INTRODUCTION
Spider silks exemplify the power of millions of years of natural selection to produce high performance materials. The life cycles of most spider species revolve around their ability to produce a range of silk fibres with different properties. The average orb-weaving spider contains five or more silk glands each producing a different silk suited to specific purposes (Vollrath and Knight, 2001). For example, an orb web contains two very different silks. The struts of the web are made from dragline silk which is remarkably strong and elastic in order to deal with the stresses involved in capturing flying objects. The capture spiral, imposed on this framework, is made from a much weaker but extremely sticky silk whose role is to ensnare any creature that contacts the web. These silks are made in separate glands, whilst yet other glands make the fine prey wrapping silk, or silks for making cocoons for example.
Each silk is composed of a unique blend of structural proteins known as spidroins. The structure of specific spidroins is thought to underlie the distinct properties of the various silks (Gatesy et al., 2001). The structures of glands and spinnerets are specific for given silks and will determine obvious features such as fineness, and also influence mechanical properties of the finished fibre (Vollrath and Knight, 2001). The glands also specify the type of coating proteins on the fibre, which determine surface characteristics such as stickiness. Spidroins from a few silks have now been identified and, in some cases, encoding genes cloned. For example, the dragline silk of Nephila clavipes contains two major spidroins, both of which are very long (coding sequences up to 7kb) and highly repetitive in structure, being composed mostly of non-essential amino acids (Guerette et al., 1996). These features, extreme length and repetitiveness seem to be common to all spidroins, and are seen in other structural proteins such as fibroins and collagens for example.
The properties of spider silks have long been the envy of materials scientists, often considerably out-performing man-made equivalents. For example, the toughness of dragline silk is between 4 and 8 times greater than that of Kevlar- the toughest man-made fibre currently available. A range of high-value applications in industry and leisure, are foreseen for spider silks both as replacements for Kevlar as well in novel applications. Applications for spider silks in medicine have been demonstrated, both in the area of cell growth matrices for tissue repair and in the formation of medical implants (Sofia et al., 2001). Because of this potential, there has been substantial investment, by industry, Dupont for example (O’Brien et al., 1998), and the US military into understanding the relationship between dragline silk composition and structure.
An impressive characteristic of spider silks is the fact that they are spun out of aqueous solutions under ambient environmental conditions. This is in contrast to most man-made fibres, which make use of high temperatures, high pressure, and unpleasant solvents to reduce viscosity and enable fibres to be spun (Vollrath and Knight, 2001). There are currently two major barriers preventing the industrial production of spider silk analogues. The first is a source of the raw materials (the appropriate silk proteins) and the second is the spinning of these proteins into high-performance fibres. Whilst considerable progress has been made with industrial spinning processes these still need improvement and more importantly a reliable cost effective source of spider silk proteins remains elusive.
Considerable effort has been devoted to producing synthetic spider silks. For example, Dupont have produced truncated recombinant versions of spidroin 1 in microbial expression systems (O’Brien et al., 1998) and spun these into fibres but of unimpressive quality. More recently, a Canadian company has produced a much larger recombinant analogue of spidroin 1 secreted into transgenic goat milk (www.nexiabiotech.com). Although protein production appears successful from transgenic goats, the overall costs and yields of protein suggest this process will not underpin a competitive industrial production process.
To understand some of the problems involved in producing silks in vitro, it is informative to look at the spinning process in spider silk glands. More than 50% of a spider’s abdominal cavity may be occupied by silk glands. In a typical spider, the largest of these is the major ampullate gland that produces dragline silk. The diagram in Fig. 1 (adapted from reference Vollrath and Knight, 1999) illustrates the complex architecture of the major ampullate gland from Nephila sp.
Fig. 1 Diagramatic representation of a major ampullate gland of Nephila senegalensis (adapted from reference Vollrath and Knight, 1999)
The A-zone is highly active in secretion, forming a two-phase dope containing small droplets in a bulk phase (Vollrath and Knight, 1999). These droplets coalesce and become elongated eventually forming canaliculi in the final silk. Secretions in the B zone do not form droplets but appear to form a coat (visible by EM throughout the secretory pathway and the final thread) on the core formed in the A zone. The final silk thread possesses a third coating which appears to be applied in 3rd limb of the duct. Within the gland duct, the major proteins are maintained in a highly concentrated liquid crystalline form adopting predominantly -helical/-turn/random coil structure (Vollrath and Knight, 2001). As the protein solution passes through the duct of the gland, water is withdrawn and co-factors subtracted and added. On demand, thread is pulled from the opening of the spinneret. Under the flow forces generated by extensional flow, the proteins refold and cross-link to form partially crystalline fibres as they leave the spinneret. There is some evidence that the spider silk proteins may be chemically crosslinked through the activity of a peroxidase before they exit the glands (Vollrath and Knight, 1999), a process that may help reinforce the final fibre properties.
Recombinant Spider Silks
Currently, the main barrier to realising the range of applications for spider silk is that the proteins are produced in low levels by their natural sources. The isolation of cDNAs encoding spider silk proteins has opened up the possibility of expressing these genes in other organisms to produce useful quantities of protein. Traditionally, microbial fermentation is the most commonly used method for the production of recombinant proteins. Such systems are ideal for the production of small high-value products like human growth factor or insulin. Microbial systems, however, suffer from a number of features that renders them unsuitable for the production of large complex proteins from eukaryotic sources. In spite of this most biotechnological approaches to producing recombinant spider silks have concentrated on producing the proteins in bacteria or yeast.
