WO2006005362A1 - Production of mammalian proteins in plant cells - Google Patents

Abstract

A method for producing Solulin or DSPAαl is provided, which method comprises growing plant cells containing an integrated sequence comprising a functional transcriptional cassette comprising a structural gene encoding for said Solulin or desmoteplase.

Description

July 13, 2004

"Production of mammalian proteins in plant cells"

The invention relates to the field of production of mammalian proteins or peptides in plant cells ("molecular farming").

For economic production of proteins and peptides it would be advantageous to use plant cells - e.g. either as cell suspension cultures or as entire plants - to produce these proteins or peptides. However, within these production systems there is an inherent risk of improper protein folding or other post translations modifications which render the produced protein unsuitable - or less effective - for the envisaged use or purpose. Thus, in the past especially attempts in producing proteins for therapeutic purposes in "molecular farm¬ ing" frequently failed.

Therefore, it is an objective to provide a method for the production of thera¬ peutically active mammalian proteins or peptides, especially of the thrombo¬ modulin analog Solulin and of the plasminogen activator of Desmodus rotun- dus (DSPA).

EXAMPLE 1 :

PRODUCTION OF A THROMBOMODULIN ANALOG IN PLANT CELLS

Thrombomodulin is an endothelial, cell-surface glycoprotein that binds thrombin and accelerates both the thrombin-dependent activation of protein C and the inhibition of antithrombin III (Esmon et al., 1982). Thrombomodulin consists of a lectin-like domain, six epidermal growth factor-like domains, an O-linked glycosylation site, a transmembrane region and a cytosolic domain (Suzuki et al., 1987). Solulin, a soluble derivative of human thrombomodulin, lacks the transmembrane and cytosolic domains and has six mutations that improve protein stability (Weisel et al., 1996). Solulin is a glycoprotein con¬ taining an 18-amino-acid N-terminal signal sequence, which promotes co- translational transport into the secretory pathway in mammalian cells (Glaser et al., 1991). The mature glycosylated protein is 487 amino acids in length has an apparent molecular mass of -75 kDa as determined by SDS-PAGE under non-reducing conditions. Thrombotic disorders are caused by clot for¬ mation in response to a stimulus like a damaged vessel wall. This triggers the coagulation cascade generating thrombin, which is converting fibrinogen to fibrin, the matrix of the clot. The systemic administration of soluble, oxida¬ tion-resistant thrombomodulin analogs such as Solulin inhibits the proco- agulant activity of thrombin by building a thrombin-thrombomodulin complex and acting as an anticoagulant by activation of protein C, which in turn at¬ tenuates the clotting cascade by inactivation of several activated cofactors (Esmon et al., 1989). This mechanism can be used to treat diseases in which thrombus formation plays a significant etiological role, e.g. myocardial infarc¬ tion, disseminated intravascular coagulation (Gonda et al., 1993), deep vein thrombosis, pulmonary embolism, septic shock, acute respiratory distress syndrome, unstable angina and other arterial and venous occlusive condi¬ tions (Glaser et al., 1991 ; Solis et al., 1994; Aoki et al., 1994).

Until now, recombinant thrombomodulin is produced in mammalian cell cul¬ ture such as Chinese hamster ovary (CHO) cells (Lin et al., 1994). However, the increasing demand on this protein for the development of various appli¬ cations in the treatment of thrombotic and vascular disease therapies re¬ quires a safe and cost-effective large-scale production system. A number of expression platforms based on plants have been developed recently, and these offer several advantages in terms of production scales, safety and economy for the production of pharmaceutical proteins (Twyman et al., 2003). We therefore assessed the feasibility of producing recombinant solu- ble thrombomodulin in tobacco plants and BY2 cell suspension cultures. To evaluate the influence of protein localization on the stability and accumula¬ tion of Solulin, the recombinant protein was fused to apoplast, protein body and vacuolar targeting sequences. The highest yields in transgenic plants were achieved by apoplast targeting, whereas the highest yields in BY2 cells were achieved using a targeting sequence for protein bodies. Active Solulin was produced in plant cells showing correct processing of the signal sequence demonstrating the potential of plants as an alternative production platform.

Materials and methods Bacteria and plants

Escherichia coli strain DH5α (Ausubel et al., 1994) was used for general cloning. Agrobacterium tumefaciens strain GV3101 (pMp90RK, GmR KmR), RifR (Koncz & Schell, 1986) was used for agroinfiltration and stable trans¬ formation. Tobacco plants Nicotiana tabacυm cv. Petite Havana SR1 and cv. Maryland Mammoth were used for agroinfiltration, and cv. Maryland Mam¬ moth was also used for stable transformation. Suspension cells originated from N. tabacum L. cv. bright yellow 2 (BY2).

Plant expression vector design and construction

The cDNA of the thrombomodulin derivative ( Solulin) including the original signal sequence (ORI) was amplified by PCR from the source plasmid pTHR525 (provided by PAION GmbH, Aachen, Germany) using the primer set Solulin-for (signal) 5'-TTT ATA AAA AAA AAA AAA ACA ATG CTT GGG GTC CTG GTC-3' and Solulin-back (stop) 5'- GCG CGA AGC TTC TCG AGT TTA CGG AGG AGT CAA GGT AGA CCC-3'. The primary PCR prod¬ uct was then used as the template in a second PCR with primers Solulin-for (5'UT) 5'-CGC GAA TTC ACA ACA CAA ATC AGA TTT ATA GAG AGA TTT ATA AAA AAA AAA AAA AC-3' and Solulin-back (stop) to fuse the Pet- roselinum crispum chalcone synthase 5' untranslated region to the 5' end of the cDNA. The product was ligated into the vector pTRAkc (a derivative of pPAM, GenBank accession number AY027531 , with an Xba\ site omitted), using the EcoRI and Xho\ sites. This vector contained the double-enhanced CaMV 35S promoter (Kay et al., 1987) and the CaMV termination sequence, preceded by a polyadenylation site (Figure 1). The resulting construct was named Solulin-ORI.

In order to direct Solulin to the apoplast, the cDNA was fused to an apoplast targeting signal, which was based on the N-terminal signal sequence of a murine antibody heavy chain (Voss et al., 1995; Vaquero et al., 1999). The DNA sequence encoding the last three amino acids of this signal was removed since the algorithm SignalP (Nielsen et al., 1997; http://www.cbs.dtu.dk/services/SiqnalP^') predicted that such action was nec¬ essary to achieve correct processing. The resulting APO^"3 sequence was amplified by PCR using the forward primer 35Sseq2 5'-ATC CTT CGC AAG ACC CTT CC-3' and the reverse primer Apo-back 5'-GCC ACC CGG CTG CGG CTC ACC TGC AGT GCC GCT G-3'. In parallel, the Solulin cDNA was amplified from plasmid pTHR525 using the primers Solulin-Apo 5'-CAG CGG CAC TGC AGG TGA GCC GCA GCC GGG TGG CAG CCA G-3' and SoIu- lin-back. The two PCR fragments were then assembled in a SOE-PCR using the primers 35Sseq2 and Solulin-back. The product of the SOE-PCR was ligated into pTRAkc using EcoRI and Xho\ restriction sites, resulting in the final construct Solulin-APO^"3.

To construct Solulin-VTS^"4, the Solulin cDNA was combined with the N-ter¬ minal vacuolar targeting sequence of Catharanthus roseus strictosidine syn¬ thase (VTS; McKnight et al., 1990). The DNA encoding the last four amino acids was omitted to ensure proper signal sequence cleavage, based on predictions by the SignalP algorithm (see above). This truncation was achieved by cutting the VTS sequence from the template vector using Asc\ and Sapl. The Solulin cDNA was amplified by PCR from the source plasmid pTHR525 using primers Solulin-VTS 5'-CTA GCT CTT CAA GCG GAG CCG CAG CCG GGT GGC AGC-3' and Solulin-back. The PCR product was digested with Xho\ and Sap\, mixed with the Asc\ and Sap\ restricted VTS^"4 sequence, and introduced into the pTRAkc vector (prepared by digestion with Asc\ and Xho\) in a triple ligation, resulting in the final construct Solulin- VTS^"4.

To enable targeting to protein bodies, Solulin was fused to the signal sequence of Pisum sativum legumin A2 (PbTS; Rerie et al., 1990). PbTS was amplified from a template vector with the primer 35Sseq2 and PbTS- back 5'-GCC ACC CGG CTG CGG TCG GCA AAG CAG CCT CCC-3'. In addition, the Solulin cDNA was amplified from plasmid pTHR525 using the primers SoIuNn-PbTS 5'-GGG AGG CTG CTT TGC CGA GCC GCA GCC GGG TGG-3' and Solulin-back. The two PCR fragments were then assem¬ bled by SOE-PCR using the primers 35S SEQ2 and Solulin-back. The prod¬ uct of the SOE-PCR was ligated into pTRAkc using EcoRI and Xho\, result¬ ing in the final construct Solulin-PbTS.

Transformation of tobacco plants and suspension cells

The plant expression vectors described above were introduced into A. tumefaciens using a Gene Pulser Il electroporation system (BioRad, Hercu¬ les, USA) according to the manufacturer's instructions. Transient expression in N. tabacum cv. Petit Havana SR1 or cv. Maryland Mammoth leaves was carried out by vacuum infiltration as previously described (Kapila et al., 1997; Vaquero et al., 1999). Transgenic N. tabacum cv. Maryland Mammoth plants were generated by the leaf disc transformation method and transgenic plants were regenerated from transformed callus (Horsch et al., 1985). Plants were grown in a greenhouse under a 16-h natural daylight photoperiod and a 25°C/day, and 22°C/night temperature regime.

Stably transformed N. tabacum cv. BY2 cells were generated by co-cultiva¬ tion with recombinant A. tumefaciens (An, 1985). BY2 cells were maintained in Murashige and Skoog (1962) medium supplemented with minimal organ- ics (BY2 medium: 0.47% (w/v) Murashige and Skoog salts, 0.15 μg ml^"1 thiamine, 0.02 μg ml^"1 KH₂PO₄ and 3% (w/v) sucrose, pH 5.2) in an orbital shaker (New Brunswick Scientific, Edison, USA) at 180 rpm and 26°C in the dark. Cultures were subcultered every week with a 5% (v/v) inoculum.

