Mechanisms of Otoconia and Otolith Development

Abstract

Otoconia are bio-crystals which couple mechanic forces to the sensory hair cells in the utricle and saccule, a process essential for us to sense linear acceleration and gravity for the purpose of maintaining bodily balance. In fish, structurally similar bio-crystals called otoliths mediate both balance and hearing. Otoconia abnormalities are common and can cause vertigo and imbalance in humans. However, the molecular etiology of these illnesses is unknown, as investigators have only begun to identify genes important for otoconia formation in recent years. To date, in-depth studies of selected mouse otoconial proteins have been performed, and about 75 zebrafish genes have been identified to be important for otolith development. This review will summarize recent findings as well as compare otoconia and otolith development. It will provide an updated brief review of otoconial proteins along with an overview of the cells and cellular processes involved. While continued efforts are needed to thoroughly understand the molecular mechanisms underlying otoconia and otolith development, it is clear that the process involves a series of temporally and spatially specific events that are tightly coordinated by numerous proteins. Such knowledge will serve as the foundation to uncover the molecular causes of human otoconia-related disorders.

1. Introduction

The vestibule of the inner ear serves as a rate sensor, detecting head movements about any axis (horizontal, vertical, or torsional). It houses three semicircular canals, which respond to rotational acceleration (head turning, for example), and two gravity receptor organs, which sense linear acceleration (accelerating in a car, for example) and gravity (Figure 1). The utricle and saccule are the two gravity receptor organs and contain otoconia, bio-crystals of calcium carbonate (CaCO₃) and proteins. Non-mammalian vertebrates have a third otolithic organ called the lagena. Otoconia and their embedding membranous structure, the otoconial membrane, lie above the kinocilia and stereocilia of hair cells in the sensory epithelium (macula) of the otolithic organs (Figure 2C). In teleost fish, the bio-crystals are solidified into three otoliths that each directly interact with an entire patch of sensory epithelium. Head tilt and linear head motion cause displacement of the otoconial complex, producing a shearing force which deflects the hair bundles and subsequently depolarizes the sensory hair cells. These electrical signals are then relayed to the central nervous system (CNS) by the afferent vestibular nerve, which jointly with other proprioceptive information, stimulate the CNS to initiate neuronal responses for maintaining body balance. The correct formation and anchoring of otoconia is essential for optimal vestibular function and balance (Anniko et al., 1988; Jones et al., 1999; Jones et al., 2004; Kozel et al., 1998; Simmler et al., 2000a; Trune and Lim, 1983a; Zhao et al., 2008b).

Figure 1.

Illustrations of the mammalian (A) and zebrafish (B) inner ear as well as the proteins known to participate in otoconia (C) and otolith (D) development. References are provided in Table 1.

Figure 2.

Development of otoconia and otoliths. (A) Otoconia are seeded across the lumen above the sensory epithelium of the utricle and saccule in mice, and seeding begins as early as E14.5. A seed has an inorganic CaCO₃ crystallite (asterisk) enveloped by an organic matrix (arrow, stained with Toluidine Blue). Without Oc90, the matrix is nearly gone (Zhao et al., 2007). (B) During growth, the individual crystallites fuse in an organized pattern. Growth begins right after seeding and continues until about a week after birth. (C) During or shortly after growth, tens of thousands of otoconia are anchored to the otoconial membrane atop the hair bundles, and some are in contact with the hair bundles (arrow). The inset shows a partially decalcified otoconium (hence abnormally shaped, arrowhead) in contact with a hair bundle (arrow). (D) Otoliths in zebrafish are also seeded across the lumen, and some seeds, while still adding matrix and CaCO₃, become anchored (arrows) to tether cells. As in mice, Oc90 is expressed very early to initiate seeding. Seeding particles are readily visible by 18.5 hpf. (E) Seeds are anchored to tether cells and aggregate in a controlled pattern to form a single large otolith on each patch of hair cells. Bead-like aggregates of Oc90 may indicate the origin of crystallite nucleation. (F) Fish otoliths continue to grow throughout the lifespan, as indicated by the daily growth rings. The arrow denotes a hair bundle underneath the otolith and arrowheads mark layers in the fish otolith. (G) Summary of mechanisms governing otoconia and otolith development. More than one mechanism drives each of the three steps in bio-crystal formation illustrated in panels A–F and are discussed within the text in more detail. hpf and dpf, hour and day post fertilization, respectively; E17.5, embryonic day 17.5; P4, postnatal day 4; 6M, 6 months old; STM, Starmaker; AcTub, Acetylated Tubulin.

Otoconia-related balance disorders are prevalent. For example, benign paroxysmal position vertigo (BPPV), a disease caused by otoconia dislocation from the utricle to the semicircular canals (Salvinelli et al., 2004; Schuknecht, 1962; Schuknecht, 1969; Squires et al., 2004), is the most common cause of vertigo in humans. Elderly people are prone to both otoconia degeneration and BPPV (Anniko et al., 1984; Igarashi et al., 1993; Ogun et al., 2014; Ross et al., 1976), suggesting that degeneration exacerbates dislocation. Otoconia abnormalities can have multiple causes such as genetic mutation, head trauma and ototoxic drugs. In order to identify the molecular etiology of vestibular disorders, it is essential to understand the mechanisms by which otoconia are formed and maintained. Over 24 mouse and about 75 zebrafish genes have recently been identified to be important for otoconia and otolith development, respectively. While zebrafish otoliths are structurally distinct from mammalian otoconia, large scale genetic studies have demonstrated that the zebrafish is a predictive model for human development and disease (Howe et al., 2013). Consequently, the study of both mouse and zebrafish model systems together will help better define how mammalian otoconia form. In this article we will provide an updated brief review of otoconial proteins, incorporating the recent findings on zebrafish homologs. We will also review the cells and cellular processes important for otoconia and otolith development.

