Mayer et al. 1

Mayer et al.

Structure and interactions of the PfEMP1 ATS domain

Structural analysis of the Plasmodium falciparum Erythrocyte Membrane Protein 1 (PfEMP1) intracellular domain reveals a conserved interaction epitope Christina Mayer, Leanne Slater, Michele C. Erat, Robert Konrat and Ioannis Vakonakis SUPPORTING METHODS Protein production and purification PfEMP1 ATS and ATS subfragments were cloned in a modified pET16 vector (Novagen) that includes a N-terminal His10-tag and a 3C-protease cleavage site. In addition, ATS variant PF08_0141 was obtained in a pET28a vector (C-terminal His6-tag), and ATS variant PFF0845c in a modified pET15 vector (Nterminal His6-tag, TEV cleavage site). KAHRP fragments have been described previously (1) and were cloned in a pGEX-6P-2 vector (GE LifeSciences) that encodes for a N-terminal GST-tag. A PFI1780w fragment spanning residues 80-280 was cloned in the same vector. All constructs were transformed into Escherichia coli strain BL21(DE3). Cells were grown at 37 oC in Luria-Bertani (LB) medium or, for NMR usage, in M9 minimal medium supplemented with 15NH4Cl and 13C6 D-glucose as necessary. Protein production was induced at cellular OD600 of approximately 0.7 by addition of 0.25 mM isopropyl β-D-1thiogalactopyranoside (IPTG), and was allowed to proceed for 5 hours at 37 oC. For protein purification of ATS and ATS subfragments under denaturing conditions the cells were resuspended in lysis buffer (6M Urea, 150 mM NaCl, 20 mM Na2HPO4 pH 7.4). Cells were lysed by repeated cycles of freeze-thawing, and DNA was sheared by brief sonication. Cell lysates were spun at 30,000 g, and lysate supernatants were incubated with TALON metal affinity resin (Clonetech) equilibrated in lysis buffer. TALON beads were washed with excess of lysis buffer, and proteins were eluted in the same buffer supplemented with 500 mM Imidazole. Protein refolding was performed by extensive dialysis against a 100 mM NaCl, 20 mM Tris-Cl pH 7.8, 1 mM DTT, 5 mM EDTA buffer, and was followed by over-night His-tag cleavage at 4 oC using 3C or TEV protease as appropriate. Samples were further purified by ion exchange chromatography over Q Sepharose columns in the same buffer and eluted using a NaCl gradient. Final purification was performed by size exclusion chromatography over Sephadex G-75 columns equilibrated in PBS (150 mM NaCl, 20 mM Na2HPO4 pH 7.4) supplemented with 1 mM DTT, or NMR buffer (50 mM NaCl, 20 mM Na2HPO4 pH 7.0, 2 mM DTT). Proteins were concentrated by centrifugal ultrafiltration. Native production of ATS variant PF08_0141 was performed as described above, but without Urea in the lysis buffer and using the pET28a clone. The structural equivalence of native or denaturing protein production methods was established by 2D 1H-15N HSQC spectra (Fig. S1). ATS variant PFF0845c was always prepared under native conditions. For KAHRP fragments and PFI1780w cells were resuspended in PBS and lysed by sonication. Cell lysates were spun at 20,000 g, and lysate supernatants were incubated with Glutathione-sepharose resin equilibrated in PBS. Resin beads were washed with the same buffer, and proteins were eluted in a 50 mM Tris-Cl pH 7.8, 12 mM reduced Glutathione buffer. The GST-tag was removed by over-night cleavage with 3C-protease, followed by reduced Glutathione removal using a Sephadex G-20 desalting column. The GST was separated from the fragments of interest by passing the samples through Glutathionesepharose resin followed by size exclusion chromatography as described above. Fluorescent labeling of ATS-FL utilized an engineered variant carrying a single amino acid substitution, Q150C, created by the QuikChange method (Stratagene). The variant was purified as described above for the wild-type protein, but lacking DTT in the final size exclusion chromatography step. Cysteine reduction was achieved by addition of tris(2-carboxyethyl)phosphine, followed by overnight incubation of the protein at 4 oC with 10-fold excess fluorescein-5-maleimide (Invitrogen). Unreacted label was subsequently removed by size exclusion chromatography over Sephadex G-75 columns equilibrated in NMR buffer. Meta-structure analysis The meta-structure concept was recently introduced as a novel theoretical framework for protein sequence analysis and provides quantitative parameters, compactness and local secondary structure about protein topology (2). The residue-specific compactness value quantifies the structural complexity of an

1

Mayer et al.


