- Email: [email protected]
Isolation and characterisation of a novel stress-inducible PDI-family gene from Aspergillus niger
Gene 193 (1997) 151–156
Isolation and characterisation of a novel stress-inducible PDI-family gene from Aspergillus niger D.J. Jeenes a,*, R. Pfaller b, D.B. Archer a a Department of Genetics and Microbiology, Institute of Food Research, Norwich Research Park, Colney, Norwich, NR4 7UA, UK b Consortium fu¨r elektrochemische Industrie GmbH, Zielstattstr. 20, 81379 Mu¨nchen, Germany Received 16 September 1996; received in revised form 12 December 1996; accepted 8 January 1997; Received by B. Dujon
Abstract Current strategies to improve the secretion of heterologous proteins in Aspergillus niger include the manipulation of chaperones and foldases specific to the endoplasmic reticulum ( ER). A family of ER-specific protein s which share active-site homology wit protein disulfide isomerase (PDI ) has been identified from other systems, many of which are inducible by agents which cause malfolding of proteins in the ER. Here we report identification of tigA from Aspergillus niger and erp38 from Neurospora crassa, two novel members of the PDI superfamily of proteins. TIGA and ERp38 show 66% identity at the amino acid level and are putative ER proteins. Both proteins show tandemly linked thiol-oxidoreductase domains followed by a functionally uncharacterised C-terminal domain. The most distal active site in TIGA is created by excision of a 66-bp intron. Although no Unfolded Protein Response elements can be seen in the tigA promoter, sequence homology has identified associated with protein trafficking (ERPTRE ) in a gene encoding the related mammalian protein, ERp72, as well as a second motif conserved amongst the glucoserelated protein family. Southern and dot blot analysis indicate that the tigA gene is present in single copy. Both the A. niger and N. crassa proteins show homology with a stress-inducible alfalfa, G1. Transcription of tigA is induced 2–3-fold after treatment with tunicamycin, an inhibitor of N-linked glycosylation. Strains overexpressing a heterologous protein show no increased tigA mRNA levels. © 1997 Elsevier Science B.V. Keywords: Filamentous fungi; Heterologous expression; Secretion; Chaperones; Tunicamycin
1. Introduction The filamentous fungus, Aspergillus niger, secretes high levels of some native proteins making it an attractive host for the production of recombinant proteins. Secreted yields of heterologous proteins from A. niger can be disappointingly low, however. Previous work has shown that post-transcriptional events are primarily responsible for the low secreted yields of heterologous proteins (Jeenes et al., 1994) and several strategies, such as the use of translational fusions or protease-deficient * Corresponding author. Tel.: +44 1603 255255; Fax: +44 1603 507723; e-mail: [email protected] Abbreviations: aa, amino acid(s); bp, base pair(s); CHO, chinese hamster ovary; ER, endoplasmic reticulum; ERPTRE, endoplasmic reticulum protein trafficking response element; kb, kilobase(s) or 1000 bp; ORF, open reading frame; PCR, polymerase chain reaction; PDI, protein disulfide isomerase; pfu, plaque forming units; TIGA, tunicamycin inducible gene A polypeptide; tsp, transcriptional start point; *, UAG stop codon. 0378-1119/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S 03 7 8 -1 1 1 9 ( 9 7 ) 0 0 0 98 - X
hosts, have been developed to overcome this problem (MacKenzie et al., 1993; Archer et al., 1994). Difficulties encountered during passage of proteins from the endoplasmic reticulum ( ER) to the Golgi are a major factor influencing secreted yields. During this step, protein folding is modulated by a battery of ER-specific proteins including protein disulfide isomerase (PDI ). Although the in vivo role of ER-specific chaperones and foldases such as PDI is poorly understood, overexpression of PDI in Saccharomyces cerevisiae can increase secreted yields of some heterologous proteins by between 10and 24-fold (Schultz et al., 1994; Robinson et al., 1994). A less dramatic increase (2-fold ) for production of some Fab∞ antibody constructs from strains of Escherichia coli overexpressing human PDI has also been shown, although yields remained low (Humphreys et al., 1996). A good correlation between PDI levels and the amount of secreted protein from a number of cell types has also been reported ( Freedman, 1984). In vitro experiments show that foldases can often act synergistically to increase both the rate and yield of folded end product
152
D.J. Jeenes et al. / Gene 193 (1997) 151–156
stress inducible, specifically by agents which cause malfolding of proteins in the ER, and show some functional overlap in vivo ( Tachibana and Stevens, 1992; Tachikawa et al., 1995). Our current aim is to dissect the roles of this protein family in the secretory pathway as part of a programme to optimise expression and secretion of recombinant proteins in A. niger. In this paper we report the isolation and characterisation of a novel stress-inducible PDI-family gene, tigA, from A. niger and the cDNA sequence of a homologous gene, erp38, from Neurospora crassa.
2. Experimental and discussion 2.1. Isolation and characterisation of tigA
Fig. 1. Nucleotide and deduced aa sequence of the A. niger tigA gene. Non-translated and intron sequences are in lower case; translated sequences are in upper case. Putative TATA- and CT-boxes are underlined. A putative ERPTRE sequence is overlined. The translational stop codon is represented by an asterisk. The polyadenylation site is underlined with a dashed line. The PDI active-site motifs are boxed and the ER-retention signal indicated by a double underline. An arrowhead refers to the probable signal sequence cleavage site. Degenerate oligonucleotide mixes used to isolate PCR fragments for cloning and partial sequencing were: 5∞-CC(CG)TGGTG( TC )GG( TC )CA(CT )TG-3∞ and 5∞-CTGAA(CT )TA(CG)AG(CT )TC(AG)TC(TC )TT-3∞. The same mixes were used to amplify a cloned PCR fragment which showed PDI homology for probing against the A. niger N402 lZAPII genomic library. 50×103 pfu were transferred onto Hybond-N+ membrane using standard protocols (Sambrook et al., 1989). Probes were labelled with 32P-dATP using the random primer Megaprime kit (Amersham, UK ) according to the manufacturer’s instructions. High stringency (65°C; 0.1×SSC ) hybridisation conditions were used. Plaques were cored and purified using a secondary hybridisation screen before plasmid rescue according to the protocol supplied (Stratagene, USA). cDNA clones were isolated in identical fashion from a lZAP cDNA library of a derivative of A. niger N402 in which the glaA gene had been deleted (kindly supplied by Dr. C.A.M.J.J. van den Hondel ). Dye terminator cycle sequencing mixes (Perkin Elmer, USA) were used with DNA purified through a Qiagen ( UK ) column to sequence the DNA using protocols supplied by the manufacturers. The nucleotide sequence data reported in this paper will appear in the EMBL, GenBank and DDBJ Nucleotide Sequence Databases under the accession No. X98748.
