SEARCH

SEARCH BY CITATION

Keywords:

  • OmpA;
  • OmpF;
  • OsmY;
  • high-throughput protein expression;
  • extracellular protein expression;
  • Escherichia coli

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and Methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Escherichia coli represents a robust, inexpensive expression host for the production of recombinant proteins. However, one major limitation is that certain protein classes do not express well in a biologically relevant form using standard expression approaches in the cytoplasm of E. coli. To improve the usefulness of the E. coli expression platform we have investigated combinations of promoters and selected N-terminal fusion tags for the extracellular expression of human target proteins. A comparative study was conducted on 24 target proteins fused to outer membrane protein A (OmpA), outer membrane protein F (OmpF) and osmotically inducible protein Y (OsmY). Based on the results of this initial study, we carried out an extended expression screen employing the OsmY fusion and multiple constructs of a more diverse set of human proteins. Using this high-throughput compatible system, we clearly demonstrate that secreted biomedically relevant human proteins can be efficiently retrieved and purified from the growth medium.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and Methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Insolubility, cell toxicity, non-native disulfide pattern, non-native folding, or aggregation problems are commonly encountered when expressing recombinant proteins of interest (“targets”) in bacterial or eukaryotic cell cultures. To overcome such issues, fusion protein approaches are commonly used to facilitate target folding and solubility in E. coli. Examples of such fusion partners are oxidoreductases such as thioredoxin, which may facilitate folding and promote solubility of the attached target in the cytoplasmic space.1 Other tags such as disulfide bond isomerase A not only direct the attached target into the more oxidative environment of the periplasmic space but can also exhibit catalytic activities facilitating the formation of structural disulfide bonds.2 While cytoplasmic and periplasmic expression systems are well established, reports regarding protein export into the growth medium of E. coli (“secretion”) remain scarce and are typically anecdotal and focused on single targets. This is in part due to the assumption that laboratory E. coli strains secrete only a few proteins, among them colicins,3, 4 whereas pathogenic E. coli strains commonly secrete a wide range of proteins such as hemolysins and other toxins.5 Recently, proteomics approaches revealed that a number of endogenous proteins are naturally exported by laboratory E. coli strains BL21 (DE3)6, 7 and K12-derived strains such as W3110.7, 8

Expression of target proteins into the culture medium of E. coli combines the advantages of periplasmic expression,9 easy recovery and the possibility of tailoring the growth conditions for preservation of recombinant protein activity and stability. Traditionally, secretory protein expression is often associated with eukaryotic systems, which require more demanding efforts, running costs, and infrastructure. Moreover, associated post-translational modifications may sometimes interfere with downstream applications, due to heterogeneous sample preparations. To allow for an easy, straightforward and highly cost-effective approach, it is highly desirable to establish E. coli based methods for this purpose as well. Thus, additional comparative information about different secretory signals and their usefulness for protein production is required.

A distinct set of reports describe the successful release of human proteins including epidermal growth factor (EGF),10 growth hormone (GH),11, 12 interleukin-2 (IL-2),13 and granulocyte colony-stimulating factor (G-CSF)6 into the culture medium by the use of certain bacterial signal peptides linked to the N- or C-terminus of the target. Interestingly, outer membrane protein F (OmpF), osmotically inducible protein Y (OsmY) and YebF fusion proteins appear to be transported into the growth medium in a specific manner without affecting outer cell membrane integrity.6, 13, 14 Notably, in the case of OmpF and OsmY the signal peptide alone does not seem to be sufficient for extracellular expression.6, 14 Here, the complete bacterial protein comprising signal peptide and mature part is required for the release into the culture medium. Alternatively, periplasmic signal peptides can be employed for a more unspecific export achieved through permeabilization of the outer cell membrane by additives such as glycine15 or by coexpressing helper proteins such as bacteriocin-release protein or colicin E1 lysin.16 Periplasmic signal peptides are widely used in phage display and scFv antibody production exploiting the favorable oxidizing environment in the periplasm of E. coli for folding of the antibody fusion proteins.17 However, their toxicity to the cell and accumulation in the periplasm may lead to a leaky phenotype with increased outer membrane permeability,18 in some cases providing the opportunity to retrieve recombinant antibodies from the culture medium.19, 20 The use of a signal peptide for a type I secretion pathway such as the hemolysin system was shown to circumvent such problems associated with periplasmic over-expression.21, 22

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and Methods
  7. Acknowledgements
  8. References
  9. Supporting Information

