|Home | About | Journals | Submit | Contact Us | Français|
Sexual reproduction is almost universal in eukaryotic life and involves the fusion of male and female haploid gametes into a diploid cell. The sperm-restricted single-pass transmembrane protein HAP2-GCS1 has been postulated to function in membrane merger. Its presence in the major eukaryotic taxa—animals, plants, and protists (including important human pathogens like Plasmodium)—suggests that many eukaryotic organisms share a common gamete fusion mechanism. Here, we report combined bioinformatic, biochemical, mutational, and X-ray crystallographic studies on the unicellular alga Chlamydomonas reinhardtii HAP2 that reveal homology to class II viral membrane fusion proteins. We further show that targeting the segment corresponding to the fusion loop by mutagenesis or by antibodies blocks gamete fusion. These results demonstrate that HAP2 is the gamete fusogen and suggest a mechanism of action akin to viral fusion, indicating a way to block Plasmodium transmission and highlighting the impact of virus-cell genetic exchanges on the evolution of eukaryotic life.
Fusion of haploid cells to form a diploid zygote is the defining event of sexual reproduction in eukaryotes (Lillie, 1913). In organisms from every eukaryotic taxon, the plasma membranes of gametes of opposite sex or mating type come into intimate contact and then fuse to form the zygote (Bianchi et al., 2014, Okabe, 2013, Sinden et al., 1976, Snell and Goodenough, 2009, Sprunck et al., 2012). In spite of the importance of gamete fusion, very little is known about the molecular mechanisms of the membrane fusion reaction between gametes, and a bona fide fusion protein has not been formally identified. The best candidate to date is the ancient gamete plasma membrane protein HAP2, whose presence in green algae, higher plants, unicellular protozoa, cnidarians, hemichordates, and arthropods (Cole et al., 2014, Ebchuqin et al., 2014, Johnson et al., 2004, Kawai-Toyooka et al., 2014, Liu et al., 2008, Mori et al., 2006, Steele and Dana, 2009) indicates it was likely present in the last eukaryotic common ancestor (LECA) (Wong and Johnson, 2010). HAP2 was first identified in a screen for male sterility in the flowering plant Arabidopsis thaliana (Johnson et al., 2004) and later—under the name GCS1 (Mori et al., 2006)—as a sperm-specific protein shown to be required at an unidentified step in sperm-egg fusion (Mori et al., 2006, von Besser et al., 2006). A screen for genes essential for gamete fusion in the green alga Chlamydomonas independently uncovered HAP2, showing that it is expressed only in minus gametes and is exclusively present on an apically localized membrane protuberance termed the minus mating structure (Liu et al., 2008) (see Figure 1A for a diagram of Chlamydomonas fertilization). Studies in Chlamydomonas and Plasmodium (the pathogen causing malaria in humans) revealed that hap2 mutant gametes in both organisms were fully capable of robust adhesion to gametes of the opposite mating type or sex, but merger of the lipid bilayers was abrogated (Liu et al., 2008). In both organisms, adhesion relies on proteins that are species-limited: FUS1 in Chlamydomonas plus gametes and its unidentified receptor in minus gametes (Misamore et al., 2003), and p48/45 in Plasmodium berghei gametes (van Dijk et al., 2001). Based on these findings, which have since been confirmed in Arabidopsis thaliana (Mori et al., 2014) and the ciliated protozoan Tetrahymena thermophila (Cole et al., 2014), a model emerged positing that HAP2, a single-pass transmembrane protein, functions after species-limited adhesion in the membrane fusion process between gametes (Liu et al., 2008). Furthermore, in all of these organisms, HAP2 is required in only one of the two gametes, raising the possibility that it may function similarly to fusion proteins of enveloped viruses (Wong and Johnson, 2010, Harrison, 2015).
To understand the function of HAP2 at the molecular level we carried out concerted bioinformatic, functional, and X-ray structural analyses of HAP2 from Chlamydomonas reinhardtii. Initial bioinformatic analyses identified weak similarity to class II fusion proteins, revealing a segment within a cysteine-rich portion of HAP2 that could potentially correspond to the fusion loop. We demonstrate by mutational analysis and fusion-blocking antibodies targeting this segment that it has elements that are essential for HAP2 function. Finally, we show that the recombinant HAP2 ectodomain is monomeric, but inserts into liposomes by concomitantly forming trimers, the X-ray structure of which revealed a class II fusion protein fold in the typical trimeric post-fusion “hairpin” conformation.
