Structural and regulatory insights into the glideosome-associated connector from Toxoplasma gondii

  1. Amit Kumar
  2. Oscar Vadas
  3. Nicolas Dos Santos Pacheco
  4. Xu Zhang
  5. Kin Chao
  6. Nicolas Darvill
  7. Helena Ø Rasmussen
  8. Yingqi Xu
  9. Gloria Meng-Hsuan Lin
  10. Fisentzos A Stylianou
  11. Jan Skov Pedersen
  12. Sarah L Rouse
  13. Marc L Morgan
  14. Dominique Soldati-Favre  Is a corresponding author
  15. Stephen Matthews  Is a corresponding author
  1. Department of Life Sciences, Imperial College London, United Kingdom
  2. Department of Microbiology and Molecular Medicine, Faculty of Medicine, University of Geneva, Switzerland
  3. Interdisciplinary Nanoscience Center (iNANO) and Department of Chemistry, Aarhus University, Denmark

Abstract

The phylum of Apicomplexa groups intracellular parasites that employ substrate-dependent gliding motility to invade host cells, egress from the infected cells, and cross biological barriers. The glideosome-associated connector (GAC) is a conserved protein essential to this process. GAC facilitates the association of actin filaments with surface transmembrane adhesins and the efficient transmission of the force generated by myosin translocation of actin to the cell surface substrate. Here, we present the crystal structure of Toxoplasma gondii GAC and reveal a unique, supercoiled armadillo repeat region that adopts a closed ring conformation. Characterisation of the solution properties together with membrane and F-actin binding interfaces suggests that GAC adopts several conformations from closed to open and extended. A multi-conformational model for assembly and regulation of GAC within the glideosome is proposed.

Editor's evaluation

The authors describe the first full–length crystal structure and solution conformation of the glideosome–associated connector (GAC) protein from Toxoplasma gondii. The data are convincing and support a model in which GAC uses multiple conformations and lipid–binding surfaces. This study is an important step towards a mechanistic understanding of glideosome assembly and function during the invasion process.

https://doi.org/10.7554/eLife.86049.sa0

Introduction

Cellular migration is an essential process that plays important roles in morphogenetic movements, immune cell trafficking, wound healing, and invasion. Interactions between cells and their environment are essential for the transmission of intracellular forces to the extracellular matrix. In multicellular eukaryotes, cell-cell adhesion ensures tissue integrity while providing footholds for the migration of cell within tissues (De Pascalis and Etienne-Manneville, 2017). Cadherins and integrins are major examples of such adhesive molecules that are coupled to the actin cytoskeleton via intracellular bridging components, such as the catenins, vinculin, and talin (Bachir et al., 2017).

Pathogenic organisms also exploit adherent junctions to facilitate movement with respect to host cells and invasion. Toxoplasma gondii is an obligate intracellular unicellular parasite and a prominent member of the Apicomplexa phylum (Kim and Weiss, 2004), which also includes Plasmodium, the causative agent of human malaria (Su et al., 1995). These parasites share a common set of specialised apical organelles, the micronemes and rhoptries, that are critical for invasion (Carruthers and Sibley, 1999). The sequential secretion of both organelles leads to the formation of a moving junction (MJ) formed between the parasite and host cell plasma membranes that participates in active penetration. Apicomplexan parasites also actively egress from infected host cells and migrate across biological barriers. Substrate-dependent, forward parasite propulsion is known as gliding motility and powered by a multiprotein structure referred to as the glideosome. A myosin motor comprising myosin A (MyoA), a class XIV myosin heavy chain and its myosin light chains, together with glideosome-associated proteins (GAPs) interact with and generate rearward translocation of actin filaments (F-actin) along the parasite (Powell et al., 2018).

The glideosome-associated connector (GAC) protein is a central bridging component of the gliding machinery (Jacot et al., 2016). This large protein composed of numerous armadillo repeats (ARMs) is highly conserved throughout the Apicomplexa phylum and links F-actin to the TRAP/MIC family of surface adhesins at the plasma membrane, which targets host cell ligands and mediates adherent anchor points. A key member of the TRAP/MIC family is the transmembrane micronemal protein 2 (TgMIC2) that supports gliding motility and host cell invasion (Huynh and Carruthers, 2006). The ectodomain of TgMIC2 comprises a von Willebrand factor A domain and multiple thrombospondin repeat domains that associate with an accessory protein (M2AP) and host cells (Huynh et al., 2015; Jewett and Sibley, 2004; Song and Springer, 2014; Tonkin et al., 2010). TgMIC2 is predicted to connect the parasite actomyosin system through an interaction between GAC and its cytoplasmic tail. GAC moves dynamically with the MJ from the parasite apical to the basal pole during gliding motility, host cell egress, and invasion. Rearward translocation of adhesins anchored to both the parasite plasma membrane and the host membrane by the inner-membrane-associated glideosome generates parasite forward movement (Carruthers and Tomley, 2008; Frénal et al., 2017a).

The initial apical location of GAC depends on the activity of an apical lysine methyltransferase (AKMT), through a yet unknown mechanism (Jacot et al., 2016). Recently ultrastructure expansion microscopy localised GAC and formin 1 (FRM1) to the preconoidal rings (PCRs; Dos Santos Pacheco et al., 2022). Importantly, FRM1 is the only and essential nucleator of actin polymerisation to drive conoid extrusion and parasite motility and invasion (Tosetti et al., 2019). As part of the conoid, the PCRs serve as platform for the assembly of the glideosome. Membrane association of GAC relies on its capacity to bind phosphatidic acid (PA), an essential lipid mediator for microneme secretion which assists in the correct engagement of GAC (Bullen et al., 2016). Despite significant efforts, high-resolution structural insight into GAC and its multiple interactions has not been available since its discovery. An initial small-angle X-ray scattering (SAXS) model predicted an elongated club-shaped conformation with the C-terminal Pleckstrin homology (PH) domain lying at the extremity of the structure (Jacot et al., 2016).

Here, we describe combined X-ray crystallography, nuclear magnetic resonance (NMR), SAXS, and hydrogen/deuterium exchange coupled to mass spectrometry (HDX-MS) with course-grained molecular dynamics (CG-MDs) analyses to illuminate the structure and conformations adopted by full-length GAC from T. gondii. Structure validation by biochemical and parasite assays provides insights into membrane and actin binding, ultimately allowing us to propose a model for assembly within the glideosome.

Results

The structure of full-length TgGAC

Well-diffracting and reproducible crystals were obtained at pH 5 for native full-length TgGAC (residues 1–2639; Kumar et al., 2022). Selenomethionine-substituted crystals were also obtained that produced sufficient anomalous signal for phase determination. The structure was solved by multiple-wavelength anomalous dispersion (MAD) to 2.7 Å resolution with an Rfree of 26% (Table 1). The electron density for residues 7–2504 was of sufficient quality to facilitate modelling for these residues (Figure 1 and Figure 1—figure supplement 1A). The remaining electron density showed evidence of peptide backbone but could not be confidently modelled and refined, indicating a degree of conformational flexibility for residues 2505–2639.

Figure 1 with 4 supplements see all
The organisation and structure of TgGAC.

(A) Schematic diagram of the apicomplexan glideosome. Key molecular components and structures are indicated, including subpellicular microtubules (SPMTs), inner membrane complex (IMC), preconoidal rings (PCRs), apical polar ring (APR), Formin-1 (FRM1), filamentous actin (F-actin), glideosome-associated protein 45 (GAP45), myosin motors (MyoA and MyoA), microneme protein 2 (MIC2), and glideosome-associated connector (GAC) protein. (B) Crystal structure of TgGAC7–2504 and representative nuclear magnetic resonance (NMR) model of TgGAC2505–2639 (inset). Helices shown in blue, β-strands in orange and loops in magenta. (C) Full-length TgGAC model comprising the combined crystal structure and NMR-validated AlphaFold2 structures in (B). Region I (R1: residues 1–1665) comprising the first 37 consecutive armadillo (ARM)/HEAT-like repeats (AHRs) is supercoiled into a three-layer pyramid structure (blue). AHR region II comprising 16 ARM repeats (RII: AHR38-53 residues 1670–2489) forms the superhelical arch (green). The C-terminal PH domain encompassing 2511–2639 of RIII is shown in orange (left). Second orientation related by a 180° rotation with even numbered AHRs shown in grey and odd number shown in blue for RI and green for RII. The R1–RII linker and AHR50 which has a helix-loop-helix insertion are shown in magenta. (D) The N/C interface between RI and RII showing key interacting AHRs. (E) Key residues specific interactions across the N/C interface. Cartoon representations coloured in light blue for RI and light green for RII.

Figure 1—source data 1

Maps and coordinates.

wwPDB X-ray structure validation; MTZ file for TgGAC X-ray diffraction data and final refined PDB file for TgGAC from X-ray diffraction.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig1-data1-v2.zip
Table 1
Data collection, phasing, and refinement statistics.
TgGAC
Space groupP 21 21 21
Cell dimensions
a, b, and c (Å)119.078, 123.605, and 221.508
 α, β, and γ (°)90, 90, and 90
PeakEdgeRemote
Wavelength (Å)0.97950.97960.9722
Resolution range (Å)110.75–2.67 (2.72–2.67)110.75–2.67 (2.72–2.67)110.75–2.67 (2.72–2.67)
Total number of reflections904,519750,918798,669
Unique reflections92,92276,60281,143
  Rpim0.036 (0.48)0.032 (0.517)0.031 (0.504)
  II14.00 (0.8)15.8 (0.73)14.70 (0.71)
Completeness (%)100 (100)100 (100)100 (100)
Multiplicity9.7 (9.7)9.7 (10.7)9.8 (10.3)
R-merge0.108 (1.627)0.096 (1.64)0.093 (1.44)
R-meas0.114 (1.698)0.101 (1.72)0.098 (1.528)
CC1/20.99 (0.58)0.999 (0.57)0.99 (0.65)
Refinement statistics
No. of reflections for refinement92,300 (9069)
No. of reflections for Rfree4502 (409)
  Rwork0.2095 (0.4036)
  Rfree0.2683 (0.3990)
  CC (work)0.904 (0.508)
  CC(free)0.910 (0.440)
No. of non-hydrogen atoms18,793
Macromolecules18,679
Solvent114
Ramachandran favoured/allowed/outliers (%)93.4/6.1/0.45
Average B-factor76.88
Macromolecules70.92
Solvent70.07
Rms bond lengths (Å)0.009
Rms bond angles (°)1.47
Clashscore17.98
  1. Values in parentheses are for highest-resolution shell.

  2. Data from a single crystal were used to solve the structure.

To gain insight into the structure of this C-terminal region, we initiated a solution NMR spectroscopy approach with a construct encompassing residues 2505–2639 (TgGAC2505–2639). NMR spectra confirmed an independently folded domain, and over 80% of the backbone resonances could be confidently assigned (Figure 1—figure supplement 1B). Several amide peaks were broadened or absent from the spectra due to conformational exchange in some loop regions. Using available nuclear Overhauser effect spectroscopy (NOESY) and assignment spectra an automated NMR structure calculation was performed using CYANA (Güntert and Buchner, 2015) implemented within the ARTINA approach (Klukowski et al., 2022). The final NMR ensemble was calculated based on 3533 nuclear Overhauser effect and superposes with an average pairwise backbone root mean square deviation (RMSD) of 0.17 Å (Figure 1—figure supplement 1C). Structure statistics are shown in Table 2. We also independently generated a structural model for TgGAC2505–2639 using AlphaFold2 (Jumper et al., 2021; Varadi et al., 2022). Excellent agreement is observed between the experimental NMR structure and predicted from the AlphaFold2 structure with a backbone RMSD of 1.5 Å (Figure 1—figure supplement 1D). The precision of the NMR ensemble is lower between residues 2548 and 2560 due to missing resonance assignments within this region.

Table 2
Nuclear magnetic resonance (NMR) structure calculation statistics.
20 lowest energy structuresTgGACPH
Assigned nuclear Overhauser effect (NOE) peaks
Intra-residue1648
Medium range (|i – j|)≤4640
Long range (|i – j|)>41260
Total3548
Average CYANA target function value12.4
Ramachandran plot
% in most favoured positions79.7±0.7
% in allowed regions20.2±0.4
% in generously allowed0.2±0.1
% in disallowed regions0.0±0.0
Atomic coordinates
Pairwise backbone RMSD secondary structures (Å)0.17±0.01

The structure of TgGAC2505–2639 adopts a PH-like domain fold comprising a seven-stranded β-barrel with three α-helices. The presence of an extended N-terminal-helix is reminiscent of the PH domains from TgAPH and TgISP (Darvill et al., 2018; Tonkin et al., 2014) and appears to be a common feature among apicomplexan PH domains. To provide an illustrative model for full-length TgGAC, the validated structure of TgGAC2505–2639 (TgGACPH) was positioned in the available electron density map for N-terminus of this region and a linker modelled with MODELLER (Fiser et al., 2000).

The crystal structure of TgGAC7–2504 comprises 169 helices (Figure 1 and Figure 1—figure supplement 2) arranged in 53 consecutive helical bundles that resemble ARM and HEAT-like repeats (AHRs; Kippert and Gerloff, 2009) linked to the mixed α/β C-terminal PH domain (Figure 1A). The architecture can be divided into three major regions (RI, RII, and RIII). RI (residues 1–1665) comprises 37 consecutive AHRs of approximately 40 residues long that are supercoiled into a three-layer pyramid structure (Figure 2A). The final ARM of region I (AHR 37) is extended with a six amino acid linker (L1) to start a second large AHR region (RII). RII comprises 16 canonical armadillo repeats (AHR38-53) and forms a superhelical arch that contacts the base of the RI pyramid helical regions. This interface encompasses a surface area of 1550 Å2, with AHR1 and AHR17-18 from RI forming one side while AHR48-50 from RII form the other (Figure 1C). The interface comprises several prominent electrostatic interactions (K2254-D11, E2436-H24, D2330-R629, D2346-K672, and D2197-R716; Figure 1D). AHR50 also has a large helix-loop-helix insertion (residues 2278–2337) between the second and third helices which creates a prominent protrusion that also stabilises the interface (Figure 1B and C). A basic AHR53 completes the region, (II) and an ordered linker from 2489 to 2510 (L2) extends to the adjacent C-terminal PH domain which comprises region III (Figure 1A and B). Residues forming these key features of the structure are highly conserved across GAC orthologues in Apicomplexa (Figure 1—figure supplement 3).

