The Functional Domain of GCS1-Based Gamete Fusion Resides in the Amino Terminus in Plant and Parasite Species

Fertilization is one of the most important processes in all organisms utilizing sexual reproduction. In a previous study, we succeeded in identifying a novel male gametic transmembrane protein GCS1 (GENERATIVE CELL SPECIFIC 1), also called HAP2 (HAPLESS 2) in the male-sterile Arabidopsis thaliana mutants, as a factor critical to gamete fusion in flowering plants. Interestingly, GCS1 is highly conserved among various eukaryotes covering plants, protists and invertebrates. Of these organisms, Chlamydomonas (green alga) and Plasmodium (malaria parasite) GCS1s similarly show male gametic expression and gamete fusion function. Since it is generally believed that protein factors controlling gamete fusion have rapidly evolved and different organisms utilize species-specific gamete fusion factors, GCS1 may be an ancient fertilization factor derived from the common ancestor of those organisms above. And therefore, its molecular structure and function are important to understanding the common molecular mechanics of eukaryotic fertilization. In this study, we tried to detect the central functional domain(s) of GCS1, using complementation assay of ArabidopsisGCS1 mutant lines expressing modified GCS1. As a result, the positively-charged C-terminal sequence of this protein is dispensable for gamete fusion, while the highly conserved N-terminal domain is critical to GCS1 function. In addition, in vitro fertilization assay of Plasmodium berghei (mouse malaria parasite) knock-in lines expressing partly truncated GCS1 showed similar results. Those findings above indicate that the extracellular N-terminus alone is sufficient for GCS1-based gamete fusion.


Introduction
Angiosperm fertilization is comprised of certain processes, from pollination to gamete fusion [1]. Each pollen grain (male gametophyte) contains a pair of sperm cells (male gametes), and elongates a pollen tube into the pistil to deliver the sperm pair towards an ovule contained in an ovary after pollination. When the pollen tube reaches the gate of the ovule (micropyle), it releases the sperm pair into an embryo sac (female gametophyte) enclosed in the ovule wall. Female gametes, namely egg and central cells, exist close by in an embryo sac and fuse with these sperm cells to produce an embryo and an endosperm, respectively (double fertilization).
In our previous study, we succeeded in identifying the novel protein GCS1 in male generative cells isolated from Lilium longiflorum pollen [2]. Most angiosperm GCS1s are composed of approximately 700 amino acid residues, and are predicted to be a single-pass transmembrane protein, because of the N-terminal signal sequence and C-terminal transmembrane domain [2][3]. It has been found that Arabidopsis GCS1 is identical to HAP2, which was previously identified as a pollen tube related factor from the hap2 phenotypes [4]. Lilium and Arabidopsis GCS1s were demonstrated to be expressed exclusively in male gametes (generative and sperm cells) and localized to the cell surface [2][3]. Furthermore, Arabidopsis GCS1 mutant pollen exhibits serious male sterility in which none of the sperm cells are able to fuse with female gametes, suggesting that GCS1 is an indispensable factor for gamete fusion [2][3].
Surprisingly, GCS1 is highly conserved and putative orthologs have been identified in various eukaryotes, e.g., protists, amoebae and invertebrates [2][3][5][6][7]. In Plasmodium berghei (a rodent malaria parasite) and Chlamydomonas reinhardtii (a green alga), it has been shown that their GCS1 is similarly expressed in the male gamete and functions in gamete fusion [5][6]. The GCS1-knockout Chlamydomonas male cannot perform gamete fusion, but does achieve attachment based on FUS1, which is a transmembrane protein expressed exclusively in the female gamete [8][9], and therefore GCS1 is expected to function in membrane fusion or in events immediately after attachment [6]. Furthermore, a recent paper reported testis-specific GCS1 expression in the hydra (a cnidarian), implying that animal GCS1s function in a similar manner [7].
Since GCS1 possesses no known functional protein functional domains, the molecular structure and central domain(s) for gamete fusion are important issues [10][11]. A recent study on Chlamydomonas GCS1 revealed GCS1 to be a glycoprotein in which two types of N-glycosylation occur, and a rapid degradation of GCS1 molecules is triggered by gamete membrane fusion so as to prevent polygamy [12]. Furthermore, Wong et al. investigated the molecular importance of N-and C-terminal sequences for the GCS1 transmembrane domain, using partially-modified AtGCS1 constructs [11]. In their study, entire deletion of either terminus leads to failure in complementation of the AtGCS1 mutation, roughly suggesting that both termini are required for the GCS1 function of gamete fusion. In addition, they indicated that the positively-charged histidine rich domains are indispensable for normal double fertilization, since reduction of the positive charge causes reduced fertility and an occasional single fertilization, where only one sperm cell fuses with the egg or central cell in an ovule [11].
In the present study, an effort was made to obtain more detailed characteristics of GCS1 structure and function using partiallymodified AtGCS1 constructs, which are based on green fluorescence protein (GFP) insertion targeted to certain characteristic AtGCS1 sequence regions. To ensure the conservation of these GCS1 characteristics, similar constructs were also produced in PbGCS1. We report that the gamete-fusion functional GCS1 domain(s) is generally in the N-terminus and the function is drastically impaired even when the N-terminus is split from the gamete membrane.

