• Open Access

Production of a recombinant full-length collagen type I α-1 and of a 45-kDa collagen type I α-1 fragment in barley seeds


*(fax +358 9 19158475; e-mail kristiina.makinen@helsinki.fi)


Recombinant DNA technology can be used to design and express collagen and gelatin-related proteins with predetermined composition and structure. Barley seed was chosen as a production host for a recombinant full-length collagen type I α1 (rCIa1) and a related 45-kDa rCIa1 fragment. The transgenic barley seeds were shown to accumulate both the rCIa1 and the 45-kDa rCIa1 fragment. Even when the amount of the rCIa1 was just above the detection threshold, this work using rCIa1 as a model demonstrated for the first time that barley seed can be used as a production system for collagen-related structural proteins. The 45-kDa rCI1a fragment expression, targeted to the endoplasmic reticulum, was controlled by three different promoters (a constitutive maize ubiquitin, seed endosperm-specific rice glutelin and germination-specific barley α-amylase fusion) to compare their effects on rCIa1 accumulation. Highest accumulation of the 45-kDa rCIa1 was obtained with the glutelin promoter (140 mg/kg seed), whereas the lowest accumulation was obtained with the α-amylase promoter. To induce homozygosity for stable 45-kDa rCIa1 production in the transgenic lines, doubled haploid (DH) progeny was generated through microspore culture. The 45-kDa rCIa1 expression levels achieved from the best DH lines were 13 mg/kg dry seeds under the ubiquitin promoter and 45 mg/kg dry seeds under the glutelin promoter. Mass spectroscopy and amino acid composition analysis of the purified 45-kDa rCIa1 fragment revealed that a small percent of prolines were hydroxylated with no additional detectable post-translational modifications.


Transgenic plants have been shown to produce an array of recombinant DNA-derived products including pharmaceuticals (insulin, lactoferrin, interferons, antibodies and oral vaccines), structural components (collagen, silk and biodegradable plastics) and industrial processing enzymes (trypsin) to deliver novel products or substitutes for animal and fossil fuel-derived products (Baez, 2005). These transgenic crops are only intended for industrial use and are not meant to be used as food. The pharmaceutical and food industries use collagen and its derivative gelatin (a denatured form of collagen) extensively in various applications. Pharmaceutical applications use over 50 000 metric tons of gelatin annually in capsules and tablets, as a plasma expander, as a stabilizer for biologics, and as a structural component in tissue-engineering applications. The traditional way of gelatin production is by extraction from collagen-rich tissues of animals. These products consist of different collagen molecules having different molecular weights, isoelectrict points, gelling properties, etc. Especially for medical applications it is important to have collagen-related products that are homogenous, origin traceable, free of animal components and having a defined composition. Using recombinant technology, it may be possible to design and express gelatin molecules with predetermined characteristics to match the requirements in a variety of pharmaceutical applications. The production of recombinant collagen and collagen fragments to match these product attributes has been demonstrated in mammalian cells, insect cell cultures, yeast, milk of transgenic animals and plant cells (e.g. Geddis and Prockop, 1993; Nokelainen et al., 1996; Vuorela et al., 1997; John et al., 1999; Ruggiero et al., 2000; Olsen et al., 2003; Ritala et al., 2008). The first experiments to express recombinant collagen-related proteins in transgenic plants were carried out in a model plant tobacco (Ruggiero et al., 2000; Merle et al., 2002; Olsen et al., 2003). The tobacco plants were shown to produce fully processed triple helical molecules of human collagen III (Olsen et al., 2003) and rCIa1 (Ruggiero et al., 2000). In this report, we describe for the first time the expression, purification and characterization of a barley seed-derived structural protein of mammalian origin, a 45-kDa collagen type I α-1 (rCIa1) fragment, and also show that the full-length rCIa1 can be expressed in barley seeds at low quantities. Recently, the accumulation and purification of full-length rCIa1 and of the same 45-kDa rCIa1 fragment described in this report was reported from transgenic corn seed (Zhang et al. 2009b).

We selected barley for the expression of recombinant collagen-related proteins because barley is an important crop plant with well-established agronomic, harvesting, transport, storage and processing practices. Barley has been adapted to a wider variety of climates than any other cereals, from sub-Arctic to sub-tropical areas, and is grown in large scale in Russia, Australia, Germany, Turkey, the Scandinavian countries and North America. Barley stores large amounts of protein and starch in its endosperm, which undergoes programmed cell death with elimination of all DNA, organelles and cytosol during the final stages of seed maturation (Young and Gallie, 2000). The desiccated mature seeds allow the storage of proteins even for several years in intact form. Barley seeds have shown their potential as a production host for various types of recombinant proteins, such as enzymes and single chain antibodies (Nuutila et al., 1999; Horvath et al., 2000; Patel et al., 2000; Schünmann et al., 2002). Promoters specific for tissue- and seed development are available for expression in barley seeds. For example, β-glucanase was expressed in homozygous seeds under the control of the D-hordein gene (Hor3-1) promoter of barley (Horvath et al., 2000). Endoplasmic reticulum (ER)-targeted expression reached on average a level corresponding to 5.4% of the extractable protein. An anti-glycophorin single-chain antibody fused to an epitope of HIV was expressed and targeted to ER using a seed-specific promoter from the wheat high-molecular weight glutelin gene (Bx17) (Schünmann et al., 2002). Using this promoter the average expression level reached 0.6% of total extractable protein, which corresponds to 151 mg single-chain antibody (scFv)/kg seeds. The use of seed-specific promoters is beneficial as the recombinant protein is expressed only in the seeds and not in any other parts of the plants. By using promoters that are active only during seed germination, expression of the recombinant protein during seed development and maturation in the field can be avoided. Another important advantage over other crops is that barley is a self-pollinating plant, and in an earlier study, it has been shown that the gene flow through pollen dispersal is extremely low in its cultivation (Ritala et al., 2002).

In this study, the feasibility of using transgenic barley seeds as an expression system for recombinant collagen-related protein products was described. A sequence encoding the 45-kDa rCIa1 fragment and the full-length rCIa1 gene optimized for monocot expression were inserted under the control of the constitutive ubiquitin (Ubi-I) promoter, the endosperm-specific GluB1 promoter or the germination-specific (α-amy) α-amylase promoter (Lanahan et al., 1992; Nuutila et al., 1999; Patel et al., 2000; Vickers et al., 2006). The recombinant proteins were targeted to the ER as ER targeting has been shown to increase the expression levels of recombinant proteins (Horvath et al., 2000; Stöger et al., 2000). In barley ER-targeted expression of F4 fimbriae from enterotoxigenic Escherichia coli under the trypsin inhibitor (TI) gene promoter yielded expression levels up to 1% of total extractable protein (Joensuu et al., 2006). The Cocksfoot mottle virus 5′-untranslated region (Mäkinen et al., 1995), called the ε-sequence (CfMVε), was used as the 5′-leader sequence in half of the constructs. This was carried out to test whether viral 5′-UTR of a barley-infecting virus could act as a translational enhancer. The purified 45-kDa rCIa1 was analyzed by mass spectroscopy (MS) and amino acid composition analysis to verify its size and to detect potential post-translational modifications.


Production of transgenic plants

The constructs used are shown in Figure 1. A schematic flow of the production of transgenic (T) and double haploid (DH) barley plants and the generation nomenclature are presented in Figure S1. Equal numbers of transformation attempts were carried out by particle bombardment with each of the six constructs containing either the Ubi-I or the GluB1 promoter (Table 1). In half of the constructs the 5′-UTR of Cocksfoot mottle virus (ε-element; Mäkinen et al., 1995) was inserted in front of the cIa1- or the 45-kDa cIa1-encoding region to test if this genetic element would improve the expression levels. Transgenic plants (T0) were obtained only from the bombardments with the plasmids pEW33, pEW44 and pJAM01 carrying the Ubi-I-cIa1, Ubi-I-ε-45kDa cIa1 and GluB1-ε-45kDa cIa1 constructs respectively. The number of T0 plants obtained and an estimate of the individual integration events with each construct are presented in Table S1. Production of the next generations of Ubi-I-cIa1, Ubi-I-ε-45kDa cIa1 and GluB1-ε-45kDa cIa1 carrying plants is summarized in Tables S3 and S4. As the transformation attempts with the plasmids pEW32, pEW45-and pJAM02 failed, evaluation of the effect of the CfMVε-element on transgene expression was impossible with these plants.