Protein extraction and analysis

The total soluble protein (TSP) from tobacco leaves was extracted as described by Fischer et al. (1999) using a 1 :2 (w/v) ratio of plant material and extraction buffer I (200 mM Tris-HCI, pH 7.4, 5 mM EDTA, 5 mM DTT, 0.1 % (v/v) Tween 20) or extraction buffer Il (150 mM NaCO₃, pH 9.5, 500 mM NaCI, 1 % (v/v) DMSO). Protein extraction from cell cultures was carried out 4-6 days after reseeding, when the packed cell volume was 25-40 %. Cells were centrifuged (500 x g, 5 min, room temperature) and the packed cells were sonicated (UW2070 with SH70G, MS-72 probe tip, 9x10%, 20%, 1 min; Bandelin, Berlin, Germany) in a 1 :2 (w/v) ratio of BY2 cells and extraction buffer I. The extracts were clarified by centrifugation prior to immunoblot analysis. Cell culture supematants were used without further treatment.

TSP concentrations were determined according to Bradford (1976) (Roti- Quant, Roth, Karlsruhe, Germany) using bovine serum albumin (BSA) as a standard. For immunoblot analysis, TSP extracted from leaves or cells was resolved by SDS-polyacrylamide gel electrophoresis and blotted onto PVDF membrane. Blotted Solulin was detected using the monoclonal anti- Solulin primary antibody 1043 (kindly provided by PAION GmbH, Aachen, Germany) diluted either 1 :1000 in PBS with 0.05% (v/v) Tween 20 (PBS-T) or 1 :1500 in PBS-T with 1% (w/v) skimmed milk powder. Binding of the primary antibody was detected using a goat-anti-mouse secondary antibody conjugated to alkaline phosphatase (Jackson lmmuno Research Laboratories, West Grove, USA) diluted 1 :5000 in PBS-T, or a goat-anti-mouse secondary antibody conjugated to horseradish peroxidase (Jackson lmmuno Research Laborato¬ ries) diluted 1 :8000 in PBS-T with 1 % (w/v) skimmed milk powder. For alka¬ line phosphatase, the signal was developed with nitroblue tetrazolium and 5- bromo-4-chloro-3-indolyl-phosphate solution (NBT/BCIP, Roth). For horse- radish peroxidase, the signal was developed using the ECL-advance kit (Amersham Biosciences, Freiburg, Germany). Solulin levels were quantified using an LAS-1000 luminescence image analyzer (Fuji, Duesseldorf, Ger¬ many) and Aida program version 2.31 (Raytest GmbH, Straubenhardt, Ger¬ many) with CHO-produced Solulin as the standard.

Protein purification and N-terminal protein sequencing

Recombinant soluble thrombomodulin was purified from suspension cell cultures by ammonium sulfate precipitation and anion exchange chromatog¬ raphy. The supernatant from a 300-ιml culture was retrieved by vacuum filtra¬ tion and was buffered with 8 mM Tris-HCI (final pH 8.5). MgSO₄ was added to a final concentration of 2 mM. To remove DNA, 500 units of Benzonase (Merck, Darmstadt, Germany) were added and the solution was incubated for 1.5 h at room temperature, followed by centrifugation (27138 x g, 20 min, 4°C). A 50% ammonium sulfate precipitation was performed, followed by a centrifugation as above to collect the precipitate. The pellet was redissolved in dialysis buffer (30 mM ammonium acetate, pH 6) and dialysed against the same buffer.

The desalted sample was processed by anion exchange chromatography using a 1-ml Sepharose Q column (Amersham Biosciences) equilibrated with dialysis buffer. The column was then washed with dialysis buffer containing 100 mM NaCI, and proteins were eluted with dialysis buffer containing 300 mM NaCI. The elution fraction was concentrated in centrifugal concentration devices (Pall Filtron, East Hills, USA) and separated by SDS-PAGE (Lammli 1970) prior to transfer onto PVDF membrane using CAPS-buffer (10 mM CAPS, pH 11 , 10% (v/v) methanol) in a semi-dry blotting apparatus (BioRad). The blot was stained with amido black, and the band correspond¬ ing to the recombinant thrombomodulin was cut out and used for N-terminal protein sequencing performed by TopLab (Martinsried, Germany). Thrombomodulin activity test

All measurements were done in a volume of 25 μl containing 5 μl sample and 20 μl assay buffer (20 mM Tris-HCI, pH 7.4, 100 mM NaCI, 2.5 mM CaCI₂, 0.15 mg ml^'1 BSA). The assays were performed in 96-well microtiter plates (Greiner, Solingen, Germany). Briefly, the 25 μl assay aliquots were incubated with 25 μl thrombin (3 nM; Haemochrom, Essen, Germany) and 25 μl protein C (0.5 μM; Merck) for 1 h at 37⁰C. Four units of hirudin (Roche, Mannheim, Germany) was then added, bringing the total volume to 100 ml, followed by incubation at 37⁰C for 15 min. Subsequently, 100 μl of the chro- mogenic substrate S-2366 (1.4 mM; Haemochrom) was added. The reaction was incubated at room temperature for two minutes and measured for 15 min at 405 nm using a kinetic protocol. Solulin standards (0, 2, 4, 6, 10 and 20 ng) derived from CHO cells were included with each test. The standards were kindly provided by PAION GmbH.

Results Table 1:

Solulin yields in transgenic tobacco plants. Recombinant Solulin levels in protein extracts from transgenic tobacco plants were determined by im- munoblot analysis using a Solulin -specific monoclonal antibody and an alka¬ line phosphatase-conjugated secondary antibody followed by NBT/BCIP staining (^a) or a horseradish peroxidase-conjugated secondary antibody fol¬ lowed by chemiluminescence based detection (^b). FW = fresh weight.

Construct Number of ana- Number of plants Maximum Solulin lysed plants producing Solulin ^a accumulation [μg g^"1

Solulin -APO^"3 40 33 115

Solulin -PbTS 65 55 38

Solulin -ORI 68 64 27 Table 2:

Solulin yields in transformed BY2 cell cultures. Recombinant Solulin levels in BY2 cell extracts and culture medium were determined by immunoblot analysis using a Solulin-specific monoclonal antibody and an alkaline phos- phatase-conjugated secondary antibody followed by NBT/BCIP staining (^a) or a horseradish peroxidase-conjugated secondary antibody followed by chemiluminescence based detection (^b). M = culture medium, C = cell fresh weight.

Construct Kanamycin- Established Solulin accumulation in elite resistant calli ^a cell cultures cell line [ng ml^"1 M or ng g^"1 C]

Solulin-APO^'3 23 M 940

C 6.570

Solulin-PbTS 27 M 500

C 27.000 Solulin-ORI 30

M 70

C 18.200

Table 3: Average Solulin activities in BY2 cell culture supernatants. TSP = total soluble protein.

Construct Active Solulin [ng TSP [μg ml^"1] ml^"1]

Solulin-APO^"3 1.300 40

Solulin-PbTS 2.100 75

Solulin-ORI 400 50 Figure 1 :

Schematic presentation of plant expression constructs. 35SS: CaMV 35S promoter with duplicated 35S enhancer region; CHS 5'UT: Chalcone syn¬ thase 5' untranslated region; SS: signal sequence; Solulin: coding sequence for Solulin; T35S: CaMV 35S terminator sequence.

Figure 2: lmmunoblot analysis of Solulin constructs transiently expressed in N. taba- cum cv. Petite Havana SR1. Total soluble proteins were separated by 12% SDS-PAGE and transferred onto a PVDF membrane. Solulin was detected using a monoclonal mouse anti- Solulin antibody as primary antibody and a goat-anti mouse secondary antibody conjugated to alkaline phosphatase, followed by NBT/BCIP staining. Size standards are indicated to the right. 1: Solulin-ORI, 2: Solulin-APO^'3, 3: Solulin-PbTS, 4: Solulin-VTS^"4, S: 200 ng purified Solulin standard from CHO cells.

Figure 3:

SDS-PAGE analysis of partially purified Solulin-PbTS. Solulin -PbTS was purified from 300 ml of BY2 suspension culture (4 d after seeding) by pre¬ cipitation and anion exchange chromatography. The elution fraction was concentrated 20-fold and 16 μl was separated by 10% SDS-PAGE. After electrophoresis, the gel was stained with Coomassie brilliant blue. Size stan¬ dards are indicated to the left. Lane 1 : 800 ng purified Solulin standard from CHO cells. Lane 2: partially purified Solulin-PbTS.

Figure 4:

Nucleic acid and protein sequence of Solulin. The entire coding region of the nucleic acid fragment used in the expression of Solulin is shown. The deducted amino acid sequence is shown below the nucleic acid sequence using the single letter amino acid code. The one letter code symbol is printed below the first nucleotide of the corresponding amino acid code. The one letter symbol is printed below the first nucleotide of the corresponding codon. The arrow indicates the first amino acid of the mature secreted protein.

Construction of the SoIuNn expression cassettes

To determine the most suitable plant cell compartment for Solulin accumula¬ tion, the cDNA of the mature Solulin was fused to different N-terminal tar¬ geting sequences to enable transport of the Solulin to the apoplast ( Solulin- APO^"3, Solulin-ORI), the vacuole ( Solulin-VTS^'4) or protein bodies ( Solulin- PbTS) (Figure 1). Importantly, the plant-derived Solulin should be identical in terms of amino acid sequence to the Solulin produced in CHO cells. To evaluate the possibility of improper signal cleavage, which would result in truncation or the presence of additional N-terminal residues, the nucleotide sequences of Solulin joined to the three non-innate signal sequences were analysed by the SignalP server, which predicts cleavage sites using a com¬ bination of several artificial neural networks (Nielsen et al., 1997). The analy¬ sis showed that the P. sativum legumin A2 signal sequence PbTS (Rerie et al., 1990) could be used without any changes. However, the last four internal amino acids of the vacuolar targeting signal from C. roseus strictosidine synthase (VTS^'4, McKnight et al., 1990), and the last three internal amino acids of the apoplastic targeting signal from the murine mAb24 heavy chain (APO^'3; Vaquero et al., 1999) had to be removed to guarantee accurate cleavage of these targeting signals.