HAP2 has 16 conserved cysteine residues with a signature distribution in the ectodomain (Figure 1B). Early alignments of HAP2 family members identified a characteristic ~50 aa domain (residues 352–399 in Chlamydomonas HAP2) with several conserved residues that was designated the HAP2/GCS1 PFAM domain (PF10699) (see http://pfam.xfam.org/family/pf10699) (Finn et al., 2016). A previous mutagenesis analysis in Chlamydomonas failed to identify functional properties in the PF10699 domain, as the mutant proteins tested either were not transported to the mating structure, or were nearly indistinguishable from wild-type in their ability to support fusion with plus gametes (Liu et al., 2015). Database searches for additional conserved regions using the HHpred protein homology detection server (Söding et al., 2005) indicated that a cysteine-rich region in the N-terminal half of the ectodomain exhibited weak similarity to class II fusion proteins. In particular, HHpred identified a polypeptide segment in C. reinhardtii HAP2 (170–204, SQVWDDTFGSSKERTRANLDCDFWSDPLDILIGRK) that fell in the fusion loop region of the flavivirus envelope protein E in the resultant amino-acid-sequence alignment (Figure S1). Analysis of HAP2 orthologs showed that the sequence in this region is highly variable, with a number of deletions and insertions and is framed at each side by relatively conserved segments: amino acids (aa) 159–167 upstream (including conserved cysteines 5–7) and aa 208–219 downstream (including conserved cysteine 9) (Figures 1B and 1C). Only amino acids R185 and C190 (in bold in the sequence above) within the identified segment are conserved, suggesting that they may play a role in HAP2 function.
To investigate the functional importance of this segment we transformed Chlamydomonas wild-type (WT) or mutant HAP2 transgenes carrying an influenza virus hemagglutinin tag (HA) into a fusion-defective, hap2 mutant strain (Liu et al., 2008) and assessed HAP2-HA expression and trafficking to the mating structure as well as fusion of the transformed hap2 minus gametes with WT plus gametes. HAP2-HA was detected in hap2 minus gametes transformed with WT HAP2-HA as the expected doublet in SDS-PAGE/immunoblotting (Figure 1D), the upper form of which was present on the cell surface as assessed by its sensitivity to protease treatment of live gametes (Liu et al., 2008). All mutant proteins were expressed at levels similar to wild-type HAP2-HA (Figure 1D), trafficked to the cell surface as assessed by their sensitivity to trypsin treatment of live gametes (examples shown in Figure 1E), and localized at the mating structure (example shown in Figure 1F). Thus, any defects in gamete fusion could be ascribed directly to the functional properties of the mutant HAP2 proteins. HAP2 with a deletion of residues 184TRA186 (HAP2-ΔTRA-HA), which includes the conserved R185 was non-functional and failed to rescue fusion in the hap2 mutant when mixed with wild-type plus gametes (Figure 1D). A mutant HAP2 with a lysine sustituted for the conserved R185 (the HAP2-R185K-HA mutant) was fully functional, whereas the HAP2-R185A-HA or HAP2-R185Q-HA mutants failed to rescue fusion (Figure 1D). A reverse-order HAP2-RA185-86 mutant (HAP2-R185A-A186R-HA) also was non-functional, indicating that a positively charged residue at position 185 is essential for the fusion activity. Finally, hap2 minus gametes expressing HAP2-F192A-W193A-HA were impaired in fusion, although fusion was not abolished by these mutations, indicating that these nearby aromatic residues also play a role in HAP2 fusion function (Figure 1D). Thus, the segment HAP2170-204, bounded by a pair of conserved cysteines, contains residues that are dispensable for protein expression, folding and localization, but are essential for the membrane fusion activity.
In an independent approach to examine the function of the HHpred-identified region, we generated a rabbit antibody against a synthetic peptide, HAP2168-190, spanning the functionally important R185 residue. The affinity-purified antibody (α-HAP2168-190) immunoprecipitated epitope-tagged HAP2-HA from lysates of HAP2-HA minus gametes (Figure 2A), confirming its reactivity with HAP2. To test whether α-HAP2168-190 interfered with gamete fusion, we incubated minus gametes with undiluted antibody, mixed them with plus gametes, and determined the percentage of gametes that had fused to form zygotes. Pre-incubation of minus gametes with α-HAP2168-190 had no effect on motility or adhesion but inhibited gamete fusion by over 75%, whereas pre-incubation with a control IgG had no effect on fusion (Figures 2B and 2C). Antibody dilution resulted in a loss of fusion-blocking activity, suggesting a low concentration of HAP2-specific antibodies in the polyclonal mixture, probably due to a low immunogenicity of the synthetic peptide. Pre-incubation of plus gametes with the antibody did not affect their ability to fuse; and the fusion-blocking activity of α-HAP2168-190 was neutralized by pre-incubation with the HAP2168-190 peptide, but not with a control peptide (Figure 2C), further documenting the specificity of the antibody. Finally, immunolocalization experiments using an anti-HA antibody showed that HAP2-HA on α-HAP2168-190-treated minus gametes remained at the mating structure (Figure 2D), indicating that α-HAP2168-190 did not alter HAP2 localization, but directly interfered with its function. These functional studies with Chlamydomonas mutant gametes and the anti-peptide antibody indicated that in its native conformation on live gametes, the 168–190 segment of HAP2 is accessible at the protein surface and its integrity and availability are essential for fusion of gametes.
