Comparative structural and evolutionary analyses predict functional sites in the artemisinin resistance malaria protein K13

Abstract

Numerous mutations in the Plasmodium falciparum Kelch13 (K13) protein confer resistance to artemisinin derivatives, the current front-line antimalarial drugs. K13 is an essential protein that contains BTB and Kelch-repeat propeller (KREP) domains usually found in E3 ubiquitin ligase complexes that target substrate protein(s) for ubiquitin-dependent degradation. K13 is thought to bind substrate proteins, but its functional/interaction sites and the structural alterations associated with artemisinin resistance mutations remain unknown. Here, we screened for the most evolutionarily conserved sites in the protein structure of K13 as indicators of structural and/or functional constraints. We inferred structure-dependent substitution rates at each amino acid site of the highly conserved K13 protein during the evolution of Apicomplexa parasites. We found two solvent-exposed patches of extraordinarily conserved sites likely involved in protein-protein interactions, one in BTB and the other one in KREP. The conserved patch in K13 KREP overlaps with a shallow pocket that displays a differential electrostatic surface potential, relative to neighboring sites, and that is rich in serine and arginine residues. Comparative structural and evolutionary analyses revealed that these properties were also found in the functionally-validated shallow pocket of other KREPs including that of the cancer-related KEAP1 protein. Finally, molecular dynamics simulations carried out on PfK13 R539T and C580Y artemisinin resistance mutant structures revealed some local structural destabilization of KREP but not in its shallow pocket. These findings open new avenues of research on one of the most enigmatic malaria proteins with the utmost clinical importance.

Figure 1

Schematic and structural representation of PfK13 and its putative function as substrate adaptor. (a) PfK13 domain annotation. Three domains of PfK13 are annotated in databases: coiled-coil-containing (CCC), BTB, and Kelch-repeat propeller (KREP). The Apicomplexa-specific N-terminal domain is predicted to exhibit a random-coil conformation. The crystal structure of BTB and KREP domains was solved; the CCC domain is expected to form two helices coiling together. (b) Proposed, simplified model of the protein complex containing PfK13, based on PfK13 domain annotation and co-immunoprecipitation experiments^,,,,. The BTB domain of PfK13 is expected to bind a scaffold Cullin protein, while the KREP domain likely binds to the substrate molecule(s) further ubiquitinated and possibly degraded by the proteasome. Importantly, the regions and sites of PfK13 involved in binding are unknown (represented as red ‘?’ symbol in the figure). For ease of representation, PfK13 was shown as a monomer although the crystal structure of PfK13 BTB-KREP was solved as a dimer.

Figure 2

Conservation level of the K13 protein, domains and codon sites. (a) Conservation level of k13 compared to those of all protein-coding genes of Plasmodium. The one-ratio PAML values, calculated for 3,256 orthologous genes among six Plasmodium species (P. falciparum, P. berghei, P. chabaudi, P. vivax, P. yoelii and P. knowlesi), were retrieved from the study of Jeffares and colleagues. Magenta and orange bars indicate the location of k13 and four other KREP protein-coding genes (PlasmoDB accession numbers: PF3D7_1022600, PF3D7_1125800, PF3D7_0724800, PF3D7_1125700) respectively. Vertical dashed lines show the 5% cutoff of the most and less conserved protein-coding genes. Among KREP protein-coding genes, only k13 belonged to the top 5% of the most conserved protein-coding genes of the Plasmodium proteome. The table shown below the histogram provides the rank of each KREP-containing protein-coding sequence in the whole dataset. (b) Conservation level of the k13 codon sites. The scatter plot shows the conservation level of k13 codon sites using the pfk13 sequence as reference for the codon numbering (starting at codon 213). White circles correspond to inter-domain positions. All codon sites were reported to evolve under strong purifying selection, with ω drastically <1. We used the ω estimates obtained under the best fitted PAML model M3 that indicates a variable selective regime among codon sites. (c) Conservation level of the annotated K13 domains. The box-whisker plot shows that BTB evolves under more intense purifying selection compared to either CCC or KREP, using non-parametric Mann-Whitney U test (p < 0.05). Box boundaries represent the first and third quartiles and the length of whiskers correspond to 1.5 times the interquartile range.

Figure 3

Conservation and structure homology of the K13 BTB fold. (a) Linear schematic representation of the BTB fold of some BTB-containing protein families. Yellow arrows and cyan cylinders represent strands and helices, respectively. Stars in magenta correspond to the 10% most conserved K13 BTB amino acid sites (based on the ranking of the λ substitution rates, FuncPatch analysis; Supplementary Dataset S1). Structural elements are labelled. (b) Maximum-likelihood unrooted phylogenetic tree of the BTB core fold using a few reference sequences per BTB-containing protein family. Each BTB-containing protein family forms a monophyletic group, identified with a colored background. K13 BTB is written in red color and clusters with the KCTD protein family of BTB-containing proteins. (c) Patch of slowly evolving amino acid sites in a three-dimensional view of PfK13 BTB. The amino acid sites are labelled using the PfK13 sequence as reference. (d) Superposition of the BTB fold of K13 with that of two members of the KCTD protein family. The most conserved amino acid sites for each protein was based on FuncPatch analysis and are shown in magenta, orange and red for K13, SHKBP1 and KCTD17, respectively. Other amino acid sites are shown in white, green, and cornflower blue for K13, SHKBP1 and KCTD17, respectively. Three common positions were identified in the BTB conserved patches of K13, SHKBP1 and KCTD17: 357 on the B2-B3 loop and 397–398 on the B4-A4 loop.

