|
| Current Opinion in Biotechnology Vol. 7, No. 5, October 1996 The expression of recombinant proteins in yeasts [Review article] Peter E Sudbery Current Opinion in Biotechnology 1996, 7:517-524. |
|
—interferon-;
The advantages of yeast as hosts for the expression of recombinant proteins from higher eukaryotes have long been appreciated (for a general review see [1]). They combine the ease, simplicity and cheapness of bacterial expression systems with the authenticity of the far more expensive and less convenient animal tissue culture systems. Like bacteria, yeast are simple to cultivate on inexpensive growth media, and there is a formidable array of techniques for the manipulation of foreign genes. However, as eukaryotes, they provide an environment for post-translational processing and secretion, resulting in a product that is often identical or more similar to the native protein. Moreover, the budding yeast Saccharomyces cerevisiae has a long history of use in the production of bread and alcoholic beverages. As a result, there is confidence that the organism is safe, and experience of industrial-scale fermentations facilitates production scale-up.
The first yeast to be employed for the production of recombinant proteins was S. cerevisiae. Foreign genes are either maintained on 2 µm-based plasmids, which maintain high copy number but require selection upon growth in large fermenters or in a low copy integrated form. Strong, unregulated promoters such as that from the PGK1 gene, or the regulatable GAL1 promoter, may be employed. A particular advantage of yeast is that the foreign protein may be directed into the secretory pathway, usually by fusion of the mature form of the recombinant protein to the prepro sequence of the mating factor (MF). The pro sequence is removed by the proteolytic action of the Kex2 enzyme, a processing step that is widely conserved in eukaryotes. During secretion, protein folding, disulphide bond formation and glycosylation takes place. N-linked glycosylation occurs at the correct sites but often differs from the native pattern in that the carbohydrate chains are often of excessive length and are of the high-mannose type. The resulting protein may therefore be immunogenic and this limits the use of glycosylated proteins as human therapeutics. This basic methodology was established by the mid 1980s and led to the successful production of the hepatitis B vaccine based on recombinant HBsAg (hepatitis B surface antigen). Recombinant human serum albumin (rHA) produced in S. cerevisiae is also in clinical trials as a plasma expander [2]. Native serum albumin contains 17 disulphide bonds but is not glycosylated. The protein is produced by yeast in completely authentic from. To be economic, it must be produced in tonne quantities at US$2–3 g-1. This will necessitate the construction of a fermentation facility of the order of 100 m3. Such an undertaking is testament to the ability of yeast to provide the dual advantages of authenticity at low cost.
Despite these successes, S. cerevisiae is often limited as an expression system by low yields. Methylotrophic yeasts such as Pichia pastoris and Hansenula polymorpha retain all the advantages of S. cerevisiae but provide a reliable means of achieving greatly elevated yields [3] [4•]. Moreover, P. pastoris may, at least partially, solve the glycosylation problem. Both H. polymorpha and P. pastoris can grow on simple, defined media utilizing methanol as the sole carbon source. The first enzyme in the methanol-utilization pathway is an alcohol (methanol) oxidase known as AOX in P. pastoris and MOX in H. polymorpha. During growth on methanol, this enzyme constitutes some 30% of the total cell protein, sequestered in peroxisomes that can occupy 80% of total cell volume. The expression of AOX/MOX is highly regulated. In both H. polymorpha and P. pastoris, the enzyme is repressed by glucose or ethanol so that it is undetectable. In both yeasts, full induction requires the presence of methanol. In H. polymorpha, but not P. pastoris, MOX is derepressed to 10–20% of the maximum during growth on a nonrepressing carbon source such as glycerol. The powerful, regulatable promoter that programs the expression of AOX/MOX has been fully exploited in both yeasts for the production of recombinant proteins. The expression cassette is normally integrated by transplacement or duplicative integration and is stable during nonselective growth. A variety of approaches are available to increase copy number, which, in some cases, can increase yield. Fermentation in a bioreactor is essential for high expression levels but useful quantities of protein can be obtained from shake flasks. Fermentation protocols are simple: very high dry-cell weights can readily be obtained and biomass production can be separated from the production phase.
