Golgi associated RAB2 interactor protein family contributes to murine male fertility to various extents by assuring correct morphogenesis of sperm heads

Sperm heads contain not only the nucleus but also the acrosome which is a distinctive cap-like structure located anterior to the nucleus and is derived from the Golgi apparatus. The Golgi Associated RAB2 Interactors (GARINs; also known as FAM71) protein family shows predominant expression in the testis and all possess a RAB2-binding domain which confers binding affinity to RAB2, a small GTPase that is responsible for membrane transport and vesicle trafficking. Our previous study showed that GARIN1A and GARIN1B are important for acrosome biogenesis and that GARIN1B is indispensable for male fertility in mice. Here, we generated KO mice of other Garins, namely Garin2, Garin3, Garin4, Garin5a, and Garin5b (Garin2-5b). Using computer-assisted morphological analysis, we found that the loss of each Garin2-5b resulted in aberrant sperm head morphogenesis. While the fertilities of Garin2-/- and Garin4-/- males are normal, Garin5a-/- and Garin5b-/- males are subfertile, and Garin3-/- males are infertile. Further analysis revealed that Garin3-/- males exhibited abnormal acrosomal morphology, but not as severely as Garin1b-/- males; instead, the amounts of membrane proteins, particularly ADAM family proteins, decreased in Garin3 KO spermatozoa. Moreover, only Garin4 KO mice exhibit vacuoles in the sperm head. These results indicate that GARINs assure correct head morphogenesis and some members of the GARIN family function distinctively in male fertility.


Introduction
Spermatozoa are generated through a specialized biological process known as spermatogenesis.This intricate process occurs in seminiferous tubules of the testis, comprising a series of discrete steps.At the first step, spermatogonia undergo mitosis, transitioning into spermatocytes.Meiosis then occurs to transform diploid spermatocytes into haploid round spermatids.Spermiogenesis follows meiosis and is characterized by sixteen steps in mice.Spermiogenesis is a complicated process that involves nuclear condensation, morphogenesis of the sperm head, and formation of the sperm tail.These events collectively ensure the integrity of spermatozoa and secure the correct interaction of spermatozoa with oocytes.
The acrosome, a distinctive Golgi-derived structure situated at the anterior aspect of the sperm head, plays an important role in sperm-egg interaction [1][2][3].The acrosome reaction, characterized as an exocytosis event releasing the enzymes stored within the acrosome, assists sperm penetration through the zona pellucida (ZP) of oocytes [1,2,4].Simultaneously, the acrosome reaction assists the migration of receptors that facilitate sperm-egg fusion [5][6][7].Spermiogenesis can be divided into 4 phases based on the morphology of the acrosome and nucleus: Golgi phase, cap phase, acrosome phase, and maturation phase [8].Starting from the Golgi phase, the Golgi apparatus actively produces vesicles containing various glycoproteins, and then, multiple vesicles fuse to form a larger vesicle which is called an acrosomal vesicle near the nucleus [9].In the cap phase, acrosomal vesicles spread over the surface and become flattened around the nucleus.After the cap phase, the skirt-like microtubule-based structure, the manchette is formed, which facilitates nuclear elongation during the acrosome phase [10].The nucleus and acrosomal vesicle then further elongate to complete spermiogenesis in the maturation phase.
Teratozoospermia is a major clinical cause of male infertility, which is defined as over 85% of the spermatozoa carrying abnormal morphology in humans [11,12].Moreover, globozoospermia, a subset of teratozoospermia, is specifically characterized by spermatozoa with abnormal acrosome morphology [11].The genetic and molecular mechanisms underlying teratozoospermia are largely unknown, and the mouse knockout (KO) model is used as an outstanding animal model to elucidate the genetic cause.In our previous study, we identified two globozoospermia-related genes: Golgi Associated RAB2 Interactor 1A (Garin1a; Fam71f2) and Garin1b (Fam71f1) [13].Garin1a and Garin1b are predominantly expressed in the testes and KO of Garin1a and Garin1b causes subfertility and infertility in males, respectively, due to abnormal acrosomal morphology [13].Furthermore, the same study showed that GARIN1B binds to both RAB2A and RAB2B which are small GTPases with high amino acid sequence similarities and are involved in vesicle trafficking.In contrast, the functions of other GARIN paralogues, namely GARIN3, GARIN4, GARIN5A, and GARIN5B, in male fertility are not clear although it has been shown recently that Garin2 (Fam71d) is not essential for male fertility in mice [14].
In this study, we generated 5 single KO mice of Garin2, Garin3, Garin4, Garin5a, and Gar-in5b and aimed to elucidate the characteristics of GARINs.we demonstrated that GARIN2, GARIN3, GARIN4, GARIN5A, and GARIN5B are all important for sperm head morphogenesis, along with GARIN3 being vital for acrosome biogenesis and male fertility.

