Molecular determinants of species-specific gamete adhesion triggering HAP2-mediated membrane fusion

Abstract

Species-specific adhesion of gametes is a prerequisite to cell fusion during fertilization, yet the underlying molecular mechanisms are not fully understood for any organism. In Chlamydomonas, the adhesion protein MAR1 chaperones the conserved fusogen HAP2 into a pre-fusion conformation on the mt− gamete and is the receptor for the adhesion protein FUS1 on the mt+ gamete. Here, we report the X-ray structure of the FUS1/MAR1 adhesion complex, showing that it forms a (FUS1/MAR1)₂ dimer. We found that mutagenesis at the observed MAR1-FUS1 interface strongly impairs adhesion, and that recombinant, soluble MAR1 binds to FUS1 on activated mt+ gametes and completely blocks HAP2 triggering for fusion. These findings uncover the molecular architecture of a cell surface receptor complex whose formation is required to trigger the membrane-fusogenic conformational change of a eukaryotic class II membrane fusion protein.

The molecular mechanisms linking gamete adhesion to membrane fusion are unknown. Here, the authors report the crystal structure of the complex between Chlamydomonas gamete adhesion proteins MAR1 and FUS1 and show that complex formation triggers fusogenic reconfiguration of conserved gamete fusion protein HAP2.

Introduction

Fusion of haploid gametes to form a diploid zygote arose early in the evolution of eukaryotes. The gamete membrane fusion reaction depends on the seamless integration of two functions—species-specific attachment of the gamete membranes and fusion of their lipid bilayers. Despite over 175 years of research on fertilization^1–3, the mechanisms underlying these concerted adhesion and fusion events remain poorly understood⁴. In the sperm of vertebrates, the membrane recognition and adhesion machinery is a conserved trimeric complex^5–7, which interacts with divergent egg proteins—JUNO in mammals and Bouncer in fish^5,8–12. A full understanding of the gamete fusion reaction in these species has been hampered because the ultimate actor in the fusion reaction, the membrane fusogen, is unknown.

The only organism for which the adhesion proteins on both gametes and the fusion protein have been identified is the unicellular, bi-ciliated, unicellular green alga Chlamydomonas reinhardtii (hereafter, Chlamydomonas)^13–16. The fusion protein is HAP2 (also termed GCS1¹⁷), a single-pass transmembrane protein conserved across the deep-rooted branches of the eukaryotic tree^14,18,19, including in plants^17,20, pathogenic protozoa, arthropods, and hemichordates. During evolution, certain sub-branches—nematodes, fungi, and chordates—appear to have lost or exchanged the HAP2 gene and lack an identified bona fide gamete fusogen^{18,19,21–24}. As in Chlamydomonas, disruption of HAP2 in the malaria agent, Plasmodium, had no effect on sperm-egg adhesion, but abrogated membrane bilayer merger and zygote formation^14,18. Moreover, antibodies against Plasmodium HAP2 blocked mosquito transmission of malaria^25–28, establishing HAP2 as a candidate for a transmission-blocking malaria vaccine.

When Chlamydomonas mating type plus and mating type minus (mt+ and mt − ) gametes are mixed to initiate fertilization, they initially interact through their cilia. Ciliary adhesion activates release of their cell walls, and both types of gametes erect from the plasma membrane in between their cilia a membrane-enclosed, tubular organelle, the mt+ and mt− mating structures²⁹. The tip of the mt+ mating structure displays the adhesion protein FUS1¹⁵, and the tip of the mt− mating structure, in addition to carrying HAP2, also includes the FUS1 binding partner, MAR1¹³. Continuous motility of the cilia-tethered gametes thrusts their mating structures into contact³⁰, enabling their adhesion through direct binding of FUS1 to MAR1¹³. HAP2 then drives fusion of the opposed lipid bilayers, which is followed almost immediately by cell coalescence into a diploid zygote. Disruption of FUS1 in mt+ gametes or of MAR1 in mt− gametes prevents mating structure adhesion and gamete fusion^13,15,16,30. Disruption of HAP2 in mt− gametes has no effect on membrane adhesion but abrogates fusion^13,14,30. Notably, MAR1 is positioned at the nexus of membrane adhesion and bilayer merger. In addition to being the receptor for FUS1, MAR1 is a chaperone for HAP2. In mt− gametes, MAR1 and HAP2 co-immunoprecipitate and, in the absence of MAR1, HAP2 expression is impaired, and the protein fails to localize properly¹³.

HAP2 is a class II membrane fusion protein (MFP)^22,31,32 related to the well-studied viral class II MFPs of yellow fever^33,34, rubella^35,36, and chikungunya³⁷ viruses, among many others. Structural orthologs were also identified in archaea³⁸ and in the FF family of fusion proteins in nematodes³⁹. The biogenesis of viral MFPs involves co-translational heterodimerization with an accompanying protein (AP), which chaperones them into adopting a metastable, acid-sensitive, pre-fusion conformation in the viral envelope⁴⁰. The viral particles enter the cell by receptor-mediated endocytosis, where the acidic environment of the endosome induces MFP/AP heterodimer dissociation, triggering a conformational change of the MFP that drives membrane fusion through irreversible trimerization^40–42. This final, post-fusion trimer is the one observed in the initial structural studies of recombinant HAP2 in several organisms^22,43–45, although a monomeric, presumably pre-fusion form of HAP2, was subsequently reported⁴⁶. In vivo, Chlamydomonas HAP2 trimer formation is not triggered by acidic pH but by a FUS1-dependent contact between the plasma membranes of the gametes³⁰. Thus, characterizing the MAR1–FUS1 recognition mode of this concerted membrane adhesion and bilayer merger reaction is central to understanding the regulation of HAP2-driven membrane fusion.

FUS1 is predicted to display seven consecutive immunoglobulin-like (IgL) domains in its ectodomain (Fig. 1a)¹³ and is an ortholog of members of the GEX2 family^13,47–49 of plant sperm adhesion proteins, for which no binding partner has yet been identified. MAR1, which lacks identified homologs outside of green algae, features a long, intrinsically disordered cytosolic tail and an ectodomain composed of two main regions: an N-terminal, membrane-distal, cysteine-rich region (CRR) and a membrane-proximal, extended proline-rich region (PRR, Fig. 1a). Here, using biochemistry and X-ray crystallography together with in vitro and in vivo mutagenesis experiments, we characterize the recognition mode of MAR1 and FUS1 and report the structural basis of their interaction. The structure of this receptor complex provides a unique entry point for understanding the mechanisms that couple gamete membrane adhesion interactions to the triggering of the HAP2 membrane-fusogenic conformational change, a step not elucidated for any non-viral system to date.

