Protein co-expression is a powerful alternative to the reconstitution in vitro of protein complexes, and is of help in performing biochemical and genetic tests in vivo. Here we report on the use of protein co-expression in Escherichia coli to obtain protein complexes, and to tune the mutation frequency of cells.
We report here that the expression of protein complexes in vivo in Escherichia coli can be more convenient than traditional reconstitution experiments in vitro. In particular, we show that the poor solubility of Escherichia coli DNA polymerase III ε subunit (featuring 3’-5’ exonuclease activity) is highly improved when the same protein is co-expressed with the α and θ subunits (featuring DNA polymerase activity and stabilizing ε, respectively). We also show that protein co-expression in E. coli can be used to efficiently test the competence of subunits from different bacterial species to associate in a functional protein complex. We indeed show that the α subunit of Deinococcus radiodurans DNA polymerase III can be co-expressed in vivo with the ε subunit of E. coli. In addition, we report on the use of protein co-expression to modulate mutation frequency in E. coli. By expressing the wild-type ε subunit under the control of the araBAD promoter (arabinose-inducible), and co-expressing the mutagenic D12A variant of the same protein, under the control of the lac promoter (inducible by isopropyl-thio-β-D-galactopyranoside, IPTG), we were able to alter the E. coli mutation frequency using appropriate concentrations of the inducers arabinose and IPTG. Finally, we discuss recent advances and future challenges of protein co-expression in E. coli.
The introduction of overexpression technologies boosted either the biochemical studies of low-copy number enzymes and the industrial production of pharmacologically active proteins (e.g., insulin). Since the advent of these technologies, significant advances have been achieved to increase the yield and quality of recombinant proteins. In addition, prokaryotic1,2 and eukaryotic3 overexpression systems were developed over the years, offering useful alternatives to the “work horse” of protein biotechnology, i.e., Escherichia coli. In particular, the availability of alternative platforms to E. coli led to the production of recombinant peptides or proteins bearing post-translational modifications. However, it should be mentioned that E. coli still represents the organism of choice for recombinant protein production. This is due to several factors, among which the most relevant can be considered: i) the availability of quite a number of overexpression systems (expression vectors and strains) for E. coli1,2; ii) the short generation times, and high biomass yields, of E. coli in a variety of rich and synthetic media; iii) the facile manipulation either at the biochemical and at the genetic level of this microorganism; iv) the isolation of strains capable of the production of toxic proteins4; v) the construction of strains featuring homogeneous induction at the population level5,6. In addition, it was recently shown that expression systems suitable for the production in E. coli of post-translationally modified proteins can be devised and constructed2.
At present, protein overexpression is mainly used to obtain monomeric or homo-oligomeric proteins, whose hypersynthesis can be performed with a single gene cloned into an appropriate plasmid. However, attention was recently paid to the construction of E. coli protein co-expression systems, challenging the production, in vivo, of hetero-oligomeric complexes2. Interestingly, early experiments of protein co-expression addressed the inter-species assembly of large and small subunits of cyanobacterial ribulose-1,5-bisphosphate carboxylase/oxygenase7,8, and the association of truncated and full-length forms of HIV-1 reverse transcriptase9. These pioneering studies demonstrated that protein co-expression represents a powerful alternative to traditional in vitro reconstitution. In addition, protein co-expression in E. coli was used to produce different proteins bearing post-translational modifications2, to obtain proteins containing unnatural amino acids2, and to increase the yield of overexpressed membrane proteins2. Moreover, the potential of protein co-expression as a tool to confer to E. coli competence in protein secretion is under active investigation2.
Two main strategies of protein co-expression in E. coli can be pursued: i) the use of a single plasmid to host the different genes to be overexpressed; ii) the use of multiple plasmids in single cells to co-express the target proteins. In the first case, the criteria for the choice of plasmid do not differ from those of traditional single protein overexpression experiments, although particular plasmids containing tandem promoter/operator elements were constructed for co-expression10. This first approach is therefore quite simple. However, it should be mentioned that the use of a single plasmid to co-express different proteins faces two major difficulties: i) the molecular mass of the vector increases with the number of hosted genes, limiting the number of co-expressed proteins; ii) when multiple genes are cloned under the control of a single promoter, polarity can decrease the expression of the genes distal from the promoter. The use of dual or multiple plasmids in single E. coli cells has to accomplish the compatibility of the vectors of choice, therefore imposing constraints to the eligible combinations of plasmids. However, this second co-expression strategy features the advantage of containing the molecular mass of vectors, and limits polarity. We recently constructed a protein co-expression system designed to facilitate the shuttling of genes between the co-expression plasmids11. In particular, we constructed the pGOOD vectors series, the relevant features of which are: i) a p15A origin of replication, to provide compatibility of the pGOOD plasmids with the commercial vectors containing the ColE1 origin (e.g., the pBAD series12); ii) a tetracycline-resistance cassette; iii) the presence of lac-derived regulatory elements, i.e., the Promoter-Operator(O1) couple and the lacIq gene. Using an appropriate pBAD-pGOOD couple, we were able to overexpress the catalytic core of E. coli DNA polymerase III, composed of three different subunits, i.e., α (the 5’-3’ polymerase), ε (the 3’-5’ exonuclease) and θ (stabilizing ε)13. In particular, we demonstrated that the co-expression of the αεθ complex was strictly dependent on the addition to the E. coli culture medium of both IPTG and arabinose, triggering overexpression from pGOOD and pBAD, respectively (Figure 1A).
