Boosting toxic protein expressionbiosynthesis: transient in vivo inactivation of engineered bacterial alkaline phosphatase

Background The biotechnology production of enzymes is often troubled by the toxicity of the recombinant products of cloned and expressed genes, which interferes with the recombinant hosts’ metabolism. Various approaches have been taken to overcome these limitations, exemplied by tight control of recombinant genes or secretion of recombinant proteins. An industrial approach to protein production demands maximum possible yields of biosynthesized proteins, balanced with the recombinant host’s viability. Bacterial alkaline phosphatase (BAP) from Escherichia coli (E. coli) is a key enzyme used in protein/antibody detection and molecular cloning. As it removes terminal phosphate from DNA, RNA and deoxyribonucleoside triphosphates, it is used to lower self-ligated vectors’ background. The precursor enzyme contains a signal peptide at the N-terminus and is secreted to the E. coli periplasm. Then, the leader is clipped off and dimers are formed upon oxidation. Results We present a novel approach to phoA gene cloning, engineering, expression, purication and reactivation of the transiently inactivated enzyme. The recombinant bap gene was modied by replacing a secretion leader coding section with a N-terminal his6-tag, cloned and expressed in E. coli in a P BAD promoter expression vector. The gene expression was robust, resulting in accumulation of His6-BAP in the cytoplasm, exceeding 50% of total cellular proteins. The His6-BAP protein was harmless to the cells, as its natural toxicity was inhibited by the reducing environment within the E. coli cytoplasm, preventing formation of the active enzyme. A simple protocol based on precipitation and immobilized metal anity chromatography (IMAC) purication yielded homogeneous protein, which was reactivated by dialysis into a redox buffer containing reduced and oxidized sulfhydryl group compounds, as well as the protein structure stabilizing cofactors Zn 2+ , Mg 2+ and phosphate. The reconstituted His6-BAP exhibited high activity and was used to develop an ecient protocol for all types of DNA termini, including problematic ones (blunt, 3’-protruding).

impermeable to charged molecules. Such a dedicated system was described as the phosphate-speci c transport system (Pst system) [3,4]. APS are almost exclusively homodimeric metalloproteins. Their common architecture includes each catalytic site containing three metal ions: two Zn 2+ and one Mg 2+ [5,6]. Furthermore, they require the adoption of a catalytically active conformation facilitated by disul de bridges, among others. In the fully active, dimeric BAP, Zn 2+ occupies active sites A and B, and Mg 2+ occupies site C, thus the enzyme has the con guration (ZnAZnBMgC)2. Four cysteine residues create disul de bridges linking the two subunits and are essential for mature BAP dimer activity [7]. Apparently, a combination of factors, including strong interactions between amino acid (aa) side chains, stabilisation of the 3D active conformation by divalent Zn 2+ and Mg 2+ , as well as the presence of disul de bridges result in an unexpected thermal stability of BAP, vastly exceeding the temperature growth range of E. coli.
The enzyme is active up to 80 o C and even undergoes reversible renaturation at 90 o C, thus a heating step could be used in enzyme puri cation [5,8]. This thermostability may play a role in the resistance of BAP to inactivation by harsh environmental conditions, present outside the cytosol. The enzymes exhibit wide substrate speci city, and catalyses: the hydrolysis of monoesters of phosphoric acid, including 5' ends of DNA, RNA, nucleotides and a transphosphorylation reaction in the presence of high concentrations of phosphate acceptors. Moreover, it hydrolyses oxyphosphate monoesters [9,10], as well as a variety of Oand S-phosphorothioates [11,12,13], phosphoramidates [10], thiophosphates and phosphates [14,15]. A minor activity of E. coli BAP -oxidation of phosphite to phosphate was also detected. Puri ed BAP catalysed the oxidation of phosphite with speci c activities approx. 1000-fold lower than phosphate ester hydrolysis. Interestingly, BAP catalyses the oxidation phosphite to phosphate and H 2 , thus it can be considered a phosphite-dependent hydrogenase that has emerged as a result of evolution [16]. Thus far, BAP was puri ed as a native or recombinant enzyme from its natural location in the periplasmic space, by weakening the outer membrane of cells, using for example osmotic shock [17] or a mutant E. coli strain [18]. BAP is widely used in molecular cloning for the removal of 5' phosphates from linearized vectors, detecting PCR products, primer labelling and immunoassays. In this study, we describe a successful alternative strategy for the cloning and high production of BAP with transiently inhibited activity and thus 'toxicity' to the recombinant E. coli host. The strategy includes the biosynthesis of the leaderless His6-tagged BAP in the E. coli cytoplasm, followed by puri cation and oxidation/renaturation of the enzyme in vitro. We also believe that the developed method will be useful for the biotechnology scale production of other periplasm residing proteins/enzymes.