Fig. 1. The MAR1/FUS1 complex.

a schematic representation of HAP2, MAR1 and FUS1 on the membranes of Chlamydomonas mt− and mt+ gametes. Proposed interactions are marked by black arrows. b SEC profile of the complex extracted by tag-affinity chromatography from the supernatant of insect cells co-transfected with the MAR1 CRR and IgL1-4 (orange line) or of IgL1-4 alone (green line). Inset (right): SDS-PAGE analysis of the elution profile (left). Representative profile from two independent experiments. c Crystal structure showing a (MAR1/FUS1)₂ dimer with protomer A displayed as ribbons. Protomer B is displayed as a surface with FUS1 in light gray and MAR1 in pale green. N-linked glycans are shown as sticks and labeled. The inset (left) schematizes a proposed model of the interaction across gametes, with dimerization induced by mutual recognition. d Orthogonal view of protomer A (rotated as indicated) of (c), highlighting the intra-protomer MAR1–FUS1 interactions. Inset (below): the groove in FUS1 into which the CRR hook bends and point insert. e Top view of the complex, with FUS1 shown in white surface and the twofold axis (white oval at the center - labeled). MAR1 CRR is shown as ribbons, ramp-colored from N- to C-terminus. Schematic representations in panels a and c are adapted from an image created in BioRender. BAQUERO SALAZAR, E. (2026) https://BioRender.com/7n0d37u.

Results

Structure determination

Heterologously expressed MAR1 and FUS1 ectodomains (see “Methods”) were secreted at high levels into the supernatants of transfected eukaryotic cells. We found that co-transfection with the MAR1 CRR (Fig. 1a) with either IgL1-7 or IgL1-4 constructs counteracted the intrinsic aggregation propensity of FUS1 observed when expressed alone, allowing purification of the resulting complex by size-exclusion chromatography (Fig. 1b). These data indicated that the direct interaction between the two proteins takes place via their N-terminal, membrane-distal halves. Obtaining diffraction-quality crystals demanded ablation of several MAR1 N-glycosylation sequons (see “Methods”). The resulting crystals of the MAR1 CRR/FUS1 IgL1-4 complex allowed the determination of the X-ray structure to 3.1 Å resolution (Supplementary Table 1 and Supplementary Fig. 1a), revealing a twofold-related (FUS1/MAR1)₂ dimer. Each dimer protomer (A and B in Fig. 1c) features one FUS1 and one MAR1 chain associated tightly. The folding topology of both proteins and the residues involved in intra- and inter-protomer contacts in the dimer are mapped along the primary sequence in Supplementary Figs. 1 and 2.

FUS1 organization

Each FUS1 IgL domain adopts a β-sandwich fold with a 4-stranded “ABED” sheet opposing a 3-stranded “CFG” sheet (Fig. 1c; Supplementary Fig.1b), related to “set C1” of constant domain immunoglobulins^50,51. FUS1 IgL1 contains two disulfide bonds through cysteines conserved in FUS1/GEX2 proteins in other Chlorophytes; the other IgLs lack disulfide bonds (Supplementary Fig. 2). Each IgL domain features three “N-loops” (connecting β-strands B to C, D to E and F to G) and three “C-loops” (connecting β-strands A to B, C to D and E to F), named for their proximity to the domain’s N- or C-termini, respectively (Fig. 1c). These loops are substantially elaborated, displaying additional secondary structure elements. In the FUS1 rod-like molecule, IgL1 exposes its N-loops at the membrane-distal end, while its C-loops interact with the N-loops of IgL2. Similarly, the IgL2 C-loops engage with the N-loops of IgL3, making a rigid IgL1-3 rod through stable interdomain interactions, with each domain rotated by roughly 180° with respect to the previous one (Supplementary Fig. 2c). At the interface between IgL3 and IgL4, the interacting C- and N-loops are shorter (Fig. 1c), resulting in a looser contact that increases the mobility of IgL4.

As expected, Foldseek⁵² identified FUS1 homology to the plant GEX2s discussed above¹³ (E-values < 10e^-21 and aa sequence identities of ∼14–16%). Structural comparison of FUS1 with GEX2 atomic models predicted by AlphaFold⁵³ revealed that FUS1 IgLs 1-3 align with consecutive GEX2 IgL domains of several plant species (Supplementary Fig. 2a). Foldseek also identified FUS1 homology to the IgL domains of filamins B and C (E-values < 1- e-17 and 15% amino acid sequence identity) and the vascular endothelial growth factor receptor (VEGFR)-related protein, VER1, from Caenorhabditis elegans (E-value: 6.06e-7, 10% aa sequence identity). VER1 is also composed of seven IgL domains and has a tyrosine kinase domain in its intracellular region⁵⁴. We note that VER1 has been proposed to work together with the class II MFP AFF1³⁹ in the fusion of glial cells during neural remodeling in C. elegans⁵⁵.

MAR1 organization

The MAR1 CRR has an overall novel fold, with two α-helices and four 2-stranded β-sheets stabilized by seven disulfide bonds (Supplementary Figs. 2b and 3a). PSI-blast identified high-confidence homologs across green algae only, which exhibited considerable insertions or deletions (Supplementary Fig. 2b), yet conserved the same structural elements and the same pattern of disulfide bonds. Foldseek identified weak similarity between the MAR1 CRR C-terminal half (residues 109–162) and the epidermal growth factor-like (EGF-like) repeats of the extracellular matrix protein, Fibulin-1 (E-value ∼10e-1; 22–26% amino acid sequence identity; Supplementary Fig. 3a).

The FUS1/MAR1 complex

We assigned the protomers of the (FUS1/MAR1)₂ dimer based on the large MAR1–FUS1 interface area (1300 Ų). These protomers associate to make a dimer with an interface area of 1200 Ų, of which only 214 Ų and 100 Ų correspond to homotypic MAR1–MAR1 and FUS1–FUS1 contacts, respectively; the rest corresponds to inter-protomer cross-interactions between MAR1 and FUS1 (Supplementary Fig. 3b). Inherent in this overall organization of the complex is that the observed dimer can form only upon FUS1–MAR1 recognition. The MAR1 CRR has the shape of a hook, with the region spanning from the bend to the point inserting within a deep groove in FUS1 to form the protomer (Fig. 1d and Supplementary Movie 1). In the (FUS1/MAR1)₂ dimer, the MAR1 CRRs pack in an antiparallel fashion, with the thicker N-terminal hook “shank” of one pairing with the point of the other (Fig. 1c, e). In this arrangement, the gape of the CRR hook embraces FUS1 within the protomer, while the shank packs parallel to FUS1 of the adjacent protomer.