Transient and stable expression of recombinant soluble thrombo¬ modulin in tobacco leaves and plants

All four Solulin constructs were transiently expressed in detached tobacco leaves of cultivars Petite Havana SR1 or Maryland Mammoth by agroinfiltra- tion. lmmunoblot analysis using a Solulin-specific monoclonal antibody dem¬ onstrated that recombinant Solulin was produced in transiently transformed tobacco leaves having a size of ~75 kDa corresponding to that of the CHO cell produced standard (Figure 2). However, some minor degradation of the protein was detectable. Three of the constructs gave rise to strong Solulin bands, but Solulin-VTS^"4 could not be detected (Figure 2, lane 4), suggesting that Solulin is not stable in the vacuole. Therefore, further experiments with the VTS^"4 construct were abandoned.

Solulin accumulation levels were similar whether the protein was targeted to the apoplast or protein bodies, but variations were seen between different infiltrations. However, higher yields of the protein were achieved using the Maryland Mammoth cultivar compared to Petite Havana SR1 (data not shown), so the former was chosen for stable transformation experiments.

The Solulin-APO^"3, Solulin-PbTS and Solulin-ORI constructs were introduced into tobacco leaf discs. Transgenic plants were recovered and tested for recombinant Solulin accumulation by immunoblot (Table 1). The maximum yields of Solulin differed according to which signal sequence was used with Solulin-APO^"3 showing the highest accumulation (115 μg g^"1 fresh weight), followed by Solulin-PbTS (38 μg g^'1 fresh weight) and Solulin-ORI (27 μg g^"1 fresh weight).

Stable expression of Solulin in BY2 cells

Transgenic BY2 calli were tested for the presence of recombinant Solulin by immunoblot and those with the highest levels of expression were used to establish cell suspension cultures. High levels of Solulin were detected in BY2 cell extracts (Table 2), although intact Solulin was also secreted and accumulated in the culture supernatant. Interestingly, the relatively high amount of Solulin-PbTS in the culture medium showed that the PbTS signal sequence mediated secretion of Solulin in BY2 cells rather than retention in the endomembrane system. The highest accumulation levels of Solulin were observed in BY2 cells producing Solulin-PbTS (27 μg g^"1 fresh weight), fol¬ lowed by Solulin-ORI (18.2 μg g^"1 fresh weight) and Solulin-APO^"3 (6.6 μg g^"1 fresh weight). Activity of plant cell produced Solulin

The activity of Solulin in BY2 cell extracts and culture medium was tested. No activity was detected in cell extracts, suggesting the presence of inhibiting compounds released upon cell disruption. This was confirmed in spiking experiments by the addition of functional, CHO-derived Solulin, whose activ¬ ity was also destroyed.

In contrast, strong and reproducible thrombomodulin activity was detected in BY2 culture supernatants. Indeed, Solulin levels calculated on the basis of activity were higher than those determined by immunoblot (see Tables 2 and 3) suggesting that only a proportion of the recombinant protein was trans¬ ferred to the membrane in immunoblot experiments. Therefore, we assume that the yields of recombinant proteins expressed in plants may be underes¬ timated when measured by immunoblot analysis alone.

The culture medium contains 50-100 times less total soluble protein than the total cell extract (40-75 μg ml^"1), resulting in a favourable recovery of active Solulin when expressed as percentage TSP (0.8-3.3%). This is a good start¬ ing point for the purification of Solulin, since the amount of contaminating proteins is low.

Purification and N-terminal protein sequencing

Solulin was purified from the BY2 culture medium. The crucial steps prior to anion exchange chromatography were DNAse treatment and ammonium sulfate precipitation with subsequent dialysis to remove DNA and other com¬ pounds interfering with the chromatography process. Although this procedure resulted in only partial purification of Solulin-APO^"3 and Solulin-PbTS (Figure 3), the yield and purity was sufficient to perform N-terminal protein sequencing. The results of these experiments revealed that the five N-terminal amino acids (Glu-Pro-Gln-Pro-Gly) were identical to those of the mature protein derived from CHO cells, demonstrating that the two targeting sequences are processed correctly in plant cells. Discussion of the results

It could be demonstrated that plant cells can produce intact and functional Solulin, with a yield of 27 μg g^"1 fresh weight in BY2 cells and 115 μg g^"1 in transgenic tobacco leaves (corresponding to 0.4% and 1.5% TSP, respec¬ tively). The levels obtained in transgenic plants were in a comparable order of magnitude as those obtained in CHO cells. However, the overall space time yield of Solulin in mammalian cell systems is much higher since produc¬ tion levels of up to several hundred μg per ml culture medium can be achieved and harvested daily. Nevertheless, the yield of recombinant soluble thrombomodulin in plants is high when compared to other thrombolytic or blood proteins. For example, recombinant hirudin was produced in trans¬ genic canola seeds (Brassica napus cv. Westar) as a fusion protein with oleosin. The expression was restricted to seeds and accumulation levels of approx. 1% of total soluble seed protein were reported (Parmenter et al., 1995). In transgenic tobacco, human protein C reached only 0.002% of TSP (Cramer et al., 1996).

To optimize Solulin accumulation, we compared several targeting signals. Although the vacuolar signal from C. roseus has been used successfully for the production of various recombinant proteins (unpublished results), Solulin- VTS^"4 could not be detected in transient transformation experiments. This suggests that Solulin is rapidly degraded in the plant cell vacuole, probably due to specific proteases only present in this compartment (Paris et al., 1996). The highest Solulin levels were obtained using the murine APO^"3 or plant PbTS signal sequences. Correct cleavage of the sequences was con¬ firmed by N-terminal sequencing, validating the in silico cleavage prediction carried out to optimize construct design. Both of these targeting signals have been used previously and successfully for recombinant protein production in plants (Vaquero et al., 1999). This supports the notion that well established heterologous targeting signals are preferable to endogenous signals that may already be present on the target protein. However, even though the original Solulin-ORI signal peptide did not perform as well as the APO^"3 and PbTS signals, it did nevertheless facilitate the accumulation of Solulin in plants and suspension cells (Tables 1 and 2).

The Solulin produced in transgenic plants and suspension cells had a molecular mass of ~75 kDa, which is the same size as the Solulin produced in CHO cells but 25 kDa larger than expected from the amino acid sequence. This deviation can be attributed to glycosylation. It is well known that N-gly- can structures slightly differ between mammalian and plant cells (Bardor et al., 1999). However, functional tests showed that thrombomodulin activity was not impaired by the presence of plant-derived N-glycans. Active Solulin was recovered from the BY2 culture supernatant but not from total BY2 cell extracts. Our suspicion that the missing activity in cell extracts reflected the presence of inhibitory compounds was confirmed when the extracts were spiked with active Solulin from CHO cells, and still lacked any sign of throm¬ bomodulin activity. The levels of recombinant protein secreted to the culture medium were low compared to amounts in cells, suggesting that Solulin is either inefficiently secreted due to its large size and size exclusion limit of the cell wall (Carpita et al., 1979; Tepfer and Taylor, 1981 ) or degraded in the culture medium. Both of these phenomena have been reported in previous studies involving the secretion of other recombinant proteins from plant sus¬ pension cells (Sharp and Doran, 2001 ; Titel and Ehwald, 1999). Neverthe¬ less, a two-step purification scheme involving salt precipitation and anion exchange chromatography was sufficient to isolate partially-purified intact Solulin from the culture supernatant.

We conclude that the production of intact and active Solulin is feasible both in tobacco plants and suspension cell cultures. Solulin yields can be increased further by selfing and backcrossing with elite plant lines (Hood et al., 2002) and by optimizing medium and culture conditions of BY2 cells (James et al., 2000). Thus, the production of Solulin in plants can become economically feasible, providing a new source to meet the high demands for this therapeutic protein.

The method according to the invention does not only provide a way how to produce Solulin but is also applicable for the production of native thrombo¬ modulin or other structural and functional derivatives of thrombomodulin or analogs thereof. It can be used for producing derivatives of thrombomodulin and solulin which are encoded by nucleic acid sequences which have at least an identity of 80 %, preferably 90% identity to the sequences of Solulin as shown in figure 4.

Examples of thrombomodulin analgos suitable for the method according to 'the invention are disclosed in US patents 5,256,770, 5,466,668, 5,827,824, 5,863,760 and 6,063,763, incorporated herein by references.

EXAMPLE 2:

PRODUCTION OF DESMOTEPLASE IN PLANT CELLS

DSPAαi is one of four plasminogen activators that have been isolated from the saliva of the common vampire bat Desmodus rotundus (Kratzschmar et al. 1991). Among these proteins, DSPAαi is the largest with a molecular weight of 50 kDa. DSPAαi is a serine protease which specifically cleaves serum plasminogen converting it into its active form, plasmin. Activated plasmin dissolves blood clots through the degradation of fibrin fibers. Its unique fibrin specificity (Schleuning et al. 1992) has led to the development of DSPAαi as a drug for the treatment of arterial thrombosis (Witt et al. 1994). Currently, recombinant DSPAαi is undergoing clinical trials for the treatment of acute ischemic stroke (Liberatore et al. 2003).

At the present time, recombinant DSPAαi is produced in transformed Chi¬ nese hamster ovary (CHO) cells (Gohlke et al. 1997; Petri et al. 1995). How¬ ever, future applications of this recombinant protein, including the treatment of stroke patients, may require a more efficient production system to provide sufficient amounts at low costs. Therefore, we explored the use of transgenic tobacco plants and suspension cells as alternative production platforms.

Transgenic plants are now becoming established as hosts for the production of valuable pharmaceutical proteins (Hood et al. 2002; Ma et al. 2003; Twy- man et al. 2003). Several plant species have been utilized, with maize and tobacco being the most common systems. Moreover, plant suspension cells have been used for protein expression under controlled conditions in biore- actors (Fischer et al. 1999; James and Lee 2001). Many molecular medi¬ cines have been synthesized in plants, including plasma proteins, enzymes, growth factors, vaccines and recombinant antibodies. These recombinant proteins can be expressed in transgenic plants on an agricultural scale, pro¬ viding a technology that can produce huge quantities of recombinant prod¬ ucts for use as diagnostics and therapeutics in modern health care and the life sciences (Gruber et al. 2001 ; Ma et al. 1998).