2. Constituent proteins in otoconia and otolith development

Otoconia and otoliths contain CaCO₃, yet morphology, crystalline structure, and protein composition vary between the two types of biominerals. With a barrel-shaped body and triplanar facets at each end, each otoconium actually contains a number of minute crystallites, which are outlined by an organic matrix (Figure 2A, 2B) (Lim, 1984; Lins et al., 2000; Mann et al., 1983; Steyger and Wiederhold, 1995; Zhao et al., 2007). In contrast, dome-shaped fish otoliths contain concentric layers of organic matrix alternating with CaCO₃ (Campana and Neilson, 1985). The crystal lattice structure also differs: from primitive to advanced fish to higher vertebrates, crystallinity has evolved from apatite to aragonite to calcite, respectively (Carlstrom, 1963; Ross and Pote, 1984). Mammalian otoconia form at late embryonic stages and may require some maintenance afterwards (Salamat et al., 1980; Thalmann et al., 2001; Lundberg, unpublished data). After initial seeding, fish otoliths quickly attach to immature hair cells known as tether cells and rapidly grow during early ear development (Figure 2D, 2E) (Petko et al., 2008; Riley and Moorman, 2000). Daily growth layers, which vary in thickness and composition depending on the environment, are subsequently added to the otolith throughout the life of the fish (Figure 2F)(Campana, 1999).

Recent data support a model in which the CaCO₃ crystalline structure and morphology in otoconia and otoliths are regulated by an organic matrix made of proteins and proteoglycans (Deans et al., 2010; Kang et al., 2008; Sollner et al., 2003; Zhao et al., 2007). Otoconial proteins are collectively referred to as otoconins, and many are essential for CaCO₃ crystallization as they bind calcium from the surrounding, calcium-poor endolymph (Endo et al., 1991; Ferrary et al., 1988; Ito et al., 1994; Marcus and Wangemann, 2009; Pisam et al., 2002; Pote and Ross, 1991; Salt et al., 1989; Verpy et al., 1999; Wang et al., 1998; Xu et al., 2010). To date, as many as nine murine otoconins have been identified: Otoconin-90 (Oc90, aka Oc95)(Wang et al., 1998; Verpy et al., 1999), Otolin-1 (Zhao et al., 2007), Keratan sulfate proteoglycan(s) (KSPGs) (Yang et al., 2011b), α-Tectorin (Legan et al., 1997), Fetuin-A (aka countertrypin; (Thalmann et al., 2006; Zhao et al., 2007), Osteopontin (aka Spp1; (Sakagami, 2000; Takemura et al., 1994; Zhao et al., 2008a), Sparc-like protein 1 (Sc1, aka hevin and Ecm2;(Thalmann et al., 2006; Xu et al., 2010), Secreted protein acidic and rich in cysteine (Sparc, aka BM-40 and osteonectin), and Dentin matrix protein 1 (DMP1) (Xu et al., 2010). These proteins and the otoconial regulators discussed in the next section are all highly conserved throughout evolution.

Oc90

Oc90 is the most abundant otoconin (Pote and Ross, 1991; Verpy et al., 1999; Wang et al., 1998) and is required for the otoconial matrix to form by specifically recruiting other components and calcium (Yang et al., 2011b; Zhao et al., 2007). While Oc90 has a similar role in seeding zebrafish otoliths, it is not the major otoconin (Sakagami, 2000). The predominant fish otolith matrix protein, OMP-1, is not required for otolith seeding but is necessary later for otolith growth (Murayama et al., 2005; Murayama et al., 2004).

Despite its structural similarity to secretory phospholipase A2 (sPLA2), Oc90 does not have the catalytic activity of the enzyme but has conserved the calcium binding capability, a function most other otoconins share (Pote and Ross, 1991; Wang et al., 1998). Oc90 is extremely acidic, with an isoelectric point measured at 2.9 (Lu et al., 2010) which enables the binding of calcium or CaCO₃. In the absence of Oc90, matrix-bound calcium on the surface of the macula is greatly reduced (Yang et al., 2011b). In both in vivo and in vitro studies, Oc90 facilitates CaCO₃ crystal nucleation and growth (Zhao et al., 2007; Lu et al., 2010). Being rich in cysteines, Oc90 may form numerous disulfide bonds to generate a rigid scaffold for such a purpose. The partial otoconia formation in Oc90 null mice may be due to the compensation of Sc1 (Xu et al., 2010). In contrast, oc90-morphant zebrafish lack otoliths (Petko et al., 2008), suggesting that Oc90 is critical during the earliest stages of zebrafish otolith formation.

Oc90 expression begins days before all other known otoconial genes in mice and zebrafish (Petko et al., 2008; Verpy et al., 1999; Wang et al., 1998). The organic matrix subsequently forms as Oc90 appears to recruit other components at the time of their expression (arrow in Figure 2A) (Zhao et al., 2007). In mice, the temporal changes in Oc90 expression levels coincide with the timing of otoconia development (Xu and Lundberg, 2012). In comparison, oc90 expression in zebrafish is relatively low but constant through the earliest stages of otolith formation (Petko et al., 2008). Oc90 is expressed in the non-sensory epithelium of the developing vestibule in mice, but in all cells in the developing otic placode in zebrafish. In the latter, punctate Oc90 seeds coalesce adjacent to immature hair cells before the otocyst forms (Figure 2D, Kramer unpublished).