We used a Drosophila expression system to produce a soluble form of C. reinhardtii HAP2 aa 23–592 (comprising almost the entire ectodomain, Figure 1B) and purified it to homogeneity (see STAR Methods; Figure S2). Analysis by size-exclusion chromatography (SEC) and multi-angle static laser light-scattering (MALLS; Figure S2) showed that the protein behaved as a monomer in solution (fraction labeled HAP2e), but had a tendency to oligomerize with time (especially under high ionic strength conditions) to elute at a volume corresponding roughly to hexamers (HAP2eh fraction in Figure S2). The purified protein from the HAP2e monomeric fraction efficiently neutralized the fusion-inhibition potential of the α-HAP2168-190 antibody (Figure 2C), indicating that at least part of the 168–190 segment is exposed in HAP2e and accessible to the antibody.
To detect membrane insertion, we incubated the recombinant protein with liposomes of a standard lipid composition (see STAR Methods) and monitored binding by co-flotation on a sucrose gradient (Figure 3E) followed by immunoblot detection of lipid-inserted HAP2e using a monoclonal antibody raised against HAP2e (mAb K3; see Figure S2; STAR Methods). We observed efficient co-flotation of the monomeric HAP2e fraction (Figure 3E), but not of the multimeric HAP2eh fraction (not shown). Electron microscopy analysis showed that HAP2e decorated the liposome surface as projecting rods about 12 nm long (Figures 3B–3D), which are similar to those formed by viral class II fusion proteins in their trimeric, post-fusion form, such as the alphavirus E1 protein (Gibbons et al., 2003). The size and shape of membrane-bound HAP2e suggested that it had also oligomerized upon membrane insertion. Of note, 3-fold symmetry was apparent in some top views of unbound proteins present in the background (Figure 3D, arrowheads). We confirmed that HAP2e had indeed trimerized upon membrane insertion by detergent-solubilizing it from the liposomes and analyzing it by native PAGE (Figure 3F) and by SEC (Figure S3). These results indicated that HAP2e behaves similarly to the alphaviruses and flaviviruses class II proteins, with membrane insertion concomitant with trimerization of a monomeric pre-fusion form (Klimjack et al., 1994, Stiasny et al., 2002), except that HAP2e did not require an acidic environment for lipid binding and trimerization. This difference is in line with the ability of HAP2 to induce gamete fusion in the extracellular environment, whereas alpha- and flaviviruses require the acidic environment of an endosome for fusion.
Because the co-flotation experiments demanded relatively high amounts of purified recombinant protein for detection, it was impractical to assess flotation inhibition by the undiluted polyclonal α-HAP2168-190 antibody. But we tested the effect of the mutations described in Figure 1 by recombinantly expressing the mutant HAP2 ectodomains. These experiments showed that the mutations that impaired gamete fusion also affected the co-flotation capacity of the mutant HAP2e (Figure 3G). The R185K HAP2e mutant co-floated with liposomes as efficiently as wild-type, whereas those in which the charge at position 185 was removed—the R185A and R185Q mutants—co-floated poorly (Figure 3G). Co-flotation with the liposomes was also essentially abrogated with the HAP2e-ΔTRA and HAP2e-F192A-W193A mutants. The combination of the in vivo (Figures 1 and and2)2) and in vitro (Figure 3) analyses indicates that HAP2 has the capacity to directly interact with membranes and that altering the conserved residues of the HHpred-identified segment affects this interaction.
We obtained crystals of HAP2e diffracting to a maximum resolution of 3.3 Å. Details of the crystallization and structure determination are described in the STAR Methods, and the crystallographic statistics are listed in Table S1. We determined the X-ray structure by the single isomorphous replacement with anomalous scattering (SIRAS) method using a PtCl4 derivative. The experimental electron density map allowed the tracing of 442 amino acids out of 569 in the HAP2e expression construct (aa 23–592; Figure 1B) from amino acids 24 to 581 with internal breaks at several disordered loops (listed in the STAR Methods).
The atomic model of C. reinhardtii HAP2 revealed a trimer with unambiguous structural homology to class II fusion proteins, featuring the three characteristic β sheet-rich domains, termed I, II, and III, arranged in a “hairpin” conformation typical of class II fusion proteins in their post-fusion form (Bressanelli et al., 2004, DuBois et al., 2013, Gibbons et al., 2004, Guardado-Calvo et al., 2016, Halldorsson et al., 2016, Luca et al., 2013, Modis et al., 2004, Willensky et al., 2016) (Figures 4 and andS4).S4). The overall shape of the HAP2e trimer is totally compatible with the projections from the liposomes observed in the electron micrographs, with the predicted membrane interacting region at the tapered end of the rods (Figures 3C and 3D). The monomeric HAP2e (Figure S2) used to grow the crystals obviously underwent the same conformational change to form trimers observed during insertion into liposomes (Figure 3). A similar oligomeric rearrangement during crystallization was reported previously for the dengue virus E protein under acidic pH conditions (Klein et al., 2013, Nayak et al., 2009). In the case of HAP2, acidification is not required; the trigger for the rearrangement to the post-fusion form is not known.
