Molecular Characteristics and Serodiagnostic Potential of Dihydrofolate Reductase from Echinococcus granulosus

The larval stage of Echinococcus granulosus causes cystic echinococcosis (CE), a neglected tropical disease that leads to morbidity and mortality in humans and livestock worldwide. Here, we identified and characterized dihydrofolate reductase (Eg-DHFR) from E. granulosus, and evaluated its potential as a diagnostic antigen for sheep CE. Comparison between mammalian (host) DHFR and Eg-DHFR indicates that 45.7% of the 35 active site residues are different. Immunolocalisation analysis showed that native Eg-DHFR was widely distributed in all life-cycle stages of E. granulosus. Recombinant Eg-DHFR (rEg-DHFR) showed typical DHFR enzymatic parameters towards substrate, and was very sensitive to inhibition by methotrexate (IC50 = 27.75 ± 1.03 nM) and aminopterin (IC50 = 63.67 ± 6.76 nM). However, inhibition of DHFR exhibited little protoscolicidal effect in vitro. As there is no reliable method to monitor sheep CE, the immunogenicity of rEg-DHFR was detected, and we developed an indirect ELISA (iELISA) for CE serodiagnosis. The iELISA exhibited diagnostic specificity of 89.58%, diagnostic sensitivity of 95.83%, and the diagnostic accuracy was 91.67% compared with necropsy. Cross-reactivity assay showed analytical specificity of 85.7%. These suggest that rEg-DHFR is an effective antigen for the diagnosis of sheep CE.


Results
Sequence analysis of Eg-DHFR. The cDNA encoding Eg-DHFR contained an open reading frame of 576 bp, corresponding to a protein of 191 amino acid residues. The predicted molecular mass of the protein was 21.9 kDa. Multiple sequence alignment revealed that Eg-DHFR had high similarity with the DHFRs from E. multilocularis (96.34%), Hymenolepis microstoma (96.34%) and Taenia solium (85.86%), and shared 41.54-42.56% overall identity with DHFRs from mammalian hosts (Fig. 1). Previous studies on mammalian DHFRs showed there are 35 active site residues in the enzyme 29 . Comparison of the active site residues between mammalian DHFRs and Eg-DHFR indicated that there are 19 conserved residues and 16 different residues (i.e., 45.7% of the total active site residues are different). Otherwise, it is worth to note the following points: (i) charge changes occurred between mammalian host and E. granulosus DHFR at four sites, including the loss of a negative charge Figure 1. Sequence alignment analysis of E. granulosus DHFR. Alignment of the deduced amino acid sequence of Eg-DHFR (GeneDB: EgrG_000572400) with homologues from other species. The percentage homology of Eg-DHFR with each DHFR is shown at the end of the alignment. Different active-site residues between mammalian (host) and E. granulosus DHFRs are indicated by green letters, and identical active-site residues are indicated by red letters. Active-site residues with charge changes are marked with blue boxes (mammalian host -E. granulosus, Asn30 -Glu30, Asp/Asn22 -Lys22, Arg/Lys33 -Thr33, and Lys68 -Pro68). The exchange of polar and nonpolar residues (Tyr34 -Phe34, Asn64 -Phe64, Val136 -Tyr135) is marked with red boxes. The NADPH binding-site is highlighted with a blue background. The predicted secondary structure of Eg-DHFR is displayed above the alignment. Accession numbers: Echinococcus granulosus DHFR (E.g-DHFR) GeneDB: (mammalian host -E. granulosus, Asn30 -Glu30), the addition of a positive charge (Asp/Asn22 -Lys23), and the loss of a positive charge (Arg/Lys33 -Thr33, Lys68 -Pro68); (ii) further changes were the exchange of polar and nonpolar amino acids (Tyr34 -Phe34, Asn64 -Phe64, Val136 -Tyr135) (Fig. 1). Although the modelled three-dimensional structure of Eg-DHFR is conserved compared with homologous DHFRs, conformation changes were predicted in some active site residues (e.g., H. sapiens -Eg-DHFR: Gly21 -Asn22, Asp22 -Lys23, Trp24 -Trp26, Phe31 -Met32, Asn64 -Phe64) (Fig. 2).