Figure 1.

 A schematic representation of the constructs used in this study.

Table 1.   The constructs used
  1. Ubi-I, maize ubiquitin promoter and first intron; ε, the 5′UTR of Cocksfoot mottle virus; rCIa1, cDNA of human collagen (I) α-1 carrying Arabidopsis basic chitinase signal sequence (ss), N-terminal telopeptide sequence (helical collagen sequence) C-terminal telopeptide sequence, and the foldon fused to the HDEL; bar, herbicide (phosphinotricin) resistance selection marker; 45kDa, 45-kDa fragment of CIaI with Arabidopsis basic chitinase ss and HDEL; GluB1, rice glutelin B1 promoter, α-amy, barley α-amylase fusion promoter 46/4-6; 35S, Cauliflower mosaic virus 35S promoter; hph-CAT1I-hph, hygromycin resistance selection marker with a castor bean catalase-1 intron.

pEW32Ubi-I-ε-cla1 + Ubi-I-bar
pEW33Ubi-I-cla1 + Ubi-I-bar
pEW44Ubi-I-ε-45kDa + Ubi-I-bar
pEW45Ubi-I-45kDa + Ubi-I-bar
pJAM01GluB1-ε-45kDa + Ubi-I-bar
pJAM02GluB1-45kDa + Ubi-I-bar
pSAM04α-amy-ε-45kDa + 35S-hph-CAT1I-hph
pSAM03α-amy-45kDa + 35S-hph-CAT1I-hph

Equal numbers of Agrobacterium-mediated transformation attempts were carried out with both α-amy-constructs (Table 1). Approximately twice as many transgenic plants (T0) were obtained with the plasmid pSAM03 carrying α-amy-45kDa cIa1 than with pSAM04 carrying α-amy-ε-45kDa cIa1 (Table S2). Analysis of seeds from these plants made it possible to evaluate the effect of the CfMVε element on the 45-kDa rCIa1 accumulation (see Recombinant protein expression levels of transgenic barley lines). Non-transgenic control material was obtained by regenerating barley plants from non-bombarded or non-infected embryos. In the progeny generations, null segregants of the transgenic plants were used as non-transgenic controls.

Fertility and seed yield of the transgenic plants

Transgenic barley lines carrying Ubi-I-CIa1

All Ubi-I-CIa1 T0 plants produced were fertile and there was no significant difference in the total seed yield when compared with the non-transgenic control plants (anova, single factor, P = 0.022; Table S1). The T1 generation (Table S3) was produced from five different T0 plants chosen on the basis of detectable rCIa1 expression (see Recombinant protein expression levels of transgenic barley lines). The fertility of these Ubi-I-cIa1 T1 plants was slightly reduced (81% fertile) when compared with the non-transgenic control plants (100% fertile). The total seed yield of the Ubi-I-cIa1-carrying T1 plants was significantly lower when compared with the control plants (anova, single factor, P = 0.0007).

Transgenic barley lines carrying 45-kDa CIa1 fragment

All Ubi-I-ε-45kDa cIa1, α-amy-ε-45kDa and α-amy-45kDa T0 plants were 100% fertile whereas the fertility of GluB1-ε-45kDa cIa1 carrying T0 plants was reduced to 72% when compared with non-transgenic control plants. The fertility of all 45-kDa CIa1 carrying T1 plants was 95%–100% (Table S3) and that of the T2 plants 100% (Table S4). There was no significant difference in the total seed yield of any T0 or T1 plant when compared with the non-transgenic controls.

Recombinant protein expression levels of transgenic barley lines

Expression levels of rCIa1 in barley seeds

Individual seeds from five T0 lines carrying the Ubi-I-cIa1 gene were analysed by enzyme-linked-immunosorbent assay (ELISA) for full-length rCIa1 accumulation. All the lines produced seeds that showed expression, but the expression levels were low (Table S1) and close to the detection limit of 3 ng/mL. Western blot analysis was used to verify the expression, revealing the presence of full-length rCIa1 similar in size to the full-length rCIa1 produced in the yeast Pichia pastoris (Figure 2a). The T2 seeds from three independent T1 lines were also analysed, but no full-length rCIa1 expression was detected by ELISA or Western blotting. Therefore, no T2 generation was produced and no attempts were made to purify the full-length rCIa1 for further product characterization.

Figure 2.

 Immunoblot analysis of full-length and 45-kDa rCIa1 in seeds. (a) Immunoblot analysis of T1 seed extracts from T0 lines carrying the Ubi-I-rCIa1 gene cassette. Twenty microliter of the seed extract was loaded on the gel and 0.5 ng of similar Pichia pastoris-derived full-length rCIa1 (labelled as PprCIa1) was used as a reference material. (b) Immunoblot analysis of T1 seed extracts from T0 line 1785 carrying the GluB1-ε-45-kDa gene cassette. The seeds had been stored at room temperature for 15 months before analysis. The loaded protein extract amounts were calculated on the basis of enzyme-linked-immunosorbent serologic assay results. The expected 45-kDa rCIa1 amount in each sample is given under the blot. (c) Analysis of the 45-kDa rCIa1 expression in seeds carrying the Ubi-I-ε-45-kDa gene cassette. Analysis of the extracts from the T1 seeds of T0 line 1755 and T2 seeds of nine T1 lines originating from line 1755 is shown as an example. Seed extracts (12.5 μL) (corresponds to 25 μg of total protein) were loaded on the gel (8%). Prior to loading, the samples were heated at 65 °C for 10 min. The similar P. pastoris-derived 45-kDa rCIa1 (labelled as Pp45-kDa) was used as a reference material. Extract from non-transgenic barley seeds (Golden Promise) was used as a negative control. The M lane was loaded with molecular weight markers.

Expression levels of the 45-kDa rCIa1 fragment in barley T1 seeds

Accumulation of 45-kDa rCIa1 was analysed from the immature T1 seeds of two T0 lines carrying Ubi-I-ε-45-kDa as well as from the mature T1 seeds of one T0 line carrying Ubi-I-ε-45-kDa and three T0 lines carrying the GluB1-ε-45-kDa transgene cassette (Table S1).

In the case of expression under the constitutive Ubi-I promoter, the average 45-kDa rCIa1 concentration was 227 ± 180 μg of 45-kDa rCIa1/g of soluble protein in the T0 line 1755. This corresponds to an accumulation level of ∼5.5 mg of 45-kDa rCIa1/kg dry seeds (Table 2). In the immature seeds analysed, the 45-kDa rCIa1 expression levels were lower, i.e. 5.8 ± 4.9 μg (line 1769) and 8.6 ± 4.6 μg (line 1770) of 45-kDa rCIa1/g of soluble protein. After 15 months of storage, the amount of the 45-kDa rCIa1 in the dry seeds of the line 1755 had not changed significantly, and was still 200 ± 185 μg/g of soluble protein. The expression of 45-kDa rCIa1 under the seed-specific GluB1-promoter was three to 25 times higher than under the constitutive Ubi-I-promoter. The T0 line 1785 was the best 45-kDa rCIa1 producer. On average 693 ± 700 μg (line 1779), 1182 ± 996 μg (line 1781) and 11 357 ± 7 143 μg (line 1785) of 45-kDa rCIa1/g of soluble protein was reached when expression was driven from the GluB1 promoter. These expression levels correspond to accumulation levels of ∼17 mg (1779), ∼43 mg (1781) and ∼142 mg (1785) of 45-kDa rCIa1/kg dry seeds (Tables 2 and S1). The 45-kDa rCIa1 content in individual seeds from the same origin varied considerably, which can be seen from the large standard deviation of the mean. In the seeds of the line 1785, high 45-kDa rCIa1 content was detected after 15 months of storage at room temperature. Western analysis of these seeds indicted no visible degradation products (Figure 2b) confirming the stability of the 45-kDa rCIa1 within seeds. The maximum 45-kDa rCIa1 concentrations in the protein samples extracted from single seeds were found to be 0.06% of extractable soluble seed protein in line 1755, 0.16% in line 1779, 0.25% in line 1781 and 2.61% in line 1785.

