BIOPHARMACEUTICAL PRODUCTION METHOD

Abstract

The present invention resides in a method for the manufacture of a disulphide-requiring biopharmaceutical having an element of at least tertiary structure using wild type E. coli.

Claims

1. A method for the manufacture of a disulphide-requiring biopharmaceutical having an element of at least tertiary structure using wild type E. coli.

2. A method according to claim 1 wherein the method comprises use of Tat-dependent export from the cytoplasm to the periplasm and subsequent extraction and purification.

3. A method according to claim 1, wherein disulphide-requiring biopharmaceutical is a proteins or fragments thereof.

4. A method according to claim 1, wherein the disulphide-requiring biopharmaceutical is an antibody fragments.

5. A method according to claim 1, wherein the disulphide-requiring biopharmaceutical is manufactured with a substantially active and/or natural conformation.

6. A method according to claim 1, wherein the disulphide-requiring biopharmaceutical adopts at least a tertiary structure before the formation of disulphide bonds.

7. A method according to claim 6, wherein the at least tertiary structure is adopted in the cytoplasm.

8. A method according to claim 1, wherein the disulphide-requiring biopharmaceutical folds into a substantially active and/or native or near-native conformation in the cytoplasm before the formation of disulphide bonds.

9. A method according to claim 1, wherein the disulphide-requiring biopharmaceutical contains two or more disulphide bonds.

10. A method according to claim 1, wherein the disulphide-requiring biopharmaceutical has a molecular weight of less than about 50 kDa, preferably a molecular weight of less than about 30 kDa, more preferably a molecular weight of about 20-30 kDa.

Description

[0018] The present invention will now be described in more detail with reference to the following examples and figures in which:

[0019] FIG. 1 is an immunoblot showing the efficient Tat-dependent export of interferon alpha 2b (IFN) and human growth hormone (hGH) by Tat in both the presence and absence of CyDisCo components. hGH and IFN were expressed with a TorA Tat signal peptide in E. coli in the presence or absence of CyDisCo components. After 3 h, cells were separated into cytoplasm/membrane/periplasm fractions (C, M, P) and samples were immunoblotted to detect the proteins. Mobilities of mature-size hGH and IFN are indicated; the mobilities of molecular mass markers are indicated (in kDa) on the left.

[0020] FIG. 2 is an immunoblot showing that scFv and VH domain antibody constructs are exported by Tat in the presence and absence of CyDisCo components. TorA-scFv and TorA-VH domain constructs were expressed in wild type or CyDisCo cells as described for IFN and hGH, and cells were fractionated to yield cytoplasm, membrane and periplasm samples (C, M, P). The samples were analysed by immunoblotting using antibodies to the His tag, and mobilities of mature scFv and VH domain protein are indicated. Asterisks denote the precursor proteins.

[0021] FIG. 3 is an immunoblot showing that export of TorA-scFv occurs exclusively by the Tat pathway. TorA-scFv was expressed in wild type and tat null mutant strains for 3 h, in either the presence or absence of CyDisCo components as indicated. After this period cells were fractionated and spheroplast and periplasm (Sp, P) samples were analysed by immunoblotting using antibodies to the C-terminal His tag on the scFv protein. The mobilities of the precursor and mature scFv forms are indicated.

[0022] FIG. 4 illustrates that TorA-scFv is efficiently exported and accurately processed. Panel A is an immunoblot showing that TorA-scFV was expressed in MC4100 cells or the tat null mutant strains in the presence or absence of CyDisCo components as indicated. The periplasmic fractions were run on Coomassie-stained gels and the putative scFv band (present in the periplasm of wild type cells but not tat null mutant cells) is indicated. Panel B shows the full mature scFv sequence with the peptides identified by MS in bold and the sequence identified by MS/MS in bold and underlined. The putative scFv band from the CyDisCo-expressing sample was excised and subjected to MALDI TOF for peptide mass fingerprinting.

