- © 2004 American Society of Plant Biologists
Abstract
In this paper, we describe a reliable method for the screening and selection of Arabidopsis transgenic seeds within minutes without germination. Expression of the Aspergillus niger β-glucosidase gene BGL1 in the plant's endoplasmic reticulum was used as a visual marker, together with 4-methylumbelliferyl-β-d-glucopyranoside (MUGluc) as a substrate. Subsequent to incubation in a solution of MUGluc at room temperature for 2 to 15 min, transgenic seeds expressing BGL1 demonstrated a distinct fluorescent signal under UV light. Optimal screening conditions at room temperature were achieved between 75 and 450 μm MUGluc, at a pH of 2.5 to 5.0 and 2 to 5 min of incubation. No significant loss of viability was detected in transgenic seeds that were redried and stored for 45 d after incubation in MUGluc solution for 2 to 150 min. Transgenic plants expressing BGL1 displayed normal phenotypes relative to the wild type. Selection frequency was 3.1% ± 0.34% for the fluorescence selection method, while kanamycin resistant selection resulted in only 0.56% ± 0.13% using the same seed batch. This novel selection method is nondestructive, practical, and efficient, and eliminates the use of antibiotic genes. In addition, the procedure shortens the selection time from weeks to minutes.
Arabidopsis is widely used as a model organism for studying the molecular and cellular biology of plants (Page and Grossniklaus, 2002). In planta methods have made gene transformation feasible and reliable without the need for tissue culture (Bechtold et al., 1993; Chang et al., 1994; Clough and Bent, 1998; Feldmann and Marks, 1987; Katavic et al., 1994). However, low transformation efficiencies necessitate screening procedures for transformed plants employing antibiotic- or herbicide-resistant genes. The selective agents employed in this method of selection quite often exert a negative effect on plant growth, and characterization of the transgenic plants can only be carried out on progenies of later generations. Moreover, public concern relating to the ecological safety of these marker genes has made them undesirable for widescale use in agriculture. Consequently, novel strategies that avoid the use of antibiotic- or herbicide-resistance genes have been developed. These include marker gene-free transformation (Gleave et al., 1999; de Vetten et al., 2003; for review, see Hare and Chua, 2002), selection with Man (Joersbo, 2001) and glycoside of the cytokinin benzyladenine glucoronide (Joersbo and Okkels, 1996), use of the inducible isopentenyl transferase (ipt; Endo et al., 2002), and genetic engineering of chloroplasts (for review, see Bogorad, 2000). Nevertheless, many of these new strategies are effective only with gene-transformation systems in which selection is achieved during the course of the regeneration process in tissue culture.
Selecting transformed Arabidopsis plants requires germinating the seeds and growing the seedlings under the stress of selective agents. Identification and selection of transgenic seedlings often takes several weeks. Once selection is achieved, an additional time period of months is required to obtain seeds which may or may not contain the transgene. A recently developed method using fluorescent proteins as visual selection markers permits the identification of mature transformed seeds by fluorescence microscopy (Jach et al., 2001; Stuitje et al., 2003). However, the use of green fluorescent protein (GFP) for selection excludes its further use for promoter analysis or subcellular localization. In this study, we report on an alternative method for the screening and selection of transformed Arabidopsis dry seeds without germination.
RESULTS
Expression of the Aspergillus niger β-Glucosidase Gene BGL1 in the Arabidopsis Ecotype Columbia
Arabidopsis plants were transformed with Agrobacterium tumefaciens LB4404 harboring the binary vector pBINPlus (van Engelen et al., 1995) containing the Aspergillus niger β-glucosidase gene BGL1 driven by the cauliflower mosaic virus (CaMV) 35S promoter and the translational enhancer Ω retrieved from pJD330 (Gallie et al., 1987). BGL1 was fused at its N terminus with the Cel1 signal peptide of Arabidopsis endo-1,4-β-glucanase (Shani et al., 1997; bases 1666–1770; GenBank accession no. X98543) and an endoplasmic reticulum (ER)-retaining tetrapeptide HDEL at its C terminus for targeting to the ER (Fig. 1A).