In this study, we compare three signal peptides with respect to their use as fusion partners in a parallel ligation-independent cloning (LIC) and expression platform. Outer membrane proteins (OMPs) are commonly secreted into the periplasm and released into the growth medium dependent on culture conditions.7, 14 Therefore, we tested if outer membrane protein A (OmpA, Swissprot P0A910, residues 1–176), outer membrane protein F (OmpF, Swissprot P02931, residues 1–352), and osmotically inducible protein (OsmY, Swissprot P0AFH8, residues 1–201) can be used as carrier proteins to secrete a set of 24 representative and biomedically interesting human target proteins including signaling molecules, signaling receptor extracellular domains, proteases, and kinases (Supporting Information Table I) into the growth medium of E. coli. The entire OmpF, OsmY, and OmpA proteins including the signal peptide and the mature part linked to a TEV cleavage site and a 6xHis-tag were fused to the N-terminus of selected human target proteins using a LIC approach (vector map, fusion protein sequences, and primers for vector construction in Supporting Information Fig. 1).

thumbnail image

Figure 1. Far Western blot detection of OsmY fusion proteins in culture medium samples (A) and in total expression (cell suspension) samples (B) using a Ni2+-NTA-HRP conjugate. Culture media were purified by Ni2+ chelating chromatography and analyzed by SDS-PAGE (C). Small-scale expression cultures of E. coli BL21 (DE3) pRARE R3 carrying pNIC-BASY constructs were grown in TB medium. A dominant protein band can be detected in most samples at an apparent molecular weight of around 28 kDa (black arrow). Expected molecular weights are summarized in Supporting Information Table I. Recombinant proteins detected in the culture medium and after Ni2+-chelating affinity chromatography purification are marked with an asterisk.

Download figure to PowerPoint

Table I. OsmY Fusion Proteins Purified from 750 mL Expression Culture Medium, Protein Yields and Mass Spectrometry Data
Target, domain fused to OsmY, residue range, number of cysteinesYield of fusion protein (μg)Expected mass without signal peptide (g/mol)aExperimental mass (amu)b
  • a

    Considering all cysteines as reduced.

  • b

    Measured using a Thermo Fisher Scientific Exactive LC-ESI-MS. The mass accuracy is expected to be lower than 0.01% of the detected mass.

  • c

    Mass difference of 302 Da corresponds to the molecular weight of glutathione.

ADAM9, EGF-like, 638-694, 617127,22727,221
FLT4, Ig-like C2-type 2, 136-330, 439043,45243,448
NTRK2, Ig-like C2-type 1, 193-284, 3 (Monomer; 6 in Dimer)109 (Monomer), 562 (Dimer)30,806, 61,61231,108c, 61,604
ACVR1, ECD, 22-123, 1014932,47832,467
NRN1, Neuritin, 28-116, 69433090030,894
CXCL9, Chemokine, 22-96, 412229,38929,385
OSTP1, Osteopontin, 17-314, 0270054,83654,836
ROR2, Kringle, 315-396, 646230,13130,125
ALK, MAM1, 251-426, 430441,05941,054
RIPK2, CARD, 428-528, 1222032,59732,597

Small-scale expression screening experiments in 96-well format using E. coli BL21 (DE3) as expression host showed that green fluorescent protein (GFP), the extracellular domain of human activin receptor type-1 (ACVR1), C-X-C motif chemokine 9 (CXCL9), growth hormone receptor (hGHR), inositol-trisphosphate 3-kinase C (ITPKC), MHC class I polypeptide-related sequence A (MICA) and neuritin (NRN1) could successfully be secreted into the culture medium using OsmY as a fusion protein and purified by immobilized metal affinity chromatography (IMAC; Fig. 1). Angiotensinogen was also detected in the culture medium but could not be purified from 1 mL culture volume. Similar results were observed for OmpA fusion proteins although these were generally less efficiently produced and secreted (Fig. 2). Notably, some OmpA fusion proteins such as OmpA-NRN1, OmpA-ACVR1, and OmpA-CXCL9 were present at relatively high levels indicating that expression of OmpA fusion proteins could be favorable dependent on the target protein. However, although expressed at similar levels as OsmY fusion proteins purification of these OmpA fusion proteins was not successful (Supporting Information Fig. 2) suggesting that structural and biochemical properties of OmpA affect binding of the recombinant proteins to the chelating resin. Of the targets mentioned above, EGF, MICA, and ACVR1 were also detected as OmpF fusion proteins in the culture medium, however expression levels were low (Fig. 2). As protein expression and secretion in prokaryotes appears to be size-limited, the higher molecular weight of OmpF (41 kDa) potentially has a negative effect on fusion protein secretion efficiency compared to the smaller OmpA (23.6 kDa) or OsmY (23.8 kDa) fusions. Furthermore, the C-terminally fused target proteins could interfere with OmpF oligomerization and the trimeric nature of OmpF itself23 could affect extracellular transport, which however contrasts with the large amounts of OmpF-β-endorphin (40 kDa) previously reported to be secreted into the culture medium of an E. coli BL21 descendant.14 We also tested expression vectors harboring the trc promoter instead of the araBAD promoter to increase expression levels. For a few constructs such as OsmY-NRN1 and OsmY-CXCL9, higher amounts of purified recombinant protein could be isolated using the trc promoter (Supporting Information Figs. 2 and 3).

thumbnail image

Figure 2. Far Western blot detection of OmpA and OmpF fusion proteins in culture medium samples after small-scale expression using a Ni2+-NTA-HRP conjugate. Cultures of E. coli BL21 (DE3) pRARE R3 carrying pNIC-BASA and pNIC-BASF constructs were grown in TB medium. The expected molecular weights are summarized in Supporting Information Table I.

