axial speciﬁcation for sensory organs versus non-sensory … · sensory organs are largely...

11Development 125, 11-20 (1998)Printed in Great Britain © The Company of Biologists Limited 1998DEV1226

Axial specification for sensory organs versus non-sensory structures of the

chicken inner ear

Doris K. Wu*, Fabio D. Nunes and Daniel Choo

National Institute on Deafness and Other Communication Disorders, Rockville, MD 20850, USA*Author for correspondence (e-mail: [email protected])

Accepted 23 October 1997; published on WWW 8 December 1997

f

an

A mature inner ear is a complex labyrinth containingmultiple sensory organs and nonsensory structures in afixed configuration. Any perturbation in the structure ofthe labyrinth will undoubtedly lead to functional deficits.Therefore, it is important to understand molecularly howand when the position of each inner ear component isdetermined during development. To address this issue, eachaxis of the otocyst (embryonic day 2.5, E2.5, stage 16-17)was changed systematically at an age when axialinformation of the inner ear is predicted to be fixed basedon gene expression patterns. Transplanted inner ears wereanalyzed at E4.5 for gene expression of BMP4 (bonemorphogenetic protein), SOHo-1 (sensory organhomeobox-1), Otx1 (cognate of Drosophila orthodenticlegene), p75NGFR(nerve growth factor receptor) and Msx1(muscle segment homeobox), or at E9 for their gross

anatomy and sensory organ formation. Our results showedthat axial specification in the chick inner ear occurs laterthan expected and patterning of sensory organs in the innerear was first specified along the anterior/posterior (A/P)axis, followed by the dorsal/ventral (D/V) axis. Whereas theA/P axis of the sensory organs was fixed at the time otransplantation, the A/P axis for most non-sensorystructures was not and was able to be re-specified accordingto the new axial information from the host. The D/V axisfor the inner ear was not fixed at the time oftransplantation. The asynchronous specification of the A/Pand D/V axes of the chick inner ear suggests that sensoryorgan formation is a multi-step phenomenon, rather thana single inductive event.

Key words: Inner ear development, Gene expression, Sensory org

SUMMARY

ot

athates,

owionre

d

)nge.

re

sde

ral

INTRODUCTION

The chick inner ear originates from a cyst-like structure knowas the otocyst, which undergoes remarkable morphologchanges during development to give rise to a complabyrinth. The molecular mechanisms underlying thdevelopment of the inner ear and the formation of its eigsensory organs are largely unknown (Knowlton, 1967). further our understanding of these developmental processeis essential to identify when positional information of inner ecomponents is specified and to sort out causal relationshbetween different morphogenetic events.

The hair cells, the signal transducing apparatus commonall of the inner ear sensory organs, are morphologically simiyet the structures of various sensory organs differ. This raian interesting question about sensory organ generation. possible scenario is that different types of sensory organsformed by a gradual specification of fates. For example, positional information may determine which part of the otepithelium should become sensory versus nonsensory tisThen, different sensory organs may be achieved by seconinductive events. Alternatively, the formation of differensensory organ types may occur independently such that oncertain region of the otic epithelium is induced to form senso

nicallexeht

Tos, itarips

tolar,sesOne aretheicsue.darytce ary

tissue, the type of sensory organ it will generate is alsspecified. Currently, there is no direct evidence thadistinguishes between these and other possibilities (forreview see Fekete, 1996). Previous studies demonstrated BMP4 is an early marker for all of the sensory organs in thchick inner ear (Wu and Oh, 1996). Two additional genep75NGFR and Msx1, are preferentially expressed in thepresumptive crista, a vestibular sensory organ. We do not knwhether any of these gene products play a role in the inductof sensory organs, or if fates of different sensory organs aalready specified by the time these genes are activated.

In addition to BMP4, several genes that have been implicatein pattern formation such as Msx1(Suzuki et al., 1991; Wu andOh, 1996), Otx1(Simeone et al., 1993; Acampora et al., 1996and SOHo-1 (Kiernan et al., 1997) are also expressed iappropriate and restricted domains early in the otic cup staFor example, Msx1, a marker for the endolymphatic duct, isexpressed in the dorsal portion of the otic cup, a location wheone would expect the endolymphatic duct to originate. Otx1, amarker primarily for the lateral wall of the cochlear duct, iexpressed in the ventral portion of the otic cup (unpublisheresults), a location appropiate for future extension of thcochlear duct. Furthermore, SOHo-1, a marker for thesemicircular canals and their cristae, is expressed in the late