Figure 2 with 1 supplement see all
Glideosome-associated connector (GAC) adopts both open and closed structures.

(A) Superimposition of the experimentally determined composite structure of TgGAC with those generated from AlphaFold2 for TgGAC and PfGAC, showing that the closed structure is conserved (left). N/C-terminal interface is highly similar but partially opened in the predicted structures of TgGAC and PfGAC. The PH domain is not displayed for clarity (right). (B) Conservation profile on a surface representation for TgGAC in three different orientation generated using ConSurf (Landau et al., 2005). Centre – full-length TgGAC in the same conformation as (A). TgGAC is split in two and each half rotated 90° to display the N/C interface. C-terminal regions RII and RIII (left) and N-terminal region RI (right). The surface is coloured according to its ConSurf conservation score, which varies from cyan which are highly variable to maroon for highly conserved. (C) Small-angle X-ray scattering (SAXS) data for GAC at pH 4.0–8.0. (D) Pair distance distribution functions (p(r)) for GAC SAXS data at pH 8.0, calculated from full-length closed TgGAC structure and calculated for the opened structure generated by molecular dynamic (MD) simulations. (E) Modelling of TgGAC solution structure at pH 8.0, calculated theoretical scattering curves for the full-length closed TgGAC structure (red), and calculated for the opened full-length TgGAC structure (blue), and a GASBOR ab initio model (orange). Structures are shown in colours corresponding to those of the scattering curves.

Figure 2—source data 1

Raw small-angle X-ray scattering (SAXS) data for TgGAC.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig2-data1-v2.zip
Figure 2—source data 2

Movie of steered molecular dynamic (MD) trajectory generating an extend TgGAC conformation.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig2-data2-v2.zip
Figure 2—source data 3

Final extended full-length TgGAC after steered molecular dynamic (MD).

https://cdn.elifesciences.org/articles/86049/elife-86049-fig2-data3-v2.zip

GAC adopts multiple extended conformations in solution

The closed conformation observed in the TgGAC crystal structure deviates significantly from the extended club-shaped structure that was proposed from previous SAXS analyses (Jacot et al., 2016), which raises concern that it could be an artefact of crystallisation. To shed light on this, we examined AlphaFold2 structures predicted for full-length TgGAC and PfGAC (Jumper et al., 2021; Varadi et al., 2022). Remarkably, the overall architecture of the predicted structures is very similar to that observed in the crystal structure of TgGAC (Figure 2A), as it is characterised by a series of tandem AHRs that are supercoiled into a large ring and is closed by specific interactions between N- and C-terminal AHRs. While the interface in the predicted TgGAC and PfGAC is partially separated (Figure 2A), the interacting regions overlap with those identified experimentally for TgGAC. The N-/C-terminal interface that characterises the closed structure is conserved at the amino acid sequence level (Figure 2B and Figure 1—figure supplement 3).

We next performed a SAXS analysis over a range of solution conditions to determine whether an open structure is indeed present in solution. GAC was measured at a concentration of 1.3 mg/mL as a function of pH from 4.0 to 8.0. We chose this concentration as a compromise between having the lowest concentration to avoid intermolecular interactions leading to aggregation whilst maintaining good signal to noise. The largest difference is observed as an increase in intensity at low scattering vector moduli, q (Figure 2B), where the intensity is proportional to the mass of the protein/complexes, suggesting that oligomerisation/aggregation is occurring and is most pronounced at pH 5.0. This is consistent with SEC profiles at pH 5 and 8 showing significantly earlier elution of TgGAC at pH 5 (Figure 2—figure supplement 1A). Furthermore, the addition of salt has no effect of the elution volume at both pHs.

To explore the structure of TgGAC in solution at pH 8.0, the mass and size were determined both by a Guinier fit (Figure 2—figure supplement 1B) and using an indirect Fourier transform (IFT) routine for calculating the pair distance distribution function, p(r) (Figure 2C). Note that due to the oligomerisation/aggregation at low pH, the overall size of the complex is too large to be resolved by our instrument setup, as the slope of the intensity at low q is too large. When the overall size cannot be established, the mass and p(r) functions cannot be reliably calculated. At pH 8.0, masses of 264 kDa (Guiner) and 286 kDa (IFT) were obtained, which are close to the expected value of 286 kDa for a TgGAC monomer. The radius of gyration, Rg, values of, respectively, 105 Å and 122 Å were obtained, which are much larger than the Rg of 48 Å calculated from the crystal structure of TgGAC. This discrepancy in size is also clearly seen when comparing the p(r) function calculated from the SAXS data and the crystal structure, where the maximum intramolecular distances are 394 Å and 138 Å, respectively (Figure 2C). The theoretical scattering curve can be calculated from the coordinates of the full-length TgGAC structure and compared to the measured data (Figure 2D). It shows that at medium to large q, the structures are relatively similar, indicating that they contain similar structural components on shorter length scales, but at medium to low q, there is a large difference showing that the overall size and shape are very different. To explore this further, an ab initio model was build using the ATSAS program GASBOR (Svergun et al., 2001), which is highly extended and fits the low q part of the experimental data (Figure 2D). A dimensionless Kratky plot also illustrates that the structure in pH 8.0 solution is much more flexible than would be predicted from the crystal structure, but also it is not completely unfolded (Figure 2—figure supplement 1). To investigate the conformation changes occurring at pH 8.0 compared to the crystal structure in solution, a series of extended models were generated by steered-MD by applying a pulling force between the C- and N-terminal side of the interface, followed by the subsequent separation of the supercoiled pyramid. The model of these that fits best with the data is the most extended, which maintain the domain structures (Figure 2D). The curve fits the data well at high and low q; however, it is significantly above the data in the intermediate range. This means that further loss of domain structure is likely occurring. We did not attempt to pursue this further as it would be too speculative to predict which regions lose their secondary structure. Note that the p(r) function and the Kratky plot of the scattering for the extended structure are shown in Figure 2—figure supplement 1. Overall, the data and the models show that even though TgGAC is primarily a monomer in solution at pH 8.0, it forms a much more extended structure than seen in the crystal structure whilst maintaining key structural elements. Collectively, it can be concluded that a closed structure is likely to be an important stable functional state as the N-/C-interface is encoded within the sequence conservation. The extended, flexible structures observed in solution may play a role is a facilitating the assembly GAC with its binding partners.

Mapping the PA-binding interface for the C-terminal PH domain of GAC

An extended patch of conserved basic residues is formed on one face of the PH domain in GAC which is reminiscent of the charge distribution on TgAPH (Darvill et al., 2018), suggesting a similar role in phospholipid binding. Using phospholipid strip assays, it had been previously demonstrated that both PfGAC and TgGAC bind specifically to PA (Jacot et al., 2016). Building upon our earlier work on the PA-binding protein TgAPH (Darvill et al., 2018), we employed CG-MDs simulations to characterise the binding of TgGAC to PA within a membrane environment. In all three simulation repeats, GAC bound to the membrane within 10 μs (Figure 3—figure supplement 1). Analysis of protein-lipid contacts highlighted two key membrane-bound regions, which localised to the PH domain and an adjacent basic protrusion from AHR53 (Figure 3).

Figure 3 with 1 supplement see all
Simulation of TgGAC binding to phosphatidic acid (PA)-containing membranes.

(A) Analysis of residues that interact with 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphate (POPA) as a percentage of contact during the simulation time. Analysis was performed for the final 5 µs of the simulation. (B) Key residues from mapped onto the TgGAC structure. (C) As in (B) in the contact of the membrane which is displayed as a transparent surface.

Figure 3—source data 1

Molecular dynamics trajectory analysis data for TgGAC.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig3-data1-v2.zip

NMR titrations were subsequently employed to quantify and comprehensively map the PA-binding site experimentally. To circumvent any issues with the missing NMR resonances for TgGACPH, we explored the equivalent region from the GAC homologue from Plasmodium falciparum, which has 53% sequence identity over the PH domain (PfGAC2471–2605) with an improved NMR spectra that enabled over 95% of the backbone residues to be assigned (Figure 4—figure supplement 1). First, 1D NMR binding assays were performed with PA-enriched large unilamellar vesicles (LUVs of ~100 nm diameter) to measure the affinity of the interaction. Titration with LUVs composed of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and POPA (50:50% POPC:POPA) caused a significant loss in PfGACPH NMR signals (Figure 4A), indicating binding to the ‘NMR invisible’ LUVs, whilst almost no loss in signal was observed upon titration with LUVs composed solely of POPC. Binding curves were generated from integration of the NMR signals and (Figure 4B) apparent dissociation constants (Kdapp) calculated (Figure 4C). The Kdapp for PfGACPH binding LUVs composed of 50% POPA was calculated to be 60±3 μM.

Figure 4 with 1 supplement see all
Nuclear magnetic resonance (NMR)-based binding assays for PfGAC2471–2605 (PfGACPH) to phosphatidic acid (PA)-enriched unilamellar liposomes.

(A) 1D 1H-NMR spectrum (9.4–6.5 ppm) of PfGACPH upon titration with increasing concentrations of large unilamellar vesicles (LUVs) composed of POPC:POPA (50:50%) LUV molar ratios: blue, free PfGACPH in solution; red 1:2; green 1:4; purple 1:7; yellow 1:15; orange 1:20; lime 1:25; black 1:30. (B) Binding curves generated from spectral integration and expressed as the fraction of bound protein for variable LUV compositions (POPC [100%] green, POPC:POPS [50:50%] purple, or POPC:POPA [50:50%] red). (C) Apparent dissociation constants (Kdapp) for binding LUVs were calculated from fitting binding curves. (D) Plot of PfGACPH paramagnetic relaxation enhancements (PREs) with PA-enriched MSP1D1H4-5 nanodiscs top (POPC:POPA:PE-DTPA-Gd3+46:40:14%) and bottom (POPC:PE-DTPA-Gd3+86:14%), against sequence number. Dashed lines represent 1 (yellow), 2 (orange), or 3 (red) SD from the mean noise (baseline). (E) PA PREs mapped onto the structure of PfGACPH, residues and coloured if greater than 2 (orange) or 3 (red). (F) Comparison of contact resides from molecular dynamic (MD) and NMR mapped on to the sequence alignment between PfGACPH and TgGACPH. (G) Electrostatic surface representation of PfGACPH revealing an extensive surface patch of positive charge surface charge with key mutated residues labelled. (H) Binding curves of PfGACPH mutants generated from 1D NMR titration with calculated Kdapp. Data are shown as mean of three replicates ±1σ. Apparent dissociation constants (Kdapp) were calculated from fitted binding curves. Data represent mean ±1σ for fitted curves.

Figure 4—source data 1

Raw and processed nuclear magnetic resonance (NMR) data for 1D small unilamellar vescicle (SUV) titration with PfGACPH.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig4-data1-v2.zip
Figure 4—source data 2

Raw and processed nuclear magnetic resonance (NMR) data for paramagnetic relaxation enhancement (PRE) measurements of PfGACPH with nanodiscs.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig4-data2-v2.zip

We next measured paramagnetic relaxation enhancements (PREs) using small PA-enriched MSP1D1ΔH4-5 nanodiscs. PA-enriched MSP1D1ΔH4-5 nanodiscs (40:60%, POPA:POPC) or nanodiscs containing no PA (100% POPC), doped with (paramagnetic nanodiscs) or without (diamagnetic nanodiscs) PEDTPA-Gd3+ paramagnetic lipid, were generated. 2D 1H-15N HSQC spectra were recorded for 15N-labelled PfGACPH in the presence of paramagnetic or diamagnetic nanodiscs. Paramagnetic induced relaxation enhancements of membrane-interacting regions were measured from reductions in signal intensities. Numerous PREs were observed for PfGACPH upon the addition of PA-enriched paramagnetic nanodiscs (Figure 4D - top) but not with 100% POPC nanodiscs (Figure 4D - bottom).

PREs mapped onto the structure of PfGACPH reveal a contiguous surface formed from residues located within the β1-strand, bordering the β1–β2 loop, within the β5–β6 loop and the loop region between β7 strand and C-terminal α-helix (Figure 4E). A comparative analysis of the PH binding residues compared to simulation with full-length TgGAC indicated that the PH domain alone binds with the same interface in the context of the whole GAC protein (Figure 4F). NMR titrations performed with TgGACPH and PA-enriched paramagnetic nanodiscs confirmed an identical PA-binding surface to PfGAC (Figure 4 and Figure 4—figure supplement 1B). Collectively, these results suggest that the solvent exposed positively charge surface in the GAC PH domain represents a PA-specific membrane interacting interface (Figure 4F). Conserved basic residue sidechains are likely to coordinate PA lipid phosphate head groups (Figure 4G and Figure 1—figure supplement 3). A series of alanine substitution mutants were then generated in PfGACPH for residues identified from NMR mapping. These mutants were tested using the 1D NMR binding assay to quantitatively assess the consequence of mutations on PA-binding residues (Figure 4H).

