Development of an expression system in Aspegiffus niduhns

338

BIOCHEMICAL SOCIETY TRANSACTIONS

Development of an expression system in Aspegiffusniduhns DAVID I. GWYNNE, FRANK P. BUXTON, SHIRLEY A. WILLIAMS, A. MICHAEL SILLS, JAMES A. JOHNSTONE, JAGDEEP K. BUCH, ZI-MIN GUO, DEBORAH DRAKE, MARGIT WESTPHAL and R. WAYNE DAVIES Allelix Inc., 6850 Goreway Drive, Mississauga, Ontario, L6Z lB5, Canada

provided for the DraI vector. Any final constructs which do not contain a direct fusion of the secretion signal sequence to the mature coding region are modified using oligonucleotidedirected mutagenesis. All vectors utilize the optimal sequences for translation initiation which have been previously described [6].

Introduction

(a)

Many human proteins with the potential to be used as therapeutics are secreted. These include growth factors, cytokines and protein hormones. The eukaryotic secretion process often involves a range of post-translational modifications which are required for correct tertiary structure and which have direct implications for biological activity as well as serum half-life. These modifications can include proteolytic processing as well as correct disulphide-bond formation and glycosylation. Since traditional prokaryotic intracellular expression systems often produce a product with aberrant disulphide bonds, as well as other changes such as a modified N-terminus, several groups interested in the production of recombinant human proteins have focused on the development of secretion systems. Fungi, in particular those of the genus Aspergillus,have been well exploited for their ability to produce high levels of extracellular proteins in an inexpensive manner [ 11. The recent developments in DNA-mediated transformation systems [2-51 have allowed the utilization of this genus as a host for the production of recombinant proteins of therapeutic interest [6, 71. Several features of Aspergillus nidulans make this an attractive organism for the expression of foreign proteins. Excellent genetics makes a wide range of defined host strains available. Recombinant A . nidulans strains contain integrated expression cassettes, often in multiple copy which are mitotically stable. Thus these strains have the advantages of the high rate of transcription associated with multiple copies as well as the ability to be grown for many generations in the absence of selection. Here we present the results of the expression of several secreted proteins utilizing an expression vector derived from the alcA gene of A . nidulans.

Hindlll

pTAWS 4742

Expression vectors

The expression vectors utilize the promoter and transcriptional regulatory sequences derived from the alcA gene of A . nidulans which encodes alcohol dehydrogenase I. It is transcribed very efficiently [8, 91 and is inducible with a wide variety of compounds. AlcA along with aldA (encoding aldehyde dehydrogenase) is under the control of a positive regulatory protein encoded by the alcR gene. The transcription of both alcA and aldA is repressed in the presence of glucose. This is mediated via the creA gene in A . nidulans. The secretion vectors also contain a synthetic secretion signal sequence [ 101.The sequence was designed to encode a short peptide with two N-terminal positively charged residues, a hydrophobic core and an optimized cleavage site as described by Von Heijne [ 101. A multiple cloning site is positioned downstream of the secretion signal sequence to allow insertion of a variety of restriction fragments. Secretion vector pALCAlS is shown in Fig. 1. A derivative expression vector pTAWTS has a DraI site or a Nael site (for blunt fragment cloning) at the 3’-end of the secretion signal sequence; one example is shown in Fig. 1. When the appropriate 5’terminus is supplied in the form ofa synthetic fragment or an Fspl-cleaved site, then a direct fusion between the end of the secretion signal and the mature sequence coding regions is

pALCA1S 4742

EcoRl

Fig. 1. Restriction maps of the alcA-based expression vectors Hatched area represents the 5’ regulatory region and promoter of the alcA gene. The signal peptide sequence is indicated by an arrow. 1989

339

627th MEETING, NOTTINGHAM Expression hosts A derivative of A. nidulans FGSC4 (T580)is utilized as an expression host for the alcA-based vectors. This recombinant strain contains multiple integrated copies of the alcR regulatory gene whose gene product is a positive transcriptional regulatory factor for the aIcA gene. The multiple copy strain T580 over-produces the transcript encoding the alcR gene product (data not shown) and probably also overproduces alcR protein. Since the concentration of alcR gene product seems to be somewhat limiting in a wild-type cell [ 1 11, multiple copies of the expression cassette integrated into the genome of the wild-type host show low transcriptional efficiency. Increased transcriptional efficiency in those strains with the multiple expression cassette can be observed in the T580 host [ 1 11. Expression offoreign genes in the alcA system It has previously been reported that biologically active human a 2-interferon and Cellulomonas fimi endoglucanase can be secreted from recombinant strains of A. nidulans T580 [6]. Recently, a gene encoding human epidermal growth factor ( E G F ) (synthetic)and a gene encoding Aspergillus niger glucoamylase have been expressed and secreted in the system. A direct fusion of human EGF to the synthetic secretion signal has produced strains which secrete human EGF at levels of 1-2 mg/l in minimal medium. Mature human EGF contains three disulphide bonds which are required for correct conformation and biological activity. The Aspergillus-secreted product seems to be correctly folded. Levels of antigenically detectable material (radioimmunoassay) are equivalent to the levels which are detectable by a receptor-binding assay. Since the receptor-binding assay measures only correctly folded product, virtually all of the secreted material is correctly folded. Biologically active glucoamylase was also secreted from the recombinant host. Optimization