Enzymatic assay of alkaline phosphatase.
A comparative enzymatic 'Glycine with Zinc Assay' of puri ed His 6 -BAP and commercially available alkaline phosphatase from E. coli (Sigma-Aldrich) was conducted as based on the rate of release of pnitrophenol from p-nitrophenyl phosphate by following the absorbance at 405 nm. One unit will hydrolyze Gel electrophoresis and protein concentration determination Agarose gels (1.0 %) for DNA analysis were prepared in TAE buffer [19]. The gels were visualized after staining with ethidium bromide using a 312 nm UV transilluminator or after staining with SYBR Green I using a 312 nm UV transilluminator and photographed with a SYBR Green gel stain photographic lter.
SDS-PAGE electrophoresis of the proteins was in 10% polyacrylamide gels [19]. The protein concentrations were measured with a Pierce™ BCA Protein Assay Kit according to the manufacturer's instructions.

Results
Rationale, design and cloning of a his6-phoA gene from E. coli The rationale behind the general concept to obtain high overproduction of a toxic, recombinant protein was to biosynthesize it in a transiently inactive state: in the cytoplasm, which provides a reducing environment. This was to be followed by refolding in vitro under mild conditions in an oxidizing environment in the presence of a reaction product or substrate, redox buffer and cofactors, if applicable.
In the example shown in this work, the modi ed His6-BAP was devoid of a secretion leader. Thus the produced truncated pre-protein was prevented from entering the periplasmatic space, which is required for disul de bridge formation. Upon isolation from the cytoplasm of recombinant E. coli, the truncated, modi ed His6-BAP was to be restored to the enzymatically active state by dialysis into a buffer with pH and ionic strength simulating in vivo conditions. For this purpose, a mixture of oxidised and reduced sulfhydryl group-containing organic compounds, the protein structure stabilizing divalent ions cofactors Zn 2+ and Mg 2+ and dephosphorylation reaction product -phosphate was used. The native phoA 3' gene fragment of 1353 bp, coding for a leaderless C-terminal ORF and a Stop codon was PCR ampli ed with a mutagenic forward primer (62-nt) ( Table 1), containing an overhang (42 nt) and mismatches, which introduced the BsaI restriction site. This arrangement removed the phoA leader-coding a 63 bp (21 aa) 5' region, replacing it with an in-frame 18 bp segment coding for: 6 histidine residues for IMAC and a 4 aa segment MPMS, protecting the His6-tag from aminopeptidases in vivo and adding more immunogenic properties to the His6-tag for the possibility of using custom made antibodies (Fig. 1). The reverse mutagenic primer (27-nt) introduced a HindIII site. Following ampli cation, the reaction product was cut with BsaI and HindIII, puri ed and cloned into NcoI-HindIII cleaved arabinose-regulated pBADmycHisA expression vector in a perfect fusion with the vector's ATG codon. Since BsaI is a Type IIS REase, its 4-nt generated cohesive ends can be of any sequence. Thus they made CATG-5' to be compatible with the NcoI-generated cohesive ends. The resulting constructs were sequenced from both ORF ends and the pBAD_BAP1 clone was selected for expression experiments (Additional les 1-4  Fig. 2A). Fig. 2A shows the time course