12

oursintll

andted

andxeds offineithre%

00aes,in

ereen, todasrateBS

nd the

6,

20)f

ut

is B,earererto

he aly, Indsehe

theeeir

D. K. Wu, F. D. Nunes and D. Choo

Fig. 1.Schematic diagram illustrating axial rotations of otocysts.Right otocysts of E2.5 (stage 16) chick embryos (hosts) werereplaced with age-matched left otocysts of donors. The axis that wasrotated in each case is boxed. No manipulation was performed on theleft otocyst before implantation for the anterior/posterior axialrotation (A/P R). In dorsal/ventral axial rotation (D/V R), the leftotocyst was rotated 180° before implantation. In medial/lateralrotation (M/L R), the left otocyst was flipped (indicated as dottedlines) so that the remaining hole in the otocyst was located on themedial, rather than the lateral side of the embryo.

wall of the otic cup. Based on these gene expression patteone may postulate that the axes of the chick inner ear mafixed by the otic cup stage. To test this hypothesis, we chanthe axes of otocysts systematically and later analyzed transplanted ears for gene expression patterns, gross anaand sensory organ formation. Our results provide evidecontrary to our hypothesis that axial patterning of the chinner ear was a later event and axial specification of sensorgans versus nonsensory structures may be different feach other.

MATERIALS AND METHODS

Transplantation procedureFertilized White Leghorn eggs (Truslow Farm, MD) were incubatat 38°C until stage 16 to 17 (E2.5, Hamburger and Hamilton, 195The right otocysts of host embryos were replaced by left otocystage-matched donors as illustrated in Fig. 1, so that only one axithe otocyst was rotated in each case. For normal controls, rotocysts from donor embryos were used. Donor embryos wtransferred to a Petri dish containing Hanks Balanced Salt Solu(HBSS). Otocysts designated for transplantation were sprinkled wDiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethyl indocarbocyaninepercholorate; Molecular Probes) crystals, dissected from surroundtissues using etched tungsten needles, and tranferred to a Petrcontaining small drops of HBSS. The location of the remainiopening in the dorso-anterior part of the otocyst and positions ofDiI crystals guided the orientation of donor otocysts durintransplantation. Right otocysts of host embryos were removed witungsten needle and replaced with age-matched donor otocysts bon the size of the remaining opening in the otocyst. After surgery, ewere sealed, returned to the incubator, and harvested at 48 h(E4.5, stage 24-25) for in situ hybridization, or 1 week later for pafilling and whole-mount immunostaining with anti-HCA (hair cespecific antigen) antibodies (Bartolami et al., 1991).

Paint-filling and whole-mount immunostaining of innerearsOperated chicken embryos were killed at E9 (stage 35, 36) processed for gross anatomical evaluation by filling the transplaninner ears with paint (Martin and Swanson, 1993; Bissonnette Fekete, 1996). For whole-mount immunostaining, embryos were fiovernight in 4% paraformaldehyde, and the membranous portioninner ears were dissected from their surrounding tissues using forceps (No. 5) under a SV-11 dissecting microscope (Zeiss) woblique fiber optic illumination. The dissected inner ears weincubated in Dulbecco’s modified Eagle’s medium containing 10fetal calf serum, 0.4% Triton X-100 and anti-HCA antibody (1:20,0dilution) at 4°C for 24 hours, followed by incubation with biotinylated anti-mouse secondary antibody (Vector Laboratori1:500 dilution) under the same conditions. Multiple washings phosphate-buffered saline (PBS) containing 0.1% Triton X-100 wperformed overnight, following each of the antibody reactions. Thinner ears were incubated with alkaline phosphatase coupledstreptavidin (Zymed, 1:500 dilution) overnight and washeextensively the next day. Detection of antibody complex wperformed using 5-bromo-4-chloro-3-indolyl phosphate as a subst(Stoker and Bissell, 1987). Color reactions were stopped with Pcontaining 1 mM EDTA.

