The structure of Plasmodium falciparum 3D7_0606800 reveals a bi‐lobed architecture that supports re‐annotation as a Venus Flytrap protein

Abstract

Plasmodium falciparum, the causative agent of malaria, employs a diverse array of surface displayed proteins to promote dissemination and establish infection in the human host. Of these, Pf3D7_0606800 is highly immunogenic and has been designated a potential top 10 candidate for inclusion in a multicomponent malarial vaccine. The role of Pf3D7_0606800 in parasite biology, however, is unknown and its characterization has been complicated by a lack of sequence identity with proteins of known structure or function. Towards elucidating Pf3D7_0606800 function, we determined its structure to a resolution of 2.35 Å using selenium single wavelength anomalous dispersion. A bi‐lobed architecture displays the core structural hallmarks of Venus Flytrap (VFT) proteins prompting us to re‐annotate Pf3D7_0606800 as PfVFT1. Structural analysis further revealed an extended inter‐lobe groove that, when interrogated by molecular docking, appears well suited to bind peptide‐based ligands. Collectively, our structural characterization of the highly antigenic P. falciparum surface protein PfVFT1 provides intriguing functional insight and establishes a structural template that could prove valuable for malaria vaccine engineering studies.

Short abstract

Interactive Figure 1, Interactive Figure 2 | PDB Code(s): 5JKQ

Introduction

Plasmodium falciparum is a member of the phylum Apicomplexa and the etiological agent of severe human malaria.1 The widespread success of P. falciparum is largely due to its complex lifecycle supported by a diverse arsenal of parasite‐specific factors. Of particular interest are surface‐exposed parasitic antigens that may be linked to human virulence.2 Recent genomic and proteomic studies have identified several parasite‐specific antigens whose functions are unknown. Since antibodies to these proteins are found in naturally infected patients and are correlated with survival, they likely play a crucial role in P. falciparum biology.2, 3, 4 Thus, structural and functional characterization of these antigens has the potential to offer valuable insight into the molecular basis of P. falciparum pathogenesis and support the design of next generation malarial vaccines.

Pf3D7_0606800 is a parasite‐specific antigen expressed by the P. falciparum 3D7 strain in both sexual and asexual stages5 and was identified as a major candidate for inclusion in a multicomponent blood‐stage malaria vaccine along with well‐known targets such as AMA1.2, 6, 7, 8 Pf3D7_0606800 is readily accessible to the immune system, may contain heat‐labile conformational epitope(s),9 and is recognized by sera of malaria‐immune adults.6, 10 The function of Pf3D7_0606800, however is unknown. Bioinformatic analysis predicts a signal peptide with no identifiable membrane anchors suggesting that it coordinates a membrane‐associated partner to enable surface localization. Consistent with this hypothesis, fractionation of detergent‐resistance membranes showed that Pf3D7_0606800 clusters with GPI‐anchored parasite proteins including Pf38, Pf92, Pf113, PfMSP‐1, PfMSP‐4, PfASP, and PfRAMA.11

Defining the structural characteristics of Pf3D7_0606800 is a key step towards revealing its role in P. falciparum biology and understanding the molecular basis for its immunogenicity. To this end, we have recombinantly produced Pf3D7_0606800 in insect cells and determined the selenium‐phased 2.35 Å resolution crystal structure. Our study represents the first detailed characterization of this highly immunogenic, yet previously understudied malaria surface antigen and offers intriguing clues as to its biological function.

Results and Discussion

Pf3D7_0606800 adopts a compact bi‐lobed architecture

Toward defining the function of Pf3D7_0606800, we first sought to establish its overall architecture. Mature forms of native and selenomethionine derivatized Pf3D7_0606800 [Fig. 1(A)] were expressed in insect cells and purified by nickel affinity and size exclusion chromatography (SEC), which showed a monomeric distribution [Fig. 1(A)]. Pf3D7_0606800 crystallized in a primitive orthorhombic space group with four molecules in the asymmetric unit (AU). Selenium single wavelength anomalous dispersion (SAD) resulted in high quality phases that enabled modeling of the complete protein sequence from the signal peptide cleavage site through the C terminus for Chain A.

Figure 1.