A. niger BGL1 gene construct (A), western blot (B), and zymogram (C) of transgenic Arabidopsis plants. P1, CaMV 35S promoter plus Ω mRNA enhancer; S, signal sequence of Arabidopsis endo-1,4-β-glucanase; BGL1HDEL, A. niger β-glucosidase gene fused to HDEL at its C terminus; T1, nopaline synthase 3′ terminator; P2, T2, promoter and terminator sequences from nopaline synthase gene, respectively; WT, wild-type plants; Angluc, purified recombinant A. niger BGL1 produced in P. pastoris.
Fifteen independent transgenic lines were obtained by selection on kanamycin (Clough and Bent, 1998). Western-blot analysis using polyclonal anti-BGL1 antibodies (Dan et al., 2000; Fig. 1B) and an in-gel β-glucosidase zymogram confirmed the presence of BGL1 protein with the expected molecular mass of the A. niger native enzyme (120 kD; Dan et al., 2000), as well as β-glucosidase activity (Fig. 1, B and C). Wild-type plants exhibited neither β-glucosidase activity nor cross-reaction with the anti-BGL1 antibodies (Fig. 1, B and C). The seeds of transgenic lines 1, 3, 6, and 11 were maintained for the optimization of seed-screening conditions.
Optimizing Seed-Screening Conditions
The excitation wavelength of 4-methylumbelliferone, released from 4-methylumbelliferyl-β-d-glucopyranoside (MUGluc) by the action of β-glucosidase, is approximately 320 or 365 nm, depending on the pH. The emission wavelength peaks at 450 nm (http://www.tecan.com/phchange_safire1.pdf). In our experiments, the wavelength of our light source ranged from 300 to 400 nm. When fluorescence is visualized using excitation filter at 320 nm and emission filter at 450 nm, the sensitivity of the selection may be improved because the background is minimized.
Both dry and water-imbibed transgenic seeds incubated with MUGluc (75 μm, pH 4.0) showed a significantly stronger fluorescent signal than the wild-type controls (Fig. 2A). As expected, wild-type seeds also demonstrated a certain amount of fluorescence due to the presence of intrinsic fluorescent compounds such as flavonoids and aromatic amino acids. Both transgenic and wild-type seeds that were imbibed overnight in water exhibited a higher fluorescent background than did dry seeds despite the fact that all treatments with MUGluc only lasted a few minutes. Therefore, the high background most likely resulted from leakage of low-molecular-weight metabolites into the surrounding solution (Bewley, 1997). It was concluded that to minimize the fluorescent background and achieve better selection conditions, dry seeds should be used for the screening procedure.
Fluorescent signals of transgenic and wild-type seeds under UV after incubation with MUGluc. A, Dry and overnight-imbibed seeds; B, incubated at different pHs with 75 μm MUGluc; C, incubated in different concentrations of MUGluc, pH4.0; WT, wild-type plant seeds; CBT, transgenic plant seeds expressing BGL1.
To determine the optimal pH range for seed screening, dry seeds were incubated in one of a series of 75-μm MUGluc solutions with increasing pHs. Wild-type seeds fluoresced, and there were no visually significant differences in signal intensity after incubation in solutions with pHs ranging between 2.5 and 5.5 (Fig. 2B). This is consistent with the results of pH-dependent activity reported for A. niger BGL1 (Shoseyov et al., 1988). Although the excitation wavelength of 4-methylumbelliferone is affected by pH (Green, 1990), the UV light source used in this experiment covered the entire excitation wavelength spectrum that is affected by pH and therefore, the fluorescent signals did not change significantly at different pHs.