[0023] FIG. 5 shows that periplasmic hGH is disulphide-bonded after Tat-dependent export in wild type or CyDisCo-expressing cells. Panel A is an immunoblot showing that TorA-hGH was expressed in the presence or absence of CyDisCo components as indicated (+/−). The periplasmic fraction was subjected to SDS-polyacrylamide gel electrophoresis in the presence or absence of reducing agent (2-mercaptoethanol) as indicated. The mobilities of reduced and oxidised hGH (Red, Ox) are indicated, as is a higher band that is possibly a dimer. Panel B is an immunoblot in which the same TorA-hGH construct was expressed in a dsbB.sup.− strain in the absence of CyDisCo components and samples of the periplasm fraction were again run under reducing or non-reducing conditions as in A. Mobilities of oxidised and reduced hGH forms are indicated (Ox, Red).

[0024] FIG. 6 shows immunoblots in which hGH is almost completely disulphide-bonded after purification from the periplasm of wild type or CyDisCo-expressing cells. TorA-hGH was expressed in wild type or CyDisCo-expressing cells and the protein was purified from periplasmic extracts by IMAC chromatography. The protein was run on SDS polyacrylamide gels in the presence or absence of 2-mercaptoethanol as indicated. Mobilities of oxidised and reduced forms are indicated.

[0025] FIG. 7 shows that complete disruption of one or both disulphide bonds in hGH does not affect export by the Tat pathway. Panel A is an immunoblot in which site-specific mutagenesis of TorA-hGH was carried out to substitute one of the Cys residues involved in the first or second disulphide bond (Cys53, Cys189, respectively) by serine. A third, double-mutant combined both substitutions. All three variants were expressed in wild type E. coli cells, together with non-mutated TorA-hGH, and the periplasmic fractions were isolated. These samples were run on SDS-polyacrylamide gels under reducing (left) or non-reducing (right) conditions. All of the mutated forms are clearly exported by Tat providing proof that prior disulphide bond formation is not required for export. Panel B is a ribbon diagram of the hGH structure showing the positions of the disulphide bonds.

MATERIALS AND METHODS

Materials

[0026] All chemicals, unless specified otherwise, were obtained from Sigma Chemical Co. Ltd. (Poole, Dorset, UK) and were of analytical grade.

Plasmids & Bacterial Strains

[0027] All constructs were amplified using Phusion high fidelity DNA polymerase (New England Biolabs) to include a 5′ NdeI site and a 3′ BamHI site immediately after a 6-His tag. The product was then digested with NdeI and BamHI (New England Biolabs) and inserted into pYU49 which had been cut similarly (Matos et al., 2014).

[0028] All plasmid purification was performed using the QIAprep spin miniprep kit (Qiagen) and all purification from agarose gels was performed using the gel extraction kit (Qiagen), both according to manufacturer's instructions. All plasmids generated were fully sequenced (see Table 1 for plasmid names and details).

Growth Conditions

[0029] Following transformation, a single colony was used to inoculate 10 ml of LB media containing 100 μg mL.sup.−1 ampicillin and grown overnight at 37° C., 200 rpm. Using the overnight culture, 50 ml of LB in 1L flasks containing appropriate antibiotics was inoculated to OD.sub.600=0.05 and grown to approximately OD.sub.600=0.5 at 37° C., 200 rpm. At OD.sub.600=0.5, cells were induced with 1 mM IPTG and left for 3 hrs at 25° C. or 30° C. After this time an amount equivalent to 1 mL of OD.sub.600=10 was collected by centrifugation (3,000 rpm, 10 minutes). The periplasmic fraction (P) was prepared by the EDTA/lysozyme/cold osmoshock procedure (Randall and Hardy (1986), Pierce et al (1997)). The pellet was then washed with buffer containing 50 mM Tris-Acetate (pH 8.2), 250 mM sucrose and 10 mM MgSO.sub.4 and centrifuged for 5 minutes at 14,000 rpm, 4° C. The resulting pellet was then re-suspended in 50 mM Tris-Acetate, 2.5 mM EDTA (pH 8.2) and sonicated on ice for 4-6×10 seconds, 8 μm amplitude with 10 seconds between sonication (Soniprep 150plus, Sanyo Gallenkamp, Loughborough, UK). The sonicate was then centrifuged for 30 minutes at 70,000 rpm, 4° C. to collect the insoluble fraction. The supernatant was removed as the cytoplasmic fraction (C) and the pellet re-suspended in 50 mM Tris-Acetate, 2.5 mM EDTA (pH 8.2) to give the membrane/insoluble fraction (M). All cell fractions were stored frozen in aliquots for further experiments as repeated freeze-thawing can influence the results obtained in a protein-dependent manner.

