Production of recombinant 1-deoxy-d-xylulose-5-phosphate synthase from Plasmodium vivax in Escherichia coli

Humanity is burdened by malaria as millions are infected with this disease. Although advancements have been made in the treatment of malaria, optimism regarding our fight against malaria must be tempered against the problem of drug resistance in the Plasmodium parasites causing malaria. New targets are required to overcome the resistance problem. The enzymes of the mevalonate-independent pathway of isoprenoid biosynthesis are targets for the development of novel antimalarial drugs. One enzyme in this pathway, 1-deoxy-d-xylulose-5-phosphate synthase (DXS), catalyzes the conversion of 1-deoxy-d-xylulose-5-phosphate to isopentenylpyrophosphate and dimethylallyl phosphate. We demonstrate the use of a step deletion method to identify and eliminate the putative nuclear-encoded and transit peptides from full length DXS to yield a truncated, active, and soluble form of Plasmodium vivax DXS, the DXS catalytic core (DXScc).


Introduction
Malaria is a devastating health problem: approximately 50% of the world's population is at risk of malaria, 216 million cases of malaria were reported in 2010, and over 650,000 people died of malaria in 2010 [1,2]. Unfortunately, the strides being made to combat malaria must be tempered with the growing problems of drug resistance in the parasites that cause malaria and insecticide resistance in the mosquitoes that spread the parasites [2][3][4][5]. Malaria is caused by an infection by one of five species of Plasmodium protozoans, specifically, P. falciparum, P. vivax, P. ovale, P. malariae, and P. knowlesi [6,7]. Infection by P. falciparum is dominant in Africa, while infection by P. vivax is found throughout Asia and the Americas [6,8,9]. Because of the high mortality rate of P. falciparum malaria [10], efforts to treat malaria have been targeted largely against P. falciparum malaria [11]. However, P. vivax malaria is debilitating, can be fatal, and is more widespread and difficult to treat than P. falciparum malaria. Arguably, P. vivax malaria may be a greater burden to humanity than P. falciparum malaria [8,11].
Enzymes involved in type II fatty acid biosynthesis [12], isoprenoid biosynthesis [13], hemozoin biosynthesis [14], and iron sulfur cluster assembly [15] have been identified as potential targets for the development of novel antimalarial drugs. Of these pathways, the enzymes required for biosynthesis of the isoprenoid precursors, isopentenylpyrophosphate (IPP) and dimethylallyl pyrophosphate (DMAPP), are particularly attractive because the plasmodial route to IPP and DMAPP is completely orthogonal to the mammalian biosynthetic route.
We report here on the use of a step wise deletion approach for the successful production of a catalytically active and soluble form of P. vivax DXS in E. coli. This catalytically active form of P. vivax DXS does not possess either the signaling or transit peptide and is designated the DXS catalytic core, DXS cc . P. vivax DXS cc has been characterized and compared to DXS enzymes from other organisms. We find that the steady-state kinetic parameters and other biochemical features of P. vivax DXS cc are consistent with data published for other DXS proteins [25,26].

Cloning of the P. vivax dxs gene from genomic DNA
The P. vivax dxs gene was extracted from strain Sal-1 using the following PCR primers: forward primer, 5 -CCTAGGATCCGATGATAATGGGAACTTCCTCT-3 and reverse primer, 5 -ATCCAAGCTTGTTGGACACCCCTCCTTGCAG-3 . The BamHI and the HindIII restriction sites are underlined. The PCR product was cloned into the BamHI and HindIII restriction sites of pET21b(+) to yield the pET21b(+)-vivax-DXS plasmid with a C-terminal His 6 tag. Cloning of the P. vivax dxs gene was confirmed by DNA sequencing.

Cloning of the P. vivax dxs gene and the constructs without the signal peptide and the transit peptide from a codon optimized gene
A synthetic, codon optimized P. vivax dxs gene with 5 -NdeI and 3 -HindIII restriction sites in a pUC57 vector was obtained from Genscript. The full length dxs gene was excised from pUC57 vector and cloned into NdeI and HindIII restriction sites of pET21b(+) vector with a C-terminal His 6 tag to yield the vector SH-1. Genes that would yield DXS proteins truncated at the N-terminus were SH-2, SH-3, SH-4, SH-5, SH-6, SH-7, SH-8, and SH-9 (N-terminal deletion of 210, 245, 252, 272, 303, 347, 357, and 395 amino acids, respectively). These were constructed by PCR extraction from the codon optimized P. vivax dxs gene and cloned into NdeI and XhoI restriction sites of pET28a(+) vector with N-terminal His 6 tag.