In the present study, we assessed the feasibility of producing recombinant DSPAαi in transgenic tobacco plants and suspension cell cultures. Since native DSPAαi is a secreted molecule, recombinant DSPAαi was targeted to the endomembrane system in transgenic plants and plant cells. The plant- derived DSPAαi was analyzed in terms of accumulation levels, integrity and functionality to establish whether plant systems could compete with the CHO cell production system. Our data showed that plants can produce functional DSPAαi . However, the plant system has to be optimized to achieve higher product levels and correct signal peptide cleavage. Finally, we provide evi¬ dence for the presence of endogenous proteases that are responsible for recombinant protein degradation in the culture medium of transgenic sus¬ pension cell cultures. This observation will help in the development of im¬ proved plant production systems for secreted proteins. Materials and Methods Plant expression vectors

The coding sequence of the mature DSPA protein was amplified from genomic DNA isolated from CHO cell line CD16.4 (provided by PAION GmbH, Aachen, Germany) using the following primers: DSPA-for (Nde\), 5^' GGT GTT CAC TCC GCA TAT GGT GTG GCC TGC AAA G 3^' and DSPA- back (stop), 5^' G CGA AGC TTC TCG AGT TTA CAG GTG CAT GTT GTC TCG AAT CC 3^'. This generated a 1340-bp product. The following plant ex¬ pression vectors, containing different signal peptide sequences for targeting the recombinant protein to the secretory pathway, were then constructed.

DSPA-APO:

LPH, the plant codon-optimized, 19-amino-acid leader peptide from the heavy chain of murine monoclonal antibody 24 (Vaquero et al. 1999), was amplified from the source vector pTRAkc-LPH (provided by Thomas Rade- macher, RWTH Aachen). The 240-bp product was fused to the 1340-bp DSPA sequence by splice overlap extension (SOE) PCR (Horton et al. 1990). The fragment was inserted at EcoRI and Xho\ sites in the plant expression vector pTRAkc, containing the T-DNA border sequences, the double enhanced CaMV 35S promoter, the 5^' untranslated region (UTR) of the chalcone synthetase (CHS) gene from Petroselinum crispum, the 35S polyadenylation signal, scaffold attachment regions and the nptW gene as a selectable marker, resulting in the final construct DSPA-APO.

DSPA-PbTS:

The PbTS leader peptide sequence (22 amino acids) is derived from legu- minA2 of Pisum sativum (GenBank^® accession X17193) and targets native leguminA2 to protein bodies in pea seeds (Hinz et al. 1995). The PbTS cod¬ ing sequence was assembled by two consecutive rounds of the PCR, using pTRAkc-LPH as a template. The forward primer pSS-LPH forw (5^' ACC ACG TCT TCA AAG CAA GTG G 3') was used for both rounds of amplification. The primers PbTS-back 1 (5^' GCA GAA AGA AAG AGA AAG TGC AAG GAG TTT AGT AGC CAT TGT TTT TTT TTT TTT TAT A 3^') and PbTS-back 2 (5^' GGC CAC ACC ATA TGC GGC AAA GCA GCC TCC CAA AAG GAG AAA GCA GAA AGA AAG AGA AAG 3^') were used successively. The final product was cleaved with EcoRI and Nde\ and ligated with the Nde\IXho\ fragment derived from DSPA-APO into the EcoRI- and X/?ol-digested vector backbone of pTRAkc, resulting in the final vector DSPA-PbTS.

DSPA-VTS ⁴:

The VTS^"4 leader sequence is derived from the strictosidine synthase gene of Catharanthus roseus (GenBank^® accession X61932) and comprises 28 amino acids. The C-terminal four serine residues from the native sequence were omitted since they would lead to incorrect cleavage as predicted by the CBS SignalP prediction server (http://www.cbs.dtu.dk/services/SignalP-2.0/) (Nielsen et al. 1997). The native enzyme is localized in the vacuoles of leaf tissue in C. roseus as well as in transgenic tobacco plants (McKnight et al. 1991). The leader sequence was assembled through three consecutive rounds of PCR using the forward primer pSS-LPH forw in all reactions in combination with one of the backward primers: VTS-back 1 , 5^' CAT CAT GGA CTT AGA TTC AGA GAA GTT TGC CAT TGT TTT TTT TTT TTT TAT A 3^'; VTS-back 2, 5^' AAG GAG AAG AAG GAA AAA CAT GAA GAA AAC TGC CAT CAT GGA CTT AG ATT C 3^'; or VTS-back 3, 5^' GGC CAC ACC ATA TGC GCT TGA AGA GCT AGA TGA AAG GAG AAG AAG GAA AAA 3^'. The final expression construct, DSPA-VTS^"4, was generated by assem¬ bling the amplified VTS^"4 and DSPA sequence with the pTRAkc backbone in a triple ligation as described above.

DSPA-ORI:

The fourth construct contained the native DSPAαi sequence including the original N-terminal signal peptide of 21 amino acids and a prosequence of 15 amino acids. The coding sequence was amplified from genomic DNA derived from CHO cell line CD16.4 in two consecutive PCRs. The two forward prim¬ ers were designed to introduce the CHS 5^'UTR: DSPAorM-for, 5^' GAT TTA TAG AGA GAT TTA TAA AAA AAA AAA AAA CAA TGG TGA ATA CAA TGA AGA 3^', and DSPAori2-for, 5^' GGA ATT CAC AAC ACA AAT CAG ATT TAT AGA GAG ATT TAT 3^'. These were used together with the DSPA(Λ/col)-back primer. The final product of 788 bp was digested with EcoRI and Λ/col and replaced with the corresponding fragment of DSPA- APO in pTRAkc, resulting in the final construct DSPA-ORI.

DSPA-coAPO, DSPA-coVTS^'4, DSPA-coPbTS, DSPA-coORI: To insert a C-terminal Hisβ tag, a Not\ restriction site was placed at the 3' end of the DSPAaI cDNA by PCR using the primer pair DSPA-for (Nde\) and DSPA-back (Λfofl) (5'-CGC GTA CTC GAG AGC GGC CGC CAG GTG CAT GTT GTC TCG-3'). Genomic DNA from CHO cells containing the DSPAaI cDNA was used as the template. To obtain the final construct DSPA-coAPO, the Λ/col- and Λ/ofl-digested PCR product, the EcoR\INot\ pTRAkc-LPH vec¬ tor backbone and the EcoRI/Λ/col fragment of the vector DSPA-APO were assembled in a triple ligation. The remaining constructs DSPA-coVTS^"4, DSPA-coPbTS and DSPA-coORI were generated by cloning the corre¬ sponding targeting sequences plus the 5' end of DSPAaI via EcoRI and Nco\ into the vector backbone of the EcoRI/Λ/col-digested DSPA-coAPO construct.

Plant transformation

All eight plant expression vectors were used to transform Agrobacterium tumefaciens GV3101 by electroporation (Dower et al. 1988). Transformed bacteria were used for transient expression studies in detached leaves of Nicotiana tabacum cv Petite Havana SR1 , cv Maryland Mammoth (obtained from IPK₁ Gatersleben, Germany) and cv NFT51 (obtained from the Bundes- anstalt fuer Zuechtungsforschung, Braunschweig, Germany) through vacuum infiltration (Kapila et al. 1997) and for stable transformation of N. tabacum cv Maryland Mammoth leaf disks (Horsch et al. 1985) or cv Bright Yellow (BY-2) cells (An 1985). Plants were grown in a greenhouse under a 16-h photoperiod. The temperature was held at 25°C during the day and 22⁰C at night.

BY-2 cells were maintained in Murashige and Skoog (1962) basal salt with minimal organics (BY-2 medium: 0.47 % (w/v) Murashige and Skoog salts, 0.15 μg/ml thiamine, 0.02 μg/ml KH₂PO₄ and 3 % (w/v) sucrose, pH 5.8) on an orbital shaker (New Brunswick Scientific, Edison, USA) at 180 rpm and 26°C in the dark. Cells were passed into fresh medium every week with a 5 % (v/v) inoculum.

Northern blot analysis

Total RNA was prepared from 200 mg of infiltrated leaves using the RNeasy plant mini kit (Qiagen, Hilden, Germany). Total RNA (10 μg) was loaded onto denaturing formaldehyde agarose gels and then capillary blotted onto nylon membranes (Hybond N⁺, Amersham Biosciences, Freiburg, Germany) using 10 χ SSC. The membrane was probed with a 770-bp Nco\/Xba\ DSPAαi fragment radiolabeled with [α³²]P-dATP (Amersham Biosciences) using the DecaLabel DNA labeling kit (Fermentas GmbH, St. Leon-Rot, Germany) according to the manufacturer's instructions. The membrane was hybridized with Church solution (1 % (w/v) BSA, 1 mM EDTA, 7 % (w/v) SDS, 0.5 M NaH₂PO₄, pH 7.0) for two hours at 65°C. After prehybridization the labeled probe was added and incubated over night at 65°C. The membrane was washed twice using 2 x SSC, 0.1 % (w/v) SDS at 65⁰C and once with 0.2 x SSC, 0.1 % (w/v) SDS at 65°C for 30 minutes each. The membrane was exposed to an X-ray film for six hours using an intensifier screen.

lmmunoblot analysis

To extract total soluble proteins from BY-2 cells, the cells were pelleted by brief centrifugation and two volumes (v/w) of BY-2 extraction buffer (50 mM Tris/HCI, 100 mM NaCI, 5 mM EDTA, 0.1 % (v/v) Tween 20, 10 mM DTT, pH 8.0) were added to the packed cell mass. This was followed by sonication (UW2070 with SH70G, MS-72 probe tip, 9 x 10%, 20%, 1 min; Bandelin, Berlin, Germany). The total soluble protein from leaf material was obtained by extraction in buffer I (200 mM Tris-HCI, 5 mM EDTA, 5 mM DTT, 0.1 % (v/v) Tween 20, pH 7.4) or buffer Il (150 mM citrate buffer, pH 4.8). Protein extracts were separated on a 12 % SDS-polyacrylamide gel using the Mini- protean system (Biorad, Hercules, CA, USA). The gels were run with a con¬ stant current of 20 mA per gel. Subsequently, gels were blotted onto nitro¬ cellulose membranes under semi-dry conditions using the standard buffer (192 mM glycine, 25 mM Tris, 20 % (v/v) methanol) and the TRANS-BLOT SD device (Biorad) at a constant current of 125 mA per gel for 40 minutes. Recombinant DSPAαi was detected using a polyclonal rabbit anti-DSPAα1 antibody (kindly provided by PAION GmbH, Aachen, Germany) at a dilution of 1 :1000 in PBS-T (phosphate-buffered saline supplemented with 0.05% (v/v) Tween 20). The secondary antibody was either an alkaline phos¬ phatase- or peroxidase-labeled goat anti-rabbit antibody (Dianova GmbH, Hamburg, Germany). Blots were developed with NBT/BCIP or the ECL™ Western Blotting Analysis System (Amersham Biosciences) according to the manufacturer's instructions.