Table 2.   Summary of accumulation levels of full-length rCla1 and 45-kDa rCla1 in the transgenic barley lines
T0 originGenesAccumulation of rClaI or 45-kDa rCIa1 in T0 plants (T1 seeds) mg/kgAccumulation of rClaI or 45-kDa rCIa1 in T1 plants (T2 seeds) mg/kg Accumulation of 45-kDa rCIa1 in T2 plants (T3 seeds) mg/kgAccumulation of rClaI or 45-kDa rCIa1 in DH0 plants (DH1 seeds) mg/kg
  1. *Expression was below the detection limit.

1730Ubi-I-cla1Avg. 0.04Udl* Udl
1732Ubi-I-cla1Avg. 0.09   
1736Ubi-I-cla1Avg. 0.01Udl Udl
1739Ubi-I-cla1Avg. 0.08   
1743Ubi-I-cla1Avg. 0.02Udl Udl
1755Ubi-I-ε-45kDaAvg. 5.5
Max. 19
Avg. 8.6
Max. 15.1
Avg. 3.07–6.38
Max. 8.62
Avg. 4.4–8.7
Max. 13.3
1779GluB1-ε-45kDaAvg. 17.2   
1781GluB1-ε-45kDaAvg. 42.5
Max. 94
Avg. 39.8
Max. 48.2
Avg. 20.5–49.3
Max. 49.3
Avg. 17.3–26.9
Max. 45.1
1785GluB1-ε-45kDaAvg. 141.7
Max. 270
Avg. 44.9
Max. 73.7
Avg. 37.8–49.6
Max. 64.5
30 Diff. linesα-amy-45kDaAvg. 1.4
Max. 11.8
19 Diff. linesα-amy-ε-45kDaAvg. 2.1
Max. 17.6

Results obtained through ELISA were verified by Western blot analysis. As an example, a Western blot analyses verifying 45-kDa rCIa1 expression in the T1 seeds of the lines 1785 and 1755 are shown in Figure 2b,c. A similar 45-kDa rCIa1 purified from the yeast P. pastoris was used as reference material. The proteins appeared to move slowly (apparent molecular weight of ∼62 kDa) relative to their calculated masses of ∼45-kDa. In general, collagenous proteins migrate ∼40% more slowly than other proteins (Butkowski et al., 1982). The protein amounts (10, 40 and 100 ng) loaded into the gel were calculated on the basis of ELISA results. Comparable signals were obtained from known amounts of yeast-derived 45-kDa rCIa1 used as a standard protein and from barley-derived proteins. This comparison indicated the accuracy of the quantification by ELISA of the barley seed-derived 45-kDa rCIa1.

Expression levels of the 45-kDa rCla1 fragment in barley T2 and T3 seeds

According to polymerase chain reaction (PCR) and expression analyses, the cloned genes were stably integrated and passed on to the T1 and T2 generations (Tables S3 and S4). The T2 seeds were analysed for accumulation of 45-kDa rCIa1. The amount of 45-kDa rCIa1 in the seeds was again higher when expressed under the GluB1 promoter than under the Ubi-I promoter, but the difference was not as high as in the T0 generation. The average 45-kDa rCIa1 content in various T1 lines varied between 511 and 2885 μg/g of soluble protein when expressed from the GluB1 promoter, and between 171 and 819 μg of 45-kDa rCIa1/g of soluble protein when expressed from the Ubi-I promoter. In the best T1 lines, the 45-kDa rCIa1 accumulated to 74 mg/kg with the GluB1 promoter and 15 mg/kg with the Ubi-I promoter of dry seeds respectively (Table 2), which represented 2885 ± 287 μg and 819 ± 189 μg of 45-kDa rCIa1/g of soluble protein. Western blot analysis of the T2 seeds expressing the 45-kDa rCIa1 under the Ubi-I is shown in Figure 2c. The T3 seeds from 24 T2 plants containing the Ubi-I-45kDa and the 16 T2 plants containing the GluB1-ε-45kDa cassette were analysed by ELISA (Table S4). The 45-kDa rCIa1 expression levels were similar to those of the T2 seeds of the T1 generation plants.

Expression levels of the 45-kDa rCIa1fragment in germinated barley T1 seeds

Altogether 146 seeds from 30 T0 lines carrying the α-amy-45kDa transgene and 62 seeds from 19 T0 lines carrying the α-amy-ε-45kDa transgene were germinated and analysed for their 45-kDa rCIa1 expression. According to a Western blot analysis the 45-kDa rCIa1 expression took place at the same time as the endogenous barley α-amylase expression (data not shown). The highest 45-kDa rCIa1 expression was observed in seeds that had almost no endosperm left. The expression levels under the α-amy promoter are shown in Table S2. In general, the expression levels were rather low and in many cases close to or below the detection limit. The average 45-kDa expression was 37 ± 89 μg of 45-kDa rCIa1/g of soluble protein (1.4 mg/kg seeds) in the seeds of the α-amy-45kDa plants, and 35 ± 69 μg of 45-kDa rCIa1/g of soluble protein (2.1 mg/kg seeds) in the seeds of the α-amy-ε-45kDa plants (Table 2). In the best T0 line (2845), the average 45-kDa rCIa1 expression was 283 ± 248 μg of 45-kDa rCIa1/g of soluble protein (n = 6). This corresponds to 12 mg of 45-kDa rCIa1/kg seeds (Table 2). By optimizing the parameters of the germination process, such as time and temperature, it could be possible to increase further the 45-kDa rCIa1 yields.

Distribution of the 45-kDa rCIa1 fragment expression in seeds

The distribution of the 45-kDa rCIa1 in seeds was studied by dissecting the seed coat, the embryo and the endosperm from the immature seeds of T2 plants carrying the Ubi-I-ε-45-kDa or the GluB1-ε-45-kDa transgene. At the time of analysis, the endosperm represented 45% of the total seed weight and contained 40% of the total extracted protein. The embryo and the seed coat represented 2.6% and 52.4% of the seed weight and contained 10%–15% and 50% of total extracted seed protein respectively. The 45-kDa rCIa1 content of the dissected seeds was analysed by ELISA. In case of expression from the GluB1 promoter, most of the 45-kDa rCIa1 (∼77%) was found in the endosperm (Figure 3), with 18% in the seed coat, and only traces of 45-rDa rCIa1 were present in the embryo (∼5%). However, some of the detected 45-kDa rCIa1 in the embryo and seed coat may originate from endosperm contamination because of the inaccuracy of the dissection procedure. In the seeds in which expression was driven from the Ubi-I promoter, the expression of 45-kDa rCIa1 was dispersed throughout the whole seed (Figure 3), whereas most of the total protein was found in the endosperm (∼42%) and seed coat (∼57%).

Figure 3.

 The 45-kDa rCIa1 expression in different parts of seeds. Immature seeds were dissected into embryos, endosperms and seed coats. Three embryos, endosperms or seed coats were pooled to provide one sample. Altogether three samples per transgenic T2 line were analysed. The 45-kDa rCIa1 expression was analysed by enzyme-linked-immunosorbent serologic assay and normalized with total protein concentration. The pattern of 45-kDa rCIa1 expression is presented as average percentages. (a) Expression in the seeds of four T2 lines originating from four different T1 lines carrying Ubi-I-ε-45-kDa. (b) Expression in the seeds of four T2 lines originating from two T1 lines carrying GluB1-ε-45kDa.

