Population genetic and biophysical evidences reveal that purifying selection shapes the genetic landscape of Plasmodium falciparum RH ligands in Chhattisgarh and West Bengal, India

Stored genomic DNAs previously isolated form peripheral blood samples collected from P. falciparum malaria infected patients admitted in Calcutta National Medical College and Hospital, Kolkata, in the year 2010 were utilized in this study [13]. Kolkata is the capital of West Bengal. Almost 10% of the total malaria cases in India are accounted in West Bengal [14]. Peripheral venous blood samples were also collected from Surguja located in Chhattisgarh, central India, during a period of 2010–2013. Surguja, incidentally a tribal dominated district, is situated in the forested area of Chhattisgarh [15]. Due to its distinct ecological and geographical conditions, Chhattisgarh contributes to ~ 12% of total malaria burden and carries the highest share of deaths (17%) in India [16]. Only blood samples from P. falciparum malaria positive patients confirmed by rapid diagnostic tests based on dual-Antigen and/or Giemsa-stained thick and thin smears were selected for this study [15]. Individuals suffering from co-infection with Plasmodium vivax and pregnant women were excluded. Additionally, the study excluded the patients suffering from chronic or severe disease conditions, such as cardiac, renal or hepatic diseases, G6PD deficiency, typhoid, measles, acute lower respiratory tract infection, bacteraemia, severe diarrhoea with dehydration, sickle cell anaemia, AIDS and cancer. Genomic DNA was isolated from the blood sample of P. falciparum malaria infected patients using QIAamp DNA Blood Midi Kit (Qiagen, Hilden, Germany) according to manufacturer’s protocol.

Oligonucleotide primers for polymerase chain reaction and sequencing of genes encoding PfRHs were designed by retrieving reference sequences of P. falciparum 3D7 from PlasmoDB (Fig. 1) [17]. Receptor binding sites of PfRH1 and PfRH4 were amplified using two pairs of overlapping primers for each [18‐20]. PfRH2a and PfRH2b possess indistinguishable ecto-domain containing the putative receptor binding site (495-860 amino acid region of PfRH2a/b in Pf3D7) [18, 19]. This identical receptor binding site of PfRH2a/b was considered for primer designing and amplification. An attempt was made to amplify the whole PfRH5 gene using four overlapping primer pairs. The primer sequences as well as cycling conditions were described in the Additional file 1: Table S1. Amplification of target regions were performed in 15μL reaction mixtures containing 1 U of GoTaq^® Flexi DNA polymerase (Promega), 0.2 mM dNTP, 1.5 mM MgCl₂, and 0.4 μM of each primer using a GeneAmp^® PCR System 9700 (Applied Biosystems). UV transillumination on gel documentation system (Biostep) was used to visualize the PCR products preceded by electrophoresis on 2% agarose gel (Promega). PCR products were purified by Qiaquick gel extraction kit (QIAGEN India Pvt. Ltd, Hilden, Germany) and sequenced using the same primers utilized in PCR. Sequencing PCR was performed using forward and reverse primers (separately) and Big Dye v3.1 dye terminator on ABI Prism 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA) [13]. Number of samples analysed for each target gene were mentioned in Table 1.

Raw sequence data files from field isolates were manually revised to exclude indefinite reads and signal noise. To compare the sequence identity, NCBI BLAST analysis was performed with all test sequences with respect to reference sequences (Gene ID of Pfrh1: PF3D7_0402300; Pfrh2a: PF3D7_1335400; PfRH2b: PF3D7_1335300; Pfrh4: PF3D7_0424200; Pfrh5: PF3D7_0424100) [21]. MEGA 7 tool was used to perform multiple sequence alignment of the nucleotide sequences and to translate them into amino acid codes. Nucleotide sequences analysed in this study were submitted to the GenBank database under the accession numbers MN937274-MN937320, MN950796-MN950967.