The quantity of recombinant DSPAαi was calculated using the LAS-1000 luminescence image analyzer (Fuji, Duesseldorf, Germany) and Aida pro¬ gram version 2.31 (Raytest GmbH, Straubenhardt, Germany) using CHO- derived DSPAαi as a standard.

Protein purification and N-terminal protein sequencing

For analytical protein purification, total soluble proteins were extracted from N. tabacum cv Petite Havana SR1 leaves that had been transiently trans¬ formed with DSPAαi constructs containing the C-terminal His₆-tag. Approxi¬ mately 20 g of infiltrated leaves were ground under liquid nitrogen using a mortar and pestle followed by the addition of two volumes of plant extraction buffer (5O mM NaH₂PO₄, 30O mM NaCI, 1O mM 2-mercaptoethanol, 1 % (w/v) polyvinylpolypyrrolidone (PVPP), pH 8.0). The pH was re-adjusted to pH 8.0 and the extract incubated for one hour at 4°C with gentle agitation. The samples were centrifuged (27,000 x g, 30 min, 4⁰C) to remove insoluble particles. Then 300 μl of equilibrated Ni-NTA matrix (Qiagen, Hilden, Ger¬ many) were added and incubated at 4⁰C on a head-over-tail rotator. The matrix was sedimented (500 x g, 5 min, 4°C) and washed once with ten bed volumes of plant extraction buffer without 2-mercaptoethanol and PVPP, and once with washing buffer (50 mM NaH₂PO₄, 300 mM NaCI, 5 mM imidazole, pH 8.0). Bound proteins were eluted with plant extraction buffer supple¬ mented with 1 M imidazole.

Samples for N-termihal protein sequencing were separated on a 10% poly- acrylamide gel without SDS. The cathode buffer was supplemented with Na₂S₂Os and gels were blotted after electrophoresis onto polyvinylidene fluo¬ ride (PVDF) membranes (Immobilon-P, Millipore GmbH, Schwalbach, Ger¬ many) using CAPS buffer (10 mM CAPS, 10 % (v/v) methanol, pH 11.0). The PVDF membrane was stained with Amido black and the recombinant DSPAαi band was excised and dried at room temperature. Protein samples were sequenced by Edman degradation using the protein sequencer 476A (Applied Biosystems, Foster City, CA, USA) at Prosequenz Bioanalytik (Ludwigsburg, Germany).

DSPA activity assay

The activity assay was carried out according to a standard operating proce¬ dure developed by Berlex Biosciences (Richmond, CA, USA). The assay is based on the liberation of a chromogenic moiety (p-nitroaniline, pNA) from the small peptide S2288™ (Ile-Pro-Arg-pNA, Haemochrom Diagnostica, Essen, Germany). For each measurement, a standard curve was included with the following concentrations of CHO-derived DSPAαi (kindly provided by PAION GmbH, Aachen, Germany): 20 μg/ml; 10 μg/ml; 5 μg/ml; 2.5 μg/ml; 1.25 μg/ml; 0.625 μg/ml and 0.312 μg/ml in assay buffer (50 mM Na₂CO₃, 0.1 % (v/v) Tween 80, pH 9.5). A 10 mM stock solution of S2288™ was diluted 1 :5 with the assay buffer immediately before use. Plant-derived and purified recombinant DSPAαi containing a C-terminal His₆-tag was desalted by gel filtration using a PD 10 column (Amersham Biosciences). The samples were diluted in assay buffer to a final volume of 100 μl and the reaction was started by the addition of 100 μl of the substrate solution in 96- well microtiter plates (Greiner Bio-One, Frickenhausen, Germany). The reac¬ tions were incubated for two minutes at room temperature and DSPAαi activity was measured at OD₄os for 5 min at 9-s intervals.

Zymography

As a substrate for proteolysis, gelatin (Merck, Darmstadt, Germany) was co- polymerized with a 12.5% SDS polyacrylamide gel at a final concentration of 0.1 % (w/v). Gels were run at 20 mA and, after electrophoresis, washed once with 2.5 % (v/v) Triton^® X-100 for 20 min at room temperature and twice in protease buffer (5O mM Tris, 5 mM CaCI₂, 100 μM ZnCI₂, pH 7.6) for 15 minutes each. The gels were incubated overnight in protease buffer at room temperature with gentle agitation. Subsequently, gels were stained with Coomassie Brillant Blue followed by destaining to reveal areas of proteolytic gelatin degradation.

Results Figure 5: lmmunoblot analysis of recombinant DSPAαi transiently expressed in N. tabacum cv Petite Havana SR1.

Recombinant DSPAαi was detected using a primary polyclonal rabbit anti- DSPAαi antiserum and a secondary goat anti-rabbit IgG conjugated to alka¬ line phosphatase; M = pre-stained broad range marker (NEB). In total, 20 μl of total soluble protein from transiently transformed tobacco leaves and 250 ng standard DSPAaI purified from CHO cells were loaded on a 12% SDS polyacrylamide gel.

Figure 6: lmmunoblot and northern blot analysis of transiently transformed tobacco leaves using different tobacco cultivars.

A) In total, 15 μl of total soluble proteins (TSP) extracted from tobacco leaves were loaded onto each lane Of a 12% SDS polyacrylamide gel. The leaves were vacuum infiltrated with DSPA-coPbTS (lanes 1 and 2, N. tabacum cv Maryland Mammoth; lanes 3 and 4, N. tabacum cv NFT 51 ; lanes 5 and 6, N. tabacum cv Petite Havana SR1 ). Non-infiltrated leaf controls are shown in lanes 7 and 8 (7, N. tabacum cv Maryland Mammoth; 8, N. tabacum cv NFT 51 ). Lane S was loaded with 250 ng purified DSPA from CHO cells. M = pre- stained broad range marker (NEB).

B) Total RNA (10 μg per lane) was extracted from leaves vacuum infiltrated with DSPA-coPbTS (1 , N. tabacum cv Petite Havana SR1 ; 2, N. tabacum cv Maryland Mammoth; 3, N. tabacum cv NFT 51) or from a non-infiltrated leaves (4, N. tabacum cv Petite Havana SR1 ; 5, N. tabacum cv Maryland Mammoth; 6, N. tabacum cv NFT 51).

i) Hybridization with DSPAαi probe; ii) Ethidium bromide stain of the nylon membrane. The arrow indicates the full-length DSPAαi transcript.

Figure 7:

Determination of recombinant DSPAαi enzyme activity. Standard DSPAαi purified from CHO cells was used in the following con¬ centrations as references: 20 μg/ml; 10 μg/ml; 5 μg/ml; 2.5 μg/ml; 1.25 μg/ml; 0.675 μg/ml and 0.312 μg/ml. The plant-derived and purified DSPA-coAPO sample was diluted 1 :5 and 1 :10 in a final volume of 100 μl.

Figure 8: lmmunoblot analysis of recombinant DSPAαi in BY-2 cells and cell culture medium. M: Pre-stained broad range marker (NEB); 1 , DSPA-APO extracted from BY- 2 cells; 2, DSPA-APO from cell culture medium; arrow indicates the full-size DSPAαi product.

Figure 9:

Analysis of DSPAaI stability in BY-2 culture medium. Culture medium from a BY-2 wild type suspension culture was spiked with 6 ng/μl DSPAaI purified from CHO cells. The samples were incubated either without proteinase inhibitors, with 10 x concentrated Complete™ (Roche) or with 5 mM EDTA. In total, 15 μl per sample were analyzed by immunoblot after 3 h or 17 h of incubation, respectively. 1 , 3 h without additives; 2, 3 h with 10 x Complete™; 3, 3 h with 5 mM EDTA; 4, 17 h without additives; 5, 17 h with 10 x Complete™; 6, 17 h with 5 mM EDTA. M = pre-stained broad range marker (NEB); arrow indicates the full-size DSPAαi product.

Figure 10:

Stabilization of standard DSPA in conditioned BY-2 culture medium by protease inhibitors

Culture medium from a BY-2 wild type suspension culture was spiked with 6ng/μl DSPA purified from CHO cells. The samples were incubated over a period of 24 h at room temperature either in the presence or absence of Complete™ (Roche) proteinase inhibitor mix at different concentrations. Lane M = pre-stained broad range marker (NEB). 1 , 10 x concentrated Complete™; 2, 5 x concentrated Complete™; 3, 2 χ concentrated Com¬ plete™; 4, 1 x concentrated Complete™; 5, no proteinase inhibitor. The arrow indicates the full-size recombinant DSPAαi .

Figure 11:

In-gel protease assay of conditioned BY-2 culture medium. BY-2 culture medium (15 μl of four-day-old cultures; wild type, DSPA-coAPO, DSPA-PbTS or DSPA-VTS^"4) was separated on a 10% SDS polyacrylamide gel (without 2-mercaptoethanol) containing 0.1% (w/v) gelatin. After electro¬ phoresis, the gel was washed twice in 2.5% (v/v) Triton^® X-100 and twice in protease buffer (50 mM Tris, 5 mM CaCI₂, 100 μM ZnCI₂, pH 7.6). Protease digestion was carried out for 17 h in protease buffer before staining the gel with Coomassie Brilliant Blue. Clear bands on the dark background indicate protease activity.

Figure 12:

Nucleic acid and protein sequence of DSPA α1. The arrow indicates the first amino acid of the mature secreted protein.

Table 4:

N-terminal sequencing of plant-derived recombinant DSPAαi . Recombinant DSPAαi proteins were purified via IMAC from transiently transformed tobacco leaves. The purified proteins were separated by SDS- PAGE and blotted onto PVDF membranes. The highest molecular weight band was cut out and used for Edman sequencing.

Expected AYGVACKDEIT sequence

DSPA-coAPO AYGVACKDEIT ratio 0.7 : 1

VACKDEIT

DSPA-coVTS ⁴ YGVACKDEIT

DSPA-coPbTS blocked Table 5:

Selection of transgenic tobacco lines and analysis of recombinant DSPAαi accumulation.

