Metalloproteases in Gonad Formation and Ovulation

Abstract

Changes in expression or activation of various metalloproteases including matrix metalloproteases (Mmp), a disintegrin and metalloprotease (Adam) and a disintegrin and metalloprotease with thrombospondin motif (Adamts), and their endogenous inhibitors (tissue inhibitors of metalloproteases, Timp), have been shown to be critical for ovulation in various species from studies in past decades. Some of these metalloproteases such as Adamts1, Adamts9, Mmp2, and Mmp9 have also been shown to be regulated by luteinizing hormone (LH) and/or progestin, which are essential triggers for ovulation in all vertebrate species. Most of these metalloproteases also express broadly in various tissues and cells including germ cells and somatic gonad cells. Thus, metalloproteases likely play roles in gonad formation processes comprising primordial germ cell (PGC) migration, development of germ and somatic cells, and sex determination. However, our knowledge on the functions and mechanisms of metalloproteases in these processes in vertebrates is still lacking. This review will summarize our current knowledge on the metalloproteases in ovulation and gonad formation with emphasis on PGC migration and germ cell development.

Introduction

The formation of a functional ovary and the release of a fertilizable oocyte during ovulation is essential for animals to reproduce. Ovulation is a dynamic tissue remodeling process, which involves cleavage of extracellular matrix (ECM), rupture of basal membranes, follicular layers, and capillary vessels, release of mature oocytes, repairing lesions; and remodeling of ECM and ovarian structures (Tsafriri, 1995; Duffy et al., 2019). Formation of a functional ovary or testis is also a vigorous remodeling process, whereby the mature organ develops from a bipotential gonad via primordial germ cell (PGC) migration, proliferation, apoptosis, transformation, and development (Ungewitter & Yao, 2013; Spiller et al., 2017; Kossack & Draper, 2019). These processes are regulated by various signaling molecules and multiple metalloproteases including matrix metalloproteases (Mmp), a disintegrin and metalloprotease (Adam) and a disintegrin and metalloprotease with thrombospondin motif (Adamts), and their endogenous inhibitors (tissue inhibitors of metalloproteases, Timp). These zinc-dependent metalloproteases play a critical role in the cleavage of ECM proteins and releasing of membrane-bound signaling molecules and receptors. Excellent reviews covering various facets of ovulation have been published over the years and readers are recommended for their examination (Ohnishi et al., 2005; Curry & Smith, 2006; Russell & Robker, 2007; Richards, 2018; Duffy et al., 2019; Takahashi et al., 2019). Metalloproteases express in a wide range of organs and tissues, including testis and have been shown to be critical to semen and sperm production, and the integrity of testicular structures and sperm in animals (Baumgart et al., 2002; Choi et al., 2015; Belardin et al., 2019). Due to limited space, readers are referred to these reviews (Le Magueresse-Battistoni, 2008; Cho, 2012; Asgari et al., 2021). The present review will focus on functions and expression of individual metalloproteases in ovarian follicle development and ovulation, conservation and difference among various species, and future directions. The review is centered around a few vertebrate models largely due to availabilities of literature. Whenever possible, I will include invertebrates and other animal species where involvements of metalloproteases in ovulation or gonad formation have been reported. I will use nomenclature conventions of fish for depictions of proteins (e.g., Adamts1), transcripts/genes (e.g., adamts1) to avoid potential confusions.

Metalloproteases are synthesized in the latent form (zymogen or proenzyme) that require in situ activation for their activities to allow physiological processes such as ovulation to occur. The activity of each metalloprotease depends on 1) de novo synthesis; and 2) in situ activation. Except for a couple of gelatinases, most metalloproteases lack a sensitive bioassay for examining their activities in vivo or in vitro. Almost all cells/tissues contain multiple metalloproteases, which also make bioassay and interpretation of in vivo results difficult. In addition, majority of metalloprotease knockouts have no clear reproductive phenotypes (Table 1). A few in vitro characterizations such as cleavage assay on the enzyme activation using recombinant proteins in test tubes are very useful for identification of substrates/ligands and cleavage sites but have limited implications in term of physiological functions and activation of each metalloprotease in native cells/tissues/processes in vivo. So, it is not surprising that majority of studies so far were focused on expression changes of metalloproteases. To include a broad range of metalloprotease candidates for their possible involvements in ovulation and gonad formation, I assessed the expression changes of each metalloprotease candidate during these two processes and attempted to include activation and knockout results when they are available. At the conclusion, I will describe limitations and suggest future directions.

Table 1.

A partial list of major reproductive processes (PGC migration, folliculogenesis, fertility, or ovulation) likely regulated or affected by representative metalloproteases in female animals.