Variants were identified by aligning the sequence reads with corresponding reference sequences. Varied number of single nucleotide polymorphisms (SNPs) was observed in both study zones. Genetic diversity parameters including (i) number of segregating (S) sites, (ii) average number of pairwise nucleotide differences within population (k), (iii) average number of observed nucleotide differences per site between any two sequences believing that the sample was randomly selected (π), (iv) Watterson’s θ (θ_w) were estimated using DnaSPv6.00 [22, 23]. The estimation of the recombination phenomena represented by the minimum number of recombination events (Rm) that happened along the isolates was also performed. Tajima’s D and Fu & Li’s statistics were deployed using DnaSP to assess the neutral theory of evolution [24‐26]. To gauge the impact of natural selection over gene sequences the ratio of non-synonymous to synonymous substitutions (dN/dS) is commonly used [27]. The method of Nei and Gojobori with the Jukes and Cantor correction was used to assess these rates of substitution within species as implemented in the MEGA7 [28, 29]. Five hundred bootstrap replications were applied to estimate p values as well as standard error. To derive an estimate of how did parasite sequence data generated in this study differ from those available in global data, particularly those from sub-Saharan Africa, Pfrh sequences were retrieved from NCBI and diversity analysis was performed (Additional file 2: Table S2). Further, the evolutionary connections among nucleotide haplotypes of each Pfrh family members were predicted using Network program [27, 30‐33].

To analyse the impact of genetic variability on ligand-receptor interaction efficacy, the most prevalent PfRH5 haplotype (rh5h1) was selected to compare with its 3D7 type counterpart (PfRH5_3D7). C203 residue in PfRH5_3D7 was replaced by Y203 in the most prevalent PfRH5 haplotype rh5h1. 4u0q.pdb protein file representing complex of rh5h1 (hereafter, mentioned as PfRH5_M) with its erythrocyte surface receptor BSG was extracted from RCSB and the interaction efficiency was compared with that of PfRH5_3D7 and BSG [34]. In order to introduce C203 residue, the side-chain of residue number 203 (Y) of ligand was removed from the 4u0q.pdb file and the residue name of the backbone parts was changed from Y to C. Thereafter, AUTODOCK tool was applied to construct the side chains of the newly modified residue [35]. The pdb file of the PfRH5_3D7 - BSG complex was thus generated. Both the interaction complexes were refined by minimizing their energies separately in equal number of steps until the difference in the energy between the last two steps is lower than 10% of the preceding step. Afterwards, the effect of the mutation was studied by comparing the residue-wise interaction patterns in the two energy minimized complexes and the H-bonding patterns of two (C/Y) residues at 203 in those structures.

To compute the residue-wise interaction pattern between ligand and receptor, partial atomic charges were assigned to all the atoms of the both complexes using CHARMM. Interaction of the atoms of individual residues in ligand with whole BSG was computed considering two components (i) electrostatic interaction and (ii) van der Waals interaction. Residue-wise interaction in the two complexes and the H-bonding patterns of the residue 203 in the two complexes were also studied.

