The molecular underpinnings of fertility: Genetic approaches in Caenorhabditis elegans

Abstract The study of mutations that impact fertility has a catch‐22. Fertility mutants are often lost since they cannot simply be propagated and maintained. This has hindered progress in understanding the genetics of fertility. In mice, several molecules are found to be required for the interactions between the sperm and egg, with JUNO and IZUMO1 being the only known receptor pair on the egg and sperm surface, respectively. In Caenorhabditis elegans, a total of 12 proteins on the sperm or oocyte have been identified to mediate gamete interactions. Majority of these genes were identified through mutants isolated from genetic screens. In this review, we summarize the several key screening strategies that led to the identification of fertility mutants in C. elegans and provide a perspective about future research using genetic approaches. Recently, advancements in new technologies such as high‐throughput sequencing and Crispr‐based genome editing tools have accelerated the molecular, cell biological, and mechanistic analysis of fertility genes. We review how these valuable tools advance our understanding of the molecular underpinnings of fertilization. We draw parallels of the molecular mechanisms of fertilization between worms and mammals and argue that our work in C. elegans complements fertility research in humans and other species.


| INTRODUCTION
A fundamental process during sexual reproduction, fertilization involves species-specific recognition, adhesion and fusion between the gametes. These processes are thought to be mediated by molecular interactions between the gametes. 1,2 In mammals, the egg coat (also called zona pellucida) contains ZP proteins that are necessary and sufficient to support recognition. 1,3 Following recognition and penetration of the egg coat, several molecules are required for spermegg binding and thus fusion, including the egg surface tetraspanin CD9 and the most recently identified sperm proteins FIMP, SOF1 and TMEM95 (Table 1 and Figure 1A). Among these molecules, the only known receptor-binding pair is the Immunoglobulin (Ig) superfamily member IZUMO1 in the sperm and the Glycosylphosphatidylinositol (GPI)-anchored JUNO in the egg. 4,10 In zebrafish, a GPI-anchored protein called Bouncer is identified as an egg surface receptor that is necessary for species-specific gamete interactions. 13 However, the binding partner for Bouncer remains unknown. In Caenorhabditis elegans, the first molecule required for fertilization was discovered as SPE-9, a transmembrane protein with EGF repeats, on the sperm surface. Mutants of spe-9 were reported in a forward genetic screen in 1988 and cloning of the gene was reported in 1998. 14,17 To date, a total of 12 C. elegans proteins have been found on the sperm and oocyte to mediate sperm-oocyte interactions including an IZUMO1-like molecule SPE-45 (Table 1 and Figure 1B). Yet, among these molecules, no receptor pairs have been identified. What additional molecules are at play and how they interact with one another is unknown.
Forward genetic screens have been a driving force in identifying the genetic regulation of a biological process. [27][28][29][30][31][32][33] Starting with mutants with a phenotype of interest, one can probe the underlying genetic cause and identify complex genetic relationships. Compared to vertebrate models in which forward genetic screens are time-, cost-and labor-intensive, [34][35][36] small model organisms have been especially useful for a forward genetic approach in developmental studies. 37,38 In fact, the majority of the fertilization molecules in C. elegans were identified through forward genetic screens. In this review, we summarize genetic screening strategies that led to identifying those fertility mutants and provide a perspective for future research that uses genetic approaches in C. elegans. We also discuss how new technologies such as next-generation sequencing and genome-editing tools help us advance our understanding of the genetic regulation of fertility.