The expression of a heterologous gene in yeast may be undertaken because the recombinant protein has intrinsic commercial value, or the engineered strain may serve as the basis to develop other products. This may come about through the study of a cellular process of medical importance, an understanding of which may ultimately lead to the development of novel therapeutics. For example, the recombinant protein may be used for structural studies leading ultimately to the rational design of a small chemical inhibitor or activator; alternatively, the engineered strain may serve as the basis for a drug screening program. Alternatively, a mammalian receptor may be expressed, so that a screen can be established to search for drugs that act as agonists or antagonists. In the latter case, the similarity of fundamental cellular processes between yeast and higher organisms, coupled with the highly developed genetics of S. cerevisiae, make this yeast an extremely powerful experimental system.
This review will be concerned with examples, published since January 1995, of recombinant proteins of biotechnological relevance that have been expressed in yeast. For convenience, proteins of intrinsic value will be separated from those where the interest lies in the construction of model systems. The technology is now essentially mature and there have been no revolutionary developments. However, useful contributions to the technical side have appeared and these will also be considered.
Production of such proteins in S. cerevisiae is limited by the cost of
13C-labelled glucose. The use of 13C-labelled methanol
reduces the cost tenfold, and could make the technique routine where previously
it was applicable only in the most exceptional circumstances. The utility of
this idea was demonstrated by the expression of tick anticoagulant peptide (TAP)
from the soft tick Ornithodoros moubata in P. pastoris [5]. The
60 amino acid TAP polypeptide is a potent and specific inhibitor of blood
coagulation factor Xa and so could be an important therapeutic agent. A
synthetic, codon-optimized gene was expressed in P. pastoris fused with
the S. cerevisiae MF prepro sequence to direct secretion. A yield of 1.7
gl-1 was obtained from a single copy insertion. In molar terms, this
represents the highest yield reported from P. pastoris and was a
sevenfold increase in productivity compared to S. cerevisiae. This system
was used to produce single and dual 13C-
and
15N-labelled protein enriched for more than 98% as determined by
one-dimensional NMR.
Hirudin is a potent thrombin inhibitor produced by the leech Hirudo medicinalis. It has been successfully expressed in both S. cerevisiae and H. polymorpha [6] [7] [8•] [9•] [10] [11] [12] [13]. The highest yield was reported in H. polymorpha using a variety of leader sequences. Yields in the gml-1 range were reported [13]. A second group made the interesting observation that yields were greatly elevated by the addition of soybean oil to the medium, which aids yeast growth by providing essential fatty acids [9•]. In S. cerevisiae, a vector based on the use of the copper-inducible metallothionein (CUP1) gene has been used [8•]. This allows selection in complex medium by the addition of cupric sulphate. In copper-sensitive hosts, the apparent plasmid stability was 100%. The selection resulted in a twofold increase in plasmid copy number. The highest yield was obtained using the CUP1 promoter to program the expression of hirudin. In such a situation, the transcription of hirudin is directly linked to the transcription of the selectable marker, so an increase in selective pressure leads directly to an increase in product yield.
Recombinant antibodies are likely to be one of the major products of the biotechnology industry. The use of murine monoclonal antibodies for diagnostic purposes is well known, but their use as therapeutic agents is limited by the likelihood of an immune response in a human subject. Monoclonal human antibodies can be produced by the powerful combinatorial library approach. Improvements in specificity and avidity can be made by clonal selection using phage display systems. Rational engineering can be used to produce novel proteins that, for example, may link a toxin or radioisotope to an antibody raised against a tumour antigen. Such antibodies may be produced in bacteria using signal sequences to direct secretion to the periplasmic space; however, yields are often disappointingly low. P. pastoris has been used to efficiently express rabbit scFv antibody fragments that recognize human leukaemia inhibitory factor isolated from a combinatorial library [14••]. Single chain (sc) Fv fragments consist of VL and VH domains joined by a short linker. The protein was tagged with five consecutive histidine residues to aid purification by Ni+-chelate affinity chromatography and secretion was directed using the S. cerevisiae MF prepro sequence. Shake flask cultures produced 100 mgl-1 of active protein, a 100-fold increase compared to E. coli. Further increases in yield may be expected when a fermenter is used.