The Disadvantages of Bacteria
In bacteria there are two major barriers to the production of recombinant silk proteins both involving the nature of the genes encoding the proteins. Spider silk proteins are very long polymers made up of a large series of repeating units. Obviously, the DNA encoding this repeat structure is also highly repetitive. Bacteria (such as E. coli) normally possess only trace quantities of repetitive DNA and attempts to introduce highly repetitive DNA, via transformation of the cells, is usually unsuccessful. Repetitive DNA produces ideal structures for homologous recombination leading to loss of the insert by rearrangements. The other major hurdle in the expression of spider proteins in E. coli lies in the differences in codon preference between these very different organisms. This is an important issue when dealing with repetitive proteins, such as those of silk, that are very rich in specific amino acids. A recent strategy has been to engineer a synthetic section of DNA in which codon usage is varied so as to limit the homology of repeats and to comply with usage in E. coli, while maintaining the repeat structure of the protein. Whilst this has proven useful for producing short silk-like proteins, these products were not long enough to produce good quality fibres, and further attempts to produce longer silks in bacteria have been unsuccessful (Vollrath and Knight, 1999). Recently, artificial silk constructs have been successfully expressed in yeast, but again the quantity and quality of protein produced render commercialisation unlikely (Vollrath and Shao, 1999).
Transgenic animals
A dramatic development in the last few years has been the production of recombinant spider silk proteins in transgenic animals. This approach has been successfully developed by Nexia Biotechnologies, a Canadian company http://nexiabiotech.com). Nexia produces recombinant silk proteins in the milk of transgenic goats and then isolate these proteins to spin into fibres. Although silks have been produced successfully in this way, the production costs are extremely prohibitive and this is unlikely to serve as the foundation for mass production.
The Advantages of Plants
In contrast to bacteria, plants have a number of distinct advantages for the expression of repetitive proteins. First, and most importantly, plant cells already normally produce a wide range of highly repetitive structural proteins, which function in the cell wall. The composition of these proteins, such as the proline-rich extensins and glycine-rich proteins, are surprisingly similar to spider silk proteins (Foelix, 1997). This means that the instability problems encountered in the use of E. coli are unlikely to occur in plants.
Second, plants and spiders share a much closer preference in codon usage than do spiders and bacteria. This is illustrated in the table below, which shows codon use for the major amino acids occurring in the repeat structure of a spider silk protein (spidroin 2). This is compared with codon use in a range of expressed genes from E. coli and tobacco plants.
These similarities make it highly probable that plant cells will be able to produce silk protein sequences. This opens up the possibility of directly using spider silk gene coding sequences for plant transformation, rather than having to engineer synthetic constructs which will have unknown properties. Once silk protein production by plant cells has been established and optimised, the process can be transferred to crop plants to provide added value and large-scale field production.
This report describes work that we have carried out towards addressing the major barriers preventing the industrial utilisation of spider silks. We will describe the production of an extensive and novel database of genes involved in the processing of silk proteins in spider silk gland. The major aim of the project was to produce transgenic tobacco plants for the production of recombinant silk proteins. Tobacco was our plant of choice because it is a non-food crop and has no wild relatives in the UK making the spread of transgenes extremely unlikely should it be approved of as a crop. Because there has been so much work in the area of spider silks, all the known silk protein genes are already largely protected by patents and we therefore decided it was important for us to establish our own IP position in the area if the project were to lead to the production of a valuable crop for UK agriculture and industry. To this end we proposed to clone a novel spider silk gene free from existing IP protection. Our choice of spider from which to clone the silk gene was carefully considered. Our colleague, Professor Fritz Vollrath at Oxford, had carried out a survey of the mechanical properties of silks from a range of different spiders and identified that the African Nursery Web Spider (Euprosthenops Sp.) produced a silk that was stronger, stiffer and less sensitive to water and other solvents than those of other spiders (Shao and Vollrath, 1999). Because of the superior quality of its dragline silk as well as the fact that at the start of the project, no genes had ever been cloned from Euprosthenops, we decided to focus on this species. We will describe the cloning and sequencing of a dragline silk protein gene from Euprosthenops and the adaptation of sequences of this gene for expression in plants.
Our original intention was to express this gene in plants under the control of a constitutive promoter through nuclear transformation. However, in June 2001 a German group published a paper detailing the expression of a previously characterised silk protein gene in tobacco and potato plants (Scheller et al., 2001). This group used methods to produce the recombinant proteins in plants that were essentially identical to our proposed approach. Whilst this publication clearly invalidated the novelty of our approach and thus severely impaired our potential IP position and thus value of our work, it also revealed to us the shortcomings of this approach. Whilst these researchers showed that the expression strategy certainly led to the production of recombinant silk proteins, the levels produce were not of sufficient magnitude to suggest that it would be cost effective. At around the same time some groundbreaking work was published that showed that extremely high levels of recombinant protein could be produced in transgenic tobacco (De Cosa et al., 2001) where the chloroplast genome (rather than the nuclear genome as is usually used) was the site of transformation and gene expression. Each cell in a plant leaf contains about 100 chloroplasts and each chloroplast contains many copies of its genome, whilst in contrast each cell typically contains two copies only of the nuclear genome. After discussing this with DEFRA officials it was agreed that a chloroplast expression strategy with our silk gene held more potential rather than following the original plan. We will describe the subcloning of the Euprosthenops silk gene into a chloroplast expression vector, the genetic transformation of tobacco cells and eventual regeneration of transgenic plants.