Compared to WT (Kdapp of 270±40 μM), mutation of basic residues reduced the affinity for PA-enriched MSP1E1 nanodiscs, indicating that these residues are important for binding PA within a membrane environment (Figure 4H). The reduction in affinity correlates with position of basic residues relative to the centre of the solvent exposed positive charges, i.e., the largest effect is observed for the RK2553–54AA mutant (Kdapp of 950±50 μM) compared to the KR2549–50AA mutant (Kdapp of 330±20 μM). The same trend is observed for K2491A (Kdapp of 770±100 μM) and K2582A (Kdapp of 710±50 μM), which are more centrally located than K2558A (Kdapp of 550±50 μM).

To establish whether the PA-binding interface identified through NMR analyses on PfGACPH mutants display a similar deficiency in full-length TgGAC, we exploited a liposome binding assay in which bound protein was quantified by sedimentation and sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE). We first tested this assay for full-length TgGAC, TgGACPH, and the known PA sensor TgAPH (Darvill et al., 2018) in the presence of liposomes with no PA and liposomes containing 50% PA. All three proteins were bound specifically to PA-enriched liposomes with most of the protein present in the pellet after ultracentrifugation, whereas in the absence of PA, the proteins were found in the supernatant (Figure 5A). Three mutants were chosen based on the NMR data on PfGAC and TgGAC and were generated in both full-length TgGAC and isolated PH domain constructs. The first mutant focuses on the major positive-charged patch that exhibited the largest effect in NMR assays (namely RK2587–88AA in TgGAC), the second is a triple mutation in the downstream basic region (KER2619–2621AAA in TgGAC) and the third combines both (RK2587–88AA/KER2619–2621AAA in TgGAC). All mutant proteins tested exhibited a significant reduction in binding to PA enriched liposomes (Figure 5B and C), with the most dramatic effects observed for the triple mutant KER2619–2621AAA and paired in quintuple mutant RK2587–88AA/KER2619–2621AAA. Interestingly, the KER2619–2621AAA had the largest effect on binding for the isolated PH domain, suggesting that other regions within full-length GAC may play a role in PA-enriched membrane binding.

Liposome binding assays of phosphatidic acid (PA) binding by glideosome-associated connector (GAC) and PH domain mutants in vitro.

(A) Liposome binding assay with the three proteins of interest. A representative gel stained by Coomassie blue (left). Quantification of the pellet fraction measured by band densitometry (n=2). n/a=no liposome. S=supernatant fraction after ultracentrifugation. P=pellet fraction after centrifugation (right). (B) Representative gel stained by Coomassie blue from liposome binding assays with full-length TgGACFL and the three mutated versions (left). Quantification of the GACFL pellet fraction measured by band densitometry (n=2–4; right). n/a=no liposome. S=supernatant fraction after ultracentrifugation. P=pellet fraction after centrifugation. RK = RK/AA mutations. KER = KER/AAA mutations. RKKER = RK/AA +KER/AAA mutations. (C) Representative gel stained by Coomassie blue from liposome binding assays with TgGACPH and the three mutated versions (left). Quantification of the GACPH pellet fraction measured by band densitometry (n=2–4; right).

Figure 5—source data 1

Full, uncropped SDS-PAGE images for liposome binding experiments.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig5-data1-v2.zip
Figure 5—source data 2

Figure and caption for liposome binding assays using uncropped SDS-PAGE images with key bands labelled.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig5-data2-v2.pdf
Figure 5—source data 3

Liposome binding data tables.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig5-data3-v2.zip

Key lipid-binding residues in the GAC PH domain are not essential for parasite invasion

T. gondii mutant lines were generated in which the PH domain was replaced at the endogenous locus with equivalent sequences containing the PA-binding mutations together with C-terminal Ty tag. Parasite lines were obtained with a wild-type PH domain (TgGACWT -Ty), with the individual targeted mutations (TgGACRK -Ty and TgGACKER -Ty), the double mutation patch (GACRKKER -Ty), and a line in which the PH domain was replaced by the P. falciparum version (TgGACPfPH -Ty). All GAC versions, wild-type, mutated and chimeric, localised correctly at the PCRs of the parasite as well as cytosolically (Figure 6A), suggesting that the PA-binding by the GAC PH domain is not critical for GACs apical localisation. Likewise, upon triggering parasite motility with 5-Benzyl-3-isopropyl-1H-pyrazolo[4,3-d]pyrimidin-7(6H)-one (BIPPO), all the strains displayed the characteristic basal accumulation of TgGAC (Figure 6B). Finally, to access if there was any fitness cost induced by the mutations, wild-type, and mutated parasites were analysed by plaque assay and revealed that the mutations were not detrimental for the parasite lytic cycle (Figure 6C). Accordingly, no clear defect in invasion could be observed in parasites expressing the mutated versions (Figure 6D). A small reduction of the plaque size and reduction in invasion capacity was observed for the TgGAC-PfPH chimera.

Phenotypical analysis of parasite bearing phosphatidic acid (PA)-binding mutations of glideosome-associated connector (GAC).

(A) GAC localisation by Immunofluorescence assay (IFA) in intracellular parasites. White arrow = apical pole. Scale bar = 2 µm. (B) GAC localisation by IFA in extracellular parasites. White arrow = apical pole. White star = basal pole. Scale bar = 2 µm. (C) Plaque assay of the different mutants analysed. (D) Red/green invasion assay (n=3). A Δku80 background T. gondii strain is used, which is deficient in nonhomologous end joining and provides highly efficient gene replacement.

Evaluation of GAC and GAC fragments binding to Toxoplasma F-actin

Association to rabbit F-actin was previously shown to involve the N-terminal 1114 amino acids of TgGAC, roughly corresponding to the first 25 AHRs of the RI region (Figure 7A and B). To map further the interaction of TgGAC with the shorter and highly dynamic actin filaments characteristic of T. gondii (Skillman et al., 2013), a series of recombinant TgGAC fragments were purified and tested for binding (Figure 7C). Both full-length TgGAC and a fragment encompassing residues GAC1–1114 co-sedimented with Toxoplasma F-actin produced from recombinant TgACT (Figure 7D and E). In contrast, a shorter N-terminal fragment, (TgGAC1–619) corresponding to the first 14 AHRs, failed to interact with F-actin. This suggests that either the binding site for F-actin lies between residues 619 and 1114 of TgGAC or that that isolated N-terminal fragment TgGAC1–619 adopts a different conformation than TgGAC1–1114 and TgGAC which prevents its interaction with F-actin.

Figure 7 with 1 supplement see all
TgGAC interaction with TgActin.

(A) Cartoon representation of the structure of TgGAC coloured according to schematic representation shown in (B). (C) Schematic representation of TgGAC structure coloured according to panel (B) with illustration of the truncated recombinant TgGAC constructs. (D) Coomassie-stained gel analysing proteins remaining in supernatant or sedimenting upon centrifugation at 100,000 g. S: Supernatant. P: Pellet. (E) Graphical representation of the percentage of TgGAC co-sedimenting alone or in the presence of TgActin filaments. Data are mean + SD of three independent experiments. p Value calculated using unpaired t-test. (F) Graph displaying deuteration levels for TgGAC1–619, TgGAC1–1114, and TgGAC-FL. Each dot represents a single peptide where the deuteration level in percentage maximal deuteration is plotted according to the residue number at the centre of the peptide (peptide centroid). (G) Comparison of H/D exchange levels for peptides spanning glideosome-associated connector (GAC) residues 1–619 between (i) TgGAC 1–619 and TgGAC1–1114 (blue dots) and between (ii) TgGAC-FL and TgGAC1–1114 (green). Each dot represents a peptide with the x-axis indicating peptide centroid as in (F). The y-axis indicates the sum of differences in number of deuterons incorporated for each peptide, measured at three deuteration times: 3, 30, and 300 s. Region showing significant differences between TgGAC1–619 and TgGAC1–1114 is indicated in red. (H) Schematic representation of the three GAC constructs studied by hydrogen/deuterium exchange coupled to mass spectrometry (HDX-MS), focussing on the region coloured in light blue (residues 1–619). Red colouring indicates the region where a difference in H/D exchange rate is observed. (I) Structure of TgGAC with fragment 1–619 shown as surface representation and highlighted in light blue. Residues 620–1114 are shown in blue, and the remaining of GAC structure is shown in light white. Region displaying a different conformation in TgGAC1–619 compared to other TgGAC constructs, with increased H/D exchange rate, is coloured in red.

Figure 7—source data 1

Full, uncropped SDS-PAGE images for F-actin binding experiments.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig7-data1-v2.zip
Figure 7—source data 2

Figure and caption for uncropped gels of F-actin binding experiments.

https://cdn.elifesciences.org/articles/86049/elife-86049-fig7-data2-v2.pdf

Investigation of GAC fragment conformations using HDX-MS

To gain a deeper understanding into the conformational requirements of the binding of GAC to F-actin, we used HDX-MS. HDX-MS is a powerful method that examines protein dynamics by monitoring the exchange rate of protein amide hydrogens with the solvent (James et al., 2022). Information on protein secondary structure and conformation is provided by measuring overall deuteration levels, and the comparison of H/D exchange rates between protein constructs reveals differences in protein dynamics and conformation. To evaluate whether the 1–619 region adopts the same conformation in TgGAC1–619, TgGAC1–1114, and TgGAC-FL, HDX-MS analyses of peptides encompassing residues 1–619 were compared for all three constructs (Figure 7F–I and Figure 7—figure supplement 1; Supplementary file 1). Most of the GAC N-terminus exhibited identical dynamics in all constructs (Figure 7F and G), and only the C-terminal extremity of TgGAC1–619 (residues 559–609) displayed higher H/D exchange rate compared to TgGAC1–1114 and TgGAC-FL (Figure 7F–I). TgGAC1–1114 and TgGAC-FL share identical dynamics within this region. This indicates that the two truncated fragments share a similar conformation as TgGAC-FL for residues 1–558, corresponding to a major portion (14 AHRs) of the first turn in the superhelical pyramid structure (19 AHRs). NMR data (Figure 7—figure supplement 1C) also confirms that this fragment is predominantly well folded. The C-terminal extremity of the shortest fragment, TgGAC1–619, is significantly more flexible which also may contribute to its inability to bind F-actin.

Discussion

Apicomplexan parasites, for which T. gondii is a model organism, propel themselves by a specialised actomyosin-dependent gliding motility that relies on a large, conserved protein to connect actin filaments with parasite plasma membrane. GAC, an abundant protein that crucially localises to the PCRs that belong to the conoid complex at the apical tip of parasite, was shown to adopt multiple conformations to regulate its function (Figure 1A). The most striking feature of the TgGAC structure is the large continuous supercoiled ARM region (residues 1–2490) which forms a ring. The closure of the ring results from an extensive interface between the ARM1 and ARMs14-16 from RI and ARMs46-48 from RII (N/C-interface - Figure 2). This closed conformation was unexpected as an earlier SAXS model suggested an extended, club-shaped molecule in solution with no evidence for a N-/C-terminal interface (Jacot et al., 2016). Despite this discrepancy, the interface formed in the closed structure is highly conserved at the protein sequence level suggesting that it is a functionally relevant state. Subsequent pH-dependent SAXS analyses reported in the present study revealed that an open structure exists at high pH, but oligomerised/aggregrated forms predominate at low pH. The nature of the N/C interface in the closed structure is hydrophilic with several electrostatic interactions, including some that are readily titratable, such H24 and H2164. It is conceivable that deprotonation of these residues at high pH removes key salt bridges that stabilise the closed conformation and facilitate a conformationally labile open structure. Our SAXS data at low pH may provide some support for this notion, as it reveals a multimerised/aggregated state. While this is unlikely to reflect oligomerisation in vivo, it could result from the reformation of these electrostatic interactions in an intermolecular manner. The ability of GAC to adopt both a flexible open conformation and transition to a closed ring structure could play a role in regulating the multiple interactions of GAC assembly on the glideosome and to the parasite membrane. This behaviour is reminiscent of the unrelated ARM-rich protein, HUWE1, which is a quality-control E3 ligase, which forms a large, closed ring from 34 tandem ARM repeats. The closed structure is believed to represent a latent state but can adopt a range of open conformations to regulate its interactions with accessory proteins (Grabarczyk et al., 2021).

GAC interacts with PA-enriched membranes that are generated through a signalling cascade in activated parasites (Bullen et al., 2016). The C-terminal PH domain within GAC presents a conserved (Figure 5 and Figure 1—figure supplement 3), positively charged surface, and while this is important for PA binding in vitro, its mutation does not cause a significant defect in GAC localisation, translocation, or parasite fitness. Other regions in GAC, such as the basic protrusion in AHR53 (Figure 3 and Figure 1—figure supplement 3), are also likely to contribute to membrane binding in vivo. Furthermore, the presence of a membrane interaction surface within the TgGACPH suggests that the binding site for the juxtamembrane region of the MIC2 cytoplasmic tail is located nearby, and these interactions may also be cooperative. Even though the PH domain represents a minor portion of the full-length GAC structure, its interaction with RII region seems crucial for GAC structural integrity, as a construct lacking the PH domain cannot fold properly.

The principal Toxoplasma F-actin binding region is localised to the first 1114 residues of TgGAC (TgGAC1–1114), which forms a large, supercoiled base of the N-terminal pyramid (TgGAC-R1:1–1656 – Figure 2) and provides a platform with several potential sites of interaction with a helical actin filament. A shorter fragment encompassing only the first turn of the supercoiled pyramid (TgGAC1–619) does not interact with F-actin. Analysis of the protein conformation by HDX-MS showed that TgGAC1–619 adopts a similar conformation for residues 1–558 as in full-length TgGAC, suggesting that the minimal actin-binding interface lies between residues 559–1114 (AHRs 15–25) in which the second supercoiled turn starts. The absence of F-actin binding for the N-terminal portion (1–619), which forms the base of GAC pyramidal structure, suggests a role in this region in stabilising the closed conformation as it makes direct contact with the C-terminal end of the RII region. Membrane association simulations for TgGAC reveal a specific membrane-binding surface involving the PH domain and RII, and this orientation places the TgGAC1–1114 region distal for interaction with the actin filament.