MUGluc solutions at different concentrations (0–450 μm, pH 4.0) were used to determine the minimum effective substrate concentration necessary for seed screening. The intensity of the fluorescent signals emitted from the transgenic seeds increased with increasing MUGluc concentrations, whereas those from the wild-type seeds did not change significantly (Fig. 2C). Although at the minimal concentration of 15 μm MUGluc, transgenic and wild-type seeds already exhibited an obvious visual difference in fluorescence level, higher substrate concentrations were found to maximize screening results during seed selection.
At 75 μm MUGluc (pH 4.0), the fluorescent signal displayed by transgenic seeds increased with incubation time (Fig. 3). A time of 2 min was sufficient to distinguish transgenic from wild-type seeds. After 15 min incubation at room temperature (RT), the fluorescent signals of the positive seeds overwhelmed those of adjacent seeds, making it difficult to distinguish the transgenic ones from the wild type. Interfering fluorescent signals from adjacent seeds may be minimized for optimal screening by adjusting the intensity of the light source.
Fluorescent signal development of transgenic and wild-type seeds incubated with 75 MUGluc, pH4.0, after 1 to 15 min.
Mixed seeds with different proportions of transgenic seeds were incubated with 75 μm MUGluc in petri dishes (300–400 seeds/plate) for 2 to 5 min. The plates were then exposed to UV (Fig. 4, A and B). Positive transformed seeds were collected with a pipette. All of the selected seeds were then sterilized and incubated with MUGluc (Fig. 4C). Individual seeds selected from the mixed populations gave strong positive signals relative to the wild-type seeds.
Fluorescent signal of transgenic seeds mixed at different ratios with wild-type seeds after incubation with 75 MUGluc, pH4.0, for 2 min. A, Full view of the mixed seeds in plates under UV; B, magnified view of the mixed seeds from the plates; C, confirmation of fluorescent signals from plate-selected seeds on individual ELISA plates.
Fluorescence Selection Versus Kanamycin Resistant Selection
The same batch of the seeds collected from Agrobacterium infected plants was divided into two groups to compare the selection frequency and reliability of fluorescence selection and kanamycin resistant selection methods. After incubation with MUGluc for 5 min, 20 plates of seeds (around 300 seeds/plate) were tested. Selection frequency was 3.1% ± 0.34%. Kanamycin resistant selection with the seeds (around 9,000 in 30 plates) from the same batch resulted in 0.56% ± 0.13% resistant seedlings, which is significantly less relative to the fluorescent selection. Furthermore, PCR tests using primers within BGL1 gene were carried out with seedlings resulted from both selection methods. All 108 seedlings resulted from fluorescent selection seeds displayed a positive PCR band as shown in Figure 5 for representative samples. Kanamycin resistant selection also resulted in 100% positive PCR reaction in all 37 selected seedlings.
PCR confirmation of transgene presence in seedlings germinated from the seeds selected by fluorescence. 1 to 9, putative transgenic seedlings; Wt, wild type; 1 kb, DNA ladder; Plasmid, pBINPlus containing BGL1.
Phenotype of Transgenic Plants
To determine the potential existence of phenotypic variance between transgenic and wild-type plants, T2 transgenic lines 1, 3, 6, and 11, as well as wild-type plants, were grown in a greenhouse. In a comparison with nontransgenic plants, no abnormal phenotype was detected among the lines of transgenic plants expressing A. niger BGL1 targeted to the ER (Table I).
Growth parameters and seed viability of wild-type and transgenic Arabidopsis plants expressing BGL1
Germination of the Screened Seeds
Transgenic seeds incubated in the MUGluc solution for 2 to 150 min were redried and stored for 45 d. Number of leaves, plant height, total seed weight, and percent of seed germination were tested. No significant loss in viability or plant-growth performance was detected following up to 150 min of incubation (Table I). Longer incubation times resulted in reduced germination rate (not shown).