To validate the result obtained from in silico analysis, the binding interaction of PfRH5 with BSG was studied in further details. A 564-bp fragment of the Pfrh5 gene encoding the 188 aa (Glutamine-35—Serine-222 residues) which encompasses receptor binding site, was PCR amplified from DNA of a P. falciparum infected blood sample possessing PfRH5_M type sequence using forward primer: CGCGGATCCCCAAGAAAATAATCTGAC and reverse primer: ACATATGACAAAGTGAAAAGTAAGCTTGCG [36]. Polymerase chain reaction was carried out in 15μL reaction mixtures having 1 U of GoTaq^® Flexi DNA polymerase (Promega), 0.2 mM dNTP, 1.5 mM MgCl₂ and 0.4 μM of each primer. The cycling conditions for PCR consisted of an initial denaturation at 94° C for 5 min, followed by 38 cycles of denaturation at 94° C for 45 s, annealing at 58° C for 45 s, extension at 72° C for 45 s, and a final extension at 72° C for 5 min. PCR product was visualized using UV transillumination on gel documentation system (Biostep) following electrophoresis on 2% agarose gel (Promega) and were purified by Qiaquick gel extraction kit (QIAGEN India Pvt. Ltd, Hilden, Germany). The PCR amplicon was digested with BamH I and Hind III (New England Biolabs, Beverly, MA) and inserted downstream of the T7 promoter in the E. coli expression vector, pET-20b(+) (Novagen, San Diego, CA) using a T4 DNA ligase (Fermentas), to obtain the plasmid rPfRH5_M- pET-20b. Site-directed mutagenesis (SDM) was performed on the plasmid rPfRH5_M- pET-20b to insert C in exchange of Y at 203 amino acid position to generate rPfRH5_3D7- pET-20b using the SDM-specific primer SDMF: 5′-GTCCTCTACATATGGAAAGTGTATAGC-3′ and SDMR: 5′-AGCATCTACAGCTATACACTTTCCAT-3′. The transcribed sequence of the recombinant plasmids (rPfRH5_3D7- pET-20b and rPfRH5_M- pET-20b) contained an additional His-tag (LEHHHHHH) at the C terminus. E. coli BL21 (DE3) cells (Novagen, San Diego, CA) were transformed with rPfRH5_3D7- pET-20b and rPfRH5_M- pET-20b and used for the expression of rPfRH5_3D7 and rPfRH5_M. Sequencing of the plasmid chimera was used to confirm that two different PfRH5 alleles were inserted in correct reading frame.

Gene expression was performed in a 100 ml culture, using Luria–Bertani medium containing 100 mM ampicillin at 37 °C. Pilot experiments determined that 150 ml bacterial culture incubated for 16 h at 37 °C gave optimal expression of recombinant parasite proteins in culture medium. Once the optical density at 600 nm reached 0.6, the culture was incubated overnight at shaking condition (160 rpm) at 37 ^°C for protein expression. Cells were harvested by centrifugation next day and the cell pellet was stored at −80 °C. The recombinant protein of the expected molecular mass (24.5 kDa) was present in a total cell lysate.

The frozen cell pellet was resuspended in 800 ml of lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole (Qiagen); pH 8.0), mixed at 4 °C for 1 h, and lysed by ultrasound technology (UP200S Ultrasonic Processor; 0.7 cycle; 80% Amplitude), followed by centrifugation at 12,000 rpm for 20 min. Supernatant containing protein was applied on 133 µl of nickel Ni–NTA resin (Qiagen) equilibrated with lysis buffer. To remove non-specifically bound proteins, 1 ml of the wash buffer (pH 7) containing 50 mM imidazole was used. Bound proteins were eluted with elution buffer containing 500 mM imidazole. Eluted fractions were analysed by SDS-PAGE and Western blotting. Concentrations of the fractions containing recombinant PfRH5 was determined by Nanodrop (Thermo Scientific) and stored at −80 °C.

Crude lysates as well as purified proteins were separated in the presence of SDS using 15% polyacrylamide gel and visualized with Coomassie^® Brilliant blue R 250 (Merck). Approximately, 30-50 ug/lane purified protein was electrophoresed by SDS-PAGE using Prism Ultra Protein Ladder (ab116028, abcam; 10–245 kDa) and transferred to polyvinylidene fluoride membrane (Pall Corporation, Port Washington, USA). Blots were treated with primary (mouse monoclonal antibody directed against 6X His tag; ab18184, Abcam) and secondary antibodies and incubated with elctrochemiluminescence (ECL) reagent (SuperSignal West Pico Chemiluminescent Substrate, Thermo Scientific) and examined on Bio-Rad-Molecular imager ChemiDoc™ XRSt with ImageLab^TM software.

The folding state of the purified proteins was evaluated by circular dichroism (CD) spectroscopy. CD spectra were recorded on a J-815 CD Spectrometer (Jasco) at 25 °C with the path length of 1 mm. The protein sample solution was diluted in PBS (pH 7.4) so that a final protein concentration was 2 μM. Secondary structural changes were recorded in the range of 190–260 nm. Each spectrum represented an average of three scans and CD data were expressed as mean residue molar ellipticity [θr].

