Get Complete Project Material File(s) Now! »

Architecture and mechanics of the hair bundle

Morphological features of the hair bundle

The hair bundle is composed of a few tens to a few hundred stereocilia that are arranged in rows of increasing height, forming a staircase pattern. Next to the tallest stereocilia, there is also sometimes a kinocilium, which is a “true” cilium comprising an axonemal core, i.e. 9 circumferential doublets of microtubules and a pair of singlet microtubules at the centre. The kinocilium is present at the early stages of hair-bundle development and may play a role in defining the polarity of the hair bundle (Denman-Johnson and Forge 1999; Tona and Wu 2020). In the mammalian and bird cochlea, the kinocilium then regresses and disappears at a later developmental stage but remains in mature vestibular organs and vestibular and auditory organs of non-mammalian species (Lewis G Tilney et al. 1986; Denman-Johnson and Forge 1999). Each bundle displays a vertical plane of mirror symmetry. The hair cell is maximally sensitive for deflections along a horizontal axis within this plane (Shotwell, Jacobs, and Hudspeth 1981). Vertebrate hair bundles, regardless of their sensory systems, are all ‘built’ after this blueprint.
(▲) and several shorter stereocilia (∆). Figure by A. Forge. (B) A frog saccular hair bundle with a few tens of stereocilia; bundle is used for detecting linear accelerations of the frog’s head. I studied this type of hair bundle during my PhD. Figures by B. Kachar. (C) A chick cochlear hair bundle with a few tens of stereocilia, from the low-frequency region of the organ. Figure by P. Barr-Gillespie and K. Spinelli. (D-E) In the mammalian cochlea, here from the rat, hair cells come in two flavours: inner hair cells (D) and outer hair cells (E). Both types of hair bundles show three prominent rows of stereocilia. In outer hair cells, the hair bundle has a characteristic V-shape, whereas those of inner hair cells are planar. (F) A bat cochlear hair bundle of an inner hair cell in the high-frequency region. The bundle shows only two rows of stereocilia, the minimum configuration required for mechanotransduction. Figures D-F by C. Hackney and D. Furness from (Fettiplace and Kim 2014).
The morphology of the hair bundle is tightly coupled to its function as a mechanosensory antenna. In neuromasts from the fish’s lateral line system at the surface of the skin, a cluster of mechanosensory hair cells senses local fluid flows. Each hair cell within a neuromast has a very long kinocilium (~15 µm) accompanied by several short stereocilia (Fig. I-5.A). A long kinocilium may serve as a flexible antenna that is very sensitive to drag forces exerted by a fluid flow along the fish, thus perpendicular to the long axis of the kinocilium. A long kinocilium is less stiff and exposes a larger area to fluid flow, thus increasing the operating range of the hair cell (Spoon and Grant 2011). In the frog’s saccular macula, a vestibular organ used for detecting linear accelerations of the head, hair bundles show kinocilia that are comparable in height to the tallest stereocilia (Fig. I-5.B). These bundles respond to shearing movements of an overlying membrane called the otolithic membrane, to which the kinociliary tips are attached, with respect to the apical surface of the hair cells (Fig. I-6). These frog hair bundles are also very cohesive (Kozlov, Risler, and Hudspeth 2007) and thus move as a single unit, maximizing the sensitivity of mechano-electrical transduction by ensuring concerted gating of the transduction channels. During my PhD, I studied the hair bundle of this type.
Figure I-6: FROG SACCULAR HAIR-BUNDLE. Light micrograph of a vertical section of the sensory epithelium of the bullfrog sacculus. The gelatinous sheath layer (GL) forms large cavities above the hair cells (HC). Each hair cell is surrounded by supporting cells and the top of the kinocilia (black arrows) are attached to the overlying gelatinous sheath layer of the otolithic membrane. Figure from (Bechara Kachar, Parakkal, and Fex 1990).