DISCUSSION
In plant molecular biology, fluorogenic chemicals such as 4-methylumbelliferyl glucuronide and 4-methylumbelliferyl galactoside have been widely used for the detection of reporter-gene proteins, mainly due to their higher sensitivity (Martin et al., 1992). In this study, the fluorogenic substrate MUGluc and the A. niger β-glucosidase gene BGL1 were employed to screen for and select dry transformed Arabidopsis seeds without the need for seed germination. With this method, screening and selection can be achieved within just minutes (Fig. 3). In addition, the selected seeds can be redried and stored for later germination without significant loss in viability (Table I). As a result, further analysis on the transgenic seeds can be done directly with these selected transgenic seeds and the primary transgenic plants at various growth and developmental stages can be characterized by simply growing the selected seeds. This may confer a significant savings in time over the antibiotic transgenic plant-selection protocols employing antibiotics or herbicides as the selective agents. The conventional selection of transformed Arabidopsis plants requires seed germination and growth under the stress of selective agents, followed by an additional interval of at least several days to distinguish between transgenic and nontransgenic plants. Although some analysis of primary transgenic seedlings at their later growth stages after selection is also possible, usually characterization of transgenic plants is conducted on transgenic progenies of later generations due to the negative impact of these selective agents on plant growth. For this reason, it is often necessary to wait additional months to obtain seeds that may or may not contain the transgene.
Parallel experiments using the same seed batch harvested from infected plants were conducted to compare fluorescence selection versus kanamycin resistance selection. Selection frequency obtained with fluorescence selection was significantly higher than that obtained with kanamycin selection. This could result from insufficient level of expression of the NPTII in some of the transgenic plants due to position effect of the T-DNA insertion site. In general, every negative selection method such as resistance to antibiotics or to herbicide is limited by the set point of the selection agent concentration, while positive selection method is limited by the sensitivity and the signal to noise ratio.
Recently Stuitje et al. (2003) used fluorescent proteins as visual selection markers for the identification of mature transformed seeds by fluorescence microscopy. GFP and its variants such as EGFP, ECFP, and EYFP (CLONTECH, Palo Alto, CA) share very high identity in their DNA sequences. The high DNA identity among those variants may limit their subsequent application in transgenic plants that already express GFP. Although coexpression of two or three GFP variant genes was obtained by Stuitje et al. (2003), it remains a challenge to avoid cosuppression of closely related gene. Furthermore, expression of more than two GFP variants in the same plants may interfere with their analysis due to overlapping emission spectra. Discrimination between GFP mutants in the same tissue requires relatively expensive equipment such as a confocal laser microscope. Our results show that the intensity of the fluorescent signals generated by the transformed seeds can be significantly increased with increasing MUGluc concentration (Fig. 2) and incubation time (Fig. 3). Therefore, using the BGL1 marker gene and its fluorogenic substrate MUGluc provides the option of increasing screening sensitivity. In addition, fluorescent sorting may be useful in the future in commercially available automatic seed sorters.
The C-terminal HDEL sequence has been reported to be sufficient for the retention of secreted recombinant proteins in the plant ER (Gomord et al., 1997). The major function of the (K/H) DEL sequence is to collect the escaping ER-resident proteins back to the ER via a special membrane-bound receptor (Lee et al., 1993). As we previously reported, in yeast (Saccharomyces cerevisiae) as well as tobacco (Nicotiana tabacum) plants (Dan et al., 2000; Wei et al., 2004), A. niger BGL1 is an N-glycoprotein with molecular mass of 120 kD. Deglycosylation results in a smaller protein band (about 100 kD), which corresponds to the predicted Mr of the translated cDNA. In tobacco plants, active and glycosylated (120 kD) A. niger BGL1 was successfully produced and targeted to the ER by fusing its N terminus with the Cell signal peptide of Arabidopsis endo-1,4-β-glucanase (Shani et al., 1997) and its C terminus with the ER-retaining HDEL sequence. In the same manner in this study, expressed BGL1 was targeted to the ER and glycosylated. Glycosylated BGL1 with a molecular mass of 120 kD was detected in transgenic Arabidopsis plants (Fig. 1B). Transgenic plants possessed a normal phenotype relative to the wild type. Recently, Arabidopsis β-glucosidases have been found to accumulate in ER bodies and to be present in seeds, cotyledons, hypocotyls, and roots (Stotz et al., 2000; Gallardo et al., 2002; Matsushima et al., 2003). Some have ER-retaining signal peptides at their C terminus and are reported to be associated with defense responses against herbivorous insects (Stotz et al., 2000; Matsushima et al., 2003). At this point of time we have not determined whether expression of BGL1 in Arabidopsis affect the above. The plant ER is not only the quality-control entry port for all proteins into the endomembrane and secretory pathways; it also hosts a large number of co- and posttranslational maturation processes of nascent polypeptides (for review, see Galili, et al., 1998). For this reason, it has been chosen as a targeted plant organelle for the expression and accumulation of foreign proteins (Conrad and Fiedler, 1998). All these findings as well as our results suggest that the ER is a suitable subcellular compartment for the expression of BGL1 as selectable marker without significantly affecting plant phenotype.