Sc1

In the absence of Oc90, a few giant otoconia form containing massively elevated levels of Sc1 (Xu et al., 2010). In normal mice, Sc1 level is extremely low, if any, in otoconia (Thalmann et al., 2006; Xu et al., 2010) but high in the brain and various types of nervous tissues (Johnston et al., 1990; Lively et al., 2007; McKinnon and Margolskee, 1996; Mendis and Brown, 1994). Currently, it is not clear what upstream genes regulate the expression of Sc1 or Sparc in the inner ear, but the up-regulation of Sc1 in Oc90 null otoconia is due to increased protein incorporation (Xu et al., 2010). The exact role of Sc1 in biomineralization has not been defined, as otoconia have not been closely examined in Sc1 null mice which have apparently normal balance function (McKinnon et al., 2000).

Sc1 and Oc90 have different sequences but share some important features. Sc1 is also an acidic glycoprotein that is rich in cysteines, and has a high content (22%) of acidic amino acids and an EF-hand motif for binding calcium and calcium derivatives (Chun et al., 2006; Xu et al., 2010). Sc1 also contains a Sparc-like region which consists of a follistatin-like, a collagen-binding, and two calcium-binding (EF-hand) domains (Maurer et al., 1995). These domains form a ‘Phe pocket’ (Hohenester et al., 2008; Sasaki et al., 1998) after binding collagenous proteins such as Otolin. Collectively, these features would allow Sc1 to compensate for the loss of Oc90.

Otolin

Otolin-1 (aka Otolin) is an inner ear-specific collagen and likely forms a collagen-like scaffold for optimal otoconia formation (Zhao et al., 2007; Deans et al., 2010; Yang et al., 2011b). It is also a secreted glycoprotein and is present in both otoconial crystals and fibrous membranes. While a mouse Otolin knockout has not been described, otoliths in otolin1a morphant fish are fused and unstable (Murayama et al., 2005).

As a member of the collagen X family and C1q super-family (Deans et al., 2010; Kishore and Reid, 1999; Yang et al., 2011b), Otolin has three collagen-like domains and a globular C1q (gC1q) domain, and both types of domains interact with Oc90 to form the otoconial matrix and sequester calcium for optimal otoconia formation. Oc90 and Otolin cooperate to synergistically facilitate matrix calcification in cultured cells (Yang et al., 2011b). In vitro, Oc90 requires the presence of Otolin to generate crystals with otoconia-like morphology (Moreland et al., 2014). Similarly, the normal growth of zebrafish otoliths requires OMP-1 and its incorporation of Otolin1a (Murayama et al., 2005; Murayama et al., 2004).

Keratan sulfate proteoglycan(s) (KSPGs)

Widely distributed at the cell surface and in the extracellular matrix, proteoglycans consist of a core protein with covalently attached glycosaminoglycan (GAG) chains. Classes of proteoglycans differ in their GAG chains and include heparan sulfate proteoglycans (HSPGs), chondroitin sulfate proteoglycans (CSPGs) and KSPGs (Kramer, 2010). Proteoglycans can form large complexes by interacting with other proteoglycans and collagen. In addition, proteoglycans have strong negative charges and can attract ions such as Ca²⁺. These features make KSPGs good candidates to mediate otoconia calcification. Deletion of different HSPGs and CSPGs causes calcification deficiencies (Hassell et al., 2002; Viviano et al., 2005; Xu et al., 1998; Young et al., 2002) which exemplifies their critical role in bone and teeth formation.

In the inner ear, KSPG(s) are the predominant proteoglycan class (Killick and Richardson, 1997; Xu et al., 2010), and are present in mouse, chinchilla and chicken otoconia (Fermin et al., 1990; Swartz and Santi, 1997; Xu et al., 2010). KSPG(s) appear to interact with Oc90 and Otolin (Yang et al., 2011b), but a detailed role of KSPG(s) in otoconia or otolith formation has not yet been illustrated, nor has a KS core protein been identified in mouse or zebrafish. Within the mouse cochlea, α-Tectorin is a KSPG (Killick and Richardson, 1997), but its expression is mostly restricted to the otolithic membrane (El-Amraoui et al., 2001).

Other minor otoconins

A few otoconial proteins that are important for bone and tooth formation in mice as well as otolith development in zebrafish are not essential for murine otoconia development. One example is Osteopontin, a small integrin-binding N-linked glycoprotein (SIBLING) protein that is abundant in the mineralized matrices of bone and teeth, and is important for organizing the bone matrix, maintaining bone strength and regulating bone crystal sizes (Boskey et al., 1993; Duvall et al., 2007; Hunter et al., 1994; Shapses et al., 2003; Thurner et al., 2010). In the vestibule, however, Osteopontin is not needed for otoconia formation or vestibular function, despite the presence of the protein in wildtype tissues (Zhao et al., 2008a).

Another member of the SIBLING family, Dentin matrix acidic phosphoprotein 1 (DMP1), is also abundant in teeth and bone where it is critical for apatite crystal seeding as well as growth (George et al., 1993; Hirst et al., 1997; MacDougall et al., 1998). In addition to severe bone defects, DMP1 null mice show imbalance behaviors (Lv et al., 2010). While the latter has been attributed to abnormal bone structure in the inner ear, otoconia also contain DMP1 (Xu et al., 2010) and may be affected as well. Indeed, zebrafish deficient in the DMP1-related protein Starmaker have otoliths with an altered crystal lattice structure (Bajoghli et al., 2009; Sollner et al., 2003).