(Rupp et al., 1994). In addition, a family of ER-specific proteins which share active-site homology with PDI is now emerging from work on a variety of systems (Freedman et al., 1994; Hayano and Kikuchi, 1995; Tachikawa et al., 1995). Several of these proteins are
Degenerate oligonucleotide primer mixes designed against the –PWCGHC– active site region of PDI and a KDEL ER-retention signal were used to generate and clone specific A. niger DNA fragments by polymerase chain reaction (PCR). One of three fragments sequenced showed strong identity to PDI sequences from a number of sources and was used to probe a lZAPII genomic library of A. niger strain N402. The majority of purified pBluescript constructs rescued shared common restriction fragments and the clone containing the largest (5-kb) insert was chosen for further study. Sequencing of the entire insert revealed one ORF, designated tigA (tunicamycin inducible gene), potentially encoding a peptide of 359 aa residues (38.7 kDa) and containing one putative 66-bp intron (Fig. 1). Potential lariat, 5∞ and 3∞ splice site signals conform well to intron excision sequences found in filamentous fungi (Gurr et al., 1987). Attempts to map the transcriptional start point using primer extension were unsuccessful; however, RT-PCR using primers which anneal between −78 to −59 (relative to the ATG initiator codon) or −133 to −114 indicate transcriptional initiation occurs between these two priming sites (data not shown). Analysis of the upstream sequence shows both TATA- and CT-boxes, motifs which are important for the transcription of many fungal promoters (Gurr et al., 1987). Several ER-specific genes contain regulatory elements which respond to the presence of unfolded proteins or agents that cause malfolding of proteins. Although no convincing Unfolded Protein Response or heat shock elements ( Kohno et al., 1993; Bush et al., 1994) could be seen in the promoter region of tigA, a 23-bp element with 70% identity to a region of the 82-bp ER Protein Traffic Response Element ( ERPTRE) found in the mammalian ERp72 gene (Srinivasan et al., 1993) was detected ( Fig. 2a). This element confers inducibility to both the Ca2+ ionophore A23187 and expression of incompletely assembled secreted forms of Ig m heavy chain in COS cells. An element with comparable identity to the same
D.J. Jeenes et al. / Gene 193 (1997) 151–156
153
2.2. tigA and erp38 ORF analysis
Fig. 2. (a) Shared homology between tigA promoter sequences and the ERp72 ERPTRE motif and (b) between tigA promoter sequences and the grp-core motif from the rat grp78 gene. Co-ordinates are numbered relative to the translational initiation codon. (c) Schematic structure of the TIGA protein: grey area, ER translocation signal (aa 1–19); open area, ORF (aa 20–359), filled area, –CGHC– motifs (aa 49–52 and 169–172); hatched area, KDEL ER-retention signal, —, intron (aa 171).
region of the ERPTRE of murine ERp72 is also seen in the A. niger pdiA gene encoding another ER-specific PDI-family gene (Ngiam et al., 1997). Both of these A. niger promoter elements are sited a similar distance upstream of the initiator ATG codon as the ERp72 motif which might suggest a regulatory function in vivo and a role in protein trafficking for the tigA gene product. A second element of 19 bp with 75% identity to part of the ‘grp core’ element conserved amongst promoters of the glucose regulated protein (grp) gene family is also present in the tigA promoter region (Fig. 2b). Interestingly, the boundaries of the homology agree almost precisely with that part of the grp core motif responsible for the elevation of basal, as opposed to inducible, levels of transcription in promoters of the human and rat grp78 genes (Li et al., 1994). This motif is a similar distance from the initiator codon in all three genes which supports an in vivo role for this element in tigA regulation. Partial cDNA clones of tigA were isolated from an A. niger N402 cDNA library to identify intron boundaries; the longest of these contained the entire ORF except for the first base of the initiator ATG codon. Sequence analysis of two independent cDNA clones confirmed the intron splice sites and defined the polyadenylation site ca. 115 bp downstream of the stop codon (Fig. 1). A schematic representation of the protein structure and intron location is shown in Fig. 2c. Southern blot analysis indicates that tigA is present as a single copy and dot blot analysis of serial DNA dilutions against a single gene copy control supports this conclusion (data not shown). Northern blot analysis confirms the predicted transcript size of approx. 1.35 kb (data not shown).
Sequence analysis using the programme SIGCLEAVE identified the most probable ER-translocation signal cleavage site (score 10.5) between aa residues 19 and 20 ( Fig. 1; von Heijne, 1986). The 340-aa mature protein contains a –KDEL carboxy-terminal motif which is likely to serve as an ER-retention signal and supports the idea that TIGA functions in the ER lumen. Interestingly, A. niger employs a different ER-retention signal (–HDEL) for the PDI protein (Malpricht et al., 1996; Ngiam et al., 1997). Binding of HDEL and KDEL ligands is sensitive to pH and shows different affinities for their common receptor in vitro ( Wilson et al., 1993) suggesting that TIGA and PDI may operate at different sites within the ER to cis-Golgi network. Two –CGHC– active-site motifs characteristic of PDI-family genes ( Freedman et al., 1994) are also present at an interval of 115 aa (Fig. 2c). The distal –CGHC– motif is created by intron excision, unique amongst those PDI-family genes reported to date, but whether this has functional importance in vivo is unknown. Spacing of these motifs at either ca. 115 aa apart or 330 aa apart is a highly conserved feature of the PDI protein family ( Fig. 3). A full length cDNA clone encoding a protein, ERp38, has also been isolated from a lgt11 library of Neurospora crassa cDNA by immunoscreening ( Fig. 4) and is a strong candidate for the N. crassa homologue of tigA with 79% similarity and 66% identity at the amino acid level (Fig. 5). Analysis of the erp38 ORF shows a similar spacing of the active sites to that of tigA, a putative N-terminal signal sequence of 18 aa (score 6.6) and a –KEEL ER retention signal. These two sequences provide evidence for a novel sub-family of ER-localised proteins in which the two thiol-oxidoreductase domains,
Fig. 3. The PDI-family of genes in the ER. References to the different proteins can be found in the Introduction (Section 1).