MATERIALS AND METHODS
Plant Growth
For plant gene transformation, Arabidopsis plants (ecotype Columbia) were planted at a density of 2 to 5/25-cm2 pot containing peat, vermiculite, and perlite (3:2:1, v/v). They were grown to the flowering stage in a shaded greenhouse, 24°C day/20°C night, 13 to 14 h light and average midday photon flux density of 200 to 250 μE m−2 s−1. To compare the phenotypes of the transgenic and nontransgenic plants, 30 transgenic plants each from independent lines 1, 3, 6, and 11, and 40 wild-type plants were grown in the afore described pots (1 plant/pot) and maintained under the same conditions. Percent germination, number of rosette leaves, plant height, and seed weight of each plant were recorded.
Construction of Chimeric Genes and Plant Transformation
The chimeric gene cassette (Ter) was constructed in the intermediate plasmid pJD330 (Gallie et al., 1987), which contained the constitutive CaMV 35S promoter, translation enhancer Ω, and nopaline synthase 3′ terminator. BGL1 cDNA (2.5 kb; GenBank accession no. AJ132386) was excised with restriction enzymes NcoI and BamHI from pETB1 (Dan et al., 2000) and inserted into pJD330 by replacing the original uidA gene with BGL1. The ER-retaining signal peptide HDEL (underlined in the antisense primer) and a SmaI restriction site were added to the BGL1 C terminus in a PCR with the following pair of primers: 5′-cagtgaccgtggatgcgacaatg-3′ and 5′-aaaacccgggttaaagttcatcatgaacagtaggcagaga-3′. A DNA fragment encoding the Cel1 signal peptide (bases 1666–1770; GenBank accession no. X98543; Shani et al., 1997) of Arabidopsis endo-1,4-β-glucanase was restricted with BamHI and XbaI from plasmid p21Ng and then fused to the N terminus of BGL1. The gene cassette in pJD330 was further isolated and inserted between the HindIII and EcoRI sites of the binary vector pBINPlus (van Engelen et al., 1995; Fig. 1A) of Agrobacterium tumefaciens and introduced into plants by floral dip transformation (Clough and Bent, 1998).
The seeds from the transfected plants were harvested and surface-sterilized with 30% bleach (2.5% sodium hypochlorite, final concentration) in 65% ethanol for 5 min, followed by 3 rinses with 100% ethanol, and dried overnight in a laminar hood. Sterilized seeds were plated on kanamycin selection media according to Clough and Bent (1998). Positive seedlings with 3 to 5 adult leaves were then transplanted and maintained in the greenhouse. Western-blot and in-gel zymogram analyses were performed as previously described (Dan et al., 2000) with crude protein extracts from rosette leaves (10 μg lane−1). Purified recombinant Aspergillus niger BGL1 produced in Pichia pastoris was used as a positive control. Fresh leaves were homogenized in liquid nitrogen, and 50 mm citrate buffer (leaf to buffer, 1:3, w/v) containing 10 mm EDTA, 4 mm dithiothreitol, and 1% (w/v) polyvinylpolypirrolidone were added to the leaf powder. The leaf homogenate was rotated for 1 h at 4°C and the soluble proteins isolated by centrifugation (17,000g for 10 min).