Metalloprotease	Affected Reproductive Processes	Reproductive phenotypes in KO (knockout)/KD (knockdown)	Species	Positive Reference	Negative Reference
Adam8	Ovulation	Normal fertility in KO mice	Mice, Zebrafish	Sriraman et al., 2008; Liu et al., 2017; Liu et al., 2018.	Zack et al., 2009*; Kelly et al., 2005*.
Adam17	Ovulation	KO mice die perinatally	Cow	Sayasith and Sirois, 2015
Adamts1	Folliculogeneis, Ovulation	Subfertile & annovulation & defects in folliculogenesis in KO mice	Brown trout, Cattle, Horse, Human, Mice, Rat, Zebrafish	Robker et al., 2000; Shindo et al., 2000*; Espey et al., 2000; Thai and Iruela-Arispe, 2002; Boerboom et al., 2003; Mittaz et al., 2004*; Brown et al., 2006 Freimann et al., 2005; Brown et al., 2006; Shozu et al., 2005*; Fortune et al., 2009; Brown et al., 2010*; Crespo et al., 2013; Sayasith et al., 2013; Mittaz et al., 2004; Rosewell et al., 2015; Baker and Van Der Kraak, 2021	Peluffo et al., 2011 (rhesus monkey); Liu et al., 2018 (zebrafish)
Adamts4	Ovulation	No any morphological abnormalities in KO mice	Mice, Rhesus monkey	Richards et al., 2005; Peluffo et al., 2011	Glasson et al., 2004*; Stanton et al., 2005*
Adamts9	PGC migration, Folliculogeneis, Female to male sex reversal, Ovulation	No PGC migration and infertility in KO C elegans. PGC mis-migration and embryonic lethal in KO Drosophila. Infertile, subfertile, annovulation, folliculogenesis defects, female to male sex reversal in KO female zebrafish. Fertile in KO zebrafish males. Embryonic lethal in KO mice.	C. elegans, Cattle, Drosophila, Monkey, Human, Zebrafish	Blelloch et al., 1999*; Blelloch and Kimble, 1999*; Nishiwaki et al., 2000*; Fortune et al., 2009; Peluffo et al., 2011; Ismat et al., 2013*; Rosewell et al., 2015; Wissing et al., 2014; Willis et al., 2017; Liu et al., 2017; Liu et al., 2018; Liu et al., 2020; Carter et al., 2019*; Carver et al., 2021*.
Adamts15	Ovulation	Male infertile but normal in KO female mice	Medaka, Rhesus monkey	Peluffo et al., 2011	MJI ref ID J:211773*
Adamts16	Testiclar development	Male sterility, small testis, lack sperm, loss germ cells in KO rats. Male fertile but with reduced testis in KO mice. XY female with complete gonadal dysgenesisin or ambiguous genitalia in human variants	Human, Mice, Rat	Abdul-Majeed et al., 2014*; Barseghyan et al., 2018; Livermore et al., 2019*.
Cathepsin L	Ovulation	High postnatal mortality. Some mutant males have both normal and atrophic seminiferous tubules and reduced sperm production in KO mice.	Mice	Robker et al., 2000	Roth et al., 2000*; Wright et al., 2003*.
Mmp1	Ovulation, Folliculogeneis	No reproductive defects reported	Chicken, Drosophila, Macaque, Human	Chaffin and Stouffer, 1999; Peluffo et al., 2011; Zhu et al., 2014; Rosewell et al., 2015	Foley et al., 2014*; Deady et al., 2015*.
Mmp2	Ovulation, Folliculogeneis	Subfertile & anovulation in KD Drosophila. Normal fertility in KO mice or KO zebrafish	Brook trout, Cat, Chicken, Ciona intestinalis, Cow, Human, Medaka, Mice, Monkeys, Rat, Zebrafish	Puistola et al., 1986; Curry et al., 2001; Curry et al., 1992; Bagavandoss, 1998; Reich et al., 1991; Chaffin and Stouffer, 1999; Liu et al., 1999; Imai et al., 2003; Jo et al., 2004; Ogiwara et al., 2005; Baka et al., 2010; Nagaraja et al., 2010; Crespo et al., 2013; Hagiwara et al., 2014; Deady et al., 2015; Fujihara et al., 2016; Matsubara et al., 2019; Kehoe et al., 2021; Wolak and Hrabia, 2021.	Itoh et al., 1997*; Riley et al., 2001; Hayashidani et al., 2003; Riley et al., 2004; Lind et al., 2006a; Sessions et al., 2009; Kok et al., 2015* Liu et al., 2018
MMP7 (Stromelysin-3)	Ovulation	Normal fertility in KO mice	Macaque	Chaffin and Stouffer, 1999	Wilson et al., 1997 *
Mmp9	Ovulation, Folliculogeneis	No reproductive defects in KO mice or KO zebrafish	Brown trout, Cat, Chicken, Ciona intestinalis, Cow, Human, Mare, Pig, Rats, Rhesus macaque, Zebrafish	Duncan et al., 1998; Hrabia et al., 2019; Kim et al., 2014; Riley et al., 2001; Curry et al., 2001; Goldberg et al., 1996; Zhao and Luck, 1996; Peluffo et al., 2011; Crespo et al., 2013; Rosewell et al., 2015; Fujihara et al., 2016; Matsubara et al., 2019; Kim and Yoon, 2020; Kehoe et al., 2021.	Vu et al., 1998*; Itoh et al., 1999*; Robker et al., 2000; Young and Stouffer, 2004; Lind et al., 2006a; Portela et al., 2009; Silva et al., 2020*.
Mmp10	Ovulation	No reproductive defects in KO mice	Rat, Rhesus monkey, Human	Lind et al., 2006a; Peluffo et al., 2011; McCord et al., 2012.	Kassim et al., 2007*.
Mmp11	Ovulation	Normal fertility in KO mice	Human	Lind et al., 2006a; McCord et al., 2012	Masson et al., 1998*; Hägglund et al., 1999
Mmp14 (Mt1-Mmp)	Ovulation, Folliculogeneis	Normal fertility in KO mice & KO zebrafish	Brown trout, Hamster, Medaka, Macaque, Rat, Human	Liu et al., 1999; Salverson et al., 2008; Crespo et al., 2013; Hagiwara et al., 2014; Puttabyatappa et al., 2014; Vos et al., 2014; Ogiwara et al., 2015.	Holmbeck et al., 1999*; de Vos et al., 2018*.
Mmp15 (Mt2-Mmp)	Ovulation	Normal fertility in KO mice	Brown trout, Medaka	Crespo et al., 2013; Hagiwara et al., 2014; Ogiwara et al., 2015; Ogiwara and Takahashi, 2019; Hagiwara et al., 2020.	Szabova et al., 2010 *
Mmp19	Ovulation	Normal fertility & increased gonad fat in KO mice.	Mice, Rat, Rhesus monkey	Hägglund et al., 1999; Jo and Curry, 2004; Peluffo et al., 2011; Nalvarte et al., 2016.	Pendás et al. (2004)*; Beck et al., 2008*.
Mmp20	Ovulation	Normal fertility in KO mice	Human	Bhyan et al., 2019	Bartlett et al., 2004*.
Mmp23	Ovulation	No defects in KO mice	Rat, Pig	Ohnishi et al., 2001; Ohnishi et al., 2005; Niu et al., 2013.
Plg/Plau	Ovulation	Decreased ovulation (25%) in double KO (Plat & Plau) mice, but normal in single KO mice	Brown trout, Cow, Human, Medaka, Mice, Sheep, Rat, Rhesus monkey.	Smokovitis et al., 1988; Tsafriri et al., 1989; Peng et al., 1993; Carmeliet et al., 1994*; Leonardsson et al., 1995; Colgin and Murdoch, 1997; Hägglund et al., 1999; Murdoch et al., 1999; Murdoch and McDonnel, 2002; Liu et al., 2004; Ohnishi et al., 2005; Curry and Smith, 2006; Peluffo et al., 2011; Crespo et al., 2013; Hagiwara et al., 2014; Ogiwara et al., 2015.	Leonardsson et al., 1995*; Ny et al., 1999*; Liu et al. (2006c)*
Timp2	Ovulation, folliculogenesis	No reproductive defects in KO mice	Brown trout, Chicken, Human, Mice.	Hägglund et al., 1999; Ogiwara et al., 2005; Lind et al., 2006b; Crespo et al., 2013; Hrabia et al., 2019.	Caterina et al., 2000*, Wang et al., 2000*; Pei et al., 2018*; Charlton-Perkins et al. (2019)*

Notes: reference with * is a study using knockout approach. No expected phenotype from knockouts is likely due to that same substrate is targeted by several metalloproteases. Loss of one or two metalloproteases is likely to be compensated in the knockouts.