154
D.J. Jeenes et al. / Gene 193 (1997) 151–156
Fig. 4. Nucleotide and deduced aa sequence of the N. crassa erp38 cDNA. The nucleotide sequence data will appear in the EMBL, Genbank and DDBJ Nucleotide Sequence Databases under the accession No. Y07562. PDI active-site motifs are boxed and the ER-retention signal indicated by a double underline. An arrowhead refers to the probable signal sequence cleavage site. Immunoscreening of a lgt11 library of N. crassa cDNA by standard protocols (Sambrook et al., 1989) used an antibody directed against gel-purified N. crassa 38-kDa proteins isolated using methods previously reported (So¨llner et al., 1989).
90–100 aa in length and centred around the active sites, are tandemly linked without an intervening sequence. This is in contrast to larger proteins in the PDI family such as PDI, EUG1, ERp61 etc. which contain a ca. 200-aa region of unknown function between the two thiol oxidoreductase domains ( Fig. 3). Database searches have also identified homology between TIGA (58% similarity, 43% identity) or ERp38 (61% similarity, 45% identity) and the product of a stress-inducible gene, G1, from alfalfa (Shorrosh and Dixon, 1992). Although the alfalfa protein does contain an ER-translocation signal, it lacks an ER-retention signal suggesting either that it may be targetted to a different subcellular location in this host or is retained as part of a heteromeric complex by interaction with subunits which do contain such a signal. Strong identity also extends beyond the two linked thiol-oxidoreductase domains into a so far functionally uncharacterised C-terminal domain ( Fig. 5). The similarity of the residues surrounding the active sites in TIGA, ERp38 and G1p to those found in PDI suggest a strongly oxidising oxidoreductase activity whilst the lack of a peptide-binding site locally rich in acidic aa argues against high isomerase activity (Noiva et al., 1993). Taken together, the peptide M , r active-site spacing and active-site composition support an isofunctional role, distinct from that for PDI, for all three peptides in protein trafficking in their respective hosts. It is also of interest to note that all three peptides are lysine-rich (>10%) although what significance this has is unclear. Functional analysis of the tigA gene product is currently in progress.
Fig. 5. Sequence comparison between the products of the tigA, ERp38 and G1 genes using PileUp from the University of Wisconsin suite of programmes. Identical residues are shaded in black, similar residues are shaded in grey, non-homologous residues are white. Similarity is defined using a reference matrix file compiled from multiple sequence alignments which considers both physicochemical similarity and the frequency of mutation between one amino acid and another.
D.J. Jeenes et al. / Gene 193 (1997) 151–156
2.3. Transcriptional regulation of tigA Like many other ER-specific proteins, several PDIfamily proteins are inducible by agents which perturb or disrupt ER function, e.g. tunicamycin, an inhibitor of N-linked glycosylation (Dorner et al., 1990; Shorrosh and Dixon, 1992; Tachibana and Stevens, 1992). A comparison of tigA expression in the presence and absence of tunicamycin shows that tigA mRNA is induced 2–3-fold after a time lag of 3 h ( Fig. 6). This relatively low level of induction agrees with the level of induction reported for PDI mRNA (3–4-fold ) but not ERp72 mRNA (10-fold ) following tunicamycin treatment of CHO cells (Dorner et al., 1990) or EUG1 mRNA (10-fold ) in yeast cells ( Tachibana and Stevens, 1992). It is also in contrast to that reported for the alfalfa G1 gene which, although not quantified, was strongly induced (Shorrosh and Dixon, 1992). It is possible that optimal induction conditions would require removal of glucose from the cultures tested as some PDI-family genes are induced under conditions of glucose starvation. Levels of actin mRNA also appear to
155
be reduced relative to the earlier time points (Fig. 6). The reason for this is unclear although dry weights indicate that the A. niger culture is still actively growing and carbon source is still detectable (data not shown). The 3–4-h lag period before tigA induction mimics that reported for the alfalfa G1 gene and suggests that tigA expression does not form part of the primary stress response. Preliminary data examining the expression of tigA in A. niger transformants overexpressing heterologous proteins show no increase in tigA mRNA (data not shown). The precise role of TIGA and other PDIfamily proteins such as ERp72, PDIR, MPD1 and P5 ( Fig. 3) in vivo remains unknown. Experiments to study phenotypic changes associated with changes in tigA expression are currently in progress.
3. Conclusions (1) The tigA gene from A. niger and erp38 cDNA from N. crassa have been cloned by PCR-based methods and antibody screening respectively. (2) TIGA and ERp38 are homologues and encode a novel potentially ER-localised member of the PDI superfamily of proteins. (3) The tigA gene is inducible by tunicamycin and TIGA, like ERp38, shows homology to a stressinducible gene product from alfalfa whose deduced amino acid sequence lacks an ER-retention signal. (4) Sequence homology analysis has identified a promoter element in tigA associated with protein trafficking in a related mammalian protein, ERp72, and a second promoter element conserved amongst chaperone proteins of the glucose-regulated protein family.
Acknowledgement
Fig. 6. Northern blot of RNA from tunicamycin-induced cultures of A. niger AB4.1. Cultures were grown on ACM/N/P as previously described (Jeenes et al., 1994) except that 1% glucose was used for carbon source. Tunicamycin (10 mg/ml ) was added at t=50 h; cultures were then harvested at 1 h, 2 h, 3 h and 4 h intervals after tunicamycin addition, the mycelium ground under liquid N and freeze dried and 2 their RNA extracted using an RNeasy extraction kit for fungi (Qiagen, UK ). 15 mg total RNA for each sample was electrophoresed through a 0.8% agarose gel containing 2.2 M formaldehyde at <3 V/cm until the bromophenol blue had run ca. 8 cm. The gel was washed through 5 changes of DEPC-treated water before transfer onto Hybond-N+ (Amersham, UK ) by standard methods (Sambrook et al., 1989). Hybridisation conditions, probe preparation and labelling were as described for library screening in the legend to Fig. 1. Induction levels represent tigA mRNA levels normalised against actin mRNA quantified using a Fuji BAS-1500 phosphorimager.