In the dominant intra-protomer FUS1–MAR1 interaction, a deep groove in FUS1 IgL2, with the ABED β-sheet as the floor and the IgL1 EF C-loop and IgL3 BC N-loop as the walls (Fig. 1d and Supplementary Fig. 1b), accommodates the bend-point region of the MAR1 hook. A key feature here is a barb at the hook’s bend made by the M141-P142 motif, where the methionine side chain and the proline ring insert into a cage of aromatic side chains of the groove floor, which could provide high recognition specificity (Figs. 1d, inset and 2a). The shank interacts with the adjacent protomer by contacting FUS1 IgL1 C-loops CD and EF as well as the CFG β-sheet (Fig. 1e and Supplementary Fig. 2), while the inter-protomer MAR1–MAR1 contacts involve the bend region (Fig. 1e). A local twofold molecular axis relating the two protomers (A and B in Fig. 1c, e) runs between the bends of the MAR1 hooks in the dimer (labeled in Figs. 1e, 2c). Complex formation involves a total of 11 hydrogen bonding interactions (10 of which are intra-protomer) between FUS1 side chains, mostly interacting with the main chain of the MAR1 hook point (Supplementary Fig. 1c, d).

Fig. 2. Soluble MAR1 rCRR blocks adhesion and gamete fusion.

a The P142 ring and M141 side chain insert between aromatic side chains of the FUS1 IgL2 ABED β-sheet floor. Dashed lines indicate the hydrogen bonds. b WT rCRR and mutants A111D and T133Q but not M141Q or P142Q co-elute with recombinant FUS1 upon co-expression. Representative immunoblot from two independent experiments. c WT rCRR and rCRR M141Q binding to FUS1 on mt+ gametes. Representative composite confocal immunofluorescence and DIC images show WT rCRR binding to activated WT mt+ but not to fus1 mt+ gametes (upper panel). rCRR M141Q binding (lower panel) is detectable, but on fewer cells and binding is of lower intensity (see Supplementary Fig. 4a). Scale bars are 10 μm. Images shown in the top and bottom panels are representative results from two independent IF experiments, respectively. d, e Gamete adhesion and fusion are blocked by WT rCRR, but not by hook barb mutants. After incubation of the indicated rCRR proteins with activated mt+ gametes, mt− gametes were added that were competent for mating structure adhesion only (hap2(−) gametes); d or for adhesion and fusion (MAR1-FLAG;mar1(−)); e and the samples were assessed for pair formation (d, lower panel) or zygote formation (e, lower panel). Representative phase-contrast images of fixed samples (d top panels, arrowheads indicate pairs; e, top panels, arrowheads indicate quadriciliated zygotes). Scale bars in all images are 10 µm. Box-and-whisker plot quantification (d, e, bottom panels) of rCRR- or buffer-treated control cell mixtures with individual biological replicates (n) shown as open circles (for sample groups on x-axis left to right; d, n = 16, 4, 4, 4, 4; e, n = 16, 4, 4, 4, 4). Box-and-whisker plot box bounds are 25th to 75th percentiles, the midline is the median, and whiskers are the data minimum and maximum values. Statistical significance was assessed using a Kruskal–Wallis test with a Dunn’s multiple comparison post-test: ∗P < 0.05, ∗∗P < 0.01, ns = not significant. P values for each comparison as well as the values for the medians, whiskers, and percentile bounds of each box & whisker plot are provided in the Source Data file.

Glycan substitutions

We observed density for the first—and in some instances the second—N-acetyl glucosamine sugar residues attached to FUS1 N41 in IgL1, N221 in IgL2, N317 and N343 in IgL3, and N387 in IgL4 and to MAR1 N127. Importantly, the glycan chain of MAR1 N127 faces MAR1 residue 56, which is an asparagine in the endogenous protein, but was mutated to eliminate glycosylation for crystallization (Supplementary Fig. 2b, c). Their proximity suggests that in the endogenous protein, the glycans attached to N127 and N56 could interact across protomers. Such a glycan–glycan interaction has the potential to stabilize the complex by lowering its dissociation rate.

Site-directed mutagenesis

The structure suggested the intra-protomer MAR1–FUS1 interaction is critical, especially the contacts mediated by the M141-P142 hook barb (Figs. 1d, 2a). We mutated these residues, together with MAR1 T133 at the intra-protomer interface (Supplementary Fig. 1c), as well as A111 at the inter-protomer MAR1–MAR1 interaction near the twofold axis (Fig. 1e, inset and Supplementary Fig. 2c). Immunoblotting of affinity purified wild-type (WT) and mutant forms of strep-tagged recombinant CRR (rCRR, MAR1 residues 26–161) from the supernatants of transfected cells co-expressing His-tagged recombinant FUS1 IgL1-4 (rFUS1) revealed that, like the WT rCRR (Fig. 1b), the A111D and T133Q rCRR mutants interacted with FUS1, counteracting its tendency to aggregate, but the barb mutants failed to interact with FUS1, and FUS1 was not immunoprecipitated (Fig. 2b). These in vitro results confirm the importance of the 141-MP-142 insertion into the aromatic pocket of FUS1 for complex formation.

WT MAR1 rCRR blocks mating structure adhesion and gamete fusion

Immunostaining showed that WT rCRR bound to the mating structures of activated mt+ gametes, but not to those of activated fus1 mt+ gametes, which lack the FUS1 protein (Fig. 2c, upper panels). M141Q rCRR bound only weakly to mt+ gametes (Fig. 2c, lower panels; Supplementary Fig. 4a). As expected, interactions between the cilia of mt+ and mt− gametes were unaffected by WT rCRR (Supplementary Fig. 4b). We examined the effects of WT and mutant forms of rCRR on mating structure adhesion by incubating activated mt+ gametes with each rCRR, followed by mixing with hap2 mt− gametes. Using fusion-defective mt− gametes enabled us to assess mating structure adhesion in the absence of fusion. Mating structure adhesion was blocked nearly completely by WT rCRR but not by rCRR M141Q (Fig. 2d, upper panel). Quantification showed that over 50% of gametes in the buffer control samples underwent mating structure adhesion, and that WT rCRR reduced mating structure adhesion tenfold, to less than 5% (Fig. 2d, lower panel). This effect was dose-dependent, with inhibition evident at 1.4 µM (Supplementary Fig. 4c). Consistent with the suboptimal interaction of rCRR M141Q with FUS1 found by immunostaining, rCRR M141 and rCRR P142Q failed to block mating structure adhesion (Fig. 2d). rCRRs with mutations near the barb, rCRR A145E and rCRR A145Q, were also ineffective (Supplementary Fig. 4e). As expected, mutations away from the barb had the opposite effect, with rCRR A111D and N56G blocking adhesion as effectively as WT rCRR (Fig. 2d, lower panel and Supplementary Fig. 4e), with rCRRs A111C, A111W, T133E, and T133Q being nearly as effective (Supplementary Fig. 4e). These findings mirrored the results obtained in vitro (Fig. 2b) and confirm the central function of the rCRR barb residues MP141-142 in adhesion between the mating structures of mt+ and mt− gametes.