Matrix metalloproteases (Mmps) are a family of zinc-dependent proteases, part of the Metzincins superfamily, can cleave extracellular proteins and signaling molecules. Most Mmps are secreted proteases, and a few are membrane anchored proteases. In addition, Mmps also can act intracellularly (Bassiouni et al., 2021). Mmps are involved in modifying matrix structure, growth factor availability and cell surface signal molecules. They then in turn affect development, growth, cellular differentiation, proliferation, and apoptosis (Page-McCaw et al., 2007). The first member, Mmp1, was discovered in 1962 by Jerome Gross and Charles Lapiere, when they observed enzymatic activity of tadpole tail in degradation of collagen matrix (Gross & Lapiere, 1962). There are 28 related enzymes belonging to Mmps in vertebrates of which 24 genes are found in humans (Murphy & Nagase, 2008). In zebrafish, 26 protease genes have been identified as Mmp-orthologs (Huxley-Jones et al., 2007; Wyatt et al., 2009). Each mature Mmp protein is initially synthesized as an inactive form (pro-enzyme) with three common domains: pro-peptide, catalytic and a hemopexin-like C-terminal domain (Nagase et al., 2005; Cui et al., 2017; Fig. 1). The pro-peptide domain contains a cysteine residue that interacts with the catalytic zinc in the active site, preventing its binding to the substrate and consequently keeping the Mmp inactive, and therefore is designated as “The Cysteine Switch” (Van Wart & Birkedal-Hansen, 1990). Once pro-peptide domain is proteolytically removed, Mmp becomes active (Nagase et al., 1990; Fig. 1). Several endogenous inhibitors including four tissue inhibitors of metalloproteases (Timp-1, Timp-2, Timp-3, and Timp-4), α2-Macroglobulin, and β-Amyloid precursor protein could also inhibit activities of Mmps via binding of hemopexin domain at c-terminal of Mmps (Murphy, 2011). These endogenous tissue inhibitors also inhibit activities of Adam or Adamts described below. Therefore, activity of each metalloprotease is regulated no less than three different levels: 1) regulation of transcription by growth factors, cytokines, and hormones; 2) activation of synthesized mature but inactive pro-enzyme (i.e. pro-domain cleavage); and 3) removal of their specific inhibitors (i.e., Timps). In some cases, Timp not only inhibits activity of individual metalloprotease, but also forms binding complex with the metalloprotease to facilitate the docking and cleavage of metalloprotease by other metalloproteases (Lu et al., 2004).

Figure 1. Schematic illustration of major domains of Mmp, Adam and Adamts.

Each mature metalloproteinase protein is synthesized in a latent form (proenzyme) that require in situ activation. Signal peptide is not shown. Activity and binding to the substrates of each mature metalloprotease is inhibited by the prodomain and a cystine switch (in the case of Mmp or Adam) and bindings of endogenous tissue inhibitors of metalloproteases (Timps, not shown), whereas Adamts in general lack the cystine switch. Cleavage of cystine and prodomain and removal of Timps are required for the activation of these metalloproteases.

Members of Adams family are single-pass transmembrane proteins and belong to the Metzincins superfamily. The metalloprotease domain of Adams is similar to Mmps metalloprotease domain, but Adams also contain a unique integrin receptor-binding disintegrin domain. Therefore, Adams were named for these two functional domains (a disintegrin and metalloprotease) (Black & White, 1998; Schlondorff & Blobel, 1999). The first member of Adam family, Adam1, was discovered as a sperm surface protein important for sperm-egg membrane fusion in the fertilization of guinea pig (Primakoff et al., 1987). The Adam family has 20 Adam members in human, whereas 22 adam genes have been identified in zebrafish (Huxley-Jones et al., 2007; Bahudhanapati et al., 2015). All mature Adam proteins share several domains including a prodomain, a metalloprotease domain, a disintegrin domain, a cysteine-rich domain, an EGF-like domain, a transmembrane domain, and a cytoplasmic tail (Black & White, 1998; Schlondorff & Blobel, 1999; Fig. 1). The Adams family has been implicated in the control of membrane fusion, cytokine and growth factor shedding, cell migration, cell adhesion, cell-cell and cell-matrix interactions, and signal transduction (Black, 2002; Liu et al., 2006). As Adamts described below, there is no sensitive bioassay for analyzing activity or activation of Adam in vivo.

The Adamts family is a subgroup of secreted zinc metalloproteases with several distinct domains separated from classical Mmp and Adam families (Tang and Hong, 1999; Apte, 2004; Porter et al., 2005). Members of the Adamts family share several distinct protein modules, including a pro-peptide domain, a metalloprotease domain, a disintegrin domain, a cysteine-rich region, and several repeats of thrombospondin type 1 (TSP1) motif. Some Adamts members have a unique C-terminal module (Fig. 1). The presence of TSP1 motifs is unique to Adamts family and is not found in Adam or Mmp family. Thrombospondin is a major constituent of platelet α-granules but is also secreted by a wide variety of epithelial and mesenchymal cells. Levels of thrombospondin are correlated with developmental changes in developing embryos and response to injury in adults (Bornstein,1992). Unlike Mmps or Adams, members of Adamts in general lack the ‘cysteine switch’ that controls activation of the pro-enzyme (Fig. 1), though Adamts15 seems to have the “cystine switch” (Kelwick et al., 2015). Interaction of propeptide domain with metalloprotease domain or binding of Timps to Adamts inhibits activities of the pro-enzyme (Kelwick et al., 2015). The first member, Adamts1, was cloned in mice and cattle as a gene suggested to be associated with inflammation and development (Colige et al., 1997; Kuno et al., 1997). The Adamts family has 19 highly structurally and sequentially conserved members in vertebrates (Brunet et al., 2015). Adamts enzymes cleave large proteoglycans such as aggrecan, versican, brevican and neurocan. They have been demonstrated to have important roles in connective tissue organization, coagulation, inflammation, arthritis, angiogenesis and cell migration (Apte, 2004; Kelwick et al., 2015).

Metalloproteases in germ cell migration and gonadal formation

Gonad formation is a remarkable tissue remodeling process. All vertebrates first form a bipotential gonad before adopting a male or female fate (McLaren, 1988; Koopman et al., 1991; Gubbay et al., 1992; Hawkins et al., 1992; Polanco and Koopman, 2007). The formation of a functional gonad begins with long-distance migration of PGCs to the presumptive gonad (Raz, 2002; Pearson et al., 2016; Piprek et al., 2018). In mammals, PGCs are specified and segregated early in gastrulation. This occurs at around six days post coitum (dpc) in mice (Lawson et al., 1999). Approximate four days later (around 10dpc), the gonadal primordium starts to develop from bilateral thickenings of the coelomic epithelium (Kusaka et al., 2010) and by 11.5 dpc proliferating PGCs have settled into the genital ridge (Harikae et al., 2013). Testis’s fate is determined by the expression of Sry, which initiates the differentiation of Sertoli cells and sequester PGCs inside testis cords by 12.5 dpc in mice (McLaren, 1988; Koopman et al., 1991). In the absence of the Sry gene, the competing ovarian signaling pathways will develop bipotential gonad into ovaries (Gubbay et al., 1992; Hawkins et al., 1992).

Zebrafish PGCs are also set aside early during gastrulation. PGCs complete their migration before 24 hours post fertilization (hpf) and migrate to the presumptive gonad located along the coelomic wall (Hartung et al., 2014; Kossack and Draper, 2019; Carver et al., 2021; Figs. 2&3). During the first week post fertilization (wpf), these PGCs (~30) are kept in a quiescent state and numbers of PGC do not change. In the following two weeks (2–3 wpf), germ cells and somatic gonad cells are actively proliferating and develop into an ovary-like bipotential gonad containing stage IA perinuclear-stage and stage IB oocytes. Two-types of stage IA divisions have been observed in medaka (Saito et al., 2007). Type I divisions generate individual daughter cells, while Type II divisions amplify and generate multiple daughter cells within a cyst. These stage IA oocytes are still connected to each other via an incompletely divided cell membrane, are forming multiple cyst-like structure, and appear in groups surrounded by pre-granulosa cells. As oocyte grow, some cysts are going to break down into individual stage IB follicle equivalent to primordial/primary follicles in mammals. Stage IB follicle will form when each oocyte is surrounded by single layer of granulosa cells (Takahashi, 1977; Saito et al., 2007; Tzung et al., 2015; Fig. 2), while some stage IA will go through programmed cell death and loss during the processes (Pepling and Spradling, 1998; Kloc et al., 2004; Saito et al., 2007; Lei and Spradling, 2013; Fig. 2). Thereafter, gonadal differentiation and primary sex determination start around three weeks when some immature oocytes (stage IA and IB follicles) begin to undergo apoptosis in presumptive males, whereas immature oocytes (stage IA & IB follicles) continue to grow (into stage II, III) and proliferate in presumptive females (Uchida et al., 2002). The transitioning phase of testis development and primary sex determination is typically over by ~ 30 days post fertilization (dpf), when majority of oocytes have been cleared from gonads that already differentiate as testes (Uchida et al., 2002; Fig. 2).