In contrast to frog (Fig. I-5.B and Fig. I-6) and chick hair bundles (Fig. I-5.C), which are directly attached to an overlying membrane, the hair bundles of inner hair cells (IHCs) in the mammalian cochlea are free-standing. These bundles adopt a planar shape (Fig. I-5.D and F) that may maximize drag and increase sensitivity to radial fluid flow hypothesized to stimulate the bundles in the cochlea. In contrast, the characteristic V-shape (or U-shape) of the hair bundles of outer hair cells (OHCs), whose tips are embedded in the overlying tectorial membrane, is believed to reduce viscous resistance to cross flow, thus increasing IHC stimulation (Ciganović, Wolde-Kidan, and Reichenbach 2017) (Fig. I-5.E). The two types of hair cells in the mammalian cochlea—inner and outer hair cells—are associated with different functions: inner hair cells are thought to be the true sensors of the organ, sending information about cochlea vibrations to the brain, whereas outer hair cells are believed to serve as mechanical amplifiers of the input to the inner hair cells (Dallos 2008). As described here, this division of labour is associated with different morphologies of the corresponding hair bundles in different modes of bundle stimulation. Interestingly, mammals tend to hear higher sound frequencies than non-mammals (Fay and Popper 1994; 1999), a property associated with hair bundles with a smaller number of rows, often down to only three and even only two in some bats (Fig. I-5.F).
Scanning electron micrographs of human hair bundles at each corresponding position (in mm from the base of the cochlea). Short hair bundles at the basal region detect high sound frequencies, whereas long hair bundles at the apical region detect lower sound frequencies. The morphological gradient is steeper for outer hair cells than for inner hair cells. Scale bar 1 µm. Figure from (Wright 1983). (B) Length of the tallest stereocilia in hair bundles from inner hair cells (IHC) and outer hair cells (OHC) along the longitudinal axis of the human cochlea (Wright 1984).
The cochlea works as a frequency analyser: high-frequency sounds elicit vibrations near the base of the cochlea, where they are detected by the hair cells residing there, whereas lower-frequency sounds are detected at progressively more apical regions of the organ. This frequency map is called the tonotopic map. One ubiquitous feature of auditory organs is that tonotopy is associated with morphological gradients of the hair bundle: the number of stereocilia, the stereociliary width, and the stereociliary height all vary monotonically along the length of the organs (Lewis G Tilney and Saunders 1983). Hair cells that detect high sound frequencies, at the basal end of the organ, are endowed with shorter hair bundles and more stereocilia than hair cells are dedicated to the detection of lower frequencies; the height and number of stereocilia in a hair bundle vary monotonically along the tonotopic axis (Fig. I-7.A). Regulation of the hair bundle size is so tight that one can locate the position of a particular hair cell by knowing only the dimensions of its hair bundle3 (Lewis G Tilney and Saunders 1983). The study of hair-bundle height in the human cochlea suggests that the precision with which the hair cell regulates the size of its hair bundle is within the range of a few nanometres because the difference in height between
3 Remarkably, it was shown that the amount of polymerized actin in the hair bundles of chicken cochlear hair cells remains nearly the same along the tonotopic axis of the cochlea (G. Tilney and Tilney 1988).
the most basal and most apical outer hair bundles, corresponding to a distance along the cochlear axis of about 3.5 cm, is only about 3-4 µm (Wright 1984) (Fig. I-7.B).
The gradient in hair-bundle height is expected to result in a passive gradient of hair-bundle stiffness along the cochlea. Indeed, the pivotal stiffness Ksp of a hair bundle ought to vary as 1/h2, if h is the hair bundle height where the force is applied (J. Howard and Ashmore 1986). For a simple harmonic oscillator, for which the natural frequency ωnatural ~ √ Ksp, it would make sense that shorter hair bundles are ‘built’ to respond best at high frequencies, and vice versa. Indeed, mechanical gradients have been observed already in the early 1980s with hair bundles from both inner and outer hair cells from the guinea pig cochlea (Strelioff and Flock 1984). A recent study confirmed the existence of such a passive mechanical gradient with hair cells from the rat cochlea (Tobin et al. 2019). However, frequency discrimination cannot happen via passive mechanical resonance alone, for which the hair bundle would build up a sensitive response by accumulating energy from cycle to cycle of the sound stimulus; the hair bundles are overdamped (Crawford and Fettiplace 1985a; Denk, Webb, and Hudspeth 1989). Still, the tight regulation of the hair-bundle size along the cochlea suggests that the morphological gradient might serve physiological purposes involving frequency discrimination by hair cells.

