.1, slightly heavier than the calculated molecular weight of 75350, were revealed in the assay (Fig. 5A). Similarly, the SDS–PAGE analysis resolved multiple bands for purified rpLF under nonreducing conditions (Fig. 5B), indicating the heterogeneity of the molecular mass of rpLF and corroborating the multiple-peaks observed by the MS assay.

Fig. (5). Molecular mass and glycosylated isoform identification of purified rpLF using MALDI-TOF and PNGase treatment. A) Purified rpLF was subjected to MALDI-TOF, and two peaks with molecular masses of 76.7 and 77.3 kDa were identified. B) Coomassie blue-stained SDS–PAGE of purified rpLF (1 μg) and rpLF, which were incubated with PNGase F at 37°C for 2 hours (rpLF+PNGase F). The molecular weight markers are indicated in kDa.

Fig. (6). VVL and PNA stained after polyacrylamide gel electrophoresis. A) Polyacrylamide gel stained with Coomassie blue; the same amount BSA (lane 1), rpLF (lane 2, indicated by arrow) and bLF (lane 3) was subjected to SDS-PAGE and stained with Coomassie blue. Lane M, molecular mass markers. B) Stained with VVL; the same protein samples stained with VVL. The arrow indicates rpLF. C) Stained with PNA; the same protein samples stained with PNA. The arrow indicates rpLF.

To characterize whether these molecular mass isoforms were caused by glycosylation, rpLF was treated with PNGase F glycosidase and subjected to SDS–PAGE analysis. The results showed that the PNGase F glycosidase-treated rpLF formed a single band with a reduced molecular mass (Fig. 5B), indicating that glycans in rpLF were removed by PNGase and revealing different degrees of glycosylation.

To confirm that rpLF contains mucin-type O-linked glycans, biotinylated Vicia Villosa Lectin (VVL), which can detect the Tn antigen (GalNAcα-Ser/Thr), and Peanut Agglutinin (PNA), which can detect the T antigen (Galβ1-3GalNAcα1-Ser/Thr), were used as both the Tn antigen and T antigen are mucin-type O-linked glycoproteins. In these assays, both rpLF and bLF showed positive signals in VVL and PNA stains (Fig. 6B and C). rpLF showed a weaker signal than bLF in the VVL stain (Fig. 6B) but a stronger signal than bLF in the PNA stain (Fig. 6C), suggesting that O-linked glycosylation is present in rpLF and bLF and that more T-antigen glycosylation appeared in rice-produced rpLF.

To confirm the antioxidant activity of rice-produced rpLF, four methods including 1) metal chelating capacity assay, 2) lipid peroxidation inhibition assay, 3) reducing power assay and 4) free radical scavenging capacity assay testing various molecular mechanisms were used.

The present study compared the expression levels of rpLF in Act-LF to 35S-LF and ActM-LF, two additional transgenic lines. The expression level of rpLF in 35S-LF was approximately one-third (1/3) that in Act-LF. Although the CaMV 35S promoter has been widely used and well characterized in transgenic plants [59, 60], including dicotyledonous plants, such as tobacco [61], and monocotyledonous plants, such as rice [62], the levels of gene expression produced by this promoter in cereals were much lower than those in dicots [63]. In contrast, the rice actin promoter was an efficient promoter in transgenic rice [52] and other cereal crops [64]. The present study showed that the expression of rpLF in transgenic rice was higher with the rice actin promoter than the 35S promoter, which is consistent with previous reports [35, 63].

Fig. (7). Antioxidant activity assays of rpLF in comparison to bovine LF (bLF) and known controls. A) The ferrous iron chelating capacity of rpLF, bLF and EDTA. B) Lipid peroxidation assay of rpLF, bLF and BHT. C) Reducing power assay of rpLF, bLF and ascorbic acid (Vit C). D) DPPH scavenging activity assay of rpLF, bLF and BHT. Each value represents the mean±SD (n=3).

Transgenes flanked with MAR sequences have been shown to be able to achieve a relatively higher level of transgene gene expression in transgenic plants [36-40]. Expression levels have been shown to increase up to 60 times using RB7 MAR in transgenic tobacco plants [36]. However, the same RB7 MAR sequences only resulted in an approximately 2.5-fold increase when applied in transgenic rice plants [40]. In the present study, the expression levels of rpLF in the transgenic rice line ActM-LF (with a vector driven by an actin promoter fused with one copy of the tobacco RB7 MAR DNA fragment) showed an approximately 2-fold increase compared to Act-LF, indicating that the effect of RB7 MAR on the expression of the porcine lactoferrin gene and the GUS gene in transgenic rice was similar, as previously observed [40].