Structural characterization of Pf3D7_0606800 reveals a bi‐lobed domain architecture. A: Top: Schematic of the Pf3D7_0606800 protein sequence. Bottom: SEC profile of native (blue line) and SeMet (black line) Pf3D7_0606800 produced in insect cells. Inset, Coomassie‐stained SDS‐PAGE gel of Native and SeMet Pf3D7_0606800 (expected molecular weight: 32 kDa). B: Cartoon representation of orthogonal views of Pf3D7_0606800 structure. C: Surface representation of Pf3D7_0606800 highlighting the central cleft (blue) with the hydrophobic (orange dots) and aromatic (yellow dots) residues. D: Orthogonal views of Pf3D7_0606800 electrostatic surface representation. An interactive view is available in the electronic version of the article.

Structural analysis revealed a bi‐lobed architecture with the lobes linked by a pair of extended beta strands [Fig. 1(B)]. Both lobes exhibit similar packing of the secondary structure elements consisting of an α/β globular domain with a central β sheet of 4–5 strands flanked by clusters of α helices. Lobe 1 is formed predominantly by the N‐terminal portion of the protein and incorporates residues Lys30 to Pro191 and His294 to Lys299 making it nearly twice the size of Lobe 2, which extends from Lys198 to Asp287 [Fig. 1(B)]. The antiparallel beta strands that connect the two lobes form a scissor‐like overlap with Lys194–His195 and Thr292–Gly293, computationally predicted12 to be high probability articulation points [Fig. 1(B)]. Analysis of the four molecules in the AU also supports the potential for inter‐lobe flexibility; the overall root mean square deviation (RMSD) range of 1.04–1.27 Å over 223–261 Cαs is greater than for Lobe 1 (0.31 Å over 136 Cαs) or Lobe 2 (0.49 Å over 85 Cαs). Moreover, a structural superimposition of the four chains also revealed that Lobe 2 rotates ∼3°–5° relative to Lobe 1 for a total displacement of ∼4 Å.

The most striking feature of the Pf3D7_0606800 molecular surface is a prominent cleft located between Lobes 1 and 2 [Fig. 1(C)]. Several aromatic residues line the cleft [Fig. 1(C)] indicating the potential for π‐stacking interactions to coordinate a molecular partner. The most solvent‐exposed aromatic residues lining the cleft are Trp (85, 100), Tyr (41, 145), and Phe (132, 146) residues from Lobe 1 [Fig. 1(C), yellow dots]. The observed inter‐lobe flexibility [Fig. 1(B)] may also promote ligand binding by stabilizing the bound form of the complex through induced fit. A second notable surface feature in Lobe 1, located outside the central cleft, is a large acidic patch comprised of ten Asp and Glu residues (Glu44, Glu49, Glu50, Asp53, Asp56, Asp123, Glu130, Asp139, Glu140, Asp141) [Fig. 1(D), right] that might also support molecular interactions to enable surface localization or function.

Structural analysis of Pf3D7_0606800 reveals a Venus Flytrap domain

To overcome the lack of sequence identity with proteins of known structure or function, we next performed a structural homology search using the DALI server.13 The DALI search resulted in a significant hit with the Venus Flytrap (VFT)‐like ATP Binding Cassette (ABC) transporter from Streptococcus thermophilus (PDB ID 3HV1). A similar result was obtained when the search was performed with individual Pf3D7_0606800 lobes consistent with an overall conserved structural fold. Despite a sequence identity of only 14%, the two proteins align with a Z‐score of 11.3 (with scores less than 2 being spurious) and an RMSD of 5.0 Å over 254 Cα positions. A list of top six DALI hits is shown in Table 2. In addition to StABC, Pf3D7_0606800 exhibits strong structural similarity to more than 50 VFT domains. Based on this similarity, we propose re‐annotating Pf3D7_0606800 as PfVFT1, which reflects both its identification as a VFT protein and, to the best of our knowledge, as the first structurally validated VFT protein from Plasmodium.

Table 2.