The success or failure of any industrial recombinant fermentation depends primarily on three things: the accumulation of a suitably large cell mass; the ability of that cell mass to produce maximal amounts of the required protein in a stable form, and in many cases, the ability of the cell to secrete the required protein. The A. niditlans alcA expression system represents an efficient way of achieving such an industrial recombinant system. The promoter is very efficient and is inducible in the absence of glucose (afcA expression is glucose repressed). This allows the construction of strains which can be grown in the presence of glucose to accumulate biomass and which can then be induced to initiate the production (and secretion) of recombinant protein. The alcA glucoamylase strains were chosen to develop the fermentation/induction protocol due to the easily assayable nature of the glucoamylase enzyme. A number of parameters were studied for the optimization of biomass accumulation, inducers and induction regimes, and the effect of fungal morphology on the secretion of proteins. As a result of this programme, a scalable system for producing recombinant proteins has been developed for A. nidulans. As seen in Fig. 2, biomass is generated in the first 30 h of fermentation in the presence of complex medium containing glucose as a carbon source. Upon exhaustion of glucose, an inducer supplemented with a carbon and energy source is added to the medium; after a 10- 15 h lag period, the recombinant protein is secreted. Fig. 2 shows the production of glucoamylase under the conditions described. About 14 g of biomass/l was generated and glucoamylase was secreted at a level of 1.2 g/I (based on the specific activity of glucoamylase). Vol. 17

\ 0

8

16 24

32 40

48 56 64

Time (h)

Fig. 2. Production of recombinant glucoamylase by A. nidulans ( + ) Concentrations of glucose in the growth medium (g/l). ( x ) Biomass (g/l). (*) Glucoamylase activity (units/l). ( # ) Concentration of total secreted protein (g/l). Inducer added at 28 h.

Currently, the growth-induction regime is being applied to the production of recombinant EGF. Molar quantities of EGF, which are equivalent to previously expressed levels of glucoamylase, should be in the 50 mg/l range and preliminary optimization data indicate that these levels should be easily attainable. Discussion Filamentous fungi of the genus A. nidulans are able to secrete large quantities of certain proteins and thus have become major sources of industrial enzymes. We have demonstrated that recombinant strains of A. nidulans can express and secrete several foreign proteins in a biologically active form. The use of secretion for high-level expression potentially avoids the intracellular accumulation of insoluble protein and allows advantage to be taken of post-translational modifications. Some of the data discussed here are derived from strains grown on minimal medium. A great number of improvements are currently being made in the yield of other recombinant proteins by utilizing some of the optimization procedures outlined above. In one particular case. expression levels have risen from 5- 10 mg/l in minimal medium to the g/l range. Clearly, each recombinant protein will have a different behaviour in the secretion pathway as well as different stabilities once in the media. It remains to be seen whether alteration of these characteristics is possible and compatible with utility. Progress is being made in other areas of optimization, such as the use of different host strains. Examples of these include mutants with low proteinase levels, as well as morphological mutants, which may provide improved yields of product. Given that large-scale fermentation techniques are well established for organisms closely related to A. nidulans, we are optimistic that the aIcA vectors will form the basis of a tractable expression system which will allow the expression of a wide variety of pharmaceutically important proteins. Barbesgaard, P. (1977) in Generics and Physiology of Aspergillus (Smith, J . E. & Pateman, J. A.. eds.), Academic Press, London Buxton, F. P., Gwynne, D. I. & Davies, R. W. (1985) Gene 37, 207-214 Tilburn, J.. Scazzocchio, C., Taylor, G. G., Zabicky-Zissman, J . H., Lockington, R. A. & Davies, R. W. (1983) Gene 26, 1370-1374 Yelton. M. M., Hamer, J . E. & Timberlake, W. E. ( 1 984) Proc. Natl. Acad. Sci. U.S.A.81, 1370-1374

340

BlOCHEMlCAL SOCIETY TRANSACTlONS

5 . Ballance, D. J., Buxton, F. P. & Turner, G. (1983) Biochern. Biophys. Res. Commun. 112,284-289 6. Gwynne, D. I., Buxton, F., Williams, S. A., Garven, S. & Davies, R. N. ( 1987)Biotechnology S,7 13-7 19 7. Upshall, A., Kuner, A., Bailey, M. C., Parker, M. D., Favreau, M. A., Lewison, K. P., Joseph, M. L., Maragonone, J. M. & McKnight, G. L. ( 1987) Biotechnology 5, 1301- I304 8. Lockington, R. A., Sealy-Lewis, H. M., Scazzocchio, C. &