Characterization of barley-produced 45-kDa rCIa1 fragment

Extraction of the 45-kDa rCIa1 fragment from transgenic barley seeds

Collagen-related proteins are acid soluble proteins, whereas many plant proteins precipitate in acidic conditions (Ruggiero et al., 2000). Therefore, in this study an acidic extraction buffer was used to extract the full-length CIa1 and 45-kDa rCIa1. As maximal extraction of these recombinant proteins is desired, we studied the efficiency of the process by determining the effect of re-extraction on the 45-kDa rCIa1 yields in the T1 seeds carrying the Ubi-I-ε-45-kDa. After 1-h of extraction on ice, the extraction procedure was repeated on the cell debris for 16 ± 2 h. The theoretical total protein content of barley seeds is 12% of the seed mass. One hour extraction released ∼14% of the theoretical protein amount (n = 66 seeds). Re-extraction released a further 6% of the theoretical protein amount. Thus, re-extraction increased the amount of extracted total protein by 43% and altogether the two extractions released 20% of the theoretical protein amount. By contrast, the re-extraction increased the 45-kDa rCIa1 amount only by 22% (n = 55 seeds). After the first extraction the average 45-kDa rCIa1 content of the analysed seeds was 6.8 mg/kg of seeds, but when the combined 45-kDa rCIa1 content of the two extractions was used in the calculations, 41% higher yields were obtained (9.6 mg/kg of seeds). As a result of the more efficient re-extraction of the total protein compared with the 45-kDa rCIa1, the 45-kDa rCIa1 amount per total protein was on average 18% lower after the second extraction (487 vs. 401 μg of 45-kDa rCIa1/g of soluble protein).

Purification and biochemical characterization of the 45-kDa rCIa1

T1 plants derived from the T0 line 1785 (GluB1-ε-45kDa) were propagated in a greenhouse to obtain sufficient seed material for purification and characterization of the 45-kDa rCIa1. Seeds from the PCR-positive plants were harvested, pooled and used for the 45-kDa rCIa1 extraction. The 45-kDa rCIa1 was purified to homogeneity by sequential ion-exchange (IEX) and gel filtration (GF) chromatography. Collected fractions were subjected to ELISA and Western blot analysis to identify and combine those fractions that contained the 45-kDa rCIa1. Electrophoretic patterns of the acid-extracted proteins (lane 2), the proteins precipitated with 80% acetone (lane 3), and the eluates from IEX (lane 4) and the GF columns (lanes 5 and 9) are shown in Figure 4. Each step increased the purity of the 45-kDa rCIa1. After the final purification step two major proteins of ∼60 and 40 kDa and one minor protein of ∼54 kDa were observed in sodium dodecyl sulphate (SDS) polyacrylamide gel (Figure 4, lanes 5 and 9). All these three protein forms were recognized by the polyclonal antibody raised against a 25-kDa rCIa1 of the helical part of the rCIa1 (Figure 4, lane 14). Therefore, the different motility of the proteins in the SDS–polyacrylamide gel electrophoresis gel suggests that three different conformations of the 45-kDa rCIa1 were present in the protein preparation. Interestingly, a difference in staining intensities of the different forms was observed between Coomassie staining, silver staining and Western blot. According to the Western blot analysis the relative amount of the protein migrating as a 40 kDa protein increased during purification.

Figure 4.

 Purification of 45-kDa rCIa1 from seeds and its biochemical characterization. The 45-kDa rCIa1 was purified by acetone precipitation, ion exchange chromatography and gel filtration. Purification of 45-kDa rCIa1 was followed by Coomassie and silver staining and by Western blotting. Pichia pastoris-derived 45-kDa rCIa1 (labelled as Pp45-kDa) was used as a reference material.

Matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) MS analysis was carried out with the purified protein preparation containing all three different forms of the 45-kDa rCIa1. Two replicate analyses revealed that the protein sample contained a single protein having a molecular mass of 44 509–44 612 Da depending on whether the protein sample was desalted using C4 or C1 columns. The obtained molecular masses correlated well with the expected theoretical molecular mass of the 45-kDa rCIa1 (44 528 Da). Thus, the obtained mass corresponds to an accurately cleaved translation product from which the N-terminal signal sequence (ss) has been removed but the ER retention signal His-Asp-Glu-Leu (HDEL) is retained. The molecular mass of the full-length protein also suggests that the major part of the barley-produced 45-kDa rCIa1 had no detectable post-translational modifications.

The 60-kDa, 54-kDa, and 40-kDa protein bands were isolated from the gel and in-gel digested with trypsin. The recovered peptides were analysed by MALDI TOF MS. The three protein bands contained all CIa1-derived non-modified tryptic peptides with molecular masses above 900 Da. The masses of these trypsin-generated peptides were 882.5, 1084.5 (weak), 1089.6 (weak), 1145.6 (weak), 1160.6 (weak), 1226.6 (weak), 1296.6, 1427.7, 1558.8 (weak), 1690.8, 1746.9, 1784.8, 1992.9, 2072.9, 2085.9, 2133.9 (weak), 2134.9, 2168.0, 2233.1 and 4016.9 Da. These peptides covered 79% of the primary 45-kDa rCIa1 fragment amino acid sequence, indicating that the analysed proteins were identical, but that for an unknown reason had different mobility in SDS-PAGE. The peptide masses correlated with data obtained from the analysis of the total mass of the protein, indicating that the majority of the barley-produced 45-kDa rCIa1 was not significantly modified post-translationally. The N- and C-terminal tryptic peptides were too small to be detected by MS. Furthermore, sequencing by Edman degradation was hampered by a blocked N-terminus. Thus, the sequence of the N- and C-termini of the 45-kDa rCIa1 could not be verified.

The MS data were analysed thoroughly to determine the possibility of a low level of post-translational modifications. Some masses could be explained by peptides modified with one or more hydroxyl groups linked to their prolines (data not shown). The peptide consisting of amino acids 7–20 is phosphorylated in the 8.5 kDa fragment of CIa1 produced in P. pastoris (Olsen et al., 2005). The non-modified peptide with molecular mass of 1226.6 Da was present in the barley-derived 45-kDa rCIa1, but no phosphorylated form was detected. Glucosyl- and galactosyl-residues are transferred to the hydroxyl groups of hydroxylysine during collagen biosynthesis (Gelse et al., 2003). The possibility of glycosylation of the purified 45-kDa rCIa1 was determined using a glycoprotein-specific stain. No glycosylation was detected (data not shown).

Amino acid composition analysis of purified 45-kDa rCIa1 is shown in Table 3. The obtained amino acid composition of the barley-derived 45-kDa rCIa1 was compared with the theoretical amino acid composition of the 45-kDa fragment based on the human rCIa1 composition (Miller, 1982) and to corn seed-derived purified 45-kDa rCIa1 (Zhang 2009b). The barley grain-derived 45-kDa rCIa1 had a similar glycine to proline ratio (1.43) to that of the human CIa1 (1.42). The amino acid analysis suggests that barley seeds contain low level of endogenous enzymatic activity capable of hydroxylating 2.8% of the rCIa1 prolines resulting in a level of prolyl hydroxylation significantly lower than in human CIa1 where typically nearly half of the proline residues are hydroxylated. A similar low level of prolyl hydroxylation to that seen in barley seed was observed in the same 45-kDa rCIa1 fragment accumulating in corn seed (Zhang et al. 2009b). Therefore, to achieve prolyl hydroxylation similar to human CIa1 the co-expression of mammalian prolyl 4-hydroxylase necessary. An 88.5% fidelity to theoretical values was observed in the amino acid analysis. For a pure protein there is typically a 95% agreement with theoretical values. This reduction in fidelity is probably caused by contaminating barley seed proteins within the sample.

Table 3.   Amino acid composition of the barley seed-derived 45-kDa rCIa1 compared with the theoretical calculation based on the same 45-kDa fragment of human type I a1 and corn seed-derived 45-kDa rCIa1
Amino acid Barley seed 45-kDa rCIa1 (%)Theoretical calculation of 45-kDa Human type I homotrimer (%) (Miller, 1982) Corn seed 45-KDa rCIa1 (%) (Zhang et al. 2009b)
Aspartic acid4.74.26.0
Glutamic acid8.27.38.1

Production and evaluation of doubled haploid plants

The zygosity state of the transgene varies in the T2 and T3 seeds. To study the effect of homozygosity on the accumulation of full-length rCla1 and the 45-kDa rCIa1, microspore cultures were initiated to obtain doubled haploid (DH) progeny from the selected transgenic barley lines. DH0 plants were produced from 23 T1 plants and from six T2 plants originating from six T0 lines (Table S5). The microspore culture regenerants that lacked the transgene were used as the non-transgenic control plants. The fertility of the DH0-plants carrying the GluB1-ε-45-kDa cassette was reduced (18% fertile) when compared with the non-transgenic control plants (100%). The total seed yields of the transgenic DH0 plants did not differ statistically from those of the non-transgenic control plants.