individual residue in the context of the 3D protein fold, with residues deeply buried in the interior of a structure exhibiting large compactness values. The meta-structure derived secondary structure parameter is defined in analogy to the well-established 13Cα chemical shift index (Fig. S6), with positive values for a α-helix and negative values indicating the presence of an extended conformation. Details of the calculation procedure have already been described (2). The availability of quantitative topology information on a per-residue basis also provides the possibility to perform protein homology searches that go beyond the primary sequence level. This aspect was exploited in the sequence-derived protein-protein interaction site analysis. In brief, sequences of about 1750 protein complexes (3500 individual sequences) were extracted from the RCSB database together with the information about the residues being part of the interaction sites. Residues located in interaction sites were defined based on the criteria that the CαCα distances to the nearest neighbor residues of the binding partner protein are below a cutoff of 8Å. The selected sequences were subjected to a meta-structure calculation. The rationale behind the prediction algorithm is the notion that there is a (limited) set of protein interaction motifs with distinct metastructural properties. Despite different protein folds or 3D structures, proteins exploit similar metastructural features to accommodate a specific (protein) binding partner. These meta-structure similarities can be quantified and used as the basis for the binding site prediction. To predict putative interaction sites the strategy was as follows: The sequence of a query protein is aligned with a template sequence from the dataset based on meta-structure similarity. Residues of the query protein aligning with template residues that are part of interaction sites are allocated weights corresponding to the alignment similarity scores. The summation over 3500 individual alignments thus provides a reliable estimation of the protein interaction propensity, which is expressed as the protein interaction index. Details together with a statistical analysis will be given elsewhere (Konrat, in preparation). Briefly, a preliminary statistical survey of the sequencederived interaction prediction results obtained for the selected sequences (in total about 3500) revealed that about 80% of residues at interaction sites have protein interaction scores larger than 100. Protein interaction indexes beyond 1000 are therefore considered to be highly significant.

2

Mayer et al.


SUPPORTING FIGURES Figure S1

Figure S1: NMR 1H-15N HSQC spectra of ATS-FL. A) Spectrum of the protein produced under native conditions. B) Spectrum of the protein produced by denaturing extraction and refolding. Resonances originating from Nε atoms of the six tryptophan sidechains are indicated by arrows. The two spectra are highly similar, although signal intensity in the native produced protein is low due to poor yields. C) Overlay of the tryptophan sidechain resonances between native (red) or denaturing preparation (green). The resonance positions, and hence the chemical environments of tryptophan residues are identical. This, coupled to the wide distribution of tryptophan residues in the ATS-FL sequence (Trp117, Trp264, Trp275, Trp289, Trp310 and Trp391) ensures that the native and refolding methods of ATS-FL preparation are equivalent.

3

Mayer et al.


Figure S2

Figure S2: Computational analysis of the ATS-FL sequence. A) Average residue compactness value (2) of ATS-FL plotted against residue number. Much of the ATS-FL sequence lies between the values calculated for proteins in the RCSB (blue line) and intrinsically unstructured molecules (red line). B) The disorder probability of ATS-FL derived from multiple algorithms is plotted against residue number. There is poor agreement between algorithms as to the extent of disorder, with DisProt (3) (green) predicting an

4

Mayer et al.


almost entirely unstructured coil, MetaPrDOS (4) (purple) predicting a mostly structured protein and RONN (5) (blue) oscillating around the 50 % disorder probability threshold. C) Meta-Structure secondary structure analysis of the ATS-FL sequence (2). Positive values correspond to α-helical elements, while negative values correspond to β-strands or extended coil. The ATS segments subsequently determined to form the structured core are highlighted in light red color in all three plots.

5

Mayer et al.


Figure S3

Figure S3: SAXS spectra and Guinier analysis for ATS-FL and its subfragments. A) X-ray scattering intensity versus angle for ATS-FL (red), ATS-25 (purple) and ATS-Core (green) in PBS buffer at 20 oC. The spectra were derived by merging data from four different protein concentrations, and have been shifted on the intensity axis by one (ATS-25) or two (ATS-FL) log-units for clarity. B) Logarithm of x-ray scattering intensity versus square of angle for ATS-FL, ATS-25 and ATS-Core. In each case a section of the data was suggested by AUTORG (6) as satisfying the Guinier approximation (black solid line). This allow us to estimate the normalized scattering intensity at zero-angle, I(0), and the particle size using the linear relationship between I(0) and molecular weight. For comparison, I(0) of BSA (66 kDa) was 13.41 in this experiment. The calculated monomeric molecular sizes are 46.2 kDa, 24.7 kDa and 9.7 kDa for ATSFL, ATS-25 and ATS-Core, respectively. Calculated radii of gyration (Rg) and quality factors for the fit are also shown for each construct.

6

Mayer et al.