Download figure to PowerPoint

thumbnail image

Figure 3. Normalized levels of all OsmY fusion proteins that could be purified from 0.5 mL supernatant of 0.96 mL expression cultures using Ni2+ chelating sepharose 6FF. Percentage densitometric values were determined using the program ImageQuantTL (GE Healthcare). The average of the densitometric values of the 62, 49, 38, 28, 14 kDa bands of the SeeBlue Plus2 pre-stained marker (Invitrogen, Carlsbad) was set to be 100%. Five microliter SeeBlue Plus2 marker and 12 μL protein samples were loaded on the lanes of NuPAGE® Novex 4-12% Bis-Tris Midi gels (Invitrogen, Carlsbad) which were run in MES buffer. Constructs marked by a diamond were used for large-scale expressions.

Download figure to PowerPoint

Based on the initial results, a broader screen was set up using OsmY fusion proteins expressed under the control of the araBAD promoter. We generated a set of multiple expression constructs for the same target by N- and C-terminal amino acid truncations as described previously.24 In this approach, largely driven by the experience of the scientist and by consideration of guidelines such as the N-end rule,25 several expression constructs are designed for the domain of interest varying in a few residues at the N- and C-termini of the predicted domain boundary. Implementation of such multiple constructs has been shown to improve the success rate in soluble protein production and crystallization using intracellular expression.24 The expanded secretory expression screen included 192 constructs encoding for 39 human proteins such as a disintegrin and metalloproteinase (ADAM), colony-stimulating factor 2 and 3 (CSF-2, G-CSF), fibroblast growth factor 18 (FGF18), Insulin-like growth factor-binding protein 2 (IBP2), interleukin 2 and 5 (IL-2, IL-5), tumor necrosis factor ligand superfamily member 10 (TNFS10) and osteopontin (OSTP-1), as well as a variety of other growth factor receptor, ribosyltransferase, nuclease, phosphatase, helicase, and extracellular ligand domains (Supporting Information Table II). OsmY fusion proteins could be isolated for 72 constructs representing 25 targets of different protein classes (Fig. 3). OsmY-OSTP-1 showed the highest expression levels; single domains of extracellular receptors such as the Ig-like C2-type domain of Ntrk2, the EGF-like domain of ADAM9 and the kringle domain of ROR2 were expressed in lower amounts. All of these extracellular domains are rich in disulfide bonds making them interesting targets for secretory expression assuming that native disulfide bonding patterns are more likely to be formed when the proteins pass the E. coli transport and chaperone machineries as compared to standard intracellular expression approaches.2, 26 Furthermore, the results demonstrate that different N- and C-terminal boundaries for the same protein domain lead to different expression results indicating that secretory expression capabilities can be optimized by applying a multiconstruct approach, similarly as described for intracellular expression.24 However, in this study, we cannot observe any clear trends regarding the properties of a successfully expressed and soluble protein construct and how this compares with those less favorable.

To verify that OsmY fusion proteins isolated from small-scale expression cultures could also be purified in a larger scale, we set up 750 mL expression cultures and purified to homogeneity recombinant proteins from the growth media by a protocol using Ni2+-chelating and size-exclusion chromatography (Fig. 4). We could obtain highly purified protein samples for different groups of human target proteins fused to OsmY, as judged by SDS-PAGE analysis. The experimentally determined masses correspond to the expected molecular weight showing that cysteine residues are present in the oxidized state forming disulfide bridges (Table I). Furthermore, mass spectrometry results show that the OsmY signal peptide comprising 28 amino acids was naturally cleaved from all purified proteins. In all expression experiments a protein with an apparent molecular weight of 28 kDa was copurified during Ni2+-chelating chromatography, likely generated from partial proteolytic cleavage of the secreted fusion protein. Fractions of the “28 kDa protein” were collected after gel filtration and subjected to MS analysis, demonstrating that in all cases the copurified protein indeed represents OsmY (without signal peptide) linked to the His-Tag and the TEV site (Supporting Information Table III).

thumbnail image

Figure 4. OsmY fusion proteins can be isolated and purified by Ni2+-chelating affinity and size exclusion chromatography. Equal amounts (1 μg) recombinant fusion proteins were analyzed by SDS-PAGE. Expected and experimental masses are summarized in Table I.

