The middle ear apparatus is composed of three endochondrial ossicles (the stapes, incus, and malleus) and two membranous bones, the tympanic ring and the gonium, which act as structural components to anchor the ossicles to the skull. Except for the stapes, these skeletal elements are unique to mammals and are derived from the first branchial arches. The ossicles form a chain linking the tympanic membrane (into which the manubrium of the malleus inserts) with the inner ear (with the stapes inserting into the oval window). Joints link the malleus to the incus and incus to the stapes. All mammals have a soft connection between the incus and the stapes, which makes the stapes a semi-independent vibrating system (Fleischer,1978). The malleus-incus articulation has a V-shape, the incus having the wedge and the malleus the notch. In many hearing disorders, the malleus and incus are malformed and/or the ossicles are fused (Mayer et al.,1997). For example, in patients with Treacher Collins syndrome (TCS), the malleus and incus are united into a single complex in almost 100% of cases (Phelps et al.,1981; Jahrsdoerfer et al.,1989; Takegoshi et al.,2000). Fusion of the malleus and incus is also observed in patients with Branchio-Oto-Renal (BOR) syndrome, although these fusions are complicated by additional defects in the inner ear (Melnick et al.,1976). Fusions of the malleus and incus result in conductive hearing loss, and therefore correct formation of this joint is very important for normal human hearing. Very little information is available with respect to formation of the malleal-incudo joint during normal development, this limitation preventing a clear picture of how defects in this structure form.
Morphology studies have shown that many joints arise from segmentation of continuous cartilage rods, as is observed in the digits. Prior to joint formation, the rounded chondrocytes flatten and become non-chondrogeneic. This non-chondrogenic region is called the interzone. The interzone may form due to the action of nearby skeletal elements where the growth of these elements would result in the flattening of interzone cells (Carey,1922) or there may be a population of prespecified cells that form this joint region (Holder,1977). Cavities form within the interzone and this separates the skeletal elements.
The malleus and incus have been shown to be a part of a larger condensation that includes Meckel's cartilage (Miyake et al.,1996). A single continuous first arch condensation was observed made up of three components: a rostral component, a core component for the major part of Meckel's cartilage, and a caudal component. Cellular arrangements differed between the three components, reflecting the different axes of growth. Cells in the caudal component were arranged concentrically along the axis of growth. Using three-dimensional imaging, Miyake et al. (1996) were able to show that the caudal component could be divided into two subcomponents that were connected at the top but then bifurcated. These two subcomponents gave rise to the malleus and incus. A separate condensation was observed for the stapes.
In the mouse, various knockout mice have been shown to have defects in the malleal-incudo joint. In addition to other ear defects, Eya 1 and Six1 mutants have fusion of the malleus and incus (Xu et al.1999; Zheng et al.,2003). Mutations in both Eya1 and Six1 have been shown to be responsible for BOR syndrome in humans, and the mouse knockouts phenocopy the human disorder (Abdelhak et al.,1997; Xu et al.,1999; Ruf et al.,2004). Six1 and Eya1 interact in the same pathway and are homologues of the Drosophila genes Sine oculis and Eyes absent (Halder et al.,1998). In the mouse, Eya1 has been shown to be expressed around the malleus and incus but is excluded from the developing joint at E15.5 (Tucker et al.,2004).
Gdf5 and Gdf6 are both expressed in the developing malleal-incudo joint (Settle et al.,2003; Tucker et al.,2004). Mutants for Gdf5 have defects in a subset of joints but the middle ear is unaffected (Storm and Kingsley,1996). In contrast, in Gdf6 mutants the malleal-incudo joint forms but the articular surface of the ossicles are disrupted and cell proliferation in these areas is reduced (Settle et al.,2003).
Bapx1, the vertebrate homologue of the Drosophila gene bagpipe, is also expressed in the developing malleal-incudo joint. Loss of this gene in non-mammalian vertebrates leads to fusion of the quadrate and articular, skeletal elements homologous to the mammalian malleus and incus (Miller et al.,2003; Wilson and Tucker,2004). However, no defect in joint formation is observed in the Bapx1 mouse knockout, although the shape of the malleus is altered (Tucker et al.,2004).