Figure S4

Figure S4: Circular dichroism (CD) spectra and thermal stability. A) Mean residue elipticity (MRE) versus wavelength spectra of ATS-FL (black) and ATS-Core (red) in PBS buffer at 10 oC. Both spectra show evidence of α-helical character (signal at 222 nm), as well as extended coil (205 nm) in ATS-FL. B) MRE at 222 nm against temperature for ATS-FL (black) and ATS-Core under the same conditions. Both constructs show high stability and cooperative transitions during thermal unfolding. C) CD spectra of engineered core constructs from ATS variants. All proteins show a largely α-helical character and (D) unfold cooperatively upon heating. E) Similar CD spectra and (F) thermal stability from PFI1780w. The spectra show evidence of strong α-helical character for this construct, courtesy of the double signal minima at 222 and 208 nm. PFI1780w is cooperatively folded and shows high thermal stability.

7

Mayer et al.


Figure S5

Figure S5: Molecular size determination for ATS-FL. A) The final UV (280 nm) absorbance of ATSFL at the end of an analytical ultracentrifugation equilibrium experiment is plotted against rotor radius. The experiment duration was 36 h at 15,000 rpm, and it was performed in PBS buffer and 4 oC. The fit to an ideal monodisperse model using the program Origin (OriginLab) is shown as solid red line, and the residuals from the fit are plotted above the graph. B) Static light scattering of ATS-FL. The light scattering intensity (blue) and UV absorbance at 280 nm (orange) is plotted against retention time in an analytical S200 size exclusion chromatography column in PBS buffer at 20 oC. Scattering intensity and differential refraction (not plotted) are used to estimate the particle molecular size (magenta, plotted on secondary Yaxis). For reference, the calculated monomeric molecular size for ATS-FL is 46.2 kDa.

8

Mayer et al.


Figure S6

Figure S6: Chemical shift deviations of ATS-FL compared to random coil. A) Per-residue plot of chemical shift deviations for Cα atoms. Stretches of positive deviations correspond to α-helical elements (7); spurious negative deviations occur in Cα atoms preceding Pro residues. B,C) Similar plots for deviations from Hα and carbonyl carbon (C') atoms, respectively. α-helical elements correspond to negative deviations, for Hα, or positive deviations, for C' atoms. The chemical shifts were derived by combining the assignments of ATS-25 with those of complementary N-terminal and C-terminal ATS constructs. The ATS segments that form the structured core are highlighted in light red color in all three plots.

9

Mayer et al.


Figure S7

Figure S7: NMR and SAXS analysis of ATS-25. A) NMR 1H-15N HSQC spectra of ATS-25 at 25 oC. Distinct resonances, corresponding to the structured part of this construct, are circled in red. Tryptophan sidechain resonances are boxed. B) Kratky representation of SAXS data from ATS-25 at 20 oC. The NMR spectra of this construct retain all the dispersed resonances seen in ATS-FL (Fig. 1B) and show a decrease in overlapped resonances, indicating reduced flexibility. This is also supported by the increased curvature of the Kratky plot compared to that of ATS-FL (Fig. 1C).

10

Mayer et al.


Figure S8

11

Mayer et al.


Figure S8: ClustalW2 alignment (8) of ATS variants in this study. The alignment was visualized in Jalview (9). The accession codes and residue numbers are shown. Residues are highlighted based on their degree of conservation; only residues above a 50 % conservation threshold are coloured. The extent of the N-terminal (blue), internal (orange) and C-terminal (green) flexible protein segments are shown as bars above the alignment, inferred from our ATS-FL analysis. The conserved ATS region at the protein centre that contains the predicted interaction epitope is indicated by a purple bar. The span of the structured protein core (red) and the secondary structure elements are also shown. Indicated by star are residues that form the packed hydrophobic core of ATS; almost all are conserved or conservatively substituted in PfEMP1 variants.

12

Mayer et al.


Figure S9

Figure S9: NMR 1H-15N HSQC spectra of additional ATS domains. Shown on the right column are spectra of full-length ATS domains from the indicated PfEMP1 variants. In all cases chemical shift dispersion is poor, which is consistent with largely flexible proteins. The Tryptophan sidechain resonances are boxed. Shown on the left column are similar spectra from the engineered structured core constructs for each variant.

13

Mayer et al.


Figure S10

Figure S10: SAXS analysis of ATS variant PFF0845c. A) X-ray scattering intensity versus angle in PBS buffer at 20 oC. The spectra were derived by merging data from four different protein concentrations. B) Kratky representation of the SAXS data, showing a similar profile to that of ATS-FL (Fig. 1C). C) Logarithm of x-ray scattering intensity versus square of angle. A section of the data was suggested by AUTORG (6) as satisfying the Guinier approximation (black solid line). This allows us to estimate the scattering intensity at zero-angle, I(0), and the particle molecular size. I(0) of BSA (66 kDa) was 73.74 in this experiment. The calculated monomeric molecular size for ATS variant PFF0845c is 39.7 kDa. The radius of gyration (Rg) and fit quality factor are also shown.