In vivo experiments mixing mt+ gametes with fusion-competent WT mt− gametes showed that the strong inhibition of mating structure adhesion by 8.6 µM WT rCRR was accompanied by a nearly complete block to gamete fusion (Fig. 2e, upper and lower panels). 80% of cells in the control mixtures of gametes and in mixtures containing rCRR M141Q or rCRR P142Q fused to form zygotes, but less than 1% of the cells in samples containing WT rCRR or rCRR A111D formed zygotes (Fig. 2e, lower panel). As with mating structure adhesion, fusion inhibition by WT rCRR was dose-dependent (Supplementary Fig. 4d). Moreover, rCRRs with other mutations external to the MAR1–FUS1 interface also blocked fusion (Supplementary Fig. 4f), whereas those near the interface, rCRR A145E and A145Q, had little if any effect on fusion (Supplementary Fig. 4f).

MAR1 barb mutations impair adhesion of mating structures in vivo

We used mar1;hap2 mt− Chlamydomonas cell lines, which express neither MAR1 nor HAP2^13,56,57, to reintroduce two FLAG-tagged versions of MAR1—a wild-type MAR1-FLAG, or a version of MAR1 with the two amino acid residues of the hook barb mutated to glutamines—MAR1-MP141-142QQ-FLAG. Immunoblot analyses of gametes expressing the WT and mutant proteins showed that they were expressed at the same level (Fig. 3a). In addition, like the WT protein, MAR1-MP141-142QQ-FLAG was susceptible to degradation by incubation with pronase (Fig. 3b), indicating that it was present at the cell surface. Immunostaining with anti-FLAG antibodies confirmed that MAR1-MP141-142QQ-FLAG was correctly localized between the bases of the two cilia, the site of the mt− mating structure (Fig. 3c). Importantly, immunostaining demonstrated that HAP2 was also properly localized at the mating structure (Supplementary Fig. 5a).

Fig. 3. The MAR1 MP141-142QQ mutation does not affect MAR1 expression but inhibits mating structure adhesion and reduces gamete fusion.

a Representative immunoblots of lysates of mt− gametes expressing MAR1-FLAG or MAR1 MP141-142QQ-FLAG and of mt+ gametes. b Sensitivity of WT MAR1-FLAG and MAR1 MP141-142QQ-FLAG protein to pronase treatment of live gametes. Tubulin was used as a loading control in (a, b). c Representative composite confocal immunofluorescence and DIC images show localization of MAR1-FLAG protein (red) at the mating structures of WT and MP141-142QQ-FLAG mt− gametes, but not in mar1;hap2 mt− gametes. d Mating structure adhesion is substantially reduced in mixtures of WT mt+ gametes with MAR1 MP141-142QQ-FLAG mt− gametes that lack HAP2. e Fusion is slightly reduced in mixtures of WT mt+ gametes with MAR1 MP141-142QQ-FLAG mt− gametes expressing HAP2. Representative SEM images of an adhering pair of mt+ and mt− gametes and a zygote, respectively, are shown to the right of plots of pair formation (d) and zygote formation (e). Individual biological replicates (n) are shown as open circles (for sample groups on x-axis left to right; d n = 7, 7, 8, 28; e n = 4, 7, 16). Box-and-whisker plot box bounds are 25th to 75th percentiles, the midline is the median, and whiskers are the data minimum and maximum values. Statistical significance was assessed using a Kruskal–Wallis test with Dunn’s multiple comparisons test: ^∗P < 0.05, ^∗∗∗P < 0.001; ns = not significant. P values for each comparison, as well as the values for the medians, whiskers, and percentile bounds of each box & whisker plot, are provided in the Source Data file. Representative blots and images shown are the results of two to four independent experiments. Scale bars are 10 µm.

To assess mating structure adhesion, we mixed fusion-defective mar1;hap2 mt− gametes expressing WT MAR1-FLAG or MAR1-MP141-142QQ-FLAG with mt+ gametes and after 10 m, stopped the reaction by addition of glutaraldehyde and quantified the percentage of cells adhering by their mating structures. In the mixtures of mt+ gametes with hap2 mt− gametes, over 40% of the gametes adhered by their mating structures (Fig. 3d, left panel; right panel shows a scanning electron micrograph [SEM] of adhering gametes), whereas, as shown previously¹³, in the mixtures of mt+ gametes with mt− gametes lacking MAR1 (mar1;hap2), mating structure adhesion failed to occur. We found that the mar1;hap2 mt− gametes expressing WT MAR1-FLAG underwent mating structure adhesion similar to control levels, while those expressing MAR1-MP141-142QQ-FLAG exhibited a 3-fold reduction in mating structure adhesion (Fig. 3d). Thus, although these results do not rule out a role for additional regions of MAR1, including the PRR, in mating structure adhesion, they demonstrate that residues MP141-142 within the CRR have a critical function in the FUS1–MAR1 interaction.

Gamete fusion proceeds despite impaired mating structure adhesion

To measure the effect of the barb mutation on gamete fusion, we used mar1 mt− gametes, whose HAP2 protein is functional. As expected¹³, in mixtures of mt+ gametes with mar1 mt− gametes, because mating structure adhesion is abrogated, zygote formation failed to occur, whereas in the mixtures with mar1 mt− gametes expressing WT MAR1-FLAG, ~75% of gametes formed zygotes (Fig. 3e, left panel; right panel shows a SEM of a quadriciliated zygote). Unexpectedly, even though the experiments with fusion-defective mt− gametes showed a threefold reduction in adhesion (Fig. 3d), in mar1 mt− gametes expressing MAR1-MP141-142QQ-FLAG, fusion was reduced by less than 30% compared to WT (Fig. 3e). These results highlight the capacity of the HAP2-mediated fusion reaction to occur under conditions of suboptimal membrane adhesion.