As well as antibodies, antibody receptors may be expressed in P.
pastoris (S Cain, B Helm, personal communication). Human IgE mediates the
allergic response. It binds to two receptors, both of which are involved in type
1 hypersensitivity reactions. The high affinity receptor (FcRI) is found on mast
cells and basophils and mediates the immune responses following cross-linking to
IgE at the cell surface. It is a tetrameric protein of the form 
. The
extracellular portion of the
subunit binds IgE. It contains two immunoglobulin-like domains, the second of
which binds IgE, and the first confers high affinity to the binding reaction.
The extracellular domain of the chain has been expressed in P. pastoris
both in its native form and as the individual immunoglobulin-like domains. After
growth in a fermenter, 200–250 mgl-1 of product was obtained from the
strain expressing the first domain, and 60 mgl-1 from each of the
strains expressing either the second domain or the total extracellular domain.
In all cases, the fragments were secreted into the culture medium. Previous
attempts to express these fragments in E. coli and Baculovirus were
disappointing, demonstrating the efficacy of this yeast system. IgE binding to
the recombinant FcRI was examined using a BIAcore Ambis scanner and found to be
indistinguishable from the native reaction. The structure of the receptor may
now be studied by two-dimensional NMR after isotopic labelling of the proteins
using 13C-labelled methanol as discussed above. The IgE constant
region has also been expressed so that the structure of the IgE–receptor complex
may be studied using X-ray crystallography.
Protein fusions are commonly employed in many different expression systems. Yield and biological activity of the recombinant protein may be increased by fusion to a rapidly folding protein such as thioredoxin to aid folding and enhance solubility, or, alternatively, to a peptide tag recognized by a commercially available monoclonal antibody or to histidine residues for binding to an Ni+-chelate affinity chromatography column. Whatever the reason for the fusion, an authentic protein requires cleavage of the chimeric protein. This is usually achieved by including a short peptide linker between the two moieties that forms a target for a protease with high specificity. Because of the increasing use of such fusions, there is an important market for the high specificity proteases. One such protease is bovine enterokinase, a serine protease that is the physiological activator of trypsinogen. The target of enterokinase is a Lys–Ile dipeptide that is preceded by a negatively charged Asp4 sequence. Enterokinase is a heterodimeric protein, consisting of a 115 kDa structural subunit that anchors a 35 kDa catalytic subunit to the intestinal brush border. The recombinant catalytic subunit produced in COS cells or E. coli retains the specificity of the holoenzyme but exhibits a 25-fold increase in activity. Shake flask cultures of engineered P. pastoris produced 6.5 mgl-1 of enzyme [15]. This enzyme was used to process thioredoxin–CAT and calmodulin–GFP fusion proteins produced in E. coli (CAT, chloramphenicol transacetylase; GFP, green fluorescent protein) and was found to be 100-fold more reactive than the native holoenzyme. Although the yield is low in comparison with other proteins, it is enough to process 1.7 g of CAT or 286.4 g of GFP fusions!
The Bm86 antigen from the cattle tick Boophilus microplus has been expressed in P. pastoris [16]. The recombinant protein was glycosylated and formed particles 17–45 nm in diameter which proved to be highly immunogenic, so that ticks engorging on vaccinated cattle were significantly damaged. An amino-terminally truncated form of outer-surface protein A from the Lyme disease spirochete Borrelia burgdorferi has been expressed in S. cerevisiae, resulting in a yield of 2.5 gl-1 [16]. Attempts to express this protein in E. coli had proven unsuccessful. Finally, Pfs25, a cysteine-rich 25 kDa protein present on the surface of Plasmodium falciparum zygotes has been expressed in S. cerevisiae [17]. When the recombinant protein was adsorbed to alum, it induced antibodies in both rodents and primates. These antibodies blocked the transmission of malaria parasites when mixed with infectious blood and fed to mosquitoes through a membrane feeding apparatus. Furthermore, unlike monoclonal antibodies to Pfs25, which block transmission only after ookinete development, antisera to Pfs25-B adsorbed to alum appeared to block the in vivo development of zygotes to ookinetes as well.
Interferon-
(IFN-)
signals the presence of an early conceptus to the maternal uterus and ensures
the continued production of progesterone by the ovarian corpus luteum. 280
mgl-1 of IFN- was
secreted in a sufficiently pure form such that, for many purposes, no further
purification is necessary.