Full-length rCIa1 accumulation was analysed from the seeds of the DH0 plants originating from three independent T0 lines. Altogether 14 DH0 lines were analysed by ELISA but no detectable full-length rCIa1 expression was observed (Table S5). A total of 73 DH0 lines accumulating the 45-kDa rCIa1 were analysed. In 56 of them the 45-kDa rCIa1 expression was under the Ubi-I promoter and in seven lines the expression was under the GluB1 promoter. When the protein was expressed under the seed-specific GluB1 promoter, the amount of the 45-kDa rCIa1 in the seeds of DH0 plants was approximately threefold higher than when the protein was expressed under the Ubi-I promoter. The expression levels achieved from the best DH0 lines were 13 mg 45-kDa rCIa1/kg dry seeds when the expression was driven by the Ubi-I and 45 mg 45-kDa rCIa1/kg dry seeds when the expression was driven by the GluB1 promoter (Table 2). These expression levels correspond to 715 ± 308 μg and 2753 ± 1534 μg of 45-kDa rCIa1/g of soluble protein respectively. The accumulation levels were comparable with those of the T2 plants having the highest 45-kDa rCIa1 accumulation levels (see Table 2).


We have previously shown that the full-length rCIa1 can be produced, albeit at low levels, in barley suspension cells (Ritala et al., 2008). In the present study also a low accumulation level of the full-length rCIa1 was observed in barley seeds when expressed under the Ubi-I promoter. These low accumulation levels may have resulted from low transcriptional activity of the Ubi-I in barley seeds (see below). Another possibility for the low accumulation levels of barley-produced rCIa1 may be the lack of suitable hydroxylation of prolines in rCIa1, which is required for formation of stable triple helical molecules (van der Rest and Garrone, 1991). Inability to form triple helical conformation correctly in the ER may lead to degradation of the rCIa1 in cytoplasm via the ER-associated protein degradation pathway. In fact, the rCIa1 produced in barley suspension cells was more susceptible to pepsin digestion than rCIa1 of human origin, indicating that the endogenous prolyl hydroxylases and other processing proteins present in barley seeds cannot compensate for the human enzymes and proteins enabling the formation of stable triple helical collagen (Ritala et al., 2008).

High accumulation of recombinant proteins is necessary for commercialization of the production system. Therefore, strong transcriptional activity is one of the key requirements for high-level expression. Three different promoters were compared for their suitability to express the 45-kDa rCIa1 in barley seeds: a seed-specific GluB1 promoter, a germination-specific barley α-amy promoter and a constitutive maize Ubi-I promoter (Table 2). In T1 seeds, the amounts of the 45-kDa rCIa1 expressed under the GluB1 were 3- to 50-fold higher than if expression was driven from the constitutive Ubi-I promoter. By contrast, the average 45-kDa rCIa1 accumulation levels in seeds from the plants transformed with the Ubi-I-ε-45kDa were 6.5 times higher than the average 45-kDa rCIa1 content in the seeds of the plants carrying the α-amy-45kDa or α-amy-ε-45kDa transgenes. No difference in the average accumulation levels between the plants containing the ε-element from CfMV and the plants containing a non-viral 5′-leader sequence in the mRNAs was observed, indicating that this viral sequence element can not be used in barley seeds to enhance expression levels. Accumulation of the 45-kDa rCIa1 was followed for three generations in the Ubi-I and GluB1 plants. The accumulation detected in the seeds of the best T0 line 1785 decreased in the next generation. However, the 45-kDa rCIa1 levels in the seeds of the T1 and T2 generations were similar, indicating that the accumulation level of the 45-kDa rCIa1 had stabilized. In the next generations of the T0 lines 1755 (Ubi-I-ε-45kDa) and 1781 (GluB1-ε-45kDa), the 45-kDa rCIa1 levels remained constant for the two generations studied (Table 2). It was also evident from these data that the 45-kDa rCIa1 content remained stable in the seeds stored for 15 months.

Transgene homozygosity is a desirable property in transgenic production lines to enhance the stability of the recombinant protein production and to facilitate further breeding of the production lines. Using traditional breeding techniques it takes nearly 10 years to reach homozygosity, whereas the DH breeding technique can develop homozygous lines in one generation. Therefore, DH production significantly reduces the time to develop new cultivars. Homozygous DH0 lines were generated from the Ubi-I and GluB1 plants. The DH0 lines expressing the 45-kDa rCIa1 under the Ubi-I had expression levels comparable with the parental T1 lines used to generate the homozygous plants. Unfortunately, we were not able to regenerate fertile DH0 plants from the T0 line 1785 having the highest expression level of the 45-kDa rCIa1. The reason for this is unknown, but it can be speculated that integration of the foreign genes affected the viability of microspores and pollen, thus giving rise either to microspore cultures with no regenerants or to non-fertile transgenic DH0 plants. To enhance the production yields, the transgenic DH barleys [Hordeum vulgare L. cv. Golden Promise (GP)] could be submitted to a breeding program in which the trait would be introduced to a barley variety with a higher seed yield.

The GluB1 promoter has previously been used to express xylanase in barley. In general, comparisons of different heterologous protein accumulation are difficult because the varying extraction conditions and properties of the protein change the ratio between the extractable total protein and the recombinant target protein. However, the estimated expression level in the highest producing line was ∼0.04% of total extractable seed protein (Patel et al., 2000). In our study, the best homozygous DH-line expressed the 45-kDa rCIa1 at a level of 0.07% of extractable protein (45.1 ± 24.0 mg/kg seeds). In a another heterologous protein accumulation study in barley seeds GluB1 promoter provided at least twofold higher recombinant protein accumulation in the seeds than the endogenous Hor3-1 promoter (Horvath et al., 2000). However, a comparative study of 18 promoters in rice showed that the 1.3 kb GluB1 promoter used in our study was among the weakest promoters studied (Qu and Takaiwa, 2004). The difference in the expression levels between the strongest promoter GluB4 and GluB1 was 22-fold. Recently an oat globulin gene promoter (AsGlo1) was shown to drive strong endosperm-specific expression in barley seeds (Vickers et al., 2006). Expression of β-glucuronidase (GUS) reached up to 10% of soluble protein. In summary, these data indicate that it is probable that the accumulation levels of 45-kDa rCIa1 could be further increased by selecting a stronger promoter driving high transcriptional activity.

The use of promoters that are transcriptionally active only during the germination of seeds precludes transgene expression in the seeds during seed maturation. Thus, the product is not present in the transgenic seeds in the field. This may improve the public acceptance of transgenic plants. The α-amy promoter, active only during germination, has been successfully used in expression of enzymes in barley. In one study, the accumulation levels of β-glucanase reached 0.1% of the total extractable protein (Horvath et al., 2000). In another study, β-glucanase accumulation reached 0.025% of the total extractable protein (Nuutila et al., 1999). In our study, the 45-kDa rCIa1 accumulation levels in the best T0 lines reached 0.028% of total extractable protein. Optimization of the germination process could further increase the accumulation levels.