Download figure to PowerPoint

CD measurements show that the purified target proteins are folded [Fig. 5(A) and Supporting Information Table IV]. Interestingly, unfolding by heating appears to be reversible for OsmY-His [Fig. 5(B)] and thus, it is possible that OsmY could also facilitate the solubility, stability, and fold of proteins attached to it by means of a rigid scaffold. As a measure of biological activity the interactions of OsmY-ACVR1 and OsmY-NTRK2 to their respective binding partners BMP-6 and BDNF were examined using surface plasmon resonance (SPR). Binding experiments show that both OsmY-ACVR1 and OsmY-NTRK2 are binding-active demonstrating that the OsmY-target fusion proteins were obtained in a biologically active state [Fig. 5(C)]. Furthermore, two of the purified OsmY fusion proteins, OsmY-NRN1 and OsmY-CXCL9, were subjected to cleavage with TEV protease showing that the fusion proteins can be cleaved and that the target protein can efficiently be collected from the flow-through of a subsequent Ni2+-chelating Sepharose chromatography step (Fig. 6). An electrophoretic mobility shift to a higher apparent molecular weight observed for the final batch of CXCL9 in non-reducing SDS-PAGE, further indicates that cysteine residues in CXCL9 are oxidized and of structural relevance.

thumbnail image

Figure 5. Characterization of proteins purified from large-scale expression cultures. Target proteins are folded as shown by CD spectroscopy (A). Thermal unfolding of OsmY as measured by the change in ellipticity is reversible (B). The same OsmY-His sample was used for CD measurements at 20°C (“native”) and at 96°C (“unfolded”). Subsequently, the sample was cooled to 20°C to check for reversibility of unfolding (“refolded”). The OsmY fusion proteins are biologically active as demonstrated by surface plasmon resonance measurements (C). Both ACVR1-OsmY and monomeric NTRK2-OsmY bind their immobilized ligands BMP-6 (left diagram) and BDNF (right diagram), respectively. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

thumbnail image

Figure 6. TEVsh protease cleavage of OsmY-NRN1 (A) and OsmY-CXCL9 (B) fusion proteins. The following samples were loaded on the lanes of NuPAGE SDS-PAGE gels (Invitrogen, Carlsbad) in (A) and (B) 1: TEVsh protease, 2: OsmY fusion protein before adding TEVsh, 3: protein mixture after TEVsh treatment, 4-6: Flow-through fractions of IMAC containing the target protein NRN1 or CXCL9 respectively, 7-9: Elution fractions of IMAC containing the OsmY-His peptide. The final protein batches were analyzed by SDS-PAGE (C). Lanes in (C) are as follows: 1-3: CXCL9, 4-6: NRN1. 3 μg (lanes 1 and 4) or 1μg (lanes 2, 3, 5, 6) of protein was analyzed. The plus and minus signs indicate the presence or absence of β-mercaptoethanol in the sample loading buffer. A mobility shift is observed for CXCL9 when analyzing the sample under nonreducing conditions.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and Methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Secretion of recombinant proteins into the extracellular environment of E. coli has been reported for a number of human proteins such as GH,11, 12 EGF,10, 27 G-CSF,6 IL-2,13 leptin,6 and IL-628; some of these proteins could be isolated in a biologically active form in yields ranging from less than 1 mg to 325 mg per 1 L culture medium in high-density fermentations. The comparatively high yields, the apparent simplicity of purifying the target proteins by affinity chromatography without the need of disrupting cells and the prospect of expressing correctly folded disulfide-rich proteins prompted us to investigate the application of secretory (extracellular) expression for biomedically relevant human proteins in a systematic manner. Outer membrane permeabilization is normally required in order to release the target protein into the growth medium when using the short signal peptides of OmpA, OmpF, PhoA, penicillinase, or other E. coli proteins naturally transported into the periplasm. However, an increased number of undesirable proteins may be released together with the protein of interest if the outer membrane is disrupted. Therefore, we aimed at avoiding the use of membrane-permeabilizing molecules such as glycine or detergents15, 29 or the coexpression of helper proteins such as kil or bacteriocin release protein11, 29 to minimize the background of native E. coli proteins and to avoid the release of potentially unfolded or aggregated target proteins.

Experiments have shown that the entire OmpF, OsmY, or YebF fusion proteins comprising signal peptide and mature part are required to export a target protein attached to the C-terminus of any of these bacterial carrier proteins.6, 13, 14 These studies demonstrated that only a minor fraction of the fusion protein released into the culture medium resulted from unspecific events such as outer membrane (periplasmic) leakage. However, an experimentally proven mechanism for the transport of recombinant proteins into the culture medium could not be provided.