Garin2-5b are expressed predominantly in mouse testes
In mice, there are 7 members of the GARIN protein family, and they all possess a RAB2-binding domain with high sequence similarity of amino acids (Figs 1A and S1A).Phylogenetic trees of mouse GARINs (S1B Fig) and human GARINs (S1C Fig) were generated by comparing amino acid sequences using Clustal Omega [15].The mouse GARIN protein family is well conserved in humans while human GARIN6 is not conserved in mice.We also generated phylogenetic trees for coding sequences of mouse Garins (S1D Fig) and human GARINs (S1E Fig) and found that GARIN5A and GARIN5B were closely related in coding sequences but not in amino acid sequences, suggesting that there may be more non-synonymous substitutions in GARIN5A and/or GARIN5B.We performed RT-PCR using cDNAs obtained from adult mouse tissues to determine the expression pattern of Garin2, Garin3, Garin4, Garin5a, and Garin5b (hereafter referred to as Garin2-5b) and found that all Garin2-5b are predominantly expressed in the testis (Fig 1B).Because the first wave of spermatogenesis in postnatal testis allows the consecutive appearance of spermatogonia, spermatocytes, round spermatids, and elongated spermatids to be analyzed [16], we analyzed the expression of mouse Garin2-5b using cDNA obtained from 10-35 days postnatal testes.All Garin2-5b showed strong signals from 28 days postnatal (Fig 1C), suggesting that GARIN2-5B may function during spermiogenesis after meiosis.These results were confirmed by published single-cell RNA-sequencing (RNA-seq) datasets [17,18], as Garin2-5b are expressed predominantly in testes and spermatids.In contrast, Rab2a and Rab2b are expressed in several tissues and do not exhibit predominant expression in late spermiogenesis (S1F and S1G Fig).

Knockouts of Garin2-5b impact male fertility with varying severity
To examine the function of GARIN2-5B in vivo, we generated 5 single KO mouse lines of each Garin2-5b with the CRISPR/CAS9 system.For Garin2, Garin3, Garin4, and Garin5b, we used pairs of guide RNA (gRNA) targeting almost the entire open reading frames (ORFs).The numbers of electroporated zygotes, transplanted embryos, pups born, and pups confirmed with a large deletion are summarized in S1 Table .The genotype of these mouse lines was confirmed by genomic PCR with indicated primers (S2A-S2C and S2E Fig) .Further, by performing Sanger sequencing of the PCR products, we confirmed 24,044 bp, 2,755 bp, 1,663 bp, 13,436 bp deletions on Garin2, Garin3, Garin4, and Garin5b, respectively.For Garin5a, we used one gRNA targeting exon 2 of the ORF (S2D Fig) and obtained mice with a 7 bp deletion, which caused the frameshift mutation.This 7 bp deletion was confirmed by Sanger sequencing of the genomic PCR product.Genotyping of Garin5a KO mice was performed by digesting the