MAR1–FUS1 binding triggers the HAP2 conformational change

We used the observation that MAR1 rCRR blocks gamete fusion to follow the triggering of HAP2 to reach its detergent-resistant, post-fusion form. In the context of wild-type HAP2, this conformational change drives membrane fusion by forcing the two membranes against each other. Because less than 10% of wild-type HAP2 forms trimers during gamete fusion³⁰, for these experiments, we used mt− gametes expressing the fh1 HAP2 mutant, with W173A and F177A mutations in the fusion loop, rendering them unable to efficiently anchor their HAP2 proteins in the membrane of the mt+ mating structure. Accordingly, fh1-HAP2 undergoes the irreversible conformational change to form post-fusion trimers without coupling it to membrane fusion. Trimers of fh1-HAP2 were shown to accumulate to high levels in conditions of mating structure adhesion as fusion fails to proceed³⁰, facilitating HAP2 trimer detection. As expected, HAP2 fh1 mt− gametes mixed with untreated mt+ gametes or with mt+ gametes incubated with rCRR-M141Q formed large amounts of SDS-resistant HAP2 trimers. On the other hand, those mixed with mt+ gametes treated with WT rCRR failed to form trimers (Fig. 4a). Trimer formation was reduced but not completely blocked by rCRR A145Q; but like WT rCRR shown in Fig. 4a, rCRR N56G fully stopped trimer formation (Supplementary Fig. 5b). We therefore interpret the trigger of the HAP2 conformational change to be MAR1 binding to FUS1, an event that releases HAP2 to transit into its post-fusion conformation (Fig. 4b).

Fig. 4. Soluble MAR1 rCRR interferes with the triggering of the HAP2 fusogenic conformational change.

a Detection of SDS-resistant HAP2 trimers by immunoblotting. Activated mt+ gametes co-incubated or not with WT rCRR or rCRR-M141Q were mixed with fh1 mt− gametes, which accumulate the HAP2 trimer of hairpins during gamete fusion, followed by immunoblot analysis of cell lysates for HAP2 trimer formation. Blots are representative of results from at least two independent experiments. b Proposed steps leading to gamete fusion. Step i: Pre-fusion form. HAP2 is represented as a rod-like molecule with domains (I, II, III), fusion loop (FL), and stem + transmembrane regions (magenta) labeled. We propose that MAR1 chaperones the metastable pre-fusion state of HAP2 (step i), and that binding to FUS1 releases HAP2 (step ii), to insert the FL into the target membrane, followed by trimerization as an extended intermediate (step iii) followed by hairpin formation to bring both membranes into apposition and hemi-fusion (step iv). Completion of the post-fusion hairpin trimer leads to the formation of a fusion pore (step v). The HAP2-HA fh1 mutant cannot insert efficiently into the mt+ gamete membrane (steps ii to v) and forms a trimer of hairpins (step v) that accumulates in the membrane of the mt− gamete, as detected in (a) as a detergent-resistant HAP2-HA trimer. Schematic representation in (b) is adapted from an image created in BioRender. BAQUERO SALAZAR, E. (2026) https://BioRender.com/7n0d37u.

Discussion

Gamete membrane adhesion and lipid bilayer merger are tightly coordinated during fertilization in unicellular and multicellular organisms. We provide here details of the molecular interactions in Chlamydomonas gamete adhesion. A crucial, direct interaction between the two mating structure adhesion proteins takes place via their N-terminal, membrane-distal halves. The crystal structure of their complex shows a dominant interaction between a groove in FUS1 and the bend-point region of the MAR1 hook. Our in vitro and in vivo experiments confirmed the functional importance of this interaction, showing that the barb motif M141-P142 of the CRR is critical for MAR1 recognition of FUS1. Additional downstream residues further stabilize this interaction, as evidenced by the list of hydrogen bonds formed between the two chains (Supplementary Fig. 1b, c) and by the observation that incubation of gametes with MAR1 rCRR A145Q, although permitting gamete adhesion and fusion, partially blocked trimer formation (Supplementary Figs. 4e, f and 5b). It was striking to observe that gamete fusion proceeds even with impaired membrane adhesion (Fig. 3d, e). Considering our previous results showing that post-fusion trimer formation by only 8% of surface-expressed HAP2 is sufficient to drive extensive fusion of wild-type gametes³⁰, and that there is low —but detectable—binding of rCRR M141Q to FUS1 on the mt+ mating structure (Fig. 2c and Supplementary Fig. 4a), our interpretation is that the impaired interaction between FUS1 and MAR1-MP141-142QQ supports only weak attachment between the membranes of the two mating structures (Fig. 3d). Yet this weak adhesion is sufficient to activate enough HAP2 molecules to drive membrane fusion and zygote formation (Fig. 3e).

In contrast to the results with cells expressing the mutant form of MAR1, we found that fusion was almost completely arrested when pretreating mt+ gametes with the WT recombinant MAR1 CRR domain. Our interpretation is that this blockade results from complete occupation of FUS1 on the mt+ gametes by the rCRR, precluding interaction with MAR1 on the mt− gamete and demonstrating that gamete recognition, specifically mediated by the MAR1–FUS1 interaction, indeed activates membrane fusion. In this context, FUS1 would displace MAR1 from its pre-fusion association with HAP2, freeing HAP2 to insert its fusion loop into the mt+ membrane while undergoing its energetically downhill fusion-driving conformational change (Fig. 4b).

The structurally related viral class II MFPs are maintained in an acid-sensitive metastable pre-fusion conformation by interaction with an accompanying protein, which dissociates from the MFP in the acidic environment of an endosome. A major difference in Chlamydomonas is that the pre-fusion complex at the mt− gamete surface is not acid sensitive³⁰, and therefore the trigger is different. We found that instead, the trigger follows a mechanism akin to that of paramyxoviruses, which fuse for entry at the plasma membrane. Although paramyxoviruses have a structurally unrelated MFP, fusion is triggered by cell receptor binding via an accompanying protein, weakly associated with the viral MFP^58–60, in a striking parallel to the Chlamydomonas HAP2/MAR1 weakly interacting pair and the recognition of FUS1 by MAR1 to form a complex across gametes. This accompanying protein mechanism prevents nonspecific membrane interactions and premature activation of the MFP due to untimely structural transitions.