of induction, where whole E. coli TOP10 [pBAD_BAP1] cells were lysed in denaturing buffer and loaded onto SDS-PAGE. Since the 4 th h after induction, His6-BAP comprises the majority of cellular proteins. Thus, taken together the high biosynthesis level and His6-tag, a simple 4-step puri cation protocol was devised, comprising of: (i) sonication of entire cells, instead of extracting BAP from the periplasm; (ii) PEI precipitation of nucleic acids and acidic proteins; (iii) ammonium sulphate fractionation and (iv) IMAC. Puri cations steps (ii) and (iii) removed the bulk of the cellular contaminants from the crude His6-BAP-containing cell extract. This allowed us to take full advantage of high speci city of the immobilised Ni 2+ interaction with the His6-tagged recombinant His6-BAP. The interaction was strong, as 40 mM imidazole-containing elution buffer eluted only trace amounts of the His6-BAP, while the bulk of the His6-BAP eluted as a sharp peak with 500 mM imidazole (Fig. 2B). Overall, the puri cation stages (ii)-(iv), each based on a different principle, were su cient to obtain a homogeneous protein (Fig.  2B). The step (ii) of PEI precipitation of nucleic acids was also selected as bene cial from the stand point of application in genomic library preparations, as no E. coli DNA should be carried over with His6-BAP to be used for dephosphorylation. To increase yields and for protective purposes, glycerol and non-ionic detergents were added to block hydrophobic patches on the His6-BAP protein surface and prevent the protein from denaturation, aggregation and adhesion. Optionally, considering the reported thermostability of BAP [5,8], a heat treatment step can be added. While heat treatment would not increase the apparent purity of the preparation, when examined on SDS-PAGE gel, for some special purposes it can be included into the presented puri cation protocol after ammonium sulphate fractionation and prior to IMAC (not shown). As judged by SDS-PAGE, fractions containing the highest His6-BAP content (Fig.2B, lanes 6-10) were pooled and dialysed against the oxidizing buffer to promote the folding of the enzyme into an active state. Next, the enzyme was dialysed against storage buffer, containing 50% glycerol and all the components of reactivation-oxidation buffer to maintain a stable redox environment. The buffered glycerol preparation was stored for over 4 years at -20 o C without an apparent loss of activity (not shown). The nal preparation was assayed in a standard colorimetric assay using p-nitrophenyl phosphate [5,21,22 ] in parallel with several available commercial preparations and their relative activities were compared. The His6-BAP exhibited a speci c activity of approx. 80% of the commercially available preparations of the highest purity (not shown). Fig. 2C shows the comparison of two BAP preparations: Sigma BAP (Fig.  2C, lane 1) and His6-BAP (Fig. 2C, lane 2), run in parallel as overloaded bands. The His6-BAP enzyme is stable, with minimal signs of degradation, when stored for over 4 years in a dedicated storage buffer with 50% glycerol (Fig. 2C, lane 2). Due to the presence of the His6-tag, the His6-BAP has a higher molecular weight than Sigma BAP. The purity of both enzymes is very similar, with two more minor bands observed in the Sigma BAP preparation.