PLOS GENETICS
PCR product with the restriction enzyme BsrBI.No obvious abnormalities were observed in the development or behavior of all Garin2-5b KO mice.
We then caged wild-type (WT) female mice with individual male Garin2-5b KO or WT mice to test their in vivo fertility.WT females mated with WT males delivered 8.2 ± 2.7 pups on average from every coitus event.The fertility of Garin2 -/-and Garin4 -/-males seemed to be lower but were not statistically significant (6.5 ± 4.0 and 6.6 ± 3.8 pups/plug, respectively) compared to WT males.In contrast, Garin5a -/-and Garin5b -/-males were subfertile with 3.0 ± 4.1 and 3.8 ± 3.6 pups/plug, respectively.Furthermore, Garin3 -/-males completely failed to sire pups although plugs were confirmed (Fig 2A).We also housed Garin3 -/-males with superovulated WT females, collected eggs 8 hours after the mating, and confirmed that Garin3 mutant spermatozoa cannot fertilize eggs in vivo (S3A and S3B Fig).
To find out the cause of impaired male fertility, we observed the testis and epididymis of Garin2-5b KO males.The testicular weights of Garin2-5b KO males were comparable with WT males (Fig 2B).Moreover, no overt abnormalities were found in testis or epididymis sections of Garin2-5b KO mice (Fig 2C).We then conducted in vitro fertilization (IVF) to further analyze the fertilizing ability of Garin2-5b KO spermatozoa.In the presence of cumulus cells and intact zona pellucida (ZP), the fertilization rates of each Garin2-5b KO spermatozoa were significantly decreased compared to WT mice (Fig 3A).Similarly, even though we removed cumulus cells before insemination, the fertilization rates of each Garin2-5b KO spermatozoa were lower than those of WT spermatozoa (Fig 3B).However, the decreased fertilization rates of each Garin2-5b KO spermatozoa were rescued by removing both the cumulus cells and ZP before insemination (Fig 3C).These results indicate that the ability to penetrate the ZP was impaired in all Garin2-5b KO spermatozoa in vitro.

Garin2, Garin3, Garin5a, and Garin5b KO mice exhibit impaired sperm motility
We next examined sperm motility by computer-assisted sperm analysis (CASA) after 10 and 120 minutes incubation in a capacitation medium as defective motility could result in impaired ZP penetration [19].Using the same CASA system, minor motility defects were found in spermatozoa with impaired ZP penetration [20].For both 10 and 120 minutes, the percentages of motile spermatozoa were significantly decreased only in Garin2 KO spermatozoa compared to WT spermatozoa (S4A and S4B Fig) .Despite the percentages of motile spermatozoa being comparable in other Garin KO mice compared to WT mice, kinetic parameters such as straight line velocity (VSL), curvilinear velocity (VCL), and average path velocity (VAP) were reduced in Garin3, Garin5a, and Garin5b KO mice after 10 minutes of incubation (Fig 3D -3F).After 120 minutes of incubation, VSL was lower in Garin5b KO spermatozoa, VCL was lower in Garin2, Garin3, and Garin5b KO spermatozoa, and VAP was lower in Garin2 and Garin5b KO spermatozoa (Fig 3G -3I).Although Garin2, Garin3, Garin5a, and Garin5b KO spermatozoa exhibited impaired motility, kinetic parameters of Garin5b KO spermatozoa (with the worst mean values) were still comparable with or higher than those of inbred C57BL/6J WT spermatozoa (S4C and S4D Fig), suggesting that the impaired ZP penetration cannot be explained only by diminished sperm motility.

Garin2-5b KO male mice show abnormal sperm head morphology
To further analyze the cause of impaired fertility in Garin2-5b KO males, we observed spermatozoa obtained from the cauda epididymis and found that sperm heads of all Garin2-5b KO males exhibited abnormal morphology (Fig 4A).WT sperm heads showed a distinctively sharp and elongated hook; however, in Garin2-5b KO males, with visual examination, 88.4%, 91.5%, 92.4%, 47.0%, and 17.9% of the sperm heads exhibited bluntness in the hook regions, respectively (Fig 4A and 4B).Notably, in Garin4 KO, approximately 33.1% of spermatozoa displayed one or multiple vacuoles in their heads (Figs 4A and S5).
To further compare the sperm head morphology of WT and Garin2-5b KO mice quantitatively, we conducted elliptic Fourier descriptors (EFDs) and principal component (PC) analysis on sperm heads to quantify the morphology [21], which have the advantage of being able to eliminate orientation errors caused by interference, image size, and starting point of the In vivo fertility test and histological analyses of Garin2-5b -/-males.(A) Pups per plug for WT males and Garin2-5b (Garin2, 3, 4, 5a, and 5b) KO males.A vaginal plug was considered as a sign of coitus.For each genotype, 5 males were tested.Pups/plug of Garin2-5b -/-males were compared with WT males (One-way ANOVA, c, P < 0.001).(B) Testis weight of WT and Garin2-5b -/-males.The difference between testis weight of WT and Garin2-5b -/-males was not significant (One-way ANOVA, P > 0.05).(C) PAS-staining of testis and epididymis sections, from WT and Garin2-5b -/-males.https://doi.org/10.1371/journal.pgen.1011337.g002whereas PC2 focuses on the length and direction of the sperm hook.By separating the sperm head morphology by PC1 and PC2, we found that for the PC1 axis, data extracted by EFDs forms 2 clusters around -0.1 and +1.0, in contrast, data was evenly distributed from -0.2 to +0.2 on the PC2 axis, indicating that the morphology of Garin2-5b and WT spermatozoa was distinguished by PC2 but not PC1 (Fig 4D ).Furthermore, all Garin2-5b KOs exhibited significantly lower values on PC2 (Fig 4E ), indicating that each KO of Garin2-5b have shortened sperm hooks.These results suggest that the ZP penetration failure can be due to abnormal sperm head morphology in Garin2-5b KO mice.In particular, Garin3 KO mice showed lower PC2 value with less variation (Fig 4E