Probes and in situ hybridizationRiboprobes for BMP4, Msx1 and p75NGFR were prepared asdescribed (Wu and Oh, 1996). The chick Otx1 riboprobe wasgenerated using RT-PCR containing E3 chick otocyst RNA adegenerate primers recognizing motifs common to members of

rns,y begedthe

tomy,nceickory

rom

ed1).

s ofs ofighteretionith

ingi dishng theg

th aasedggs

Otx gene family (5′ and 3′primers correspond to amino acids 28-3and 176-184 of the mouse Otx1 gene, respectively; Simeone et al.1993). Only clones with highest homology to the mouse Otx1sequences were obtained (number of clones sequenced = suggesting that only Otx1is expressed in the chick otocyst. One othe PCR clones was used to generate a riboprobe for Otx1. Riboprobesfor SOHo-1were generated using the 3′ untranslated region of SOHo-1 cDNA (Deitcher et al., 1994). In situ hybridization was carried oas described (Wu and Oh, 1996).

RESULTS

Anatomy of a normal inner earA paint-filled membranous portion of a right inner ear at E9shown in a lateral and a postero-lateral view in Fig. 2A andrespectively. By this age, the gross anatomy of the inner was fairly well-formed and resembled that of a mature innear (Bissonnette and Fekete, 1996). Similar to othvertebrates, the chick inner ear can be roughly divided indorsal vestibular and ventral auditory components. Tauditory component of the chick, the cochlear duct, wasrelatively straight tube that extended medially and ventralrather than a coiled structure as in higher vertebrates.addition, both the proximal (cd in Fig. 2A) and the distal en(cd in Fig. 2B) of the cochlear duct pointed towards thposterior part of the body, forming an arc-shaped structure. Tvestibular component consisted of two connecting sacs, utricle and the saccule (u and s in Fig. 2A,B), thresemicircular canals (ssc, psc and lsc in Fig. 2A,B) and th

13Axial patterning in the chicken inner ear

andne.ittalht

(pscsc inwn theneeo

euchre

ghtrntater,larostscseint

ryuntntionndedureralandC).ns

arct),ig.-

Fig.ral

00vedbeinystalre

nsyere

Fig. 2. Normal anatomy of an E9 right inner ear. Membranousportions of inner ears were either filled with paint (A,B) or stainedwith anti-HCA antibodies (C,D,E), and are shown in lateral (A),postero-lateral (B,C), medial (D) and an anterior (E) views. Theinner ear in B is tilted dorsally to better reveal the lateral canal. Inand B, there are two ampullae at the anterior part of the inner earand la) and only one in the posterior (pa). The superior ampulla (sis the only one in a vertical position and the other two are in ahorizontal position (A). The superior semicircular canal (ssc) is in sagittal plane, the posterior semicircular canal (psc) is in a transvplane and the lateral semicircular canal (lsc) is in a horizontal plan(A and B). The posterior canal (psc) has the most superior andanterior insertion point on the commom crus (cc in A). The supericanal (ssc) has a posterior and a slightly more ventral point ofinsertion on the common crus and the lateral canal has the mostventral and posterior point of insertion. The endolymphatic duct (ehas a dorso-posterior projection and the cochlear duct (cd) has aventro-posterior projection. All eight sensory organs revealed byanti-HCA staining are shown in C, D and E: superior crista (sc inC,D,E), lateral crista (lc in C,E), posterior crista (pc in C,D), macuutriculi (mu in C,D,E) and sacculi (ms in C,D), basilar papilla (bp iC), lagena (lg in C) and macula neglecta (mn in D). The superior in E) and the posterior (pc in C,D) cristae have a W-shaped patterand the lateral crista (lc in E) has a V-shaped pattern. The patch ostaining above the W-shaped pattern of the superior crista (sc) in an artifact resulting from trapping within the gelatinous material ofthe cupula. D, dorsal; A, anterior; L, lateral; ed, endolymphatic dula, lateral ampulla; pa, posterior ampulla; s, saccule; u, utricle. Scbars, 100 µm; bar in A also applies to C.

corresponding ampullae, which contain the cristae (sa, la pa). Each semicircular canal was situated in a different plaThe superior canal, the largest of the three, was in a sagplane (ssc in Fig. 2A,B). The posterior canal formed at a rigangle with the superior canal and was in a transverse plane in Fig. 2A,B). The lateral canal was in a horizontal plane (lin Fig. 2A,B). Each canal ended with the common crus (ccFig. 2A) on one side and a swelling on the other side knoas the ampulla. There were two ampullae, the superior andlateral, located anteriorly (sa and la in Fig. 2A), and oposterior ampulla, situated posteriorly (pa in Fig. 2A,B). Thsuperior ampulla was positioned vertically while the other twampullae were positioned horizontally (Fig. 2A,B). Thsuperior and posterior canals orignated as a common outpostarting at E4. Over the next 2 days, the primordial structufor the posterior canal gradually became situated at a riangle to the primordial superior canal. Most likely, this pattewas achieved by differential growth of the otic epithelium. Aa similar time in development, the two opposing epithelia the center of each primordial canal structure came togethfused and were reabsorbed, thus forming the two semicircucanals. After reabsorption, the posterior canal has the msuperior and anterior insertion point on the common crus (pin Fig. 2A,B); slightly more ventral and posterior to that wathe insertion for the superior canal (ssc in Fig. 2A,B). Thlateral canal has the most ventral and posterior insertion po(lsc in Fig. 2A,B).