As the available pellicular space between the parasite inner membrane complex and the plasma membrane is insufficient for the open structure of GAC (i.e. with RII and RIII extended) to bridge F-actin to the plasma membrane lengthways, the closed structure is likely to represent a functional important state of cytosolic GAC or during GACs initial engagement within the glideosome. Other actin-membrane bridging proteins, like the ERM (ezrin, moesin, and radixin) family of proteins, crosslink cortical actin to plasma membrane, and full engagement is achieved by a reorganisation of actin-binding regions by cooperative interactions with phosphatidylinositol 4,5-bisphosphate (PIP2; Ben-Aissa et al., 2012). It is therefore conceivable that PA binding and MIC2 recruitment at the plasma membrane by GAC contribute to conformational changes to GAC structure, i.e., a transition from closed to open and extended structures.

The function of GAC in bridging parasite cytoskeleton to the host cell substrate is reminiscent of that for the mammalian catenins within adherens junctions, which comprise several components. The C-terminal AHR arch region (AHRs 38–53) displays significant structural similarity with the armadillo repeat region (ARM) of the ß-catenins, superimposing with an RMSD of 4.2 Å over the backbone of 348 equivalent residues (Figure 1—figure supplement 4). Nominally, α-catenin crosslinks F-actin to β-catenin, while β-catenin establishes the connection to the E-cadherin tails. The single chain of GAC carries out both these roles, i.e., RI acts like the actin-binding α-catenin and is linked via L1 to RII ARM arch structure, which like β-catenin tethers the system to the cell surface adhesins. The ß-catenins bind extended cadherin tails emerging from the plasma membrane via a superhelical surface formed by 12 tandem ARM repeats (Choi et al., 2009; Huber and Weis, 2001; Ishiyama et al., 2010). As TgGAC likely binds the cytoplasmic C-terminal tails of the plasma membrane adhesin TgMIC2 (Jacot et al., 2016), it is tempting to speculate that a similar interface formed in ß-catenin/E-cadherin complexes is used by TgGAC to link TgMIC2 to the glideosome machinery (Figure 1—figure supplement 4).

Intriguingly, the extensive interface within and between the catenins and E-cadherin facilitates molecular transduction via mechanical strain (Angulo-Urarte et al., 2020; Bush et al., 2019; Valenta et al., 2012). Mechanical force induces a rearrangement of binding interfaces that result in a strengthening of its interaction with F-actin, and this effect is dependent on the direction of applied force (Mei et al., 2020; Xu et al., 2020). Such catch bond behaviour may also be relevant for GAC function. Importantly, part of the 559–1114 F-actin binding region is not fully accessible in the closed conformation; therefore, it is also conceivable that optimal F-actin binding requires significant structural rearrangement of the closed structure. The stabilised closed conformation would be able to both resist and sense the significant inter parasite–substrate forces generated by MyoA translocation of F-actin. Such force acting along parasite–host cell interface could induce opening of the GAC structure and unveil additional cryptic binding sites that strengthen F-actin binding and ensure a coordinated direction of motion. The lower affinity of full-length TgGAC compared to TgGAC1–1114 for F-actin supports this hypothesis (Figure 7E).

Structural similarity also exists between the C-terminal ARM region of GAC (TgGAC1670–2639) and the family of myosin-specific chaperones which possess ARM-rich UCS (UNC-45/Cro1/She4) domains (Hellerschmied and Clausen, 2014) that interact with myosin motors domains (Figure 1—figure supplement 4). While T. gondii possesses a dedicated UCS chaperone for TgMyoA (TgUNC; Bookwalter et al., 2014), which is critical for successful folding of all the parasite myosin motors (Frénal et al., 2017b), the similarity between the GAC C-terminal ARM region and the UCS chaperones is intriguing. Myosin chaperones also have the propensity to multimerise and, in some cases, form chains that assist myosin assembly on the filament (Gazda et al., 2013). Although no direct evidence for an interaction between TgGAC and TgMyoA has been found, it is tempting to speculate that TgGAC may assist TgMyoA organisation on F-actin.

Materials and methods

Protein expression and purification

Request a detailed protocol

Full-length TgGAC gene with TEV cleavable N-terminal 6xHis-tag has been cloned into the pET28a vector as previously described (Jacot et al., 2016; Kumar et al., 2022). Constructs for PfGACPH (PfGAC2471–2605) and TgGACPH (TgGAC2505–2639) were constructed with a 6His purification tag and an additional SUMO tag for soluble expression of PfGACPH. A Q5 site-directed mutagenesis kit (NEB) and the manufacturers protocol were used to generate and PfGACPH mutants with standard primers. For protein expression, plasmids were transformed into BL21 (DE3; NEB) or Rosetta2 (Novagen) Escherichia coli strains. Expression was carried in minimal medium supplemented with 15NH4Cl and/or 13C-glucose for NMR isotopic labelling. Purification for His-tagged constructs was carried by Ni2+ affinity chromatography. Removal of the SUMO tag for PfGACPH samples was carried out by incubation with purified ULP1 protease. Further purification for all constructs was achieved by size-exclusion chromatography.

NMR spectroscopy

Request a detailed protocol

Samples of purified 15N/13C-PfGACPH and 15N/13C-TgGACPH were prepared and supplemented with D2O. All NMR spectra were acquired at 298 K on Bruker Avance-III DRX 800 and Avance-III 600 spectrometers. Triple resonance HNCA, HNCACB, HNCO, and HN(CO)CA spectra (Sattler et al., 1999) were recorded and analysed to obtain backbone assignments, which was assisted using MARS program (Jung and Zweckstetter, 2004). Chemical shift assignment and analysis were performed using an in-house version of NMRview (Marchant et al., 2008). HBHA(CO)NH, H(CCO)NH, CC(CO)NH, and HCCH-TOCSY spectra were recorded for use in side-chain chemical shift assignment. 15N-NOESY and 13C-NOESY spectra were recorded and used as distance restraints in structural calculation/validation.

For PfGACPH and TgGACPH 1D 1H NMR LUV binding assays performed with LUVs containing an increasing proportion of POPA (POPA Mol% value). To form LUVs, a lipid suspension was sonicated until transparent and then clarified by centrifugation. The supernatant containing LUVs were prepared at an 8 mM total lipid concentration for use in 1D NMR LUV-binding assays. Dynamic light scattering analyses show homogeneous LUV diameters of ~100 nm. For each LUV composition titration series, a separate Kdapp value was calculated for each replicate (n=3), and a mean Kdapp value was calculated. PfGACPH and TgGACPH titration PRE experiments were carried out with increasingly POPA-enriched MSP1D14-5 nanodiscs. MSP1D1H4-5 MSP was expressed and purified by nickel affinity chromatography using a well-established protocol (Ritchie et al., 2009). Peak elution fractions containing pure nanodiscs (assessed by SDS-PAGE) from size-exclusion chromatography were pooled, concentrated, and stored (at –80°C). Relative NMR signal reductions for amide resonances in 1H-15N HSQC spectra were determined and mapped. For 1D 1H-NMR LUV binding assays, the fraction of bound protein at each titration point was averaged between replicates and plotted as the mean. Error bars for binding curves represent 1σ from this mean value. Error bars shown for Kdapp values represent 1σ for mean Kdapp from fitting binding isotherms.

X-ray data collection and processing

Request a detailed protocol

Diffraction data from a single native crystal were collected on beamline i04 of the Diamond Light Source (DLS), UK. Data were processed with CCP4 dials (Beilsten-Edmands et al., 2020; Winn et al., 2011; Winter, 2010; Winter et al., 2018) and scaled using dials.scale (Evans, 2006) within the Xia2 package (Winter et al., 2013). Multi-wavelength anomalous diffraction (MAD) data from a single SeMet labelled crystal were collected on beamline i04 of the DLS at the following wavelengths: peak = 0.9795 Å, inflection = 0.9796 Å, and remote = 0.9722 Å. Data were processed initially by AutoProc (Vonrhein et al., 2011). Substructure definition and initial model building were performed using AutoSHARP (Vonrhein et al., 2007). This was followed by manual building in Coot (Emsley et al., 2010) and further refinement using Phenix Refine (Adams et al., 2010). Data collection statistics have been published previously (Kumar et al., 2022). The structure has been deposited in wwPDB under accession code: PDB ID 8C4A (Berman et al., 2007).

Small angle X-ray scattering

Request a detailed protocol

Samples for SAXS measurements were prepared by concentrating samples from SEC using a centrifugal spin device with a molecular weight cut-off of 100 kDa. For all experiments, a buffer of 25 mM Tris, 5 mM TCEP, and pH 8.0 was used, which was adjusted to a pH value between 4.0 and 8.0 by addition of 1 M HCl leading to a salt concentration of maximum 30 mM in the case of pH 4.0. SAXS data were measured using a laboratory based flux-optimised Bruker AXS Nanostar with a gallium liquid metal jet source (Schwamberger et al., 2015) and scatterless slits (Li et al., 2008), a detector distance of 867 mm, and a q range of 0.0098–0.42 Å–1. The entire q range could be probed with this setup in a single measurement. More information about the optimised instrument can be found here (Lyngsø and Pedersen, 2021). All data were measured for 1800s at 20°C. SAXS data are plotted as an intensity as a function of q, which is the modulus of the scattering vector and defined as q = (4πsin[ϴ])/λGa, where 2ϴ is the scattering angle between the incident and scattered beam and λGa = 1.34 Å. Data were background subtracted and converted to absolute scale using the software package SUPERSAXS (CLP Oliveira and JS Pedersen). The mass was calculated using M=(I0×NA )/(c×Δρm2), where I(0) is the intensity extrapolated to q=0, NA is Avogadro’s number, c is the protein concentration in g/mL, and Δρm is the scattering contrast per mass that can be estimated to 2.0×1010 cm/g for a typical protein. I(0) was determined both by a Guinier fit analysis (using the intercept with the y-axis) and from an IFT routine (Glatter, 1977). The theoretical scattering curve was calculated using the program wlsq_PDBx (Steiner et al., 2018). An ab initio reconstruction of the protein structure was performed using GASBOR from the ATSAS package (Svergun et al., 2001) where the number of amino acids is given, and each amino acid is represented by a dummy residue. The optimisation was performed with the real space option due to the large size of GAC.

Molecular dynamics simulations

Request a detailed protocol

The TgGAC crystal structure was used to generate coarse-grained protein:membrane simulation systems in which TgGAC was rotated randomly in respect to the membrane patch to generate the first set of coordinates. CHARMM-GUI Martini Maker was used to generate a system in which GAC was translated 12 nm in the z direction from the centre of the membrane (Qi et al., 2015). The membrane composition used was 50% POPC: 50% POPA to replicate simulations performed in Darvill et al., 2018. For two independent replicates, TgGAC was rotated again (0, 90) and (90, 0). Simulations were performed using the gromacs biomolecular software package version 2021.3 (Hess et al., 2008) and the MARTINI3 forcefield with ElNeDyn restraints (Souza et al., 2021). The v-rescale thermostat (tau 1.0 ps; Bussi et al., 2007) and the Parrinello–Rahman barostat (tau 12.0 ps; Parrinello and Rahman, 1981) were used to maintain temperature (303.15 K) and pressure (1 bar). Production simulations were 10 μs in length. Analysis was performed using gromacs tools, VMD (Humphrey et al., 1996), and the ProLint server (Sejdiu and Tieleman, 2021).

To generate a model for a TgGAC open conformation, we used steered-MD. The TgGAC crystal was first used to generate a coarse-grained model, using the CHARMM-GUI Martini Maker (Qi et al., 2015) with the MARTINI3 forcefield and the ElNeDyn restraints (Souza et al., 2021). The restraint network at the N-/C-terminal interface was removed using an in-house Python script. A small and optimised pulling force was applied (pull rate = 0.01 nm/ns, k=100 kJ/mol.nm2) between a selected helix on either side of the interface to drive its separation and the extension of the structure. A series of structures along the trajectory were generated over the course of an averaged 2.5 μs simulations in length. The procedures were repeated to drive the separation of the three supercoiled pyramid interfaces (TgGAC1–619, TgGAC620–1114, and TgGAC1114–1661). The snapshots were back mapped to atomistic models using Martini to All-atom Converter (Wassenaar et al., 2014) and used in the fits of the SAXS data.

Plaque assay

Request a detailed protocol

Human foreskin fibroblasts (HFFs) monolayers were infected with freshly egressed parasites and incubated for 7 days at 37°C. Cells were then fixed with 4% paraformaldehyde (PFA)/0.05% glutaraldehyde for 10–15 min. After neutralisation with 0.1 M glycine/PBS, cells were stained using crystal violet. For quantification, pictures were taken, and the plaque area was determined using ImageJ.

Immunofluorescence assay

Request a detailed protocol

For vacuoles images, parasites were inoculated on an HFF monolayers previously seeded on a glass coverslip. The parasites were grown for 16–24 hr at 37°C. For extracellular parasites, freshly egressed parasites were seeded on gelatin-coated coverslips, and media containing BIPPO was used to stimulate motility. The coverslips were then fixed with 4% PFA/0.5% glutaraldehyde (PFA-Glu) during 10 min. The fixative agent was then neutralised using 0.1 M glycine/PBS for 10 min. Cells were then permeabilised for 20 min using 0.2% TritonX100/PBS and blocked using 5% BSA/PBS for 20 min before incubation with primary antibodies in 2% BSA/0.2% TritonX100/PBS for 1 hr. Coverslips were then washed three times for 5 min using 0.2% TritonX100/PBS. Secondary antibodies were incubated for 1 hr similarly to the primary antibodies. Finally, the coverslips were washed three times with PBS before mounting on glass slides using DAPI-containing Fluoromount. For immunofluorescence analysis, the secondary antibodies Alexa Fluor 488 and Alexa Fluor 594 conjugated goat α-mouse/rabbit antibodies (Molecular Probes) were used.