Garin3 KO spermatozoa display abnormal acrosomal morphology
We have been shown that each KO of Garin1a and Garin1b results in abnormal sperm head and acrosomal morphology [13].Therefore, we observed the acrosome of Garin2-5b KO spermatozoa with PNA staining.Although the majority of the acrosome from Garin2-5b KO spermatozoa seemed normal, some of the acrosomes of Garin3 KO spermatozoa appeared to expand (Fig 5A ), consistent with Garin1a and Garin1b KO spermatozoa [13].We quantified the PNA-positive area in WT and Garin2-5b KO spermatozoa and found that the acrosome area of Garin3 KO spermatozoa was significantly larger than that of WT spermatozoa (Fig 5B ), suggesting that abnormal head morphology of Garin3 KO spermatozoa may be caused by impaired acrosome biogenesis.In addition, we observed nuclear vacuoles in the Hoechst staining of the Garin4 KO spermatozoa (Fig 5A ), consistent with the phase contrast observation (Fig 4A ).
Because acrosomal morphology was abnormal in Garin3 KO spermatozoa, we measured acrosome reaction rates after 10 minutes and 4 hours incubation in a capacitation medium, and also after treatment with the Ca 2+ ionophore A23187 [23].There were no significant differences in acrosome reaction rates between WT and Garin3 KO spermatozoa in these conditions (Fig 5C ), supporting the idea that the ZP penetration failure is caused by abnormal sperm head morphology but not impaired acrosome reaction.

Deletion of Garin3 affects sperm head plasma membrane proteins
To further analyze the effects of Garin3 deletion, which caused the severest impairment in male fertility among Garin2-5b KOs, we performed mass spectrometry (MS) analyses of mature spermatozoa collected from cauda epididymis.We found that not only GARIN3 was completely absent in Garin3 KO spermatozoa, but also 18 proteins such as ADAM4, ADAM6A, and ADAM6B were significantly downregulated in Garin3 KO spermatozoa (S2 Table ).We performed gene ontology (GO) analysis on downregulated proteins (Fig 6A ) and found that proteins categorized in the sperm head plasma membrane and male gonad development were significantly downregulated in Garin3 KO spermatozoa.It has been shown that the expression levels of ADAM4 and ADAM6 are dramatically reduced in Adam2 KO and Adam3 KO spermatozoa, and ADAM6 forms a complex with ADAM2 and ADAM3 [24].Therefore, we analyzed the amounts of ADAM proteins for those antibodies that were available for us and found that ADAM2, ADAM3, and ADAM32 were downregulated in Garin3 KO spermatozoa (Fig 6B).As a negative control, the amount of IZUMO1, a transmembrane protein localized in the acrosomal membrane, was not affected.These results suggest that not only head morphology but also the amounts of plasma membrane proteins were affected in Garin3 KO spermatozoa.Moreover, in Garin2-5b KO spermatozoa, only Garin3 KO spermatozoa showed downregulated ADAM3 (S6A Fig) , suggesting the functional differences of GARIN3 among GARIN2-5B.
Because it has been shown that in the absence of ADAM2, ADAM3, or ADAM6, the ZP binding ability is severely impaired [25][26][27], we analyzed ZP binding ability of Garin3 KO spermatozoa (Fig 6C).Although the ZP binding ability of Garin3 KO spermatozoa is significantly reduced, 5.5 spermatozoa on average could still bind to the ZP, which is about 31.3% of the binding ability compared to WT spermatozoa (Fig 6C and 6D).These results suggest that the combinatorial effect of impaired sperm head morphology and ZP binding ability resulted in ZP penetration failure in Garin3 KO spermatozoa.