Construction of Modified AtGCS1s
As previously described, AtGCS1 is composed of an N-terminus signal sequence (SS), which probably leads to its cell membrane localization, a long body sequence containing the HAP2-GCS1 domain that is highly conserved among GCS1-possessing organisms, a hydrophobic transmembrane (TM) domain and a highly basic C-terminal histidine-rich (HR) sequence ( Figure 1A) [2][3]11]. Unlike the previous study, in which long sequences of AtGCS1 were deleted or exchanged mainly [11], we used a different strategy that partly disrupts characteristic AtGCS1 domains with a fluorescence tag insert so that we could see the importance of each domain, and normal expression of the modified AtGCS1s as well. We first disrupted, separated, or exchanged AtGCS1 domains described above with the GFP cDNA insertion at appropriate restriction-enzyme-recognizing sites. As a result, 6 modified AtGCS1 (mAtGCS1) constructs were obtained (designated as GDD, GAA, GAH, GHH, GHP and GPP). The products expected from these constructs are shown in Figure 1B. In GDD, the HAP2-GCS1 domain has been disrupted. The GFP insert in GAA and GHH separates the entire N-terminal AtGCS1 sequence and HR from TM, respectively. GAH and GHP have lost TM and HR, respectively. In GPP, HR is followed by the GFP insert. These constructs were produced on the basis of a genomic AtGCS1 clone, which was successfully used in GCS1 knockout rescue in a previous study [2], and introduced into heterozygous Arabidopsis GCS1 mutant (+/gcs1) plants to assess which of the constructs maintains or loses function, namely which domain(s) is critical to gamete fusion. The +/gcs1 line was used in our previous study and the gcs1 pollen is completely male-sterile [2]. Normal transcription and translation of each construct were confirmed in reverse transcription (RT)-PCR assays and by pollen observation, respectively (Figures 2A and 2B). From all results above, all constructs proved to work desirably and give no notable defects to the pollen development.