Detection of Proteins by Immunoblotting

[0030] After SDS-PAGE and transfer, PVDF membranes to be immunoblotted with C-terminal His antibodies were blocked with PBS-T containing 5% (w/v) dried skimmed milk powder for at least 1 hour. The membranes were washed in PBS-T before incubation with PBS-T containing the primary antibody (Anti-His (C-term), Life Technologies, CA, USA) for 1 hour. The membranes were washed before incubation with the secondary antibody (Anti-Mouse IgG (H+L), HRP Conjugate, Promega, WI, USA) for another hour. The membranes were washed and immunoreactive bands were detected using ECL (enhanced chemiluminescence) kit (BioRad, Herts, UK) according to the manufacturer's instructions. Membranes were developed using a BioRad chemiluminescence imager and corresponding software.

Mass Spectrometry: In-Gel Digestion of Proteins

[0031] Excised, diced protein bands from Coomassie-stained SDS gels were washed, reduced and S-alkylated essentially as described in Fox et al ((2011) J. Microbiol. Methods. 84:243-250). A sufficient volume of 2 ng/μL of trypsin (modified sequencing grade, Promega, Southampton, UK) in 25 mM ammonium bicarbonate was added to cover the gel pieces and digestion performed overnight at 20° C. The digests were then acidified by the addition of a 0.5 volume of 50% acetonitrile with 5% formic acid prior to MS and MS-MS analysis.

MS and MS-MS Analysis

[0032] The sample (1 μL of the above peptide digest) was placed on the sample target (AnchorChip standard, 800 μm) and dried. Subsequently 0.5 μL of matrix was added and dried. The matrix was α-cyano-4-hydroxy-cinnamic acid (α-CHCA, 0.7 mg mL.sup.−1 in 85% acetonitrile, 15% H.sub.2O, 0.1% TFA and 1 mM NH.sub.4H.sub.2PO.sub.4). For external calibration in the protein mass range, Peptide Calibration Standard I (Bruker) standards were used. MALDI TOF MS and MALDI TOF-TOF MS-MS analysis was performed (in the positive ion mode) using a Bruker UltrafleXtreme. The spectra were obtained in reflector mode with an acceleration voltage of 25 kV and a pulse ion extraction time of 80 ns. The mass range for MS was generally between 700 and 3500 m/z. The number of laser shots summed in MS was 3500. The number of laser shots summed in MS-MS was 3000. The software flexAnalysis (Bruker, Bremen, Germany) was used for peak picking prior to using the standard Mascot search engine. The peptide mass fingerprint was searched against the non-redundant Swiss-Prot protein database (all organisms) placed in the public domain by UNIProt and modified by the addition of sequences corresponding to the recombinant fusion proteins being studied in this manuscript. flexAnalysis takes advantage of the isotopic envelope available from high resolution MS-MS spectra in peptide identification. Mascot is available from Matrix Science Ltd, London, UK.

TABLE-US-00001 TABLE 1 Strains and constructs used in this study Strain/Plasmid Description Source/Reference DH5α F.sub.——80lacZ_M15_(lacZYA-argF) Invitrogen, U169 deoR recA1 endA1 hsdR17 Carlsbad, CA, (RK_, mk_) gal-phoA supE44 thi-1 USA gyrA96 relA1 W3110 F- mcrA mcrB IN(rrnD-rrnE)1 ATCC, Manassas, lambda− VA, USA MC4100, AraR, FaraD139DlacU169 rpsL150 Bolhuis et al Ara.sup.R relA1 flB5301 deoC1 ptsF25 rbsR (2000) FEBS Letts. 472: 88-92 ΔTatABCDE, Like MC4100 Ara.sup.R; ΔtatABCDE Wexler et al Ara.sup.R (2000) J. Biol. Chem. 275: 16717-16722, Bolhuis et al (2000) (supra) K12, ΔdsbB Δ(araD-araB)567 ΔLacZ4787(::rrnB- Baba et al (2006) 3) λ.sup.− dsdB774::kan rph-1 Δ(rhaD- Mol. Syst. Biol. 2: rhaB)568 hsdR514 2006.008 BL21(DE3) fhuA2 [Ion] ompT gal (λ DE3) [dcm] Miroux and ΔhsdS λ DE3 = λ sBamHIo ΔEcoRI- Walker (1996) J. B int::(lacI::PlacUV5::T7 gene1) i21 Mol. Biol. 260: Δnin5 289-298 pYU49 pET23 based vector with pTac Matos et al 2014 promoter expressing TorA.sub.sp IL-1β (supra) scFv, codon-optimised Erv1p and mature codon-optimised hPDI pHAK1 As above with mature hGH-6His This study replacing IL-1β scFv pHAK2 As above with mature α2b IFN-6His This Study replacing IL-1β scFv pHAK7 As above with β-galactosidase This study, scFv-6His replacing IL-1β scFv Martineau et al (1998) J. Mol. Biol. 280: 117-127 pHAK12 As above with VH domain-6His This study replacing IL-1β scFv pHAK14 As pHAK1 but without Erv1p and This study PDI pHAK15 As pHAK2 but without Erv1p and This study PDI pHAK17 As pHAK7 but without Erv1p and This study PDI pHAK23 As pHAK12 but without Erv1p and This study PDI pKWK1 As pHAK1 with C53S mutation This study pKWK2 As pHAK1 with C189S mutation This study pKWK3 As pHAK1 with C53S C189S This study mutation