This work was partly funded by an EC Biotechnology programme grant, BIO2 CT-942045 (D.J.J. and D.B.A) and by the Sonderforschungsbereich 184 (project B1; R.P.) We thank M. Kiebler for experimental advice and greatly appreciate the support from Walter Neupert at the Institut fu¨r Physiologische Chemie, Mu¨nchen, during the course of the ERp38 cDNA isolation.
References Archer, D.B., Jeenes, D.J., MacKenzie, D.A., 1994. Strategies for improving heterologous protein production from filamentous fungi. Antonie van Leeuwenhoek 65, 245–250. Bush, K.T., Hendrickson, B.A., Nigam, S.K., 1994. Induction of the FK506-binding protein, FKBP13, under conditions which misfold proteins in the endoplasmic reticulum. Biochem. J. 303, 705–708. Dorner, A.J., Wasley, L.C., Raney, P., Haugejorden, S., Green, M.,
156
D.J. Jeenes et al. / Gene 193 (1997) 151–156
Kaufman, R.J., 1990. The stress response in chinese hamster ovary cells: regulation of ERp72 and protein disulfide isomerase expression and secretion. J. Biol. Chem. 265, 22029–22034. Freedman, R.B., 1984. Native disulfide bond formation in protein biosynthesis: evidence for the role of protein disulfide-isomerase. Trends Biochem. Sci. 9, 438–441. Freedman, R.B., Hirst, T.R., Tuite, M.F., 1994. Protein disulphide isomerase: building bridges in protein folding. Trends Biochem. Sci. 19, 331–336. Gurr, S.J., Unkles, S.E., Kinghorn, J.R., 1987. The structure and organisation of nuclear genes of filamentous fungi. In: Kinghorn, J.R. ( Ed.), Gene Structure in Eukaryotic Microbes. IRL Press, Oxford, pp. 93–139. Hayano, T., Kikuchi, M., 1995. Molecular cloning of the cDNA encoding a novel protein disulfide isomerase-related protein (PDIR). FEBS Lett. 372, 210–214. Humphreys, D.P., Weir, N., Lawson, A., Mountain, A., Lund, P.A., 1996. Co-expression of human protein disulphide isomerase (PDI ) can increase the yield of an antibody Fab∞ fragment expressed in Escherichia coli. FEBS Lett. 380, 194–197. Jeenes, D.J., MacKenzie, D.A., Archer, D.B., 1994. Transcriptional and post-transcriptional events affect the production of secreted hen egg white lysozyme by Aspergillus niger. Transgenic Res. 3, 297–303. Kohno, K., Normington, K., Sambrook, J., Gething, M.J., Mori, K., 1993. The promoter region of the yeast KAR2 (BiP) gene contains a regulatory domain that responds to the presence of unfolded proteins in the endoplasmic reticulum. Mol. Cell. Biol. 13, 877–890. Li, W.W., Sistonen, L., Morimoto, R.I., Lee, A.S., 1994. Stress induction of the mammalian GRP78/BiP protein gene: in vivo genomic footprinting and identification of p70CORE from human nuclear extract as a DNA-binding component specific to the stress regulatory element. Mol. Cell. Biol. 14, 5533–5546. MacKenzie, D.A., Jeenes, D.J., Belshaw, N.J., Archer, D.B., 1993. Regulation of secreted protein production by filamentous fungi: recent developments and perspectives. J. Gen. Microbiol. 139, 2295–2307. Malpricht, S., Thamm, A., Khanh, N.Q., 1996. Cloning of cDNA for the protein disulfide isomerase from Aspergillus niger strain NRRL3 using PCR. Biotechnol. Lett. 18, 445–450. Ngiam, C., Jeenes, D.J., Archer, D.B., 1997. Isolation and characterisation of a gene encoding protein disulfide isomerase, pdiA, from Aspergillus niger. Curr. Genet. 31, 133–138.
Noiva, R., Freedman, R.B., Lennarz, W.J., 1993. Peptide binding to protein disulfide isomerase occurs at a site distinct from the active sites. J. Biol. Chem. 268, 19210–19217. Robinson, A.S., Hines, V., Wittrup, K.D., 1994. Protein disulfide isomerase overexpression increases secretion of foreign proteins in Saccharomyces cerevisiae. Bio/Technology 12, 381–384. Rupp, K., Birnbach, U., Lundstro¨m, J., Van, P.N., So¨ling, H.-D., 1994. Effects of CaBP2, the rat analog of ERp72, and of CaBP1 on the refolding of denatured reduced proteins: comparison with protein disulfide isomerase. J. Biol. Chem. 269, 2501–2507. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular cloning. A Laboratory Manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Schultz, L.D., Markus, H.Z., Hofmann, K.J., Montgomery, D.L., Dunwiddie, C.T., Kniskern, P.J., Freedman, R.B., Ellis, R.W., Tuite, M.F., 1994. Using molecular genetics to improve the production of recombinant proteins by the yeast Saccharomyces cerevisiae. Ann. N.Y. Acad. Sci. 721, 148–157. Shorrosh, B.S., Dixon, R.A., 1992. Molecular characterisation and expression of an alfalfa protein with sequence similarity to mammalian ERp72, a glucose-regulated endoplasmic reticulum protein containing active site sequences of protein disulphide isomerase. Plant J. 2, 51–58. So¨llner, T., Griffiths, G., Pfaller, R., Pfanner, N., Neupert, W., 1989. MOM19, an import receptor for mitochondrial precursor proteins. Cell 59, 1061–1070. Srinivasan, M., Lenny, N., Green, M., 1993. Identification of genomic sequences that mediate the induction of the endoplasmic reticulum stress protein, ERp72, by protein traffic. DNA Cell Biol. 12, 807–822. Tachibana, C., Stevens, T.H., 1992. The yeast EUG1 gene encodes an endoplasmic reticulum protein that is functionally related to protein disulfide isomerase. Mol. Cell. Biol. 12, 4601–4611. Tachikawa, H., Takeuchi, Y., Wataru, F., Miura, T., Gao, X.-D., Fujimoto, D., Mizunaga, T., Onodera, K., 1995. Isolation and characterisation of a yeast gene, MPD1, the overexpression of which suppresses inviability caused by protein disulfide isomerase depletion. FEBS Lett. 369, 212–216. von Heijne, G., 1986. A new method for predicting signal sequence cleavage sites. Nucleic Acids Res. 14, 4683–4690. Wilson, D.W., Lewis, M.J., Pelham, H.R., 1993. pH-dependent binding of KDEL to its receptor in vitro. J. Biol. Chem. 268, 7465–7468.