Expression and identification of rEg-DHFR. The Eg-DHFR gene was successfully inserted into the pET28a (+) expression vector and a soluble protein was expressed in Escherichia coli BL21 (DE3) cells following induction. The purified recombinant protein showed a single band in 12% SDS-PAGE, which was similar in size to the predicted ~26 kDa (including a His-tag). In western blotting analysis, a single ~26 kDa band was recognised using both rabbit anti-rEg-DHFR IgG (positive control) and sheep anti-E. granulosus serum (experimental group), which suggested strong reactivity and good antigenicity of this recombinant protein (Fig. 3). In addition, anti-rEg-DHFR rabbit IgG recognised a ~22 kDa protein from extracts of E. granulosus protoscoleces (PSCs), corresponding to the size of native Eg-DHFR. No band was recognised by native (preimmune/uninfected) serum from rabbit and sheep.

Figure 2.
Comparison of the three-dimensional structure of DHFR between E. granulosus and Homo sapiens. The 3D structure of Eg-DHFR was modelled based on the crystal structure of Schistosoma mansoni DHFR (PDB accession code: 3vco.1). The global and per-residue model quality of Eg-DHFR has been assessed using the QMEAN scoring function (GMQE and QMEAN value were 0.76 and −0.14, respectively). 3D structure of Eg-DHFR (deepsalmon) was aligned to DHFR of H. sapiens (cyan; PDB accession code: 3gyf.1), and shown as cartoon (A) and surface (B). Important structural difference were enlarged. rEg-DHFR; lane 3, purified rEg-DHFR probed with anti-rEg-DHFR rabbit serum; lane 4, purified rEg-DHFR probed with native (preimmune) rabbit serum; lane 5, purified rEg-DHFR probed with the serum of E. granulosus infected sheep; lane 6, purified rEg-DHFR probed with native (healthy) sheep serum; lane 7, the total protein from protoscoleces probed with anti-Eg-DHFR rabbit serum; lane 8, the total protein from protoscoleces probed with native (preimmune) rabbit serum.

Immunolocalisation of endogenous Eg-DHFR. Endogenous Eg-DHFR was localised in different
life-cycle stages of E. granulosus by an immunofluorescence method using rabbit rEg-DHFR antibody. In PSCs, the fluorescence signals were mainly localised in the parenchymal region, while weak signals were also detected in the tegument tissues (Fig. 4). Fluorescent signals were observed in the germinal layer. In adult worms, Eg-DHFR was widely distributed in the parenchymal region and tegument tissue (Fig. 4), and especially distributed in the rostellum and suckers of the scolex (see Supplementary Fig. S1). No fluorescence signals were detected in the negative controls. Enzymatic activity of rEg-DHFR and inhibitory effect of antifolates. The kinetic properties of rEg-DHFR and human DHFR were investigated. By using the Michaelis-Menten curve fit, we calculated the K m and V max value of rEg-DHFR and human DHFR. The K m and V max value of rEg-DHFR were 18.74 ± 0.5651 μΜ and 2.384 ± 0.01874 μmol/min/mg, respectively; the K m and V max value of human DHFR were 14.61 ± 0.6590 μΜ and 3.370 ± 0.02327 μmol/min/mg, respectively. Methotrexate, aminopterin, trimethoprim and pyrimethamine were inhibitors of DHFR, which share structural similarities with the substrate of DHFR (DHF). The effect of inhibitors on the enzymatic activity were examined. Methotrexate and aminopterin were potent inhibitor with an IC 50 of 27.75 ± 1.03 nM and 63.67 ± 6.76 nM for rEg-DHFR respectively, while pyrimethamine and trimethoprim showed lower inhibition effect with IC 50 of 0.86 ± 0.0935 μM and 30.81 ± 2.09 μM, respectively (Fig. 5B-E and Table 1).