Organization of actin filaments in the core of stereocilia

A stereocilium is made of a parallel network of a few hundred heavily cross-linked actin filaments that stretch from a few nanometres below the stereociliary tip to a dense disorganized network of actin filaments called the cuticular plate underneath the apical surface of the hair cell (Fig. I-8.A-B). There are two types of actin filament packing in the stereociliary core: liquid and hexagon. the liquid packing of actin filaments has been observed in mouse inner hair cells (Mogensen, Rzadzinska, and Steel 2007), in mouse utricular hair cells (Krey et al. 2016), and lizard cochlea hair cells (L G Tilney, Derosier, and Mulroy 1980) whereas chick cochlea and utricles exhibit hexagonal packing (L G Tilney, Derosier, and Mulroy 1980). In the shaft region of a stereocilium, adjacent actin filaments are spaced by about 10 nm (L G Tilney, Derosier, and Mulroy 1980) (Fig. I-18.C-D). Near their insertion into the apical surface of the hair cells, in a region that spans a micrometre or so above the surface, the stereocilia taper, i.e. their diameter progressively decreases. A few tens of actin filaments with denser packing penetrate the cuticular plate, where they form a so-called ‘rootlet’ (Fig. I-8.B and I-9). As a result, when the hair bundle is deflected a stereocilium behaves like a stiff rod that pivots about its insertion point; the stereocilium does not bend (Crawford and Fettiplace 1985b).
Figure I-8: ORGANIZATION OF ACTIN FILAMENTS IN THE CORE OF STEREOCILIA. (A) Each actin-filled stereocilium is connected by an oblique proteinaceous tip link (arrow heads) that gates mechanosensitive ion channels located at the stereociliary tip near the lower end of the tip link. Transmission electron micrograph by A. J. Hudspeth from (Pascal Martin 2007). (B) Each stereocilium is made of a parallel network of actin filaments. From the stereociliary taper, some of the filaments form a rootlet that inserts into the cuticular plate. The densely-packed stereociliary rootlet (arrow head) can be seen in a vertical section along the stereociliary height. Scale bar 250 nm. (C) The transverse section of the stereociliary core above its taper reveals cross-sections of actin filaments. Scale bar 60 nm. (D) The 2D Fourier transform of the image shown in (C) shows a ring-like pattern indicating that the actin filaments in the stereociliary core are organized according to a liquid packing with an inter-filament spacing of ~10 nm. Scale bar 10 nm-1. Figures B-C from (Mogensen, Rzadzinska, and Steel 2007).