| C . ELEGANS AS A MODEL TO STUDY REPRODUCTION
C. elegans is a widely used model organism because of its ease of culture, low cost, short life cycle and the availability of genetic tools. 33 These advantages together with their hermaphroditic mode of reproduction make it relatively convenient to isolate mutants by chemical mutagenesis. Temperature-sensitive (ts) mutations generally exist more frequently in C. elegans than in other multicellular organisms. 39 Additionally, the fertility of the animal is inherently ts, higher at permissive and lower at restrictive temperatures. These features make it possible to find and maintain homozygous fertility mutants. C. elegans are transparent and so the whole reproductive tract can be observed in live animals. Unfertilized oocytes are readily distinguishable from unhatched embryos, allowing us to tell fertilization defects apart from embryonic lethality. 40 Hermaphrodites can self-fertilize to produce self-progeny but can also be mated with males to produce out-cross progeny. The hermaphroditic mode of reproduction ensures that sterility in an unmated mutant hermaphrodite is likely due to defects in the gametes and not mating behaviors or copulation. These features make C. elegans a good model to study fertilization. Studying fertility T A B L E 1 Fertilization molecules in vertebrates and worms in C. elegans as a nematode species will also inform our understanding of reproduction in other nematodes, such as parasitic nematodes that cause human disease or cause losses in both plant and animal agriculture. 41 C. elegans fertilization takes place in the spermatheca in an assembly-line fashion. 42 Oocytes move from distal to proximal gonad as they enter prophase of meiosis I. When oocytes get close to the spermatheca, they receive maturation signals from the major sperm proteins (MSP) secreted by the sperm. 43 Matured oocytes are ovulated into the spermatheca, fertilized and then pushed into the uterus where they finish meiosis and start embryogenesis (Figure 2A- 42 Oogenesis starts from young adulthood and continues into later adult stages. Before making oocytes, the germline of a hermaphrodite makes a finite number of spermatids during the last larva stage. The first ovulation pushes the spermatids into the spermatheca, where spermatids undergo a post-meiotic differentiation process called sperm activation or spermiogenesis. 44 Activation transforms spherical and non-motile spermatids into ameboid and motile spermatozoa with pseudopods ( Figure 2C ). 40

| SCREENING FOR FERTILIZATION MUTANTS: STRATEGIES
The first report of a forward mutagenesis screen for fertility mutants in C. elegans, by Hirsh and Vanderslice, came shortly after Sydney Brenner first introduced C. elegans as an experimental model and described its mutagenesis and genetics. 33,52 Hirsh and Vanderslice looked for ts sterile mutants in a classic genetic screen ( Figure 3A). 38 Parental worms (P0) were mutagenized by ethyl methanesulfonate (EMS) and allowed to produce F1 and F2 generations of self-progeny.
Individual F2 worms were singled out and allowed to produce the F3 generation at permissive temperature ( Fer mutants demonstrated the power of mutagenesis screens in dissecting the process of spermatogenesis and offered key insights into fertilization and sperm development.
To characterize spermatogenesis, L'Hernault and coworkers designed a screen for spe mutants. 17,21 They described two types of strategies. One strategy is similar to the one described above. All the generations were grown at restrictive temperatures. They allowed several F1s to be picked or laid and grown on a plate. F2s were examined for the appearance of lots of early oocytes. In the other strategy, they used a starter strain with one or multiple morphological markers.
F I G U R E 2 Hermaphrodite reproductive tract and sperm. A, DIC image of a live worm. Themiddle section of the worm is shown here. Arrow is pointing at sperm in the spermatheca. B, A diagram illustrating the middle section of the reproductive tract. Arrow is pointing the direction at which oocytes move. C, Images of spermatids (left) and spermatozoa (right). Scale bar is 10 μm Mutagenized P0s were crossed with wild-type males. The F1 generation were singled out onto individual plates, and F2 larvae were shifted to restrictive temperatures (Table 2). Plates with lots of early unfertilized oocytes were analyzed for linkage and for fertility when outcrossed with wild-type males. Any mutations that appeared linked with the morphological marker likely affected genes on the same chromosome as the marker. The initial outcross not only helped with establishing linkage but also reduced the number of extraneous mutations thus lowering false-positive rates. With these strategies, they concentrated on Chromosome I and identified 23 ts and nonconditional mutations belonging to 11 complementation groups.
Poisson analysis of mutant frequency suggested that their screen nearly reached saturation for Chromosome I mutations. The phenotypes of these mutants covered various stages of spermatogenesis and sperm functions, including early spermatogenesis, sperm activation, sperm-oocyte interactions and paternal contribution to embryogenesis. 17 Among the mutants, the spe-9 and spe-13 mutants make spermatids that are morphologically normal, can differentiate into spermatozoa, can migrate, but fail to fertilize the oocytes 14,55 and Krauchunas et al. in preparation). This spe-9 class sperm phenotype is seen in both hermaphrodites and males. The gene spe-9 encodes a single-pass transmembrane protein with multiple EGF repeats and is thought to mediate signaling or adhesion with the oocytes. 14,16 Since spe-9 is the first gene found to regulate sperm-egg interactions, it defines the spe-9 class, 56 which now comprises 10 genes.