We highlight that the close apposition of gamete membranes mediated by the MAR1–FUS1 interaction may also bring together additional factors, including lipids, that are required to interact with components of the fusion machinery and modulate the HAP2 conformational transition or the kinetics of the fusion reaction. An interesting possibility is that the (FUS1/MAR1)₂ dimers observed in the crystals, which can only form upon MAR1/FUS1 recognition, may have a biological relevance in the gamete fusion mechanism. Ligand-induced receptor dimerization is a widely used mechanism for activation of intracellular signaling pathways in plants⁶¹ and animals⁶². Although future studies are needed to understand the biological relevance of the (FUS1/MAR1)2 dimer and HAP2/MAR1 interactions, the presence in Chlamydomonas of multiple gamete-specific protein kinases⁶³, along with multiple predicted phosphorylation sites in the cytoplasmic domains of HAP2 and MAR1, raises the intriguing possibility that FUS1-induced MAR1 dimerization could provide an additional layer of regulation to gamete adhesion and fusion.

We conclude by noting that the mechanistic insight on Chlamydomonas cell–cell fusion derived from this study will serve as a paradigm for understanding related molecular mechanisms of gamete fusion in agriculturally significant plants and insects, as well as in the globally devastating unicellular pathogens Toxoplasma⁶⁴, Leishmania⁶⁵, Babesia⁶⁶, Cystoisospora⁶⁷, and Plasmodium¹⁴, all of which rely on HAP2 for fertilization.

Methods

Chlamydomonas cells and cell culture

Cells were grown vegetatively in liquid TAP media at 22 °C in a 13:11 light–dark cycle. Gametogenesis was induced by the transfer of cells to nitrogen-free medium, followed by overnight agitation on a shaker in the presence of continuous light. Chlamydomonas strains used in experiments were 21gr(+), fus1::FUS1-HA(+), hap2(−), hap2::HAP2-HA(−), mar1(−), mar1;hap2(−), mar1;hap2;MAR1-FLAG(−), and hap2::fh1-HAP2-HA mt. New strains generated for and used in this study were made by transformation of the aforementioned strains and include mar1::MAR1-FLAG(−), mar1::MAR1-FLAG-MP141-42QQ(−), and mar1;hap2::MAR1-FLAG-MP141-142QQ(−). Names denote strain origins and genotype; lowercase italics indicates the presence of a mutant gene; uppercase italics indicates the presence of the WT gene; double colons (::) indicate strains where transgene introduction was done through direct transformation by electroporation; semi-colons (;) indicate the presence of a transgene in a progeny strain that was introduced through genetic crosses of a previously transformed parental strain; and (−) and (+) indicate mating type.

Chlamydomonas transformation

Chlamydomonas cells were transformed by electroporation⁶⁸ using SbfI-linearized plasmid DNA encoding MAR1¹³ with or without the site-directed mutations indicated (Genscript) and a C-terminal 3× FLAG tag. A pre-existing MAR1 plasmid¹³ was provided to Genscript for site-directed mutagenesis (see Supplementary Data file for sequence). Briefly, mid-log phase vegetative cells were electroporated in the presence of DNA in a 0.4 cm cuvette using Bio-Rad Gene Pulser Xcell at 25 μF and 0.8 kV with no shunt resistor. Genscript mutant MAR1-FLAG plasmid DNAs used for transformation were confirmed by sequencing analysis. Positive transformants were selected for further analysis based on their growth on TAP-agar plates containing zeocin (10 μg/mL) and their expression of the FLAG-tagged MAR1 protein in gamete lysates.

Cloning, protein expression, and purification of recombinant rMAR1 and rFUS1

Codon-optimized DNA sequences encoding the full-length ectodomains of MAR1 (CRR + PRR) and FUS1 (IgL1-7) were synthesized by GenScript and cloned into a modified pMT/BiP plasmid containing a metallothionein-inducible promoter, in-frame with the BiP signal sequence. Shorter ectodomain variants including MAR1 CRR (residues 26–161) and FUS1 IgL 1-4 (residues 17–479) were generated using standard PCR methods. MAR1 constructs featured a C-terminal enterokinase site followed by a double strep-tag, while in FUS1 constructs, the strep-tag was replaced by a 6xHis-Tag.

To reduce glycosylation heterogeneity in the CRR, which significantly hindered crystallization of the complex, the N residues at positions 56, 64, 72, and 75 were mutated to Q. These mutations had no impact on protein yield or binding to FUS1 constructs. Point mutations at residues 56, 111, 133, 141, 142, and 145 were introduced into the WT CRR for pull-down assays to test MAR1–FUS1 interaction in solution. Mutations were introduced using standard PCR methods. All plasmid and primer sequences are listed in the associated Supplementary Data file.

Drosophila S2 cells (ATCC CRL-1963) were cultured at 28 °C in serum-free media (HyClone SFM4Insect, Cytiva) supplemented with 50 units/ml of penicillin and 50 units/ml of streptomycin. Cells were co-transfected with different combinations of MAR1 and FUS1 constructs along with the pCoPuro plasmid for puromycin selection of stable transfectants, using the Effectene transfection reagent (Qiagen) according to the manufacturer’s instructions. Cell cultures were further supplemented with 7 μg/ml of puromycin at 3 days post-transfection.

For pull-down assays, a FUS1 stable cell line was initially generated with puromycin selection. New cell lines were generated by transfecting this FUS1 cell line with equal amounts of plasmids encoding CRR WT or the mutants listed above along with pCoBlast plasmid for Blasticidin selection. Selection of stable transfectants was performed by supplementing cell culture medium with 7 μg/ml of Blasticidin at 3 d post-transfection.

For large-scale protein production, cell cultures were expanded in 3 L spinner flasks until reaching cell densities of ~1 × 10⁷ cells/ml. At this point, protein expression was induced by activating the metallothionein promoter with 4 μM CdCl₂. At 5 d post-induction, cells were sedimented, and the supernatant was concentrated to 50 ml on a VivaFlow ultrafiltration device (10 kDa MWCO, Sartorius). CRR- and FUS1-interacting constructs were then purified by affinity chromatography on a StrepTactin column (Cytiva), followed by size-exclusion chromatography on a Superdex 200 10 300 column pre-equilibrated with buffer containing 10 mM Tris-HCl pH 8.0 and 100 mM NaCl. Eluted fractions were analyzed by SDS-PAGE under reducing conditions, and those fractions containing CRR alone or FUS1 co-expressed with CRR constructs were concentrated and stored at −80 °C until use.