Small peptides can be difficult to express in yeast. However, expression of human insulin-like growth factor II (IGFII; 67 amino acids) and Xenopus laevis magainin (23 amino acids) was found to be facilitated in H. polymorpha by fusion to the carboxyl terminus of amine oxidase [18••]. High-level synthesis of the fusion proteins, exceeding 20% of total cell protein, was obtained in methanol-limited chemostat cultures. After synthesis, the peptides could be readily separated from amine oxidase through factor Xa cleavage of a target sequence engineered into the linker. Although amine oxidase is normally targeted to the peroxisome via the PTS2 (PTS, peroxisomal targeting sequence) import pathway, the fusion proteins remained cytosolic. However, the addition of a carboxy-terminal SKL (PTS1) sequence to the AMO–IGFII construct did direct the recombinant protein to the peroxisome. This is an important finding, because it demonstrates a novel route for the production of recombinant proteins in methylotrophic yeasts. Peroxisomal targeting may reduce any toxic effects of the recombinant protein and, because peroxisomes can readily be separated from cell debris, it may facilitate the recovery and purification of an internally expressed protein.
As well as providing products of direct value, the expression of proteins in
yeast can be used to establish experimental systems for the development of other
pharmaceutical products. For example, the expression of mammalian receptor
proteins in yeast can form the basis of screens for small chemical drugs that
may act as agonists or antagonists of receptor function. An important and
elegant example of this is provided by the expression of rat somatostatin
receptor subtype 2 (SSTR2) in S. cerevisiae [19••]
(Fig.
1). This is a G-protein-coupled receptor with seven transmembrane
domains. G-proteins are heterotrimeric proteins consisting of , and
subunits. The
subunit binds GTP and interacts with both the receptor and intracellular
effectors and so defines the activity of the G-protein complex. Yeast mating
pheromones also bind to G-protein-coupled receptors with seven transmembrane
domains. Upon pheromone binding, the trimeric complex dissociates, relieving
negative regulation by the subunit on the dimer. This dimer then stimulates a
mitogen-activated protein kinase (MAPK) cascade, which has two consequences:
firstly, the activation of FAR1 and consequent cell cycle arrest in G1;
and secondly, the transcriptional activation of FUS1. In order to create
a yeast cell that can respond to somatostatin activation, the following changes
were engineered in the his3 host: firstly, SSTR2 was expressed from a
GAL1 promoter on a multicopy plasmid; secondly, the FAR1 gene was
deleted, preventing cell-cycle arrest upon activation of the mating pheromone
pathway; thirdly, the FUS1 gene was replaced with a fusion of the
FUS1 pheromone responsive elements to the HIS3 coding sequences,
so making His3 expression dependent upon activation of the mating pheromone
pathway; fourthly, the GPA1 gene encoding the yeast G
subunit was deleted, to be replaced by a gene fusion consisting of the
amino-terminal G
interacting domain from GPA1 and the carboxy-terminal
receptor-interacting domain of the rat Gi2,
thus allowing recombinant SSTR2 to stimulate the yeast pheromone response
pathway; and fifthly, the yeast SST2 gene was deleted, increasing
sensitivity to mating pheromone. As a result of these changes, somatostatin
induced a dose-dependent increase in growth on media lacking histidine. Thus,
the strain provides a sensitive bioassay to examine the molecular nature of
interactions between the ligand and receptor and between the receptor and its
cognate heterotrimeric G-protein complex. Moreover, the system should be
generally adaptable for the discovery of novel, therapeutically active ligands
for receptors of medical importance.
 |
|
Figure 1 Somatostatin-dependent growth of yeast. (a) |
Receptor-type tyrosine kinases, such as Src and Abl, play a key role in regulating cell proliferation and other processes in mammalian cells. They form part of the signal transduction pathway. Upon ligand binding to the extracellular domain, the receptors dimerize, resulting in autophosphorylation and consequent activation. The Src family of receptor kinases are negatively regulated by a discrete group of tyrosine kinases called the Csk family. These phosphorylate members of the Src family at a conserved tyrosine (Tyr527). This maintains the receptors in a low level of activity in the absence of ligand binding. Tyrosine kinases present an attractive target for small drug action to regulate cell proliferation. Two papers describe yeast-based screens for regulators and antagonists of their action [20••] [21••].