Figure 4

Structural organization of amino acid conservation across K13 KREP. (a) Structure-based amino acid alignment of the six blade (or kelch) repeats of PfK13 KREP. The x axis shows the position in the structure-based amino acid alignment of the six blades; the left y axis shows the amino acid blade number, followed by the first and last amino acid (AA) positions of the corresponding blade. The strands of PfK13 KREP are colored in grey; the AB and CD loops forming the top face of the KREP are colored in cyan; the BC and DA loops architecting the bottom face of the KREP are colored in yellow and green respectively. A consensus sequence of the blades was defined and is shown below the alignment: strict consensus amino acids are shown in bold capital letters and highly conserved amino acids are shown in standard lowercase. A mapping of the different strands and loops onto the three-dimensional structure of PfK13 KREP is shown as surface representation above the structure-based amino acid alignment: side view (left), bottom view (middle) and top view (right). The outline of the shallow pocket surface was delineated with a black line on the bottom view structure (middle) and the amino acid sites forming the shallow pocket surface were surrounded and written in bold in the structure-based amino acid alignment. (b) Mapping of the site-specific substitution rates λ onto the structure-based amino acid alignment of the six blade repeats of PfK13 KREP. Upper graph: heat map showing the λ substitution rate for each amino acid site. Black boxes correspond to the gaps in the structure-based amino acid alignment. The median of the λ substitution rates was first calculated (white) to produce a scale ranging from the lowest (magenta) to the highest (green) site-specific substitution rate λ. Lower graph: Plot of the mean and 95% confidence interval of the λ substitution rates along the structure-based amino acid alignment of the six blades. The positions including one or more gaps were discarded.

Figure 5

Conservation level and electrostatic potential across the KREP structures of K13 and other BTB-Kelch proteins. (a) Location of the 10% most conserved amino acid sites (magenta) on the three-dimensional structure of PfK13 KREP. The conservation level of positions was defined using the site-specific substitution rates λ estimated with FuncPatch (Supplementary Dataset S1). The KREP structure is shown from the side view as cartoon (left structure) and from the bottom view as surface (right structure). The amino acid sites forming the surface of the shallow pocket and belonging to the 10% most conserved sites are labelled. (b) Electrostatic surface potential of the PfK13 KREP structure, estimated with the APBS method. Electrostatic potential values are in units of kT/e at 298 K, on a scale of −8 kT/e (red) to +8 kT/e (blue). White color indicates a neutral potential. The missing charges were added using the Add Charge function implemented in USCF Chimera. (c) Box plots showing the distribution of root-mean-square fluctuations (RMSFs; Supplementary Dataset S2) for the PfK13 KREP shallow pocket positions (shallow pocket group, green) and the remaining PfK13 KREP positions (other group, white). RMSFs were calculated through a molecular dynamics simulation for a duration of 100 ns. Box boundaries represent the first and third quartiles and the length of whiskers correspond to 1.5 times the interquartile range. The difference between groups was evaluated by non-parametric Mann-Whitney U test. (d) Location of the 10% most conserved amino acid sites and electrostatic potential for KEAP1, KLHL2, KLHL3 and KLHL12 KREP structures. The color code and structure orientation are the same as for PfK13 in panels a and b. For KEAP1, KLHL2 and KLHL3, the key amino acids interacting with their respective protein substrates are labelled. The PDB codes and protein substrates are provided above each KREP structure.

Figure 6

Results of molecular dynamics simulations for WT and ART-R mutant PfK13 KREP structures. Molecular dynamics simulations were carried out on the KREP structure using GROMACS during 100 ns at a temperature of 300 K in an all-atom system. The first ns (0 to 5 ns) correspond to an equilibration phase. (a) Mapping of PfK13 ART-R mutations onto the KREP structure. The positions associated with a validated or a candidate ART-R mutation are colored in red and yellow respectively. Validated and candidate ART-R mutations are defined on the basis of the last WHO status report on ART-R. Amino acids forming the shallow pocket are colored in purple. The mutations studied by molecular dynamics simulations are labelled, including the A578S, which is not associated with ART-R. (b) Root-mean-square fluctuation (RMSF) values of WT (black) and mutant ART-R (red) PfK13 KREP structures. RMSF per position was calculated on the backbone Cα atoms (excluding the first five ns, corresponding to the equilibration phase). (c) Dynamical cross-correlation maps (DCCMs) of WT (bottom right) and mutant ART-R (top left) systems. In DCCMs, positive correlation for a pair of residues (red) implies that the two residues move in the same direction, while negative correlation (blue) indicates that the two residues move in opposite directions. Dashed boxes correspond to the differential movements between the WT and mutant systems. Maps were generated using Bio3D in R. (d) Local impacts of C580Y and R539T ART-R mutations on PfK13 KREP structure. Blade number and location of the shallow pocket are indicated. Inter-atomic distances are expressed in Å.