Sparc is present in tissues undergoing remodeling (Bolander et al., 1988; Hohenester et al., 1997; Sage and Vernon, 1994) and may be critical for initiating bone mineralization (Bianco et al., 1985; Termine et al., 1981). Its function can be attributed to its high affinity for calcium and collagen (Bolander et al., 1988; Hohenester et al., 2008; Maurer et al., 1995). While Sparc is present at an extremely low level and may not be important for murine otoconial development (Xu et al., 2010), it is necessary for zebrafish otolith formation (Kang et al., 2008). Despite being up-regulated in Oc90 null mice, its family member Sc1 appears to be the preferred matrix protein when Oc90 is deleted (Xu et al., 2010).

3. Direct regulators of otoconia and otolith development

Bio-crystal development in the ear is regulated by non-constituent proteins that are not incorporated into otoconia and otoliths. These regulatory proteins are critical to establish an appropriate milieu for crystal seeding and growth by (1) influencing otoconin secretion or functional modification as well as (2) generating spatial and temporal gradients of calcium and other ions.

NADPH oxidase 3 (Nox3) and associated proteins

The membrane-bound enzymes NADPH oxidases (Noxs) produce reactive oxygen species that have both normal and pathological roles (Bedard and Krause, 2007). The activities of Noxs are modulated by partners such as p22^phox, Nox organizers (Noxo1, p47^phox and p40^phox), and Nox activators (Noxa1 and p67^phox). Among the seven mammalian Nox members, Nox3 is primarily expressed in the inner ear and is located in the cytoplasmic membrane. Nox3 and its regulators p22^phox and Noxo1 are all essential for otoconia formation and balance function in mice (Kiss et al., 2006; Nakano et al., 2007; Nakano et al., 2008; Paffenholz et al., 2004).

Just how Nox and associated proteins mediate otoconia formation is not understood. If otoconins are oxidized by Nox3, conformational changes could ensue to facilitate crystal nucleation (Xu et al., 2012). Alternatively, electrons generated by Nox3 could depolarize the apical plasma membrane and increase local calcium concentration, as proposed by Nakano et al., 2008. Also, the resultant superoxide may increase local pH by reacting with protons, thus promoting CaCO₃ crystal formation and maintenance. To date, Noxs have not been described in the zebrafish ear, which could help clarify the mechanism.

Otopetrin-1 (Otop1)

Otop1 is required for otoconia and otolith development possibly by regulating protein secretion and by mobilizing calcium (Hughes et al., 2004; Hurle et al., 2003). Otop1 has multiple transmembrane (TM) domains, and single-point mutations in these domains can cause absent otoconia as in the two mutant mice, tilted (tlt) and mergulhador (mlh). This is similar to the phenotype observed in the zebrafish backstroke mutant where otop1 is defective (Sollner et al., 2004).

In zebrafish, otop1 is expressed in hair and supporting cells before otolith seeding, but only in hair cells during otolith growth (Hurle et al., 2003; Sollner et al., 2004). Its expression pattern is opposite to that of Oc90 in mice. A recent study (Kim et al., 2010) showed that Otop1 is located in the apical region of supporting cells and some transitional cells. Disruption of this apical localization of Otop1 in tlt and mlh mutant epithelia may be the cause of absent otoconia (Kim et al., 2011). This suggests that Otop1 may mediate otoconin secretion, as the otoconin Starmaker is not secreted into the otocyst lumen but retained in the otic epithelia of otop1-mutant zebrafish (Sollner et al., 2004). Calcium mobilization may also be regulated by Otop1, as the protein can increase cytosolic calcium concentrations by inhibiting purinergic receptor P2Y (Hughes et al., 2007; Kim et al., 2010). Therefore, Otop1 may regulate multiple aspects of otoconia and otolith development.

PMCA2

PMCAs are Calmodulin-sensitive plasma membrane calcium-ATPases. PMCA2a is an essential source of calcium for otoconia formation. These pumps are Ca²⁺/H⁺ exchangers which extrude Ca²⁺ from hair cell stereocilia, thereby increasing the [Ca²⁺] near the plasma membrane. All four PMCAs are expressed in the mammalian cochlea, but only PMCA2a, a protein encoded by Atp2b2 gene, is present in vestibular hair bundles (Crouch and Schulte, 1996; Dumont et al., 2001; Furuta et al., 1998; Yamoah et al., 1998). Otoconia and otoliths are absent in Atp2b2 knockout mice and knockdown zebrafish, respectively (Blasiole et al., 2006; Kozel et al., 1998).

Pendrin

More than 150 mutations in Slc26a4 result in a combination of congenital hearing loss, balance, and thyroid disorders known as Pendred syndrome (Dai et al., 2009; Luxon et al., 2003). The Slc26a4 gene product is Pendrin, an anion transporter primarily expressed within the endolymphatic duct and sac, the vestibular transitional epithelia, and the cochlear external sulcus (Everett et al., 1999). Slc26a4 knockout mice have significantly lower endolymphatic pH, which results in the formation of fewer but larger crystals in both the utricle and saccule (Everett et al., 2001; Kim and Wangemann, 2010; Kim and Wangemann, 2011; Nakaya et al., 2007). While the closely related protein Prestin is an electromechanical transducer, a missense mutation within Slc26a4 specifically impairs anion transport activity and phenocopies the Slc26a4 knockout mice (Dror et al., 2010). Consequently, Pendrin regulates otoconia formation through its ability to transport anions such as HCO₃⁻ and by buffering the endolymphatic pH. HCO₃⁻-ATPase and Cl⁻/HCO₃⁻-exchangers also mediate the trans-epithelial transport of HCO₃⁻ to the endolymph and regulate otolith formation (Tohse and Mugiya, 2001).