Isolation and characterisation of a novel stress-inducible PDI-family gene from Aspergillus niger D.J. Jeenes a,*, R. Pfaller b, D.B. Archer a a Department of Genetics and Microbiology, Institute of Food Research, Norwich Research Park, Colney, Norwich, NR4 7UA, UK b Consortium fu¨r elektrochemische Industrie GmbH, Zielstattstr. 20, 81379 Mu¨nchen, Germany Received 16 September 1996; received in revised form 12 December 1996; accepted 8 January 1997; Received by B. Dujon
Abstract Current strategies to improve the secretion of heterologous proteins in Aspergillus niger include the manipulation of chaperones and foldases specific to the endoplasmic reticulum ( ER). A family of ER-specific protein s which share active-site homology wit protein disulfide isomerase (PDI ) has been identified from other systems, many of which are inducible by agents which cause malfolding of proteins in the ER. Here we report identification of tigA from Aspergillus niger and erp38 from Neurospora crassa, two novel members of the PDI superfamily of proteins. TIGA and ERp38 show 66% identity at the amino acid level and are putative ER proteins. Both proteins show tandemly linked thiol-oxidoreductase domains followed by a functionally uncharacterised C-terminal domain. The most distal active site in TIGA is created by excision of a 66-bp intron. Although no Unfolded Protein Response elements can be seen in the tigA promoter, sequence homology has identified associated with protein trafficking (ERPTRE ) in a gene encoding the related mammalian protein, ERp72, as well as a second motif conserved amongst the glucoserelated protein family. Southern and dot blot analysis indicate that the tigA gene is present in single copy. Both the A. niger and N. crassa proteins show homology with a stress-inducible alfalfa, G1. Transcription of tigA is induced 2–3-fold after treatment with tunicamycin, an inhibitor of N-linked glycosylation. Strains overexpressing a heterologous protein show no increased tigA mRNA levels. © 1997 Elsevier Science B.V. Keywords: Filamentous fungi; Heterologous expression; Secretion; Chaperones; Tunicamycin
1. Introduction The filamentous fungus, Aspergillus niger, secretes high levels of some native proteins making it an attractive host for the production of recombinant proteins. Secreted yields of heterologous proteins from A. niger can be disappointingly low, however. Previous work has shown that post-transcriptional events are primarily responsible for the low secreted yields of heterologous proteins (Jeenes et al., 1994) and several strategies, such as the use of translational fusions or protease-deficient * Corresponding author. Tel.: +44 1603 255255; Fax: +44 1603 507723; e-mail: [email protected] Abbreviations: aa, amino acid(s); bp, base pair(s); CHO, chinese hamster ovary; ER, endoplasmic reticulum; ERPTRE, endoplasmic reticulum protein trafficking response element; kb, kilobase(s) or 1000 bp; ORF, open reading frame; PCR, polymerase chain reaction; PDI, protein disulfide isomerase; pfu, plaque forming units; TIGA, tunicamycin inducible gene A polypeptide; tsp, transcriptional start point; *, UAG stop codon. 0378-1119/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S 03 7 8 -1 1 1 9 ( 9 7 ) 0 0 0 98 - X
hosts, have been developed to overcome this problem (MacKenzie et al., 1993; Archer et al., 1994). Difficulties encountered during passage of proteins from the endoplasmic reticulum ( ER) to the Golgi are a major factor influencing secreted yields. During this step, protein folding is modulated by a battery of ER-specific proteins including protein disulfide isomerase (PDI ). Although the in vivo role of ER-specific chaperones and foldases such as PDI is poorly understood, overexpression of PDI in Saccharomyces cerevisiae can increase secreted yields of some heterologous proteins by between 10and 24-fold (Schultz et al., 1994; Robinson et al., 1994). A less dramatic increase (2-fold ) for production of some Fab∞ antibody constructs from strains of Escherichia coli overexpressing human PDI has also been shown, although yields remained low (Humphreys et al., 1996). A good correlation between PDI levels and the amount of secreted protein from a number of cell types has also been reported ( Freedman, 1984). In vitro experiments show that foldases can often act synergistically to increase both the rate and yield of folded end product
152
D.J. Jeenes et al. / Gene 193 (1997) 151–156
stress inducible, specifically by agents which cause malfolding of proteins in the ER, and show some functional overlap in vivo ( Tachibana and Stevens, 1992; Tachikawa et al., 1995). Our current aim is to dissect the roles of this protein family in the secretory pathway as part of a programme to optimise expression and secretion of recombinant proteins in A. niger. In this paper we report the isolation and characterisation of a novel stress-inducible PDI-family gene, tigA, from A. niger and the cDNA sequence of a homologous gene, erp38, from Neurospora crassa.
2. Experimental and discussion 2.1. Isolation and characterisation of tigA
Fig. 1. Nucleotide and deduced aa sequence of the A. niger tigA gene. Non-translated and intron sequences are in lower case; translated sequences are in upper case. Putative TATA- and CT-boxes are underlined. A putative ERPTRE sequence is overlined. The translational stop codon is represented by an asterisk. The polyadenylation site is underlined with a dashed line. The PDI active-site motifs are boxed and the ER-retention signal indicated by a double underline. An arrowhead refers to the probable signal sequence cleavage site. Degenerate oligonucleotide mixes used to isolate PCR fragments for cloning and partial sequencing were: 5∞-CC(CG)TGGTG( TC )GG( TC )CA(CT )TG-3∞ and 5∞-CTGAA(CT )TA(CG)AG(CT )TC(AG)TC(TC )TT-3∞. The same mixes were used to amplify a cloned PCR fragment which showed PDI homology for probing against the A. niger N402 lZAPII genomic library. 50×103 pfu were transferred onto Hybond-N+ membrane using standard protocols (Sambrook et al., 1989). Probes were labelled with 32P-dATP using the random primer Megaprime kit (Amersham, UK ) according to the manufacturer’s instructions. High stringency (65°C; 0.1×SSC ) hybridisation conditions were used. Plaques were cored and purified using a secondary hybridisation screen before plasmid rescue according to the protocol supplied (Stratagene, USA). cDNA clones were isolated in identical fashion from a lZAP cDNA library of a derivative of A. niger N402 in which the glaA gene had been deleted (kindly supplied by Dr. C.A.M.J.J. van den Hondel ). Dye terminator cycle sequencing mixes (Perkin Elmer, USA) were used with DNA purified through a Qiagen ( UK ) column to sequence the DNA using protocols supplied by the manufacturers. The nucleotide sequence data reported in this paper will appear in the EMBL, GenBank and DDBJ Nucleotide Sequence Databases under the accession No. X98748.