Emx2 is a vertebrate homologue of the Drosophila gene Empty spiracles. Emx2 mutant heterozygotes have a defect in the joint articulation between the malleus and the incus with the incus being much smaller creating a circular rather than a double V-shape. The homozygote completely lacks an incus and the articulation for the missing incus on the malleus is defective (Rhodes et al.,2003).
In humans and mice, fusion of the malleus and incus results in hearing loss. In guinea pigs and Chinchilla, however, fusion of the malleus and incus occurs naturally (Fleischer,1978; Asarch et al.,1975; Judkins and Li,1997). As animals, such as the chinchilla, have very sensitive hearing (Strother,1967), fusion in these cases cannot be thought of as degenerative as it is in the human and mouse. In these animals, the malleus and incus function as one unit. The development of the malleal-incudo complex in such animals has not previously been studied. The lack of a joint in the adult might, therefore, be due to failure of the joint to form, or due to a physical pushing together of the ossicles after initial joint formation. This report will investigate normal malleal-incudo joint formation in the mouse during development and compare this to that observed in the guinea pig embryo.
Initiation of the Malleal-Incudo Joint
In the digits, the joints form within an initially continuous Alcian blue staining cartilaginous condensation. In contrast, when the malleus and incus are stained with Alcian blue from E14.5 onwards, they always appear as two separate elements (Fig. 1A,B). From the work of Miyake et al. (1996), however, the presumptive malleus and incus appear fused into a single condensation at the most rostral end at Theiler stage 21.31, approximately E13. Alcian blue, which stains components of the extra cellular matrix associated with cartilage development, does not stain the caudal part of Meckel's cartilage prior to E14.5, and so we used in situ hybridisation for early cartilage markers and type II collagen lacZ reporter mice to investigate the development of the ossicles and joint region at earlier stages. Type II collagen is a marker of the onset of overt cartilage development (von der Mark et al.,1976). Skeletal elements express collagen type II while the joint interzone is collagen type II negative (Craig et al.,1987). Type II collagen expression is positively regulated by Sox9 (Bell et al.,1997). At E14.5, in situ for Sox9 and type II collagen shows the malleus and incus as two separate elements (Fig. 1C,D). At E13.5, however, a clear bridge can be seen between the rostral parts of the two ossicles uniting them at the top, creating a inverted U-bend shape (Fig. 1E). The joint separating the ossicles is thus initiated between E13.5 and E14.5. This inverted U-shape is even visible at E12.5 by in situ for type II collagen (Fig. 1F). The development of the ossicles can be clearly viewed in whole mount using type II collagen beta galactosidase reporter mice stained with lacZ. At E13.5, the malleus and incus can be seen united at the rostral ends, while the second arch derived stapes appears as a separate condensation (Fig. 1G,H). Again the inverted U-shape is clearly observed in sections through these stained heads (Fig. 1I).
Expression of Joint Markers
Although a morphological joint is not visible until E14.5, the region where the presumptive joint will form may have already been specified. Bapx1 and Gdf5 are clearly expressed in the middle ear joint at E15.5, once the joint region is established (Fig. 2E–G) (Tucker et al.,2004). We, therefore, investigated whether these joint markers were expressed prior to any morphological sign of a joint at E13.0. Gdf5 could be seen expressed between the malleus and incus marking the presumptive joint region (Fig. 2A,B). Bapx1 was expressed around the malleus but also weakly in the presumptive joint region (Fig. 2C). The expression of Gdf5 and Bapx1, therefore, precedes the split between the malleus and incus. Bmps are expressed in developing joints in the limbs. Bmp2 is expressed strongly in the digit joints as the interzone forms and remains in the joint during the start of cavitation (Francis-West et al.,1999). We, therefore, also looked at the expression of Bmp2 during middle ear joint development. Before and after initiation of the joint, Bmp2 was expressed around the ossicles but was not evident in the joint region (Fig. 2D,H). The role of Bmps in the digit joints may, therefore, not be mirrored in the middle ear.