Overexpression of full length and truncated P. vivax DXS proteins in E. coli
Both full length and truncated P. vivax DXS proteins were expressed in E. coli BL-21 B(DE3) or Rosetta B(DE3) cells. A fresh colony of E. coli containing the appropriate plasmid was cultured at 37 • C in LB broth containing 100 μg/mL ampicillin for SH-1 and 50 μg/mL kanamycin for SH-2 to SH-9 supplemented with 0.8% glucose and 25 mM potassium phosphate at pH 7.2. The cultures were grown for 2-3 h until the OD 600 reached 0.3, then diluted 100-fold, incubated at 37 • C until the OD 600 reached 0.6, and then cooled to 25 • C. Expression was induced by the addition of 0.5 mM isopropyl β-d-1thiogalactopyranoside (IPTG), incubated for 5-6 h with shaking, the cells harvested by centrifugation, and the cell pellets stored at −80 • C.

DXS and DXR purification by Ni-NTA affinity chromatography
Cells were thawed and all the purification steps were performed at 4 • C. Cells were resuspended in the binding buffer (20 mM Tris pH 7.5, 500 mM NaCl, 20 mM imidazole, and 10 mM BME) supplemented with 1 mM PMSF, 4 μg/mL leupeptin, 2 μg/mL pepstatin, sonicated, and centrifuged at 16,000 × g for 20 min. The supernatant from the cell lysate was applied to the Ni-NTA resin (1.5 cm × 5 cm, equilibrated with the binding buffer) at a rate of 1 mL/min and the non-bound proteins were eluted from the column by washing with 5 column volumes of the binding buffer followed by 20 column volumes of the wash buffer (20 mM Tris pH 7.5, 500 mM NaCl, 60 mM imidazole, and 10 mM BME. DXS SH-1 and SH-2 were eluted using elution buffer (20 mM Tris pH 7.5, 500 mM NaCl, 250 mM imidazole, and 10 mm BME), fractions containing active protein were combined and the protein concentrated for size exclusion chromatography. The truncated DXS protein, DXS SH-2, will be referred to as the DXS catalytic core, DXS cc . Additional details of our experimental methods and the materials can be found in the Supplementary material .

Expression of full length P. vivax DXS using genomic DNA
The expression of proteins from Plasmodium in E. coli has always been difficult, even more so for proteins targeted to the apicoplast. This problem arises from the differences in codon bias between E. coli and Plasmodium species [27,28] and from the N-terminal bipartite targeting sequence of apicoplast proteins (Fig. 1). The enzymes of the MEP are located in lumen of the apicoplast [29]; thus, such enzymes from Plasmodium are difficult to express in E. coli. Not surprisingly, only two MEP enzymes from Plasmodium have been expressed in E. coli: DXR [30] and 2-C-methyl-d-erythritol 2,4-cyclodiphosphate synthase [31]. Both were from P. falciparum and were expressed with the N-terminal amino acids deleted from a putative catalytic core. None of MEP enzymes from P. vivax have been expressed in E. coli, to date.
The open reading frame corresponding for the full length dxs gene was isolated from P. vivax genomic DNA by PCR (Fig. S1). Expression of full length P. vivax DXS in E. coli using the dxs gene isolated from genomic DNA was first performed as a proof of concept. A final yield of 40 μg of partially purified P. vivax DXS per liter of culture from BL-21 cells and 80 μg per liter of culture using Rosetta B(DE3) cell line was obtained. Full length P. vivax DXS was active and exhibited a specific activity of 1.2 ± 0.1 μmol/min/mg.

Expression of full length P. vivax DXS using a synthetic codon optimized dxs gene
The expression of low levels of full length, active P. vivax DXS in E. coli using the genomic dxs gene was encouraging. A synthetic P. vivax dxs gene codon optimized for E. coli expression was prepared to improve upon the low expression yield. Expression yields of full P. vivax DXS cc from E. coli improved to ∼1.5 mg/L of culture when using the codon optimized gene, but a low molecular weight protein contaminant was always present with the full length DXS (Fig. 2). We varied an array of experimental parameters, including induction temperature, induction time, IPTG concentration, and protease inhibitor combinations, but were unable to eliminate the low molecular weight contaminant from full length P. vivax DXS.