FTIR was carried out using PerkinElmer spectrum 100 FT-IR Spectrometer at a resolution of 0.5 cm⁻¹ in the wavenumber range 1000–4000 cm⁻¹. 10 ul of solution containing 1ug/ul protein was utilized for each FTIR analysis.

Synchronous fluorescence patterns of PfRH5_3D7, PfRH5_M, PfRH5_3D7–BSG and PfRH5_M–BSG were observed in different Δλ at 25 °C [37, 38]. Since, in synchronous fluorescence, maximum peak intensity was shown at 379 nm while Δλ was 70, fluorescence anisotropy of BSG, PfRH5_3D7, PfRH5_M, PfRH5_3D7 –BSG (1:1), and PfRH5_M–BSG (1:1) were performed in the range between 309 and 379 nm (Δλ = 70) for 300 s with both excitation and emission bandwidth of 5 nm using a Varian Cary Eclipse (USA) spectrofluorimeter [39].

Isothermal calorimetry is a quantitative tool used to evaluate thermodynamic profile of interaction between two or more molecules in solution, while finding out equilibrium binding constant (K_a) [40]. In this study, BSG was added drop by drop in a fixed volume with 1 min time interval in the sample cell containing either PfRH5_3D7 or PfRH5_M in a VM2: Cell - ITC 200 (micro cal) titration micro-calorimeter. Concentration of BSG was 10 times higher than PfRH5. Calorimetric reference cell was loaded with 200 μl of Dulbecco phosphate buffer saline solution (pH 7.4). Titration curve for PBS-BSG interaction was subtracted from the heat of binding reaction of PfRH5–BSG to obtain the effective heat of binding. The resulting titration curves were fitted for one set of binding site using MicroCal Origin software.

To infer the nature of the evolutionary forces shaping the genetic landscape of Pfrh members, fragments encoding Pfrh1, Pfrh2a, Pfrh2b, Pfrh4 and Pfrh5 were amplified from genomic DNA isolated from peripheral blood samples of malaria patients. Overlapping sequences from each sample were assembled before estimating the genetic diversity parameters.

To understand the exact role of C203Y, a ubiquitously observed nonsynonymous variation, the efficacy of interaction of BSG with PfRH5 encoded by the most prevalent (frequency = 0.63) haplotype, rh5h1 (hereafter mentioned as rPfRH5_M) was compared to that encoded by the reference Pf3D7 allele (rPfRH5_3D7). Protein file, 4u0q.pdb, representing a complex of rPfRH5_M with BSG was extracted from RCSB protein data bank. A pdb file containing only rPfRH5_M was produced using AUTODOCK followed by generation of rPfRH5_3D7 pdb from the one containing rPfRH5_M. Ribbon models constructed using pdb files of rPfRH5_3D7-BSG and rPfRH5_M-BSG complexes indicated that the Y203 was closer to BSG than C203 (Fig. 3a, b). The interaction efficiencies of two PfRH5 variants with BSG estimated using CHARMM showed that the presence of Y203 resulted in a lowering of total self-energy of rPfRH5_M-BSG complex (-11200 kcal/mol) compared to that of rPfRH5_3D7 -BSG (−11127 kcal/mol) indicating that the former was a more stable complex. This was validated by estimation of interaction energy of individual residues of rPfRH5_3D7 and rPfRH5_M in the energy minimized complexes of rPfRH5_M/rPfRH5_3D7 with BSG (−10.9 kcal/mol for PfRH5_3D7 -BSG and -29.9 kcal/mol for PfRH5_M –BSG) (Fig. 3c, d). Residue-wise interaction energies varied not only at amino acid position, 203, but in neighbouring positions as well. This signified different conformational arrangement of rPfRH5_3D7 and rPfRH5_M at the time of interaction with BSG (Fig. 3c, d).