(Rupp et al., 1994). In addition, a family of ER-specific proteins which share active-site homology with PDI is now emerging from work on a variety of systems (Freedman et al., 1994; Hayano and Kikuchi, 1995; Tachikawa et al., 1995). Several of these proteins are
Degenerate oligonucleotide primer mixes designed against the –PWCGHC– active site region of PDI and a KDEL ER-retention signal were used to generate and clone specific A. niger DNA fragments by polymerase chain reaction (PCR). One of three fragments sequenced showed strong identity to PDI sequences from a number of sources and was used to probe a lZAPII genomic library of A. niger strain N402. The majority of purified pBluescript constructs rescued shared common restriction fragments and the clone containing the largest (5-kb) insert was chosen for further study. Sequencing of the entire insert revealed one ORF, designated tigA (tunicamycin inducible gene), potentially encoding a peptide of 359 aa residues (38.7 kDa) and containing one putative 66-bp intron (Fig. 1). Potential lariat, 5∞ and 3∞ splice site signals conform well to intron excision sequences found in filamentous fungi (Gurr et al., 1987). Attempts to map the transcriptional start point using primer extension were unsuccessful; however, RT-PCR using primers which anneal between −78 to −59 (relative to the ATG initiator codon) or −133 to −114 indicate transcriptional initiation occurs between these two priming sites (data not shown). Analysis of the upstream sequence shows both TATA- and CT-boxes, motifs which are important for the transcription of many fungal promoters (Gurr et al., 1987). Several ER-specific genes contain regulatory elements which respond to the presence of unfolded proteins or agents that cause malfolding of proteins. Although no convincing Unfolded Protein Response or heat shock elements ( Kohno et al., 1993; Bush et al., 1994) could be seen in the promoter region of tigA, a 23-bp element with 70% identity to a region of the 82-bp ER Protein Traffic Response Element ( ERPTRE) found in the mammalian ERp72 gene (Srinivasan et al., 1993) was detected ( Fig. 2a). This element confers inducibility to both the Ca2+ ionophore A23187 and expression of incompletely assembled secreted forms of Ig m heavy chain in COS cells. An element with comparable identity to the same
D.J. Jeenes et al. / Gene 193 (1997) 151–156
153
2.2. tigA and erp38 ORF analysis
Fig. 2. (a) Shared homology between tigA promoter sequences and the ERp72 ERPTRE motif and (b) between tigA promoter sequences and the grp-core motif from the rat grp78 gene. Co-ordinates are numbered relative to the translational initiation codon. (c) Schematic structure of the TIGA protein: grey area, ER translocation signal (aa 1–19); open area, ORF (aa 20–359), filled area, –CGHC– motifs (aa 49–52 and 169–172); hatched area, KDEL ER-retention signal, —, intron (aa 171).
region of the ERPTRE of murine ERp72 is also seen in the A. niger pdiA gene encoding another ER-specific PDI-family gene (Ngiam et al., 1997). Both of these A. niger promoter elements are sited a similar distance upstream of the initiator ATG codon as the ERp72 motif which might suggest a regulatory function in vivo and a role in protein trafficking for the tigA gene product. A second element of 19 bp with 75% identity to part of the ‘grp core’ element conserved amongst promoters of the glucose regulated protein (grp) gene family is also present in the tigA promoter region (Fig. 2b). Interestingly, the boundaries of the homology agree almost precisely with that part of the grp core motif responsible for the elevation of basal, as opposed to inducible, levels of transcription in promoters of the human and rat grp78 genes (Li et al., 1994). This motif is a similar distance from the initiator codon in all three genes which supports an in vivo role for this element in tigA regulation. Partial cDNA clones of tigA were isolated from an A. niger N402 cDNA library to identify intron boundaries; the longest of these contained the entire ORF except for the first base of the initiator ATG codon. Sequence analysis of two independent cDNA clones confirmed the intron splice sites and defined the polyadenylation site ca. 115 bp downstream of the stop codon (Fig. 1). A schematic representation of the protein structure and intron location is shown in Fig. 2c. Southern blot analysis indicates that tigA is present as a single copy and dot blot analysis of serial DNA dilutions against a single gene copy control supports this conclusion (data not shown). Northern blot analysis confirms the predicted transcript size of approx. 1.35 kb (data not shown).
Sequence analysis using the programme SIGCLEAVE identified the most probable ER-translocation signal cleavage site (score 10.5) between aa residues 19 and 20 ( Fig. 1; von Heijne, 1986). The 340-aa mature protein contains a –KDEL carboxy-terminal motif which is likely to serve as an ER-retention signal and supports the idea that TIGA functions in the ER lumen. Interestingly, A. niger employs a different ER-retention signal (–HDEL) for the PDI protein (Malpricht et al., 1996; Ngiam et al., 1997). Binding of HDEL and KDEL ligands is sensitive to pH and shows different affinities for their common receptor in vitro ( Wilson et al., 1993) suggesting that TIGA and PDI may operate at different sites within the ER to cis-Golgi network. Two –CGHC– active-site motifs characteristic of PDI-family genes ( Freedman et al., 1994) are also present at an interval of 115 aa (Fig. 2c). The distal –CGHC– motif is created by intron excision, unique amongst those PDI-family genes reported to date, but whether this has functional importance in vivo is unknown. Spacing of these motifs at either ca. 115 aa apart or 330 aa apart is a highly conserved feature of the PDI protein family ( Fig. 3). A full length cDNA clone encoding a protein, ERp38, has also been isolated from a lgt11 library of Neurospora crassa cDNA by immunoscreening ( Fig. 4) and is a strong candidate for the N. crassa homologue of tigA with 79% similarity and 66% identity at the amino acid level (Fig. 5). Analysis of the erp38 ORF shows a similar spacing of the active sites to that of tigA, a putative N-terminal signal sequence of 18 aa (score 6.6) and a –KEEL ER retention signal. These two sequences provide evidence for a novel sub-family of ER-localised proteins in which the two thiol-oxidoreductase domains,
Fig. 3. The PDI-family of genes in the ER. References to the different proteins can be found in the Introduction (Section 1).