Figure 2. Schematic illustration of conserved processes for folliculogenesis and programmed cell death of ovarian follicles in animals.

Metalloproteases induce changes of intrafollicular structures and/or signaling which can lead to survival or programmed cell deaths of ovarian follicles in all stages of folliculogenesis. Major events or processes of oogenesis/folliculogenesis, and times of occurrence in zebrafish, mice and human are illustrated. For simplicity, only three representative metalloproteases are used as examples. Certain transcriptional or growth factors are well known to function in these processes (not shown). Adamts9 is expressed in the early germ cells. Adamts9 knockout (adamts9^−/−) caused female-to-male sex conversion in adult zebrafish (see Carter al., 2019 for detail). Under normal conditions, an excessive programmed cell death of ovarian follicles is required for male sex determination in zebrafish that occurs around 4 weeks post fertilization (wpf) in early juveniles (around the red Stop sign in the figure). Thereafter, the gonadal sex is maintained throughout the life in wildtype (adamts9^+/+) zebrafish. However, disproportionate loss of ovarian follicles during the development, in the juveniles or adults all could lead to females-to-males sex conversion in zebrafish (see Dranow et al., 2013; Dai et al., 2015; Tzung et al., 2015 for detail). In our Adamts9 knockout (adamts9^−/−) zebrafish, unwarranted losses of ovarian follicles were still evident in the ovaries collected from middle-juvenile stage of zebrafish at eight-week post fertilization (wpf). A red arrowhead (in image D) indicates a leftover huge hole likely due to loss of a germ cell. A white arrowhead indicates that a stage IB oocyte is undergoing programmed death (image D). (A-C), Representative confocal images of Adamts9 expression in the early germ cells of gonads from an established transgenic zebrafish line Tg (adamts9:EGFP, Carver et al., 2021). A). a representative image of a pair of gonads from four wpf zebrafish; B). a part of six wpf ovary; C). a stage IB oocyte at six wpf (also see Carver et al., 2021 for additional detail). Similar expression of native Adamts9 protein in the germ cells of early stages was also confirmed by immunostaining using specific antibodies (data not shown). (D & E), Representative confocal images of ovaries at eight wpf. D), a representative image of Adamts9 knockout (−/−) ovary from adamts9^−/− zebrafish at eight wpf; E), a representative image of a normal ovary from wildtype (+/+, adamts9^+/+) sibling at 8wpf, in which stage I were developed further into stage II oocytes. In contrast, all follicles in Adamts9 knockout were arrested at early stage of stage IB. (F). Representative ovaries from 3-month old adult Adamts9 knockout (−/−) or wildtype (+/+) sibling, which were raised in enriched environments with low fish density and enhanced feeding. Enrichment completely rescued minor deficiencies in growth and development, but still could not rescue the deficiency in folliculogenesis and fertility. Ovaries in Adamts9 knockout (−/−) remain underdeveloped, Adamts9 knockout females are anovulatory and infertile, while ovaries in their siblings of Adamts9 wildtype (+/+) developed normally and ovulated. White arrowheads indicate ovulated ooctyes. Scale bars, A&B, 200 μm; C, 10 μm; D&E, 50 μm; F, 2mm. The apoptosis carton within the figure was modifed from an original carton with a permission from Dr. Vinita Bharat at Stanford who was the original creator of the carton.

Figure 3. Delay in primordial germ cells (PGCs) migration in Adamts9 knockout.

However, Adamts9 knockout (adamts9^−/−) has no effect on the survival and proliferation of PGC during early development in zebrafish (see Carver et al., 2021 detail).

It is evident that migration of PGCs, degradation of unnecessary parts of the bipotential gonads, and formation of testes or ovaries require various signaling molecules and modification of ECM. Germ cells and gonadal somatic cells not only contribute to the content of the ECM but can also receive signals from the ECM through mechanical and chemical means, which are essential for cells to respond to their environments and key for gonadal sex development and differentiation (Fröjdman et al., 1989; Mackay, 2000). There is some evidence that metalloproteases are expressed in developing gonads in human (Robinson et al., 2001), mice (Chen et al., 2007; Díez-Torre et al., 2013; Piprek et al., 2018), chicken (Huh & Jung, 2013), and zebrafish (Mazzoni & Quagio-Grassiotto, 2020). The secretion of Mmp2, Mmp9 and four Timps was found in developing testes and ovaries in human fetuses. Mmp1, Mmp2, Mmp9 and Timps were localized to the interstitium and the tubules of the testis. Mmp9 and Timp4 were abundant in both Sertoli cells and gonocytes. In the ovary, all four Timps and Mmp1, Mmp2 and Mmp9 were localized to the oogonium/oocyte cytoplasm in the gonads of human fetuses (Robinson et al., 2001). Higher expression of Mt1-Mmp (Mmp14), Mmp2, Mmp11, Mmp9, and Timp were also found in migrating PGCs and genital ridge somatic cells in mice (Díez-Torre et al., 2013). Piprek and colleagues studied the expression of ECM enzymes during the sexual differentiation in mice. They found that most of these enzymes were expressed in supporting and interstitial/stromal cells, and a few enzymes were expressed in the germ cells in mice (Piprek et al., 2018). In zebrafish, at least three Mmps were found to be expressed in the somatic cells that were located ahead of the PGCs in the migration route after organogenesis and maintained throughout the mesentery but not expressed in the PGCs in zebrafish. Once PGCs reached the gonadal ridge, both the PGCs and somatic cells expressed Mmps (Mazzoni & Quagio-Grassiotto, 2020). Several matrix metalloproteases (Mmps) had also been implicated in testis development in chicken (Huh and Jung et al., 2013). Treatment of Mmps inhibitors caused accumulation of ECM and led to cells dispersion and disappearance of testis cords, whereas Mmps activator 4-aminophenylmercuric acetate (APMA) caused ECM loss and a complete disruption of the gonad structure in mouse gonads. In addition, these inhibitor or activators also increased apoptosis and the loss of germ cells, decreased Sertoli cell markers Sox9 and Amh, but did not cause sex reversal in mice (Piprek et al., 2019). However, functional studies of metalloproteases in gonad formation are still rare. A few natural mutations have been identified affecting ECM leading to defects in folliculogenesis, gonadal development, or sex determination, presumably due to 1) these natural mutations could not be passed to offspring due to embryonic lethality or parents’ infertility; 2) compensation from other members of metalloproteases.