Sensory organs of a chick inner earAll eight sensory organs in the chick inner ear, one auditoorgan and seven vestibular, can be revealed with whole-moanti-HCA antibody staining of only the membranous portioof an inner ear. Examples of inner ears after such a preparaare shown in postero-lateral (Fig. 2C), medial (Fig. 2D) aanterior (Fig. 2E) views. Sensory organs can be dividroughly into two groups, anterior and posterior. The fosensory organs in the anterior group, the superior and latcristae (sc and lc) and the maculae utriculi and sacculi (mu ms), can be seen clearly in a postero-lateral view (Fig. 2The posterior group also contained four sensory orgaincluding the posterior crista (pc in Fig. 2C,D), the basilpapilla (bp in Fig. 2C, the sensory organ of the cochlear duthe lagena (lg in Fig. 2C) and the macula neglecta (mn in F2D). Furthermore, HCA staining of hair cells showed a Wshaped pattern in the superior and posterior cristae (sc in 2E and pc in Fig. 2C,D), and a V-shaped pattern in the latecrista (lc in Fig. 2E).

Inner ear transplantationAnterior/posterior rotationsOtocyst transplantations were carried out in more than 3embryos. On the average, about 25% of the embryos survithe transplantation and paint-filling procedures and could analyzed (Table 1). Among the eight surviving controls, which the right otocyst of a host was replaced by a donor otocoriented properly, six of these inner ears displayed normanatomy as indicated by the paint-filling technique (compaFig. 3A with Fig. 2B) and correct topology of sensory orgaby anti-HCA staining (compare Fig. 3E with Fig. 2C). In manof the rotated transplantations incomplete inner ears w

A (saa)

aersee

or

d)

laen(scnfE is

ct;ale

14

ndear inite

ngryxialchare

rynging

ares

ansCryW-culi).

lso inerthear

theime

fill,oveateIning 2).


Table 1. Summary of structures formed in rotated, paint-filled inner ears

No.

CystA/P

Trans- No. withSCC† Ampulla† Orientation

planted Evaluated Cyst CD ED 1 2 3 1 2 3 CD N P.R. N.D.

CTRL 30 8 0 1 7 0 1 6 0 0 7 8 7 0 1A/P R 73 20 1 0 16 1 5* 13 0 4 15 18 0 17 3D/V R 78 24 2 4 5 4 8 6 4 8 6 20 7 0 17M/L R 44 18 4 7 10 5 2 0 2 3 0 13 0 0 18

†Specimens were grouped by the number of semicircular canals or ampullae formed. *One of the specimens had two partially formed canals.Cyst, rudimentary inner ear with no differentiation. Cyst with CD, rudimentary inner ear with only a ventral cochlear extension. A/P orientation: N, normal, P.R., A/P orientation partially reversed, N.D., orientation could not be determined.ED, endolymphatic duct; SCC, semicircular canals; CD, cochlear duct.

formed; the number of inner ear structures obtained after etype of rotation are summarized in Table 1.

Three examples of A/P rotated inner ears are shownlateral (Fig. 3B) and postero-lateral (Fig. 3C,D) viewAlthough these particular inner ears looked fairly normal acontained three semicircular canals and a cochlear duct, cexamination indicated that there were now two ampullaeand la in Fig. 3B,C,D) in the back of the inner ears and oone ampulla (pa in Fig. 3B,C,D) in the front. Furthermore, vertical position of one of the ampullae in the back of the inear suggested that it may be a superior ampulla (sa in 3B,C,D). Taken together, these results suggest that ampullae are formed according to the axis befotransplantation.