In the present report, we illustrate how protein co-expression can efficiently solve difficulties linked to the poor solubility of a protein complex subunit. In addition, we show how in vivo protein complementation tests can be performed, and we finally report on the use of protein co-expression to tune mutation frequency in E. coli. To this aim, we used pGOOD-pBAD couples suitable to illustrate relevant examples of each case study.
1. Isolation of E. coli Co-transformants
2. Co-expression of α and ε Subunits of E. coli DNA Polymerase III
3. Co-expression of α Subunit of Deinococcus radiodurans DNA Polymerase III and ε Subunit of E. coli DNA Polymerase III
4. Gel Filtration Chromatography
5. Mutation Analysis of Populations Co-expressing the Wild-type ε Subunit of E. coli DNA Polymerase III and the Mutagenic εD12A Variant
The ε subunit of E. coli DNA polymerase III consists of 243 amino acids and features poor solubility17,18, unless the residues 187-243 are deleted17. However, we have previously shown11 that the co-expression of full-length α, ε, and θ subunits yields soluble DNA polymerase III catalytic core (Figure 1). In particular, using the pBAD-pGOOD co-expression system, we demonstrated that: i) the overexpression of α and ε subunits can be independently controlled by arabinose and IPTG, respectively (Figure 1A); ii) full-length ε subunit can be detected in soluble protein extracts isolated from E. coli cells overexpressing α, ε, and θ and subjected to gel-filtration (Figure 1B). In this case, exonuclease activity was distributed into 3 major peaks (Figure 1B, filled circles), and the peak centered at fraction 23 was shown to contain the αεθ catalytic core (Figures 1C and 1D). The stability of ε is greatly increased upon binding to α and θ. When we previously overexpressed full-length ε in E. coli, a low amount of free ε was detected by western blotting in soluble protein extracts19 (Figure 2A). It is interesting to note that, under the same conditions, we always detected higher amounts of free C-ter truncated forms of ε (ε-234, ε-228, ε-213, or ε-186), among which ε-186 performed better19 (Figure 2A). We did also show that the proteolysis of ε C-terminal domain is prevented by the binding to α and θ subunits19. The effect, if any, of co-expression on solubility of ε can be easily tested. As reported in Figure 2B, soluble protein extracts isolated from cells overexpressing α, ε, or α and ε can be analyzed by SDS-PAGE, and conveniently compared with total protein extracts. In this case, the electrophoretic analysis indicates that, when overexpressed alone, ε subunit features poor solubility (compare lanes 1 and 2), while the co-expression of α greatly enhances the concentration of ε in soluble protein extracts (compare lanes 3 and 4).
Co-expression can also be used to perform inter-species complementation tests. Accordingly, we thought it might be of interest to evaluate the interaction between subunits belonging to the replicative machineries of Gram+ and Gram– bacteria. Deinococcus radiodurans is a Gram+ bacterium, featuring a large α subunit (1,335 amino acids, 150 kDa) and a putative ε subunit containing 197 amino acids. Interestingly, D. radiodurans ε subunit is devoid of the C-terminal domain present in its E. coli counterpart. Nevertheless, the identity and homology between E. coli and D. radiodurans ε subunits are equal to 29% and 59%, respectively (considering the region 1-178, D. radiodurans coordinates). To assay the competence of D. radiodurans polymerase III α subunit (αDr) in binding E. coli ε subunit, we have cloned into the pBAD vector a synthetic gene coding for αDr and we have co-transformed E. coli with pBAD-αDr and pGOOD-ε. The simultaneous addition of IPTG and arabinose to the culture medium triggers the co-expression of αDr and ε (Figure 3A). The association of these 2 proteins was tested by gel filtration. Cells co-expressing the DNA polymerase III αDr and ε subunits were collected by centrifugation (5,000 x g, 20 min), suspended in lysis buffer, disrupted by sonication, and the soluble proteins were immediately loaded onto a Superdex 200 column (Figure 3B). As Figure 3C shows, when the collected fractions were subjected to 3’-5’ exonuclease and to DNA polymerase activity assays, the activity peaks were detected at significantly shifted positions, suggesting that αDr is not competent in binding E. coli ε subunit. This was confirmed when aliquots of fractions 32, 36, and 52 were concentrated and analyzed by SDS-PAGE. Fraction 36 was found to contain αDr and to be devoid of E. coli ε subunit, while fraction 52 contained E. coli ε but was devoid of αDr (Figure 3D).