Evaluation of His6-BAP performance in DNA dephosphorylation for molecular cloning The major purpose of this work was to provide large amounts of ultra-pure alkaline phosphatase for molecular cloning approaches, where thermolabile phosphatases do not perform well, namely, in preparation of blunt-and protruding 3' termini of vector's DNA. For this purpose, we tested the His6-BAP in a vector dephosphorylation assay. Fig. 3 shows the general cloning vector pUC19 cleavage/dephosphorylation/self-ligation assay. Clearly, His6-BAP e ciently dephosphorylates the vector's DNA. In the nal range tested (from 5x10 -3 to 5x10 -2 colorimetric units) both Sigma BAP and His6-BAP e ciently dephosphorylated KpnI-generated 3' sticky ends and SmaI-generated blunt ends. For cloning, it is more convenient to use a DNA dephosphorylation activity unit, which is different than the de nition of the activity unit used in the classical p-nitrophenyl phosphate assay. One should note that the phosphatase unit de nition may vary, depending on manufacturer, but in general it is based on dephosphorylation of a set amount of DNA (such as 1 mg) or pmols of 5'ends (such as 1 pmol) in a set reaction time, temperature and buffer composition. As 1 pmol of DNA ends is a little over about 1 µg of a 3 kb plasmid, these de nitions are not far away from each other. Typically they use 10-30 min reaction time and 37°C. For the purpose of this work we have adopted and modi ed the unit de nition used by New England Biolabs  (Fig. 3A, lane 8), one can observe a mixture of ligated and non-ligated plasmid forms, indicating than only partial dephosphorylation was obtained. For the 3'-protruding, KpnI-linearized pUC19 DNA, very similar results were obtained, with slightly lower dephosphorylation e ciency, as faint ligated bands can be observed (Fig. 3A, lane 1). Thus, for a maximum reduction of the cloning background, we recommend to use a 2-fold or higher excess of His6-BAP, namely a minimum of 10 'cloning units' per 1 mg of the linearized plasmid of approx. 3 kb. The same results were obtained both for Sigma BAP and His6-BAP.

Discussion
Here, we propose a novel approach to the high level biosynthesis of toxic proteins by expressing them in an inactive state. This is followed by the use of an in vitro gentle reactivation method, not employing harsh protein unfolding compounds, such as guanidinium chloride, but rather based on a tertiary structure recovery stimulation by enzyme cofactors, reaction substrates, products and disul de bridge formation in a buffer resembling in vivo conditions. Further, this apparently allowed correct arrangement of disul de bridges formation. All of these factors help the distant, catalytically active, polypeptide segments to come together in a native arrangement. This is an alternative and/or complementary approach to our previous publications concerning the expression of high GC content toxic genes. In that work we designed, synthesized and cloned the entire recombinant genes, encoding thermophilic restriction endonucleases-methyltransferases (REases-MTases), 'toxic' to a recombinant E. coli host. For this purpose we used a modi ed 'one amino acid-one codon' optimization method combined with weighting toward low GC content codons. This approach allowed for a signi cant expression increase of a thermophile gene in the recombinant host [25]. Together with the post-optimization sequence scanning for mRNA secondary structures, codon clusters and the local codon environment, the nal synthetic gene became 'E. coli friendly', allowing for a one-order of magnitude increase in taqIIRM gene expression. In another approach, we used a biased 'codon randomization' method, which besides ORF optimization, apparently sped up the translation kinetics, which proved to be critically important to allow recombinant TthHB101RM REase-MTase to fold into an active state [26]. All the above approaches, including His6-BAP, can be combined to e ciently biosynthesize secreted 'toxic' proteins from a thermophile or other bacteria/archaea with GC-rich genomes. As the expression of the wt phoA gene is induced in E. coli (when grown under phosphate-limiting conditions) and results in moderate production of BAP, it can be considered as already naturally 'optimized' with regard to codon usage context, mRNA sequence and structure among other factors. Thus there was no need for applying gene optimisation procedures to obtain a high expression level, when natural regulation of phoA gene was replaced by the vector's precisely regulated P BAD promoter [20], as we obtained in our hands as massive natural codon-containing gene expression as for best optimised. This system is approaching the levels of expression as known from highest robustness T7-lac