Stereociliary rootlets and pivotal stiffness of stereocilia

The stereociliary rootlet is a densely packed array of actin filaments that extends from around the stereociliary taper into the cuticular plate underneath the cell apical surface. In transmission electron micrographs (TEM) of longitudinal sections of the stereocilia, the osmophilic rootlets appear as a dark region (Fig. I-9.B). In transverse sections near the stereociliary insertion, the rootlet appears as a region with more tightly-packed actin filaments than in the periphery (Fig. I-9.C-D). Removing the hair bundle by either pushing with an eyelash or blowing pressurized air reveals the imprints of the stereocilia at the insertion site (Fig. I-9.E). A detailed in the result section, I have extensively used this trick in my work to characterize the size of the rootlet insertions as well as their organization.
The pivotal stiffness of a stereocilium is thought to be dictated by elastic resistance to bending of actin filaments in the stereociliary rootlet over a short region near its insertion (J. Howard and Ashmore 1986; Jonathon Howard 2001). In a simple description, this region can be modelled as a cylindrical beam of radius a and length l, the latter being much shorter than the length L of the whole stereocilium (L >> l). Introducing the flexural rigidity EI and radius r of a single actin filament and assuming that the actin filaments are crosslinked, the pivotal stiffness of a stereocilium is then estimated as where n is the number of actin filaments at the insertion of each stereocilium (Fig. I-9.F). Within this framework, the pivotal stiffness of a stereocilium is largely determined by the number n of actin filaments at the insertion and the height of the stereocilia L. Any change in these parameters would result in a dramatic change of the hair bundle’s pivotal stiffness. The pivotal stiffness Ksp was reported in the range of 200 to 650 µN•m−1 for the saccular hair bundle of the American bullfrog (Jaramillo and Hudspeth 1993; Marquis and Hudspeth 1997) and 150 to 550 µN•m−1 for the rat cochlear hair bundles (Tobin et al. 2019). Assuming that ≃ 50 stereocilia contribute equally to this measured stiffness, the estimate of the pivoting stiffness of a single stereocilium is ≃ 4 − 15 µN/m, corresponding to a rotational stiffness = 2 ≃ 0.06 − 0.35 fN∙m•rad-1 if = 5 µm is an average length for the stereocilia in these hair bundles. Using typical parameter values, = 4 10−26 ⋅ 2 , = 50 , = 5 , = 1 µ , = 5 µ , and = 30, leads to theoretical estimates ≃ 5 µ • −1 and ≃ 0.12fN∙m•rad-1 that are consistent with experimental estimates. Note that in the absence of crosslinking, the stereocilia would be much softer, with = /( 2), corresponding to stiffness that is about 100-fold smaller than that with cross-linking.
Figure I-9: THE STEREOCILIARY ROOTLETS. (A) A transmission-electron micrograph of a vertical section of a rat OHC bundle reveals the stereociliary rootlets. They appear as dark regions extending from around the taper region of the stereocilia into the cuticular plate beneath the hair cell’s apical surface. Scale bar 200 nm. Figure adapted from (David N. Furness et al. 2008). (B) Schematic representation of the rootlet structure of a single stereocilium. The stereociliary rootlet consists of a few tens of densely packed actin filaments inserted into the cuticular plate. Figure adapted from (Pacentine, Chatterjee, and Barr-Gillespie 2020). Transmission-electron micrographs of the transverse sections of stereocilia from guinea pig cochlea (C) at the taper region and (D) at the stereociliary insertions. (C) Near the stereociliary insertion, actin filaments appear as dots that are more tightly packed in the rootlet (arrow head) than within the peripheral region. Scale bar (A-B) 100 nm. Figures adapted from (Itoh 1982). (E) Scanning-electron micrograph of the apical surface of a frog saccular hair cell after the hair bundle was broken at its base by pushing against it with an eyelash or by blowing pressurized air. The stereociliary imprints reveal the stereociliary rootlets at the site of insertions. (F) Schematic representation of a single stereocilium that behaves like a cantilevered beam with a flexible base where a displacement X is imposed at the tip of the beam. The inset shows the stereociliary rootlet of a single stereocilium at the site of insertion. Figure adapted from (J. Howard and Ashmore 1986).
The pivoting stiffness of a hair bundle is not only imposed by the stereociliary rootlets. Horizontal lateral links interconnecting the stereocilia, as well as the oblique tip links, actually contribute a significant fraction of the hair-bundle stiffness (Bashtanov et al. 2004). In intact hair-bundles from the bullfrogs’ sacculus, the experimentally measured hair-bundle stiffness is in the range of 680 to 1,200 µN•m−1 (Marquis and Hudspeth 1997; P. Martin, Mehta, and Hudspeth 2000). The contribution of tip links the total hair-bundle stiffness is about 20% on average but can go up to 50% in rat cochlea hair bundles (Tobin et al. 2019), 80% in Mongolian gerbils (Chan and Hudspeth 2005), and around 50-80% in the frog saccular hair bundles (Marquis and Hudspeth 1997). Having the tip links contribute a large fraction of the hair-bundle stiffness is advantageous to convey the energy of the stimulus to the transduction machinery associated with the tip links (Chan and Hudspeth 2005).

READ  Identification of 5mC and 6mA in the genomic DNA of G. intestinalis

The mechano-electrical transduction machinery

The molecular identity of the ion channel that mediates mechanoelectrical transduction in the hair cell—the transduction channel—has been resisting identification and extensive research efforts for many years. This is in part due to the small number of hair cells in auditory organs (only a few thousand) and of transduction channels per hair cells (2-4 per tip link (Beurg et al. 2009; 2018), corresponding to a few hundred molecules per hair cell). However, over the past 30 years, genetic studies of deafness in mammals helped to identify several molecules implicated in mechanotransduction of hair cells (see the reviews of (Michalski and Petit 2015) and (W. Zheng and Holt 2021) regarding the genetic studies of the mechano-electrical transduction machinery).
The mechano-electrical transduction (MET) machinery consists of the tip link, the transduction channel itself located near the lower end of the tip link, and the myosin motor complex located at the upper end of the tip link. Conventional Transmission Electron Microscopy reveals that both ends of the tip link connect to dark osmophilic regions called the upper tip-link density (UTLD) and the lower tip-link density (LTLD) (Figure I-8.A).
Figure I-10: MECHANOTRANSDUCTION MACHINERY OF THE HAIR CELLS. (A) A high-resolution image of the tip link using freeze-etch electron microscopy reveals its right-handed helical structure. The tip link is about 5 nm in diameter and 180 nm in length and is made of the heterophilic association of dimers of unconventional calcium-dependent adhesion proteins, one dimer of protocadherin-15 and one dimer of cadherin-23. Scale bar 50 nm. Figure from (B. Kachar et al. 2000). (B) Schematic representation of the mechano-electrical transduction (MET) machinery. The MET machinery comprises the tip link connected in series with the transduction channel itself and the myosin motor complex, located at the lower and upper end of the tip link, respectively. Figure from (James O. Pickles 2012).