Crystallization and X-ray data collection

The complex between FUS1 IgL1-4 and the CRR glycosylation mutant was treated with EndoH glycosidase (New England Biolabs, NEB) for 24 h at room temperature, followed by enzymatic digestion of the C-terminal tags by incubation with enterokinase (NEB) for an additional 24 h at 4 °C. The protein solution was subsequently passed through Strep-tactin (Cytiva) and Ni-NTA (Cytiva) columns to retain the uncleaved protein complex. The column flow-through was concentrated and injected in a Superdex 200 10 300 column under the same conditions described above. The central fractions of the chromatographic peak corresponding to the protein complex were pooled and concentrated to 8 mg/ml. Crystals of the complex were obtained by the sitting drop method mixing 0.2 µl of protein solution with an equal volume of the crystallization solution containing 1.2 M potassium sodium tartrate tetrahydrate, 0.1 M Tris-HCl pH 8.5, and 1 µM Anderson-Evans polyoxotungstate (Jena Bioscience). Hexagonal salt crystals appeared at d 5 and served as nucleation agents for rectangular, thin plates that appeared 15 days after⁶⁹. These crystals were fished in crystallization solution supplemented with 15% v/v of glycerol and subsequently flash-frozen in liquid nitrogen. X-ray diffraction data were collected on the Proxima 2 beamline at the SOLEIL Synchrotron (Saint-Aubin, France). The data were reduced using the XDS-package and the structure was solved by molecular replacement with the Phaser Software using predicted models of MAR1 and FUS1 generated with Alphafold 2. The model was manually corrected in COOT and refined with phenix.refine.

SDS-PAGE and immunoblotting

For SDS-PAGE, 2.5 × 10⁷ gametes in 225 μL of N-free medium were lysed by mixing with 75 μL of 4× SDS-PAGE protein sample buffer (160 mM Tris, 4.0% SDS, 20.0% glycerol, 4 mM EDTA, and 0.0006% Bromophenol blue) with reducing agent (TCEP, 5 mM) 5 m at 95 °C for detection of MAR1-FLAG proteins, or without reducing agent for 5 m at 42 °C for detection of HAP2-HA homotrimers³⁰. For immunoblotting, 4.2 × 10⁶ cell equivalents were subjected to electrophoresis on SDS polyacrylamide tris-glycine gels (7%), transferred to PVDF membranes, blocked in 5% milk + TBST buffer, and probed with either rat anti-HA (1/1000) or mouse anti-FLAG (1/5000) primary antibodies, followed by either goat-anti-rat HRP or goat anti-mouse HRP (1/5000) secondary antibodies diluted in 0.5% milk + TBST buffer. Tubulin was used as a loading control and was detected by probing with mouse anti-α-tubulin primary antibodies (1/5000; Sigma, B512). Immunoblots were developed with SuperSignal West Femto (Thermo Scientific) and scanned with a LI-COR C-Digit blot scanner. Images were processed using LI-COR Image Studio software and Adobe Illustrator (V28.5). A white space present between lanes in some images indicates the cropping of an empty lane.

Pull-down assays

Protein expression was induced in stable cell lines co-expressing FUS1 IgL1-4 and WT or mutated CRR. 2 mL of cell culture supernatant was collected and treated with 3 μL of Biolock (IBA Lifesciences) per mL of medium. The supernatant was incubated with MagStrep type 3 XT beads (IBA Lifesciences) for 1 h at 4 °C with constant agitation. The beads were harvested by centrifugation and 3 washes with buffer containing 10 mM Tris (pH 8) and 100 mM NaCl for 30 m at 4 °C. Protein elution was performed using washing buffer containing 2.5 mM desthiobiotin. The eluted fraction was analyzed by SDS-PAGE and immunoblotting, using HRP-conjugated anti-6xHis-tag mouse antibodies (Proteintech) and anti-strep mouse antibody.

Protease treatments

Protease treatments to assess cell surface localization of WT and mutated MAR1-FLAG on live gametes were performed as described previously^13,70 with minor modifications. Briefly, 2.5 × 10⁷ live mt− gametes were incubated in the presence and absence of 0.05% pronase (1.0 mg/mL) for 20 m followed by three washes in cold N-free media containing 1 mM PMSF and protease inhibitor cocktail (Sigma) and 1% BSA (Sigma), then lysed in SDS-PAGE sample buffer at 95 °C for 5 m with 5 mM TCEP prior to SDS-PAGE and immunoblotting.

Indirect immunofluorescence microscopy

Immunofluorescent staining of gametes was performed as described previously¹³ with minor modifications. Briefly, approximately 1 × 10⁷ live gametes in N-free medium were allowed to settle and adhere for 10–15 m onto 0.1% poly-L-lysine coated 22 × 22 mm coverslips. Excess media and non-adhered cells were gently removed by pipetting before adding 100 μL of 4.0% paraformaldehyde for 10 m. Coverslips were then plunged into ice-cold methanol and incubated at −20 °C for 20 m, washed 3× with PBS at 20 °C (5 m/wash), washed 2× with PBSt (PBS + 0.5% Triton-X 100), and blocked for 1 h with blocking serum (5% goat serum, 0.1% BSA, 0.5% Triton-X 100, 1% cold-water fish gelatin in PBS). Coverslips were then incubated with anti-FLAG antibodies (1/500 dilution) in blocking serum for 2 h at 22 °C or overnight at 4 °C, washed 3× with PBSt, incubated for 1 h at 20 °C in the dark with Alexa Fluor 594-conjugated goat anti-mouse secondary antibodies diluted 1/500 in blocking serum, washed 3× with PBSt, and coverslips were mounted using Prolong Gold antifade reagent. These methods were also used to determine the localization of rCRR protein bound to WT mt+ and fus1 mt+ gametes with the following modifications: activated, live gametes were incubated with 150 μg/mL of rCRR protein for 15 m, before 3× washing with N-free media and fixation of gametes in 4% paraformaldehyde for 10 m. Fixed gametes were attached to the 0.1% poly-L-lysine coated coverslips. An anti-STREP mouse antibody (1/100) was used to detect the rCRR protein in these samples, followed by an Alexa Fluor 488-conjugated goat anti-mouse secondary antibody (1/500). Immunofluorescence images of gametes were captured on a Leica Stellaris 8 FALCON laser scanning confocal microscope with a Leica 63×/1.4 NA oil objective lens. Images are individual z-sections or z-stack composites as indicated in the figure legends and were processed with Leica LASAF software and Adobe Illustrator (V28.5).

Phase-contrast and differential interference contrast (DIC) microscopy

Phase-contrast microscopy was used in bioassays to quantify adhesion and fusion and to capture images of quadriciliated zygotes and gamete pairs adhered by their mating structures. For phase-contrast imaging, a Zeiss Universal Microscope equipped with a trinocular-mounted CMOS 13″ LCD Display Digital Camera and Zeiss ×40 objective lens was used to capture still images from glutaraldehyde-fixed, mixed gamete samples containing an ~1/5 dilution of Lugol’s iodine solution. DIC microscopy was used to assess ciliary adhesion between mt+ and mt− gametes with an inverted Axiovert 135 microscope equipped with an WF16X eyepiece adapter for image and video capture by a mobile device and a ×5/0.15 NA DIC Epiplan NEOFLUAR objective lens.