Davies, R. W. (1986) Gene 33,137-149 9. Gwynne, D. I., Buxton, F. P., Sibley, S., Davies, R. W., Lock-

ington, R. A., Scazzocchio, C. & Sealy-Lewis, H. M. (1987) Gene 51,205-216 10. Von Heijne, G. (1983) Eur. J. Biochem. 133, 17-21 11. Gwynne, D. I., Buxton, F. P., Gleeson, M. A. & Davies. R. W. (1987) in Prorein Purification: Micro to Macro (Burgess, R., ed.),pp. 355-365. Alan R. Liss, New York

Received 23 June 1988

Secretion of recombinant proteins into the culture medium by Escherichia coli and Staphylococcus aureus + + +

+ +

-

MATHIAS U H m N and LARS ABRAHMSGN Department of Biochemistry and Biotechnology, The Royal Institute of TechnoloD, S-100 44 Stockholm, Sweden

-36

Synopsis The ability of staphylococcal protein A (SPA) to bind to the Fc part of IgG has been used for the purification of a number of heterologous gene products as fusion proteins. Both the SPA promoter and signal sequence function in Escherichia coli, as well as in a number of Gram-positive bacteria, which facilitates comparisons of the expressed specific products in different hosts. The expression system developed for E. coli yields excretion of the fusion protein to the growth medium, which makes E. coli a competitive alternative to Gram-positive bacteria for the expression of secreted products. The human peptide hormones insulinlike growth factors (IGF) I and I1 were expressed using the protein A system in E. coli and Staphylococcus aureus. Despite a high degree of structural homology, large differences in the yields were observed in the two hosts. This underlines the importance of investigating different bacterial hosts for a particular protein product.

-1 0 SerValThrLeuGl yThrLeuLeulleSerGl yG1y V a l T h r P r o A l a A 1a hydrophobic b

Introduction SPA is a cell-wall associated protein of S. aureus having a number of immunological properties [l].The structure is remarkably repetitive with five highly homologous IgG-binding domains, of about 58 amino acids each, followed by the cell-wall anchoring part, region X [2]. The first part of region X is composed of an octapeptide repeated 12 times. The Cterminal part is a membrane anchor sequence of 20 hydrophobic residues followed by three arginines [21. In this paper we describe a strategy for analysing the signals for secretion in SPA. We also analyse the unexpected observation that small fragments, subcloned parts, of SPA are translocated to the culture medium E. coli. It is shown that this phenomenon may be used for the export of heterologous gene products to the growth medium. Confirmation of the proposed signal peptide The nucleotide sequence of the cloned gene revealed a putative signal peptide of 36 amino acids (Fig. 1).The overall composition of the signal peptide is the same as for known Gram-negative counterparts [31, except for the length, most pronouncedly of the charged N-terminal part of eleven residues. To investigate the signal peptidase cleavage of SPA Abbreviations used: SPA, staphylococcal protein A; IGF, insulinlike growth factor.

-30

-20

MetLysLysLysAsnlleTyrSerlleArgLysLeuGlyValGlylleAla -basic

A

-1 +I +10 A s n A l a A1aG1nHi sAspGlu A1aGlnGl n A s nA1aPheTyrGl n V a l Leu

-unique-

-consensus-

Fig. 1. Structure of the SPA signal sequence The deduced amino acid sequence [2] and relevant characteristics are shown. Unique and consensus indicate the part of the region E not homologous and homologous, respectively, to the other IgG-binding regions [3]. derivatives in different recombinant hosts, a plasmid, pSPAl6 [4], containing a truncated SPA gene was introduced into E. coli, Bacillus subtilis, S. aureus and Staphylococcus xylosus and the resulting gene products N-terminally sequenced by Edman degradation [ 3 ] .The results indicate a common processing site after the proposed signal peptide (Fig. 1). When a similar construct was introduced into Streptomyces lividans, the same processing site was observed (G. Emis, personal communication). This processing site follows the rules by von Heijne for signal peptidase cleavage [ 51. These results demonstrate that the SPA signal sequence is functional in a large variety of bacterial hosts and that the mature protein obtained after signal peptidase cleavage is independent of the host.

Evaluation of sequences relevant for secretion to the culture medium by E. coli The five IgG-binding domains of SPA are about 80% homologous on the protein level [6]. The most striking divergence is in the first domain, domain E, in which the Nterminal part differs considerably from the other domains. To evaluate the sequences relevant for secretion in E. coli, different SPA fragments were expressed in E. coli using recombinant DNA technology and the localization of the products was determined 13, 71. The results summarized in Table 1, demonstrate that the first part of region E seems to be adapted to secretion since this N-terminal domain could not be substituted by a homologous domain (region B) without interfering with the secretion mechanism. Unexpectedly, some of the smaller subcloned fragments of SPA (E, EB and EE) were found to be exported to the growth medium when expressed in E. coli (Table 1). The 1989