Site-Specific Mutagenesis

[0033] hGH mutants were made using the Agilent site-directed mutagenesis protocol. The primers used for the C53S mutation were (5′ to 3′): CCCCCAGACCTCCCTCTCTTTCTCAGAGTCTATTCCGAC (SEQ ID NO:1) and GTCGGAATAGACTCTGAGAAAGAGAGGGAGGTCTGGGGG (SEQ ID NO:2). For the C189S mutant the following primers were used: GTGGAGGGCAGCTCTGGCTTCCATCATCATCATCATCAC (SEQ ID NO:3) and GTGATGATGATGATGATGGAAGCCAGAGCTGCCCTCCAC (SEQ ID NO:4). For the double C53S C189S mutant, the C189S mutant was used as a template with the C53S primers (described above)

Results

[0034] IFN, hGH, an scFv and a VH Domain Construct are all Efficiently Exported by Tat in the Absence of Prior Disulphide Formation

[0035] Previous studies (DeLisa et al (supra); Matos et al 2014 (supra)) showed that several disulphide-bonded proteins, including PhoA, a phytase AppA, an scFv construct and a Fab fragment, were exported in E. coli when a Tat signal peptide was present at the N-terminus, provided that disulphide formation could occur in the cytoplasm. In one case this was achieved by expression in Δgor/ΔtrxB cells. This strain passively enables the formation of disulphide bonds in proteins in the cytoplasm by the removal of the two naturally occurring reducing pathways (Prinz et al (1997) J. Biol. Chem. 272(25):15661-15667). In the other case the proteins were expressed in ‘CyDisCo’ strains that express a thiol oxidase and protein disulphide isomerase in the cytoplasm, actively promoting efficient disulphide bond formation. The same constructs were not exported in wild type strains, with the Tat pathway clearly identifying these constructs as ‘inappropriately folded’ and rejecting them.

[0036] The primary aim of this study was to test whether the Tat system can export other disulphide-bonded therapeutic proteins with high efficiency and human growth hormone (hGH), human interferon α2b (IFN), an scFv raised against the omega peptide of β-galactosidase (described in Martineau et al (1998) J. Mol. Biol. 280:117-127) and an antibody VH domain construct (Dudgeon et al (2008) Prot. Eng. Des. Sel. 22:217-220) were chosen as targets. Constructs containing C-terminal His tags were used to aid identification of the proteins and all proteins bore N-terminal TorA signal peptides (as used in Matos et al 2014 (supra)) to direct export by the Tat pathway. The precursor proteins were expressed on a pET23-based plasmid either alone or together with the CyDisCo components Erv1p and mature human PDI.

[0037] FIG. 1 shows the results of export assays using TorA-hGH and TorA-IFN. After a 3 h induction of expression, the cells were fractionated to yield cytoplasm, membrane and periplasm samples (C, M, P) which were immunoblotted using antibodies to the His tags on the target proteins. The figure shows that most of the TorA-IFN is exported to the periplasm (P) and processed to the mature size (20.7 kDa). A small amount of precursor protein is apparent in the membrane fraction and some mature-size protein is also present in the cytoplasm. This is commonly found during export of Tat substrates, probably because the signal peptide is prone to proteolytic clipping (see e.g. Matos et al (2012) Biotechnol. Bioeng. 109:2533-2542). The export of TorA-hGH is even more efficient with the vast majority of protein being found in the periplasm and the precursor form being barely detectable in the cytoplasm.