Carbonic Anhydrase (CA)

CA likely regulates otoconia development and maintenance also by providing HCO₃⁻ and maintaining appropriate pH. In the mammalian inner ear, CA is widely expressed in the sensory and non-sensory epithelia (Lim et al., 1983; Pedrozo et al., 1997), especially in the embryonic endolymphatic sac. Inhibition of CA activity decreases HCO₃⁻ concentration and pH in the endolymphatic sac (Kido et al., 1991; Tsujikawa et al., 1993), and causes abnormal and reduced otoconia in the developing chick embryos (Anken et al., 2004; Kido et al., 1991). Alteration of macular CA is also associated with changes in zebrafish otolith growth (Anken et al., 2004). The isoform CA2 and Pendrin may cooperate to maintain the normal function of the endolymphatic sac, where the two are co-expressed (Dou et al., 2004). Otoconia are also absent within the utricle of zinc transporter 4 (SLC30A4)-mutant mice, possibly as the result of decreased CA activities (Huang and Gitschier, 1997).

Other ion channels

A family of transient receptor potential vanilloids (TRPVs) that selectively transport calcium and magnesium may mediate endolymph homeostasis in the inner ear. While all mouse TRPVs (TRPV1-6) and zebrafish TRPV4 are expressed in vestibular and cochlear sensory epithelia, TRPV4 is also present in the endolymphatic sac, and TRPV5 along with TRPV6 are found in vestibular semi-circular canal ducts (Amato et al., 2012; Ishibashi et al., 2008; Kumagami et al., 2009; Takumida et al., 2009; Yamauchi et al., 2010). The low endolymphatic pH in Pendrin-deficient mice appears to inhibit acid-sensitive TRPV5/6 calcium channels, increasing endolymphatic calcium that likely contributes to the formation of abnormal otoconia crystals (Nakaya et al., 2007). TRPV4 knockout mice have hearing defects, and both TRPV5 and TRPV6 mouse knockouts as well as a trpv6 mutant zebrafish have defective bone formation (Chen et al., 2014; Hoenderop et al., 2003; Tabuchi et al., 2005; Vanoevelen et al., 2011). However, TRPV-deficient mice and zebrafish have not been associated with significant otoconia/otolith abnormalities.

4. Anchoring proteins in otoconia and otolith functional development

Otoconia crystals are anchored to the sensory epithelia by the otoconial membrane, a honeycomb-like structure (Figure 2C) that plays a crucial role in mechanotransduction. Compromised otoconial membrane can cause otoconia to detach and dislocate, resulting in vertigo and balance problems. While the zebrafish otolith also appears to be embedded within a gelatinous membrane, interaction of the otolith and hair cell kinocilium is distinct: the zebrafish kinocilium is embedded in the early otolith while the mouse kinocilium indirectly interacts with most otoconia through the otoconial membrane (Figure 2C and 2F). Nevertheless, homologous proteins appear to mediate attachment in both model systems.

The otoconial membrane is composed of the collagenous protein Otolin as well as five non-collagenous glycoproteins (Otogelin, Otogelin-like, α-Tectorin, β-Tectorin and Otoancorin) and KSPG(s). Otogelin is a glycoprotein that is only present in the inner ear acellular membranes (Cohen-Salmon et al., 1997). It is expressed in supporting cells and the expression continues into adulthood in the vestibule but not in the cochlea (El-Amraoui et al., 2001). Otogelin is required for the anchoring of the otoconial membrane and cupula to the corresponding sensory epithelia. Otogelin mutant mice have displaced acellular matrices and severe vestibular deficits (Simmler et al., 2000a; Simmler et al., 2000b). These mice are also deaf due to defects in the organization of the fibrillar network in the tectorial membrane.

Murine Otogelin and Otogelin-like have an amino acid identity of 33.3% (56.0% similarity), but the latter is expressed at lower levels in a variety of human tissues (Yariz et al., 2012). Mutations in the Otogelin-like gene also cause mild to moderate human hearing loss (Bonnet et al., 2013; Yariz et al., 2012), and knocking down otogelin-like in zebrafish leads to sensorineural hearing loss (Yariz et al., 2012). From human to zebrafish, Otogelin-like is present in all three inner ear acellular membranes, similar to that of Otogelin. However, the cells expressing it are slightly different: in addition to the supporting cells that express Otogelin, Otogelin-like is also expressed in cochlear outer hair cells and saccular hair cells. The expression level is highest during embryonic development, and is gradually reduced from neonatal to adult stages. Therefore, Otogelin and Otogelin-like have similar but non-redundant roles during inner ear development.

α- and β-Tectorin are major non-collagenous glycoproteins of the mammalian tectorial (Legan et al., 1997) and otoconial membranes but are not present in the cupula (Goodyear and Richardson, 2002; Xu et al., 2010). α-Tectorin is a glycosylphosphatidylinositol (GPI)-linked protein that appears to be a KSPG (Legan et al., 1997; Kramer unpublished data). Null mice of α-Tectorin have reduced otoconial membrane and a few scattered giant otoconia (Legan et al., 2000). Mutations in α-Tectorin cause both autosomal dominant and recessive profound hearing loss in humans (Alloisio et al., 1999; Mustapha et al., 1999).

