Solution conformation, backbone dynamics and lipid interactions of the intrinsically unstructured malaria surface protein MSP2

Abstract

Merozoite surface protein 2 (MSP2), one of the most abundant proteins on the surface of the merozoite stage of Plasmodium falciparum, is a potential component of a malaria vaccine, having shown some efficacy in a clinical trial in Papua New Guinea. MSP2 is a GPI-anchored protein consisting of conserved N- and C-terminal domains and a variable central region. Previous studies have shown that it is an intrinsically unstructured protein with a high propensity for fibril formation, in which the conserved N-terminal domain has a key role. Secondary structure predictions suggest that MSP2 contains long stretches of random coil with very little alpha-helix or beta-strand. Circular dichroism spectroscopy confirms this prediction under physiological conditions (pH 7.4) and in more acidic solutions (pH 6.2 and 3.4). Pulsed field gradient NMR diffusion measurements showed that MSP2 under physiological conditions has a large effective hydrodynamic radius consistent with an intrinsic pre-molten globule state, as defined by Uversky. This was supported by sedimentation velocity studies in the analytical ultracentrifuge. NMR resonance assignments have been obtained for FC27 MSP2, allowing the residual secondary structure and backbone dynamics to be defined. There is some motional restriction in the conserved C-terminal region in the vicinity of an intramolecular disulfide bond. Two other regions show motional restrictions, both of which display helical structure propensities. One of these helical regions is within the conserved N-terminal domain, which adopts essentially the same conformation in full-length MSP2 as in corresponding peptide fragments. We see no evidence of long-range interactions in the full-length protein. MSP2 associates with lipid micelles, but predominantly through the N-terminal region rather than the C terminus, which is GPI-anchored to the membrane in the parasite.

Figure 1

Domain structure of FC27 and 3D7 MSP2 isoforms based on amino acid sequence analysis (GenBank accession numbers J03828, M28891, M28892, M59765-68).,,

Figure 2

Circular dichroism spectroscopy of MSP2 in aqueous solution. (a,b) Mean residue ellipticity (°cm²dmol⁻¹) plotted as a function of wavelength (nm) for 3D7-6H (panel A) and FC27-6H (panel B) solubilised at an initial concentration of 0.15 mg/mL in buffer at pH 3.4 and t = 0 (dashed line), pH 3.4 and t = 168 h (solid line), pH 6.2 and t = 0 (solid line + circle), pH 6.2 and t = 168 h (solid line + triangle), pH 7.4 (dash-dot-dot line), and pH 7.4 and t = 168 h (long-short-short dashes). (c) Change in ellipticity, Δε(M⁻¹cm⁻¹) plotted as a function of wavelength (nm) for FC27-6H at pH 3.4 and t = 0 (circles) and t = 7 days post incubation at room temperature (diamonds). The nonlinear least squares best fits employing the CDPro software package with the SDP48 reference set and CONTINLL algorithm are shown as solid lines. The t = 0 data yielded a nonlinear best-fit to 7.4% α-helix, 1.6% β-strand, 8.9% turn and 82.1% unordered structure. By contrast, the t = 7 days data resulted in a nonlinear best-fit to 27.9% α-helix, 16.8% β-strand, 20.6% turn and 34.7% unordered structure

Figure 3

Sedimentation velocity analyses of 3D7-6H and FC27-6H MSP2. (a,b) The absorbance at 235 nm is plotted as a function of radius (cm) at 12 min intervals for 3D7-6H at pH 3.4 (panel a) and FC27-6H at pH 3.4 (panel b). The raw absorbance data (symbols) is overlaid with the nonlinear least-squares best-fits (solid lines) to a single non-interacting species (panel a) and a mixture of 4 species (panel b) fitted using the program SEDFIT. All nonlinear fits yielded rmsd values <0.0061 and Runs test Z values < 7.40. Top panels –Residuals for the nonlinear least squares best fits described below plotted as a function of radial position (cm) from the axis of rotation. (c) The absorbance at 235 nm is plotted as a function of radius for freshly prepared FC27-6H (i.e. at t = 0) at pH 3.4 and FC27-6H incubated at room temperature and pH 3.4 for t = 24 h, t = 48 h, t = 72 h and t = 168 h as indicated. The data shown represents the initial absorbance profile as a function of radius from the axis of rotation in the XL-I analytical ultracentrifuge at 40,000 rpm. (d) The average absorbance signal due to monomeric MSP2 plotted as a function of time following incubation at room temperature of 3D7-6H at pH 7.4 (open squares + dashed line), 3D7-6H at pH 6.2 (open triangles + dashed line), 3D7-6H at pH 3.4 (open circles + dashed line), FC27-6H at pH 7.4 (solid squares + solid line), FC27-6H at pH 6.2 (solid triangles + solid line), and FC27-6H at pH 3.4 (solid circles + solid line). Each data point represents the mean absorbance value at 235 nm averaged over radii values of 6.1 to 7.1 cm (as shown for example in panel c for FC27-6H at pH 3.4) enclosed by the standard deviation (error bars).

Figure 4

¹H-¹⁵N HSQC (800 MHz) for FC27 MSP2 in 10 mM HOAc at 25 °C. Peaks are labeled with residue numbers in the native protein; overlapped resonances are labeled with numbers separated by slashes.

Figure 5

Secondary chemical shifts HN (a), H^α (b), C^α (c), and CO (d) along the sequence of FC27 MSP2 in 10 mM HOAc at 25 °C. Dotted lines represent threshold values (the residues with chemical shift deviations beyond threshold values have a propensity to adopt ordered secondary structure). The positions of the disulfide bond and helical regions inferred from the CSI consensus results are shown on the top panel. Random coil chemical shift references are from Schwarzinger et al.; we used these random coil values as our spectra were obtained under acidic conditions. We also undertook a CSI analysis using another set of random coil chemical shifts, which were derived under non-denaturing conditions and only for proton atoms. The result was consistent with the analysis documented in this figure, except that additional residues in the N-terminal region now appeared to be involved in helical structure (not shown).

Figure 6

¹⁵N relaxation parameters of FC27 MSP2 in 10 mM HOAc at 25 °C. (a) R₁, (b) R₂, and (c) steady-state {¹H}-¹⁵N NOE. Relaxation parameters measured at 600 and 800 MHz are shown in filled and open circles, respectively. A schematic representation of helical regions inferred from the CSI consensus plot and the disulfide bond location is shown at the top.

Figure 7

Reduced spectral density functions of FC27 MSP2. (a) J(0), (b) J(ω_N), and (c) J(0.87ω_H) calculated from relaxation parameters measured at 600 MHz (filled circles) and 800 MHz (open circles) using Eqs 3–6. A schematic representation of helical regions inferred from the CSI consensus plot and the disulfide bond location is shown at the top.

Figure 8

Superposition of backbone amide region of ¹H-¹⁵N HSQC spectra (500 MHz) of full-length FC27 MSP2 (red, 2048×256 points) with that of the N-terminal 28-mer peptide MSP2_1-25 (green, 1024×128 points), both in 10 mM HOAc at 25 °C. The peaks labelled with negative numbers were residues from the vector prior to the start of the native sequence. In addition, Lys6 was not observed in the spectrum of full-length protein.

Figure 9

Superposition of ¹H-¹⁵N HSQC spectra (500 MHz) of FC27 MSP2 at pH 4.9 in the absence (green) and presence (red) of 100 mM DPC at 25 °C. (a) Peaks that showed the largest shifts and/or broadening are labelled. The upper panel (b) shows the spectral region containing resonances from side chains of Asn, Gln and other residues, the arrows highlight side chain peaks that appear in the presence of DPC.