[0038] FIG. 1 also shows similar assays carried out in wild type cells, i.e. in the absence of CyDisCo components. Surprisingly, export of both proteins is again highly efficient and both proteins are found predominantly in the periplasm as mature size proteins. Next, the export of the scFv and VH domain constructs were tested and the results of export assays in wild type and CyDisCo strains are shown in FIG. 2. The data again show efficient export of both substrates in that a substantial proportion of the scFv and the majority of the VH domain construct are found in the periplasm in both CyDisCo-expressing and wild type cells.

TorA-scFv is Exported Exclusively by the Tat Pathway

[0039] The data shown in FIGS. 1 and 2 are surprising because the Tat-dependent export of other disulphide-bonded proteins has shown an absolute dependence on prior cytoplasmic disulphide bond formation, as explained above. Therefore it was considered possible that the substrates tested in this study may in fact be exported by the Sec pathway, since Sec-type signal peptides are similar in overall structural terms to Tat signal peptides. To test this possibility directly, some of the same constructs were expressed in tat null mutant cells and the data for TorA-scFv are shown in FIG. 3.

[0040] The results show that the scFv is not exported in the tat mutant strain to any significant extent, either in the absence or presence of CyDisCo components. Indeed, all of the protein is found in the cytoplasm or membrane fractions (C, M). In control assays, export is again observed in both wild type and CyDisCo-expressing strains, confirming that the export observed in FIGS. 1 and 2 can be attributed solely to the Tat pathway. Identical results were obtained for TorA-IFN and the TorA-VH domain protein (data not shown).

TorA-scFv is Efficiently Exported and Accurately Processed to the Mature Size.

[0041] The immunoblots shown in FIGS. 1 and 2 show that the four proteins appear to be exported with high efficiency by the Tat pathway, in the sense that the majority of detected protein is in the periplasm. In order to assess the actual export flux, and to additionally test whether the proteins are correctly processed to the mature size, the samples were analysed by Coomassie staining of the periplasm and mass spectrometry of the exported scFv. FIG. 4A shows a Coomassie-stained gel of the periplasmic fractions after export of TorA-scFv in wild type MC4100 cells or the tat null mutant strain, in either the presence or absence of CyDisCo components. The immunoblot (lower panel) shows that the scFv protein is present in the periplasm of the wild type cells and the Coomassie-stained gel (upper panel) shows that this corresponds to an abundant 28 kDa protein that is absent from the periplasm of the tat mutant strain. The presumed scFv band was excised and subjected to trypsin digestion and MALDI TOF as described in Materials and Methods. This generated the peptide sequences shown in FIG. 4B which confirm the protein to be scFv. Importantly, the N-terminal peptide sequences correspond precisely to the sequences immediately following the predicted signal peptidase cleavage site, with no evidence of incorrect processing or large-scale proteolytic clipping. These results provide strong evidence that the precursor protein is accurately processed to the mature size following export to the periplasm, and the relative abundance of the exported scFv, after only a 3 h export assay, shows that the protein is exported at a high rate in both the presence and absence of CyDisCo components.

Tat-Exported hGH and scFv Acquire Disulphide Bonds in the Periplasm

[0042] The four test proteins tested in this study are clearly exported by Tat in the absence of disulphide bond formation and the next study tested whether one of the proteins acquires disulphide bonds in the periplasm. The DsbABCD system catalyses disulphide bond formation in the periplasm (reviewed in Kadokura and Beckwith (2010) Antioxid. Redox Signal 13:1231-1246), usually with proteins that have been exported by the Sec pathway in an unfolded state. In contrast, the Tat pathway almost certainly exports proteins in either a fully folded or near-native state and under such circumstances it is by no means clear whether the Dsb system can access the required thiol groups in order to catalyse disulphide bond oxidation. This issue was investigated by analysis of the periplasmic proteins on reducing vs non-reducing SDS polyacylamide gels—disulphide-bonded proteins usually migrate differently under non-reducing conditions because the presence of disulphide bonds prevents full SDS-dependent unfolding of the proteins.