The comparison with class II fusion protein trimers of known structure defines the membrane-facing side (top in Figure 4) and the membrane distal end (bottom side) of the HAP2 trimer, in agreement with the shape observed in the electron micrographs (Figures 3B–3D). The Dali server (Holm and Park, 2000) yielded Z scores ranging from 9 to 16 for up to ~343 Cα atoms of class II fusion proteins of viral and cellular origin (Table 1), in the same range as those obtained in previous comparisons among known class II fusion proteins (Pérez-Vargas et al., 2014).
Domain I in HAP2 is a β sandwich of about 200 aa with ten β strands in the two apposed β sheets: the A0B0I0H0G0 sheet is buried and the J0C0D0E0F0 sheet is exposed in the trimer, and we refer to them as inner and outer sheets (Figures 5A and 5D). A specific feature of the HAP2 domain I is that the A0B0 β-hairpin is long and projects out of the inner sheet to augment the outer sheet of the adjacent subunit, where strand A0 runs parallel to J0 (Figure S5). These additional inter-subunit main chain β-interactions are likely to confer extra stability to the post-fusion trimer. At the membrane-distal, bottom face of the trimer, the outer β sheet projects very long loops, which are disordered in the crystal (dashed tubes in Figures 4, ,5,5, ,S4,S4, and andS5).S5). The C0D0 loop projection displays six additional cysteine residues (presumably forming three disulfide bonds) in an insertion that is present essentially only in algal HAP2 (Liu et al., 2008) (Figure 5D); the E0F0 loop projection is also an insertion with four potential N-linked glycosylation sites.
Domain III is made of about 130 aa and has an immunoglobulin-like fold, with seven β strands in two sheets (labeled ABE and GFCC’; Figures 5A, 5D, and andS4).S4). Although the β sheets are longer than in other class II fusion proteins (Figure 4), giving domain III a more elongated bean shape, it is in a similar location at the side of the trimer (Figure 4) as in other class II fusion proteins of known structure. Domain III packs against two adjacent protomers of the central trimer “core,” which is composed of domains I and II arranged parallel to each other and interacting along their length about the 3-fold molecular axis. The latter central trimer interaction is postulated to exist during formation of an “extended trimeric intermediate” state of the fusogenic conformational transition (Liao et al., 2010), in which the fusion loops are inserted in the target membrane (plus gamete here), while the C-terminal TM segment is anchored in the minus gamete mating structure, at the opposite end. The final collapse into the post-fusion hairpin conformation of each protomer in the trimer brings domain III to the sides of this core, projecting the downstream stem and TM regions toward the fusion loop (Liao and Kielian, 2005), in a fusogenic rearrangement of the protein analogous to an umbrella folding inside out. In this final, post-fusion location, domain III buries an area of about 2,100 Å2 of its surface, divided roughly equally in contacts with the same and with the adjacent subunit (intra- and inter-subunit contacts). The observed contacts therefore can only form after the assembly of the central trimer core, in line with the proposed clamping role of domain III, and resulting in irreversible trimerisation, as proposed for other class II fusion proteins (Liao and Kielian, 2005, Pérez-Vargas et al., 2014).
Domain II is the largest domain (roughly 250 aa in total) and is made by two distinct segments emanating from domain I: the D0E0 and H0I0 strand connections of the outer and inner sheets of domain I, respectively (Figure S4). As in all class II proteins, the domain I proximal region of domain II has a central β sheet (aefg; Figures 5 and andS4)S4) flanked by additional short helices. The distal “tip” of domain II contains β sheet bdc, with the β strands running parallel to the molecular 3-fold axis at the distal end of the D0E0 segment. The connection between β strands c and d at the tip of domain II (the cd loop) was shown to be the fusion loop in the viral proteins (reviewed in Kielian and Rey ). The bdc sheet normally packs against the ij β-hairpin (the distal end of the H0I0 segment), which in HAP2 maps to the conserved HAP2/GCS1 PFAM domain PF10699) (Figure 5A). Although in HAP2 β strands i and j are absent, we still refer to this region as the “ij loop” (Figures 5 and andS4S4).
The cd loop in HAP2 is 40 aa long and has an intervening short α-helix in the middle (α0), in contrast to the standard class II fusion proteins from the arthropod-borne viruses such as flaviviruses and alphaviruses (10–15 residues long). In this respect, HAP2 resembles the rubella virus class II fusion protein E1, which has 50 aa in between stands c and d, with a couple of short intervening α helices as well as an additional β strand which results in two separate fusion loops (DuBois et al., 2013). The presence of the α0 helix in HAP2 also results in two loops (loops 1 and 2 in Figures 5B and 5C), which project outward. Although disordered in the crystal, these two loops are in position to project non-polar residues into the target membrane. The HHpred alignment was indeed quite close and pointed correctly to the cd loop (Figure S1). The HAP2168-190 peptide used for immunization spans loop 1 all the way to the end of helix α0, and the fact that the resulting polyclonal antibody blocks fusion is in line with this region being exposed at this end of the molecule. The HAP2 crystal structure is therefore compatible with the mutagenesis data in this region, and with the effect of the α-HAP2168-190 antibody. Moreover, a peptide derived from the corresponding region of Tetrahymena thermophila HAP2 was found to display properties typical of a fusion loop (see related paper in Current Biology, Pinello et al., 2017), in line with our findings with C. reinhardtii.