154
D.J. Jeenes et al. / Gene 193 (1997) 151–156
Fig. 4. Nucleotide and deduced aa sequence of the N. crassa erp38 cDNA. The nucleotide sequence data will appear in the EMBL, Genbank and DDBJ Nucleotide Sequence Databases under the accession No. Y07562. PDI active-site motifs are boxed and the ER-retention signal indicated by a double underline. An arrowhead refers to the probable signal sequence cleavage site. Immunoscreening of a lgt11 library of N. crassa cDNA by standard protocols (Sambrook et al., 1989) used an antibody directed against gel-purified N. crassa 38-kDa proteins isolated using methods previously reported (So¨llner et al., 1989).
90–100 aa in length and centred around the active sites, are tandemly linked without an intervening sequence. This is in contrast to larger proteins in the PDI family such as PDI, EUG1, ERp61 etc. which contain a ca. 200-aa region of unknown function between the two thiol oxidoreductase domains ( Fig. 3). Database searches have also identified homology between TIGA (58% similarity, 43% identity) or ERp38 (61% similarity, 45% identity) and the product of a stress-inducible gene, G1, from alfalfa (Shorrosh and Dixon, 1992). Although the alfalfa protein does contain an ER-translocation signal, it lacks an ER-retention signal suggesting either that it may be targetted to a different subcellular location in this host or is retained as part of a heteromeric complex by interaction with subunits which do contain such a signal. Strong identity also extends beyond the two linked thiol-oxidoreductase domains into a so far functionally uncharacterised C-terminal domain ( Fig. 5). The similarity of the residues surrounding the active sites in TIGA, ERp38 and G1p to those found in PDI suggest a strongly oxidising oxidoreductase activity whilst the lack of a peptide-binding site locally rich in acidic aa argues against high isomerase activity (Noiva et al., 1993). Taken together, the peptide M , r active-site spacing and active-site composition support an isofunctional role, distinct from that for PDI, for all three peptides in protein trafficking in their respective hosts. It is also of interest to note that all three peptides are lysine-rich (>10%) although what significance this has is unclear. Functional analysis of the tigA gene product is currently in progress.
Fig. 5. Sequence comparison between the products of the tigA, ERp38 and G1 genes using PileUp from the University of Wisconsin suite of programmes. Identical residues are shaded in black, similar residues are shaded in grey, non-homologous residues are white. Similarity is defined using a reference matrix file compiled from multiple sequence alignments which considers both physicochemical similarity and the frequency of mutation between one amino acid and another.
D.J. Jeenes et al. / Gene 193 (1997) 151–156
2.3. Transcriptional regulation of tigA Like many other ER-specific proteins, several PDIfamily proteins are inducible by agents which perturb or disrupt ER function, e.g. tunicamycin, an inhibitor of N-linked glycosylation (Dorner et al., 1990; Shorrosh and Dixon, 1992; Tachibana and Stevens, 1992). A comparison of tigA expression in the presence and absence of tunicamycin shows that tigA mRNA is induced 2–3-fold after a time lag of 3 h ( Fig. 6). This relatively low level of induction agrees with the level of induction reported for PDI mRNA (3–4-fold ) but not ERp72 mRNA (10-fold ) following tunicamycin treatment of CHO cells (Dorner et al., 1990) or EUG1 mRNA (10-fold ) in yeast cells ( Tachibana and Stevens, 1992). It is also in contrast to that reported for the alfalfa G1 gene which, although not quantified, was strongly induced (Shorrosh and Dixon, 1992). It is possible that optimal induction conditions would require removal of glucose from the cultures tested as some PDI-family genes are induced under conditions of glucose starvation. Levels of actin mRNA also appear to
155
be reduced relative to the earlier time points (Fig. 6). The reason for this is unclear although dry weights indicate that the A. niger culture is still actively growing and carbon source is still detectable (data not shown). The 3–4-h lag period before tigA induction mimics that reported for the alfalfa G1 gene and suggests that tigA expression does not form part of the primary stress response. Preliminary data examining the expression of tigA in A. niger transformants overexpressing heterologous proteins show no increase in tigA mRNA (data not shown). The precise role of TIGA and other PDIfamily proteins such as ERp72, PDIR, MPD1 and P5 ( Fig. 3) in vivo remains unknown. Experiments to study phenotypic changes associated with changes in tigA expression are currently in progress.
3. Conclusions (1) The tigA gene from A. niger and erp38 cDNA from N. crassa have been cloned by PCR-based methods and antibody screening respectively. (2) TIGA and ERp38 are homologues and encode a novel potentially ER-localised member of the PDI superfamily of proteins. (3) The tigA gene is inducible by tunicamycin and TIGA, like ERp38, shows homology to a stressinducible gene product from alfalfa whose deduced amino acid sequence lacks an ER-retention signal. (4) Sequence homology analysis has identified a promoter element in tigA associated with protein trafficking in a related mammalian protein, ERp72, and a second promoter element conserved amongst chaperone proteins of the glucose-regulated protein family.
Acknowledgement
Fig. 6. Northern blot of RNA from tunicamycin-induced cultures of A. niger AB4.1. Cultures were grown on ACM/N/P as previously described (Jeenes et al., 1994) except that 1% glucose was used for carbon source. Tunicamycin (10 mg/ml ) was added at t=50 h; cultures were then harvested at 1 h, 2 h, 3 h and 4 h intervals after tunicamycin addition, the mycelium ground under liquid N and freeze dried and 2 their RNA extracted using an RNeasy extraction kit for fungi (Qiagen, UK ). 15 mg total RNA for each sample was electrophoresed through a 0.8% agarose gel containing 2.2 M formaldehyde at <3 V/cm until the bromophenol blue had run ca. 8 cm. The gel was washed through 5 changes of DEPC-treated water before transfer onto Hybond-N+ (Amersham, UK ) by standard methods (Sambrook et al., 1989). Hybridisation conditions, probe preparation and labelling were as described for library screening in the legend to Fig. 1. Induction levels represent tigA mRNA levels normalised against actin mRNA quantified using a Fuji BAS-1500 phosphorimager.
This work was partly funded by an EC Biotechnology programme grant, BIO2 CT-942045 (D.J.J. and D.B.A) and by the Sonderforschungsbereich 184 (project B1; R.P.) We thank M. Kiebler for experimental advice and greatly appreciate the support from Walter Neupert at the Institut fu¨r Physiologische Chemie, Mu¨nchen, during the course of the ERp38 cDNA isolation.