In the knockout of Adamts9 ortholog, gon-1 in C. elegans, germ cells did not migrate, and the gonad developed as a disorganized mass of somatic and germ line tissues (Blelloch et al., 1999; Blelloch & Kimble, 1999; Nishiwaki et al., 2000). Gon-1 helped migration of distal tip cells by degrading extracellular matrix components. In Gon-1 mutants, the adult gonad was severely disorganized with no arm extension and no recognizable somatic structure. The developmental defects in gon-1 mutants were limited to the gonad. Other cells, tissues, and organs developed normally in C. elegans (Blelloch et al., 1999; Blelloch & Kimble, 1999; Nishiwaki et al., 2000). In Drosophila, the knockout of Adamts9 ortholog, AdamtTS-A caused the mis-migration of collective cells, including germ cells (Ismat et al., 2013). These studies of Adamts9 orthologs in C. elegans and Drosophila suggest this protease may release a signal or clear a path for PGC to migrate correctly in invertebrates. However, partly due to embryonic lethality in the knockouts of Drosophila and mouse models (Enomoto et al., 2010; Ismat et al., 2013; Nandadasa et al., 2015; Tharmarajah et al., 2018), the underlying mechanisms of Adamts9 in PGC migration, gonad development and formation is still unclear.

We found that Adamts9 was expressed in the germ cells of stage I immature follicles, but the expression was greatly reduced or disappeared in the germ cells of the growing or mature follicles (stages III or V) in zebrafish (Carver et al., 2021; Fig. 2). During the embryonic development, zygotic expression of adamts9 increased significantly around eight somite stage (~10 hpf) in wildtype zebrafish. A few hours later, we found a delay in PGC migration at 15 and 24 hpf in the Adamts9 knockout (adamts9^−/−) zebrafish (Carver et al., 2021; Fig. 3). In wildtype siblings, all PGCs completed their migration to gonadal ridge around 24hpf (Weidinger et al., 2002; Carver et al., 2021; Fig. 3), whereas PGCs would take 48 hours to migrate to the gonadal ridge in Adamts9 knockout zebrafish (Carver et al., 2021; Fig. 3). Our results suggest a conserved function of Adamts9 in germ cell migration among vertebrates and invertebrates. Our results also suggest that Adamts9 is not essential in the germ cell migration in the zebrafish as demonstrated in C. elegans (Blelloch et al., 1999; Blelloch & Kimble, 1999; Nishiwaki et al., 2000). The role of Adamts9 in germ cell migration in zebrafish is likely to be shared by other proteases since members of metalloproteases are dramatically expanded in vertebrates (Brunet et al. 2015).

It is well established that the number of germ cells is important for gonadal development and sex determination in zebrafish (Dranow et al., 2013; Dai et al., 2015; Tzung et al., 2015). A threshold number of germ cells is required for ovarian development. The depletion of germ cells or ovarian follicles in the development or the adult would cause female-to-male conversion (Dranow et al., 2013; Dai et al., 2015; Tzung et al., 2015). Intriguingly, most Adamts9 knockout adult zebrafish developed as males (Carter et al., 2019; Figs 2&4). Therefore, we hypothesized that the unusual development of ovaries and severe male biased sex ratio found in Adamts9 knockout zebrafish was due to deficiencies in the germ cell migration, apoptosis, proliferation of PGCs or germ cells during primary sex determination. Surprisingly, we found Adamts9 had no effects in the numbers of germ cells during PGC migration (Carver et al., 2021). Thus, the prospect of PGC migration defects as a possible cause for sex reversal in Adamts9 knockout is low. Further research in our lab identified another surprising fact: Adamts9 knockout has no effects on sex ratio after the primary sex determination, but significantly reduced the number and size of stage IB follicles in Adamts9 knockout at 5 wpf (right after primary sex determination) and affected long-term viability of stage IB follicles in zebrafish (Fig. 2; Carver and Zhu, unpublished). Therefore, the likely cause of severe male sex ratio biases in Adamts9 knockout adult zebrafish is not due to PGC migration or primary sex determination, but due to defects in early folliculogenesis, which lead to reduced viability of ovarian follicles, programmed germ cell death, and finally sex reversal in adult zebrafish. The processes and signaling that control folliculogenesis, especially in early stages of development, are highly conserved among different species (Fig. 2). Therefore, our Adamts9 knockout provide a new opportunity for examining functions and mechanisms of metalloproteases in folliculogenesis, germ cell development, follicle losses and sex reversal in animals. Elucidation of underlying mechanisms of Adamts9 in these processes and additional proteases in PGCs migration are ongoing studies in the lab.

Figure 4. A proposed model for metalloproteases’ actions in ovarian follicles.

I used Adamts9 as an example for illustration. Metalloproteases can promote or inhibit intrafollicular signaling via cleavage of ECM proteins including cell surface receptors and/or their ligands. Regulation ECM and intrafollicular signaling determines survival and programmed cell deaths of ovarian follicles.

Metalloproteases in ovulation

Ovulation is a well-studied physiological process that releases a fertilizable oocyte from surrounding follicular cells. This process is initiated by luteinizing hormone (LH) that triggers multiple signaling pathways including progestin and prostaglandin signaling, and finally activates downstream metalloproteases, eventually leading to changes in ECM and rupture of follicular layers (Richards, 2018; Takahashi et al., 2019). The components of ECM have been studied extensively in the ovaries of various mammalian species and a few fish species using in situ, immunohistochemistry and PCR (Irving-Rodgers & Rodgers 2006, Berkholtz et al. 2006, Curry & Smith 2006; Thomé et al. 2010; Kato et al. 2010). Various ECM proteins including collagen, aggrecan, versican, fibronectin and laminin were found in the ovary of various mammalian and fish species (Rodgers et al. 2003; Horiguchi et al., 2008; Kato et al., 2010; Takahashi et al., 2019). Collagen I, IV, versican and laminin were thought to be major constituents of ECM proteins in ovaries of mammals and fish (Lind et al. 2006b; Santos et al. 2008; Thomé et al. 2010; Kato et al. 2010). Regulation of the ECM is critically important not only for the regulation of ovarian cell morphology and release of mature oocytes, but also for paracrine and endocrine signaling, differentiation, proliferation, and apoptosis of gonad somatic cells and germ cells (Woodruff and Shea, 2007). At ovulation the follicular basal lamina and extracellular collagens were degraded, presumably by metalloproteases (Smith et al., 1999; Curry & Smith, 2006). The involvement of metalloproteases including members of Mmp, Adam, and Adamts families, have been examined in various species (Robker et al., 2000; Richards, 2005, 2018; Brown et al., 2006; Sriraman et al., 2008; Peluffo et al., 2011; Takahashi et al., 2013, 2019; Duffy et al., 2019). Most of our knowledge on the candidates and involvements of metalloproteases are from studies of hormonal regulations or changes of these metalloproteases during ovulation in various animal species. It is important to note that I will focus on the most likely metalloproteases that may have critical roles in the ovulation in various species, and to discuss whether their roles in the ovulation are conserved across different species.