The structure shows that the functionally important residue R185 is at the N terminus of the α0 helix and its side chain points away from the two exposed loops and toward the core of the HAP2 trimer. R185 makes a salt bridge and bidentate hydrogen bonds with the side chain of the strictly conserved E126 in β strand b (Figures 5B and 5C). Furthermore, the R185 side chain is at the core of a network of interactions stabilizing the main-chain conformation of the ij loop. This network involves main chain atoms of F376, G382 and R385, together with the strictly conserved Q379 side chain, which are part of the PF10699 signature segment that allowed the identification of HAP2 in widely disparate organisms. The structure now shows that the conserved pattern of disulfide bonds of this signature element is required for the ij loop to adopt a convoluted fold acting as a framework underpinning the cd loop by interaction with the conserved R185 so that the two fusion loops project out at the membrane-interacting region. The latter region, in contrast, is variable and has multiple deletions and insertions in the various orthologs, some insertions being quite long (Figure 1C). It is possible that the differences in membrane-interacting regions of HAP2s across the broad spectrum of eukaryotic organisms reflect evolutionary adaptations required for fusion with different target gametes. We note that the fusion loop in viral class II proteins is in general the most conserved segment of the protein within orthologs from a given virus genus, most likely because the same residues are also required for inter-subunit interactions in their pre-fusion form on infectious particles. But this comparison is not necessarily informative, since the analyzed HAP2 proteins span eukaryotic taxa that are much more distantly related than are viruses within a given genus. The pre-fusion conformation of HAP2 remains unknown, however, and it is possible that the non-polar residues of the two fusion loops are maintained unexposed until the time of fusion. The conserved interaction of the R185 side chain with that of the E126 residue in the tip domain may allow exposure of the two fusion loops only after a conformational change. In this context, the fusion-inhibiting α-HAP2168-190 antibody would bind only after a conformational change that exposes the fusion loop, but this remains to be explored.
The identical topological arrangement of secondary, tertiary, and quaternary structure elements of HAP2 with the viral class II proteins can only be explained by postulating a common ancestor. Indeed, the probability of convergence to the observed complex fold from independent origins, to result in two proteins displaying exactly the same topological arrangement throughout the entire ectodomain (as shown in Figure S4) is extremely low and can be considered negligible. Nature is parsimonious, and once a protein required for a complex function such as membrane fusion becomes available, the corresponding gene is used over and over again, most likely transferred via horizontal gene exchanges. This concept is supported by the observation that only three structural classes of viral fusion proteins have been observed so far, in spite of the enormous variety of known viruses. As one example, the membrane fusion proteins of the herpesviruses, rhabdoviruses and baculoviruses, which are otherwise totally unrelated viruses, were shown to be homologous by structural studies (Heldwein et al., 2006, Kadlec et al., 2008, Roche et al., 2006). These fusion proteins most certainly derived from a distant common ancestor (class III proteins) (Backovic and Jardetzky, 2011), whose genes must have been acquired via horizontal exchanges. Within eukaryotic organisms, only a few other cell-cell fusion proteins have been positively identified. Although the myoblast myomaker, a seven-pass transmembrane protein that governs fusion of myoblasts to form myotubes (Millay et al., 2013, Millay et al., 2016), has no obvious relation to viruses, the proteins involved in the two other cell-cell fusion events that have been characterized at the molecular level are virus related. Cytotrophoblast fusion in mammals during placenta formation (Blaise et al., 2003, Holm and Park, 2000) and epidermal cell-cell fusion in nematodes to form syncytia (Mohler et al., 2002) both use fusion proteins also found in viruses. In the first case, the class I fusion protein involved is clearly derived from an endogenous retroviral element (Denner, 2016). It is also possible that a similar process involving retrotranscription may have taken place in the case of the class II Caenorhabditis elegans fusion protein EFF-1 (Pérez-Vargas et al., 2014), as retroviruses of nematodes have been found to have an envelope protein related to that of the phleboviruses (Frame et al., 2001, Malik et al., 2000), which have class II fusion proteins (Dessau and Modis, 2013). These observations suggest that retro-transcription, followed by integration into the genome, may have been an important pathway for gene exchanges between viruses and eukaryotic cells.
HAP2 is present in organisms responsible for several of the globe’s most devastating human diseases, including Plasmodium, Trypanosoma, and Toxoplasma. Many arthropods that are vectors of human diseases or are agricultural pests, such as insects and ticks, also possess HAP2. A strategy that used recombinant Plasmodium HAP2 fragments to induce transmission-blocking immunity in mice has been reported previously (Blagborough and Sinden, 2009, Miura et al., 2013), but the expression systems were not sufficient for viable clinical developments. Our results suggest that the use of a peptide spanning the HAP2 fusion loops as immunogen might be sufficient to induce transmission-blocking immunity, similar to the antibodies we obtained here against Chlamydomonas HAP2.