Expression of the Src tyrosine kinase is lethal in S. pombe, so cells expressing Src from a repressible thiamine promoter are unable to grow in the absence of thiamine. When such cells are transformed with a cDNA library, derived from either SV40 large T transformed human lung fibroblasts or human Burkitt lymphoma cells, only clones expressing negative regulators of Src are able to grow in the absence of thiamine. This screen isolated two classes of regulators [21••]. The first corresponded to the Csk family of Src regulators. The second were tyrosine phosphatases that inactivated Src itself or dephosphorylated its targets. Although both classes of regulators were already known, the results clearly validate the methodology. It is possible that these regulators are over-represented in the cDNA libraries used. Alternatively, there may be no more regulators left to identify.
Further members of the Src family can be isolated by screens based on conserved sequences. However, there is little sequence conservation in the ligands to which they bind. A screen in S. cerevisiae to identify such ligands has been developed. There are only very small amounts of proteins phosphorylated on tyrosine residues in a normal yeast cell. However, heterologous expression of an Src-family kinase leads to an increase that is readily detected by antiphosphotyrosine antibodies. The level is dramatically increased by the simultaneous expression of a known ligand. A yeast cell expressing the fibroblast growth factor (FGF) receptor gene was transformed with a Xenopus cDNA library and screened by colony immunoblotting. Six novel sequences were recovered. Of particular interest were two genes, FRL1 and FRL2, which are distantly related to epidermal growth factor (EGF) and angiotensin respectively. Both genes activated the FGF receptors in Xenopus oocytes. Overexpression of both genes induces mesoderm and neural specific genes in explants, and they are expressed at different stages during development. FRL1 is broadly expressed during gastrulation and neuralation; FRL2 is expressed principally in the axial mesoderm and brain at later stages.
The overexpression of P-glycoproteins in human cancers has been shown to confer resistance to a variety of unrelated chemotherapeutic drugs. Such proteins are members of the ATP-binding cassette (ABC) superfamily of membrane transporters. Other members of this group include the yeast a-mating pheromone transporter Ste6, and the human cystic fibrosis transmembrane conductance regulator (CFTR) protein, mutations of which result in cystic fibrosis. Expression of human multidrug resistance associated protein (MRP), or the mouse MRP Mdr3, in an anthracycline supersensitive yeast restored resistance to adriamycin, and conferred resistance to the unrelated antifungal agents FK506 and valinomycin [22••]. Radiolabelling experiments suggested that adriamycin was removed by an active transport mechanism. Human MRP also partially complements a sterile ste6 allele, suggesting that although MRP is only weakly homologous in structure to Ste6p, it can mediate the transport of the a-mating pheromone. Yeast may therefore prove to be extremely useful for the biochemical and pharmacological characterization of eukaryote ABC transporters.
Cytochrome P450 and NADPH–cytochrome P450 oxidoreductase (OR) are key enzymes in the metabolism of xenobiotics, therapeutically active drugs and carcinogens. The heterologous expression of human P450s in yeast provides a valuable model system for their study. The expression of human P450s in yeast is limited by the activity of yeast OR (yOR) and the availability of internal membranes in which they are anchored. The lack of yOR may be overcome by the fusion of CYPIAI, the human gene encoding P450AI, to human OR (hOR) and the expression of the chimeric protein in yeast [23••]. This results in the expression of a catalytically self-sufficient, hybrid protein correctly located in internal membranes. The quantity of internal membranes may be increased by the overproduction of chimeric proteins consisting of CYPIAI fused to HMG2/1, which comprises the seven transmembrane domains of the S. cerevisiae 3-hydroxy-3-methylglutaryl-CoA reductase isozymes 1 and 2 [24••]. The hybrid protein was catalytically active and correctly located in the internal membranes. Replacement of the amino-terminal membrane anchor domain of human NADPH–cytochrome P450 OR by the HMG2/1 peptide also resulted in a functional fusion enzyme that was able to interact with the HMG2/1–CYPIAI fusion and the yeast endogenous P450 enzyme lanosterol-14-demethylase.
There are many examples in which the expression of proteins in yeast is
apparently limited by bottlenecks in the secretory pathway. Two recent papers [25] [26]
suggest ways in which this may be overcome. The secretion of human leukocyte
protease inhibitor was found to be elevated three- to
fourfold by simultaneous overexpression of ubiquitin from a chromosomal
UBI4 gene under the control of the GAL1 promoter [25].
Similarly, the overexpression of protein disulphide isomerase resulted in a
10-fold increase in the levels of secretion of human platelet-derived growth
factor and a fourfold increase in the levels of acid phosphatase [26].