Although Ubi-I is considered to be a rather weak promoter (Qu and Takaiwa, 2004), three recombinant proteins produced in transgenic corn under the Ubi-I are currently commercially available: avidin, GUS and aprotinin. Commercial production of plant-derived recombinant avidin is based on transgenic corn seeds in which constitutive expression and targeting to extracellular space resulted in accumulation levels of 5.7% of extractable protein (Kusnadi et al., 1998). By contrast, commercially available plant-produced GUS is derived from plants that accumulated GUS at much lower levels, 0.7% of extractable protein (Kusnadi et al., 1998). The third commercially produced recombinant protein, serine protease inhibitor aprotinin, reaches accumulation levels of 0.07% of extractable protein (Zhong et al., 1999). Constitutive expression from the Ubi-I promoter and ER targeting of a scFv in rice resulted in maximal accumulation levels of 0.1% of total extractable protein in rice, whereas the expression levels in wheat were 20-fold lower (Stöger et al., 2002).) In our study, the best Ubi-I-45kDa-expressing DH0-line accumulated 45-kDa rCIa1 up to similar levels as corn seeds expressing aprotinin (0.07% of soluble protein).

Expression of the target protein in certain compartment of the seed allows for the possibility that specifically that fraction of the seed can be collected and used as the starting material for product purification. In our study, the distribution pattern of the 45-kDa rCIa1 varied depending on the promoter used. Earlier studies on the expression of recombinant avidin and GUS in transgenic corn seeds have suggested that the Ubi-I promoter has some tissue preferences (Hood et al., 1997; Kusnadi et al., 1998). The protein accumulation was mainly observed in the embryo/germ, but the distribution also depended on the expressed proteins. Avidin was mainly found in the embryo/germ (93%), whereas 55% of GUS was found in the germ and 22% in the endosperm (Hood et al., 1997). No GUS was detected in the hulls (Kusnadi et al., 1998), whereas 7% of the avidin was found in the hulls. In our study, the 45-kDa rCIa1 fragment expressed under the Ubi-I promoter accumulated evenly in all parts of the barley seed. However, expression profiles varied between the studied T2 lines. This may have resulted from the differences in the developmental stages of the immature seeds used in the analysis. In rice, the Ubi-I promotes heterologous protein expression in young and metabolically active tissues but not in older, less active tissues (Cornejo et al., 1993). The rice GluB1 promoter has been shown to confer endosperm-specific expression of recombinant proteins in barley (Patel et al., 2000; Qu and Takaiwa, 2004). At the early stages of seed development, the GluB1 promoter is transciptionally most active in the aleurone and subaleurone as well as in the region of endosperm that is located close to the embryo (Qu and Takaiwa, 2004). At later stages (17-day after flowering), the transcriptional activity also increases in the inner parts of endosperm (Qu and Takaiwa, 2004). We found that when the GluB1 promoter was used the majority of the expressed 45-kDa rCIa1 accumulated in the endosperm.

Collagens are acid soluble, and therefore an acidic extraction buffer was used to extract the seeds. As an added advantage, under acidic conditions the majority of the barley seed proteins are insoluble. Glutelins represent a group of barley seed proteins that are soluble in dilute acid or alkali. Linko et al. (1989) reported that glutelins constitute ∼38% of the total protein content of the seeds. A 1-h extraction released approximately one-third of the assumed acid-soluble fraction of the proteins, whereas further extraction improved the total protein yield to almost 60% of the expected acid-soluble proteins. The re-extraction experiment indicated that it is beneficial to repeat the extraction procedure at least twice to purify the maximal amount of 45-kDa rCIa1 from barley seeds. Optimization of the extraction conditions could further increase the yield of 45-kDa rCIa1. Extraction strategy based on one extraction step resulting to maximal yield would be desirable for building an efficient production process.

In mammalian cells during synthesis and maturation collagen undergoes several post-translational modification. For example certain proline and lysine residues are hydroxylated, and some lysines undergo oxidative deamination. Limited glycosylation also takes place, but no phosphorylation has been detected in collagen (Gelse et al., 2003). Protein glycosylation in plants differs from that of mammalian cells (Faye et al., 2005). However, ER retention should result in human-like non-immunogenic recombinant proteins, as the high-mannose-type N-glycans produced in the ER are structurally common among plants and mammals. The glycoprotein-specific staining indicated that the barley-produced 45-kDa rCIa1 is not glycosylated. MALDI-TOF analysis of the purified 45-kDa rCIa1 and its tryptic peptides further confirmed the glycoprotein staining result showing that the majority of the barley-produced 45-kDa rCIa1 had no detectable post-translational modifications. Some peptide masses obtained from the minor peaks indicated potential hydroxylation of prolines. The presence of low amounts of hydroxyprolines was verified by the amino acid composition analysis. This shows a difference between barley and yeast expression systems, as no proline hydroxylation can be detected in yeast-derived product (Olsen et al., 2003). The impact of this low level of prolyl hydroxylation on the functionality of the barley-derived 45-kDa rCIa1 can not be estimated until the protein is tested as a component of specific products.

In conclusion, we have successfully expressed both the full-length and the 45-kDa fragment of human collagen (I) α-1 chain in barley seeds. The best productivity was obtained with the seed-specific GluB1 promoter. The average yield of barley in Finland is 3280 kg/ha. Using the 45-kDa rCIa1 accumulation data from the seeds of the best DH0-line 2692 (45.1 ± 24.0 mg/kg seeds), a production level of ∼150 g 45-kDa rCIa1/ha can be achieved. At this accumulation level, committing 53 000 ha of the barley cultivation to 45-kDa rCIa1 production could yield over 5 metric tons annually. This area equals to 10% of the land dedicated to barley production in Finland. A low production cost could be achieved by co-production of the recombinant protein with starch, fibres, fatty acids, phytochemicals and distiller’s grains for use as feeds and for conversion to biofuels.

Experimental procedures

Constructs with the human recombinant collagen (I) α-1 (rCIa1) and the 45-kDa rCIa1 fragment for barley expression

The cDNA for 3.2-kb coding sequence of the Homo sapiens gene for collagen I α-1 chain (cIa1, pUO005, NCBI accession no. NP000088) was obtained from Fibrogen Inc., San Francisco, CA, USA. Codon usage of the gene had been optimized for monocot expression resulting in a GC content above 60%. The amino acids 372–869 of Cla1 are included in the 45-kDa rCIa1. The constructs used in this study are listed in Table 1 and a schematic drawing is shown in Figure 1. A gene cassette carrying the Arabidopsis thaliana basic chitinase ss, N-terminal telopeptide sequence, the human collagen I α-1 chain cDNA (full-length CIa1), C-terminal telopeptide sequence, and the foldon domain gene obtained from bacteriophage T4 fibritin protein fused to the HDEL sequence (this cassette was termed cIa1) and a gene cassette carrying the chitinase ss, the 45-kDa rCIa1 and the HDEL sequence (this cassette was termed 45kDa) were constructed. The details of cloning the foldon domain, telopeptide, chitinase ss and ER retention signal to cIa1 are described in Ritala et al. (2008). In some of the constructs the ε-element of CfMV (Mäkinen et al., 1995) was inserted in front of the cIa1 or 45-kDa rCIa1 encoding region with the aim of improving the expression levels. The chitinase ss and the sequence for ε-element or 45-kDa rCIa1 were fused by overlapping extension polymerase chain reaction (PCR) with Pfu turbo polymerase (Stratagene, La Jolla, CA, USA) with the following primers. The forward primers used were FWA 5′-GTTGC TCG AG TGA TAA TAG TGC GAA GAA AGA C-3′ for XhoI-ε-ss or primer FWB 5′-GAC CGC TCG AGA TGG CTA AGA CTA ATC TTT TTC-3′ for XhoI-ss and a reverse primer 5′-GGC ACC TGG CAG TCC TTC GGC CGA GGA TAA TG-3′. The forward primers contained XhoI restriction site (underlined) and FWB nucleotides GCT downstream from the ATG to create an optimal translation initiation codon for monocots (Lukaszewicz et al., 2000). This PCR step generated a 3′-overhang complementary to the 5′-end of the 45-kDa rCIa1. The 45kDa-‘HDEL’-stop-BamHI with a 5′-overhang complementary to 3′ of chitinase ss was prepared by PCR with the forward primer 5′-CAT TAT CCT CGG CCG AAG GAC TGC CAG GTG CCA AAG GAC-3′ and reverse primer 45kDa-HDEL-REV 5′-CGG GAT CCT CAA AGC TCA TCA TGT GGC GAA CCA TCA CGT CC-3′) using pUO005 as a template. The reverse primer contained BamHI site (underlined). The two PCR fragments were joined by overlap extension PCR with 45kDa-HDEL-Rev primer and FWA or FWB forward primers. Then the 45-kDa rCIa1 expression cassettes were placed under the maize ubiquitin promoter and the first intron (Ubi-I) by introducing blunted 45-kDa rCIa1 expression cassettes into pAHC25 (Christensen and Quail, 1996) within SmaI and EcoRV restriction sites. GluB1 promoter was cloned from genomic DNA of Rice cv. VNIIR-17 using primer sequences 5′-CAA CTG CAG TTT TGA GGA ATT TTA GAA GTT GAA CAG-3′ and 5′-CGC GGA TCC TCG AGC TTA AGC TAA TTG ATG TGA GTT C-3′ designed according to the gene bank sequence X54314. The sequences of PCR products showed 97% identity to data bank sequence. The GluB1-promoter was fused as an EcoRI-Xho1 fragment to 45kDa to obtain the pJAM-plasmids. The pEW- and pJAM-plasmids contained a bialaphos resistance gene (bar) under the control of Ubi-I.