Adamts1 in ovulation

Proteoglycans such as aggrecan and versican were identified as main substrates of Adamts1 (Kuno et al., 2000; Rodriguez-Manzaneque et al., 2002; Rodríguez-Baena, 2018). Versican isoforms were the main proteoglycans regulated by follicle stimulating hormone (FSH) and testosterone but not by nuclear progestin receptor (Pgr) in rat and mice ovaries (Russell et al., 2003). Adamts1 could also cleave thrombospondin-1 and thrombospondin-2 (Lee et al., 2006), and ligands of the epidermal growth factor (EGF) receptor such as pro-HBEGF (heparin-binding EGF-like growth factor) or pro-amphiregulin (Liu et al., 2006). Novel substrates such as syndecan 4, TFPI-2, semaphorin 3C, nidogen-1, −2, desmocollin-3, dystroglycan, mac-2, gelatin and TGF-α were also identified (Satz-Jacobowitz & Hubmacher, 2021). Adamts1 was implicated in inflammatory processes (Kuno et al., 1997; Rodríguez-Baena, 2018). Studies using Adamts1 knockout mice showed that this protease was important for normal growth, organogenesis and fertility (Shindo et al., 2000; Mittaz et al., 2004; Brown et al. 2010; Table 1).

Expression studies of adamts1 have suggested the involvement of Adamts1 in ovulation and fertility in mammals. Transcripts of adamts1 were upregulated in the preovulatory follicles of several mammalian species including mice (Robker et al., 2000; Shindo et al., 2000; Shozu et al., 2005), human (Freimann et al., 2005), horse (Boerboom et al., 2003; Brown et al., 2006), and cattle (Fortune et al., 2009; Sayasith et al., 2013, Table 1). This upregulation was partly dependent on the expression of Pgr in the granulosa cells (Robker et al., 2000). However, increase of adamts1 transcript was only observed after ovulation in rhesus monkey, which suggests Adamts1 may not play a role in ovulation for them (Peluffo et al., 2011). The studies of Adamts1 in non-mammalian species is limited. A recent study in zebrafish found that adamts1 transcript increased significantly in the preovulatory follicular cells. However, the expression of adamts1 transcript was not affected in pgr^−/− zebrafish in comparison to those in their wildtype sibling. The adamts1 transcripts were highest in stage I immature follicles and gradually decreased as follicles grew, which suggests additional roles for Adamtsl in early folliculogenesis (Liu et al., 2018). A more recent study also confirmed that progestin did not stimulate the expression of adamts1 transcripts in zebrafish follicles (Baker & Van Der Kraak, 2021). Interestingly, adamts1 expression was upregulated by prostaglandins (PGE2) specifically via a prostaglandin receptor subtype (EP4) (Baker & Van Der Kraak, 2021).

Knockout of Adamtsl in mice suggests an important role of this protease as a downstream effector of Pgr in ovulation. Homozygous Adamtsl knockouts were subfertile, producing litters four-to five-fold less than control littermates, partially due to failed rupture of some large follicles (Shozu et al., 2005). It should be noted that the subfertility was observed in the global Adamtsl knockout mice. Retarded growth and follicular cell development were also observed in this global Adamtsl knockout mice, which also likely contributed to the observed subfertility (Thai and Iruela-Arispe, 2002; Shozu et al., 2005). Nevertheless, Adamtsl knockout mice were subfertile and had less severe phenotypes than Pgr knockout mice, which could not ovulate and therefore had no litters. This indicates that other Pgr regulated proteases may also contribute to ovulation.

Adamts9 in ovulation

In COS-l cells transfected with human adamts9, this protease could proteolytically cleaved bovine versican and aggrecan core proteins at the Glu^44l–Ala⁴⁴² bond of versican Vl and the Glu^l77l–Ala^l772 bond of aggrecan, respectively (Somerville et al., 2003). Recently, Adamts9 was also found to bind fibronectin dimers and fibrils directly through multiple sites in both molecules. An Adamts9 cleavage at Gly2196-Leu2197, which was also target of other proteases, was identified (Wang et al., 2019). In Adamts9 deficient mice, reduced cleavage of versican was observed (Kern et al., 2010; Dubail et al., 2014). Aggrecan, versican and fibronectin are currently considered to be the main substrates of Adamts9 (Somerville et al., 2003; Wang et al., 2019; Satz-Jacobowitz and Hubmacher, 2021). As more novel substrates have been identified for Adamts1, it is likely that many new substrates of Adamts9 will be discovered in the future.

Adamts9 is another member of Adamts family likely to be critical for ovulation, as Adamts9 was regulated by LH and progestin, and dramatically increased its expression during ovulation both in fish and mammalian species (Peluffo et al., 2011; Wissing et al., 2014; Liu et al.,2017, 2018, 2020; Table 1). Transcripts of adamts9 were found to be upregulated in preovulatory follicles or granulosa cells following hCG treatment in macaque and human (Peluffo et al., 2011; Wissing et al., 2014). A synthetic Pgr agonist, R5020, could reverse the inhibitory effect of mifepristone (RU486) on the LH induced expression of adamts1 and adamts9 mRNA in granulosa cells of cattle (Fortune et al. 2009; Willis et al., 2017). In zebrafish, we found that adamts9 was the downstream target of LH and progestin. Expression of adamts9 in ovarian follicles was regulated by LH and progestin both in vivo and in vitro. Transcripts and proteins of Adamts9 increased dramatically in preovulatory follicular cells during ovulation, and the expression of Adamts9 proteins and transcripts were significantly low in Pgr knockout zebrafish (Liu et al., 2017, 2018, 2020; Carver et al., 2021).

The knockout of Adamts9 ortholog, gon-1 in C. elegans survived normally, but led to infertility mainly due to defects in primordial germ cells migration and formation of non-functional gonads (Blelloch et al., 1999; Blelloch & Kimble, 1999; Nishiwaki et al., 2000). Recently, male biased gonadal sex ratio and female infertility were reported in Adamts9 knockout zebrafish (Carter et al., 2019, Figs 2&4). However, Adamts9’s role in ovulation in vertebrates have not been firmly established by the knockout studies, due to Adamts9 knockout mice dying before gastrulation (Enomoto et al., 2010). Furthermore, overwhelming adult Adamts9 knockout zebrafish were males, and a few Adamts9 knockout females have underdeveloped ovaries and infertile (<7% females in Adamts9 KO zebrafish vs. 33% females in wildtype and heterozygous siblings, Carter et al., 2019, Fig. 2).

Mmp2 in ovulation

Mmp2, also known as gelatinase A, has a wide range of substrates including collagen, elastin, endothelin, fibroblast growth factor, Mmp9, Mmp13, plasminogen, and Tgf-β (Bassiouni et al., 2021; DeCoux et al., 2014).