Expression of Ossicle Markers
As the malleus and incus are initially derived from a single condensation, it is of interest to identify whether the two ossicles have a separate identity early on. In the mouse, Emx2 has been shown to play a role specifically in the formation of the incus, as the incus is lost in the Emx2 knockout (Rhodes et al.,2003). The expression pattern of Emx2 in the middle ear, however, has not been shown. We, therefore, followed the expression pattern of Emx2 in the middle ear Emx2 at E13.5 to E15.5. At 13.5, Emx2 is expressed strongly in the developing incus but not in the malleus, despite the fact that these two ossicles are united at this point in development (Fig. 3A,C). Genetically, the malleus and incus are, therefore, distinct at a stage when morphologically they appear as a single entity. At later stages, the expression of Emx2 becomes downregulated in the differentiated cartilage but remains at high levels in the surrounding undifferentiated cartilage (Fig. 3D,F).
To test whether the joint region falls within the Emx2 expression domain, we looked at expression of Gdf5, on serial sections to that of Emx2. At this stage, the presumptive joint region, as indicated by expression of Gdf5, was located within the Emx2 expression domain (Fig. 3A). This was also true at E14.5 (data not shown) and E15.5 (Fig. 3E).
Middle Ear Development in the Guinea Pig
The malleus and incus start out as a single element. In mammals where a single malleal-incudo complex is observed in the adult, a joint may simply never form to separate this structure during embryonic development. In the guinea pig, the malleus and incus are found as a single complex in the adult in contrast to the situation observed in the mouse (Fig. 4A,B). The guinea pig has a gestation period of between 55 to 75 days, compared to 19 to 20 days in the mouse. At E31, the guinea pig is at an equivalent stage to the mouse E14.5–E15.5, as indicated by the development of the other craniofacial structures, such as the teeth and salivary glands. At this stage in the mouse, the ossicles are clearly separated (Fig. 5A) and a very similar situation is observed in the guinea pig when stained in whole mount with alcian blue (Fig. 5B). At the same stage in section, a type II collagen negative joint interzone can be observed developing between the malleus and incus in the both the mouse and the guinea pig (Fig. 5C,D). The guinea pig joint region also expressed Gdf5, in a similar manner to the mouse (Fig. 5E–H). A joint, therefore, starts to form in the guinea pig at a similar equivalent stage to the mouse. The expression of Eya1 around the ossicles, but excluded from the joint region, was also similar in both species (Fig. 5I,J). As the mouse develops, the joint region between the two ossicles becomes narrower (Fig. 6A) and a very similar process is seen to occur in the guinea pig (Fig. 6B). In section at E37 in the guinea pig, equivalent to E18 to newborn in the mouse, the malleus and incus are distinct but pushed up against each other to give the impression of a single element (compare Fig. 6C with Fig. 6D). No cavitated joint region is observed between the two cartilages (Fig. 6F). This is in contrast to the situation observed in the mouse newborn, where the malleus and incus are separated by a thin joint region devoid of cells due to cavitation (Fig. 6E).
Under the light microscope, the fused malleus and incus in the adult appear as a single unit. When viewed by scanning electron microscope, however, a suture line can be observed running between the incus part and malleus part of the united single element running completely around the ossicles (Fig. 4C). The remnants of the initial joint region is thus still in evidence. Of particular interest is the fact that this suture line is not straight but forms a V-shape, akin to that seen in the normal mouse articulation.