GARIN3 binds to both inactive and active forms of RAB2A/B
To investigate the cause of abnormal acrosome morphology in Garin3 KO spermatozoa, we analyzed proteins involved in acrosome formation.We found that the amounts of GOPC, ZPBP1, and SPACA1 were not affected in Garin2-5b KO testes (S6B Fig) .Our previous study on another GARIN protein family member, GARIN1B, demonstrated that N-linked glycosylation of SPACA1 in the Garin1b KO testis was impaired [13].In contrast, WB using an anti-SPACA1 antibody showed normal bands for Garin3 KO testis (S6B Fig) .This result suggests that although GARIN1B and GARIN3 are members of the same protein family and deletion of either gene caused sperm head malformation and infertility, their functions could be different.
RAB2A/B is a GTPase that is only active in the GTP-bound form and is inactive in the GDP-bound form [28]. RAB2A/B can be mimicked into constitutively active (CA; GTPbound) and constitutively negative (CN; GDP-bound) forms, by introducing the Q65L and S20N amino acid substitutions, respectively [29].One feature of GARIN1B is that it only interacts with the CA form but not with the CN form of RAB2A/B.In contrast, co-IP analyses using HEK293T cells showed that GARIN3 could bind to both CA and CN forms of RAB2A/B (Fig 7A and 7B).Moreover, further analyses showed that, among GARIN2, GARIN4, GAR-IN5A, and GARIN5B, GARIN5A could bind to both CA and CN forms of RAB2A although GARIN2/4/5B could bind to the CA form strongly and bind to the CN form weakly (S6C Fig) .Moreover, GARIN5B could bind to both the CA and CN forms of RAB2B, but only GARIN2/ 4/5A binds to the CA form of RAB2B preferentially (S6D Fig) .These results suggest that GARIN3, GARIN5A, and GARIN5B have different characteristics among all members of the GARIN protein family.
Previously, we have shown that GARIN1B is localized in the Golgi-apparatus when expressed in COS-7 cells [13].In contrast, we found that GARIN3 was localized in the cytoplasm of COS-7 cells (Fig 7C).However, when RAB2A or RAB2B was co-expressed with GARIN3, GARIN3 was localized in the Golgi-apparatus, suggesting that RAB2A/B recruits GARIN3 to the Golgi-apparatus.It has also been shown that RAB2B recruits GARIN5A to the Golgi apparatus [30].These results also indicate that GARIN3 and GARIN5A have different characteristics from that of GARIN1B.
Sperm head formation is important for male fertility [33][34][35].Because each mutant of Garin2-5b causes sperm head malformation, we performed EFDs and PC analysis to quantify head morphology.EFD was first introduced in 1982, and it is used to describe the morphology by generating a series of ellipses that approximate the shape [36], which has been applied to evaluate various biological samples [21,37].The analyses showed that PC2 values that focused on the length and direction of the sperm hook were lower in all Garin2-5b KO mice compared to WT mice.Indeed, Garin4 KO spermatozoa show the lowest average PC2 value among all of the Garin2-5b KO spermatozoa; however, the standard deviation was also the largest.Spermatozoa with high PC2 values in Garin4 KO mice may be responsible for better fertilizing ability in both in vitro and in vivo compared to Garin3 KO mice.EFDs and PC analysis can be a powerful tool for quantifying sperm head morphology, which may also be used to diagnose teratozoospermia, a condition of infertile men who exhibit abnormally shaped spermatozoa.Garin1b KO mice exhibit sperm head malformation, abnormal N-linked glycosylation of SPACA1, defective acrosome biogenesis with impaired acrosome reaction, and male sterility [13].In contrast, the mutation in Garin3 causes sperm head malformation and male sterility, but does not cause severely impaired acrosome biogenesis or impaired acrosome reaction.Moreover, Garin3 KO spermatozoa exhibited normally processed SPACA1.Instead, the amounts of ADAM2, ADAM3, and ADAM6, which are crucial for sperm-ZP binding and