The protoscolicidal activity of antifolates in vitro. Based on the IC 50 concentrations of antifolates derived from the enzyme activity inhibition assays, we treated the PSCs with these antifolates at different concentrations. Praziquantel was demonstrated to exhibit protoscolicidal activity, which was used as a positive control 30 . The survival of PSCs after exposure to antifolates were shown in Fig. 6. Control PSCs were not altered and remained viable (82.65 ± 6.11%) after 7 days of incubation. At all three concentrations, methotrexate and aminopterin showed slight effect on PSCs after 7 days of incubation. In the group treated with the highest concentration (800 μM) of pyrimethamine, the percentage vitality of PSCs was only 20.67 ± 2.49%. Trimethoprim showed a weaker protoscolicidal effect than pyrimethamine, with 38.67 ± 6.18% (3000 μM) of PSCs remaining viable in culture after 7 days. Establishment of indirect ELISA. Based on the checkerboard titration protocol, the optimal concentration of antigens and serum dilution were determined ( Table 2). To determine the cut-off value of the indirect ELISA, 24 negative serum samples from sheep were tested in the optimised conditions. The mean OD 450 value of rEg-DFHR ELISA was 0.255 and the standard deviation was 0.0488. Thus, the cut-off value was 0.4014 (mean + 3 SD). The cut-off values of other antigen ELISAs were listed in Table 2.

Discussion
DHFR catalyses the NADPH-dependent reduction of DHF to THF, which is involved in subsequent reactions such as thymidylate and purine nucleotide biosynthesis 36 . DHFR has been studied extensively in a variety of organisms, including mammals 37 , protozoa 38 , bacteria 39 and several insects 27,40 . Moreover, DHFR can be used as a drug target in the treatment of cancer and bacterial and parasitic infections 25 . However, the DHFR gene has not been described in tapeworms. This study characterised E. granulosus DHFR, and established an iELISA method based on the recombinant protein to diagnose CE in sheep.
The potential for using Eg-DHFR as a drug target for the treatment of E. granulosus has been analysed in our study. A comparison of the active site residues was performed between the DHFR sequences of mammalian hosts and E. granulosus, and they showed a difference of 45.7% (16/35). Among these different active site residues was the "REDMTFFS" motif, the region with the largest number of active site residues, which contained many potentially significant changes with respect to the design of non-mammalian specific DHFR inhibitors. As observed in previous studies of non-mammalian DHFRs, there were also residue charge changes at the NADPH binding site between H. sapiens DHFR (negative charge) and Eg-DHFR (positive charge), which results in a different direction and size of the loop on the NADPH site of the two enzymes. This charge and conformation change raise the possibility of designing specific inhibitors directed towards E. granulosus by introducing a negatively charged group at this position 27,29 . Simultaneously, it is interesting to note that modelled three-dimensional structure changes between mammalian hosts and E. granulosus DHFR at two positions, where the α-helix is replaced as random coil (Asn64 -Phe64) or the random coil is replaced as α-helix (Trp24 -Trp26) ( Fig. 2A). These differences in both residue type and three-dimensional structure lead to the appearance of a deep, narrow groove in the surface of E. granulosus DHFR, but not in the surface of human DHFR. This is a very attractive feature as its great significance for the design of inhibitors with high selectivity for E. granulosus DHFR. In addition, other potential significant differences are in the glutamate sub-site and the p-aminobenzamide binding region, where residues Met32 and Phe64 in E. granulosus are changed to be Phe31 and Asn64 (Fig. 2B). Referring to previous studies, this could expose a larger hydrophobic binding pocket in Eg-DHFR and provide a significant reference for the design of inhibitors 27,29 .
Recombinant Eg-DHFR exhibited enzymatic parameters towards the substrate DHF similar to those of other DHFRs (K m of 18.74 ± 0.5651 μM and V max of 2.384 ± 0.01874 μmol/min/mg) 40,41 . Native Eg-DHFR is difficult and expensive to prepare. However, rEg-DHFR retains enzymatic activity that can be used to further screen newly designed drugs to inhibit E. granulosus, which greatly reduces the difficulty of the screening of inhibitors. The inhibitors of DHFR belong to antifolates, which share structural similarities with the substrate of DHFR (DHF). The effect of antifolates on the enzymatic activity of rEg-DHFR were investigated in this study. Methotrexate and aminopterin were shown to be stronger inhibitors of both rEg-DHFR and human DHFR enzymes than pyrimethamine and trimethoprim. It is attributed to the fact that methotrexate and aminopterin have a pteridine ring and glutamic acid, which are the most similar to the normal substrate of DHFR (DHF), and lead to the effective inhibition (Fig. 5). Due to the inhibitory effect of on the activity of the rEg-DHFR enzyme, it is interesting to explore whether these antifolates could affect PSCs in vitro. Consequently, even at concentrations that are orders of magnitude higher than the IC 50 for EgDHFR, there is little effect on protoscolex viability. Therefore, although the antifolates were very effective against EgDHFR, inhibition of DHFR is of little consequence to protoscolex viability in vitro.