Top 6 DALI Hits of PfVFT1

PDB ID	Z‐score	Function
3HV1	11.3	Polar amino acid ABC uptake transporter substrate
2V25	10.9	Major cell binding factor
5HMT	10.4	Cyclohexadienyl dehydratase
3KBR	10.3	Cyclohexadienyl dehydratase
2VHA	10.2	Periplasmic binding transport protein
3K4U	10.1	Binding component of ABC transporter

While the overall VFT architecture if conserved, PfVFT1 harbors several unique secondary structure elements, particularly in the N‐ and C‐termini [Fig. 2(A)]. Specifically, two helices (α1 and α2) in Lobe 1 extend from the N‐terminus, whereas their equivalents (α6 and α7) in all other VFTs are located at the C‐terminus [Fig. 2(A)]. This rearrangement in PfVFT1 appears to contribute to a more condensed packing of Lobe 1 as a 35‐residue loop connecting the C‐terminus of α2 to the N‐terminus of α3 wraps tightly around α1 and α2. Additionally, a helix located between β6 and β7 in Lobe 2 of most VFTs is absent in PfVFT1 [Fig. 2(A), gray rectangle with yellow border]. This helix, which plays a role in substrate recognition in select VFTs,17, 18 is replaced by a hydrophobic loop in PfVFT1 that forms a part of the hydrophobic cleft and may play a similar role in ligand recognition. The cleft of PfVFT1 is also larger (surface area: 928 Ų; volume: 2004 ų) than that of StABC (surface area: 434 Ų; volume: 404 ų) as shown by surface pocket analysis with CastP19 consistent with the ability to coordinate a larger ligand.

Figure 2.

Pf3D7_0606800 adopts a Venus Flytrap (VFT) domain fold and is re‐annotated as PfVFT1. A: Secondary structure elements that are conserved (gray) and that diverge (purple, green, and blue) between PfVFT1 and StABC (PDB ID 3HV1). Topology diagram of PfVFT1 highlighting unique N‐terminus (blue components prior to β1, deep purple) and C‐terminus (green). Starburst and dotted connector indicate a disulfide bond. Gray dotted indicates lobe 2 helix (yellow outline) that is common to most VFTs, but absent in PfVFT1. B: A phylogenetic tree assembled using structure‐based (RMSD) alignments of PfVFT1 with two representative members for every cluster, obtained from previous evolutionary analysis on VFTs14 using PDBefold.15, 16 For simplicity, the recently identified Cluster G14 was not included. Escherichia coli (Ec); Homo sapiens (Hs); Yersinia pesits (Ys); Halomonas elongate (He); Ruegeria pomeroyi (Ru); Streptococcus thermophiles (St); Synechocystis spp. (Ss). EcFhuD‐1PSZ; EcFnbp‐2CHU; EcLBP‐1USK; EcLIV‐1Z16; EcMBP‐4MBP; HsMBP‐2ZVK; EcPBP‐1IXH, YsPBP‐2Z22; HeSBP‐2VPO; RuSBP‐3FXB; StABC‐3HV1; SsCmpa‐2I48; SsNrta‐2G29. An interactive view is available in the electronic version of the article.

Since VFT proteins are clustered into several different subgroups and often based on the nature of their ligands,14, 20 we next sought to determine in which subgroup PfVFT1 clusters. However, the sequence diversity of PfVFT1 relative to other VFTs significantly complicates a multiple sequence alignment‐based approach. We thus leveraged the PfVFT1 structure and generated a distance tree based on structural comparisons [Fig. 2(B)]. Though the tree cannot be statistically validated using bootstrap values, the topology of the tree is similar to that previously described for VFTs.14 Based on our analysis, PfVFT1 aligns with cluster F [Fig. 2(B)], which includes VFTs that bind amino acid ligands such as l‐glutamate, l‐histidine, l‐lysine, and l‐ arginine.14

The PfVFT1 cleft displays structural features consistent with peptide coordination

Homologs of PfVFT1 are not observed across all Plasmodium species, but instead predominate in Plasmodium parasites that cause malaria in apes (PgVFT1 and PrVFT1) and birds (PgalVFT1 and PrelVFT1). A sequence comparison of these VFT1 homologs reveals a clear bifurcated evolutionary profile [Fig. 3(A), left]. To better understand the implications of these sequence differences, we mapped them onto the structure of PfVFT1 using ConSurf23 [Fig. 3(A), right]. While the surface generally shows extensive divergence throughout the Plasmodium spp., the hydrophobic cleft is notably well conserved consistent with an important functional role. Additionally, a structure‐based sequence alignments with VFTs and mapping the conserved residues on PfVFT1 using consurf23 revealed no residues that appeared to be conserved in the hydrophobic cleft suggesting that the hydrophobic cleft may be unique to Plasmodium spp.

Figure 3.