male fertility [25][26][27], and the amounts of other ADAM family members, ADAM4 and ADAM32, were decreased in Garin3 KO spermatozoa.Among these ADAM family members, it has been suggested ADAM2-ADAM3-ADAM4, and ADAM2-ADAM3-ADAM6 may form complexes [38].GARIN3 may interact with some proteins of the complexes as well as ADAM32 to regulate the processing of these membrane proteins.Notably, despite the decreased amount of ADAMs, Garin3 KO spermatozoa still maintain 31.3%ZP binding ability compared to WT spermatozoa.Because a transgenic study has shown that a small amount of ADAM3 is able to fully rescue the fertility of Adam3 KO male mice [39], reduced amounts of ADAMs alone may not be sufficient to exterminate male fertility, and the shape of the sperm head may also be involved.
RABs are involved in membrane trafficking and vesicle fusion/fission by interacting with numerous proteins including its effector which interacts with the active form of RABs [40,41].Indeed, GARIN1B and GARIN3 possess a RAB2-binding domain in the N-terminus and bind to RAB2A/B, however, in the WB study, we show that GARIN3 interacts with both the CA and CN forms of RAB2A/B in HEK293T cells (Fig 7A and 7B), and GARIN3 is present in mature spermatozoa (Fig 6B).Given the fact that GARIN1B only interacts with the CA form of RAB2A/B and was absent in mature spermatozoa [13], GARIN3 may function differently from GARIN1B.This difference is possibly due to the varied length and low similarity of the regions other than the RAB2-binding domain.It should be noted that we analyzed the interaction and localization of GARIN2-5B and RAB2A/B using a heterologous expression system as we have no functional GARIN2-5B antibodies for IP or immunofluorescence.Further analyses of their interaction and co-localization in testes would reveal the functional relationship of GARIN2-5B and RAB2A/B.Furthermore, studying the binding affinity of GARINs and the CA and CN forms of RAB2A/B using surface plasmon resonance or biolayer interferometry would reveal the binding specificity of each GARINs with RAB2A/B in more detail.
Our previous study indicates that RAB2A/B is localized in the Golgi apparatus of the round spermatids and localized in the acrosome of the elongating spermatids and that GARIN1B may assist RAB2A/B in fulfilling their vesicle transportation from the Golgi apparatus to the acrosome [13].GARIN3 may have similar functions to some extent since Garin3 KO mice show morphological abnormalities in the acrosome and sperm heads, although not as severe as in Garin1b KO mice.However, in the present study, we show that RAB2A/B recruited GARIN3 to the Golgi-apparatus (Fig 7C ), and the amounts of membrane proteins, namely ADAMs, were decreased in Garin3 KO spermatozoa.Taken together, GARIN3 may also function in a RAB2A/B-dependent manner in the Golgi apparatus for membrane protein trafficking, vesicle anchoring, and/or processing.KO of Garins affects male fertility and sperm head morphogenesis to various extents (Fig 7D ), notably, Garin4 KO sperm displays vacuoles in their head.Therefore, other than GARIN3, some members of the GARIN protein family may also possess unique functions that are different from other GARINs.
Among GARIN2-5B, we demonstrated that GARIN3 is essential for male fertility, and all GARIN2-5B are indispensable for sperm head morphogenesis.GARINs are all conserved in humans and GARIN2 is found under positive selection in Europeans [42].Therefore, GARINs may play important roles in humans, and inactivation of them may lead to teratozoospermia.Moreover, it is known that nuclear vacuoles are present in the spermatozoa of fertile and infertile men and related to male fertility potential, however, the origin and the formation process of nuclear vacuoles are not well characterized [43].Further analysis of GARINs may lead us to a better understanding of sperm head morphogenesis, teratozoospermia, and male infertility.