There were no research on the location of DHFR in parasitic helminths. Our results revealed that Eg-DHFR was widely distributed in the larva, germinal layer, and adult worm of E. granulosus. In particular, a high level of Eg-DHFR protein was detected in the rostellum and suckers (scolex) of adult worms, but not in PSCs. The rostellum and suckers of adult worms are involved in absorption of nutrition and adhesion to the walls of the small intestine of the definitive host, which are of major importance for the survival of the parasite 42,43 .
CE is a worldwide economic and public health problem. Currently, detection and surveillance of E. granulosus infection in livestock relies on necropsy and macroscopic observation procedures in abattoirs 44,45 , and these detections without histological examination have a high error rate (15.4%) 46 . It is important to establish an inexpensive, accurate immunodiagnostic assay as a monitoring tool for the detection of CE in live animals 17 . The antigen B (8 kDa) from hydatid cyst fluid has high diagnostic sensitivity and specificity (over 90%) to detect CE in sheep 18,19 . However, it is hard to purify and standardise large amounts of natural E. granulosus hydatid cyst fluid antigens. Thus, recombinant protein antigens have become an important way to enhance immunodiagnosis of CE in sheep. Previous research using the recombinant EG95 oncosphere protein and recombinant AgB protein as antigens in ELISA exhibited poor diagnostic sensitivity, 1.6% and 28% for sheep CE, respectively 15,21 . Here, an iELISA diagnostic method based on recombinant Eg-DHFR was successfully established and optimised to detect E. granulosus infection in sheep. Compared with iELISAs using other recombinant antigens of E. granulosus (rEg-cystatin, rEg-annexin, rEg-tetraspanin-1, rEg-prohibitin and hydatid fluid), recombinant Eg-DHFR had high analytical sensitivity (1:6400) and analytical specificity (85.7%). There was no cross-reaction detected with C. cerebralis-positive serum. However, other antigens showed three to seven cross reactions with C. cerebralis-positive serum. DSN, DSP and diagnostic accuracy were 95.83% (23/24), 89.58% (43/48) and 91.67% (66/72) compared with the results of necropsy, which suggested that the recombinant Eg-DHFR could clearly react with E. granulosus-specific IgG antibodies. Recombinant Eg-DHFR had a similar diagnostic performance as rEg-glutaredoxin-1 compared with our previous report 35 . Although a lower cross-reaction ratio of rEg-DHFR was observed when against taeniid cestodes positive sera compared with rEg-glutaredoxin-1. However, the difference is not significant because of the small sample size. Therefore, the cross-reaction ratio should be calculated more accurate by further study using large sample number.

Conclusion
We identified a new E. granulosus diagnostic antigen candidate distributed in all life-cycle stages of the parasite. The bioinformatic, tissue distributions and enzymatic characteristics of Eg-DHFR was made in this study. The results demonstrate that Eg-DHFR is a potential drug target for control of the important cestode parasite E. granulosus. Furthermore, we established an iELISA based on rEg-DHFR, which might be a promising tool for diagnosis and serosurveillance of E. granulosus infections in sheep.

Parasites and animals.