Liposome binding assay

Request a detailed protocol

Proteins were expressed in bacteria and purified by the PPR2P platform (University of Geneva). The liposomes were prepared in-house as follow: lipids were mixed in glass vials with the following proportions: 10% DOPE +X%DO-PA + qsp%DOPC (percentage by weight). The mix was slowly dried using nitrogen flow. The lipids were dried further in a dessicator for 30 min. The dried lipids were then resuspended in ‘lipid buffer’ (50 mM Hepes pH 7.5/100 mM NaCl/5% glycerol) to reach a 5 mg/mL concentration. Resuspension was then ensured by vortexing the mixes for 5 min. Then, seven freeze–thaw cycles were performed (20 s in liquid nitrogen followed by 90 s in a 33°C water bath). Lipid mixes were extruded by passing them through 0.1 µm filters for 21 times. Liposomes were then aliquoted in microtubes (50 µL), flash frozen in liquid nitrogen, and stored at –80°C until use.

For the binding assay, proteins were resuspended in ‘protein buffer’ (20 mM Tris pH 7.4/200 mM NaCl/1 mM DTT), centrifuged at 100,000 g for 30 min at 4°C to remove any precipitate and measure the exact protein concentration in the supernatant. Protein concentration was then adjusted to 0.2 mg/mL using protein buffer. In parallel, the liposomes were diluted at 2 mg/mL in ‘lipid buffer’ (50 mM Hepes pH 7.4/100 mM NaCl/1% glycerol) and re-diluted twofold using water to achieve a concentration of 1 mg/mL. For the ‘no lipid’ conditions, lipid buffer was simply diluted twofold in water. For the reaction, 40 µL of protein and 40 µL of liposomes were mixed and incubated on ice for 1 hr. The mix was then centrifuged at 120,000 g for 30 min at 4°C. Supernatant and pellets were separated, resuspended in Laemmli buffer, and boiled 10 min at 95°C. Equal volumes were then ran on polyacrylamide gels and stained using Coomassie blue. Quantification was performed by band densitometry.

HDX-MS sample preparation and data analysis

Request a detailed protocol

HDX-MS experiments were performed at the UniGe Protein Platform (University of Geneva, Switzerland) following a well-established protocol with minimal modifications (Wang et al., 2018). Details of reaction conditions and all data are presented in Supplementary file 2 and Supplementary file 3. HDX reactions were done in 50 μL volumes with a final protein concentration of 2.4 μM of GAC protein. Briefly, 120 pmol of the protein in 10 µL final volume were pre-incubated 5 min at 22°C before the reaction.

Deuterium exchange reaction was initiated by adding 40 µL of D2O exchange buffer (20 mM Tris pH 8, 150 mM NaCl, and 5 mM DTT in D2O) to the protein–peptide mixture. Reactions were carried-out at room temperature for three incubation times (3 s, 30 s, and 300 s) and terminated by the sequential addition of 20 μL of ice-cold quench buffer 1 (4 M Gdn-HCl, 1 M NaCl, 0.1 M NaH2PO4 pH 2.5, 1% formic acid [FA], and 200 mM TCEP). Samples were immediately frozen in liquid nitrogen and stored at –80°C for up to 2 weeks. All experiments were repeated in triplicate.

To quantify deuterium uptake into the protein, samples were thawed and injected in a ultra-performance liquid chromatography (UPLC) system immersed in ice with 0.1% FA as liquid phase. The protein was digested via two immobilised pepsin columns (Thermo #23131), and peptides were collected onto a VanGuard precolumn trap (Waters). The trap was subsequently eluted, and peptides were separated with a C18, 300 Å, 1.7 μm particle size Fortis Bio 100×2.1 mm column over a gradient of 8–30% buffer C over 20 min at 150 μL/min (buffer B: 0.1% FA; buffer C: 100% acetonitrile). Mass spectra were acquired on an Orbitrap Velos Pro (Thermo), for ions from 400 to 2200 m/z using an electrospray ionisation source operated at 300°C, 5 kV of ion spray voltage. Peptides were identified by data-dependent acquisition of a non-deuterated sample after MS/MS, and data were analysed by Mascot. All peptides analysed are shown in Supplementary file 2 and Supplementary file 3. Deuterium incorporation levels were quantified using HD examiner software (Sierra Analytics), and quality of every peptide was checked manually. Results are presented as percentage of maximal deuteration compared to theoretical maximal deuteration. Changes in deuteration level between two states were considered significant if >20% and >2 Da and p<0.01 (unpaired t-test).

Strains generation – GAC mutations at the endogenous locus

Request a detailed protocol

The pGST-GAC_PH domain vector used to produce the PH domain of GAC was used as a base to generate the points mutation. The points mutations were inserted using the Q5 mutagenesis kit (New England Biolabs) and specific primers. The mutations in the PH domains were checked by sequencing. Then, the 5’UTR_UPRT-pT8-DDGACΔPH-Ty-DHFR-3’UTR_UPRT was linearised with SgrAI, while the mutated PH domain from the pGST vector was amplified by PCR using primers containing homology regions. The mutated PH domain was inserted in the linearised 5’UTR_UPRT-pT8-DDGACΔPH-Ty-DHFR-3’UTR_UPRT by Gibson assembly. The insertion of the PH domain in the receiving vector was checked by analytical digestion and sequencing. The mutated PH domain of newly generated 5’UTR_UPRT-pT8-DDGACmut-Ty-DHFR-3’UTR_UPRT vector (therefore without introns) was amplified by KOD PCR using specific primers containing homology regions (on one side with the sequence preceding the PH domain, and on the other with the 3’UTR of the GAC gene). A double gRNA targeting the region preceding the PH domain and the 3’UTR of the GAC gene was generated in parallel by Q5 mutagenesis. For transfection, the equivalent of 100 µL of KOD PCR and 40 µg of gRNA was used. 48 hr after transfection, the parasites were fluorescence-activated cell sorted (FACS) to select clones expressing the gRNA-Cas9-YFP construct. Then, the individual clones were amplified, genomic DNA was extracted, and integration PCR of the GAC PH domain region was checked (the PCR ensured the loss of the introns, generating a smaller PCR amplicon in modified parasites compared to wild types). In addition, the PH domain of the PCR positive clones was fully sequenced. Primers are shown in Supplementary file 4.

Materials and data availability

Request a detailed protocol

Reagents generated in this study will be made available upon request. All data generated or analysed during this study are included in the manuscript and supplementary files. DNA primers are listed in Supplementary file 4. The structure has been deposited in wwPDB under accession code: PDB ID 8C4A (Berman et al., 2007). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2022) partner repository with the dataset identifier PXD039335.

Data availability

Diffraction data have been deposited in PDB under the accession code 8C4A.The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD039335.All data generated or analysed during this study are included in the manuscript, figures and supplementary files.

The following data sets were generated
    1. Kumar A
    2. Morgan RML
    3. Matthews SJ
    (2023) RCSB Protein Data Bank
    ID 8C4A. Structural and regulatory insights into the glideosome-associated connector from Toxoplasma gondii.
    1. Kumar A
    2. Morgan RML
    3. Matthews SJ
    (2023) PRIDE
    ID PXD039335. Structural and regulatory insights into the glideosome-associated connector from Toxoplasma gondii.

References

    1. Emsley P
    2. Lohkamp B
    3. Scott WG
    4. Cowtan K
    (2010) Features and development of coot
    Acta Crystallographica. Section D, Biological Crystallography 66:486–501.
    https://doi.org/10.1107/S0907444910007493
    1. Evans P
    (2006) Scaling and assessment of data quality
    Acta Crystallographica Section D Biological Crystallography 62:72–82.
    https://doi.org/10.1107/S0907444905036693
    1. Winter G
    2. Lobley CMC
    3. Prince SM
    (2013) Decision making in xia2
    Acta Crystallographica. Section D, Biological Crystallography 69:1260–1273.
    https://doi.org/10.1107/S0907444913015308

Decision letter

  1. Olivier Silvie
    Reviewing Editor; Sorbonne Université, UPMC Univ Paris 06, INSERM, CNRS, France
  2. Amy H Andreotti
    Senior Editor; Iowa State University, United States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Structural and regulatory insights into the glideosome– associated connector from Toxoplasma gondii" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Amy Andreotti as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Although the structure of GAC is overall well described it requires further polishing in terms of model quality and accuracy (as it stands it appears like a structure at an early stage of refinement). Please carefully address the reviewer comments as detailed below in the recommendations for the authors.

2) Alternative interpretation of the SAXS data should be considered concerning the oligomerization state of GAC induced by pH.

3) The manuscript should be carefully reviewed to correct numerous errors in the text, figures and legends.

4) The model presented in Figure 8 is too speculative and the figure should be removed.

Reviewer #1 (Recommendations for the authors):

– The crystal structure model a long way from being good enough in quality and appears like a structure at an early stage of refinement. It has a large number of Ramachandran and bond length and angle outliers. Looking at the model, there are also large areas in which the model does not fit the map. I have not mentioned this in the public report, to be kind, but this really must be fixed, or it will be embarrassing. A structure at this resolution should really have no outliers (except perhaps for a few Ramachandran outliers if the density shows them to be genuine) of these types and should fit the maps.

– Figure 1 does not make clear what is a model and what an experimentally derived structure is. I would show the NMR and crystal structures as the first two panels and only then the composite. In the era of AlphaFold, it is important to distinguish clearly between experimentally determined structures and models.

– There are quite a lot of errors in the text and figures and legends.

– Error in Figure 4 legend – two (e) and no (f).

– Error in Figure 5 legend – there is no (d) or (e) in the figure.

– Comment on the differential effect of mutations – for example, KER – on full–length and PH domains.

– Figure 7d/e – what are GAC N100 and GAC N70? What is peptide centroid? Why is this not shown for the full protein length when GAC fl has been studied in this way? What is e?

– It felt to me that the data in Figure 7d/e would fit better with the SAXS data and make more sense to readers here when describing the flexibility of GAC.

– Personally I would remove Figure 8 as it is extremely speculative.

Reviewer #2 (Recommendations for the authors):

1. Structure (X–ray and NMR) description.

The full–length and quite large (2639 residues) structure was solved in a composite way. The largest part (7–2504) solved Se–MAD crystallography revealing 53 modules resembling armadillo/heat–like repeats and a small (2505–2639) and dynamic pleckstrin homology–like domain C–terminal domain solved by NMR assisted with AlphaFold2 modelling.

The structure is well–described and illustrations are relatively clear to get a sense of the modular organization of this large modular protein. Figure S2 in particular is to be mentioned as the Authors took the time to generate a SS mapping of the entire large structure on the sequence.

This reviewer is a crystallographer and inspected the density map in its entirety in COOT using the PDB and two maps provided 2Fo–Fc and Fo–Fc (thank you for the PDB and maps). It is a large protein so I am aware it represents a considerable amount of work but the repetitive modular organization is actually helpful to guide building (especially with other related structures at hand). This structure still requires some significant polishing in some areas where things can be improved and clearly cleaned up especially for this resolution range considering it was phased by MAD.

A lot of seleno methionines need to be adjusted better in the maps. Especially the Selenium atoms are electron–dense so it should be easy to fix. Quite a few side chains are built in but have no density, might want to do something about that and a few missing could be built in or readjusted.

I listed down some of the most obvious problems.

SeMet 30

R84

SeMet 89

S120

Q125

E128

T160 <->Q203

Loop 275-285. Discontinuous there but strong residual density

Fragment 326-329 is problematic

Q403

E420

E437 backbone trace?

Y488

K525 and R560 no sidechain visible

E745

D787-V790 check the backbone. This is wrongly traced.

H810 area is weak. Bad geometry of the backbone around that turn. Please inspect.

Q957/Q959

Q981

K1041

T1069 rotamer?

R1166

E1183

D1221

K1321

SeMet 1408 is plain broken in my coot?

E1510

E1530

SeMet 1609

D1663

R1667

SeMet 1707

E1749

K1752

R1832

H1988

SeMet 2010

Segment L2052-T2056 is problematic

SeMet 2343 and 2349

E2412

SeMet 2416

I see a disulfide bond between residues C2420 and C2467 but that looks really wrong. It is a cytoplasmic protein in the parasite and was expressed in E. coli. The maps show that the C2420 SH and C2467 point outwards and would not be engaged into an S-S bridge that looks wrong in the Fo-Fc negative maps. So recheck this and refine accordingly. C2420 looks towards T2470 while C2467 would rather look towards the backbone carbonyl of T2466.

Would these cysteines be conserved in Plasmodium?

Recheck your waters. Some dense peaks could be assigned to waters.

The statistics for crystal structure resolution (i.e., Se-MAD phasing) are not reported accurately. This should be remedied.

Phasing and refinement.

Data quality and refinement/phasing.

Outer shell statistics with Rmerge and Rmeas of over 100% with an I/σ(I) of only 0.71 are a bit extreme even to me. I know that the contributions of weak reflections are weighted down with all the maximum likelihood refinement procedures but still. Please report the multiplicity/redundancy of the overall data set and the outer shell in this present version of the study.

Table 1. Since no phasing information is reported there, it would be adequate to remove phasing from its title. It is only data quality and refinement statistics.

That said it is expected to scientifically and rigorously report the phasing statistics since no phasing statistics can be found anywhere in this manuscript. I am sure the first and last authors will agree with me on this. Citing the other article *, which however does report multiplicity, is of no interest in this case.