While the transformation technology for producing recombinant pLF in transgenic rice has been well researched [31], the process to isolate pure rpLF for functional characterization and a robust manufacturing method have not been fully developed. In the present study, high salt crude extraction, ammonium sulfate cutting and column purification were utilized to achieve high levels of rpLF purification. The crude extraction with 20 mM Tris and 500 mM NaCl yielded total extractable protein at approximately 1% of the grain weight, which was less than the theoretical ~7% of total protein content, and the amount of rpLF determined by ELISA was approximately 0.12% of the total extractable proteins, which was 8 times less than the amount (approximately 1%) reported previously [31]. The low yield of rpLF and total extractable protein in the present study could be attributed to using a larger sample scale and different extraction buffer than in previous studies [31]. After crude extraction, purification was performed with 4-55% ammonium sulfate cutting, a series of empirical results, and a heparin-Sepharose column to reach nearly 100% purity with approximately 33% recovery from the crude extraction step. The chromatography step in this purification scheme removed most nontarget proteins and enhanced the purity of rpLF from 0.78% to 100% at the cost of approximately 55% loss in rpLF. In summary, the chromatography step significantly enhanced the purity of rpLF and shortened the purification period. Based on this purification scheme, a standardized purification protocol could be further developed as a basis for potential commercial production.

The present study demonstrated that the signal peptide of rpLF could be precisely removed and that rpLF is a glycoprotein in transgenic rice. The MS/MS assay further showed that two potential N-glycosylation sites located at N365 and N472 were glycosylated in the purified rpLF. Similarly, N-linked glycosylation and the glycan structures of a rice-derived recombinant human lactoferrin (rhLF) have also been characterized [18, 65]. Although the core glycan structure of rhLF is different from natural hLF, their biochemical and biophysical properties are similar [18]. In addition, rice-derived rhLF retains many of its biological activities [32]. A previous study also demonstrated that rice-derived rpLF retained iron-binding and antimicrobial activities [31]. All these observations support the notions that the post-translation process and N-glycosylation of recombinant rLFs, regardless of their origins, occurred in transgenic rice and that the biological activities of natural LF could be retained in rLFs.

In the present study, purified rpLF showed two different molecular mass peaks in the MALDI-TOF-MS assay and multiple bands in SDS–PAGE under nonreducing conditions, indicating the heterogeneity of the molecular mass of rpLF. These molecular mass differences were possibly caused by various degrees of post-translational modifications, such as glycosylation in transgenic rice. After treatment with PNGase F glycosidase, a single band with a reduced molecular mass was observed, suggesting that different degrees of rpLF glycosylation existed in transgenic rice. Taken together, the results indicate that the signal peptide of rice-produced rpLF could be correctly processed and that different degrees of N-glycosylated rpLF existed in the transgenic rice, causing molecular mass heterogeneity.

Although mucin-type O-glycosylation has not been described in plants [66], a report showed that protein subpopulations from alcohol-extracted rice grains were recognized by biotinylated VVL and PNA, suggesting that mucin-type O-glycosylation was present in rice [67]. A recombinant human GM-CSF derived from suspension-cultured rice cells showed various O-glycans attached to three O-glycosylation sites [68]. In the present study, the observations of rpLF with VVL and PNA blotting suggested that mucin-type O-glycosylation was present in rice and could imply that rice grains have the machinery to perform GalNAc O-glycosylation. However, since the detection of O-linked glycoprotein by VVL and PNA staining methods was indirect, the observations could be possible artifacts. Confirming whether rpLF contains mucin-type O-glycans and whether rice grain possesses machinery to perform GalNAc O-glycosylation will require further investigation.

Although LFs from different animal sources, such as humans [48], bovines [45-47] and camels [49], displayed antioxidant activity, whether recombinant LF produced from transgenic plants also retains antioxidant activity has not yet been fully demonstrated. In the present study, the Trolox equivalent antioxidant capacity, DPPH free radical scavenging capacity, ferrous iron chelating capacity and reducing the power of rpLF were analyzed. The reducing DPPH radical-scavenging activity of purified rpLF was compared to that of bLF and BHT. rpLF showed the same radical-scavenging activity as bLF while possessing approximately 49% of the antioxidant properties of BHT. Further experimentation also showed that rpLF had a nearly equivalent capacity to bLF in terms of reducing power, iron chelating activity and lipid peroxidase activity.

The present study therefore infer that rpLF produced from transgenic rice can be considered as containing equivalent antioxidant activity as bovine LF and recognized as a plant-based antioxidant to be used as a functional additive in animal feed and for the food industry.