Expression and purification of P. vivax DXS cc
Expression of full length P. vivax DXS in E. coli using the synthetic dxs gene was complicated by the presence low molecular weight protein (Fig. 2). This low molecular weight contaminant is, most likely, a truncated form of DXS resulting from the proteolytic removal of the N-terminal bipartite pre-sequence by an E. coli protease. Our solution to this problem was to directly express the P. vivax DXS cc in E. coli by eliminating the bipartite targeting peptide from the optimized full length dxs gene. Bioinformatic tools to predict the signal peptide generally work well and predicted a signal peptide compromising the first 23-30 amino acids in full length P. vivax DXS ( Table 1).
The accurate prediction of the transit peptide, which serves in the movement of a protein into the lumen of the apicoplast, is not possible because transit peptides vary considerably in length and sequence. In addition, the cleavage site to remove the transit peptide from the mature protein apicoplast is not well understood. Thus, there is no consensus sequence for the transit peptide.
In order to express the soluble and active P. vivax DXS cc without the signaling and transit peptides, sequences of E. coli, Deinococcus radiodurans, and P. vivax DXS were compared (Fig. S2). Based on sequence similarities between the DXS enzymes from E. coli and D. radiodurans, we designed clones SH-7 to SH-9 to eliminate large sections of the N-terminus from full length enzyme. A similar strategy was used for the expression of active forms of P. falciparum DXR and IspF in E. coli [30,31]. Clones SH-7, SH-8, and SH-9 yielded only microgram levels of soluble, but inactive protein per liter of culture medium.
DXS sequences from five different species of Plasmodium species were aligned to better capture the nature of the transit peptide. This comparison revealed little homology in the first 200 N-terminal amino acids; highlighting the difficulty in predicting the transit peptide of P. vivax DXS. Given the paucity of guidance from the sequence comparisons on how to best design a gene coding only for DXS cc , we generated a series of clones, each with a progressively longer Nterminal deletion in the final P. vivax DXS-derived protein (Fig. S3). This step deletion approach ultimately led to clone SH-2, with an N-terminal deletion of the first 210 amino acids. Expression of the protein derived from SH-2, DXS cc , lead to a soluble, homogenous, and active enzyme in good yield when expressed in E. coli. All the other truncated DXS proteins discussed herein produced microgram quantities of soluble, but inactive protein upon expression in E. coli.
An investigation of the transit peptide for the P. falciparum acyl carrier protein showed that the transit peptide is disordered or adopts a low occupancy helix [32]. Thus, one validation of the predicted transit peptide sequence for P. vivax DXS comes from the calculation of sequence entropy for residues 1-250 using DisEMBL [33]. Analysis of the sequence entropy indicated that residues from 1 to 250 are highly disordered (Fig. S4), providing further evidence that the transit peptide for P. vivax DXS is defined within the N-terminal 250 amino acids of the full length protein. Since the signal peptide is, most likely, encompassed by the first ∼25 amino acids (Table 1), we propose that the transit peptide of full length P. vivax DXS extends from about residue 30 to about residue 250 (Fig. 3).
Nickel affinity chromatography was used as a first step in the purification, followed by size exclusion chromatography (Fig. 4(A)). Two major protein peaks were observed in the size exclusion chromatogram. The first peak, representing a protein with a molecular weight corresponding to the DXS cc tetramer or high molecular weight aggregates, exhibits little to no DXS activity. The second peak, representing a protein with a molecular weight of the DXS cc dimer, has relatively high DXS activity (Fig. 4(B) and (C)). The final P. vivax DXS cc preparation, obtained using nickel affinity and size-exclusion chromatography, was >99% pure, as judged by SDS-PAGE and Western blot analysis (using anti-His 6 antibody). The final yield of dimeric P. vivax DXS cc was approximately 30%, yielding 1 mg/L of enzyme per liter of E. coli culture ( Table 2).