Replication fidelity primarily relies on the ability of DNA polymerases to discriminate against erroneous base pairings during DNA extension. Nevertheless, mismatched deoxynucleotides can be incorporated at 10-4 frequency20, and the action of 3’-5’ exonucleases is essential to proofread the newly synthesized DNA strand. Moreover, it was estimated that this DNA editing action increases replication fidelity by 3 orders of magnitude20. Accordingly, mutator strains can be constructed by impairing DNA proofreading, e.g., by conditionally expressing a mutagenic variant of E. coli DNA polymerase III ε subunit. This can be obtained constructing a variant of ε competent in binding α but devoid of its proofreading activity, and controlling the production of mutagenic ε by appropriate expression systems. Using this strategy, it was shown that E. coli mutation frequency increases under conditions inducing mutagenic variants of ε11,21. However, no fine tuning of the mutation frequency was accomplished by this means. Protein co-expression can be used to obtain a better control of mutator strains. To this aim, we co-transformed E. coli with pBAD-ε and pGOOD1-εD12A, in order to induce the expression of the wild-type and the mutagenic D12A variant of ε with arabinose and IPTG, respectively. As a phenotypic test, we determined the appearance of β-glucosidase activity, which is cryptic (Bgl–) in wild-type E. coli22,23. Spontaneous mutants able to utilize β-glucosides as carbon sources (Bgl+) can be enriched or isolated using liquid23 or solid media24, respectively. When cells were induced to express the D12A mutagenic variant, β-glucosidase activity was acquired in ca. 20 generations (Figure 4A). It should be noted that a similar trend was observed in non-induced cells (Figure 4A), most likely because the transcriptional control exerted on pGOOD1 is rather leaky11. Nevertheless, when the wild-type ε subunit was induced, alone or in conjunction with the D12A variant, β-glucosidase activity was acquired by E. coli at moderate levels (Figure 4A). The β-glucosidase activity of Bgl+ mutants was reported to range between 3 and 40 μM/min23. The level determined here in E. coli populations subjected to the expression of εD12A is much lower, but it should be considered that we did not attempt to isolate Bgl+ mutants on solid media.
Figure 1. Co-expression of E. coli DNA polymerase III catalytic core. (A) SDS-PAGE of total proteins extracted from E. coli TOP10 / pBAD-α1160 / pGOOD-ε243 grown in the absence of inducers, in the presence of arabinose only, of IPTG only, or in medium supplemented with both arabinose and IPTG (lanes 1-4, respectively). (B) Absorbance (empty circles) and exonuclease activity (filled circles) of fractions isolated by subjecting to gel-filtration soluble proteins extracted from E. coli TOP10 overexpressing α, ε, and θ subunits of E. coli DNA Pol-III. (C) SDS-PAGE of fractions 6, 12, 23, 31, and 41 (lanes 1-5, respectively) reported in Panel B. (D) SDS-PAGE of a concentrated aliquot of fraction 23, whose exonuclease activity and electrophoretic pattern are reported in Panels B and C, respectively. Reprinted with permission from Biotechnology Letters, 33, Conte et al.: “pGOODs: new plasmids for the co-expression of proteins in Escherichia coli”, 1815-1821 (2011). Please click here to view a larger version of this figure.
Figure 2. Proteolysis of E. coli DNA polymerase III ε subunit. (A) Degradation of monomeric as revealed by western blotting of proteins extracted from E. coli TOP10 overexpressing full-length or truncated forms of ε; arrowheads indicate the position of 25 and 19 kDa pre-stained molecular markers. Reprinted with permission from Elsevier from Biochimica et Biophysica Acta Proteins and Proteomics, 1794, Bressanin et al.:”Proteolysis of the proofreading subunit controls the assembly of Escherichia coli DNA polymerase III catalytic core”, 1606-1615 (2009). (B) Co-expression of α and ε subunits of E. coli DNA polymerase III catalytic core. SDS-PAGE of total (lanes 1 and 3) and soluble (lanes 2 and 4) protein extracts isolated from cells overexpressing ε subunit (lanes 1 and 2), or α and ε subunits (lanes 3 and 4).