Peptide coordination in the inter‐lobe cleft. A: Left: Plasmodium spp. VFT1 polypeptide sequences were aligned using Clustal Omega21 and a phylogenetic tree rooted to the divergent PfVFT1 homolog in T. gondii was generated using MEGA22 with 500 bootstrap replicates (See methods for accession numbers). Right: mapping of conserved (burgundy) and variable (cyan) residues of PfVFT1 homologs onto the PfVFT1 surface using ConSurf23; conservation of cleft region indicated by black border. B: Interactions between the N‐terminal tail of Chain A and the inter‐lobe cleft of Chain B. Inset: residues from Chains B and A are labeled green and yellow, respectively. C: Left: Weblogo of the consensus sequence of the 500 best‐scored tripeptides and respective docking result to PfVFT1. Right: Energy minimized docking pose of the consensus peptide.

Analysis of the conserved cleft amongst the four PfVFT1 monomers in the AU revealed that the N‐terminal tail of Chain A binds in the cleft of the neighboring Chain B. The Chain A N‐terminal tail (²²GVEEVVNNK³⁰) is stabilized through intermolecular hydrogen bonds between E24, E25 of the tail with Y62 and W105 of the cleft [Fig. 3(B)], and a salt bridge between E25 and K103 (cleft) [Fig. 3(B)]. To get an insight into the functional relevance of PfVFT1 cleft region, we compared the electron density of the cleft region with previously studied ligand bound VFTs by structurally overlaying PfVFT1 and amino acid bound VFTs belonging to the cluster F. While the N‐terminal peptide (from Chain B‐ PfVFT1) models into the electron density very well, structural overlays revealed no suitable amino acid modeling into the electron density clearly suggesting that the binding partner for PfVFT1 is not similar to classical VFTs.

With the electron density comparison failing to give any insight into the functional relevance of PfVFT1, we decided to investigate the relevance of the peptide–PfVFT1 bimolecular complex. Towards this, we synthesized the N‐terminal peptide and performed native gel shift binding assays and thermal stability assays with PfVFT1. No definitive interaction was measured, suggesting that the peptide may be too small or have an incorrect consensus sequence that is only able to bind in the artificially high protein concentration of a crystal lattice. To further probe the potential for a peptide ligand to bind in the hydrophobic PfVFT1 cleft, we screened in silico 8000 tripeptides without defining any constrained interactions. The energetically most favored 500 tripeptides bound to the hydrophobic patch of the binding pocket of Lobe 1. A consensus scoring for each of the positions was performed and displayed in Weblogo24 format [Fig. 3(B), left]. The consensus sequence was then docked against PfVFT1 to represent a potential binding mode of an enriched peptide sequence [Fig. 3(B), right]. Notably, the in silico generated consensus sequence is not anchored by specific interactions, indicating that shape complementarity may be a primary driving force for coordination within the PfVFT1 cleft.

Materials and Methods

Cloning, protein production, and purification

A construct encoding the mature form of Pf3D7_0606800 (Val23 to Lys299) with three predicted N‐linked glycosylation sites mutated to NxA was codon optimized for insect cells and synthesized by GenScript. The construct was sub‐cloned into a modified pAcGP67b vector with an N‐terminal His₆ tag and produced in insect cells using previously established protocols.25 Recombinant PfVFT1 was purified by nickel‐affinity and SEC (Superdex 16/600 75) in 20 mM HEPES pH 7.5, 150 mM NaCl. Selenomethionine (SeMet) labeled protein was generated using established protocols.26

Bioinformatics

Phylogenetic analysis for Plasmodium VFTs based on the sequences in PlasmoDB,5 NCBI,27 and Genbank28 was performed in MEGA 6.0.22 A neighbor‐joining tree with 500 bootstrap replicates was constructed and rooted to the PfVFT1 homolog in Toxoplasma gondii (NCBI: XP_002365273.2). Accession numbers: P. falciparum VFT1 (PfVFT1), Pf3D7_0606800 (PlasmoDB); Plasmodium reichenowi VFT1 (PrVFT1), PRCDC_0605400 (PlasmoDB); Plasmodium gaboni VFT1 (PgVFT1), XP_018642701.1 (NCBI); Plasmodium gallinacuem VFT1 (PgalVFT1), CRG93648.1 (Genbank); Plasmodium relictum VFT1 (PrelVFT1), CRH01066.1 (Genbank). Sequences were aligned in Clustal Omega21 and surface mapped using ConSurf.23 Two representative structures of each cluster were pairwise structurally aligned to PfVFT1 with PDBefold15, 16 and the RMSD scores were input to the kitsch program in phylips package29 to generate the distance tree.