Ethics statement
Mouse experiments were approved by the Animal Care and Use Committee at the Research Institute for Microbial Diseases, Osaka University (#Biken-AP-H30-01).

Animals
Mice were purchased from CLEA Japan (Tokyo, Japan) or Japan SLC (Shizuoka, Japan).All mice were maintained under specific-pathogen-free conditions with ad libitum feeding, under an artificial 12-h light/12-h dark cycle.All gene-modified mice generated in this study will be available through either the RIKEN BioResource Research Center or the Center for Animal and Development (CARD), Kumamoto University.

Protein extraction of HEK293T cells, mouse testes, and spermatozoa
HEK293T cells were harvested with 1% Triton lysis buffer (1% Triton X-100, 50 mM Tris-HCl pH 7.5, 150 mM NaCl) with 1% protease inhibitor cocktail (Nacalai Tesque, Kyoto, Japan) and incubated for 2 hours on ice.Mouse testes and epididymis were dissected by forceps and ophthalmologic scissors under a dissection microscope.Then, seminiferous tubules inside the testes were pulled out and homogenized in 1 mL urea lysis buffer (6 M urea, 2 M thiourea, 2% sodium deoxycholate) with a homogenizer.Mature spermatozoa were extracted from the cauda epididymis and resuspended in 1 mL phosphate-buffered saline (PBS).After centrifuge at 300 × g at 4˚C for 10 minutes, the supernatant was discarded and then resuspended with 0.1 mL urea lysis buffer.Both testes and sperm lysates were incubated on ice for 2 hours.The lysate was then centrifuged at 15,000 × g at 4˚C for 20 minutes and the supernatant was subjected to immunoprecipitation or Western blotting.
B6D2F1 females at 7-weeks-old were superovulated with 0.1 mL CARD HyperOva (Kyudo, Saga, Japan) and followed by injection of 5 units human chorionic gonadotropin (hCG; ASKA Pharmaceutical, Tokyo, Japan) after 48 hours.Females were then mated with B6D2F1 males and sacrificed to collect the 2 pronuclei (2PN) oocytes.Electroporation was performed to introduce the complex of CAS9, CRISPR RNA (crRNA), and trans-activating crRNA (tracrRNA) into 2PN oocytes [46].Next, the oocytes were cultured in potassium simplex optimization medium (KSOM) to the 2-cell stage [47] and then transplanted into the oviducts of 0.5-day pseudopregnant ICR females.The F0 pups were naturally delivered or obtained by Cesarean section.For the F0 and F1 generations, the deletion of coding sequences was confirmed by genomic PCR and Sanger sequencing.Genotyping of F2 and later generations was only performed by genomic PCR.For Garin5a -/-mice, the genotyping was performed by BsrBI digestion of the PCR product.KOD Fx Neo DNA polymerase was used for PCR.The primers used in genomic PCR are shown in S5 Table.

In vivo fertility test
Each KO or WT male was caged with three 7-week-old females for 8 weeks.During these 8 weeks, vaginal plugs were checked as a sign of successful coitus, and the number of pups born was counted.After 8 weeks, males were removed from the cage, then the number of pups was counted for 3 weeks as the females could give birth to the last litters.

In vitro fertilization (IVF)
Female mice were superovulated with CARD HyperOva and hCG as previously stated before IVF.Male mice were dissected and mature spermatozoa were collected from cauda epididymis, then spermatozoa were released in Toyoda Yokoyama Hoshi (TYH) medium and incubated for 2 hours in the condition of 37˚C, 5% CO 2 , to induce capacitation [48].During these 2 hours, oocytes were extracted from the oviductal ampulla, and a portion of them was treated with 330 μg/mL of hyaluronidase (Wako, Osaka, Japan) or 1 mg/mL of collagenase (Sigma-Aldrich) to remove cumulus cells or the ZP, respectively.Then, spermatozoa were incubated with the oocytes at a concentration of 2 × 10 5 sperm/mL in 0.1 mL TYH drops.After 6 hours of insemination, 2PN embryos were counted and the rate of 2PN formation was determined as the fertilization rate.

Acrosome reaction analysis
Cauda epididymal spermatozoa were released in TYH drops and incubated at 37˚C, 5% CO 2 , for 10 minutes or 4 hours.After incubation, a portion of spermatozoa was spread onto glass slides and air-dried.Another portion was treated with A23187 (Merck Millipore, Darmstadt, Germany) in the TYH drops to induce the acrosome reaction for 10 minutes, then spermatozoa were spread onto glass slides and air-dried.4% paraformaldehyde (PFA) in PBS was added to the glass slides for 10 minutes.After 3 quick washes with PBS, the slides were blocked with a blocking buffer consisting of 5% bovine serum albumin (BSA) and 10% goat serum in PBS for 1 hour at room temperature, then treated with anti-IZUMO1 antibody [49] in blocking buffer overnight at 4˚C.After 3 washes with PBS, the samples were incubated with fluorophore-conjugated secondary antibody for 1 hour at room temperature, washed 3 times with PBS, then incubated with 0.02% Hoechst 33342 (Thermofisher) in PBS for 10 minutes.Following 3 washes, slides were mounted using Immu-Mount (Thermofisher) before being observed with an Olympus BX-53 microscope (Tokyo, Japan).