Recombinant DSPAαi accumulation levels in extracts from transgenic tobacco plants were determined by immunoblot analysis using the DSPAaI- specific polyclonal antiserum and an alkaline phosphatase-conjugated sec¬ ondary antibody followed by NBT/BCIP staining (^a) or a horseradish peroxi¬ dase conjugated secondary antibody, chemiluminescence substrate and three different amounts of DSPA standard (0.5, 1 , 1.5 ng). Levels were determined using a phosphorimager (^b). Only the signal representing the full- size DSPAαi protein was used to determine the protein accumulation level. FW = fresh weight.

Con¬ Number of Number of Maximum struct analyzed plants producing DSPAαi accu¬ plants DSPAcrf ^a mulation [μg/g

FW| ^b

DSPA- 167 99 34

APO

DSPA- 136 95 38 VTS^"4

DSPA- 108 69 27 PbTS

Transient expression of recombinant DSPAαi in tobacco leaves

Detached leaves of N. tabacum cv Petite Havana SR1 plants were tran¬ siently transformed by vacuum infiltration with each of the eight DSPAαi constructs that are described under Materials and Methods. In each case, recombinant DSPAαi was detected by immunoblot (Figure 5) and with the expected size (about 49 kDa). In addition to these full-size bands, and irre¬ spective of the construct used, degradation bands could be detected indi¬ cating that degradation is independent of the signal peptide used for recom¬ binant protein targeting. Transient expression experiments were repeated several times and showed no major differences between the constructs.

In addition to the cultivar Petite Havana SR1 , two other cultivars - Maryland Mammoth and nicotine free tobacco (NFT) 51 - were tested for accumulation of recombinant DSPAαi . The three different tobacco cultivars were com¬ pared using the construct DSPA-coPbTS. Immunoblot analysis of transiently transformed leaves revealed that the different tobacco cultivars showed dif¬ ferent accumulation levels of the recombinant protein, but significant differ¬ ences in degradation patterns were not observed. The highest level of recombinant protein accumulation was achieved with cultivar Maryland Mammoth followed by Petite Havana SR1 and NFT 51 (Figure 6A). Four independent experiments were performed using different DSPAαi constructs leading to the same result.

Northern blot analysis was carried out to investigate whether or not the increased recombinant protein accumulation in Maryland Mammoth is based on increased steady state transcript levels. These experiments confirmed a correlation between transcript and recombinant protein levels (Figure 6B). The highest transcript levels were observed in vacuum infiltrated Maryland Mammoth leaves followed by the cultivars Petite Havana SR1 and NFT 51. Moreover, northern blot analysis proved that degradation of recombinant DSPAαi was not based on aberrant transcripts. Only high molecular weight RNA could be detected, indicating the presence of readthrough transcripts. Based on the outcome of these investigations we chose to use Maryland Mammoth for the generation of stably transformed tobacco plants. Enzymatic activity of transiently expressed recombinant DSPAαi

Spiking experiments revealed that DSPAαi enzymatic activity is inhibited by the tobacco plant extract. Therefore, His₆-tagged DSPAαi constructs were transiently expressed in tobacco leaves and recombinant proteins were puri¬ fied by immobilized metal ion affinity chromatography (IMAC) avoiding the time-consuming establishment of conventional purification protocols for the non-tagged proteins. IMAC purification for DSPA-coAPO, DSPA-coPbTS and DSPA-coVTS^"4 achieved sufficient purity and product quantity for subsequent analysis. Purification of DSPA-coORI was possible but the purity and yield was low (data not shown).

All four Hisβ-tagged DSPA proteins were tested for enzyme activity using the chromogenic substrate S2288™. In each case, enzyme activity was clearly detected. Figure 7 shows as an example the enzymatic activity of DSPA- coAPO. According to this activity, we calculated an accumulation level of 5.5 μg DSPA-coAPO per gram of transiently transformed tobacco leaf tissue. The levels for the remaining proteins - DSPA-coVTS^'4, DSPA-coPbTS and DSPA-coORI - were 3.1 μg, 5.6 μg and 0.4 μg per gram of leaf material, respectively. This agreed with the levels determined by immunoblot (data not shown) indicating that the purified proteins are predominantly functional. The low value for DSPAcoORI is at least partially due to a less efficient Ni-NTA purification. However, in general, the accumulation of DSPAcoORI and DSPA-ORI in plant tissue were low. Therefore, both constructs were excluded from further characterization.

Signal peptide processing

To investigate the processing of the N-terminal signal peptide, transiently expressed and purified His₆-tagged proteins DSPA-coAPO, DSPA-coVTS^'4 and DSPA-coPbTS were analyzed by N-terminal protein sequencing. The plant-derived DSPA-coAPO were present in two different forms (Table 4). One form started with an alanine residue as expected, but the second started with a valine, missing the first three amino acids of the mature DSPAαi protein. The ratio of the two forms was 0.7: 1. DSPA-coVTS^"4 was processed uniformly but the initial alanine residue was missing. The mature recombinant protein starts with a tyrosine residue. Sequencing of DSPA- coPbTS failed, probably due to a blocked N-terminus.

Recombinant DSPAαi production in transgenic tobacco plants N. tabacum cv Maryland Mammoth was transformed with DSPA-APO, DSPA-VTS^"4 and DSPA-PbTS, and kanamycin resistant T₀ plants were ana¬ lyzed by immunoblotting. The highest accumulation level of intact recombi¬ nant protein was detected in the case of DSPA-VTS^"4 (38 μg/g leaf material) followed by DSPA-APO (34 μg/g) and DSPA-PbTS (27 μg/g) (Table 5). Elite plant lines showed very similar accumulation levels, and the degradation pattern was the same as that observed for the transient expression system. The differences in accumulation levels between the three proteins were not regarded as significant in terms of the potential influence of the different sig¬ nal sequences.

Recombinant DSPAαi production in transgenic BY-2 cells

The rational for DSPAαi expression in tobacco BY-2 cells is the facilitated purification of the recombinant protein when targeted for secretion into the culture medium (Sijmons et al. 1990). DSPA-APO, DSPA-VTS^"4 and DSPA- PbTS constructs were used for stable transformation of tobacco BY-2 cells, and approximately 35 kanamycin resistant calli were tested for DSPAαi accumulation by immunoblotting. Transgenic BY-2 calli showing the highest level of recombinant protein accumulation were selected and used to estab¬ lish cell cultures. These were tested for recombinant DSPAαi accumulation by activity assays and immunoblotting. Transgenic BY-2 suspension lines producing DSPA-APO and DSPA-PbTS proteins showed the highest accu¬ mulation levels (1.5 μg/g wet weight) whereas DSPA-VTS^"4 accumulated to lower levels (0.3 μg/g). These results show that intact recombinant DSPAαi was 18 to 127 times less abundant in BY-2 cells than in transgenic plants. Moreover, distinct degradation bands were visible, similar to those observed in transient expression experiments in leaves and in stably transformed plants (Figure 8, lane 1). In contrast to leaves, DSPAαi activity was not inhibited in BY-2 cell extracts. Therefore, results obtained from immunoblot analysis and activity assays could be compared and demonstrated that almost all of the produced recombinant protein was functional (data not shown). However, intact DSPAαi was not detected in the culture medium, only degraded proteins with MWs smaller than 30 kDa (Figure 8, lane 2). Since degradation products could be detected in the transgenic BY-2 cul¬ tures producing DSPA-APO, DSPA-PbTS and DSPA-VTS^'4 we concluded that all three signal peptides enabled recombinant protein targeting to the plant cell apoplast and culture medium.

Characterization of recombinant DSPAcrt degradation

The lack of intact recombinant DSPAαi in the BY-2 culture medium might be due to rapid degradation of the full-size molecule by extracellular plant pro¬ teases or the retention of the full-size molecule by the plant cell wall. Spiking experiments using DSPAαi from CHO cells revealed that instability was not caused by components in the Murashige and Skoog plant cell culture medium. The DSPAαi standard was stable in fresh culture medium for at least 20 hours without any degradation (data not shown). In contrast, the DSPAαi standard was rapidly degraded in medium taken from a wild type BY-2 cell suspension culture (Figure 9, lanes 1 and 4). The rapid proteolysis of DSPAαi could be reduced significantly by the use of a protease inhibitor blend for the simultaneous inhibition of proteases from different classes. A 10 x concentrated Complete™ proteinase inhibitor mix (Roche) was the most effective in reducing the degradation of DSPAαi (Figure 9, lanes 2 and 5; Figure 10). It was therefore assumed that proteases in the cell culture super¬ natant were responsible for the observed degradation of DSPAαi . This pro- teolysis-inhibiting effect could also be achieved by the addition of EDTA at a concentration of 5 mM (Figure 9, lanes 3 and 6). In contrast, polyvinylpyr¬ rolidone (PVP) 360 and gelatin, which have also been shown to stabilize recombinant proteins in plant cell culture media (LaCount et al. 1997; Mag- nuson et al. 1996) did not have any stabilizing effect on secreted DSPAαi . This finding indicated that metalloproteases are involved in the DSPAαi deg¬ radation process since these proteases require the presence of divalent cations (Yang et al. 1996). To confirm the presence of proteases in the BY-2 medium, culture supernatants from different cell lines were analyzed by zy- mography. As shown in Figure 11 , several bands with gelatin-degrading ac¬ tivity were detected by this assay. Besides three major bands several addi¬ tional minor bands were also observed. In total, about a dozen proteases were detected with this zymogram assay. Whether each band belongs to a different protease remains to be determined because it has been shown that certain proteases adhere to gelatin during electrophoresis (Hummel et al. 1996). Nevertheless the presence of at least one endogenous protease was clearly demonstrated.

Discussion of the results

Our studies have demonstrated that the production of functional DSPAαi is feasible in transgenic tobacco plants and suspension cells. Product accu¬ mulation was significantly higher in transgenic plants resulting in a maximum yield of 38 μg/g leaf material corresponding to 0.4 % of total soluble protein (TSP). To our knowledge, human protein C (hPC) is the only other recombi¬ nant proteolytic enzyme involved in blood hemostasis or thrombolysis that has been produced in transgenic plants thus far and the maximum level achieved for this protein was 0.002 % of TSP (Cramer et al. 1996). Although recombinant DSPAαi yields were 200-fold higher, the plant expression sys¬ tem we described currently can not compete with the levels achieved in transgenic CHO cells, which reach 60 μg ml^'1 d^"1 (Petri et al. 1995). However, only T₀ plants have been analyzed so far and there is a high probability that recombinant DSPAαi yields could be increased in subsequent generations. For example, recombinant protein accumulation has been enhanced 4-fold in Ti plants by selfing elite To lines (Zimmermann et al. 1998) and as much as 150-fold in eight generations using selection and a backcross program with elite germplasm (Hood et al. 2002). Another proteolytic enzyme, trypsin, has been produced on a commercial scale in transgenic maize kermels (Woodard et al., 2003). Yields of 58 mg kg^"1 have been achieved in the fifth generation.