In conclusion, our data now open the way to a full mechanistic characterization of gamete fusion induced by HAP2 and raise new questions, including the identification of the trigger for the HAP2 fusogenic conformational change, the structure of the pre-fusion form(s) of HAP2, and its organization on the mating structure of the minus gamete membrane. Evolution through hundreds of millions of years may have led the different taxa to develop alternative solutions, and the metastable pre-fusion form of HAP2 may be organized differently in the multiple organisms in which it is present, as shown for viruses of different families. In contrast, the post-fusion conformation described here appears as a universal feature of class II fusion proteins.
Requests for resources and reagents should be directed to Félix A. Rey (firstname.lastname@example.org).
Wild-type Chlamydomonas strains 21 gr (mating type plus; mt+; CC-1690), 40D4 (hap2 mating type minus mutant (Liu et al., 2015)) and HAP2-HA (HAP2-HA plasmid-rescued hap2 strain (Liu et al., 2015)) were used and are available from the Chlamydomonas Culture Collection. Vegetative growth of cells and induction of gametogenesis in gamete medium (M-N) were as described before (Liu, et al., 2008). Gametes were activated with dibutyryl-cAMP (db-cAMP) by incubation with 15 mM db-cAMP and 0.15 mM papaverine for 0.5 hr in N-free medium (Ning et al., 2013). Gamete fusion was assessed by determining the number of cells that had formed zygotes after being mixed with wild-type plus gametes for 30 min and was expressed as percent fusion using the following equation: (2 × number of zygotes)/[(2 × number of zygotes)+(number of unfused gametes)] × 100 (Liu et al., 2015).
A polyclonal antibody against HAP2 peptide SSSQVWDDTFGSSKERTRANLDC (aa 168-190) was made in rabbits by Yenzym Antibodies (San Francisco, CA; α-HAP2168-190), under oversight by their Institutional Review Board. The antibody was purified on a peptide-conjugated affinity column prepared by the company. To assay for antibody inhibition of gamete fusion, gametes that had been activated with dibutyryl-cAMP for 30 min (Misamore et al., 2003) were washed once with gamete medium and the activated gametes (2 × 107 cell/ml) were incubated with 110 ug/ml peptide antibody (pre-dialyzed in gamete medium) for 2 hr followed by mixing with gametes of the opposite mating type gametes for the times indicated in the figure legends. The number of zygotes (detected as cells with 4 rather than 2 cilia) was determined by phase-contrast microscopy. At least 100 cells were counted each time with at least two counts per sample in at least two separate experiments. A rabbit IgG (Sigma) was used as a control antibody.
To test the capacity of peptides or HAP2e to neutralize the fusion-inhibiting properties of α-HAP2168-190, the antibody (220 ug/ml) that had been dialyzed in gamete medium was incubated overnight with control peptide CTQPPRPPWPPRPPPAPPPS (200 ug/ml) (this peptide is encoded by Cre03 g176961 whose transcripts are specific to minus gametes and upregulated during gamete activations (Ning et al., 2013), HAP2168-190 peptide (200 ug/ml), or HAP2e protein (200 ug/ml) overnight and then added to activated gametes as above to determine percent gamete fusion.
Immunofluorescent staining of gametes was performed as described previously (Belzile et al., 2013). The gametes were fixed in ice cold methanol; the primary antibody was rat anti-HA (Roche) and the secondary was Alexa Fluor 488 Goat Anti-Rat (Invitrogen). Images were captured in DIC or FITC channels as described previously (Liu et al., 2015). For HAP2-HA localization after pre-incubation in α-HAP2168-190 (see below), db-cAMP-activated gametes were incubated with the antibody for 2 hr, washed three times with gamete medium, and stained with HA antibody as above.
For immunoprecipitation, gametes were disrupted by incubation at 4°C for 30 min in RIPA buffer (20 mM Tris, 150 mM NaCI, 1% NP-40, 0.5% deoxycholate, 0.1% SDS) containing a proteinase inhibitor cocktail (Roche). The samples were centrifuged at 12000 rpm for 30 min, the supernatants were subjected to immunoprecipitation with α-HAP2168-190 and protein A agarose beads using methods described previously (Liu et al., 2010) and the immunoprecipitated proteins were subjected to SDS-PAGE and immunoblotting (Liu et al., 2015). For trypsin treatment 5x107 live gametes/ml were incubated in 0.05% freshly prepared trypsin for 20 min at room temperature, diluted 10-fold with N-free medium, centrifuged, and resuspended in fresh N-free medium containing 0.01% chicken egg white trypsin inhibitor. For immunoblotting, the remaining cells were washed twice more with N-free media containing 0.01% chicken egg white trypsin inhibitor before analysis by SDS-PAGE and immunoblotting (Misamore et al., 2003).