References Archer, D.B., Jeenes, D.J., MacKenzie, D.A., 1994. Strategies for improving heterologous protein production from filamentous fungi. Antonie van Leeuwenhoek 65, 245–250. Bush, K.T., Hendrickson, B.A., Nigam, S.K., 1994. Induction of the FK506-binding protein, FKBP13, under conditions which misfold proteins in the endoplasmic reticulum. Biochem. J. 303, 705–708. Dorner, A.J., Wasley, L.C., Raney, P., Haugejorden, S., Green, M.,
156
D.J. Jeenes et al. / Gene 193 (1997) 151–156
Kaufman, R.J., 1990. The stress response in chinese hamster ovary cells: regulation of ERp72 and protein disulfide isomerase expression and secretion. J. Biol. Chem. 265, 22029–22034. Freedman, R.B., 1984. Native disulfide bond formation in protein biosynthesis: evidence for the role of protein disulfide-isomerase. Trends Biochem. Sci. 9, 438–441. Freedman, R.B., Hirst, T.R., Tuite, M.F., 1994. Protein disulphide isomerase: building bridges in protein folding. Trends Biochem. Sci. 19, 331–336. Gurr, S.J., Unkles, S.E., Kinghorn, J.R., 1987. The structure and organisation of nuclear genes of filamentous fungi. In: Kinghorn, J.R. ( Ed.), Gene Structure in Eukaryotic Microbes. IRL Press, Oxford, pp. 93–139. Hayano, T., Kikuchi, M., 1995. Molecular cloning of the cDNA encoding a novel protein disulfide isomerase-related protein (PDIR). FEBS Lett. 372, 210–214. Humphreys, D.P., Weir, N., Lawson, A., Mountain, A., Lund, P.A., 1996. Co-expression of human protein disulphide isomerase (PDI ) can increase the yield of an antibody Fab∞ fragment expressed in Escherichia coli. FEBS Lett. 380, 194–197. Jeenes, D.J., MacKenzie, D.A., Archer, D.B., 1994. Transcriptional and post-transcriptional events affect the production of secreted hen egg white lysozyme by Aspergillus niger. Transgenic Res. 3, 297–303. Kohno, K., Normington, K., Sambrook, J., Gething, M.J., Mori, K., 1993. The promoter region of the yeast KAR2 (BiP) gene contains a regulatory domain that responds to the presence of unfolded proteins in the endoplasmic reticulum. Mol. Cell. Biol. 13, 877–890. Li, W.W., Sistonen, L., Morimoto, R.I., Lee, A.S., 1994. Stress induction of the mammalian GRP78/BiP protein gene: in vivo genomic footprinting and identification of p70CORE from human nuclear extract as a DNA-binding component specific to the stress regulatory element. Mol. Cell. Biol. 14, 5533–5546. MacKenzie, D.A., Jeenes, D.J., Belshaw, N.J., Archer, D.B., 1993. Regulation of secreted protein production by filamentous fungi: recent developments and perspectives. J. Gen. Microbiol. 139, 2295–2307. Malpricht, S., Thamm, A., Khanh, N.Q., 1996. Cloning of cDNA for the protein disulfide isomerase from Aspergillus niger strain NRRL3 using PCR. Biotechnol. Lett. 18, 445–450. Ngiam, C., Jeenes, D.J., Archer, D.B., 1997. Isolation and characterisation of a gene encoding protein disulfide isomerase, pdiA, from Aspergillus niger. Curr. Genet. 31, 133–138.
Noiva, R., Freedman, R.B., Lennarz, W.J., 1993. Peptide binding to protein disulfide isomerase occurs at a site distinct from the active sites. J. Biol. Chem. 268, 19210–19217. Robinson, A.S., Hines, V., Wittrup, K.D., 1994. Protein disulfide isomerase overexpression increases secretion of foreign proteins in Saccharomyces cerevisiae. Bio/Technology 12, 381–384. Rupp, K., Birnbach, U., Lundstro¨m, J., Van, P.N., So¨ling, H.-D., 1994. Effects of CaBP2, the rat analog of ERp72, and of CaBP1 on the refolding of denatured reduced proteins: comparison with protein disulfide isomerase. J. Biol. Chem. 269, 2501–2507. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular cloning. A Laboratory Manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Schultz, L.D., Markus, H.Z., Hofmann, K.J., Montgomery, D.L., Dunwiddie, C.T., Kniskern, P.J., Freedman, R.B., Ellis, R.W., Tuite, M.F., 1994. Using molecular genetics to improve the production of recombinant proteins by the yeast Saccharomyces cerevisiae. Ann. N.Y. Acad. Sci. 721, 148–157. Shorrosh, B.S., Dixon, R.A., 1992. Molecular characterisation and expression of an alfalfa protein with sequence similarity to mammalian ERp72, a glucose-regulated endoplasmic reticulum protein containing active site sequences of protein disulphide isomerase. Plant J. 2, 51–58. So¨llner, T., Griffiths, G., Pfaller, R., Pfanner, N., Neupert, W., 1989. MOM19, an import receptor for mitochondrial precursor proteins. Cell 59, 1061–1070. Srinivasan, M., Lenny, N., Green, M., 1993. Identification of genomic sequences that mediate the induction of the endoplasmic reticulum stress protein, ERp72, by protein traffic. DNA Cell Biol. 12, 807–822. Tachibana, C., Stevens, T.H., 1992. The yeast EUG1 gene encodes an endoplasmic reticulum protein that is functionally related to protein disulfide isomerase. Mol. Cell. Biol. 12, 4601–4611. Tachikawa, H., Takeuchi, Y., Wataru, F., Miura, T., Gao, X.-D., Fujimoto, D., Mizunaga, T., Onodera, K., 1995. Isolation and characterisation of a yeast gene, MPD1, the overexpression of which suppresses inviability caused by protein disulfide isomerase depletion. FEBS Lett. 369, 212–216. von Heijne, G., 1986. A new method for predicting signal sequence cleavage sites. Nucleic Acids Res. 14, 4683–4690. Wilson, D.W., Lewis, M.J., Pelham, H.R., 1993. pH-dependent binding of KDEL to its receptor in vitro. J. Biol. Chem. 268, 7465–7468.