Plant-derived DSPAαi is enzymatically active but inhibited by components present in total soluble protein extracts from intact plants. No major influence of the targeting signal on enzyme activity or accumulation levels were observed with the exception of the native signal sequence (ORI) from Des- modus rotundus wh\ch reduced yields approximately ten-fold. The N-terminal ORI signal peptide consists of a 21-amino-acid signal sequence and a 15- amino-acid propeptide (Kratzschmar et al. 1991 ). It is possible that incorrect processing of the ORI signal resulted in reduced secretion efficiency and low product yield. However, N-terminal signal peptide processing was not verified since DSPA-ORI levels were too low to provide sufficient material for N-ter¬ minal protein sequencing. Only DSPAαi, containing the N-terminal APO leader, showed correct signal peptide cleavage. Nevertheless, 60% of this plant-derived protein lacked the three N-terminal amino acids (Table 4). Although cleavage of the VTS sequence has been analyzed in silico and optimized by omitting the four C-terminal serine residues (VTS^"4), plant- derived mature DSPAαi lacked the N-terminal alanine residue. Whether cleavage of the additional amino acids was due to incorrect signal peptide processing or subsequent protein degradation in vivo or during extraction was not investigated.

Recombinant DSPAαi produced in tobacco plants and suspension cells was partially degraded irrespective of the signal peptide used to target the protein to the endomembrane system. Our results indicated that these degradation products are due neither to the presence of aberrant transcripts nor to auto- proteolysis of DSPAαi , through its own serine protease activity (Kratzschmar et al. 1991). The latter was supported by spiking experiments demonstrating that DSPAαi derived from CHO cells was stable in plant extracts and freshly prepared plant cell culture medium. Since the degradation pattern of the recombinant DSPAαi is similar to that of all tested constructs (Figure 5) we assume that proteolysis took place in the same plant cell compartment. This is supported by the observation that all three targeting sequences APO, PbTS and VTS^"4 led to secretion of recombinant DSPAαi fragments into the culture medium of transgenic BY-2 suspension cells (data not shown).

Proteolytic degradation was most severe in conditioned BY-2 cell culture medium. Several groups have reported the instability of recombinant proteins secreted into the plant cell culture medium (Bateman et al. 1997; LaCount et al. 1997; Lee et al. 2002; Tsoi and Doran 2002). In some cases, recombinant proteins were stabilized by addition of polymers such as PVP or gelatin leading to increased levels of intact product (LaCount et al. 1997; Lee et al. 2002). However, these polymers did not stabilize recombinant DSPAαi in the plant cell culture medium. In contrast, full-size DSPAαi was detectable after the addition of a protease inhibitor mix or 5 mM EDTA indicating that proteases present in the culture medium are responsible for DSPAαi degra¬ dation. Indeed we showed the presence of several active proteases in condi¬ tioned culture medium (see Figure 11 ). Our ability to inhibit this protease activity through the addition of EDTA indicates that metalloproteases are responsible for the degradation of DSPAαi . In contrast to the aspartate, serine, threonine and cysteine proteases, metalloproteases require a divalent metal ion cofactor for their activity (Mayne and Robinson 1996). The most common cofactor is Zn²⁺ but other divalent cations including cobalt or nickel have also been described as cofactors. Therefore, metalloproteases can be inhibited effectively by chelating agents such as EDTA or 1 ,10-phe- nanthroline (Belozersky et al. 1990).

Whether or not these proteases are actively secreted remains to be deter¬ mined since we cannot exclude passive secretion (i.e. through cell lysis) or localization within the plasma membrane of the plant cell as alternative explanations. Metalloproteases are poorly characterized in the plant king¬ dom. There are only few detailed reports describing plant metalloproteases. For example, the metalloprotease GmMMP2 from soybean is localized in the plasma membrane and is most likely a component of the defense machinery against pathogens (Liu et al. 2001 ). The At2-MMP metalloprotease from A. thaliana is thought to play a role in plant growth and development (Golldack et al. 2002).

We have demonstrated that plants and plant cells can be used as alternative platforms for production of functional DSPAαi . It is likely that product yields can be further increased by selfing and crossing of elite plant lines. Although production levels in transgenic BY-2 suspension cells were lower, our initial experiments clearly indicate that these levels can be significantly increased by the optimization of medium composition, culture conditions and process technologies. The most effective strategy for elevating production levels appears to be the elimination or inhibition of proteases in the culture medium. The advantage of protease-deficient strains for recombinant protein production has been demonstrated for Escherichia coli (Meerman and Geor- giou 1994). The development of protease-deficient plants and plant cells will certainly significantly improve the plant production platform, especially when the recombinant protein can be recovered from the culture medium reducing the costs for extraction and purification.

The method according to the invention does not only provide a way how produced DSPA α1 but is also applicable for the production of other Des- moteplase isoforms as disclosed in US 6,008,019 and US 5,830,849 incor¬ porated herein by references or for plasminogen activators with a structural identity to DSPA α1 of at least 75 %, preferably 80 through 90 % (compared with the amino acid sequence as shown in figure 12. References

An G (1985) High efficiency transformation of cultered tobacco cells. Plant Physiology 79: 568-570.

Aoki Y, Takei R, Mohri M, Gonda Y, Gomi K, Sugihara T, Kiyota T, Yamamoto S, lshida T & Maruyama I (1994) Antithrombotic effects of recombinant human soluble thrombomodulin (rhs-TM) on arteriovenous shunt thrombosis in rats. American Journal of hematology 47: 162-166.

Ausubel FM, Brent R & Kingston RE (1994) Current Protocols in Molecular Biology. New York, Wiley Interscience.

Bardor M, Faye L & Lerouge, P (1999) Analysis of the N-glycosylation of recombinant glycoproteins produced in transgenic plants. Trends in Plant Science 4: 376-380.

Bateman KS, Congiu M, Tregear GW, Clarke AE, Anderson MA. 1997. Baci¬ tracin significantly reduces degradation of peptides in plant cell cultures. Biotechnology and Bioengineering 53(2):226-231.

Belozersky MA, Dunaevsky YE, Voskoboynikova NE. 1990. Isolation and properties of a metalloproteinase from buckwheat (Fagopyrum esculen- tum) seeds. Biochem J 272(3):677-82.

Berg DT, Grinnell BW. 1991. Signal and propeptide processing of human tissue plasminogen activator: activity of a pro-tPA derivative. Biochem Biophys Res Commun 179(3): 1289-96.

Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye bind¬ ing. Analytical Biochemistry 72: 248-254.

Carpita N, Sabularse D, Montezinos D & Delmer DP (1979) Determination of the pore size of cell walls of living plant cells. Science 205: 1144-1147.

Cramer CL, Weissenborn DL, Oishi KK, Grabau EA, Bennett S, Ponce E, Grabowski GA, Radin DN. 1996. Bioproduction of human enzymes in transgenic tobacco. Ann N Y Acad Sci 792:62-71. Dower WJ, Miller JF, Ragsdale CW. 1988. High efficiency transformation of E. coli by high voltage electroporation. Nucleic Acids Res 16(13):6127- 45.

Esmon CT (1989)The roles of protein C and thrombomodulin in the regula¬ tion of blood coagulation. Journal of Biological Chemistry 264: 4743- 4746.

Esmon CT, Esmon NL & Harris KW (1982) Complex formation between thrombin and thrombomodulin inhibits both thrombin-catalyzed fibrin for¬ mation and factor V activation. Journal of Biological Chemistry 257: 7944-7947.

Fischer R, Emans N, Schuster F, Hellwig S, Drossard J. 1999. Towards molecular farming in the future: using plant-cell-suspension cultures as bioreactors. Biotechnology and Applied Biochemistry 30(Pt 2): 109-112.

Fischer R, Schumann D, Zimmermann S, Drossard J, Sack M & Schillberg S (1999) Expression and characterization of bispecific single-chain Fv fragments produced in transgenic plants. European Journal of Biochem¬ istry 262: 810-816.

Glaser CB, Morser MJ & Light DR (1991 ) Oxidation resistant thrombo¬ modulin analogs. International Patent Application Publication No. WO91/15514.

Gohlke M, Nuck R, Kannicht C, Grunow D, Baude G, Donner P₁ Reutter W. 1997. Analysis of site-specific N-glycosylation of recombinant Desmodus rotundus salivary plasminogen activator rDSPA alpha 1 expressed in Chinese hamster ovary cells. Glycobiology 7(1 ):67-77.

Golldack D, Popova OV, Dietz KJ. 2002. Mutation of the matrix metallopro- teinase At2-MMP inhibits growth and causes late flowering and early senescence in Arabidopsis. Journal of Biological Chemistry 277(7):5541- 5547.

Gonda Y, Hirata S, Saitoh K, Aoki Y, Mohri M, Gomi K, Sugihara T, Kiyota T, Yamamoto S, lshida T & Maruyama I (1993) Antithrombotic effect of recombinant human soluble thrombomodulin on endotoxin-induced dis- seminated intravascular coagulation in rats. Thrombosis Research 71 : 325-335.

Gruber V, Berna PP, Arnaud T, Bournat P, Clement C, Mison D, Olagnier B, Philippe L, Theisen M, Baudino S and others. 2001. Large-scale produc¬ tion of a therapeutic protein in transgenic tobacco plants: effect of sub¬ cellular targeting on quality of a recombinant dog gastric lipase. Molecu¬ lar Breeding 7(4):329-340.

Hinz G, Hoh B, Hohl I, Robinson DG. 1995. Stratification of storage proteins in the protein storage vacuole of developing cotyledons of Pisum sativum L. Journal of Plant Physiology 145(4):437-442.

Hood EE, Woodard SL & Horn ME (2002) Monoclonal antibody manufactur¬ ing in transgenic plants - myths and realities. Current Opinion in Bio¬ technology 13: 630-635.

Horsch RB, Fry JE, Hoffmann NL₁ Eichholtz D, Rogers SG, Fraley RT. 1985. A simple and general method for transferring genes into plants. Science 227(4691): 1229-1231.