14

Mayer et al.


Figure S11

Figure S11: Ensemble fit of the ATS-25 SAXS data, and comparison to NMR calculation. A) An initial random coil ensemble of molecular conformations was optimized to fit the SAXS data (10), shown here as a blue line. The calculated scattering pattern from the final ensemble is overlaid (magenta) on the original data. The final ensembles fit the experimental data well to a resolution of approximately 4 Å. B) The Rg probability distribution from the initial random coil ATS-25 ensemble (blue line), and following optimization against the SAXS data (magenta). Shown in green bars is the Rg probability distribution of the ATS-25 ensemble calculated from NMR restraints.

15

Mayer et al.


Figure S12

Figure S12: NMR analysis of the reported ATS-KAHRP interaction. A-C) Overlay of 1H-15N HSQC spectra of 15N labelled ATS-FL alone at a concentration of 50 μM (red) or in the presence (green) of (A) 50 μM unlabeled KAHRP K1, (B) 300 μM unlabeled KAHRP K2 or (C) 300 μM unlabeled KAHRP K3. All spectra were recorded in NMR buffer and 25 oC. The limited solubility of KAHRP K1 prevented us from exploring higher concentrations of this construct. D,E) Overlay of 1H-15N HSQC spectra of 50 μM 15 N labelled K1A (D) or K2A (E) alone (red) or in the presence of 500 μM ATS-FL (green). We observed no significant resonance perturbations in any of the NMR titrations.

16

Mayer et al.


SUPPORTING TABLES Table S1: Structure quality statistics for ATS-Core Experimental restraints NOE Intraresidue (i-j=0) Sequential (i-j=1) Short range (i-j 0.3 Å 0 o Dihedral angle violations > 5 0 RMSDs from idealized geometry Bonds (Å) 0.0023±0.0003 Angles (deg.) 0.49±0.02 Impropers (deg.) 0.39±0.02 Ramachandran statistics (%) Orderedb Corec Most favored regions 93.1 87.1 Additionally allowed 5.0 9.3 Generously allowed 1.8 2.6 Disallowed regions 0.0 0.9 Structure precisiond Backbone atoms (Å) 0.41±0.08 0.61±0. All heavy atoms (Å) 0.90±0.06 1.01±0.

Structure with lowest rms to average 0.0194 0.48 1.2 0.81 0.57 0.48 0.03 0 0 0.0023 0.48 0.38 Orderedb Corec 93.7 87.3 4.8 8.5 1.6 2.8 0.0 1.4

a) The RDC R-factor was calculated as suggested by Clore and Garrett (11). b) Mobile residues ({1H–}15N NOE < 0.6) were excluded. Included are ATS residues 112-136, 245-248 and 258-291. c) Excluding the cloning artifact and the ATS-Core flexible N-terminus. Included are ATS residues 112-136 and 245-291. d) RMS deviations from the average structure

17

Mayer et al.


REFERENCES 1. Waller, K. L., Cooke, B. M., Nunomura, W., Mohandas, N., and Coppel, R. L. (1999) J Biol Chem 274, 23808-23813 2. Konrat, R. (2009) Cell Mol Life Sci 66, 3625-3639 3. Obradovic, Z., Peng, K., Vucetic, S., Radivojac, P., and Dunker, A. K. (2005) Proteins 61 Suppl 7, 176-182 4. Ishida, T., and Kinoshita, K. (2008) Bioinformatics 24, 1344-1348 5. Yang, Z. R., Thomson, R., McNeil, P., and Esnouf, R. M. (2005) Bioinformatics 21, 3369-3376 6. Petoukhov, M. V., Konarev, P. V., Kikhney, A. G., and Svergun, D. I. (2007) J Appl Crystallogr 40, 223-228 7. Wishart, D. S., and Case, D. A. (2001) Methods Enzymol. 338, 3-34 8. Larkin, M. A., Blackshields, G., Brown, N. P., Chenna, R., McGettigan, P. A., McWilliam, H., Valentin, F., Wallace, I. M., Wilm, A., Lopez, R., Thompson, J. D., Gibson, T. J., and Higgins, D. G. (2007) Bioinformatics 23, 2947-2948 9. Waterhouse, A. M., Procter, J. B., Martin, D. M., Clamp, M., and Barton, G. J. (2009) Bioinformatics 25, 1189-1191 10. Bernado, P., Mylonas, E., Petoukhov, M. V., Blackledge, M., and Svergun, D. I. (2007) J Am Chem Soc 129, 5656-5664 11. Clore, G. M., and Garrett, D. S. (1999) Journal of the American Chemical Society 121, 9008-9012

18