Sperm-ZP binding assay
As described above in the methodology of IVF, cumulus-free oocytes were inseminated with capacitated spermatozoa at a concentration of 2 × 10 5 sperm/mL in 0.1 mL TYH drops.After 30 minutes, oocytes were removed from TYH drops and fixed with 0.1% PFA in TYH for 10 minutes.Oocytes were then moved to a new TYH drop, and then images were captured using an Olympus IX73 microscope.

Immunocytochemistry of COS-7 cells
COS-7 cells were fixed with 4% PFA in PBS for 15 minutes, then blocked with 3% BSA in PBS blocking buffer.After blocking, slides were incubated with primary antibodies diluted in blocking buffer overnight at 4˚C.After 3 quick washes with PBS, slides were probed with fluorescence protein conjugated secondary antibodies diluted in blocking buffer for 1 hour at room temperature.After secondary antibody incubation, the slides were incubated with 0.02% Hoechst 33342 in PBS for 10 minutes, washed 3 times with PBS, and then mounted using Immu-Mount.Slides were observed with a Nikon Eclipse Ti microscope.

Statistical analysis
Data are presented as mean ± standard deviation (SD).Data consisting of two groups was statistically analyzed with the unpaired two-tailed Student's t-test.For data consisting of more than two groups, data was analyzed by one-way ANOVA.Statistical significance was defined as P < 0.05 (a, P < 0.05; b, P < 0.01; c, P < 0.001).

Fig 1 .
Fig 1. GARINs show predominant expression in mouse testes and interact with RAB2A/B.(A) Schematic representation of GARIN family members.The RAB2-binding domain is shown in blue.(B) RT-PCR for GARIN family members was performed with cDNA obtained from mouse tissues.Actb was used as a loading control.BR, brain; TH, thymus; LU, lung; HE, heart; LI, liver; SP, spleen; KI, kidney; TE, testis; OV, ovary.(C) RT-PCR was performed with cDNA obtained from postnatal testes with primers for GARIN family members.Actb was used as a loading control.(D) Schematic representation of the domains of RAB2A/B used in Fig 1E-I.The PA tag is shown in green and the Small GTP-binding protein domain is shown in magenta.(E-I) Co-IP of GARINs and RAB2A/B.Anti-FLAG and anti-PA antibodies were used to detect GARINs-FLAG and PA-RAB2A/B, respectively.In Fig 1E, the input PA bands were saturated because the signals were strong.Anti-ACTB antibody was used to detect ACTB as an endogenous control.https://doi.org/10.1371/journal.pgen.1011337.g001

Fig 6 .
Fig 6.KO of Garin3 led to the loss of ADAMs and aberrant ZP-binding ability.(A) GO analysis of proteins significantly downregulated in Garin3 KO spermatozoa (S2 and S6 Tables).MF, CC, and BP are abbreviations for molecular function, cellular component, and biological process, respectively.(B) Western blotting analyses revealed the amounts of ADAM2, ADAM3, and ADAM32 decreased in Garin3 KO spermatozoa.IZUMO1 and ACTB were detected as endogenous controls.(C) Representative image of ZP-binding analyses of WT and Garin3 KO spermatozoa.(D) Numbers of ZP-bound spermatozoa per egg.Garin3 KO spermatozoa exhibited lower ZP-binding ability compared to WT spermatozoa.For Garin3 -/-and WT mice, 3 males were used for the ZP-binding assay (Student's t-test, c, P < 0.001).https://doi.org/10.1371/journal.pgen.1011337.g006

Fig 7 .
Fig 7. Differences in the interaction between GARIN1B/3 and RAB2A/B.(A and B) Co-IP of GARIN1B/3 and constitutively active (CA)/constitutively negative (CN) forms of RAB2A/B.GARIN3 interacts with both CA and CN forms of RAB2A/B, while GARIN1B only interacts with the CA form of RAB2A/B.(C) Localization of GARIN3-FLAG (green) with or without PA-RAB2A/B (yellow) in COS-7 cells.The Golgi apparatus and nuclei were stained with anti-GOLGA2 (magenta) and Hoechst33342 (Blue), respectively.(D) Summary of phenotypes observed on male reproduction for all Garins KO male mice.https://doi.org/10.1371/journal.pgen.1011337.g007