Generally, the terminal branches of the secretion machineries in gram-negative E. coli can be classified into five to six categories dependent on the underlying mechanism.30, 31 Most of the mechanisms such as type I (hemolysin) or type III secretion are not found in nonpathogenic laboratory strains of E. coli. Genes encoding for proteins part of the general secretion (Gsp, type II) or outer transmembrane pore (type V, auto-transportation) pathways are however present in laboratory E. coli strains and potentially part of the protein export machinery.32 OMPs like OmpA and OmpF exhibiting an anti-parallel β-barrel structure23, 33 insert into the outer membrane following expression as soluble, non-native intermediates,34, 35 often in complex with molecular chaperones such as Skp36 or the oxidase DsbA.37 Folding into the native state and insertion into the outer membrane is triggered by contact with lipopolysaccharides (LPS) forming the outer leaflet of the outer membrane.34, 38 Crowding of the periplasmic space as a result of outer membrane protein over-expression could lead to an unspecific release of surplus protein either directly or through vesicles (blebbing) and interestingly, association of OMPs with foldases such as Skp seems to facilitate their incorporation into vesicles.39, 40 It has previously been shown that less than 1% of total outer membrane protein, including OmpA and OmpF, are released in the form of vesicles under normal growth conditions in gram-negative E. coli.41 However, this portion could well be increased as a consequence of cellular stress, such as during protein over-expression.

OsmY, the carrier protein finally chosen for our extended expression studies, is structurally and functionally different to OmpA and OmpF (Supporting Information Fig. 4) and likely exerts its function by contacting the phospholipid interfaces at membranes surrounding the periplasmic space.42, 43 Unlike the OMPs, OsmY consist of two bacterial OsmY and nodulation (BON) homology domains. Recently, the NMR structure of the outer membrane protein Rv0899 of Mycobacterium tuberculosis was solved revealing that its BON homology domain follows a αββαβ-fold,44 which is in agreement with our CD measurements of OsmY [see Fig. 5(B)]. Other recent results suggest that another carrier protein, YebF, which has previously been used to secrete human IL-2 into the growth medium of E. coli,13 could also fold into a mixed α-helical and β-stranded structure.45 In conclusion, it seems likely that the fusion proteins with OsmY and OmpA/OmpF are released into the culture medium by different mechanisms due to structural and biochemical differences. Further experiments are needed to elucidate the exact mechanisms allowing extracellular transport of OsmY fusion proteins in E. coli.

Some target proteins expressed and purified as OsmY fusion proteins in our study could not be obtained in soluble form before using standard E. coli expression methods. The C-X-C chemokine CXCL9 (MIG), which elicits important functions in immune response, can reportedly be isolated from E. coli inclusion bodies46 and using CHO expression systems.47 We demonstrate that human CXCL9 can be produced in soluble form as OsmY fusion protein. CXCL9 isolated after TEV cleavage of OsmY-CXCL9 is highly pure and biochemical examinations indicate that it was produced in a biologically relevant form. Recent results show that the CXCL9 we purified binds streptococcal inhibitor of complement (SIC; W. Streicher and M. Wikström, personal communication), which is a protein naturally secreted by pathogenic Streptococcus pyogenes counteracting the antibacterial activity of chemokines such as CXCL9.48 Furthermore, using the trc promoter instead of the araBAD promoter yields of up to 80 mg fusion protein per 1 L of growth medium could be achieved. Future experiments should show if other chemokines of the C-X-C family are also accessible using the approach presented here.

NRN1 (CPG15, neuritin) is a naturally secreted ligand involved in embryonic and adult neurogenesis.49 Fusing full-length mature human NRN1 to OsmY provided soluble protein that could be cleaved and purified to homogeneity whereas the expression of NRN1 in the cytoplasm of E. coli was not successful (results not shown). Similar results could be achieved for the extracellular domain of ACVR1, which can be isolated from Sf9 insect cells50 but not from Pichia pastoris.51 To our knowledge, soluble expression and purification of biologically active ACVR1 from E. coli has not been reported so far.

In addition to these three proteins, which have a low molecular weight (CXCL9: 8249 Da, NRN1: 9760 Da, ACVR1: 11338 Da), other extracellular human proteins of higher molecular weight such as GHR (28398 Da), ITPKC (31139 Da) and MICA (32687 Da) could be isolated suggesting that the OsmY fusion is useful for other human target proteins as well. In an extended screen a variety of other biologically interesting proteins could be expressed and purified in small-scale. Some of these targets were scaled-up depending on the amount of purified protein detected after small-scale IMAC purifications. Not all of the selected, scaled-up proteins could be isolated in the yields suggested from small-scale expression and IMAC purification experiments (see Table I and Fig. 3). Discrepancies were observed for OsmY-ROR2 and OsmY-RIPK2 yielding 0.5 mg and 2.2 mg, respectively, in the scale-up experiment. Notably, the full-length human phosphoglycoprotein OSTP1 (SPP1, osteopontin) could be purified in high amounts and several constructs tested showed minor deviations in expression and purification levels. Taking into consideration the relatively high molecular weight of the OsmY-OSTP1 fusion protein of 57.7 kDa this was somewhat surprising since smaller proteins such as OsmY-ADAM9-EGF-like (30.1 kDa) and OsmY-ACVR1 (35.4 kDa) were obtained in lower yields (see Table I) indicating that in addition to the molecular weight other factors related to the biochemical properties of the target protein, such as amino acid composition and the presence of disulfide bridges, contribute to extracellular expression levels. Mass spectrometric analysis of purified OsmY-OSTP1 shows that the protein is not phosphorylated, which correlates to literature reports.52–54