The expression of proteins in yeast is often undertaken for reasons of fundamental research. Many investigations require the ectopic expression of a protein under the control of promoters programming different levels of expression. A convenient set of vectors has been developed that allows the constitutive level of proteins to be expressed over a range of three orders of magnitude [27] [28]. The gene to be expressed is cloned into a multiple cloning site situated between the promoter and the CYC1 transcriptional terminator. The promoter may be the weak CYC1 promoter, the ADH promoter or the stronger TEF1 and GPD promoters. The set is based on centromeric or 2 µm vectors.
Because stable replicating vectors are not available for methylotrophic yeasts, genes must be chromosomally integrated. This can make it difficult to obtain strains harbouring a high copy number of the heterologous gene. In leu2 S. cerevisiae strains, the copy number of plasmids carrying the leu2d allele is elevated compared to plasmids bearing a wild-type LEU2 sequence. This is thought to be due to the inefficient expression of Leu2 leading to selection for cells with more copies of the plasmid. A similar process can be used to increase the copy number of chromosomally integrated genes in H. polymorpha [29•]. When a ura3 strain is transformed with a plasmid carrying the S. cerevisiae URA3 gene, the plasmid replicates in a highly unstable fashion. Presumably, this is due to a fortuitous ARS sequence in the URA3 gene. After prolonged growth in nonselective conditions, the replicating form of the plasmid disappears and is replaced by multiple copies (5–30 per cell), integrated head-to-tail in tandem arrays at a chromosomal site that is not the homologous URA3 locus. The high copy number probably reflects inefficient expression of the heterologous URA3 gene, leading to selection for increased copy number. It is not clear whether a single copy of the plasmid integrates first, followed by integration events of further plasmids at the same site, or whether multimerization of the plasmid occurs prior to a single integration event.
A similar phenomenon was observed during the expression of recombinant
haemoglobin (rHbA) in H. polymorpha [30],
using the S. cerevisiae LEU2 gene for selection. The construct carrying
the and
chains
was chromosomally integrated. After 40 generations in a chemostat, amplification
was observed. The pressure for amplification is presumed to arise from the
inefficient expression or function of the heterologous LEU2 gene. In some
cases, the LEU2 gene was amplified but the globin expression cassettes
were lost. In other cases, the expression cassettes were also amplified and
there was a corresponding increase in the yield of rHbA. Such isolates
maintained the increase when grown in fed-batch fermentation, producing fully
functional rHbA.
Selection based on an integrated copy of the homologous H. polymorpha LEU2 gene can also be unstable [31•]. Despite vectors lacking any ARS sequence, unstable replicating plasmids arise that can retransform E. coli. Apparently, rearrangements result in the acquisition of a chromosomal ARS sequence. The same group have also described an elegant procedure for the targeted integration of an expression cassette to the MOX locus [32•]. It relies upon the TRP3 gene, which lies immediately downstream of the MOX locus. A strain is generated in which the TRP3 gene is disrupted by insertion/deletion with the S. cerevisiae LEU2 gene. The expression cassette is constructed to carry 5' MOX promoter sequences at one end and the TRP3 sequences at the other, extending beyond the sequences replaced by the LEU2 gene. Transplacement of the disrupted locus with the expression cassette thus regenerates a Trp+ strain that can readily be selected.
Where the expression of recombinant protein is the only objective, methylotrophic yeasts are now the preferred option. There are now many examples in which proteins have been expressed at significantly higher yields in methylotrophic yeasts, or even in which the successful expression in a methylotrophic yeast of proteins that completely failed to be expressed in S. cerevisiae has been achieved. Apart from the advantages discussed above, a well-designed commercial kit based upon P. pastoris is now available from InVitrogen. This should facilitate the use of such systems for those new to the field. The only note of caution is that there is a long history of the safe use of products derived from S. cerevisiae. The relative newness of methylotrophic systems may make the acceptance of recombinant products by regulatory authorities more difficult. Where expression of a recombinant protein is to be the basis of a model system to be used to develop other products, for example, in a drug screen, the exquisite genetic system of S. cerevisiae is likely to ensure that it will remain the predominant system for the foreseeable future.
authentic processing of three different preprohirudins.
FUS1
strain together with a modified G