The tip links

The oblique tip link is made of the heterophilic association of one dimer of protocadherin-15 (PCDH15) and one dimer of cadherin-23 (Cdh23), which together form a right-helical strand interconnecting two stereocilia in adjacent stereociliary rows, from the tip of the shorter stereocilia to the flank of its taller neighbour (Fig. I-10.A). The tip link seen by Scanning Electron Microscopy is ~ 8-11 nm wide and ~150-300 nm long. The short stereocilium contributes a dimer of protocadherin-15 molecules with 11 extracellular (EC) domains, while the longer stereocilium contributes a dimer of cadherin-23 molecules with 27 EC domains, corresponding to about 1/3 and 2/3 of the tip link, respectively (B. Kachar et al. 2000). Because cadherin proteins are calcium-dependent adhesion proteins, the tip link can be disrupted by applying calcium chelators, such as BAPTA4, to lower the Ca2+ concentration and disconnect protocadherin 15 from cadherin 23 (Assad, Shepherd, and Corey 1991) (Fig. I-11.B).
Figure I-11: TIP LINKS. (A) Schematic representation of the molecular constituents of the tip link. The tip link is made of two parallel protofilaments forming cis-homodimers of protocadherin-15 at the lower portion of the tip link and cadherin-23 at the upper portion. The opposing extracellular (EC) repeats of protocadherin-15 and cadherin-23 interact through a ‘molecular handshake’, forming a trans interaction at the last two-terminal domain (EC1-EC2) of the cis-homodimers. (B) Surface representation of protocadherin-15 (in purple and pink) and cadherin-23 (in blue and cyan) at the site of the molecular handshake. Figures B-C adapted from (Sotomayor et al. 2012). (C) Tip links are sensitive to chemical disruption by tetracarboxylic calcium chelators, such as BAPTA. Assad et al. demonstrated that tip links and mechanotransduction can be chemically disrupted by replacing a high Ca2+ standard perilymph (Right panel) with a 5 mM BAPTA solution for 10 s (Left panel). Scale bar 500 nm. Figures from (Assad, Shepherd, and Corey 1991).
The tip link was initially proposed to be the ‘gating spring’ that pulls on the transduction channel, as introduced in the gating spring model of mechanoelectrical transduction (D. Corey and Hudspeth 1983). However, molecular dynamic (MD) simulations based on structural data later indicated that the tip link is likely too rigid to be the gating spring (Sotomayor, Corey, and Schulten 2005; Sotomayor et al. 2010; 2012), and thus that the ‘gating spring’ may be composed of other elastic elements connected in series with the tip link. Nevertheless, the conclusion was recently re-evaluated based on MD simulations (Araya-Secchi, Neel, and Sotomayor 2016) and force spectroscopy of a portion and the whole tip link (Bartsch et al. 2019; Mulhall et al. 2021). Within a physiological tension range, a single protocadherin-15 molecule shows a low stiffness but the stiffness can go up to ~ 10 mN•m-1 at very high tension as a result of strain stiffening (Bartsch et al. 2019). These recent studies bring back the question regarding the identity of the tip link as the long-sought ‘gating spring’.
Disrupted tip links can regenerate. Normal transduction is mostly restored within 24 hours (Zhao, Yamoah, and Gillespie 1996). Regeneration happens first with Pcdh15-Pcdh15 links at ~12 h. Then Cdh23 replaces Pcdh15 at the upper end of the tip link at ~36 h (Indzhykulian et al. 2013). Pulling on the tip-link connection containing extracellular (EC) domains EC1-5 has recently revealed a mean lifetime of 9 s for the tip-link bound at a resting tension of 10 pN (Mulhall et al. 2021) but the lifetime reduces to about 500 µs at the tension of 40 pN, a high-tension value that may be physiological in the cochlea (Tobin et al. 2019). The tip-link bond seems to be much more dynamic than originally thought. Interestingly, it was also recently shown that applying a sinusoidal force to the hair bundle or pushing the bundle towards the negative direction to facilitate encounter of the tip link fragments evoked tip-link recovery within seconds after disruption (Alonso et al. 2020). These remarkable observations suggest that mechanical stimulation may facilitate the reconstitution of functional tip links, which may help explain why normal hearing is restored after temporary hearing loss due to loud sound exposure.

