Gram-Negative Host

One o f the major considerations when deciding on an alternative host system for expression of a particular gene is the probability that the transcription and translation regulation regions o f the heterologous gene are recognised by the new host. This

concern increases the higher the degree o f diversity between the two systems. Thus, a gene from a Gram-positive organism is less likely to be correctly expressed in a Gram­ negative host than in another Gram-positive host. In general, it has been shown that genes from Gram-positive bacteria are more likely to be successfully expressed in a Gram-negative bacterium than vice versa. It is thought that the RNA polymerase enzymes o f Gram-positive bacteria are more stringent in their requirements for promoter recognition than those of E. coli. A barrier to gene expression also exists at the translational level. Thus Gram-positive genes, and in particular B. subtilis, have shown a requirement for a more extensive degree of homology between the mRNA Shine-Delgamo sequence and the 3' end of the 16s rRNA than that required by E. coli

(Murray and Rabinowitz, 1982). It is, however, known that the spa gene from S. aureus Cowan I is expressed in E. coli and produces a functional protein, as the initial detection of the cloned spa gene from an E. coli host relied upon the production of a protein exhibiting IgG binding activity (Duggleby and Jones, 1983). The levels of functional SpA produced in E. coli (0.2% total protein) were 10-fold less than those produced by the chromosomally located spa gene in S. aureus Cowan I, despite being located on pBR328 (a multicopy plasmid derived from pBR322). This may be caused by a number of factors, including the poor recognition of the spa transcription or translational regulatory sequences, or may result from the incorrect processing of the SpA polypeptide, yielding a high proportion of inactive SpA.

The use of E. coli for expression of heterologous proteins is popular because of the ease with which the genome can be manipulated. The wide range of E. coli host strains and plasmid vectors available allows a high degree of control over factors such as plasmid stability, the production of some proteases, plasmid copy number, promoter efficiency and regulation, translation initiation strength, selection o f marker for plasmid propagation and m aintenance, and a w ell characterised origin o f replication. Transformation of E. coli is highly efficient and relatively simple to perform, while other laboratory techniques used in the manipulation of recombinant DNA in E. coli are well documented (Maniatis et al„ 1982, Rodriguez and Tait, 1983 and Walker, 1984).

Therefore, E. coli may prove a highly suitable host for the generation of SpA in order to study the properties of the gene, as it allows the easy manipulation of the various structural and regulatory regions of that gene.

A problem that is likely to be encountered in using E. coli as a host for SpA production, is the final cellular location o f the heterologous protein. The cell wall of E. coli, a Gram-negative bacterium, is substantially different to that of a Gram-positive bacterium such as S. aureus, (Figure 1.2). In Gram-positive bacteria, proteins are located in the cytoplasm, in or on the cytoplasmic membrane or exported to the growth medium, whereas in Gram-negative bacteria, proteins can also be localised to the periplasmic space or to the inner or outer membrane. This difference in cell wall structure is reflected by the fact that E. coli in common with most enterobacteria excrete very few proteins to the exterior of the cell. Those that are fully exported are mainly toxins: (e.g., haemolysin, enterotoxins and bacteriocins such as colicin) and do not appear to rely on a common pathway for transport across the outer membrane. As is shown in the summary below, each of the above toxins utilises a different mode of exit from the cell. The heamolysin determinant consists o f four genes, hlyC, hlyA, hlyB and hlyD of which only hlyA codes for the heamolysin protein, with the export being dependent on the products of hlyB and hlyD. These helper proteins are believed to interact with the C-terminus o f the hlyA protein, allowing A TP-dependent transport through a transenvelope channel (Felmlee et at., 1985a and b; Mackman et al„ 1985). There are two E. coli enterotoxins. The heat stable enterotoxin is a monomer, and secretion is thought to be dependent on the small size (5 kDa) o f the molecule. The heat labile toxin is a multimeric complex composed of an A subunit and five B subunits. The A subunit requires the presence o f the B subunit for release (Yamamoto and Yokota, 1982), which is thought to be facilitated by the conditions of low pH, low oxygen and bile salts present in the intestine (Hirst et al., 1984). Colicins are encoded on naturally occurring plasmids, which generally encode a lysis protein required for colicin release.

a) Gram-positive cell envelope

b) Gram-negative cell envelope

Exported proteins

Figure 1.2 Schematic represen tat ion of the cell envelope of a typical Gram-negative and Gram-positi ve bacteria, showing the localisation of proteins

LPS: lipopolysaccharide chain. SP: surface proteins, consisting o f additional surface layers such as protein matrices and capsules; and surface appendages such as fimbriae, pili and flagellae. AD: adhesion zone: an alternative route for secretion of proteins via a transenvelope structure rather than through the periplasm (Bayer, el a!., 1982).

This type of release is not considered to be true secretion as it is nonspecific and causes envelope damage by activating a phospholipase, resulting in the release of free fatty acids and alteration in outer membrane permeability (Pugsley and Schwartz, 1984).