In conclusion, we have generated a versatile high-throughput compatible E. coli based secretory system, which has proven efficient to produce “hard-to-get” human proteins of medical and biological importance in a biologically relevant form. Secretion into the culture medium of expression cultures simplifies the purification process and enables tailoring of the growth medium to keep the recombinant protein of interest in a soluble and active form, for example by adding stabilizing molecules. Our study clearly shows that the OsmY fusion system is applicable to a large number of biomedically relevant and normally challenging human proteins.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and Methods
  7. Acknowledgements
  8. References
  9. Supporting Information

E. coli strains, cloning material, chemicals, and culture media

Cloning experiments were carried out using E. coli Mach1 T1 cells (Invitrogen, Carlsbad). Restriction enzymes and deoxynucleotides were purchased from NEB (Ipswich). Chemicals were from Sigma-Aldrich (St. Louis). Entry clones of the human target proteins serving as templates for amplification of DNA inserts were acquired from cDNA libraries including the mammalian gene collection (MGC; Imagenes, Germany), ORFeome collection (Imagenes, Germany) and Ultimate ORF collection (Invitrogen, Carlsbad). Plasmid DNA and PCR purification kits were from Macherey-Nagel (Düren, Germany) and DNA maxi preparations were carried out using the JetStar Maxi kit from Genomed (Löhne, Germany). Protein expression experiments were done with E. coli BL21 (DE3) pRARE2 R3.55 Depending on the experiment, cells were grown in LB, TB, or M9 minimal medium56 supplemented with the appropriate antibiotics and with 8 g/L (TB) or 9.5 g/L (M9) glycerol as carbon source, respectively.

Expression vectors

Expression vectors for excretory expression are based on the pNIC28-Bsa4 vector (GenBank accession code EF198106) containing the T7lac promoter, N-terminal His-tag, TEV protease cleavage site, and a SacB negative marker flanked by two BsaI sites. Using site-directed mutagenesis, an AflII site was introduced upstream of the T7lac promoter cassette. Using this AflII site and the existing NdeI site, DNA fragments encoding the araBAD promoter or the trc promoter cassettes, respectively, were introduced thereby replacing the T7lac promoter cassette. The DNA sequences encoding OmpA, OmpF, and OsmY were amplified from E. coli BL21 (DE3) genomic DNA and cloned into the NdeI site of the expression vector. Constructs harbouring promoters and excretory signals in correct orientations were sequence verified. Primers, sequences of the OmpA, OmpF, and OsmY tags, a vector map and a generalized plasmid construct can be found in Supporting Information Fig. 1.

Ligation-independent cloning

Domain boundaries of the target proteins were chosen based on secondary structure predictions.57, 58 Primers for amplification of the DNA fragments of interest from entry clones were designed using appropriate software58 and were flanked by LIC sites (5′ LIC sequence: 5′-TACTTCCAATCCATG; 3′ LIC sequence: 5′-TATCCACCTTTACTGTCA). The expression vectors were cleaved with BsaI, treated with T4 DNA polymerase in the presence of dGTP and used for LIC of the target DNA fragments as described previously.55 Single E. coli Mach1 T1 (Invitrogen, Carlsbad) colonies harbouring the LIC constructs were analyzed by colony PCR and overnight cultures of colonies containing an insert of correct size were used for plasmid mini preparations. Finally, the constructs were transformed into the expression host, and glycerol stocks were prepared and stored at −80°C.

Small scale expression and purification

Overnight cultures were grown from glycerol stocks in a micro-expression shaker incubator (Glas-Col, Terre Haute) at 1000 rpm and at 30°C. The GlasCol shaker allows the growth of microliter cultures to high optical densities facilitating the identification of proteins that can successfully be expressed in large-scale fermentations.59 Small-scale expression cultures were set up in 96-deep square well (ABgene, Epsom, UK) or 24-deep round well plates (Whatman, Kent, UK) containing 0.96 mL or 5 mL TB medium per well, respectively. The expression cultures were inoculated with 1/100 (v/v) overnight culture and grown at 37°C and 1000 rpm (96-well plate) or 400 rpm (24-well plate). At an optical density at 600 nm of around 1.0 the Glas-Col shaker was cooled to 18°C. After 1 h, at optical densities at 600 nm between 1.5 and 2.0, protein production was induced by adding 0.05% (w/v) L-(+)-arabinose and continued at 18°C overnight. Optical densities of all plate wells were measured at induction and after overnight expression using a plate reader. Samples of cell suspensions and expression culture media were used for SDS-PAGE on E-PAGETM 48 8% gels (Invitrogen, Carlsbad). The gels were blotted using the iBlot® and regular midi gel nitrocellulose transfer stacks (both from Invitrogen). Detection of His-tagged protein was carried out using Ni2+-NTA-HRP conjugate (Pierce, Rockford) and SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford), visualized with a Bio-Rad Fluor-S® MultiImager, or using 3,3′,5,5′-Tetramethylbenzidine (TMB 1-C, KPL, Gaithersburg) for colorimetric staining and subsequent scanning. The BenchMarkTM His-tagged protein standard (Invitrogen, Carlsbad) was loaded as positive control.