Complement Assay of the +/gcs1 Plants with the mAtGCS1 Constructs
When an mAtGCS1 construct is hemizygously introduced into +/gcs1 plants, the pollen from the transformants are categorizable into 4 genotype groups, namely wild type pollen (I), gcs1 pollen (II), mAtGCS1-possessing pollen (III) and mAtGCS1-possessing gcs1 pollen (IV) (Figure. 3A). The T-DNA insert in the gcs1 pollen is linked with the kanamycin-resistance gene (Kan R ), while all of the mAtGCS1 constructs are linked with the gentamycin-resistance gene (Gen R ). Of these groups, the pollen from groups I and III is  [3]. The exons coding the signal sequence (SS), HAP2-GCS1 domain, transmembrane domain (TM) and histidine-rich domain (HR) are differently colored. SS, HAP2-GCS1 and TM sequences correspond with those of previous study by von Besser et al [3]. HR region ranges from the head of H1 to the end of H3 in a previous study by Wong et al [11].
The T-DNA insert in the AtGCS1 mutant is indicated (gcs1). N-and Cterminal regions (AtN and AtC respectively), which were deleted in the previous study by Wong et al [11], are represented. The endogenous restriction enzyme sites used for mAtGCS1 construction are indicated with abbreviations; HindIII (Hd), AflII (A), HpaI (H), PmlI (P). (B) The structure of mAtGCS1 products. The top bar is the normal Arabidopsis GCS1 protein structure (AtGCS1). The other bars indicate mAtGCS1 protein structures with a GFP insertion. In the GDD, GAA, GAH, GHH, GHP and GPP constructs, the GFP cDNA insert is located in their Hd, A, A-Ha, Ha, Ha-P and P-P sites respectively. doi:10.1371/journal.pone.0015957.g001 evidently fertile because of normal GCS1 expression. On the other hand, the group II pollen is infertile because of the GCS1 mutation, and therefore, Kan R is never inherited by the offspring plants paternally. However, if an mAtGCS1 construct complements the GCS1 mutation, both Kan R and Gen R in group IV should be inherited by the offspring plants paternally. In this case, one-third (,33.3%) of the offspring should survive in the Kan-and Gencontaining media, while the remaining ones (,66.7%) should not be alive, because they are from groups I and III in which the wildtype plants were pollinated with the transformant pollen. Based on this expected scenario, wild type females were pollinated with pollen from each of the mAtGCS1 transformants, and the resulting offspring seeds were sown in Kan-and Gen-containing media to assess which construct complements the GCS1 mutation ( Figure 3B). As a positive control, +/gcs1 AtGCS1 lines, in which a wild-type genomic AtGCS1 clone linked with Gen R (AtGCS1 construct) is hemizygously expressed, were used. As a result, when the +/gcs1 AtGCS1 pollen was used, approximately 33.3% of the offspring plants survived in the media (chi-square, P.0.5), indicating that complementation by the construct was almost completely successful ( Figure 3C, the top bar chart). On the other hand, almost all of the plants from the +/gcs1 GDD and +/gcs1 GAH pollen died, while that of the +/gcs1 GAA pollen exhibited weak complementation ( Figure 3C). The difference between the results for +/gcs1 GAA and +/gcs1 AtGCS1 is statistically significant (chisquare, P,0.01). Furthermore, the results for the +/gcs1 GHH , +/ gcs1 GHP and +/gcs1 GPP pollen displayed obvious complementation and were comparable with to +/gcs1 AtGCS1 (chi-square, P.0.5) ( Figure 3C). At the same time, these results for GHH, GHP and GPP constructs indicate that their C-terminal GFP inserts did not dominantly affect the fertility of wild-type pollen.

Sperm Activity in +/gcs1 GAA Plants
The weak gcs1 complementation by GAA led to the expectation that the +/gcs1 GAA sperm cells occasionally fail to fuse. To observe the sperm behavior in an ovule, we produced a +/gcs1 GAA line whose sperm cells express HTR10-RFP fusion as a sperm nucleus marker [13], in addition to the GFP signal from GAA (Figures 4A-C). When isolated ovules were observed at 24 h after pollination (HAP), a pair of unfertilized sperm cells was occasionally detected in an ovule as a GFP signal pair ( Figure 4D). Most of these sperm cells also expressed RFP signals and were detected in the vicinity of the border between the egg and central cells, indicating that GAA is less functional than the other mAtGCS1 constructs which are successful in complementation, such as GHH, GHP and GPP (Figures S1, 4E and 4F). Less frequently, it was observed that one sperm cell was fusing with an egg cell, while the remaining one of the pair was left unfertilized and retained the GFP signal To trace fertilization products derived from such incomplete gamete fusion, an Arabidopsis line that express H2B-RFP and DFWA-GFP as egg and central cell division marker respectively [14][15][16], was produced and pollinated with +/ gcs1 +/GAA line pollen, in which sperm cells homozygously express GAA ( Figures 5A-C). As a result, when the ovules at 30 HAP were observed, single fertilization products were occasionally detected ( Figures 5D-G). Taken together, we conclude that GAA is not stably functional and the GFP insert separating the entire Nterminus from TM might have given an altered feature to the GCS1 molecule.