From our early expression data and use of type II collagen lacZ reporter mice, it is clear that the malleus and incus are initially united at the rostral end. They are then split into two ossicles by the formation of a joint that develops at around E14.0. The forming joint can be viewed by the loss of expression of Sox9 and type II collagen in the developing interzone. Even before Sox9 and type II collagen expression are lost in the developing joint interzone, expression of the joint markers Gdf5 and Bapx1 are observed in the presumptive joint region. Expression of these genes in the presumptive joint, therefore, precedes any overt sign of joint initiation in the middle ear. This is interesting as the expression of Gdf6 in the middle ear has previously been reported as only occurring once the joint is established (Settle et al.,2003). Thus, the two Gdfs may play subtly different roles in the middle ear joint.
The transcription factor Emx2 is expressed in the developing incus from an early stage. In the Emx2 knockout, skeletal defects are also observed in the scapula and illium of the body (Pellegrini et al.,2001). This is interesting as these structures are the most proximal parts of the limbs. In keeping with this, the incus can be thought of as the most proximal part of the first arch derived skeletal elements. Emx2, therefore, may play a role in determining the identity of the most proximal elements. Although Emx2 has been shown to be necessary for formation of the scapula, illium, and incus, its expression may not be sufficient for the formation of these structures. Overexpression of Emx2 in the chick forelimb does not result in formation of a duplicated or enlarged scapula, the only defect being the development of an additional posterior digit (Prols et al.,2004). Emx2, therefore, appears to play a role in determining the relative position of skeletal elements along the proximal-distal axis.
The presumptive joint region also expresses Emx2. Emx2, therefore, appears to identify not only the incus but the joint region as well. This fits in with the fact that in the Emx2 knockout, in addition to loss of the incus, the malleus lacks its articulating surface (Rhodes et al.,2003). In a similar fashion to the articulation of the incus with the malleus, the illium of the hind limb normally articulates with the sacral vertebrae. In the Emx2, mutant loss/reduction of the illium is accompanied by loss of the articulatory surface on these vertebrae. This was proposed to be due to loss of an inductive signal that would normally be produced by the illium. This signal would potentially induce formation of the articulatory processes of the sacral vertebrae (Pellegrini et al,2001). In contrast, our data suggest that Emx2 may be expressed not only in the illium but also in the presumptive joint region, and it is loss of this expression domain that leads to the defect in the sacral vertebrae.
As the guinea pig has a single malleal-incudo complex, it provides a good model for identifying how fusions in these ossicles can arrive during embryonic development. In this study, we have shown that in the case of the guinea pig middle ear, a joint starts to form in between the malleus and incus in a similar manner to the mouse. The developing joint turns off type II collagen and switches on Gdf5 in a similar fashion to that observed in the mouse. The joint-forming process, however, is aborted prior to cavitation and the two ossicles are pushed together into a single complex. The force that leads to the pushing together of the two ossicles is unclear, but it may be that if the joint fails to cavitate, the cartilages on either side end up falling in on each other. The suture line where the two ossicles abut is still visible in the adult when viewed under a scanning electron microscope. This suture traces a V structure reminiscent of the articulation of the malleus and incus observed in most other mammals.
In the mouse and human, loss of Eya1 leads to fusion of the malleus and incus and so it was possible that the fusion of the malleus and incus observed in the guinea pig was due to reduced Eya1 signalling. The expression of Eya1, however, in the guinea pig matched that of the mouse. We cannot rule out, however, that the protein levels are different in the two species.
Guinea pigs and humans have a freely mobile middle ear, whereby the malleus is not fixed in place by fusion to the tympanic ring via the gonial. Instead, the malleus is linked to the tympanic via a ligament. Mice, in contrast, have a microtype middle ear arrangement with a fixed malleus and well-developed articulation between the malleus and incus (Fleischer,1978). Fusion of the malleus and incus is only observed in those mammals with a freely mobile middle ear arrangement. In the mouse, the fixation of the malleus has to be accompanied by a well-developed articulation to allow for a degree of flexibility in the middle ear. In the guinea pig, as the malleus itself is more flexible, the articulation between the malleus and incus is of reduced importance allowing for the fusion of these ossicles in the adult. The arrangement and degree of flexibility of the middle ear may affect the hearing range of the animal. The guinea pig and chinchilla ear is sensitive to low-frequency hearing, while the mouse ear is able to hear at high frequency (Miller,1970; Smith,1975). Fusion of the malleus and incus is seen in an extreme case in the African mole rat, Heliophobius. Here the malleus and incus are fused both at the site of articulation and along the long arm of the incus (Fleischer,1978). The mole rat lives underground for its entire life and, thus, hearing may not be of particular importance to this species.