To better understand sperm-oocyte interactions, the Singson lab developed a screening strategy ( Figure 3B) based on screens that looked for maternal-effect embryonic lethal mutants. [57][58][59][60][61] The goal of this screen is to find ts mutants in which sterility results from defects in sperm-oocyte interactions. To facilitate mutant selection, we use a starter strain that carries a sem-2 mutation and an embryonic gut marker 22 ( Figure 3B). The sem-2 mutation leads to defects in muscles that control egg laying but do not affect vulva opening. 59,62 Fertilized eggs hatch inside of the mother, causing her to form a "bag of worms" and die. 38 In this genetic background, any sterile F2 worms would appear normal and crawl on the plate whereas fertile worms would form a bag of worms and die. The gut lineage marker elt-7p::gfp is turned on early in the embryo and helps select against any mutants that are maternal-effect embryonic lethal. 22 We grow the F2s at restrictive temperature and select candidates that are non-baggers with a GFP-negative uterus ( Ste is widely used to describe a broad spectrum of fertility phenotypes that in most cases do not impact fertilization.
misclassified as embryonic lethal and sometimes discarded. For example, the spe-49 gene was initially named let-479 because the phenotype was thought to be embryonic lethality. 21 Therefore, community mutation collections could be a rich source of uncharacterized fertility mutants.
The screen in the Singson lab described above is being continuously performed in our lab, in the hope that we come across fertilization-defective egg mutants. Each time when we perform the screen, instead of attempting to characterize every single mutant, we prioritize our characterization of mutants based on their phenotypes.
This allows us to focus on understanding their underlying biology and identifying new genes. Meanwhile, we continue the screen in the lab so that we keep adding new mutants that potentially define new genes. This approach is a move away from the traditional strategy of completing a saturation mutagenesis before moving to any molecular analysis.

| NEW PERSPECTIVES FOR A FORWARD GENETIC APPROACH
We have so far discovered 10 proteins that are required for sperm function during fertilization. What binding partners are on the oocyte surface has long been a question in the field. In C. elegans, EGG-1/ EGG-2 is a semi-redundant pair of LDL receptor repeat-containing proteins that are required in the oocytes for fertilization. 25 Originally identified as candidate genes that encode oocyte surface proteins with ligand-receptor binding domains, EGG-1/2 were later shown to not bind to SPE-9 in cultured cells (Singson lab unpublished data).
These observations support the hypothesis that additional molecules on the oocyte surface exist to mediate recognition and adhesion with the sperm. One could argue that the reason our screen has recovered far fewer egg mutants is the tight link between fertilization and egg activation. It is true that oocyte maturation, fertilization and egg activation are a set of precisely regulated, tightly linked and continuous events, which possibly involve some shared genes and genetic regulation. However, defects in one or more of these events should lead to the same oocyte-laying phenotype. Thus, the tight link between these events should not interfere with gene discovery by a screen. Because forward genetic screens remain a productive method in discovering missing pieces of this puzzle, we propose modifying our current screening strategy to put more emphasis on egg-sterile mutants as described above.
Screening for ts mutants of essential genes has been a useful strategy because it allows for easy maintenance of mutants and sometimes offers unique insights of protein functions. 16 However, ts alleles are relatively rare, with some genes not mutable to a ts phenotype. 65 Based on our own and others' observations, only 5% to 10% of sterile mutants are ts. Thus, one reason that we see a lot more ts Spe mutants than egg mutants could be that the egg genes do not tend to mutate to a ts phenotype. Thus, broadening our screening to include non-conditional mutants might facilitate our search for egg-sterile good-looking sperm and oocytes. For example, a germline specific cell membrane marker and/or a histone marker will allow us to observe the morphology of gametes. These markers together with microscopy will also help filter out gametogenesis and embryonic lethal mutants.