Conservation of GCS1 functional domains
To confirm whether the importance of the GCS1 domain is generally conserved in other organisms, a similar GCS1 modification assay was performed in P. berghei GCS1 (PbGCS1). In this assay, we prepared 3 partially-modified PbGCS1 constructs, designated -HG, -C and -TM/C, besides the normal PbGCS1 construct (PbGCS1 Full). The expected product from each construct is shown in Figure 6A. In the -HG, -C and -TM/C peptides, the HAP2-GCS1 domain, the entire C-terminal region from TM, and a sequence covering the TM domain and the entire C-terminal region, were respectively deleted. Each construct was introduced into P. berghei parasites by double crossover homologous recombination, by which the endogenous PbGCS1 gene is replaced with each construct. Normal expression of the transgene in these knock-in (KI) parasites was confirmed with RT-PCR assay ( Figure 6B).
Since the in vitro fertilization assay has been established in rodent malaria parasites [17], we used it to assess the male fertility of each KI parasite, based on the appearance of ookinates, zygotic cells resulting from successful gamete fusion. As a result, obvious ookinate conversion was frequently detected in -C KI-parasites with similar efficiency to that of PbGCS1 Full KI parasites (chisquare, P.0.5). This result indicates that the homologous recombination itself does not affect the male fertility and that the C-terminus of PbGCS1 less contributes to gamete fusion ( Figures 6C and 6D). On the other hand, no ookinates were detected in -HG-and -TM/C-KI parasites, suggesting that the highly conserved N-terminal domain and TM of PbGCS1 are similarly critical to parasite gamete fusion ( Figures 6C and 6D). We, therefore, conclude that the domain-deleting strategy is also applicable to investigation of molecular GCS1 functions and the results are consistent with those of AtGCS1.

Discussion
The C-terminus of GCS1 is Dispensable for Gamete Fusion In a previous study, we showed that angiosperm GCS1 is a membrane-associated protein that is localized in the male gamete membrane [2]. Consistent with this, AtGCS1 without TM (GAH) exhibited impaired GCS1 function, suggesting that the TM domain is required for membrane localization of GCS1 ( Figure 3C). A similar result was also obtained in -TM/C KI parasites, in which GCS1 lacks both TM-and entire C-terminalregions ( Figure 6). Since -C KI parasites, in which GCS1 lacks only the entire C-terminus, exhibited normal fertility, its TM is similarly critical to GCS1 localization.
Next question was which of the N-and C-termini to the TM is critical for the gamete fusion function of GCS1. Wong et al. recently reported 3 major histidine-rich regions (H1-3) in the Cterminus of AtGCS1, and demonstrated that deletion of the Cterminal sequence covering them or an exchange of their positively charged residues, resulted in an obvious reduction of GCS1 function [11]. However, our data did not show any functional deficiency in GHP, the H1-3 regions of which were all exchanged with the GFP sequence ( Figures 1B and 3C). Since the GFP insert has no positive charge corresponding to the original histidine-rich regions, the hypothesis of the C-terminus contribution would appear to be controversial. In addition, the C-terminus-deleted GCS1 was also functional in gamete fusion in the malaria parasite ( Figure 6). This indicates that the C-terminus contribution to gamete fusion is at least not conserved as a general GCS1 feature. As in the case of GHP, functional deficiency was not observed in either GHH or GPP, in which the GFP insert is flanked by HR, indicating that the GFP structure itself did not inflict any damage on the C-terminus. These markers are useful to track GCS1 behavior during gamete fusion, and some are currently being used for the purpose of a live imaging of Arabidopsis gamete fusion (unpublished data). As an explanation for our results conflicting with the hypothetical C-terminus contribution, C-terminal domain(s) may be associated with functional localization and/or stability of AtGCS1 molecules on the sperm membrane. Indeed, the functional GHP keeps endogenous 34 amino acids between TM and the GFP insert ( Figure 1B), while one of their C-terminusdeleted AtGCS1s, in which 13 amino acids following TM were kept, was not functional [11]. This fact may suggest that the 34 residues are critical to precise AtGCS1 localization, even though this hypothesis is not applicable to the -C construct of PbGCS1.