Figure 3. Co-expression of DNA polymerase III α and ε subunits from Deinococcus radiodurans and Escherichia coli, respectively. (A) SDS-PAGE of total protein extracts isolated from cells not induced (lane 1), or induced to overexpress αDr, ε, or αDr and ε (lanes 2-4, respectively). (B) Chromatogram (empty circles, left axis) and protein concentration of fractions (filled circles, right axis) eluted form a gel filtration column (Superdex 200) loaded with soluble proteins extracted from E. coli overexpressing αDr and ε subunits. (C) Proofreading (empty circles, left axis) and DNA polymerase (filled circles, right axis) activities of the fractions reported in Panel B. (D) SDS-PAGE of concentrated aliquots of fractions 32, 36, and 52. Please click here to view a larger version of this figure.
Figure 4. Mutation frequency of E. coli cells subjected or not to co-expression of the wild-type and the mutagenic variant D12A of DNA polymerase III ε subunit. The map of the new plasmid pFINE is also shown. (A) β-glucosidase activity of E. coli TOP10 not induced (No Ind) or subjected to overexpression of wild-type ε subunit (Ara), the mutagenic variant D12A (IPTG), or both wild-type and D12A ε subunits (Ara + IPTG). The β-glucosidase activity is reported as a function of generations. (B) Map of the new expression vector pFINE. The multiple cloning site (MCS) of this plasmid is identical to those present in the compatible pBAD and pGOOD vectors, letting facile shuttling of genes among them.
Proteins can be intrinsically disordered, featuring regions whose tertiary structure is not restricted to a limited number of conformations25. These disordered proteins are usually prone to aggregation25, and their isolation and characterization might represent a difficult task. The ε subunit of E. coli DNA polymerase III features two distinct domains26,27, namely: i) the N-ter domain, bearing the 3’-5’ exonuclease activity, and competent in binding the θ subunit; ii) the C-ter domain, responsible for the binding to the α (polymerase) subunit. The C-ter domain of ε is known to confer poor solubility to this protein, promoting its aggregation. It was indeed demonstrated that ε-186, a truncated form of the protein devoid of the C-ter domain, is soluble, and can be purified with high yields17. However, when the interaction of ε with α has to be quantitatively studied (e.g., to determine the KD of the αε complex), the purification of full-length ε is required, implying a difficult biochemical task. In this frame, protein co-expression provides qualitative estimations of the binding of ε to α. We have shown that high yields of the αεθ complex can be obtained in vivo by co-expression (Figure 1), bypassing the poor solubility of ε subunit. Remarkably, the purification of overexpressed αεθ complex has been reported28,29. The co-expression strategy can also be used to test whether or not the insertion of site-specific mutations in one of the interacting partners impairs complex formation. Accordingly, protein co-expression represents a convenient and rapid tool to perform qualitative tests of protein-protein interaction. It should also be noted that our co-expression system relies on 2 different inducers, i.e., arabinose and IPTG. Therefore, appropriate controls can be easily performed when testing the formation of a protein complex, e.g., omitting one inducer from the bacterial growth medium (Figure 1A).
We presented here an optimized procedure for the co-expression of the α and ε subunits of E. coli DNA polymerase III. When this protocol was designed and tested, the following parameters were identified as critical to the success of co-expression experiments: i) the temperature at which induction is performed; ii) the time-length of induction; iii) the concentration of inducer(s). In particular, we were unable to obtain soluble αεθ complex11 from cells induced at 37 °C, independently of the induction time. We therefore suggest to test different temperatures for the induction step, and to check the overexpression of proteins as a function of time. In addition, the parallel evaluation of co-expression in different E. coli strains can be of help when facing moderate yields of the target proteins.
We have reported here a representative experiment performed to test the binding of proteins from different species, and we have shown that protein co-expression is useful to rapidly test the association of 2 heterologous proteins into a hybrid functional complex. Gel filtration chromatography was used here to evaluate complex formation (Figure 3). However, the insertion of an appropriate tag to one of the interacting partners could facilitate this evaluation, letting the use of affinity capture methods. To avoid interference of the tag with protein-protein interactions, the gene coding for one of the interacting partners could be inserted in parallel into pBAD/His and pBAD/Myc-His vectors, respectively conferring a hexahistidine motif at the N- and the C-terminus of the target protein.