ΔTRA184-6, R185A, R185K, R185Q, RA185-6AR and FW192-3AA mutant forms of Chlamydomonas HAP2 (GI:288563868) were performed using standard PCR methods. The PCR fragments were inserted into the HAP2-HA plasmid at BglII/NruI sites using In-fusion Dry-down PCR Cloning Kit (Clontech). All plasmids were confirmed by DNA sequencing. Plasmids were transformed into the 40D4 hap2 mutant using electroporation and selected on paromomycin plates (Liu et al., 2015). Colonies were selected based on PCR identification of the transgene and confirmation of HAP2-HA expression by immunoblotting (Liu et al., 2008).
Codon-optimized synthetic cDNA corresponding to a soluble C-terminally truncated version of the HAP2 ectodomain (HAP2e) comprising residues 23-592 from Chlamydomonas reinhardtii was cloned into a modified Drosophila S2 expression vector described previously and transfection was performed as reported earlier (Krey et al., 2010). For large-scale production, cells were induced with 4 μM CdCl2 at a density of approximately 7x106 cells/ml for 8 days, pelleted, and the soluble ectodomain was purified by affinity chromatography from the supernatant using a StrepTactin Superflow column followed by size exclusion chromatography using a Superdex200 column in 10mM bicine pH9.3. Pure protein was concentrated to approximately 20 mg/ml.
Purified HAP2e ectodomain was subjected to SEC using a Superdex 200 column (GE HealthCare) equilibrated with the indicated buffers. Separations were performed at 20°C with a flow rate of 0.5 mL min−1. Online MALLS detection was performed with a DAWN-HELEOS II detector (Wyatt Technology, Santa Barbara, CA, USA) using a laser emitting at 690 nm. Online differential refractive index measurement was performed with an Optilab T-rEX detector (Wyatt Technology). Data were analyzed, and weight-averaged molecular masses (Mw) and mass distributions (polydispersity) for each sample were calculated using the ASTRA software (Wyatt Technology).
BALB/c mice were immunized subcutaneously with 10 μg of recombinant HAP2e in Complete Freund’s adjuvant and boosted five times with the same antigen dose in Incomplete Freund’s adjuvant. Mice splenocytes were fused to P3U1 myeloma cells and growing hybridomas were selected in an ELISA test on plates coated with 1 μg/ml HAP2e. Specific IgG producing hybridomas were further subcloned by limiting dilution. Monoclonal antibody K3 was purified using Protein G HiTrap columns (GE Healthcare) according to the manufacturer’s instructions followed by SEC in PBS using a Sdx200 column.
DOPE (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine), DOPC (1,2-dioleoyl- sn-glycero-3-phosphocholine), cholesterol and sphingomyelin were purchased from Avanti Polar Lipids. Liposomes were freshly prepared by the freeze–thaw and extrusion method using molar ratios of 1/1/3/1 of DOPE/DOPC/cholesterol/sphingomyelin. 0.7 μM purified HAP2e were mixed with 8 mM liposomes and incubated 1h at 25°C in 100 μL PBS. Samples were then adjusted to a final concentration of 20% sucrose, overlayed with a 5%–60% sucrose gradient (in PBS) and centrifuged overnight at 4°C at ~152.000 x g. Fractions from the top, middle and bottom of the gradient were analyzed by immunoblotting using specific anti-HAP2 monoclonal antibodies and the bands quantified using the GeneTools Syngene software. The percentage of HAP2e in either fraction was calculated as the ratio between HAP2e in individual fractions and total HAP2e (sum of HAP2e in top and bottom fractions).
Purified HAP2e (C. reinhardtii) mixed with liposomes was spotted on glow discharged carbon grids (CF300, EMS, USA), negatively stained with 2% phosphotungstic acid (PTA) pH 7.4, analyzed with a Tecnai G2 Bio-Twin electron microscope (FEI, USA) and imaged with an Eagle camera (FEI, USA). For cryo-electron microscopy liposomes alone or liposomes mixed with purified HAP2e were applied on a glow discharged Lacey Carbon grid (Agar Scientific, UK). Samples were plunge-frozen in liquid ethane using an automated system (Leica EMGP, Austria) and visualized on a Tecnai F20 electron microscope operating at a voltage of 200 kV. Image frames were recorded in low-dose mode on a Falcon II direct electron detector (FEI, USA).
Purified recombinant HAP2e in PBS and HAP2e from the top fraction of the sucrose gradient solubilized with 4% CHAPS were independently purified by size-exclusion chromatography using a Superdex 200 Increase column. Elution fractions corresponding to the respective proteins were further analyzed on a 4%–16% native gradient gel using the nativePAGE Novex Bis-Tris gel system (Invitrogen) followed by silver staining.
Crystals of HAP2e were obtained using in situ proteolysis as described before (Dong et al., 2007). Briefly, subtilisin dissolved in 10mM Tris pH8, 30mM NaCl at 10mg/mL was added to protein solution (12-14 mg/ml in 10 mM bicine, pH 9.3) on ice immediately prior to crystallization trials in a 1/100w:w ratio. Crystals of HAP2e were grown at 293K using the hanging-drop vapor-diffusion method in drops containing 1 μL protein/protease solution mixed with 1 μL reservoir solution containing 100 mM HEPES pH7.5, 2% 2-Propanol, 100 mM sodium acetate, and 12%–14%w/v PEG 8000. Diffraction quality rod-like crystals appeared after 1 week and were flash-frozen in mother liquor containing 30% (v/v) MPD.