The N-terminus of GCS1 Contains Functional Gamete Fusion Domain(s)
The HAP2-GCS1 domain was recently determined to be made up of highly conserved residues [2,11]. In particular, cysteine distribution in the domain is conserved in all of the GCS1possessing eukaryotes known to date, suggesting that it serves as an indispensable structure for gamete fusion. Besides the C-terminusmodified AtGCS1 constructs, we produced two N-terminus constructs as well, namely GDD and GAA. In GDD, the GFP insert has resulted in a change in the cysteine residue positions in the HAP2-GCS1 domain, resulting in drastically reduced GCS1 function ( Figure 3C). Also in PbGCS1, the deletion of the HAP2-GCS1 domain shows that it is vital for its function (Figures 6C and  6D). On the other hand, GAA, in which the entire N-terminal sequence is separated from TM by the GFP insert, did retain some GCS1 function. There are no cysteine residues between the GFP insert and TM in GAA, and therefore, the spatial N-terminal structure might be, at least partly, maintained and functional, even though the distance between the N-terminus and TM is extensive. Besides, a prediction of GCS1 orientation based on Hidden Markof Modeling and a gamete fusion blocking with N-terminus recognizing antibodies suggest that the N-terminus may be extracellular [11,18]. Taken together, the central functional domain(s) of GCS1 probably resides in conserved N-terminal regions, such as the HAP2-GCS1 domain, and the putative extracellular N-terminus alone is sufficient for gamete fusion.

Incomplete Fertilization in gcs1 GAA Sperm Cells
To date, incomplete double fertilization (i.e. single fertilization), where only one sperm cell fuses with a female partner cell in an ovule, has been reported in various Arabidopsis male gametic mutants and transgenic lines [16,[19][20][21]. In all such lines, the single fertilization occurs in both sperm-egg and sperm-central combinations, resulting in fertilization products of the embryo or the endosperm alone. Also in +/gcs1 GAA plants, both fertilization patterns were observed, (Figures 5E and 5F). When ovules of +/ gcs1 GAA were observed at 24 HAP, sperm nuclei in karyogamy were occasionally detected ( Figure 4H). Since a previous study showed that karyogamy between sperm and egg/central cells is completed within 7-8 HAP [22], such karyogamy at 24 HAP is extremely late. Since wild-type Arabidopsis lines homozygously expressing GAA showed no abnormalities in their seed sets (data not shown), the +/gcs1 GAA phenotypes might not be attributed to neofunctionalization by GAA products. We, therefore, conclude that the original gcs1 phenotype, namely no gamete fusion, was converted into delayed gamete fusion by the introduction of GAA. Nevertheless, the fact that some gcs1 sperm cells were complemented by GAA indicates that slightly delayed gamete fusion is not crucial for double fertilization, while extensively delayed fertilization results in a failure of seed development ( Figure 3C).

Conclusion
GCS1 is highly conserved and putative orthologs have been identified in various eukaryotes, e.g., protists, amoebae and invertebrates, and therefore, its molecular structure and function are important to understanding the common molecular mechanics of fertilization in these organisms. In addition, understanding of malaria parasite fertilization based on GCS1 may be applicable to novel strategies attacking parasites [18,23]. Since GCS1 is a novel transmembrane protein with no known functional domains, it is still difficult to analyze its molecular function during gamete fusion by means of studies using GCS1 alone. If new GCS1-related factors were to be identified, whether they are male or female ones, the understanding of the importance of each GCS1 domain would likely be accelerated. In the present study, we demonstrated that the N-terminus of GCS1 by itself is sufficient for gamete fusion and a highly conserved region, such as the HAP2-GCS1 domain, is critical to successful fertilization in both plants and parasties. This suggests that such a domain possesses a fundamental function indispensable for gamete fusion in GCS1possessing organisms. Further investigations, focusing on the Nterminus by itself, may help achieve the isolation of GCS1-related

Ethics Statement
Studies with experimental animals were approved by Animal Care and Use Committee of Gunma University, and followed guidelines of this committee. The transgenic P. berghei was generated under the guidelines of the recombinant DNA experiments committee of Gunma University. The assigned ID for above experiments is 10-007.