Mmp2 is another metalloprotease that most likely causes follicular rupture during ovulation in various animal species. Increases of mmp2 transcripts or type IV collagenolysis were reported in periovulatory follicles of women (Puistola et al., 1986; Baka et al., 2010), rats (Reich et al., 1991; Curry et al., 1992, 2001; Bagavandoss, 1998; Liu et al., 1999), monkeys (Chaffin and Stouffer, 1999), medaka (Ogiwara et al., 2005), mice (Nagaraja et al., 2010), cat (Fujihara et al., 2016; Kehoe et al., 2021), and Ciona intestinalis (an Ascidiacea, Matsubara et al. 2019). There was a progressive increase in follicular Mmp2 production during the periovulatory period in sheep. Mmp2 was localized within connective tissue strands that extend into the substance of the corpus luteum and form an infrastructure for thecal invasion and neovascularization (Gottsch et al., 2000). The transcripts, proteins and/or activities of Mmp2 were also upregulated by FSH or LH in growing or preovulatory follicles in chicken (Wolak et al., 2021), brook trout (Crespo et al., 2013), rat (Jo et al., 2004), and cow (Imai et al., 2003). Knockdown Mmp2, but not Mmp1, caused anovulation and subfertility in Drosophila (Deady et al., 2015). However, other studies suggested that Mmp2 was not the key acute regulator for the changes in follicle shape immediately prior to ovulation (Riley et al., 2004; Sessions et al., 2009). No changes of Mmp2 were found throughout follicular development in mares, although the predominant gelatinase in follicular fluid was Mmp2 (Riley et al., 2001). We found that mmp2 transcripts were high in immature follicles, but low in preovulatory follicles and decreased prior to ovulation in zebrafish (Liu et al., 2018). We also found the expression of mmp2 transcript was not different in ovarian follicles between Pgr knockout and wildtype zebrafish (Liu et al., 2018), It was also reported that Mmp2 was not the target of Pgr during ovulation in mice (Robker et al. 2000). An obligate function for Mmp2 in ovulation may be species dependent. Indeed, mice or zebrafish deficient in Mmp2 had normal fertility (Itoh et al., 1997; Hayashidani et al., 2003; Kok et al., 2015, Table 1).

Mmp9 in ovulation

Mmp9, known as gelatinase B, also have a wide range of substrates including collagen, elastin, aggrecan, gelatin, laminin and a variety of non-ECM substrates such as casein, plasminogen, and Tgf-β1 (Bassiouni et al., 2021; Skjøt-Arkil et al., 2012; Kobayashi et al., 2014).

Transcripts of mmp9 and its gelatinase activity increased in the preovulatory follicles of chicken (Hrabia et al. 2019), pigs (Kim et al., 2014), mares (Riley et al., 2001), rats (Curry et al., 2001), cattle (Goldberg et al., 1996; Zhao and Luck, 1996), rhesus macaque (Peluffo et al., 2011), women (Duncan et al., 1998), and Ciona intestinalis (Matsubara et al. 2019). LH and FSH were shown to be important for maintaining and increasing of Mmp9 transcript, protein, and its enzymatic activities in vitro in porcine granulosa cells (Kim and Yoon, 2020). We also found that mmp9 transcript was high in immature follicles, and low in mature follicles in zebrafish (Liu et al., 2018). Interestingly, the expression of mmp9 transcript increased significantly in the follicular cells of preovulatory follicles, and this increase was significantly suppressed in Pgr knockout zebrafish (Liu et al. 2018).

In contrast, it was found that LH suppressed Mmp9 proteins in rhesus monkeys (Young and Stouffer, 2004). Both FSH and IGF1 inhibited mmp2 and mmp9 mRNA expression in vitro in bovine granulosa cells. The specific down regulation by the gonadotropic hormones FSH and IGF1 in vitro suggests that excess Mmp2 and Mmp9 expression is neither required nor desired for follicle development in cattle (Portela et al., 2009). The mmp9 transcript was induced by Tgf-β1 but not by FSH, LH, progesterone, or estrogen in chicken (Zhu et al., 2014). The expression of Mmp9 in ovaries of PGR knockout mice was like that in their wild-type littermates (Robker et al., 2000). Both Mmp9 knockout mice and zebrafish were fertile (Itoh et al., 1999; Silva et al., 2020, Table 1), which suggest Mmp9 is not essential for ovulation in mice and zebrafish.

Other proteases and their endogenous inhibitors in ovulation.

Adam8 was found to be a downstream target of Pgr in murine ovary (Sriraman et al., 2008). Both Adam8 mRNA and protein were expressed in granulosa cells and cumulus cells of periovulatory follicles. The adam8 expression was also significantly stimulated in the follicles by hCG and greatly reduced in Pgr KO mice. Basal activities of adam8 promoters could be increased with co-expression of Pgr and its co-activators (Sriraman et al., 2008). We found a significant increase of adam8b expression in preovulatory follicles of wildtype zebrafish, while this increase was blocked in Pgr KO (pgr^−/−) zebrafish (Liu et al., 2017, 2018, Table 1) that failed to ovulate (Zhu et al., 2015). Transcripts of adam 17 showed a transit increase in granulosa and theca interna cells of bovine preovulatory follicles following treatments of hCG or forskolin, and this increase was dependent on the response elements of forskolin in promoter region of adam17 (Sayasith & Sirois, 2015). The mmp19 and timp1 transcripts were localized in the granulosa and thecal-interstitial cells of large preovulatory or ovulating follicles, and significantly increased (5–10 fold) following hCG treatment in mice. It was suggested that the increase Mmp19 was involved in ECM degradation that occurred during follicular rupture and that Timp1 could have a role in terminating Mmp19 activity after ovulation (Hägglund et al., 1999). Similar gonadotropin induced mmp19 expression was also reported in at granulosa cells of rats and rhesus macaques (Jo & Curry, 2004; Peluffo et al., 2011), but not theca-interstitial cells in rats, suggesting a role of this proteinase during follicular growth, ovulation, and luteal regression (Jo & Curry, 2004). Estrogen is considered as an inhibitory factor for oocyte maturation and ovulation. Interestingly, Mmp19 was identified as a downstream target of ERβ, and downregulated in ERKO mouse ovaries (Nalvarte et al., 2016.). The expression of mmp23 (a membrane-anchored Mmp) transcripts was spatially and temporally regulated by gonadotropin during follicular development in rats (Ohnishi et al., 2001). However, the functions of Mmp23 in folliculogenesis and ovulation are still unclear (Ohnishi et al., 2005, Table 1).

An increase in mmp1, mmp2, mmp7, timp1, timp2 transcripts was found within 12 hours after hCG administration in macaque. Interestingly, mmp9 mRNA did not increase until 36 hours after hCG treatment, immediately prior to ovulation (Chaffin and Stouffer, 1999). Trilostane, an inhibitor of 3β-hydroxysteroid dehydrogenase could inhibit hCG stimulated expression of mmp1, mmp2, mmp7, timp1, timp2 transcripts, while transcripts of mmp1 and timp1 were comparable to those in control in combined treatments of progestin (R5020), hCG and trilostane (Chaffin and Stouffer, 1999). Administration of hCG in dominant follicles rapidly increased transcripts of mmp1, mmp10, mmp19, adamts4, adamts9, and adamts15 in rhesus monkey (Peluffo et al., 2011). No changes in Mmp2, Mmp9, or Timp2 protein expression were found in dominant follicles during the ovulatory phases in women (Lind et al., 2006a). Transcripts of mmp10 increased significantly between the preovulatory and early ovulatory periods and remained elevated at the late ovulatory phase in granulosa cells in human ovulatory follicles. In contrast, transcripts of mmp11 decreased significantly after hCG treatment in both granulosa and thecal cells in human (McCord et al., 2012). Increases of mmp1, mmp19, adamts1, and adamts9 transcripts were also found in the granulosa cells in human in response to hCG stimulation (Rosewell et al., 2015).