Characterization of P. vivax DXS cc : pH optimum, molecular weight, and Mg(II) stoichiometry
At saturating concentrations of pyruvate and GAP, a pH optimum of 7-7.5 was determined for P. vivax DXS cc (Fig. S5). The pH optimum for P. vivax DXS cc is comparable to the values reported for DXS from Rhodobacter capsulatus (7.0) [34] and Agrobacterium tumefaciens (8.0) [26].
The molecular mass of active P. vivax DXS cc estimated by gel filtration chromatography under non-denaturing conditions was 260 kDa (Fig. S6) while the molecular weight of the enzyme estimated by SDS-PAGE analysis was 115 kDa (Fig. 4(C)). These data suggest the P. vivax DXS cc exists as a homodimer. DXS from D. radiodurans and R. capsulatus have been reported to exist as dimers, as well [24,34]. MALDI-MS was further used to confirm the molecular weight of the  monomeric unit. The results of MALDI-MS analysis showed peaks for M, M/2, and 2M peak with the highest intensity observed for the M peak (Fig. S7), and the molecular weight for the M peak corresponding to 103,822 Da with an accuracy of 0.1% (99 Da from the theoretical 103,921 Da). DXS catalysis requires thiamin pyrophosphate (TPP) to transfer an acetyl unit from pyruvate to GAP in forming DXP. TPP is observed in the crystal structures of both E. coli DXS and D. radiodurans DXS [24]. Like other TPP-dependent enzymes, DXS also requires a divalent metal ion for catalysis. DXS from A. tumefaciens [26], R. capsulatus [34], Streptomyces [35], and Mycobacterium tuberculosis [36], require Mg(II) or Mn(II) for maximum activity.
We found that P. vivax DXS cc was completely active even in the absence of any exogenously added TPP or divalent metal ion to the assay solutions or to any of the buffers employed during the purification of the enzyme (data not shown). These data suggest that P. vivax DXS cc binds both TPP and metal with relatively high affinity and both were acquired from the culture medium. In order to test the activity of DXS cc in the absence of TPP and metal, the protein was dialyzed against EDTA to remove the bound cofactors. DXS cc was completely active after exhaustive dialysis against EDTA. Incubation of the enzyme with acidic, saturated ammonium sulfate (pH 3.5) lead to the complete removal of TPP, with high recovery of activity upon reconstitution with TPP (Fig. S8). However, the required metal ion remained bound to the enzyme as added Mg(II) had no effect on reconstitution of DXS cc activity. Preparation of metal-free DXS cc proved difficult as exhaustive dialysis against EDTA in pH 7.0 buffer resulted in completely active enzyme. Incubation of the enzyme with saturated ammonium sulfate and 5.0 mM EDTA at pH 3.5 led to the complete loss of DXS cc activity, with only a 25% recovery of activity upon the addition of Mg(II). Inductively coupled plasma resonance mass spectrometry (ICP-MS) was then used to quantify the amount of Mg(II) in DXS cc . Molar ratio of Mg(II) to DXS was found out to be 0.97 ± 0.01. These results suggest that Mg(II) may serve both a structural and catalytic role in DXS cc , that Mg(II) is bound to DXS cc with a higher affinity than TPP, and that the bound Mg(II) with another divalent metal might be achieved only by culturing the P. vivax DXS cc -expressing E. coli cells in minimal media supplemented with a divalent metal ion of interest.

Kinetic characterization of P. vivax DXS cc
Steady state kinetic parameters for pyruvate and GAP for P. vivax DXS cc were calculated using the DXS-DXR coupled assay (Fig. 5). The steady-state kinetic parameters for DXS cc were comparable to DXS from other bacterial species [39,40,50]. The relatively small differences in the steady-state kinetic parameter for P. vivax DXS cc when compared to similar values for DXS R. capsulatus and A. tumefaciens, may result from the different pH optima of the enzyme from these organisms (Table 3). β-Fluoropyruvate (F-Pyr), a pyruvate analog and a known competitive inhibitor of DXS [25] was used to further characterize P. vivax DXS cc . F-Pyr was competitive with respect to pyruvate with a K i value of 200 ± 100 μM (Fig. 6), a value lower than the K i of 1.0 ± 0.1 mM reported for DXS from R. capsulatus [25].
In conclusion, we report here on the successful expression of a soluble and catalytically active form of P. vivax 1-deoxy-d-xylulose-5-phosphate synthase in E. coli. P. vivax DXS cc exists as a homodimer, contains one-bound Mg(II) per enzyme molecule, and exhibits a For P. vivax DXScc, the apparent kinetic constants for pyruvate were measured by varying the initial concentration of pyruvate at 0.1 mM GAP. b For P. vivax DXScc, the apparent kinetic constants for GAP were measured by varying the initial concentration of GAP at 4.0 mM pyruvate. c No error analysis was reported for the kinetic constants measured for A. tumefaciens DXS. steady-state kinetic constants similar to other DXS enzymes. Apo-DXS cc is inactive and insoluble suggesting that the bound Mg(II) has both a catalytic and structural role in the enzyme. We also provide some insight concerning the bipartite pre-sequence N-terminus to the mature catalytic core of P. vivax DXS. While not conclusive, our data suggest that the bipartite pre-sequence spans from amino acid 1 to ∼250 in the full length protein. We propose that the first ∼25 amino acids represent the signal peptide while amino acids ∼25 to ∼250 represent the transit peptide. A proposed length of ∼225 amino acids for transit peptide of P. vivax DXS is within the range reported for the transit peptides of other proteins characterized from Plasmodium (Table S1). The results presented herein will foster the development of DXS inhibitors to treat P. vivax malaria and provide a better understanding of the mechanisms used in Plasmodium for the transit of proteins coded in the nucleus to the lumen of the apicoplast.