Experimental Materials
The Arabidopsis GCS1 mutant seeds were obtained as previously described. All Arabidopsis plants (Columbia ecotype) were grown in an incubator on a 16-light/8-h dark at 20uC. Plasmodium berghei (ANKA clone 2.34) was intraperitonealy injected into a Balb/c female mouse.

The gcs1 Complementation Assays
Emasculated wild-type pistils were pollinated with pollen from the +/gcs1 transformants described above. The resulting offspring seeds were sown in the same media as in the transformant selection. Surviving and dead seedlings were counted in the plate media.
Observation of Incomplete Double Fertilization in +/gcs1 GAA The +/gcs1 plants were crossed with the transgenic Arabidopsis expressing HTR10-RFP (sperm nucleus marker), and the F 1 +/ gcs1 plants hemizygously expressing the marker were obtained. The F 1 plants were transformed with the GAA construct and the T 1 +/gcs1 GAA plants homozygously expressing HTR10-RFP were selected in the media as above. Their hand-pollinated pistils were disassembled at 24 HAP and the developing ovules were isolated and observed under fluorescence microscopy. The Arabidopsis lines, expressing H2B-RFP and DFWA-GFP driven by the pEC1 and pFWA promoters respectively, were crossed, and the F 2 plants homozygously expressing both markers were recovered. The obtained plants were pollinated with +/gcs1 +/GAA pollen and the fertilized ovules were observed at 30 HAP.

Production of mPbGCS1 Constructs and mPbGCS1-Expressing P. berghei
For production of the PbGCS1 Full construct, two PbGCS1 gene fragments covering a region from its own promoter (0.6 kb) to terminator (1.5 kb), and only the terminator (0.6 kb), were prepared using genomic PCR with two primer sets, PbGCSF(KpnI)/PbGCSR(ClaI) and PbGCS3UTRF(EcoRV)/ PbGCS3UTRR(XbaI), respectively. Each PCR fragment was cloned into each side of the selectable marker gene (TgDHFR-ts) in pBS-DHFR [24]. For production of the -HG, -C and -TM/C constructs, inverse PCR was employed to remove each domain from the PbGCS1 Full construct with the three primer sets, -HAP2/GCS1InvF/-HAP2/GCS1InvR, -CInvF/-CInvR, and -CInvF/-TMR, and the PbGCS1 Full construct as a template. The PCR products were treated with DpnI to digest the PbGCS1 Full template plasmid, and the 59 end was phosphorylated and then self-ligated. The resultant constructs were sequenced and cut by KpnI and XbaI. The linearized constructs were then introduced into the parasite genome by double crossover homologous recombination so that the constructs come under the control of the endogenous PbGCS1 promoter. The parasite transfection, pyrimethamine selection, and dilution cloning were repeated tree times for each construct, as described previously [25]. Three independent parasite clones were obtained from each of the three transfections and used for the experiment. RT-PCR was performed to confirm the successful elimination of the HAP2/ GCS1, C and TM/C domains in each transgenic parasite using the two sets of primers, AF/AR and BF/BR.

Assessment of the Fertility of mPbGCS1-expressing P. berghei
Thin blood films prepared from mice infected with transgenic parasites were stained with Giemsa, and then parasitemia and the female gametocyte ratio were calculated. The infected blood was mixed with fertilization medium and cultured for 16 h. The thin blood films were prepared and the ookinate ratio was calculated. The efficiency of fertilization was expressed as the ratio of females which were fertilized with males and transformed into ookinetes. The assay was repeated three times for each clone. Thus, the assay was repeated 9 times for each transgenic parasite. Figure S1 Precise observation of unfertilized gcs1 GAA sperm cells. (A-C) Detailed observation of an unfertilized +/ gcs1 GAA sperm pair. In the +/gcs1 GAA line, in which the sperm nuclei are labeled with HTR10-RFP, an unfertilized sperm pair is occasionally visible in an ovule. In a visualized embryo sac under differential interference contrast (DIC) microscopy (A), such sperm pairs were detected in the vicinity of female gametes as GFP and RFP signal pairs (B and C, respectively).