DNA sequence representing the receptor binding region of PfRH5 which spanned amino acids 141 to 756 of the protein and harboured Y203 (rPfRH5_M) allele was PCR amplified from genomic DNA isolated from one P. falciparum infected blood sample. The PCR product was cloned downstream to T7 promoter in the E. coli expression vector pET-20b (+) in the correct reading frame. C203 (rPfRH5_3D7) allele was obtained by site-directed mutagenesis using the Y203 containing clone. The chimeric plasmids were mentioned as rPfRH5_3D7- pET-20b and rPfRH5_M- pET-20b. Recombinant proteins (rPfRH5_3D7 and rPfRH5_M) of expected molecular mass (24.5 kDa) detected in cell lysates were purified by one-step affinity chromatography on Ni–NTA resin (Fig. 4a). Fractions eluted with 500 mM imidazole were separated by SDS-PAGE and detected by Coomassie staining and Western blotting using anti-His tagged antibody (Fig. 4b, c). Yields of purified PfRH5 proteins were 1 mg/ml of E. coli culture.

Synchronous fluorescence of rPfRH5_3D7 and rPfRH5_M were measured at a range of wavelength differences (Δλ) (Additional file 4: Fig. S2.). Synchronous fluorescence of both proteins showed peak with maximum intensity at Δλ₇₀. The spectral pattern at Δλ₇₀ did not show any differences between the two variants (Fig. 4d). FTIR spectra of rPfRH5_3D7 and rPfRH5_M also did not differ from each other (Fig. 4e). Variation in folding states of purified rPfRH5 proteins detected as CD data were analysed by SELCON3 programme in Dichroweb (Fig. 4f, g) [45]. CD spectra of rPfRH5_M showed an increase in the proportion beta-sheet (49.6% vs 34.2%) at the expense of alpha-helix (2% vs 10.9%) and random coil (22.7% vs 34.2%) compared to that of rPfRH5_3D7. Secondary structural components of rPfRH5_M differed significantly than that of rPfRH5_3D7 (p = 0.02) which appeared to be a result of C to Y replacement.

To explore the conformation of two ligand-receptor complexes, FTIR, synchronous fluorescence, fluorescence anisotropy UV CD spectroscopy were applied. No significant differences in backbone conformation could be identified when PfRH5_3D7–BSG and rPfRH5_M–BSG complexes were subjected to FTIR spectroscopy (Fig. 5a). However, contrary to its unbound state, rPfRH5_M–BSG exhibited a higher fluorescence peak intensity at Δλ₇₀, indicating that either Y203 change or exposure of aromatic amino acids in the rPfRH5_M–BSG complex was responsible for the observed difference (Fig. 5d). Estimated fluorescence anisotropy indicated a more compact conformation of rPfRH5_M–BSG complex (0.176) compared to rPfRH5_3D7–BSG (0.198) (Fig. 5e). CD spectra of rPfRH5_3D7–BSG and rPfRH5_M–BSG complexes also showed differences in the percentage of beta-sheet and random coil contents of rPfRH5_3D7–BSG (50.6% beta-sheets, 25.3% random coil) in comparison to rPfRH5_M –BSG (43.4% beta-sheet, 33.3% random coil) (Fig. 5b, c).

Finally, isothermal titration calorimetry was employed to compare the thermodynamic aspects of complexes formed between rPfRH5s and BSG. ITC yielded negative heat deflection, indicating the binding between rPfRH5_M and BSG to be an exothermic event. The association constant (K_a) of rPfRH5_M with was found to be 2.77 fold higher (Ka = 3.63E⁶ ± 2.02E⁶ M⁻¹) than that between rPfRH5_3D7 and BSG (Ka = 1.31E⁶ ± 1.21E⁶ M⁻¹). This validated a distinct functional impact of C203Y polymorphism in relation to ligand-receptor interaction (Fig. 5f, g). Gibbs free energy of binding (ΔG) was also lower for rPfRH5_M–BSG (−8944) than rPfRH5_3D7–BSG (− 8340).