promoter-containing expression vectors [27]. The native BAP gene, containing a secretion leader, was cloned previously into bacteriophage lambda and further subcloned into the pBR322 vector derivative [28,29] and into a derivative of pBluescript-SK(-) vector [30]. Those constructs were not for overproduction and directed BAP into the periplasm. However, an interesting observations were made by others that concern E. coli BAP production [31] on overproduction of the cloned native BAP precursor, where formation of cytoplasmic inclusion bodies was observed. Accumulation of BAP precursor at a high rate indicates that its synthesis can be uncoupled from secretion [31]. It may also suggest, that the presence of leader secretion peptide in the BAP precursor impairs its folding, thus leading to the inclusion bodies formation and providing additional natural route of transient toxicity suppression. In contrast, we cloned the mature, leaderless BAP with added His6-tag. The recombinant His6-BAP was biosynthesized in the E. coli cytoplasm in an inactive form, thus nontoxic. This allowed for massive biosynthesis, comprising over 50% of cellular proteins of the recombinant host. We believe that this strategy will be useful for biotechnology industry production. The isolation/puri cation is fairly straightforward, with only one chromatographic step on Ni 2+ -chelating metal a nity resin, proceeded by two rapid precipitation steps involving PEI and ammonium sulphate, yielding a homogeneous protein. Further, we developed a simpli ed and mild BAP reactivation protocol, based on oxidation of cysteines in the presence of correct fold-stimulating agents (in this case Mg 2+ , Zn 2+ , phosphate, detergents, physiological pH and salt concentration). While native (= secreted and puri ed from periplasm) BAP was subjected to denaturation and successful refolding into an active state previously, these approaches were based on harsh conditions, such as the use of concentrated solutions of guanidinium hydrochloride [32,33,34]. Further, in those previous works, concerning denaturation/renaturation of phosphatases, there were used natively matured enzymes, thus already containing disul de bridges in correct arrangements [32,33,34]. The overproduction, puri cation and reactivation procedure of His6-BAP is easily scalable. We repeated the described procedure in various variants, using from 0.5 g up to 50 g of the induced bacterial cells. Comparative assays of the His6-BAP (conducted in parallel with the highest activity available commercially BAP preparation) showed that His6-BAP has approx. 10% higher speci c activity. Previously it was found that BAP is a thermostable enzyme in a mesophilic, cytoplasmatic environment [18]. This does not necessarily mean that there is an evolutionary relationship with 'truly' thermophilic enzymes, rather points to convergent protein structure evolution of mesophilic enzyme to withstand the harsher than cytoplasmatic conditions of an external environment. This feature can be exploited in adding an extra puri cation step to the His6-BAP protocol presented here, if desired for special application purposes. The thermal stability of BAP imposes linearized vector DNA puri cation upon dephosphorylation. Using other thermolabile phosphatases, such as calf intestinal phosphatase [35] or oceanic shrimp phosphatase [36], is convenient in DNA manipulations, as they can be heat-inactivated. However, high BAP thermostability may be of practical use in DNA manipulation methodologies in some applications, such as dephosphorylation of GC-rich blunt ends of DNA molecules, as DNA end 'breathing' at elevated temperatures makes 5'-phosphates more accessible to the enzyme.

Conclusions
1. E. coli phoA gene was engineered by removal of the secretion leader and its replacement by a His6tag and cloned under the P BAD promoter control, which resulted in the cytoplasmatic biosynthesis of the recombinant His6-BAP.

2.
A very high level of inactive His6-BAP overproduction was achieved as a result of combination of the strong P BAD promoter and transient suppression of His6-BAP toxicity by preventing essential disul de bridge formation in the reducing cytoplasmatic environment.
3. A simple protocol of puri cation to homogeneity of His6-BAP, suitable for biotechnology production, was developed.
4. An in vitro His6-BAP activity reactivation protocol was developed, based on folding and gentle oxidation in redox buffer, supplemented with cofactors (Zn 2+ , Mg 2+ ), reaction product (phosphate) and non-ionic detergents.
5. An e cient dephosphorylation protocol was developed for all types of DNA termini: blunt, 3' protruding and 5' protruding.

Competing interests
The authors declare that they have no competing interest.