β-Tectorin is expressed in the striolar region of the utricule and saccule, whereas α-Tectorin is expressed in the transitional zone and a region of the roof (Legan et al., 1997; Rau et al., 1999). Disruption of the tectorial membrane is associated with low-frequency hearing loss in Tectb null mice (Russell et al., 2007), yet no vestibular defects have been reported. Nevertheless, knockdown of tectb in zebrafish affected the position and number of otoliths (Yang et al., 2011a). Other otoconins, including Otogelin and α-Tectorin, possess several von Willebrand factor type D domains that contain the multimerization consensus site CGLC (Mayadas and Wagner, 1992), suggesting that this structural feature might result in the formation of heterogeneous, higher order structures.

At the interface between the sensory epithelia and their overlaying a cellular membranes of the inner ear is another GPI-anchored protein, Otoancorin (Zwaenepoel et al., 2002). While mutations in Otoancorin result in autosomal recessive human hearing loss (Walsh et al., 2006; Zwaenepoel et al., 2002), the tectorial membrane remains attached to the sensory hair cells in mice that are Otoanchorin-deficient (Lukashkin et al., 2012). Balance defects have not been described in humans or mice defective in Otoanchorin; and otoanchorin does not appear to be expressed in the zebrafish inner ear (Kramer unpublished data), together suggesting that Otoanchorin is not critical within the vestibular system.

5. Cells and cellular processes involved in otoconia and otolith formation

Participating cell types and cellular structures

Otoconia and otoliths overlay the sensory epithelium, yet all epithelial cells in the utricle and saccule participate in otoconia and otolith formation by synthesizing various otoconins and regulatory proteins: Oc90 is expressed in all non-sensory epithelia, Otolin, Sc1 and Otop1 in hair and supporting cells, Nox3, Noxo1 and PMCA2 in hair cells, p22^phox (a 3^rd component of the Nox complex) in the endolymphatic sac and duct, Pendrin in transitional cells, and various carbonic anhydrases in the sensory and non-sensory epithelia (references are listed in each corresponding sub-sections). Morphologically and functionally similar to marginal cells of the stria vascularis, dark cells in the mammalian utricle express a few otoconial proteins (Oc90, α-Tectorin and carbonic anhydrases) and actively transport ions (including Ca²⁺, HCO₃⁻, Cl⁻ and H⁺) through the expression of numerous ion channels and transporters (Pendrin, KCNQ1, KCNE1, ATP1B2, ATP1A1, and SLC12A2). Expression of ion channels and transporters in the mouse endolymphatic sac or zebrafish semicircular canals also help optimize the environment for otoconia formation and maintenance (Abbas and Whitfield, 2010; Dou et al., 2004; Grunder et al., 2001; Lee et al., 2001; Salt, 2001; Stankovic et al., 1997; Taguchi et al., 2007; Wangemann et al., 1996).

To preserve the balanced ionic composition of the endolymph in both mouse and zebrafish, the otocyst must be maintained as a sealed vesicle surrounded by polarized epithelia that are connected with tight apical junctions. Apical junctional complexes are abnormal in the inner ear of zebrafish that carry a mutation in the transcription factor Grhl2 (Han et al., 2011). Otoliths are significantly reduced, the otocyst is expanded, and expression of claudin b and epcam are dramatically reduced. Remarkably, co-injection of claudin b and epcam can rescue the mutant phenotype. Mutation of epcam or the tight junction protein claudin j also results in zebrafish with smaller otoliths and impaired balance (Hardison et al., 2005; Slanchev et al., 2009). Because seeding particles abnormally persist in claudin j-mutant zebrafish, endolymph composition appears to be critical in otolith growth. Several claudins are expressed in the mouse vestibular epithelium, and mutations in multiple claudins are associated with hearing loss in mouse and human (Ben-Yosef et al., 2003; Kitajiri et al., 2004; Nakano et al., 2009). Surprisingly, no obvious vestibular disorders have been observed in any Claudin-deficient mice or humans, suggesting that there is functional compensation by other tight junction proteins. However, recent data from analysis of the tricellular tight junction protein MarvelD2 instead suggest that otoconia formation and vestibular function can tolerate some change in ionic composition (Nayak et al., 2013). Mutations in MarvelD2 are associated with autosomal recessive nonsyndromic deafness in humans and mice (Nayak et al., 2013; Riazuddin et al., 2006). Despite having ultrastructurally-abnormal tight junctions in both cochlear and vestibular epithelia, the vestibular function appears normal in MarvelD2-mutant mice (Nayak et al., 2013). Perhaps additional research into how tight junctions regulate endolymph composition will help better reconcile the perceived difference between zebrafish and mouse.