Cysts of E. granulosus were obtained from a slaughterhouse in Qinghai Province, China. The PSCs were separated in sterile conditions and washed three times in phosphate-buffered saline (PBS). Rinsed PSCs were diluted to a concentration of 1,500 mL −1 and maintained as previously described 33,47 . Briefly, PSCs were cultured in RPMI 1640 medium (Hyclone) with 10% fetal bovine serum (Hyclone), 10% hydatid   Expression and purification of rEg-DHFR. The region encoding mature Eg-DHFR was amplified from E. granulosus cDNA using the primers 5′-CGCGGATCCATGGGGCTGAAGCGTCT-3′ and 5′-CCGCTCGAGATGATCATTAAGGGGATGCG-3′, and then ligated into the BamHI/XhoI restriction sites of vector pET-28a(+) (Novagen, Madison, WI, USA). Recombinant protein was expressed and purified as previously described 33 . Briefly, the recombinant plasmid was transformed into E. coli BL21 (DE3) cells (Invitrogen, Carlsbad, CA, USA) and expression was induced with 1 mM isopropyl-thio-β-D-1-thiogalactopyranoside at 37 °C for 5 h. The bacterial cells were harvested and suspended in lysis buffer, followed by ultrasonic lysis. The recombinant protein was purified by Ni 2+ affinity chromatography (Bio-Rad, Hercules, CA, USA) following the manufacturer's instructions. Recombinant protein was detected by 12% SDS-PAGE and the concentration of protein was estimated using a bicinchoninic acid protein assay kit (Pierce, Rockford, IL, USA).
Serum and antigen preparation. Positive serum against E. granulosus (24 samples) was collected from naturally infected sheep in Ganzi Tibetan Autonomous Prefecture, Sichuan Province, China. Goat serum positive against Cysticercus tenuicollis (7 samples) and sheep serum positive against Coenurus cerebralis (7 samples) was also obtained from farms in Sichuan Province. Negative serum (72 samples) was collected from 72 healthy sheep with no cysts (determined by autopsy) from Xichang, Sichuan Province. The polyclonal antibody against rEg-DHFR was obtained as previously described 33 . Briefly, each rabbit was inoculated subcutaneously with 200 μg rEg-DHFR emulsified in Freund's complete adjuvant (Sigma, St. Louis, MO, USA), followed by three repeat inoculations every 2 weeks with 200 μg rEg-DHFR emulsified in Freund's incomplete adjuvant (Sigma). Rabbit anti-rEg-DHFR serum was detected by ELISA and purified using HiTrap Protein A affinity chromatography (Bio-Rad). The rEg-cystatin, rEg-annexin, rEg-tetraspanin-1 rEg-prohibitin and rEg-glutaredoxin-1 (the gene accession number are EgrG_000849600, EgrG_000041300, FJ384717, KT149769 and EgrG_000124800, respectively) were all produced similarly as described above. The crude hydatid fluid antigen was extracted from the hydatid cysts with 20 ml syringe. Then the fluid was clarified by centrifugation at 10,000 rpm for 20 min and filtered through a 0.22 μm membrane 19, 49 . Immunoblot and immunohistochemical analyses. The total proteins of PSCs were obtained with a mammalian protein extraction kit (CWBIO, Beijing, China). Purified rEg-DHFR and total PSC proteins were separated by 12% SDS-PAGE and transferred onto nitrocellulose membranes. The membranes were blocked with 5% (w/v) skim milk for 2 h at room temperature. Subsequently, the membranes were incubated with E. granulosus-infected sheep serum or anti-rEg-DHFR rabbit IgG (1:200 v/v dilutions) overnight at 4 °C. Following washing, the membranes were incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG or   rabbit anti-sheep IgG (1:3,000 dilutions; Boster Bio-project Co, Wuhan, China) for 1 h, respectively. Signals were visualised using an Enhanced HRP-DAB Chromogenic Substrate Kit (Tiangen, Beijing, China). Negative controls were performed with serum of healthy sheep and preimmune rabbit. Immunolocalisation studies were performed as previously described with some modifications 33 . Briefly, fresh adult worms, PSCs and germinal layer were fixed and embedded in paraffin. The sections were prepared with 0.01 M citrate buffer (pH 6.0) for 12 min for heat-induced antigen retrieval. Then the sections were probed with purified rabbit anti-rEg-DHFR IgG (1:200 dilution) and reacted with fluorescein (FITC)-conjugated goat anti-rabbit IgG (H+L) (1:200 dilution in 1% Evans Blue; Bethyl Laboratories, Montgomery, TX, USA) for 1 h at 37 °C in darkness. The nucleus was stained with 4′,6-diamidino-2-phenylindole (DAPI; Sigma, USA) for 5 min. The sections were imaged with a fluorescence microscope (Nikon, Tokyo, Japan). Negative controls were performed with preimmune rabbit serum.