Kumar A, Zhang X, Vadas O, Stylianou FA, Dos Santos Pacheco N, Rouse SL, Morgan ML, Soldati–Favre D, Matthews S (2022) Secondary Structure and X–rayCrystallographic Analysis of the Glideosome–Associated Connector (GAC) from Toxoplasma gondii. Crystals 12

Refinement. In light of what I saw while inspecting the whole structure and ED Fourier difference maps, it might be useful to consider anisotropic TLS refinement of atomic displacement parameters, especially in light of the 'subdomain' structures suggested by SAXS/HDX and MD simulation data. Coloring the structure by the B factor on the backbone is quite indicative actually.

2-SAXS analysis.

The authors use SAXS to characterize the solution–extended conformation of TgGAC and show that it is different from the crystal conformation.

Reading Kumar et al. (2021) one notices that crystals were grown at pH=5 using a protein initially purified at pH=8.0.

There is a piece of data that could be shown by the Authors in that study. It would be the superimposed SEC profiles of purified TgGAC at pH 8 and pH 5 under a diluted regime maybe (but with some salt around) see my comment below.

Authors show SAXS profiles at different pHs in a diluted regime (Figure 2b), how is pH change performed? They notice that oligomerization seems the most pronounced at pH5 (the crystallization pH) but we will also make the observation that the SAXS buffer has virtually no salt (25 mM Tris) and these are not ideal conditions to limit intermolecular interactions and maybe aggregation independently of molecular compaction (closing of the structure as in the crystal conformation).

The P(r) compared are different. The solution curve was obtained at pH 8 but the calculated curve was done on a structure obtained at pH 5. The N/C interface being so charged it would not be surprising that it is sensitive to pH changes. In this case, why not measure and calculate the experimental P(r) function at pH5 would it fit (or not, like what is observed in pH 8) the crystal structure? The SAXS study of Jacot et al. (2016) mentions a range of concentrations from 1.2 to 4.6 mg/ml for the equivalent construct (FL–TgGAC) on a wider angular range it seems compared to what is described here.

Since the authors present P(r) analysis, in addition to the large difference in RG it would be meaningful to emphasize in the figures the maximum intramolecular distance (Dmax) calculated with the X–ray structure (~145 Å from what I see in Figure S3b/S2c) to the 400Å distance estimated by Fourier inversion of the experimental scattering curve. This clearly shows that the structure seems more elongated in solution providing that there is no significant aggregation or oligomerization.

Concerning the ab initio reconstruction approach, since the authors have a crystal structure could they not model a few subsets of extended structures using simple rigid body motions or the models later discussed in Fig8 and assess the quality of the corresponding I(q) fit with programs such as Crysol.

Figure S3a.

Small typo in the figure, it should be eight instead of eight and the X–axis unit legend is partially occulted for the scattering angle q2 (Å–2).

For the Guinier analysis. Fitting a straight line on a scattering curve is usually done indicating the corresponding q.RG range of points (only 8 points here based on what is reported).

Thus the q.RG range would be going up to ~ 0.017x105=1.8 which is rather generous but not unheard of. I understand this was done with a high–flux in–house instrument (described in Lyngso and Pedersen J. Appl. Cryst. 2021, 54), not on a synchrotron beamline. Could the distance between the sample and detector be specified and also the resulting angular range(s) used to measure in the Guinier (low q) and P(r) regimes (high q), were two curves merged or was it a single I(q) curve used for the Fourier inversion?

Figure S3b which is also Figure 2c? For the pair–distance distribution function P(r) derived from experimental scattering curves, it is unclear to me if the error bars are included in the graphs. The concentration range seems quite low to obtain a scattering curve with a good s/n ratio in the large q, high–resolution regime.

There is also a typo. MW should be 286 kDa instead of mg/ml

I agree overall with the conclusions of the SAXS section despite an issue with the role of pH and the buffer conditions used here. Besides the few errors that need to be corrected in the figures and some experimental details that could be added in the methods section.

Page 6. "collectively, it can be concluded that a closed structure…"

Would there be a way to illustrate this then? Using a sequence co–conservation projection coloring projected on the two surfaces (N/C) involved.

The PH domain of TgGAC binds to membranes enriched in PA through two basic patches as shown by nice MD simulations (Figures3 and S3) and NMR studies using also the corresponding PfGAC PH domain using elegant (at least to me) relaxation methods to map and somehow quantify interactions between PH domain and either nanodiscs or LUVs which they should define as large unilamellar vesicles somewhere in their manuscript although they also assimilate them to liposomes (Figures4, S4 and S5).

I would suggest the authors state the sequence conservation (similarity or identity) % based on PH structure–guided sequence alignments for Pf and Tg. Sequence conservation is mentioned in the introduction but not 'quantified' explicitly, as a matter of fact, the information would be useful for the full–length structure. This brings back my comments about my surface conservation analysis for Apicomplexa.

While Pf is obviously also of 'interest' (maybe more so than Tg) and it emphasizes conservation of the process throughout the phylum Apicomplexa, is there a practical/technical reason to jump between Tg and Pf systems for the PH–lipid interaction studies by NMR? It sometimes gets a bit confusing to jump back and forth.

The mapping of the PA binding site on the PH domain is well described and is validated by site–directed mutagenesis on isolated TgGAC–PH domain but also on full–length TgGAC (Figure 5). Although the liposome binding assay is a simple centrifugation assay not a classical flotation assay with gradient centrifugation separation in sucrose, the results are rigorously presented and support the conclusion about the role played by the PH domain during GAC preferential binding to PA–enriched membranes

Page 9. The title 'Targeted mutations in GAC PH domain are not fitness conferring in vivo' confuses me. Is conferring truly what the authors intended? I would think they meant targeted mutations do NOT affect fitness or virulence. Am I missing something?

The next two sections of the manuscript (but not the discussion) lack clarity and require some substantial rewriting/detailing to help the reader follow the intellectual thread developed by the Authors.

H/D exchange analysis and Figure 7 as a whole

While panels b and c are clear and show some endogenous interaction between TgGAC and endogenous parasitic actin. The rest of the section associated with Figure 7 needs some serious rewriting and explaining. What are GAC N70 and GAC N100? Despite figure S3 I genuinely do not understand.

I have absolutely no clue as to what I am looking at in Figures 7d and 7e and furthermore, panel 7e is not captioned either so please clarify this (at least to me) in the figure and the main text of the manuscript.

Also in Fig7 panel a. Something seems wrong or misleading between the helical cartoon model on the left and the schematic on the right. Why is there magenta on the top of the schematic? Is that the linker, if so it should be colored differently and in a thinner line, then the linear protein schematic below should be edited accordingly.

Page 10/11. Modelling of TgGAC binding to MIC2.

I would suggest precision (again) about what MIC2 is, some kind of adhesin. It is only briefly mentioned once in the introduction on page 3, but that is about it. Please help us (re)connect some dots there. Not everyone is an addict to glideosome machineries.

Figure 8 should be annotated more rigorously. In the legend.

8a left side should be captioned a bit more carefully and rigorously GAC ARII–ARIII it seems, the rest is omitted. MIC2 tail (red I guess).

The description becomes increasingly unclear and we are left trying to guess what we are looking at. Page 11.

De facto these are not superimpositions. Correct me if wrong.

You have a model (green GAC arch 39–42 + red MIC2 ?) of your complex (left panel) and shown next to them in the same relative orientation as the corresponding other structurally/functionally related complexes p120, catenin, etc to emphasize peptide/ARM repeats interactions. E–cadherin labels could be colored in red and it suddenly becomes crystal clear.

Panel 8b is similarly and annoyingly cryptic.

8b left side…what is the magenta–pink thing? An F actin filament…

8b right side…the boxed structure is a crystal or cryoEM structure of? At least a PDB code and a reference in the legend would be required.

The discussion is clearly written and puts things in perspective summarizing the roles of PH in PA binding and protein stabilization although this latter issue is mentioned but not really explained.

Since it seems that the PH domain is not required for fitness and membrane binding seems dispensable, the authors state that its presence is required for proper folding of the full–length construct… Has there been any attempt to generate parasites lacking the c–terminal PH domain?

I have one main objection. The oligomerization state induced by pH as seemingly observed by SAXS.

Having done a fair amount of X–ray and neutron scattering, I am somehow puzzled by the experimental buffer conditions chosen by the authors (the same applies to the study by Jacot et al. that I looked at). There is no salt present, even under the relatively low protein concentration described here it is reasonable to think that they could be non–specific interactions between molecules.

I would temper the interpretation of the data at pH 5 with that so–called oligomerization that could be aggregation in disguise.

I quote Kumar et al. in Crystals 2021

Reproducible protein crystals were obtained in 100 mM magnesium acetate, 100 mM sodium acetate, 6% PEG8000, and pH 5.0. These were manually optimised by screening over sodium acetate pH ranges of 4.0 to 5.0 in one dimension and a PEG8000 concentration gradient of 4%–10% in the second dimension. Crystallization was set up at a concentration of 5–60 mg mL−1.

Obviously, the protein must not be that unhappy at pH 5 to crystallize but it is in presence of some significant amount of soft salts.

I would really be curious to see a SEC of FL–GAC profile at ph 5 and 8 in presence of a decent amount of salt like we crystallographers like to use when purifying proteins. Ideally, the corresponding SAXS in more 'salty conditions' might be of interest.

The authors put their work in the perspective of understanding glideosome assembly and mechanism of action to promote adhesion and the generation of movement/gliding of parasites as they enter or egress the host cell. The availability of the first full-length atomic structure of an apicomplexan GAC is an important step towards a mechanistic understanding of glideosome assembly and action. Identification of protein-protein and protein–lipid interactions can provide new ways to design anti-parasitic drugs.

Reviewer #3 (Recommendations for the authors):

There are several areas in which the manuscript could be improved.

– A schematic diagram placing the GAC in the context of the parasite would be a valuable addition to figure 1, as would including the sequence diagram from Figure 7a in Figure 1.

– More information is required to judge the reproducibility of binding data in the relevant figure panels. How many times were binding experiments repeated? Do repeats represent technical repeats with the same protein preparation or different preparations? Do error bars represent SEM? Please add this information to relevant figure legends.

– Figure S1. It is difficult to judge concordance between experimental NOE data and NOE count from the AlphaFold model. Can a chart of experimental NOEs be shown for comparison? Panel E requires colour key.

– Sequence conservation of the N/C interface is mentioned in the text several times, but it would be useful to show this plotted on the structure.

– The AlphaFold prediction of the interface between GAC and MIC2 is interesting but not convincing in its current form. A PAE plot is required to judge the confidence of the GAC–MIC2 interface. pLDDT scores (ideally plotted on the structure) are required to judge confidence in the conformation of the MIC2 peptide.

– Does AlphaFold predict TgGAC to be fully closed, as in crystal structure, or partially separated, as in the PfGAC prediction?

– p9. The section titled "Targeted mutations in GAC PH domain are not fitness conferring in vivo" is confusing. Key lipid–binding residues in the GAC PH domain are not essential for parasite invasion.

– Figure 8b, middle panel. The basis for how the GAC is placed with respect to actin is unclear – it would seem that there are many other ways it could be placed that are equally compatible with the data.

– There are numerous typographical issues, including mis-numbering of the supplemental figures, which make the manuscript difficult to read.

https://doi.org/10.7554/eLife.86049.sa1

Author response

Essential revisions:

1) Although the structure of GAC is overall well described it requires further polishing in terms of model quality and accuracy (as it stands it appears like a structure at an early stage of refinement). Please carefully address the reviewer comments as detailed below in the recommendations for the authors.

The authors have been progressing the refinement and the current model is now significantly improved. See comments below and validation report.

2) Alternative interpretation of the SAXS data should be considered concerning the oligomerization state of GAC induced by pH.

We have toned down the discussion of the pH dependent SAXS data (section GAC adopts multiple extended conformations in solution). The choice of ‘no salt’ in our ideal buffer came out of numerous screens for GAC stability, homogeneous behaviour in size exclusion chromatography, reproducibility in crystallisation. We have we now included size exclusion profiles in Figure 2—figure supplement 1, which confirm oligomerisation at pH 5 and that salt has no effect of the behaviour of TgGAC at either pH.

3) The manuscript should be carefully reviewed to correct numerous errors in the text, figures and legends.

The manuscript has been thoroughly checked for textual mistakes and these have fixed.

4) The model presented in Figure 8 is too speculative and the figure should be removed.

The speculative models in Figure 8 have been removed. The structural comparison of the C-terminal arch of TgGAC with the β-catenin ARM structure is shown as a supplementary figure (Figure 1—figure supplement 3), without the modelled structure of TgGAC/MIC2 as that requires further validation. We now comment of the structural similarity of the β-catenins in the discussion (lines 424-429).

Reviewer #1 (Recommendations for the authors):

– The crystal structure model a long way from being good enough in quality and appears like a structure at an early stage of refinement. It has a large number of Ramachandran and bond length and angle outliers. Looking at the model, there are also large areas in which the model does not fit the map. I have not mentioned this in the public report, to be kind, but this really must be fixed, or it will be embarrassing. A structure at this resolution should really have no outliers (except perhaps for a few Ramachandran outliers if the density shows them to be genuine) of these types and should fit the maps.

Agreed and we thank the reviewer for his expert analysis of the GAC crystal structure. The authors have been progressing the refinement and made significant modifications and improvements to the model. Individual residues that the reviewer kindly earmarked have been addressed. The Ramachandran outlier statistics have been improved (<1%). For some solvent exposed loop regions, the authors have attempted to model the residues as best possible, but a few Ramachandran outliers persist in those regions where the electron density is poor.