For the pSAM-plasmids, vector pWBVec8 (kindly provided by CSIRO Plant Industry, Canberra, Australia; Wang et al., 1997, 1998) was used as a backbone. The barley amylase fusion promoter 46/4-6 (α-amy, kindly provided by John Rogers, Washington State University, Pullman, WA, USA) was fused as a HindIII-NcoI fragment to the 45-kDa rCIa1-encoding sequence. The α-amy-45kDa was introduced as a HindIII fragment to a unique HindIII restriction site of pWBVec8. The pSAM-plasmids contained a hygromycin resistance gene (hph) with a castor bean catalase-1 intron (CAT1I) under the control of Cauliflower mosaic virus 35S-promoter.

Gene transfer and production of transgenic plants

The particle bombardment, culture, selection and regeneration of the bombarded half-embryos of barley (H. vulgare L. cv. GP) followed the procedure of Wan and Lemaux (1994). The constructs used in bombardments are shown in Table 1. After four rounds of selection on solid callus induction medium (CIM) supplemented with 2.5 mg/L of Dicamba (Sigma-Aldrich, St. Louis, MO, USA) and 5 mg/L of Bialaphos (Duchefa, Haarlem, The Netherlands), the viable cell lines were transferred to regeneration medium (FHG) supplemented with 1 mg/L of BA (Sigma-Aldrich) and 5 mg/L of Bialaphos. The regenerants were rooted on CIM supplemented with 3 mg/L of Bialaphos.

The Agrobacterium-mediated transformation, selection and regeneration of transformed barley embryos (H. vulgare L. cv. GP) followed the procedure of Tingay et al. (1997). Agrobacterium strain AGL0 (Rhizobium radiobacter ATCC BAA-100; Lazo et al., 1991) was used with a co-cultivation period of 3 days at +22 °C. The constructs used in infections are shown in Table 1. After three rounds of selection on solid CIM supplemented with 2.5 mg/L of Dicamba, 50 mg/L of Hygromycin (Sigma-Aldrich) and 150 mg/L of Timentin (Duchefa), the viable cell lines were transferred to regeneration medium (FHG) supplemented with 1 mg/L of BA, 50 mg/L of Hygromycin and 150 mg/L of Timentin. The regenerants were rooted on CIM supplemented with 50 mg/L of Hygromycin and 150 mg/L of Timentin.

The rooted plants were potted in soil mix (Vermiculite : peat : soil, 2 : 1 : 1) and grown in a greenhouse (22/13 °C day/night, 19 h light 300 μmol/m2/s, 40%–50% relative humidity). The plants were fertilized weekly (Kukkien Y-lannos, Maatalouspalvelu Oy, Finland: Biolan S.M.3, Biolan Oy, Finland; and Puutarhan täyslannos, Kemira Oy, Finland) according to the manufacturer’s instructions. A schematic illustration of the production of transgenic barley plants is shown in Figure S1.

PCR analysis of transgenic plants

Total genomic DNA from the leaves of the regenerated plants was isolated using the Cetyl trimethyl ammonium bromide (CTAB) method (Murray and Thompson, 1980). Putative transgenic plants were screened for the presence of the transgene by PCR reactions containing 100 ng of plant DNA, 100 μm dNTPs, 0.5 mm MgCl2, 1× Taq DNA polymerase buffer and 1.25 U Taq DNA polymerase (Perkin Elmer, Waltham, MA, USA) in a final volume of 50 μL. The primers 255 (5′-GGT CCC GCT GGA AAG GAT-3′) and 253 (5′-GCT CCA GGA GGA CCT GCA GA-3′) were used to amplify an internal 797-bp fragment of cIa1 or 45kDa. Thirty-five cycles were performed under the following conditions: 30 s denaturation at 95 °C, 1 min annealing at 61 °C, and 1 min 30 s extension at 72 °C. PCR products were analysed in agarose gels.

Production of doubled haploid lines

The homozygosity was induced in transgenic barley lines by microspore culture (Figure S1). The procedure followed in detail that one described in Ritala et al., 2005. Spikes of the transgenic T1 or T2 generation plants carrying microspores in the late uni-nucleate stage were cold pretreated at 4 °C for 4 weeks. Teflon rod maceration, with 0.3 m mannitol as isolation medium, and a plating density of 100 000 viable microspores were used. The regenerated and rooted plantlets were treated with colchicine (Sigma-Aldrich) by soaking the tip-cut roots in 0.05% (w/v) colchicine solution for 5 h prior to potting in soil. The plants were screened by PCR as described in PCR analysis of transgenic plants.

Harvesting and phenotyping of the transgenic plants

The mature seeds of the transgenic and the DH plants were harvested individually. Fertility of the plants was estimated and categorized as fully fertile, carrying some seeds or sterile. The weight of 100 seeds and the total seed-set (yield) were measured and an estimate of total seed number was calculated (total weight/100 seed weight × 100). The seeds were stored at +14 °C. Transgenic seeds carrying 45kDa gene under the α-amy-promoter were surface-sterilized and germinated in darkness at 14 °C. Those seeds that germinated were analysed after 6–12 days of germination. At the time of the analysis, the roots were ∼1.5–3 cm long.

Western blot and ELISA

Total protein was extracted from seeds by grinding the individual seeds with a pestle and mortar in the presence of fine sea sand. Proteins were extracted on ice for 1 h with an extraction buffer containing 0.1 m HCl, 2.5 mm ethylenediaminetetraacetic acid (EDTA), 1 μm pepstatin A (Sigma) and protease inhibitor coctail (Complete, Mini, EDTA-free; Roche Diagnostics GmbH, Mannheim, Germany). Lysates were cleared by centrifugation at 10 000 g for 10 min at +4 °C. Total protein content of extracts was analysed by a bicinconic acid (BCA) method (Pierce, Rockford, IL, USA).

Cell extracts were analysed by SDS-PAGE followed by Western blotting as described by Ritala et al., 2008. Membranes were probed with polyclonal antibodies against a 25-kDa fragment of the helical part of rCIa1 (anti-25kDa Cla1 antibody, CA725; FibroGen Inc., San Francisco, CA, USA). The secondary antibody used was horse radish peroxidase (HRP)-goat anti-rabbit IgG (Promega, Madison, WI, USA) or alkaline phosphatase (AP)-goat anti-rabbit IgG (Promega). Antigen-antibody complexes were visualized by ECL chemiluminescent reagents (GE Healthcare, Uppsala, Sweden) or staining with Western Blue stabilized substrate for AP (Promega).