Enzyme assay. The enzymatic characteristics of rEg-DHFR and human DHFR enzyme (DHFR assay kit; Sigma) were analysed using a DHFR assay kit (Sigma) following the manufacturer's instructions 41 . Briefly, a range of purified rEg-DHFR concentrations was added to the assay buffer and mixed well and the optimal concentration was determined. Reaction progress was measured based on the decrease in absorbance at 340 nm. Data was recorded at 25 s intervals for 2.5 min. The temperature was controlled at 22 °C. The extinction coefficient for the DHFR reaction was 12.3 mM −1 cm −1 at 340 nm. One unit was defined as the amount of enzyme required to convert 1 μmol of DHF per min to THF at 22 °C. Thereafter, to estimate the kinetic parameters maximal velocity (V max ) and Michaelis constant (K m ), different concentrations (ranging from 0 μM to 120 μM) of DHF and a constant concentration (60 μM) of NADPH were added into the reaction system. Kinetic parameters (V max and K m ) were calculated using GraphPad Prism ® software (San Diego, CA, USA; http://www.graphpad.com) based on a Michaelis-Menten curve fit.
Determination of IC 50 for antifolates. Different concentrations of antifolates (Sigma) were used to measure the 50% inhibition of the enzyme reaction (IC 50 ) containing 50 μM dihydrofolic acid and 60 μM NADPH. The IC 50 was calculated using GraphPad Prism software by nonlinear curve-fitting.

Establishment of the indirect ELISA.
ELISAs were performed essentially as described 52,53 . Briefly, the optimal concentration of antigens (rEg-DHFR, rEg-cystatin, rEg-annexin, rEg-tetraspanin-1, rEg-prohibitin and hydatid fluid antigens) and serum were assessed by standard checkerboard titration procedures. The ELISA plates were coated with six different concentrations overnight at 4 °C. After washing, the plates were blocked with 5% skim milk for 1.5 h at 37 °C. After the blocking solution was discarded, the plates were incubated with sheep serum samples in twofold dilutions ranging from 1:20 to 1:2560. Subsequently, the plates were incubated with a 1:200 dilution of HRP-conjugated rabbit anti-goat/sheep IgG (Boster Bio-project Co) at 37 °C-for 1 h. Following washing, antibody binding was detected with 100 μL of 3, 3′, 5, 5′-tetramethylbenzidine (Tiangen). After stopping the reaction, the absorbance was determined at 450 nm in a microplate reader (Thermo Scientific, Pittsburgh, PA, USA). An OD 450 value for positive serum close to 1.0 and the highest P/N value between positive and negative serum were regarded as optimal. The cut-off value was calculated as the mean OD 450 absorbance value for the 24 negative serum samples plus three standard deviations (3 SD). Intraplate repeatability was evaluated by the coefficient of variation (CV %) of every serum sample. Three separate assays were used to evaluate the intraplate repeatability.
Analytical sensitivity and specificity of indirect ELISA. To evaluate the analytical sensitivity of the indirect ELISA, three E. granulosus-positive sheep serum samples were diluted in twofold series from 1:100 to 1:102,400. The minimum detection limit of the positive serum samples was determined as the analytical sensitivity of the indirect ELISA by comparing with the cut-off value 54 . To evaluate the analytical specificity of the indirect ELISA, cross-reactions were tested with C. cerebralis-positive serum and C. tenuicollis-positive serum. The percentage analytical specificity was calculated as indirect ELISA negative × 100/true negative.
Estimates of the indirect ELISA. The DSN, DSP and diagnostic accuracy of the indirect ELISA was evaluated using 72 sheep serum samples, including 24 positive sera and 48 negative sera. The DSN, DSP and accuracy of the iELISA were calculated based on the following formulae 54 : DSN = true-positive/(true-positiv e + false-negative) × 100, DSP = true-negative/(true-negative + false-positive) × 100 and diagnostic accuracy = (true-positive + true-negative)/total number × 100. Each serum sample was tested three times. Statistical analysis. All data are presented as the mean ± SD. Statistical analyses were performed by t-test and one-way ANOVA for comparison between groups using the software package GraphPad Prism. P values < 0.05 were considered significant.