For small-scale purifications, 50 μL of a 50% slurry of Ni2+ Chelating Sepharose Fast Flow (GE Healthcare) in binding buffer (30 mM Hepes, 300 mM NaCl, 10% (v/v) glycerol, 10 mM imidazole), was added to the wells of a Millipore 96-well filter plate (1.2 μm). Subsequently, 150 μL culture medium from an overnight expression was added to a Millipore 96-well filter plate (0.65 μm) and filtrated into the 1.2 μm-plate containing the resin by applying vacuum. The 1.2 μm-plate was sealed with a CrystalClear seal (Hampton Research, Aliso Viejo) and placed in a vertical position in a Glas-Col shaker (400 rpm, 18°C, 15 min.). Subsequently, the liquid was filtrated into a waste plate and the filtration and binding steps were repeated with another 150 μL culture medium. For washing, 250 μL washing buffer (30 mM Hepes, 300 mM NaCl, 10% (v/v) glycerol, 30 mM imidazole) was added to the resin. The suspension was mixed in a Glas-Col shaker (400 rpm, 18°C, 2 min.) and the liquid was removed by centrifugation (300g, 4°C, 2 min.). The washing step was repeated two times. Finally, 40 μL elution buffer (30 mM Hepes, 300 mM NaCl, 10% (v/v) glycerol, 500 mM imidazole) was added to the wells. After an incubation period of 15 min. on a horizontal shaker the eluate containing the recombinant protein was recovered by centrifugation (300g, 4°C, 2 min.). Protein samples were analyzed by SDS-PAGE using NuPAGE® Novex 10% Bis-Tris Midi gels (Invitrogen, Carlsbad). The same procedure was applied for the purification of culture supernatants from 5 mL expression cultures except that a total of 4 mL culture supernatant was purified using 200 μL chelating resin treated with 1 mL washing and 300 μL elution buffers in the order mentioned above.

Large-scale cultivation and purification

An overnight culture in 10 mL TB medium supplemented with 50 μg/mL kanamycin and 34 μg/mL chloramphenicol was inoculated with a glycerol stock of E. coli BL21 (DE3) pRARE2 R3 harboring the expression construct and grown in an Infors shaker incubator (Infors HT, Bottmingen, Switzerland) at 150 rpm and 30°C. The expression culture in 750 mL M9 minimal medium56 or TB medium supplemented with the appropriate antibiotics was inoculated with 1/100 (v/v) overnight culture and grown in a shaker at 150 rpm and 37°C. At an optical density at 600 nm of 0.7–0.8 the cultivation flask was transferred to another shaker pre-cooled to 18°C and grown for 1 h. Subsequently, 0.05% (w/v) L-(+)-arabinose was added and the culture was further incubated at 18°C overnight. Cells were harvested by centrifugation at 4°C and frozen for storage. The supernatant was supplemented with 1 tablet Complete EDTA-free protease inhibitor cocktail set (Roche, Indianapolis) per 1 L culture medium. After adjusting the pH value to ph = 7.5 the solution was filtrated through a 0.22 μm bottle top filter (TPP). Subsequently, 5 mL of a 50% suspension of Ni2+ chelating Sepharose 6FF (GE Healthcare) in binding buffer (30 mM Hepes, 300 mM NaCl, 10% (v/v) glycerol, 10 mM imidazole) was added to the filtrate. After incubating on a horizontal shaker at 4°C for 1 h the suspension was filtrated over an Econo glass column (Bio-Rad, Hercules) and the resin was washed with two times 12 CV binding buffer followed by two times 12 CV washing buffer (30 mM Hepes, 300 mM NaCl, 10% (v/v) glycerol, 20 mM imidazole). His-tagged proteins were eluted by 2 CV elution buffer (30 mM Hepes, 300 mM NaCl, 10% (v/v) glycerol, 0.5 M imidazole).

A 26/60 SD75 prep grade size exclusion column (GE Healthcare) was equilibrated with running buffer (30 mM Hepes, 300 mM NaCl, 10% (v/v) glycerol, pH 7.5) at a flow rate of 2 mL/min. The eluates from the affinity chromatography step (5 mL) were applied to the column and the collected fractions were analyzed by SDS-PAGE and pooled accordingly. The proteins were concentrated using Millipore centrifugal concentrators (3.5 kDa cutoff), aliquoted and stored at −80°C.