Scanning electron microscopy

mt+ (21gr) and mt− (hap2::HAP2-HA(-)) gametes (each at 2 × 10⁷ cells/mL in N-free medium) were separately activated by incubation for 15 m with ciliary ectosomes prepared as described previously⁷¹, mixed together for 30 m, washed 6× in sterile-filtered N-free media, fixed 60 m in Parducz fixative (1 part saturated HgCl₂, 6 parts 2% OsO₄)⁷², and transferred onto 0.1% poly-L-lysine coated coverslips for an additional 30 m. Coverslips were washed 10× in ddH₂O (2 m/wash), and dehydrated by three sequential washes (2 m/wash) in 70%, 95%, and 100% ethanol solutions. Coverslips were critical point dried from CO₂ and mounted on stubs over double-sided tape before being sputter-coated with gold–palladium (60%:40%) (Balzers Med 010) and observed using a scanning electron microscope (Hitachi S-4700 FESEM) at an accelerating voltage of 5 kV in high vacuum.

Chlamydomonas bioassays

Assays to quantify gamete activation, mating structure membrane adhesion, and cell–cell fusion were performed as previously described^13,45. Briefly, gamete activation was assayed by cell wall loss quantified by the amount of OD₄₃₅-detectable chlorophyll released into the supernatant upon treatment of samples with a detergent buffer^73,74. Equal numbers of mt+ and mt− gametes of the indicated genotypes were mixed for 10 m at a concentration of 5 × 10⁷ cells/mL, then added to 1.6 volumes of 4 °C N-free medium containing 0.075% Triton-X 100 and 5 mM EDTA (pH 8), vortexed, centrifuged at 8700×g for 30 s, and the OD₄₃₅ of the supernatant was measured using a Nanodrop 2000 spectrophotometer. The total amount of detectable chlorophyll (OD₄₃₅max) for individual samples was obtained from the supernatants of samples disrupted by a 12 s sonication with a Microson XL sonicator. Percent gamete activation was calculated as: (OD₄₃₅exp/ OD₄₃₅max) × 100.

Mating structure adhesion was assayed in mt+ and mt− gametes mixed in equal numbers at a concentration of 5 × 10⁷ cells/mL for 10 m, followed by fixation in an equal volume of 5% glutaraldehyde. Ciliary adhesions were disrupted by pipetting 20× with a 200 μL pipette tip. Glutaraldehyde fixation coupled with the gentle agitation from pipetting disrupts ciliary adhesions, leaving pairs of mt+ and mt− gametes attached only by their mating structures^45,75. The percent of gametes attached by their mating structures was calculated as: [(2 × number of pairs)/(2 × number of pairs + number of single gametes)] ×  100. Gamete fusion was assayed by the microscopic enumeration of quadriciliated cells from equal numbers of mt+ and mt− gametes mixed for 10 m and fixed in an equal volume of 5% glutaraldehyde. The percent of gamete fusion was calculated as: [(2 × number of quadriciliated cells)/(2 × number of quadriciliated cells + number of single gametes)] × 100. For adhesion and fusion bioassays, fixed samples were stored at 4 °C overnight, and at least 200 events (quadriciliated cells, single cells, or pairs) were counted in each biological replicate.

To assess the effects of rCRR on the ability of mt+ gametes to undergo mating structure adhesion and gamete fusion, mt+ gametes (5.0 × 10⁷ cells/mL in N-free medium) were activated by incubation for 1 h in db-cAMP buffer (15 mM db-cAMP, 0.15 mM papaverine, and 40 mM HEPES, pH 7.0¹⁵), followed by incubation in 150 µg/ml WT or mutant rCRR or buffer (10 mM Tris, 50 mM NaCl, pH 8.0) for 20 m, and mixture with equal numbers of mt− gametes for 10 m (0.1 mL samples at 2 × 10⁶ cells/mL) before fixation by addition of an equal volume of 5.0 % glutaraldehyde.

Statistical analysis

Data were analyzed using Prism 10.2.3 software (GraphPad). Results are shown as box and whisker plots with the box midline representing median, box size representing standard deviation, and whiskers showing data minima and maxima. All individual biological replicates for each sample are overlaid on plots as open circles. Significance was defined as a P value of <0.05. A nonparametric ANOVA, the Kruskal–Wallis test, and Dunn’s post-test were used to test for significant differences among samples. The sample sizes (n values) are indicated in the figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

J.F.P., H.H., and W.J.S. conceived the in vivo experiments. J.F.P. and H.H. generated the MAR1 mutant Chlamydomonas strains. J.F.P., H.H., S.M., M.T., S.P.V., T.T.G., and R.D. performed in vivo assays. A.F.-I., F.A.R., and E.B. conceived the in vitro experiments with recombinant MAR1 and FUS1. E.B. made constructs for recombinant expression of MAR1 and FUS1, expressed and crystallized the complex, collected and processed the X-ray diffraction data, and determined the initial model. P.L. assisted with model completion and structure refinement. A.F.-I. performed mutagenesis of rMAR1 and the interaction assays with recombinant proteins. F.A.R., W.J.S., E.B., and J.F.P. supervised the experimental work and wrote the manuscript with input from all other authors.

Peer review

Peer review information

Nature Communications thanks Pablo Aguilar, Shaogeng Tang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

F.A.R. discloses support for the research of this work from the ERC Advanced grant project (Health) CelCelFus; from grants ANR-18-CE92-0006, ANR-22-CE35-0004, and ANR-10-LABX-62-10 IBEID; from ANRS-MIE Project EMERGEN - ANRS 0149 - PRODEVAR and from program CAPNET as well as the Wellcome Trust collaborative grant UNS2208. F.A.R also used general support from Institut Pasteur and CNRS. W.J.S. discloses support for the research of this work from the National Institutes of Health [grant #R01GM146162], and J.F.P. discloses support for the research of this work from the National Institutes of Health [grant #K99/R00GM146112]. Support for the purchase of the Leica Stellaris 8 (A. Beaven) used for the research of this work was from the National Institute of Health [grant# 1S10ODO34260].

Data availability

Materials generated in this study including plasmids and Chlamydomonas strains are available upon request or by direct purchase from the Chlamydomonas Resource center (https://www.chlamycollection.org/). No new code or large-scale datasets were generated in this study. The accession number for the atomic coordinates and structure factors reported in this paper is PDB: 9QS0 https://doi.org/10.2210/pdb9QS0/pdb. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76990-7.