| LIMITATIONS OF FORWARD GENETIC SCREENS
Although a powerful and unbiased way to identify genetic regulation of biological processes, a forward genetic method has its own limitations. A blind spot of forward genetic screening is functional redundancy, where paralogous genes have overlapping functions. Loss of one paralog often is not sufficient to cause a phenotype, due to compensation by another paralog. In this case, these genes could be identified only through certain dominant alleles. 63,[67][68][69][70] Multiple examples of redundant genes exist in C. elegans spermatogenesis, 71,72 oogenesis, 67,68 ovulation, 43 fertilization, 25 and oocyte-to-embryo transition 49,50 and these genes were identified through reverse genetic or biochemical approaches. It is estimated that 30% of C.
elegans genome encode proteins with one or more paralogs. [73][74][75] Although it is not known to what extent these paralogs have redundant functions, this level of redundancy poses a challenge in gene discovery with forward genetic approaches.
Other than redundancy, pleiotropy is another potential limitation of forward genetic screens. One gene product may regulate multiple processes, with different timing or in different tissues. Phenotypes of loss of function alleles may represent only one specific function but mask others. An example of this redundancy during C. elegans spermatogenesis is spe-6. Loss-of-function alleles showed that SPE-6 played roles in completing meiosis and organizing and assembling the sperm cytoskeleton MSPs during early spermatogenesis. 76 However, hypomorphic alleles of spe-6 revealed its later role in coordinating sperm activation. 77 Similar to redundancy, pleiotropic genes could be uncovered during a forward genetic screen only by rare and specific alleles.  Another advancement that brings changes to our query into fertilization molecules is in transcriptomics. In C. elegans, germline-enriched and/or sex-biased gene expression profiles have been reported. [82][83][84][85][86][87] Using mutants in the sex-determination pathway that cause the hermaphrodites to only produce sperm or oocytes, these transcriptomic datasets yield lists of genes that are specifically expressed in the sperm vs oocytes. [82][83][84] The availability of transcriptomics data has fueled reverse genetic approaches that overcome some of the limits of genetic screens. Several genes during C. elegans fertilization and egg activation were identified through reverse genetic approaches, such as spe-45 and egg-1/egg-2 that regulate fertilization, and egg-3 and egg-4/egg-5 that regulate egg activation. 24,25,[49][50][51] The improved sample processing methodology and increased sequencing sensitivity means that we can look at global transcription in a finer temporal and spatial scale, even at a single-cell level. [85][86][87] Due to constraints on sequencing sensitivity, some subtle or transient changes in gene expression may be lost. 87 Yet, as the technology evolves, this method may strengthen our ability to discover genes that regulate fertilization and egg activation.
The Crispr-based genome editing tools have expanded our abilities to access protein functions by genetics. 88,89 Crispr-based methods have made it more efficient to generate null and conditional alleles, to engineer customized point mutations, and to tag endogenous proteins. [90][91][92][93] The mouse fertilization molecule SOF1 was identified in an effort to test the functions of a number of testis-specific genes using Crispr-based knockout alleles. 5,94,95 In C. elegans, some groups have reported using Crispr to examine the functions of genes that regulate fertility. 96,97 A few groups have had success in tagging and visualizing proteins that exert functions in the germline. 98,99 Sperm proteins especially those that function at fertilization are usually expressed at low levels that makes visualization challenging.
Crispr-based editing provides a yet another way of localizing proteins in addition to traditional antibody-based fluorescent labeling. Overall this genome editing tool when combined with transcriptomic datasets will allow us to evaluate gene functions, genetic relationships and protein dynamics at an unprecedented level.
Although genetics-based fertility gene discovery has been fruitful, biochemical and proteomic studies are also viable approaches. For C.
elegans, both sperm and oocytes can be isolated in bulk as starting material for biochemical analysis. 100,101 These purifications methods can be used with mutants of C. elegans that only make sperm or oocytes, which will be powerful in determining the sperm or oocyte proteome. Several proteomics studies were published before, some of which generate gene/protein lists that are relevant for our search of fertilization molecules. 72,102,103 In addition to purifying whole sperm or oocytes, isolating subcellular compartments such as the MOs of the sperm for proteomics will provide insights into specific proteins that are transported to the surface of the sperm during activation. Additionally, cultured Drosophila S2 cells can be used as a system to express worm proteins for other biochemical assays. Many C. elegans biologists have a preference for genetic analysis. However, it should be remembered that many well-worn and newly developed biochemical and proteomic methodologies will be important for both additional gene discovery and gaining molecular mechanistic insights into fertilization.