Mass spectrometry

Purified proteins were desalted and concentrated using ZipTip C4 pipette tips (Millipore, Billerica) according to a protocol provided by the manufacturer. The eluted proteins, dissolved in 50% (v/v) methanol supplemented with 0.1% formic acid, were analyzed on an Exactive nLC-MS system (Thermo Scientific Waltham) using direct infusion. Deconvoluted mass spectra were obtained using the Xtract program within the Xcalibur software package and the program ProMass Deconvolution (Novatia, Monmouth Junction).

Circular dichroism

Purified proteins were dialyzed against 10 mM potassium phosphate buffer pH 7.2 and centrifuged prior to circular dichroism measurements performed on a Jasco J-815 spectrometer (Jasco, Easton) equipped with a temperature control device. The sample volume in the 1 mm micro cuvette (Hellma, Müllheim, Germany) was 200 μL. For each sample five spectra were acquired for a wavelength range between 300 nm and 185 nm in continuous scanning mode (100 nm/min, bandwidth 1 nm). Spectra were averaged and a blank measurement (dialysis buffer) subtracted. Protein unfolding was assessed by monitoring the ellipticity at 222 nm while heating the sample from 20°C to 96°C (heating rate: 2°C min−1, steps of 1°C). Spectra at 96°C were recorded as described. The sample was then cooled to 20°C and spectra were acquired to examine whether unfolding is reversible. The mean residue ellipticity (MRE) was calculated according to the following formula: MRE = E × 100/(c × d × N) [deg cm2 dmol−1] (E: ellipticity in deg, c: concentration in mol L−1, d: path length in cm, N: number of amino acid residues). To obtain spectra of the target proteins the CD spectrum of the OsmY-His peptide was subtracted from the CD spectrum of the OsmY-target fusion protein. Melting temperatures were determined using a sigmoidal curve fit.

Surface plasmon resonance

Analyte proteins were diluted with HBS-EP+ buffer (GE Healthcare) to the required concentration. Measurements were carried out on a Biacore T100 system (GE Healthcare) using HBS-EP+ as running buffer. Ligands were purchased from R&D systems, dissolved or diluted in 10 mM sodium acetate pH 4.5 and immobilized on CM5 sensor chips by direct amine coupling. Association and dissociation times were set to 360 s. The flow rate was 10 μL/min. Regeneration of the biosensor surface was accomplished by washing with 10 mM glycine pH 2 or 4 M MgCl2 for 2 min. Simple binding experiments were carried out by perfusing with 200, 400, and 800 nM solutions of the analyte (OsmY fusion protein) in running buffer. Sensorgrams were corrected by subtracting the response from a reference flow channel without immobilized ligand. As a control solution of 200, 400, and 800 nM OsmY-His peptide (fusion tag after TEV cleavage) were analyzed in parallel showing a very low or baseline response compared to the measurement with the fusion protein.

TEV cleavage of OsmY fusion proteins

TEVsh protease60 (2 mg mL−1) was added in a molar ratio of 1:20 (TEVsh:OsmY fusion protein) to the protein pools obtained after IMAC and gel filtration. The mixture was incubated at 4°C over night and purified using a 1 mL HiTrap Ni2+ Chelating Sepharose column. The target protein was retrieved from the column flow-through while TEVsh protease and the OsmY-His peptide remained bound to the chelating resin. Protein bound to the resin was eluted with running buffer (30 mM Hepes, 500 mM NaCl, 10% glycerol, 5 mM imidazole, pH 7.5) containing 0.5 M imidazole. Fractions were combined according to SDS-PAGE analysis. The protein pools were dialyzed against 30 mM Hepes, 500 mM NaCl, 10% glycerol, pH 7.5 and analyzed by SDS-PAGE and mass spectrometry.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and Methods
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank P. W. Thulstrup at the Faculty of Life Sciences, University of Copenhagen, for giving access to the CD spectrometer.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and Methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and Methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
PRO_593_sm_suppinfofigure1.tif1294KSupporting Information Figure 1
PRO_593_sm_suppinfofigure2.tif3098KSupporting Information Figure 2
PRO_593_sm_suppinfofigure3.tif199KSupporting Information Figure 3
PRO_593_sm_suppinfofigure4.tif1012KSupporting Information Figure 4
PRO_593_sm_suppinforeferences.doc30KSupporting Information References
PRO_593_sm_suppinfotable1.doc47KSupporting Information Table 1
PRO_593_sm_suppinfotable2.doc87KSupporting Information Table 2
PRO_593_sm_suppinfotable3.doc44KSupporting Information Table 3
PRO_593_sm_suppinfotable4.doc86KSupporting Information Table 4

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.