In contrast to the above examples, transport across the E. coli cytoplasmic membrane of most E. coli proteins follows the "general" secretion pathway involving the sec genes and the presence of a signal sequence at the N-terminus of the polypeptide, as described in Section 3.3.3. In general terms the periplasmic export of proteins in E. coli and the secretion to the growth medium of proteins by Gram-positive bacteria are considered an almost identical process, requiring similar transportation signals. Therefore, any signal sequence present in the SpA polypeptide which would normally allow secretion to the cell wall or growth medium in S. aureus, even if recognised in E. coli may only allow the transport o f the SpA polypeptide across the inner membrane to the periplasmic space. The final cellular location in E. coli of a heterologous protein normally found in the cell membrane or secreted to the growth medium o f a Gram-positive organism appears to vary according to the individual protein expressed. Thus the a-amylase gene from Bacillus coagulans encodes for an exoprotein in its natural host, however, when expressed in E. coli the protein is secreted to the periplasmic space (Cornelis et al.,

1982). In contrast, Sarvas and Palva (1983) reported the export of the penicillinase encoded by the Bacillus licheniformis penicillinase (penP) gene to the outer membrane fraction o f E. coli. Here it is present as an am phiphilic protein with properties indistinguishable from those o f the B.licheniformis membrane penicillinase, however,

E. coli appears to lack the proteolytic enzymes necessary to cleave the membrane- bound lipoprotein form of the penicillinase to give the soluble exopencillinase found in the natural host.

The cellular location of a heterologous protein produced by E. coli can dramatically affect the final levels of protein recovered from a cell culture, both quantitatively and qualitatively. It has been shown to have a direct affect on the ratio of active to inactive forms of the protein, the degradation rate, the ease of final recovery of the product and

finally on the growth rate and viability of the host organism itself. A protein that is normally located outside of the cytoplasm in its natural host, if retained in the cytoplasm when expressed in a heterologous host such as E. coli, will often have little or none of the normal enzymatic activity. This is thought to be due to the incorrect folding of the protein brought about by the reducing environment of the cytoplasm. A high degradation rate is often seen for non-cytoplasmic proteins retained in the cytoplasm. This may be caused by a combination o f two factors, the heterologous protein may not possess N- and C-terminal sequences resistant to the action of cytoplasmic proteases and the fact that there are a greater number of proteolytic enzymes present in the cytoplasm than in the rest of the cell. Eight separate soluble proteolytic acitvities have been reported in E. coli (Swamy and Goldberg, 1981). Five were located in the cytoplasm, two in the periplasm and one distributed evenly between the two compartments. Talmadge and Gilbert, (1982) have shown dissimilar rates of degradation for a hybrid proinsulin molecule when located in different cellular compartments. When secreted to the periplasm, the hybrid protein exhibited a 10-fold longer survival time than that remaining in the cytoplasm. Synthesis of high levels of some recombinant proteins in E. coli have resulted in the production of intracellular inclusion bodies (K lier et al., 1982; W eis et a l., 1983; W illiams et a!., 1982; Schoemaker, et al. 1985). The sequestering of the heterologous protein into dense insoluble inclusion bodies has been reported to protect a high percentage of the protein against proteolysis (Weis et al., 1983; Kleid et al., 1981; Cheng et al., 1981). Inclusion bodies can be advantageous in increasing the level of recovery of the protein by making the isolation and purification stages easier. However although some inclusion bodies are easily solubilised to yield large amounts of conformationally correct protein that retain their enzymatic activity, it is more common for the inclusion bodies to be composed of aggregates of the protein which are very difficult to solubilise and purify. Schoemaker e t al. (1985) showed that in the case of recombinant prochymosin, aggregation was caused by the interlinkage of intermolecular disulphide bonds to form protein multimers. Though disulphide bonds are usually unable to form in the

cytoplasm, it is suggested that they may be able to exist in an inclusion body, as the molecules are very closely packed and are therefore more likely to react with one another and as such are protected from the reducing environment of the cytoplasm. The presence o f cysteine residues, and thus the potential for disulphide linkages, is a common feature in proteins seen to accumulate as inclusion bodies when present in high concentrations.

Secretion to the periplasm or outer membrane of E. coli would appear the preferable location for non-cytoplasmic heterologous proteins, in terms of reducing degradation and en co u rag in g the correct po st translational processing and fin al protein conformation while avoiding the formation of inclusion bodies. However, Emr et al.

(1978; 1980) have reported that at high levels of protein production, secretion of the protein to the periplasm can be highly toxic to the cell, resulting in either reduced growth or cell lysis. In these studies cell lysis was brought about by the production of large amounts of a lamB -lacZ hybrid protein and the cells inability to efficiently export the hybrid protein to the periplasm, resulting in a lethal jamming of the normal cellular protein export machinery. Similar results were obtained by Rose and Shafferman (1981) when expressing the vesicular stomatitis glycoprotein. In this case cellular lethality was prevented by removing the hydrophobic core region o f the signal sequence thereby almost certainly preventing its insertion into the cytoplasmic membrane. Brosius, (1984) revealed similar results when expressing a hybrid rat insulin gene, but also reported detrimental effects to the cell of a non-excreted form of the rat insulin gene when fused to an extremely strong prom oter (rrnB). These latter observations indicate that other, more specific, effects may contribute to cellular lethality in individual gene cases. In such circumstances the formation of cytoplasmic inclusion bodies may serve to protect the host cell against the toxic effects of overproduction (Weis et al. 1983).

Despite there being problems of poor efficiency, poor yields and toxicity to cells when expressing a heterologous protein in E. coli these are only potential problems and might

not arise in the case of SpA. Even if one or more were found to be rate limiting factors, the versatility that is possible using different E. coli hosts and plasmid systems would offer a greater chance of dealing with such a problem, than if it was found to occur in the less well understood Gram-positive systems that are currently available. Therefore

E. coli was selected as the preliminary host for the development of a high expression system for SpA.