The transduction channels

The MET channel is located at the lower end of the tip link where (Beurg et al. 2009). TMC1/2, TMIE, LHFPL5, and CIB2 were identified as components of the transduction channel (Fig. I-12). TMC1/2 are believed to work as pore-forming proteins (Pan et al. 2013; Kurima et al. 2015; Jia et al. 2020). Another transmembrane protein, TMIE is likely to be a subunit of the transduction channel and binds to LHFPL5 (Cunningham et al. 2020). LHFPL5, which binds to the lower end of the tip link through protocadherin-15, may transmit tension in the tip link to the transduction channels (Xiong et al. 2012; Beurg et al. 2015). CIB2 has an affinity for Ca2+, a prominent ionic signal in the hair cells, and was reported to bind to the MET channel (Giese et al. 2017). Based on their homologs in Caenorhabditis elegans, CIB2 was also proposed to link the stereociliary actin core to the transduction channel through ankyrin proteins (Tang et al. 2020). However, direct evidence of such arrangement (TMC1/CIB2/Ankyrin) has yet to be confirmed in vertebrate hair bundles.
Transmembrane channel-like proteins 1 and 2 or TMC1/2 have been shown to be pore-forming proteins and are accepted as the main determinant of the conductive pathway for the transduction channel (Jia et al. 2020; Pan et al. 2013). Immunofluorescence and immunogold labelling localize the two proteins at the tip of the stereocilia shorter rows of mice hair bundles, where the mechanotransduction channels are located, but not at the tallest (Kurima et al. 2015). Both TMC1 and TMC2 are required for mechanotransduction and both are detected at early postnatal stages in mouse hair cells. TMC2 expression reaches its peak during the first postnatal week, around the onset of the mechanotransduction, and then declines to be undetectable at P10. During this time, TMC1 expression, which starts 2 – 3 days later than TMC2 expression, overcomes TMC2 expression around P6 – P7 (Kawashima et al. 2011). TMC1 expression then reaches saturation. The developmental switch between TMC2 and TMC1 is tightly coupled to properties of mechanotransduction and expression of important actin-binding proteins (Krey et al. 2020), as will be described further.
TMC1 molecules assemble as dimers (Pan et al. 2018). The transduction channels of OHCs and IHCs in the mouse cochlea show multiple conductance states in 50-pS increments (Beurg et al. 2018). In addition, it was shown in the same study for OHCs that the number of TMC1 molecules per MET complex increases from ~ 8 at the apex to ~ 20 at the base of the cochlea. Thus, there seems to be a varying number of channels per tip link and each MET complex is composed of multiple TMC1 molecules, with a gradient along the tonotopic axis of auditory organs for OHCs (but not for IHCs).
In mammalian hair cells, the single-channel conductance of the transduction channel was reported to be in the range of 50-300 pS (Beurg et al. 2018; 2006; Géléoc et al. 1997; K. X. Kim et al. 2013; Pan et al. 2018; 2013) while the single-channel conductance of purified TMC1 and TMC2 in liposomes is 40 pS and 35 pS (Jia et al. 2020), respectively. TMC1 and TMC2 also show distinct permeation properties. Hair cells expressing only TMC1 have a smaller single-channel conductance and lower Ca2+ permeability than those expressing only TMC2 (K. X. Kim and Fettiplace 2013). The developmental switch from TMC2 to TMC1 at the onset of the mechanotransduction may result in reducing the Ca2+ influx in the mature hair cells.

Table of contents :



Related Posts