Otolithic seeding particles normally attach to kinocilia, a specialized, immotile cilium on hair cells in zebrafish and higher vertebrates (Spoon and Grant, 2011). Additionally, zebrafish have motile cilia that are located near the early zebrafish hair cells (tether cells), require a dynein regulatory complex for their function, and contribute to the asymmetric otolith shape (Colantonio et al, 2009; Wu et al., 2011). In zebrafish that lack hair cells, an otolithic mass aggregates and floats untethered in the otocyst (Millimaki et al., 2007). When hair cells are present but cilia formation is blocked, seeding particles attach to and otoliths grow on the hair cell surface (Stooke-Vaughan et al., 2012). These results suggest that the otolithic membrane (or a critical component) is secreted by hair cells and is the primary mediator of otolith attachment, akin to the otolithic membrane’s role in mammalian otoconia. Nevertheless, cilia are critical to normal zebrafish otolith formation: defective otoliths are observed in embryos that are deficient in any one of at least 16 cilia-related genes, including dyneins, centrosomal proteins, and tubulin-modifying enzymes (Colantonio et al., 2009; Lee et al., 2012) (Figure 1). Since many of these genes have overlapping roles in cilium biogenesis, vesicle transport, and cell-cell signaling, it is possible that other cellular functions are affected as well.

Critical molecular processes

A recent proteomic analysis comparing vestibular hair cells with non-sensory cells found that approximately 50% of the hair cell enriched proteins are involved in protein trafficking, while just 4% of the proteins enriched in non-sensory cells are associated with intracellular transport (Herget et al., 2013). The most abundant protein that Herget et al identified in their hair cell fraction was Otoferlin, an essential regulator of synaptic vesicle exocytosis and neurotransmitter release in hair cells (Dulon et al., 2009). While trafficking to the basolateral synapse is important for sensory relay, intracellular transport to the apical membrane is critical for both otolith and cilia assembly. Proteins that mediate trafficking from the Golgi apparatus to the apical plasma membrane include Otop1 and several adaptor proteins of coated-vesicle transport (reviewed in (Hughes et al., 2006)). Otoconia are absent or decreased in mice with mutations in the adaptor proteins Mocha, Muted, and Pallid (Falcon-Perez et al., 2002; Rolfsen and Erway, 1984; Trune and Lim, 1983b), and otoliths fail to form in zebrafish with a mutation in a Pallid homolog (Amsterdam et al., 2004). Remarkably, expression of the same adapter proteins and secretion of otoconins are concomitantly decreased in two mouse strains with mutations in the autophagic proteinases Atg4b and Atg5 (Marino et al., 2010). Because zebrafish with mutations in atg5 also have defective otoliths (Hu et al., 2011), the connection between autophagy and secretion is likely conserved in the inner ear. Furthermore, expression of several proteins specifically in the tips of stereocilia is conserved in both cochlear and vestibular hair cells (Grati and Kachar, 2011), suggesting that the mechanisms for correctly trafficking proteins in cochlear and vestibular hair cells might be conserved as well.

Several cell-cell signaling molecules, their receptors, and downstream transcription factors affect otoconia and otolith formation (Figure 1 and Table 1). Because these molecules are known to regulate differentiation of both neuronal and non-neuronal cells within the inner ear, the observed effects on otoconia and otoliths may be indirect or secondary. Thus, this review will not delve into the details of how these molecules regulate formation of the cell types necessary for otic bio-crystal formation. Interested readers are encouraged to examine several recent summaries (Groves et al., 2013; Maier et al., 2014). Briefly, FGF, TGFβ, Hedgehog, BMP4, Wnt5b, retinoic acid, and Notch signaling together serve as extrinsic factors that induce and pattern the inner ear precursors at different developmental stages (Esterberg and Fritz, 2009; Hammond et al., 2003; Leger and Brand, 2002; Louwette et al., 2012; Pujades et al., 2006; Radosevic et al., 2011; Sapède and Pujades, 2010). These molecules turn on intrinsic transcription factors includes Sox2, Neurog1 and Atoh1 (Bermingham et al., 1999; Zheng and Gao, 2000; Millimaki et al., 2007; Radosevic et al., 2011; Raft et al., 2007; Sapède et al., 2012). At later stages of development, Atoh1 is important for stereocilia formation and hair cell survival (Cai et al., 2013; Chonko et al., 2013). Additionally, microRNAs appear to modulate hair cell formation, maturation, maintenance, and survival (Friedman et al., 2009; Weston et al., 2011). All these functions would conceivably affect bio-crystal formation and maintenance.

Table 1.

References for zebrafish proteins listed in Figure 1 that are not discussed in the main text. Knockdown or mutation of all these proteins is associated with defects in otolith formation.