The SPE molecules at the fertilization synapse discovered thus far, likely represent only part of the picture. Mechanistically, how they interact with one another and with potential partners on the oocyte surface is still elusive. Protein interaction studies can help us identify these interactions. As a community resource, our lab has established the SPE interactome using a membrane yeast two hybrid system. 104 This interactome offers some insights into the landscape of the sperm surface. For example, SPE-38 is a four-pass transmembrane protein of the spe-9 class and is required for the correct localization of another component, SPE-41, at the sperm surface. 18,19 We hypothesize that it may play a similar role as the tetraspanin CD9 on the mammalian egg surface, serving as a molecular raft organizing other molecules. 11,12,115 Consistent with this hypothesis, SPE-38 has been shown to interact with multiple SPE proteins in our interactome. 104 As more mutants are discovered and the molecular identities of the genes become known, the interactome will continue to grow and give us the opportunity to generate new hypotheses.

| FERTILIZATION SYNAPSE IN NEMATODE PROVIDES INSIGHTS INTO THAT OF MAMMALS
The spe-45 gene in C. elegans was identified not only in our mutant hunt, but also independently by a reverse genetic approach searching for single-Ig domain containing, sperm-surface molecules like IZUMO1. 24 In addition to identical phenotypes of in both the worm and mouse, chimeric SPE-45 with the mouse IZUMO1 Ig domain expressed as a transgene partially rescued the spe-45 mutant fertility, suggesting conserved functions of the Ig domain. 24,105 The discovery of C. elegans spe-51 as an Ig-like fold containing protein further highlights important functions of Ig domains at the fertilization synapse.
The fact that shared protein domains are utilized to achieve the same reproductive goal in two evolutionarily distant species suggests a deep conservation of protein structures at the gamete interface despite rapid co-evolution of gamete-interacting proteins. 106,107 The complexity of the fertilization synapse on the sperm side in C. elegans implies that a great deal of information is still unknown for all species, which may be even more complex. 2 Recent identification of four sperm surface proteins, SOF1, SPACA6, TMEM95 and FIMP1 in mice supports the complexity of the mouse fertilization synapse. [5][6][7][8] Loss of functions for each of these molecules does not affect the level or localization of IZUMO1, suggesting a hierarchy in the assembly and dynamics of the fertilization synapse. Similarly, in C. elegans, loss of certain components of the fertilization synapse does not necessarily affect the localization of others suggesting differential roles of these molecules during gamete interactions. 19,108 These observations lead us to think that proteins on the sperm side of the synapse may or may not each have their own binding partner on the egg surface, with some of them playing a scaffolding rather than a direct interacting role. In mice, the JUNO-IZUMO1 interaction is not sufficient for sperm-egg fusion, as shown by cell binding assays and structural analyses, suggesting additional molecules are involved. [109][110][111][112][113] In C.
elegans, additional egg surface receptors are also to be identified and these questions will be the focus of future research.

| CONCLUSIONS
Our ongoing journey using C. elegans as a paradigm to understand the fertilization process has complemented our understanding of mammalian fertilization. Our findings also provide future perspective on the control of reproduction in other species including parasitic nematodes.
We argue that robust efforts in a variety of model systems will be the most effective way to understand nature's mysteries of conception.
Forward and reverse genetic approaches to study reproductive processes have been reported for fruit fly, 114 zebrafish 115 and mice. [116][117][118] Although these efforts in these diverse model systems have identified interesting mutants that impact stem cell biology, meiosis and gametogenesis, we hope that they continue to identify key molecules of fertilization. With the advancement of new technologies, our forward screens, complemented by a reverse genetic and biochemical approaches, will give us a better understanding of the protein interactions at the fertilization synapse, and a better picture about fertilization and fertility in C. elegans and other animals including humans.