E. coli mutator strains feature mutation frequencies higher than their wild type counterparts. Mutator strains are of interest in biotechnology30, e.g., to rapidly evolve a target strain towards novel, desired, phenotypes. We provided here evidence that co-expressing the wild-type and a mutagenic ε subunit of E. coli DNA polymerase III, the mutation frequency of the bacterial host can be tuned with sufficient convenience. To further improve this technology, tightly-regulated expression vectors are necessary to repress as much as possible the expression of the strongly mutagenic D12A variant of ε subunit. In particular, it would be interesting to test the mutation frequency in E. coli transformed with the pBAD-εD12A and pGOOD1-ε, i.e., the configuration alternative to that presented here. The pBAD vectors are indeed known to be tightly regulated12, and this feature could help in keeping the basal concentration of εD12A at low levels.
The examples of protein co-expression reported here testify the multifaceted nature of this technique. However, it is important to note that using a multiplicity of compatible plasmids in single E. coli cells can greatly expand the challenges which can be taken up by protein co-expression. In recent years, intense work was devoted to expand the repertoire of compatible plasmids to be used for protein co-expression. In particular, a ternary system relying on compatible plasmids bearing the ColE1, p15A, and pSC101 origins of replication was used to co-express either bacterial and mammalian proteins31. Recently, a quaternary co-expression system was reported. In this case, 13 heterologous genes were co-expressed in E. coli co-transformed with 4 compatible expression vectors, containing the ColE1, p15A, CloDF13, and RSF origins of replication32. This co-expression system was successfully used to produce in E. coli functionally active soluble hydrogenase I from Pyrococcus furiosus32. To enlarge our binary system, we constructed a new expression vector, pFINE, containing the pSC101 origin of replication, a neomycin-resistance cassette, and the lac regulatory elements (Figure 4B). This plasmid contains the same polylinker present in pBAD and pGOOD vectors, thus facilitating gene shuttling among the 3 components of the co-expression system. Although pFINE is a low copy number plasmid, its stability was found to be satisfactory when tested in rich medium. We are currently engaged in further expansion of our ternary co-expression system. To this aim, we constructed a pBAD derivative containing a chloramphenicol-resistance cassette, and we are on the way to exchange the ColE1 origin of replication with one compatible with ColE1, p15A, and pSC101. It is indeed our opinion that the use of multiple plasmids in single E. coli cells represents a powerful tool to challenge the expression in vivo of protein complexes composed of a multitude of different subunits.
The authors have nothing to disclose.
The permission by Springer and Elsevier to reprint figures is greatly acknowledged.
Name of Material/ Equipment | Company | Catalog Number | Comments/Description |
Agar | Sigma-Aldrich | A1296 | |
Ampicillin | Sigma-Aldrich | A9518 | |
Chloroform | Sigma-Aldrich | 288306 | |
EDTA | Sigma-Aldrich | EDS | |
Glycerol | Sigma-Aldrich | G5516 | |
INT | Sigma-Aldrich | I8377 | |
IPTG | Sigma-Aldrich | I5502 | |
KCl | Sigma-Aldrich | P9541 | |
L-Arabinose | Sigma-Aldrich | A3256 | |
MgCl2 | Sigma-Aldrich | M2670 | |
NaCl | Sigma-Aldrich | 31434 | |
PMSF | Sigma-Aldrich | P7626 | |
PNP-gluc | Sigma-Aldrich | N7006 | |
pNP-TMP | Sigma-Aldrich | T0251 | |
Tetracyclin | Sigma-Aldrich | 87128 | |
Trizma base | Sigma-Aldrich | T1503 | |
Tryptone | Sigma-Aldrich | 95039 | |
Yeast extract | Fluka | 70161 | |
Acrylamide solution 30% | BioRad | 161-0158 | For gel electrophoresis |
Ammonium persolphate | BioRad | 161-0700 | For gel electrophoresis |
Glycine | BioRad | 161-0718 | For gel electrophoresis |
SDS | BioRad | 161-0302 | For gel electrophoresis |
TEMED | BioRad | 161-0800 | For gel electrophoresis |
Tris | BioRad | 161-0719 | For gel electrophoresis |
Cuvettes 0.1 cm | BioRad | 1652089 | For electroporation |
EQUIPMENT | |||
Centrifuge 5415R | Eppendorf | ||
Centrifuge Allegra 21R | Beckman | ||
Chromatography apparatus GradiFrac | Pharmacia Biotech | ||
Gene Pulser II electroporation | BioRad | ||
Microplate Reader 550 | Biorad | ||
MiniProtean 3 cell | BioRad | ||
Power Supply | BioRad | ||
Sonicator 3000 | Misonix |