Optimizing Seed-Screening Conditions
Seeds which had been imbibed overnight with distilled, sterilized water were rinsed three times in sterilized water. A series of solutions with varying pHs or increasing concentrations of MUGluc (M3633, Sigma, St. Louis) were prepared in 25 mm citric buffer. Unless otherwise specified, the substrate solutions used in the experiments contained 75 μm MUGluc at a pH of 4.0. To facilitate seed dispersion, all substrate solutions contained 0.1% (w/v) agarose.
Drops of substrate solutions (95 μL each) were placed in rows in 25-cm petri dishes. To each drop, 5 to 10 seeds were added and quickly mixed with the aid of a Pasteur pipette. The petri dishes containing the seeds were then immediately moved into the dark chamber of an M-2 UV transilluminator (UVP, Upland, CA) and maintained for a predetermined incubation period. To minimize exposure of the seeds to UV during the photographic process, the UV light was activated for only approximately 5 to 10 s.
Mixed-Seed Screening
Transgenic and wild-type plant seeds were weighed and vortexed to obtain seed mixtures of 1%, 10%, and 30% CBT in the wild type. The seeds (about 5 mg, accounting for about 300 seeds) were then quickly mixed with 10 mL of MUGluc solution in a petri dish and kept there for 2 to 5 min. Then, the seeds were exposed to UV light and the positive ones were picked out by Pasteur pipette. To test seed-selection reliability, the selected as well as wild-type seeds were sterilized and placed in the wells of an ELISA plate, one per well; 100 μL of 75 μm MUGluc solution (25 mm citric buffer, pH 4.0) was then added to each well, and the plate was incubated at 37°C for 4 h, after which the plate was exposed to UV light.
Fluorescence Selection Versus Kanamycin Resistant Selection
The same batch of dry seeds collected from Agrobacterium infected plants after flower-dip was divided into two groups. One group for kanamycin resistant selection was sterilized, dried, and plated on selective media as described above. Three weeks later, seedlings with green leaves and roots were transplanted into peat, vermiculite, and perlite (3:2:1, v/v) and kept in the greenhouse. The other group subjected to MUGluc fluorescent selection was treated and screened as described in the mixed-seed screening section. The selected seeds were rinsed with sterilized water three times and plated on wet Whatman paper in petri dish. After 2 d at 4°C, seeds were moved into a growth chamber with 22°C, 16 h of photoperiod, and photon flux density of 50 μE m−2 s−1. The resultant seedlings were then transplanted and kept in the greenhouse. The numbers of selected seeds and seedlings over total number of seeds or seedlings tested were calculated as selection frequency.
PCR Confirmation of Transgenic Plants in the Selected Seeds
Plant genomic DNA extracts obtained from selected seedlings grown in the greenhouse were used as a template in PCR reaction. A sense primer (5′-ggggagaagcccgcccagttacgaccaccgtccggacttctac-3′) and an antisense primer (5′-gtcaggttcctgcgggcacctaggtttccgtc-3′) within BGL1 gene were used to amplify a PCR product with expected size near 1,100 bp.
Seed Germination
Transgenic and wild-type seeds were incubated in a 75-μM MUGluc solution (pH 4.0) containing 1% agarose for different periods (2 min–6 h) and then exposed briefly to UV. These seeds were moved onto a Whatman filter paper and dried with hair dryer (33°C) for 5 min and then kept in a hood overnight. Redried seeds were sealed with parafilm in petri dishes (25 cm) and kept at RT for 45 d. For germination, seeds were imbibed with sterile water. The seeds were maintained for 2 d at 4°C, then at RT for an additional 3 d. Germination percentages for all treatments (about 300 seeds/treatment) were calculated.
Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession numbers X98543 and AJ132386.
Footnotes
-
↵1 This work was supported by the Eugine and Edith Schhoenberger Foundation.