[0043] FIG. 5A shows immunoblots of periplasmic hGH after export in CyDisCo-expressing or wild type cells. The samples were run under reducing or non-reducing conditions (the latter samples were run in duplicate in case reducing agent diffused to the adjacent lane). hGH run under non-reducing conditions clearly migrates faster than the fully reduced sample, confirming that this protein sample contains disulphide bonds. Importantly, the vast majority of the protein from both CyDisCo and wild type cells runs with this increased mobility (denoted ‘hGH (ox)’), confirming the presence of disulphide bonds and providing strong evidence that hGH exported in wild type cells acquires disulphide bonds in the periplasm. A small proportion of the protein runs further up the gel, possibly indicating formation of some dimer under these conditions.

[0044] It seemed likely that the exported, reduced hGH would acquire disulphide bonds using the DsbABCD machinery that catalyses disulphide bond formation in exported Sec substrates and this was tested using the same approaches. TorA-hGH was expressed in a dsbB.sup.− strain that was previously used to study disulphide bond formation in other exported substrates (Matos et al 2014 (supra)) and the data are shown in FIG. 5B. In the absence of CyDisCo, hGH is exported and the samples migrate as the usual 19 kDa band under reducing conditions (‘hGH (red)’). Importantly, the samples still run primarily with the same mobility under non-reducing conditions, with only a minority migrating more rapidly like the oxidised form (hGH (ox)). These data have been interpreted to indicate that, in the absence of DsbB, the bulk of the exported hGH is unable to form disulphide bonds properly, strongly suggesting that the DsbABCD machinery catalyses the formation of the bonds in the experiment shown in FIG. 1. A minor proportion of exported hGH does appear to contain disulphide bonds, and it is speculated that these may form as a result of spontaneous oxidation.

[0045] Larger amounts of hGH were also purified from periplasmic samples after export in CyDisCo and wild type cells. The protein was then analysed under reducing and non-reducing conditions. FIG. 6 shows that the purified hGH samples from CyDisCo and wild type cells both run as the expected 19 kDa band on reducing gels, whereas the faster-migrating form (containing disulphide bonds) was observed under non-reducing conditions. Importantly, the latter form predominates to an equal extent after purification from either wild type or CyDisCo cells, showing that the periplasmic protein can become essentially fully folded and disulphide-bonded whether the disulphide bonds are formed in the cytoplasm or periplasm. Taken together, the data from FIGS. 5 and 6 indicate that the Tat system can export these proteins even when they lack disulphide bonds and that these bonds can form after entry into the periplasm.

The Tat System can Transport hGH Even when Disulphide Bond Formation is Blocked by Substitution of the Cys Residues Involved

[0046] The above studies imply that the Tat system is transporting a range of substrates in a reduced form, possibly because they are able to attain a near-native structure in the absence of disulphide bond formation. This is an important issue and it was considered important to confirm that a substrate can definitely be transported in the absence of prior disulphide formation. To do this site-specific mutagenesis was used to block disulphide bond formation in the hGH substrate. hGH contains two disulphide bonds: one between Cys53 and Cys165 (a long-range bond between residues remote from each other in sequence), and the other between Cys182 and Cys189 (a short range bond) (FIG. 7A). Substitution of Cys53 by serine (variant C53S) prevents formation of the first bond whereas substitution of Cys189 by serine (C189S) blocks formation of the second. Export of the C53S and C189S single mutants, together with the C53S/C189S double mutant, was tested and the data are shown in FIG. 7. The positions of the above disulphide bonds in hGH are also illustrated.

[0047] The blot shows periplasmic samples from the export assays and the reduced samples (on the left) show the presence of very similar levels of exported periplasmic protein in each case. None of the substitutions adversely affect translocation competence. Analyses of the samples under non-reducing conditions are shown on the right hand side of the Figure. The wild type protein shows the characteristic increase in mobility that accompanies disulphide bond formation and the same increase in mobility is observed with the C189S mutant. On the other hand, the C53S and double mutant do not show this increase in mobility and instead migrate with the same mobility as the reduced samples. From this it may be concluded that the first disulphide bond (disrupted in C53S) is solely responsible for the shift in mobility in SDS polyacrylamide gels whereas disruption of the second bond (in C189S) does not have a significant effect. This is consistent with the protein structure (FIG. 7B). The C182-C189 disulphide bond is a short-range bond whereas the C53S-Cys165 bond is a long range bond that effectively ties together distant regions of the protein. It is therefore unsurprising that the presence of this bond has a major effect on protein mobility when the sample is run under denaturing, non-reducing conditions. Notably, the Tat system exports all these variants and thus appears to recognise the proteins as correctly folded, despite the absence of one or both disulphide bonds.