Crystallization and data collection

Crystals of selenomethionine‐labeled PfVFT1 were grown in 0.1M Bis–Tris:HCl pH 6.5, 2.0M ammonium sulfate using sitting drops. Crystals were cryoprotected in paratone or saturated lithium sulfate and flash cooled in liquid nitrogen. Diffraction data were collected on beamline 7–1 at the Stanford Synchrotron Radiation Lightsource (SSRL) using a wavelength of 1.127 Å, or an optimized wavelength of 0.9792 Å for the f” Se edge.

Data processing, structure solution, and refinement

Diffraction data were processed to 2.50 Å (PfVFT1‐SeMet SAD) or 2.35 Å (PfVFT1‐SeMet) resolution using Imosflm30 and Aimless.31 The structure of PfVFT1‐SeMet SAD was solved by SeMet SAD with 13 high confidence Se sites identified using ShelxC/D32 and refined using Phenix.AutoSol (Supporting Information Fig. S1).33 High quality phases enabled building and registering of ∼80% of the backbone using Phenix.autobuild.34 The higher resolution PfVFT1‐SeMet structure was solved by molecular replacement using the Se‐phased model in Phaser.35 COOT36 was used for model building and selection of solvent atoms and the model was refined in Phenix.refine.37 Structural validation was performed with Molprobity38 with greater than 98% of residues in the most favored conformations. Data collection and refinement statistics are presented in Table 1. Atomic coordinates and structure factors of PfVFT1 have been deposited in the PDB with accession code 5JKQ.

Table 1.

Data Collection and Refinement Statistics

	PfVFT1‐SeMet SAD	PfVFT1‐SeMet
A. Data collection statistics
Space group	P21212	P21212
a, b, c (Å)	84.62, 201.01, 77.53	84.77, 201.43, 77.56
α, β, γ (deg.)	90, 90, 90	90, 90, 90
Wavelength	0.9792	1.127
Resolution range (Å)	84.62–2.50 (2.59–2.50)	84.77–2.35 (2.42–2.35)
Measured reflections	667,898 (65,916)	240,477 (19,464)
Unique reflections	46,651 (4505)	55,998 (4514)
Redundancy	14.3 (14.6)	4.3 (4.3)
Completeness (%)	100.0 (100.0)	99.8 (99.8)
I/σ(I)	15.8 (5.1)	8.6 (2.3)
R merge	0.162 (0.603)	0.125 (0.546)
B. Refinement statistics
Resolution (Å)		78.18–2.35
R work/R free		0.210/0.263
No. of atoms
Protein (A/B/C/D)		2260/2265/2201/2126
Sulfate		30
Solvent		243
B‐values (Å2)
Protein (A/B/C/D)		25.9/29.6/23.6/26.8
Sulfate		38.6
Solvent		22.8
r.m.s. deviation from ideality
Bond lengths (Å)		0.004
Bond angles (deg.)		0.721
Ramachandran statistics (%)
Most favored		98.1
Allowed		1.9
Disallowed		0.0

Values in parentheses are for the highest resolution shell.

In silico peptide docking

In silico peptide‐mimetic screens were performed using ROCS.39 All 8000 possible tripeptides were generated using OMEGA2 allowing 500 conformers per tripeptide to exhaustively sample rotational space. The library was then docked using FRED and visually inspected with VIDA. A consensus fingerprint was derived from the proximity of each docked amino acid to a defined point in space. Equidistant points along the binding pocket were assigned and residues within 1.5 Å were assigned to that particular position. The final consensus peptide was then docked to the PfVFT1 structure.

Conclusion

Severe human malaria, caused by P. falciparum, remains a major global human health concern. Improved therapeutic strategies are desperately needed as parasites continue to develop resistance to front‐line drug treatments and evade the development of a highly effective vaccine. In this study, we report the first structural characterization of PfVFT1 (previously Pf3D7_0606800), a P. falciparum protein identified as a top candidate for a multicomponent malaria vaccine and a putative factor in human virulence.2, 6 Intriguingly, while the VFT domain is generally well‐characterized, the PfVFT1 structure presents a novel topological organization, thereby expanding our understanding of this important family of ligand‐binding proteins. Structural analysis and in silico studies reveal a possible peptide binding function for PfVFT1 and provide a high‐resolution framework for future epitope mapping studies and functional characterization.

Supporting information

Supporting Information Figure 1.