Class	Protein
Pumps/transporters	Atp1a1a.1	(Malicki et al., 1996; Whitfield et al., 1996)
	Atp1b2b	(Wang et al., 2008)
	Atp2b1a	(Cruz et al., 2009)
	Atp6v1e1b	(Golling et al., 2002)
Adhesion	Cdh11	(Clendenon et al., 2009)
	Lypd6b	(Ji et al., 2012)
Signaling	Lepa	(Liu et al., 2012)
	Lepr	(Liu et al., 2012)
	S1pr2	(Hu et al., 2013)
	Sbno1	(Takano et al., 2011)
	Sbno2a	(Takano et al., 2011)
	Tmem2	(Totong et al., 2011)
Nuclear	Chd7	(Patten et al., 2012)
	Ctr9	(Langenbacher et al., 2011)
	Dlx3b	(Hans et al., 2007)
	Eya1	(Kozlowski et al., 2005)
	Foxi1	(Malicki et al., 1996)
	Foxj1b	(Yu et al., 2011)
	Hmx2	(Feng and Xu, 2010)
	Hmx3	(Feng and Xu, 2010)
	Hnf1ba	(Lecaudey et al., 2007)
	Lmx1bb	(Schibler and Malicki, 2007)
	Nfe2	(Williams et al., 2013)
	Orc1	(Bicknell et al., 2011)
	Pes	(Simmons and Appel, 2012)
	Pou5f3	(Golling et al., 2002)
	Prdm1a	(Birkholz et al., 2009)
	Rerea	(Asai et al., 2006)
	Sox10	(Whitfield et al., 1996)
	Supt6h	(Malicki et al., 1996)
Cilia	Ahi1	(Simms et al., 2012)
	C15orf26	(Austin-Tse et al., 2013)
	C21orf59	(Austin-Tse et al., 2013)
	Ccdc65	(Austin-Tse et al., 2013)
	Cep131	(Wilkinson et al., 2009)
	Cep41	(Lee et al., 2012)
	Cep70	(Wilkinson et al., 2009)
	Daw1	(Gao et al., 2010)
	Dnaaf1	(Colantonio et al., 2009)
	Dnaaf3	(Colantonio et al., 2009)
	Dnah9	(Colantonio et al., 2009)
	Dzip1	(Yu et al., 2011)
	Gas8	(Colantonio et al., 2009)
	Ift172	(Zhao et al., 2013)
	Lrrc6	(Serluca et al., 2009)
	Nin	(Dauber et al., 2012)
	Ofd1	(Ferrante et al., 2009)
	Pcm1	(Stowe et al., 2012)
	Ttc8	(May-Simera et al., 2010)
	Ttll6	(Lee et al., 2012)
	Zmynd10	(Zariwala et al., 2013)
Intracellular	Gnptab	(Flanagan-Steet et al., 2009)
	Mib	(Malicki et al., 1996)
	Obscnl	(Raeker et al., 2010)
	Ptena	(Croushore et al., 2005)
	Ptpn11a	(Razzaque et al., 2007)
	Raf1a	(Razzaque et al., 2007)
	Rgs18	(Louwette et al., 2012)
	Rhoab	(Zhu et al., 2006)
	Rpl11	(Chakraborty et al., 2009)
	Sec61a1	(Golling et al., 2002)
	Stat3	(Liang et al., 2012)
	Stxbp3	(Amsterdam et al., 2004)

6. Degeneration and regeneration of otoconia

Little has been done on issues related to otoconia degeneration and regeneration. As mentioned in the Introduction, otoconia degeneration is prevalent in elderly humans (Anniko et al., 1984; Igarashi et al., 1993; Ross et al., 1976) and rodents (Jang et al., 2006; Lim, 1984). Aging could bring about a variety of adverse changes in the inner ear, such as disruption of the homeostasis of endolymphatic ions (Gratton et al., 1997), which would affect otoconia. There would be insufficient expression of otoconial proteins (Xu and Lundberg, 2012) and fragility of anchoring filaments, leading to degeneration and dislocation. Re-expression of otoconial proteins, such as by adenoviral delivery of respective genes or by some chemical induction, may be able to regenerate otoconia or slow down degeneration.

Since hair cells express critical otoconial proteins and channels and pumps and are important for secreting components of the anchoring membrane, otoconial degeneration may accompany hair cell loss commonly seen in the elderly (Merchant et al., 2000). In addition to aging, hair cells are also susceptible to injury from trauma and aminoglycoside ototoxicity, while supporting cells are less vulnerable (Lee et al., 2013). Since these factors also contribute to BPPV, protecting or regenerating sensory hair cells are possible strategies to treat vestibular dysfunction (Cabrera Kang and Tusa, 2013). Indeed, adenoviral delivery of Atoh1 into the supporting cells of the inner ear has recently been approved for human clinical trials to treat auditory and balance disorders (Géléoc and Holt, 2014). Mammalian vestibular hair cells have also recently been generated from pluripotent stem cells by recapitulating the developmental timing of BMP, TGFβ, and FGF signaling (Koehler et al., 2013). Unlike those of the mammalian ear, zebrafish can regenerate their hair cells, and the underlying mechanisms have been the focus of much research (Steiner et al., 2014). However, data also demonstrate that aminoglycosides can directly degrade otoconia in vitro (Walther et al., 2014). Whether aminoglycoside-induced vestibular disorders arise from lost hair cells or directly from degrading otoconia, it will be of particular interest to see what effect hair cell regeneration has on otoconia and otolith structure.

Summary and future directions

Great progress has been made in the research in otoconia and otolith development in recent years. It is now clear that, despite their apparently simple structures, the processes governing the development of these crystals are complex and involve a series of temporally and spatially well coordinated cellular and extracellular events. As summarized in Figure 2G, several molecules from different pathways (channel/pump, enzymatic and trafficking processes) work in concert to increase the local concentrations of Ca²⁺ and HCO₃⁻ to form CaCO₃ crystallites. At the same time or beforehand, Oc90 selectively recruits other otoconins to form an initial matrix to facilitate the seeding process. Continuous addition of organic and inorganic components and fusion of several minute crystallites ensure proper growth. The otoconial membrane is critical to the correct anchoring of the growing crystals. Thus, multiple mechanisms drive each of the three steps in bio-crystal formation illustrated in Figure 2.

Much more work is needed to define the roles of many newly discovered proteins. Murine models with targeted disruption of many genes are not yet available, and double mutants could yield important details on the shared pathways of otoconial proteins. At least 70 additional mutant zebrafish have been identified with defective otoliths, but the responsible genes are yet unclear. Additional studies are needed to uncover what drives otoconia formation in specific locations, and how to prevent and treat degeneration. These types of studies are the foundation required to design future therapies to treat otoconia-related vertigo and balance disorders.