Discussion

[0048] A number of previous reports (DeLisa et al (supra); Matos et al 2008 EMBO J. 27: 2055-2063; Matos et al 2014 (supra)) have shown that the Tat system has an effective, although poorly understood, quality control system that enables it to identify and reject proteins that are misfolded. This feature is important for the normal functioning of the system: many of its substrates are cofactor-containing proteins and it is important that these are only exported in a folded, assembled state. However, the feature does not only operate during the export of natural Tat substrates, since this selectivity operates towards a variety of heterologous proteins when expressed with Tat signal peptides, including some mammalian disulphide-bonded proteins.

[0049] The Tat pathway has considerable potential for the industrial production of biopharmaceuticals, since it is known to be capable of exporting a number of ‘Sec-incompatible’ proteins and is furthermore capable of exporting them at high rates in fermenter systems (Matos et al 2012 (supra)). Moreover, the inbuilt quality control feature should enable the system preferentially to export proteins that are correctly folded, which is a potentially valuable trait for biotechnological applications. It was believed that the export of disulphide-bonded proteins would be problematic since these cannot form native structures in the cytoplasm, but recent studies have shown that CyDisCo strains offer potential as a means of presenting the Tat system with prefolded, disulphide-bonded substrates that it can export to the periplasm (Matos et al 2014 (supra)). The main aim of the present study was to determine whether this CyDisCo-Tat combination could be used for the production of other disulphide-bonded proteins, especially biotherapeutics.

[0050] This issue was approached by testing for export of human interferon α2b, hGH, an scFv and a VH domain construct. It has been shown that all are exported by the Tat system with high efficiency. Indeed, the export efficiencies observed with some of these substrates are higher than those observed with many native Tat substrates, despite expressing these heterologous substrates at relatively high levels using pET23-based plasmids. However, the major surprise from this study is that none of these substrates require prior disulphide bond formation for export: the proteins are all exported with similar, usually undiminished, efficiencies in wild type cells where the reducing cytoplasm prevents disulphide bond formation prior to export. The likely explanation is that the proteins adopt structures that are sufficiently native-like to be accepted by the Tat system's proofreading system. Indeed, one of the substrates, hGH, is known to adopt a near-native structure in the absence of disulphide bond formation and is even active in this state (Youngman et al (1995) J. Biol. Chem. 270:19816-19822; Bewley et al (1969) Biochemistry 8, 4701-4708).

[0051] These experiments were carried out in a very similar manner to those described in Matos et al 2014 (supra) and the question is why this set of substrates shows such differing requirements for export. Without wishing to be bound by theory, it is believed that the most likely reason relates to the complexity of the proteins studied in previous reports by DeLisa et al (supra) and Matos et al 2014 (supra). PhoA is a 51 kDa protein with two sequential disulphide bonds, while AppA is a 53 kDa protein with 4 disulphide bonds. The dimeric Fab construct studied in the DeLisa et al (supra) report is particularly complex with both intra-chain and inter-chain disulphide bonds. In contrast, the proteins tested in the present study are smaller, with molecular masses of 20.7 kDa (IFN), 23.6 kDa (hGH), 28.9 kDa (scFv) and none contain more than two disulphide bonds. One possibility is that the simpler 3-dimensional structure of the proteins enables them to acquire native, or near-native, structures in the absence of disulphide bond formation.

[0052] Finally, the data have implications for the mechanism of the Tat proofreading system. While this system has a remarkable ability to detect whether substrates are folded, it is clearly not perfect and the four substrates analysed in this study were able to ‘fool’ the translocation pathway and undergo transport. The first point is that the Tat system clearly cannot detect whether disulphide bonds have formed in substrates, at least in these proteins. The second point is that, assuming that all four substrates are not 100% correctly folded in the absence of disulphide bond formation, there appears to be some margin for error in the proofreading mechanism.