Horton RM, Cai ZL, Ho SN, Pease LR. 1990. Gene splicing by overlap extension: tailor-made genes using the polymerase chain reaction. Bio- techniques 8(5):528-535.

Hummel KM, Penheiter AR, Gathman AC, Lilly WW. 1996. Anomalous esti¬ mation of protease molecular weights using gelatin-containing SDS- PAGE. Analytical Biochemistry 233(1): 140-2.

James E, Lee JM. 2001. The production of foreign proteins from genetically modified plant cells. Advances in Biochemical Engineering/Biotechnology 72:127-156.

James EA, Wang C, Wang Z, Reeves R, Shin JH, Magnuson NS & Lee JM (2000) Production and characterisation of biologically active human GM- CSF secreted by genetically modified plant cells. Protein Expression and Purification 19: 131-138.

Kapila J, De Rycke R, Van Montagu M & Angenom G (1997) An Agrobacte- m/m-mediated transient gene expression system for intact leaves. Plant Science 122: 101-108. Kay R, Chan A, Dayly M & McPherson J (1987) Duplication of CaMV 35S promoter sequences creates a strong enhancer for plant genes. Science 236: 1299-1302.

Koncz C & Schell J (1986) The promoter of TL-DNA gene 5 controls the tissue specific expression of chimeric genes carried by a novel type of Agrobacterium binary vector. Molecular and General Genetics 204: 383- 396.

Kratzschmar J, Haendler B, Langer G, Boidol W, Bringmann P, Alagon A, Donner P, Schleuning WD. 1991. The plasminogen activator family from the salivary gland of the vampire bat Desmodus rotυndus: cloning and expression. Gene 105(2):229-237.

LaCount W, An GH, Lee JM. 1997. The effect of polyvinylpyrrolidone (PVP) on the heavy chain monoclonal antibody production from plant suspen¬ sion cultures. Biotechnology Letters 19(1):93-96.

Lammli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227: 680-685.

Lee JH, Kim NS, Kwon TH, Jang YS, Yang MS. 2002. Increased production of human granulocyte-macrophage colony stimulating factor (hGM-CSF) by the addition of stabilizing polymer in plant suspension cultures. Jour¬ nal of Biotechnology 96(3):205-211.

Liberatore GT, Samson A, Bladin C, Schleuning WD, Medcalf RL. 2003. Vampire bat salivary plasminogen activator (desmoteplase) - A unique fibrinolytic enzyme that does not promote neurodegeneration. Stroke 34(2):537-543.

Lin JH, Wang M, Andrews WH, Wydro R & Morser J (1994) Expression effi¬ ciency of the human thrombomodulin-encoding gene in various vector and host systems. Gene 147: 287-292.

Liu Y, Dammann C, Bhattacharyya MK. 2001. The matrix metalloproteinase gene GmMMP2 is activated in response to pathogenic infections in soy¬ bean. Plant Physiology 127(4): 1788-97. Ma JK, Drake PM, Christou P. 2003. Genetic modification: The production of recombinant pharmaceutical proteins in plants. Nature Reviews Genetics 4(10):794-805.

Ma JK, Hikmat BY, Wycoff K, Vine ND, Chargelegue D, Yu L, Hein MB, Lehner T. 1998. Characterization of a recombinant plant monoclonal secretory antibody and preventive immunotherapy in humans. Nature Medicine 4(5):601-606.

Magnuson NS, Linzmaier PM, Gao JW, Reeves R, An G, Lee JM. 1996. Enhanced recovery of a secreted mammalian protein from suspension culture of genetically modified tobacco cells. Protein Expression and Purification 7(2):220-228.

Mayne J, Robinson JJ. 1996. Purification and metal ion requirements of a candidate matrix metalloproteinase: A 41 kDa gelatinase activity in the sea urchin embryo. Biochemistry and Cell Biology-Biochimie Et Biologie Cellulaire 74(2):211-218.

McKnight TD₁ Bergey DR, Burnett RJ, Nessler CL. 1991. Expression of Enzymatically Active and Correctly Targeted Strictosidine Synthase in Transgenic Tobacco Plants. Planta 185(2): 148-152.

McKnight TD, Roessner CA, Devagupta R, Scott Al & Nessler CL (1990) Nucleotide sequence of a cDNA encoding the vacuolar protein stricto¬ sidine synthase from Catharanthus roseus. Nucleic Acids Research 18:4939.

Meerman HJ, Georgiou G. 1994. Construction and characterization of a set of E. coli strains deficient in all known loci affecting the proteolytic stabil¬ ity of secreted recombinant proteins. Biotechnology (N Y) 12(11): 1107- 10.

Murashige T & Skoog F (1962) A revised medium for rapid growth and bioassay with tobacco tissue cultures. Physiologia Plantarum 15: 473- 497.

Nielsen H, Engelbrecht J, Brunak S and von Heijne G (1997) Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleav¬ age sites. Protein Engineering 10: 1-6. Paris N, Stanley CM, Jones RL & Rogers JC (1996) Plant cells contain two functionally distinct vacuolar compartments. Cell 85: 563-572.

Parmenter DL, Boothe JG, van Rooijen GJH, Yeung EC & Moloney MM (1995) Production of biologically active hirudin in plant seeds using oleo- sin partitioning. Plant Molecular Biology 29: 1167-1180.

Petri T, Langer G, Bringmann P, Cashion L, Shallow S, Schleuning WD, Donner P. 1995. Production of vampire bat plasminogen activator DSPA alpha 1 in CHO and insect cells. Journal of Biotechnology 39(1):75-83.

Rerie WG, Whitecross Ml & Higgins TJ (1990) Nucleotide sequence of an A- type legumin gene from pea. Nucleic Acids Research 18: 655.

Schleuning WD, Alagon A, Boidol W, Bringmann P, Petri T, Kratzschmar J, Haendler B, Langer G, Baldus B, Witt W and others. 1992. Plasminogen activators from the saliva of Desmodus rotυndus (common vampire bat): unique fibrin specificity. Annals of the New York Academy of Sciences 667:395-403.

Sharp JM & Doran PM (2001 ) Characterization of monoclonal antibody frag¬ ments produced by plant cells. Biotechnology & Bioengineering 73: 338- , 346.

Sijmons PC, Dekker BM, Schrammeijer B, Verwoerd TC, van den Elzen PJ, Hoekema A. 1990. Production of correctly processed human serum albumin in transgenic plants. Bio-Technology 8(3):217-221.

Solis MM, Vitti M, Cook J, Young D₁ Glaser C, Light D, Morser J, Wydro R, Yu S, Fink L & Eidt JF (1994) Recombinant soluble human Thrombo¬ modulin: a randomized, blinded assessment of prevention of venous thrombosis and effects on hemostatic parameters in a rat model. Throm¬ bosis Research 73: 385-394.

Suzuki K, Kusumoto H, Deyashiki Y₁ Nishioka J₁ Maruyama I₁ Zushi M₁ Kawahara S₁ Honda G, Yamamoto S & Horiguchi S (1987) Structure and expression of human thrombomodulin, a thrombin receptor on endothe¬ lium acting as a cofactor for protein C activation. EMBO Journal 6: 1891- 1897. Tepfer M and Taylor IEP (1981) The permeability of plant cell walls as measured by gel filtration chromatography. Science 213, 761-763.

Titel C & Ehwald R (1999) Permeability limits of cell walls in suspension cultures: biotechnological and physiological aspects. Russian Journal of Plant Physiology 46: 739-744.

Tsoi BM, Doran PM. 2002. Effect of medium properties and additives on antibody stability and accumulation in suspended plant cell cultures. Biotechnology and Applied Biochemistry 35(Pt 3):171-80.

Twyman RM, Stoger E, Schillberg S, Christou P & Fischer R (2003) Molecu¬ lar farming in plants: host systems and expression technology. Trends in Biotechnology 21 : 570-578.

Vaquero C, Sack M, Chandler J, Drossard J, Schuster F, Monecke M, Schill¬ berg S & Fischer R (1999) Transient expression of a tumor-specific sin¬ gle-chain fragment and a chimeric antibody in tobacco leaves. Proceed¬ ings of the National Academy of Sciences of the United States of Amer¬ ica 96: 11128-11133.

Voss A, Niersbach M, Hain R, Hirsch HJ, Liao YC, Kreuzaler F & Fischer R (1995) Reduced virus infectivity in N. tabacum secreting a TMV-specific full-size antibody. Molecular Breeding 1 : 39-50.

Weisel JW, Nagaswami C, Young TA & Light DR (1996) The shape of Thrombomodulin and interactions with Thrombin as dete^'rmined by elec¬ tron microscopy. Journal of Biological Chemistry 271 : 31485-31490.

Witt W, Maass B, Baldus B, Hildebrand M, Donner P, Schleuning WD. 1994. Coronary thrombolysis with Desmodus salivary plasminogen activator in dogs. Fast and persistent recanalization by intravenous bolus admini¬ stration. Circulation 90(1 ):421 -426.

Woodard, S.L., J. M. Mayor, M. R. Bailey, D.K. Barker, RT. Love, J. R. Lane, D.E. Delaney, J. M. McComas-Wagner, H.D. Mallubhotla, E.E. Hood, L.J. Dangott, S. E. Tichy, and J.A. Howard. 2003. Maize (Zea mays)-derived bovine trypsin: characterization of the first large-scale, commercial pro¬ tein product from transgenic plants. Biotechnology and Applied Bio¬ chemistry 38:123-30. Yang XY, Mackins JY, Li QJ, Antony AC. 1996. Isolation and characterization of a folate receptor-directed metalloprotease from human placenta. Journal of Biological Chemistry 271 (19): 11493-9.

Zimmermann S, Schillberg S, Liao YC, Fisher R. 1998. Intracellular expres¬ sion of TMV-specific single-chain Fv fragments leads to improved virus resistance in Nicotiana tabacum. Molecular Breeding 4(4):369-379.

Claims

Claims

1. A method for producing Solulin, which method comprises growing plant cells containing an integrated sequence comprising a functional transcriptional cassette comprising a structural gene encoding for said Solulin.

2. A method for producing desmoteplaseαi (DSPAαi), which method comprises growing plant cells containing an integrated sequence com¬ prising a functional transcriptional cassette comprising a structural gene encoding for said desmoteplaseαi .