Seed extracts were analysed for their CIa1-related protein content by ELISA using the polyclonal anti-25 kDa rCIa1 antibody (FibroGen Inc.) as the primary antibody and HRP-conjugated anti-rabbit IgG (Promega) as the secondary antibody. The purified non-hydroxylated P. pastoris-produced rCIa1 (PprCIa1) and 45-kDa rCIa1 (Pp45kDa) were used as a standard protein (FibroGen Inc.). The ELISA was performed as described by Ritala et al., 2008 with slight modification. In short, wells of the ELISA plates (Costar High Binding, Corning, NY, USA) were coated overnight with 5 ng per well of heat-denatured non-hydroxylated PpCIa1 in phosphate buffer solution (PBS) at 4 °C. After washings (10 mm PBS, 0.05% Tween 20, 0.05% Proclin 300, pH 7.0), the plates were blocked with 0.1% bovine serum albumin (BSA) in PBS for 1 h at room temperature. Samples and the standard proteins were heated at 65 °C for 30 min and cooled on ice and diluted in the assay buffer [100 mm phosphate-buffered saline (PBS), 0.05% Tween 20, 0.05% Proclin 300, 0.1% BSA, pH 7]. Dilutions (50 μL) were pipetted into the wells together with 50 μL of the polyclonal anti-25kDa antibody (CA725; Fibrogen Inc.) diluted 1 : 4000 in the assay buffer. After 1-h incubation at 4 °C, wells were washed three times with the wash buffer. The HRP-conjugated anti-rabbit antibody (Promega) was diluted in the assay buffer (1 : 5000) and applied into the wells (100 μL/well). The plate was then incubated for 1 h at room temperature. Wells were washed three times, after which TMB substrate solution (TMB single solution, Zymed; Invitrogen, Carlsbad, CA, USA) was added. The reaction was stopped after 30 min incubation with 1 m HCl and the absorbance was read immediately with a micro-plate reader (model 680; Biorad, Hercules, CA, USA) at 450 nm. Relative concentrations of the recombinant proteins within the plant extracts were determined using dilution series of PpCIa1 or Pp45kDa. The full-length rCIa1 ELISA had the detection limit of 3 ng/mL and that of the 45-kDa rCIa1 ELISA was 6 ng/mL. If the signal obtained from ELISA indicated that either the rCIa1 or 45-kDa rCIa1 level was below the detection limit, seeds were regarded as non-expressing.

Characterization of the barley-produced 45-kDa rCIa1

Approximately 100 g of dry seeds from the T1 plants originating from the T0 line 1785 expressing the 45-kDa rCIa1 under the GluB1 promoter were frozen in liquid nitrogen and milled into a fine powder in an analytical mill (A11 basic; IKA-Werke GMBH & CO.KG, Staufer, Germany) for 1 min. Further homogenization was performed with a pestle in a mortar containing 25 mL of extraction buffer [0.1 m HCl, 2 mm EDTA, five pills of protease inhibitor mix (Complete, Mini, EDTA-free; Roche Diagnostics), 4 μm Leupeptin (Sigma), 0. 4 mm phenylmethylsulphonylfluoride (PMFS; Sigma)] for 10 min on ice. After 1- h incubation on ice, the slurry was centrifuged at 30 600 g for 20 min at 4 °C. The extraction was repeated with the pellets. The pooled supernatants were adjusted to contain 20% (vol/vol) acetone, precipitated overnight at −20 °C, and centrifuged at 30 600 g for 30 min at 4 °C. Next, acetone concentration in the supernatants was adjusted to 80% and proteins were precipitated for 1 h at −20 °C. Finally, precipitated proteins were collected by 30 min centrifugation at 36 600 g at 4 °C and dissolved in 50 mL of IEX-loading buffer (50 mm sodium acetate, pH 4.6, supplemented with three pills of protease inhibitor 1.5 mm PMFS, 0.5 μm of pepstatin). The protein solution was clarified by centrifugation at 36 600 g for 1 h, and the supernatant was ultrafiltrated with steriflip (Millipore, Billerica, MA, USA). An IEX-column (SP-sepharose fast flow resin; GE-Healthcare 17-072910, Uppsala, Sweden) was packed, connected to ÄKTAprime (Amershamn Pharmacia Biotech, Uppsala, Sweden) and equilibrated according to manufacturer’s instructions with 50 mm sodium acetate running buffer, pH 4.6. Samples were injected into the column in two separate runs through a super loop (40 mL, 3 mL/min). After 100 mL wash with the running buffer, the column was eluted with 400 mL of running buffer with a linear gradient of sodium chloride (NaCl, 0–0.5 m). Fractions (4 mL) were collected and total protein contents were analysed with a BCA-kit (Pierce).

The 45-kDa rCIa1 concentration in the collected fractions was assayed by ELISA, and fractions containing the highest amounts of 45-kDa rCIa1 (elution fractions containing 0.12–0.17 m NaCl) were pooled and subjected to 40% ammonium sulphate precipitation. After centrifugation (26 900 g), pellets were dissolved in water supplemented with the protease inhibitor mix (one pill to 25 mL) on ice. Salts were removed by a Sephadex G-25 column (PD-10; GE-Healthcare) and the proteins were eluted with 3.5 mL of water and concentrated by lyophilization. Pellets were resuspended in 2 mL of 50 mm sodium acetate buffer containing 0.15 m NaCl. Samples were loaded to a HIPREP 16/60 Sephacryl S-200 GF column (Amersham Pharmacia Biotech) connected to ÄKTAprime. Buffer flow was adjusted to 0.7 mL/min and 2 mL fractions were collected. Total protein and 45-kDa rCIa1 content of the fractions were analysed as described above. The fractions containing 45-kDa rCIa1 were pooled, lyophilized and resuspended in 2.5 mL of 10 mm Tris-(hydroxymethyl)aminomethane–HCl (Tris–HCl), pH 6.8, 1 mm PMFS. Once again, salts were removed by the Sephadex G-25 column, the eluate was lyophilized, and the proteins were resuspended in 0.1 mL of water.

The protein preparation was subjected to reverse-phase chromatography on a C4 (C4, 5 cm, 300 Å; Phenomenex, Torrance, CA, USA) or C1 (C1, 10 cm, 250 Å; Tosoh Corp, Tokyo, Japan) column and eluted with a linear gradient of acetonitrile (0%–100% in 60 min) in 0.1% trifluoroacetic acid. Fractions containing proteins were analysed by SDS-PAGE. The molecular masses of the isolated proteins were determined by MALDI-TOF MS. Coomassie Brilliant Blue R250-stained protein bands of interest were cut out of the gel and in-gel digested essentially as described by Shevchenko et al., 1996. Proteins were reduced with dithiotreitol and alkylated with iodoacetamide before digestion with trypsin (Sequencing Grade Modified Trypsin; Promega). After desalting, the recovered peptides were subjected to MALDI-TOF analysis. MALDI-TOF mass spectra for mass fingerprinting were acquired using an Ultraflex TOF/TOF instrument (Bruker-Daltonik GmbH, Bremen, Germany). GelCode Glycoprotein Stain (Pierce) was used according to the GelCode Glycoprotein Stain Protocol provided by the supplier to detect whether barley seed-derived 45-kDa rCIa1 was a glycoprotein.

The purified 45-kDa rCIa1 sample was subjected to acid hydrolysis (6 n HCL, 1% Phenol for 24 h at 110 °C). After hydrolysis the sample was reconstituted in 20 mm HCl and transferred to an high performance liquid chromatography (HPLC) vial then loaded onto an HPLC where the hydrolyzates were derivitized (Agillent Techologies, Santa Clara, CA, USA) using o-phthalaldehyde for primary amino acids and 9-fluorenylmethyl chloroformate for secondary amino acids. Primary amino acids were detected using a UV detector at 366 nm. Secondary amino acids were detected with a Fluorometer using 266 nm excitation and 305 nm emission. Peak areas of individual amino acids in the sample were compared with corresponding peak areas in an amino acid standard curve (Pierce Standard H supplemented with Hydroxyproline). Protein composition was calculated by comparing the relative amounts of the amino acids in the sample.


The authors are grateful for the technical assistance of Sheri Almeda and Frank Buschman (FibroGen Inc.), Marica Björkman, Anne Heikkinen, Sirkka-Liisa Kanervo and Jaana Rikkinen. Funding provided by the Technology Development Agency of Finland (decision nos: 40680/01 and 40014/04) and Ministry of Agriculture and Forestry (decision no. 4778/501/2004) is gratefully acknowledged.