Data collection was carried out at the Swiss Light Source (PX I), the ESRF (ID30A-3), and the Synchrotron Soleil (Proxima1). Data were processed, scaled and reduced with XDS (Kabsch, 1988), Pointless (Evans, 2006) and programs from the CCP4 suite (Collaborative Computational Project, 1994). A single-wavelength anomalous dispersion (SAD) dataset was collected from a single crystal of HAP2e from C. reinhardtii soaked for 6 hr in 2 mM K2PtCl4 solution in cryo buffer. Data were collected at the LIII edge of Platinum (1.072 Å) on a single crystal using low-dose (0.5 MGy per 360°), high-redundancy (5 × 360°) fine -sliced collection strategy using five crystal orientations by means of a high-precision multi-axis PRIGo goniometer (Weinert et al., 2014). An initial set of experimental phases was obtained by the Single Isomorphous Replacement method using autoSHARP (Vonrhein et al., 2007) with the Platinum derivative and a highly isomorphous native dataset. Starting phases were improved by consecutive cycles of manual building and combination with phases derived from molecular replacement using the partial model as search model in Phaser (MR-SAD) (McCoy et al., 2007). After building an initial poly-alanine model accounting for ~50% of the Cα atoms these phases were further refined using the anomalous signal of a highly redundant Sulfur-SAD dataset collected at a wavelength of 2.06641 Å on crystals of the native protein following a similar collection strategy as mentioned above (Weinert et al., 2014). Model building was performed using Coot (Emsley et al., 2010), and refinement was done using AutoBuster (Bricogne et al., 2010) with repeated validation using MolProbity (Chen et al., 2010). The final model includes amino acids 24 to 581 (see linear diagram in Figure 1B), with internal breaks at loops 69-97, 152-156, 167-182, 194-205, 238-283 and 330-345, corresponding to disordered loops that are marked with a gray background in the C. rheinhardtii sequence in Figure 5D (top sequence) and as dashed tubes in the ribbon diagrams (Figures 4 and and5A–5C).5A–5C). Clear electron density was observed for one N-linked and one O-linked glycan chain (attached to N497 and T577 in domain III).
Data are presented as mean ± SD unless otherwise indicated in figure legends and experimental repeats are indicated in figure legends.
The accession number for the atomic coordinates and structure factors reported in this paper is PDB: 5MF1.
W.J.S. and Y.L. conceived the in vivo experiments with full-length HAP2, and T.K. and F.A.R. conceived the in vitro experiments with the recombinant HAP2 ectodomain. J.P. and N.G. performed the HAP2 homology search and generated alignments. W.J.S., Y.L., J.P., and N.G. designed the mutant HAP2 proteins and selected the peptide used to produce the fusion-blocking antibody. W.J.S., Y.L., and W.L. designed the experiments with Chlamydomonas gametes; Y.L. and W.L. performed the gamete fusion assays with HAP2 mutants and antibodies and interpreted them with W.J.S. J.F. and A.M. made the constructs, expressed and crystallized the HAP2e protein, and J.F. optimized the crystals for diffraction. F.T. raised and isolated monoclonal antibodies targeting C. reinhardtii HAP2. J.F. and T.K. collected data and determined the HAP2 X-ray structure and G.B. helped devise a collection strategy for derivative X-ray data. J.F. and M.A.T. demonstrated insertion of HAP2 into liposomes, and G.P.A. performed electron microscopy on liposomes. T.K., F.A.R., and W.J.S. supervised experimental work and wrote the manuscript with input from all other authors.
We thank Francis-André Wollmann and Sandrine Bujaldon for discussions and help with initial experiments; Fredrick Grinnell, Saikat Mukhopadhyay, Michael Henne, and Margaret Phillips of UT Southwestern for helpful discussions; the crystallization platform of the Pasteur Proteopole for technical help; Remi Fronzes for help with native PAGE experiments; Vincent Olieric and the staff of synchrotron beamlines PX-III at the Swiss Light Source, Proxima-1 and -2 at SOLEIL and ID29 and ID30-3 at the European Synchrotron Radiation Facility for help during data collection; members of the Rey and Snell labs for discussions; and M. Nilges and the Equipement d’excellence CACSICE for providing the Falcon II direct electron detector. F.A.R. acknowledges funding from the ERC Advanced grant project (340371) CelCelFus for this work, which also used general support from Institut Pasteur and CNRS. W.J.S. was supported by a grant from the NIH (GM56778) and acknowledges funding from the UTSW HI/HR Program. N.V.G. is funded, in part, by a grant from the NIH (GM094575) and the Welch Foundation (I-1505). J.F. benefitted from an Allocation ministérielle pour l'Ecole Polytechnique AMX.
Published: February 23, 2017
Supplemental Information includes five figures and one table and can be found with this article online at http://dx.doi.org/10.1016/j.cell.2017.01.024.