– Figure 1 does not make clear what is a model and what an experimentally derived structure is. I would show the NMR and crystal structures as the first two panels and only then the composite. In the era of AlphaFold, it is important to distinguish clearly between experimentally determined structures and models.

Suggested additional figure panels are included in Figure 1 to clearly separate the crystal structure and the NMR-structure/AlphaFold model of the PH domain. The Figure 1 caption now explained the full-length structure is a hybrid of both.

– There are quite a lot of errors in the text and figures and legends.

The manuscript has been thoroughly checked for textual mistakes and these have fixed.

– Error in Figure 4 legend – two (e) and no (f).

Fixed.

– Error in Figure 5 legend – there is no (d) or (e) in the figure.

Fixed.

– Comment on the differential effect of mutations – for example, KER – on full–length and PH domains.

Agreed this is interesting and we now comment on this and the implication for other regions in full-length GAC contributing to membrane binding (line 281-284).

– Figure 7d/e – what are GAC N100 and GAC N70? What is peptide centroid? Why is this not shown for the full protein length when GAC fl has been studied in this way? What is e? It felt to me that the data in Figure 7d/e would fit better with the SAXS data and make more sense to readers here when describing the flexibility of GAC.

Figure 7 and the H/D exchange section has been completely reworked to improve clarity and accuracy. In summary, the HDX-MS in Figure 7 establishes that the conformation of the TgGAC1-619 and TgGAC1-1114 have similar conformations to when in context of full-length TgGAC by probing dynamics and accessibility of amides to exchange. We also show a supplementary Figure S6c of the NMR spectrum for TgGAC1-619 indicating that it is well folded.

– Personally I would remove Figure 8 as it is extremely speculative.

The speculative models in Figure 8 have been removed. The structural similarity of the C-terminal arch of TgGAC with the β-catenin ARM structure is shown as a supplementary figure (Figure 1—figure supplement 3),, without the modelled structure of TgGAC/MIC2 as this requires further validation. We now comment of the structural similarity of the β-catenins in the discussion (line 426-430).

Reviewer #2 (Recommendations for the authors):

1. Structure (X–ray and NMR) description.

The full–length and quite large (2639 residues) structure was solved in a composite way. The largest part (7–2504) solved Se–MAD crystallography revealing 53 modules resembling armadillo/heat–like repeats and a small (2505–2639) and dynamic pleckstrin homology–like domain C–terminal domain solved by NMR assisted with AlphaFold2 modelling.

The structure is well–described and illustrations are relatively clear to get a sense of the modular organization of this large modular protein. Figure S2 in particular is to be mentioned as the Authors took the time to generate a SS mapping of the entire large structure on the sequence.

Figure S2 (now Figure 1—figure supplement 2 is now explicitly mentioned in the text (line 115))

This reviewer is a crystallographer and inspected the density map in its entirety in COOT using the PDB and two maps provided 2Fo–Fc and Fo–Fc (thank you for the PDB and maps). It is a large protein so I am aware it represents a considerable amount of work but the repetitive modular organization is actually helpful to guide building (especially with other related structures at hand). This structure still requires some significant polishing in some areas where things can be improved and clearly cleaned up especially for this resolution range considering it was phased by MAD.

We thank the reviewer for this comment. Significant improvements have now been made to the model as stated in the response above.

The statistics for crystal structure resolution (i.e., Se-MAD phasing) are not reported accurately. This should be remedied.

We thank the reviewer for his observations on the statistics presented in the manuscript. A revised table has been provided to address the lack of phasing statistics. The improved refinement statistics are also provided.

Phasing and refinement.

Data quality and refinement/phasing.

Outer shell statistics with Rmerge and Rmeas of over 100% with an I/σ(I) of only 0.71 are a bit extreme even to me. I know that the contributions of weak reflections are weighted down with all the maximum likelihood refinement procedures but still. Please report the multiplicity/redundancy of the overall data set and the outer shell in this present version of the study.

Table 1. Since no phasing information is reported there, it would be adequate to remove phasing from its title. It is only data quality and refinement statistics.

That said it is expected to scientifically and rigorously report the phasing statistics since no phasing statistics can be found anywhere in this manuscript. I am sure the first and last authors will agree with me on this.

A revised table has been provided to address the lack of phasing statistics.

Refinement. In light of what I saw while inspecting the whole structure and ED Fourier difference maps, it might be useful to consider anisotropic TLS refinement of atomic displacement parameters, especially in light of the 'subdomain' structures suggested by SAXS/HDX and MD simulation data. Coloring the structure by the B factor on the backbone is quite indicative actually.

We thank the reviewer for their advice in refinement procedures. We have applied both mixed anisotropic TLS refinement parameters to our refinements and the R factors have improved (19% and 26%) as well as the resulting electron density maps.

2-SAXS analysis.

The authors use SAXS to characterize the solution–extended conformation of TgGAC and show that it is different from the crystal conformation.

Reading Kumar et al. (2021) one notices that crystals were grown at pH=5 using a protein initially purified at pH=8.0.

There is a piece of data that could be shown by the Authors in that study. It would be the superimposed SEC profiles of purified TgGAC at pH 8 and pH 5 under a diluted regime maybe (but with some salt around) see my comment below.

We have included a supplementary figure of the SEC profiles (Figure 2—figure supplement 1A) which show that as expected TgGAC behaves significantly larger at pH 5 than pH 8, and salt does not influence the elution volume (lines 158-160).

Authors show SAXS profiles at different pHs in a diluted regime (Figure 2b), how is pH change performed? They notice that oligomerization seems the most pronounced at pH5 (the crystallization pH) but we will also make the observation that the SAXS buffer has virtually no salt (25 mM Tris) and these are not ideal conditions to limit intermolecular interactions and maybe aggregation independently of molecular compaction (closing of the structure as in the crystal conformation).

The pH was adjusted from pH 8.0 to the desired pH by addition of the appropriate amount of 1 M HCl, which gives rise to a maximum salt concentration of 30 mM at pH 4.0. This is now added to the Materials and methods section. As the increase in ionic strength increases the screening of intermolecular interactions, lowering the pH should decrease rather than increase the tendency towards oligomerization for GAC. We presume the oligomerization observed at lower pH must stem from conformation changes rather than this small difference in ionic strength. We present our SEC profiles at high/low pH and salt concentration and demonstrate and no beneficial effect of salt on elution.

The P(r) compared are different. The solution curve was obtained at pH 8 but the calculated curve was done on a structure obtained at pH 5. The N/C interface being so charged it would not be surprising that it is sensitive to pH changes. In this case, why not measure and calculate the experimental P(r) function at pH5 would it fit (or not, like what is observed in pH 8) the crystal structure?

It is true that the p(r) functions are from, respectively, a structure at pH 5.0 and solution data at pH 8.0. Ideally, we would like to compare the structures at pH 5.0 in solution with the crystal structure, but as GAC oligomerises/aggregates at pH 5.0, we are not able to determine an overall size accurately. The absence of aggregation and the ability to determine overall size is a prerequisite for calculating the p(r) as one needs to estimate the Dmax. We have now stressed this point in the manuscript to make it clearer. We compare SAXS data at two pH values to demonstrate the presence of a much larger (aggregated) structure at a higher pH relative to the crystal structure.

The SAXS study of Jacot et al. (2016) mentions a range of concentrations from 1.2 to 4.6 mg/ml for the equivalent construct (FL–TgGAC) on a wider angular range it seems compared to what is described here.

We measure at relatively low concentration (1.3 mg/mL), as we want to go as low as possible to avoid any affect from intermolecular interactions that could promote aggregation. On the other hand, we cannot go significantly lower as the statistics on the data will then be considerable worse, so 1.3 mg/mL turned out to be the best compromise. We have added a comment about this in the Results section “GAC adopts multiple extended conformations in solution”. SAXS data in Jacot et al. (2016) are shown as a function of s (equivalent to q) on a linear scale, where we show data on a logarithmic s/q scale. We believe a logarithmic scale is more suitable as it more clearly shows the data at low q, where one would see effects from intermolecular interactions. On a linear s/q scale, this region is very small, and these important effects can be more difficult to detect. As data are on either a logarithmic or linear scale, it is not straightforward to compare the scales. Jacot et al. (2016) do not give a q range in the article, but we have copied the figure and estimated the position of the first and last point, which give a q range of around 0.012 – 0.43 Å-1 (see Author response image 1 for estimation of first and last point). Apparently, the two curves at the top have a larger range, however, this is in fact extrapolated model fit curves and not data. In our data we have a q range of 0.0098 – 0.42 Å-1 (which is also mentioned in the Materials and methods section now), so the range is similar to Jacot et al. (2016).

Author response image 1

Since the authors present P(r) analysis, in addition to the large difference in RG it would be meaningful to emphasize in the figures the maximum intramolecular distance (Dmax) calculated with the X–ray structure (~145 Å from what I see in Figure S3b/S2c) to the 400Å distance estimated by Fourier inversion of the experimental scattering curve. This clearly shows that the structure seems more elongated in solution providing that there is no significant aggregation or oligomerization.

Thank you for pointing this out. We have now emphasized this in the text to make it clearer (section ‘GAC adopts multiple extended conformations in solution’).

Concerning the ab initio reconstruction approach, since the authors have a crystal structure could they not model a few subsets of extended structures using simple rigid body motions or the models later discussed in Fig8 and assess the quality of the corresponding I(q) fit with programs such as Crysol.

We have tried modelling extended structures using simple rigid body motions and we devised another approach in which steered-MD was used with the application of a pulling force between residues on either side of the interface to drive its separation and the extension of the structure. The structure that fit best was the most extended generated from MD. These fits are presented in Figures 3 and described in the SAXS section ‘GAC adopts multiple extended conformations in solution’.

For the Guinier analysis. Fitting a straight line on a scattering curve is usually done indicating the corresponding q.RG range of points (only 8 points here based on what is reported).

Thus the q.RG range would be going up to ~ 0.017x105=1.8 which is rather generous but not unheard of.

We have now indicated it on the structure. It is 0.0155 Å-1 x 105 Å = 1.6.

I understand this was done with a high–flux in–house instrument (described in Lyngso and Pedersen J. Appl. Cryst. 2021, 54), not on a synchrotron beamline. Could the distance between the sample and detector be specified and also the resulting angular range(s) used to measure in the Guinier (low q) and P(r) regimes (high q), were two curves merged or was it a single I(q) curve used for the Fourier inversion?

Yes, the distance is 867 mm, the q range is 0.0098 – 0.42 Å-1, and one single curve was used where the whole q range could be probed in one measurement. This information is available in the article references for more information about the instrument, but we have now also added it to the Materials and methods section.

Figure S3b which is also Figure 2c? For the pair–distance distribution function P(r) derived from experimental scattering curves, it is unclear to me if the error bars are included in the graphs. The concentration range seems quite low to obtain a scattering curve with a good s/n ratio in the large q, high–resolution regime.

Error bars were not included initially but they are now and are small. We chose as low a protein concentration as possible while maintaining a good signal to noise ratio. Usually, such good statistics cannot be obtained with a concentration of 1.3 mg/mL, but keep in mind that GAC is a very large protein (286 kDa) and that the intensity scales with protein mass, so therefore much better data can be obtained at low concentrations for GAC compared to proteins with masses of 30-50 kDa.

I agree overall with the conclusions of the SAXS section despite an issue with the role of pH and the buffer conditions used here. Besides the few errors that need to be corrected in the figures and some experimental details that could be added in the methods section.

Fixed.

Page 6. "collectively, it can be concluded that a closed structure…"

Would there be a way to illustrate this then? Using a sequence co–conservation projection coloring projected on the two surfaces (N/C) involved.

Fixed and we now include a surface representation of the interface with Consurf conservation scoring in Figure 2b and a full sequence alignment in Figure 1—figure supplement 2.

The PH domain of TgGAC binds to membranes enriched in PA through two basic patches as shown by nice MD simulations (Figures3 and S3) and NMR studies using also the corresponding PfGAC PH domain using elegant (at least to me) relaxation methods to map and somehow quantify interactions between PH domain and either nanodiscs or LUVs which they should define as large unilamellar vesicles somewhere in their manuscript although they also assimilate them to liposomes (Figures4, S4 and S5).

LUVs and their size are now defined. Further details of LUV and nanodisc preparations are described (line 221 and in the Material and Methods).

I would suggest the authors state the sequence conservation (similarity or identity) % based on PH structure–guided sequence alignments for Pf and Tg. Sequence conservation is mentioned in the introduction but not 'quantified' explicitly, as a matter of fact, the information would be useful for the full–length structure. This brings back my comments about my surface conservation analysis for Apicomplexa.

This is now stated (53% identity) in the appropriate section on line 217. We also include a full annotated sequence alignment as Figure 1—figure supplement 2.

While Pf is obviously also of 'interest' (maybe more so than Tg) and it emphasizes conservation of the process throughout the phylum Apicomplexa, is there a practical/technical reason to jump between Tg and Pf systems for the PH–lipid interaction studies by NMR? It sometimes gets a bit confusing to jump back and forth.

Yes, it was a practical and technical reason for the Tg/Pf switch for the NMR-based lipid binding studies. To summarise, our NMR assignment for TgGACPH, was incomplete with several surface exposed amide NMR resonances between residues 2548-2560 absent from spectra. While this did not prevent our NMR structure validation, we felt that potential interacting residues could be missed in titrations with TgGACPH. We therefore first explored PfGACPH which displayed improved NMR spectra and near complete assignment was possible, which facilitated a comprehensive map of the PA binding surface. These results were then confirmed and validated with equivalent NMR experiments on TgGACPH. We have rewritten this explanation to improve clarity (from line 214).

Page 9. The title 'Targeted mutations in GAC PH domain are not fitness conferring in vivo' confuses me. Is conferring truly what the authors intended? I would think they meant targeted mutations do NOT affect fitness or virulence. Am I missing something?