Studies of plasmin/plasminogen also suggest their involvements in the ovulation, activation of other Mmps, and degradation of ECM in mammals (Curry and Smith, 2006; Liu et al., 2004; Murdoch and McDonnel, 2002; Ohnishi et al., 2005; Peluffo et al. 2011; Ogiwara et al., 2012). However, double knockout of two main plasminogen activators only decreased ovulation by 26%, suggesting that plasmin/plasminogen contribute to ovulation but are not essential (Leonardsson et al., 1995, Table 1). Cathepsin L, a lysosomal cysteine protease member of the papain family was also shown to be induced by progestin in preovulatory follicles of rodents (Robker et al., 2000).

In medaka, at least five distinct proteases (Plau1, plasmin, Mmp2, Mmp14 and Mmp15) was shown to be responsible for the degradation ECM during ovulation. However, Mmp15 was the only protease that is drastically induced in the granulosa cells of preovulatory follicles by LH (Ogiwara et al., 2012; Hagiwara et al., 2014). Transcripts of mmp2, mmp9, adamts1,plg, and timp2 were also found to be up-regulated by LH in vitro in brown trout (Crespo et al., 2013). Significant increase of adam8b, adamts1, adamts8a, adamts9, and mmp9 transcripts in the follicular cells were also observed in the preovulatory follicles of zebrafish (Liu et al., 2017, 2018). Takahashi and colleagues (Takahashi et al., 2019) proposed a “two-step ECM hydrolysis mechanism’ model for follicle rupture based on changes of expression and activities of proteases during ovulation in medaka. In this model, two key proteases (Plaul & Mmp2) are constantly synthesized and released, but their activities are relentlessly blocked by protease inhibitors, Pai1 for Plau1 and Timp2b for Mmp2. In the 1^st step of this ECM hydrolysis model, expression and secretion of Pai1 (inhibitor for Plau1) from granulosa cells decrease considerably approximately 7 hours prior to the ovulation, thus releasing Plau1 from the inhibition. Active Plau1 starts converting membrane-bound plasminogen to the active enzyme plasmin, which in turn hydrolyzes laminin, a major ECM component of basement membrane in follicles. Around 3 hours before ovulation with sufficient degradation of laminin, the activity of Plau1 and conversion of plasminogen to plasmin are blocked again by Pai1 due to resumed synthesis and secretion of Pai1. Mmp2 is constitutively synthesized in the oocyte as its inactive precursor proMmp2. ProMmp2 is then activated by membrane-bound Mmp14 (also known as Mt1-Mmp) which is expressed constitutively on the cell surface of the oocyte. However, Timp2b, which is also produced and secreted from the oocyte to the extracellular space of the follicle, binds to Mmp2 to suppress the enzyme activity until 3 hours prior to ovulation. In the 2^nd step of this model, Timp2b in the extracellular space of the periovulatory follicle is drastically reduced around 3 hours before ovulation. Removing inhibition of Timp2b on Mmp2 permits the active Mmp2 to degrade collagen type IV in the basement membrane, and Mmp15 (also known as Mt2-Mmp) to cleave collagen I in the extracellular space. Interaction of Mmp15 with collagen I is achievable only after the breakdown of the basement membrane by Plasmin and Mmp2 (Takahashi et al., 2019).

It is clear that interplay of multiple proteases for degrading various ECM is required for the rupture of follicular layers and extracellular membranes in order to release ovulated oocytes. The biological effects of these proteases are dependent on de novo production, proteolytic activation and endogenous tissue inhibitor of matrix metalloprotease (Timp) concentrations. Studies of changes and hormonal regulations of various proteases in different animal species have increased our knowledge on the candidates of proteases involved. However, significant number of these studies only examined changes of transcripts, which may not reflect changes of proteins and enzymatic activities. So far, knockout studies of most metalloproteases in mice or zebrafish have provided little information on the functions of these proteases (Table 1), mainly due to knockouts either dying in utero or exhibiting non-observable defects in ovulation (Carmeliet et al. 1993; Peschon et al. 1998; Vu and Werb 2000; Kelly et al. 2005; Enomoto et al. 2010).

Summary and Future Directions

Intensive interests and research in past decades have shown changes of various metalloproteases during ovulation in various species. Our understanding is also advanced by hormone regulation of these metalloproteases and differential expression in the knockouts vs. wildtype for metalloproteases including Adamts1, Adamts9 and others. Unlike 100% anovulation and infertility found in Pgr KOs in the rat, mouse, and zebrafish (Lydon et al., 1995; Zhu et al., 2015; Kubota et al., 2016; Tang et al., 2016), none of metalloprotease knockout could completely block ovulation. A few of these metalloprotease knockouts are embryonic lethal, while no obvious reproductive defects in the ovulation or fertility were reported for most metalloprotease knockouts (Table 1). It is well known that same substrate or even same cleavage site could be the target of several metalloproteases (Wang et al., 2019). Loss of one or even two metalloproteases could be compensated by other metalloproteases. Therefore, no observable deficiency in the knockouts only suggests that specific metalloprotease is not essential for the process; however, does not suggest that the protease has no function in the reproductive process such as ovulation. On the other hand, subfertility observed in Adamts1 global knockout mice also needs to be interpreted with caution. Most of these metalloproteases including Adamtsl are expressed broadly in various tissues, which also have effects in the processes other than reproduction such as growth and development of other vital organs including the brain and heart. To further explore roles and mechanisms of these metalloproteases in ovulation, additional tools such as conditional knockouts, specific antibodies for ECM proteins, transgenic reporter lines and sensitive in vivo bioassay that allow studying dynamic changes of ECM and signaling molecules need to be developed.

Our knowledge on the roles and mechanisms of metalloproteases in PGC migration, gonad formation and sex differentiation, especially in vertebrates is still lacking. The fast generation of knockouts, availability of transgenic lines that label PGCs, gonad somatic cells and specific ECM molecules with fluorescent proteins and nearly transparent embryos made zebrafish an excellent vertebrate model for studying germ cell migration at a high resolution within a live organism (Raz, 2002; Dumstrei et al., 2004). Transgenic zebrafish lines in different genetic backgrounds in addition to Adamts9 knockout, and additional new research reagents will permit studying the roles and mechanisms of various metalloproteases and signaling molecules in PGC migration and gonad development starting from the earliest stages of their development through gonadal differentiation or sex reversal in adult animals. New research reagents, new models and new knowledge for the gonadal formation and ovulation are likely to be generated from these future studies.

Highlights.

• Adamts9 has a role in PGC migration in zebrafish.

• Adamts9 is critical for the development and maintenance of ovarian follicles.

• Metalloprotease expression including Adamts1, Adamts9, Mmp2, and Mmp9 change during gonadal development and ovulation.

• Post-translational activation of metalloproteases is critical for their functions.

• Lack of clear reproductive phenotypes in most metalloprotease knockouts is likely due to multiple proteases complementing each other in their ECM targeting.