In the human and mouse, fusion of the malleus and incus is observed in various syndromes and is associated with loss of hearing. It is unclear whether the fusions observed are due to a failure of the original joint to form, or whether the joint initially forms and then fails to progress, as observed here in the guinea pig. This is an important distinction to make, as it influences possible treatment of the defect observed. An analysis of ossicle fusion in mouse mutants that are models of human hearing disorders is, therefore, the next goal.
Haemotoxylin and Eosin/Alcian Blue and Chlorantine Fast Red
Sections were fixed in 4% paraformaldehyde and dehydrated to 100% ethanol. Paraffin wax sections were cut at 8 μm and split over 5 to 6 slides. Serial sections were stained with hematoxylin/eosin (Sigma) to visualize the tissue morphology or Alcian blue and chlorantine fast red to look at bone and cartilage.
Lac z Staining
Transgenic Collagen type II mice expressing a β galactosidase reporter were stained at E12.5 and E13.5 as described in Zhou et al. (1995).
In Situ Hybridisation
Slides were then prepared for S35 in situ hybridisation as described by Tucker et al. (1999). Serial sections were stained with haemotoxylin and eosin (Sigma) to visualise tissue morphology.
Probes used include Sox9, Collagen type 11, GDF5, Bmp2, and Emx2. Mouse Sox9 was linearised with Ecor-1 and transcribed with T3. Collagen type ll was linearised with Ecor-1 and transcribed with T3. GDF5 was linearised with Hind lll and transcribed with T7. Emx2 was linearised with BamHl and transcribed with T7. Bapx1 was linearised with Hind III and transcribed with T7. Bmp4 was linearised with Ecor1 and transcribed with SP6. Eya1 was transcribed with T3. Bmp2 was linearised with HindIII and transcribed with T7.
Skeletal staining of E14.5 middle ears was carried out using Alcian blue and Aliziran red to visualise the cartilage and bone, respectively, in a method devised by Kessel and Gruss (1991).
Type II collagen immunohistochemistry was performed using 11-116B3 antibody (Developmental Studies Hybridoma Bank) on paraffin wax sections. To enhance the signal, slides were microwaved in 0.01M citrate buffer (Shi et al.,1991) treated with chondroitinase ABC 0.25 U per ml and Hyalurondase 1.45 U per ml at 37°C for 45 min (Sigma). The collagen antibody was used at a dilution of 1:100.
Scanning Electron Microscopy
The middle ear ossicles of a guinea pig adult were dissected out and fixed in paraformadlehyde for several days. The ossicles were then washed in water and left to dry for a week before being prepared for SEM by the Centre for Ultrastructural Imaging at King's College London. Photographs were taken at a pressure of 0.72 Torr. Magnitude 175X, HV 2KV, WD 9.9 mm.
Thanks to Mathew Bradman for his assistance with the animals used in this report; to Tony Brain for his help with the SEMs and Eva Matalova for her help with the discussion; and to Benoit de Chrombrugghe for the gift of the Type II collagen lacZ reporter mice. The Eya1 probe was a gift from Pin-Xian Xu, the Emx2 probe was a gift from Antonio Simeoni, the Bapx1 probe was a gift from Robert Hill, the Gdf5 probe was a gift from Frank Luyton, the Sox9 probe was a gift from Robin Lowell-Badge, and the type II collagen probe was a gift from David Rice. The type II collagen antibody, developed by T.F. Linsenmayer, was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. S.A. was funded by the MRC.