This has been rephrased to remove confusion.

The next two sections of the manuscript (but not the discussion) lack clarity and require some substantial rewriting/detailing to help the reader follow the intellectual thread developed by the Authors.

These sections has been rewritten to improve clarity and comprehension (sections Evaluation of GAC and GAC fragments binding to toxoplasma F-actin and Investigation of GAC fragment conformations using hydrogen/deuterium exchange coupled to mass spectrometry).

H/D exchange analysis and Figure 7 as a whole.

Figure 7 and the H/D exchange section has been reworked to improve clarity and accuracy. In summary, the HDX-MS in Figure 7 establishes that the conformation of the TgGAC1-619 and TgGAC1-1114 have similar conformations to when in context of full-length TgGAC by probing dynamics and accessibility of amides to exchange. We also show a supplementary Figure 7—figure supplement 1C of the NMR spectrum for TgGAC1-619 indicating that it is well folded.

Page 10/11. Modelling of TgGAC binding to MIC2.

I would suggest precision (again) about what MIC2 is, some kind of adhesin. It is only briefly mentioned once in the introduction on page 3, but that is about it. Please help us (re)connect some dots there. Not everyone is an addict to glideosome machineries.

TgMIC2 is now introduced properly within the introduction (from line 45) and refreshed where mentioned in the discussion.

Figure 8 should be annotated more rigorously. In the legend.

8a left side should be captioned a bit more carefully and rigorously GAC ARII–ARIII it seems, the rest is omitted. MIC2 tail (red I guess).

The description becomes increasingly unclear and we are left trying to guess what we are looking at. Page 11.

There is significant consensus on the speculative nature of Figure 8, so it has been removed as suggested. We have however included the structural comparison of the C-terminal arch of TgGAC with the β-catenin ARM structure in a supplementary figure S7, without the modelled structure of TgGAC/MIC2 as that requires further validation. We now comment of the structural similarity with the β-catenins in the discussion (paragraph starting line 415).

De facto these are not superimpositions. Correct me if wrong.

See above point, the legend in the now Figure 1—figure supplement 3has been reworded accordingly.

You have a model (green GAC arch 39–42 + red MIC2 ?) of your complex (left panel) and shown next to them in the same relative orientation as the corresponding other structurally/functionally related complexes p120, catenin, etc to emphasize peptide/ARM repeats interactions. E–cadherin labels could be colored in red and it suddenly becomes crystal clear.

See above point.

Panel 8b is similarly and annoyingly cryptic.

8b left side…what is the magenta–pink thing? An F actin filament…

8b right side…the boxed structure is a crystal or cryoEM structure of? At least a PDB code and a reference in the legend would be required.

See above point.

The discussion is clearly written and puts things in perspective summarizing the roles of PH in PA binding and protein stabilization although this latter issue is mentioned but not really explained.

Since it seems that the PH domain is not required for fitness and membrane binding seems dispensable, the authors state that its presence is required for proper folding of the full–length construct… Has there been any attempt to generate parasites lacking the c–terminal PH domain?

Yes, there has been multiple attempts to generate GAC-ΔPH in the parasites either expressed as second copy or by deletion in the endogenous locus but all have failed. Expression appears to be very low, and the mutation is lethal. Concordantly, recombinant expression of GAC-ΔPH in bacterial/insect cells also failed suggesting that the mutant protein is not folded properly and unstable. This prompted us to focus on point mutants instead. See Author response image 2 for reviewer information.

Author response image 2
Production and purification of recombinant GAC-FL and GAC-∆PH.

Both GAC contract constructs are well expressed in BL-21 bacteria, showing an intense band at the expected MW. Purification using a combination of affinity, anion-exchange and size-exclusion chromatography (SEC) shows excellent result for GAC-FL. Although some protein elultes in the void volume 8ml on Superdex200 SEC, most of GAC-FL comes off the column as a symmetrical peak at 9.7ml suggested of monomeric protein. Purification of GAC-∆PH following a similar procedure as for GAC-FL shows many degradation products and absence of the intact protein on the SEC column. The purifications described here are representative of several attempts with no GAC-∆PH protein that could be isolated. The results indicate that C-terminal PH domain is essential for recombinant GAC stability, with removal of the C-terminal PH domain leading to degradation of GAC.

I have one main objection. The oligomerization state induced by pH as seemingly observed by SAXS.

Having done a fair amount of X–ray and neutron scattering, I am somehow puzzled by the experimental buffer conditions chosen by the authors (the same applies to the study by Jacot et al. that I looked at). There is no salt present, even under the relatively low protein concentration described here it is reasonable to think that they could be non–specific interactions between molecules.

I would temper the interpretation of the data at pH 5 with that so–called oligomerization that could be aggregation in disguise.

The buffer choice came out of numerous screens for GAC stability, homogeneous behaviour in size exclusion chromatography, reproducibility in crystallisation and we concluded that high pH, low or no salt and reducing agent is where GAC behaved best. Agreed the low pH behaviour this could be an in vitro artifact of aggregation, so the discussion has been toned down.

Reviewer #3 (Recommendations for the authors):

There are several areas in which the manuscript could be improved.

– A schematic diagram placing the GAC in the context of the parasite would be a valuable addition to figure 1, as would including the sequence diagram from Figure 7a in Figure 1.

This is now included.

– More information is required to judge the reproducibility of binding data in the relevant figure panels. How many times were binding experiments repeated? Do repeats represent technical repeats with the same protein preparation or different preparations? Do error bars represent SEM? Please add this information to relevant figure legends.

These are now included in caption and methods.

– Figure S1. It is difficult to judge concordance between experimental NOE data and NOE count from the AlphaFold model. Can a chart of experimental NOEs be shown for comparison? Panel E requires colour key.

We took the opportunity to calculate an NMR structure from our available NOESY data and made a comparison with the Alphafold model. Figure S1 has been modified to show the calculated NMR ensemble and illustrate the comparison (paragraph from line 91)

– Sequence conservation of the N/C interface is mentioned in the text several times, but it would be useful to show this plotted on the structure.

We now include a surface representation of the interface with Consurf conservation scoring in Figure 2b and a full sequence alignment in Figure 1—figure supplement 2.

– The AlphaFold prediction of the interface between GAC and MIC2 is interesting but not convincing in its current form. A PAE plot is required to judge the confidence of the GAC–MIC2 interface. pLDDT scores (ideally plotted on the structure) are required to judge confidence in the conformation of the MIC2 peptide.

The confidence scores are reasonable for the MIC2 peptide however due to it speculative nature we now removed Figure 8 as suggested, but keep the basic comparison of the C-terminal ARM arch of GAC with the catenins and display this in Figure 1—figure supplement 3.

– Does AlphaFold predict TgGAC to be fully closed, as in crystal structure, or partially separated, as in the PfGAC prediction?

We have calculated an Alphafold model for TgGAC and like PfGAC the N/C interface is also partially separated. This is now shown together in Figure 2.

– p9. The section titled "Targeted mutations in GAC PH domain are not fitness conferring in vivo" is confusing. Key lipid–binding residues in the GAC PH domain are not essential for parasite invasion.

The title has been reworded as suggested.

– Figure 8b, middle panel. The basis for how the GAC is placed with respect to actin is unclear – it would seem that there are many other ways it could be placed that are equally compatible with the data.

See above but the figure has removed.

https://doi.org/10.7554/eLife.86049.sa2

Article and author information

Author details

  1. Amit Kumar

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Formal analysis, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  2. Oscar Vadas

    Department of Microbiology and Molecular Medicine, Faculty of Medicine, University of Geneva, Geneva, Switzerland
    Contribution
    Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3511-6479
  3. Nicolas Dos Santos Pacheco

    Department of Microbiology and Molecular Medicine, Faculty of Medicine, University of Geneva, Geneva, Switzerland
    Contribution
    Formal analysis, Investigation, Visualization, Methodology, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1959-194X
  4. Xu Zhang

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  5. Kin Chao

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Formal analysis, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  6. Nicolas Darvill

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Investigation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  7. Helena Ø Rasmussen

    Interdisciplinary Nanoscience Center (iNANO) and Department of Chemistry, Aarhus University, Aarhus, Denmark
    Contribution
    Formal analysis, Investigation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8384-656X
  8. Yingqi Xu

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Investigation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  9. Gloria Meng-Hsuan Lin

    Department of Microbiology and Molecular Medicine, Faculty of Medicine, University of Geneva, Geneva, Switzerland
    Contribution
    Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  10. Fisentzos A Stylianou

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  11. Jan Skov Pedersen

    Interdisciplinary Nanoscience Center (iNANO) and Department of Chemistry, Aarhus University, Aarhus, Denmark
    Contribution
    Formal analysis, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  12. Sarah L Rouse

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Formal analysis, Investigation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  13. Marc L Morgan

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Formal analysis, Investigation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  14. Dominique Soldati-Favre

    Department of Microbiology and Molecular Medicine, Faculty of Medicine, University of Geneva, Geneva, Switzerland
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    Dominique.Soldati-Favre@unige.ch
    Competing interests
    Senior editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4156-2109
  15. Stephen Matthews

    Department of Life Sciences, Imperial College London, London, United Kingdom
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    s.j.matthews@imperial.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0676-0927

Funding

Leverhulme Trust (RPG_2018_107)

  • Stephen Matthews

Biotechnology and Biological Sciences Research Council (BB/W001764/1)

  • Stephen Matthews

Swiss Re Foundation (10030_185325)

  • Dominique Soldati-Favre

Swiss Re Foundation (CRSII5_198545)

  • Dominique Soldati-Favre

Biotechnology and Biological Sciences Research Council (BB/D524840/1)

  • Stephen Matthews

Wellcome Trust (202926/Z/16/Z)

  • Stephen Matthews

Independent Research Fund Denmark (8021-00133B)

  • Stephen Matthews

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. For the purpose of Open Access, the authors have applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.

Acknowledgements

This work was supported by a BBSRC and Leverhulme Trust awards to SJM (BB/W001764/1 and RPG_2018_107), the Independent Research Fund Denmark through grants 8021-00133B (JSP), and the Swiss National Science Foundation to DSF (10030_185325 and CRSII5_198545). The crystallisation facility at Imperial College London is supported by the Biotechnology and Biological Sciences Research Council (BB/D524840/1) and Wellcome Trust (202926/Z/16/Z). We are grateful to staff at Diamond Light Source beamline I04 for their help with data collection. We are thankful to Rémy Visentin at the Protein Platform of the University of Geneva for assistance with recombinant protein purification.

Senior Editor

  1. Amy H Andreotti, Iowa State University, United States

Reviewing Editor

  1. Olivier Silvie, Sorbonne Université, UPMC Univ Paris 06, INSERM, CNRS, France

Version history

  1. Received: January 10, 2023
  2. Preprint posted: January 23, 2023 (view preprint)
  3. Accepted: April 3, 2023
  4. Accepted Manuscript published: April 4, 2023 (version 1)
  5. Version of Record published: April 24, 2023 (version 2)

Copyright

© 2023, Kumar et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,205
    Page views
  • 189
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Amit Kumar
  2. Oscar Vadas
  3. Nicolas Dos Santos Pacheco
  4. Xu Zhang
  5. Kin Chao
  6. Nicolas Darvill
  7. Helena Ø Rasmussen
  8. Yingqi Xu
  9. Gloria Meng-Hsuan Lin
  10. Fisentzos A Stylianou
  11. Jan Skov Pedersen
  12. Sarah L Rouse
  13. Marc L Morgan
  14. Dominique Soldati-Favre
  15. Stephen Matthews
(2023)
Structural and regulatory insights into the glideosome-associated connector from Toxoplasma gondii
eLife 12:e86049.
https://doi.org/10.7554/eLife.86049

Further reading

    1. Microbiology and Infectious Disease
    Qibin Geng, Yushun Wan ... Fang Li
    Research Article

    SARS-CoV-2 spike protein plays a key role in mediating viral entry and inducing host immune responses. It can adopt either an open or closed conformation based on the position of its receptor-binding domain (RBD). It is yet unclear what cause these conformational changes or how they influence the spike's functions. Here we show that Lys417 in the RBD plays dual roles in the spike's structure: it stabilizes the closed conformation of the trimeric spike by mediating inter-spike-subunit interactions; it also directly interacts with ACE2 receptor. Hence, a K417V mutation has opposing effects on the spike's function: it opens up the spike for better ACE2 binding while weakening the RBD's direct binding to ACE2. The net outcomes of this mutation are to allow the spike to bind ACE2 with higher probability, mediate viral entry more efficiently, but become more exposed to neutralizing antibodies. Given that residue 417 has been a viral mutational hotspot, SARS-CoV-2 may have been evolving to strike a balance between infection potency and immune evasion, contributing to its pandemic spread.

    1. Biochemistry and Chemical Biology
    2. Microbiology and Infectious Disease
    Rui-Qiu Yang, Yong-Hong Chen ... Cheng-Gang Zou
    Research Article

    An imbalance of the gut microbiota, termed dysbiosis, has a substantial impact on host physiology. However, the mechanism by which host deals with gut dysbiosis to maintain fitness remains largely unknown. In Caenorhabditis elegans, Escherichia coli, which is its bacterial diet, proliferates in its intestinal lumen during aging. Here, we demonstrate that progressive intestinal proliferation of E. coli activates the transcription factor DAF-16, which is required for maintenance of longevity and organismal fitness in worms with age. DAF-16 up-regulates two lysozymes lys-7 and lys-8, thus limiting the bacterial accumulation in the gut of worms during aging. During dysbiosis, the levels of indole produced by E. coli are increased in worms. Indole is involved in the activation of DAF-16 by TRPA-1 in neurons of worms. Our finding demonstrates that indole functions as a microbial signal of gut dysbiosis to promote fitness of the host.