In contrast, the semicircular canals were mostly pattercorrectly according to the body axis of the host. In all thexamples shown (Fig. 3B,C,D), the superior canal (ssc*) the largest and situated in a sagittal plane. Presumably, twere posterior canals that had assumed the pattern of supcanals in the host. Similarly the posterior canals (psc*), whwithout rotation would have been superior canals, were nsituated in a transverse plane like posterior canals. Whereapatterns of these two semicircular canals were usually coraccording to the host, only a few inner ears displayed a nopattern of canal insertion on the common crus (n=4, Fig.and C). In most A/P rotated inner ears, the insertion pointthe superior and posterior canals on the common crus wsymmetrical (Fig. 3D). In each of the A/P rotated ears tcontained a lateral canal (lsc*), the canal was always situin a horizontal plane (compare Fig. 3B,C,D with Figs 2Aand 3A), but its insertion point on the common crus was nventral and anterior instead of the normal ventral and poste(lsc* in Fig. 3B,C,D), suggesting that some aspects of caformation were fixed at the time of transplantatioFurthermore, the cochlear duct was only partially respecifi

Table 2. Distribution of senso

NumberCrista†

analyzed 1 2 3 >3

A/P R 7 0 0 7 0D/V R 11 1 1 4 5M/L R 13 4 5 1 1

†Specimens were grouped by the number of cristae and maculae form*Papilla and lagena sensory organs combined as one, see Fig. 3H.

ach

ins.nd

loser (sanlythenerFig.there

nedreewasheseeriorichows therect

rmal 3Bs ofere

hatated,Bowriornaln.ed.

Most of the cochlear ducts in A/P rotated ears failed to extefully (cd in Fig. 3B). In ones that have a reasonable cochlduct, the distal end of the duct extended posteriorly asnormal inner ears (compare Fig. 3C,D to Figs 2B, 3A), despthe fact that the proximal end of the duct was now pointianteriorly (compare Fig. 3G,H to 2D). Another nonsensostructure that was also patterned according to the ainformation of the host was the endolymphatic duct whipointed towards the posterior part of the inner ear (compFig. 3F with 2C; 3G with 2D).

A more detailed investigation of the organization of sensoorgans in A/P rotated ears using HCA antibody stainiverified that all the sensory organs were patterned accordto the body axis of the donor. Examples of such inner earsshown in a postero-lateral (Fig. 3F) and two medial view(Fig. 3G,H). In these ears, the anterior group of sensory orgwas now located posteriorly (compare Fig. 3F with Fig. 2and 3E; Fig. 3G,H with Fig. 2D). Three of the four sensoorgans within the anterior group, the superior crista (sc, shape), the lateral crista (lc, V-shape) and the macula utri(mu) are shown clearly in the postero-lateral view (Fig. 3FThe fourth sensory organ, macula sacculi, which was alocated in the back of the inner ear, is shown more clearlythe two medial views (ms in Fig. 3G and H). On the othhand, the posterior crista (pc) and the proximal end of cochlear duct (cd) were located in the front of the inner e(bp in Fig. 3G and bp+lg in 3H). These data showed that A/P axis for the sensory organs was already fixed at the tof transplantation.

Among 20 A/P rotated ears that were analyzed by paint-17 showed the partially reversed A/P pattern described ab(Table 1). The remaining three inner ears did not differentienough to determine their axial orientation (Table 1). addition, all seven inner ears processed for anti-HCA stainshowed a reversed A/P pattern for sensory organs (Table

ry organs in rotated inner earsMacula†

Basilar Papilla1 2 papilla Lagena + lagena*

7 6 2 2 35 6 2 2 211 2 0 2 0

ed.


sedhe

re

Taken together, these data showed that while the A/P axithe sensory organs was fixed at the time of transplantationA/P axis for most of the nonsensory structures was not was able to be re-specified according to their new axinformation.

The BMP4gene expression pattern of A/P rotated inner eat 48 hours after transplantation showed that all presump

Fig. 3.Anatomy of A/P rotated inner ears and controls. (A) and (Eotocyst from another embryo. The saccule (s) and the cochlear ducontained an endolymphatic duct which also did not fill with paint.either filled with paint (A-D) or stained with anti-HCA antibodies (Eand H are medial views of right inner ears. The inner ear in A is tilIn A/P rotated ears, there are two ampullae in the back of the inneaddition, only one ampulla is found in the front of A/P rotated ears(ssc*) and the posterior (psc*) canals have the correct insertion pasuperior and posterior canals are correct in D, their points of inser(lsc*) shown in B, C and D has a ventral and anterior point of inseas in the control (lsc in A). Anti-HCA staining revealed that the anlateral crista (V shape, lc in F), maculae of the utricle (mu in F,G,H(E), whereas the proximal end of the papilla (bp in G, and bp+lg inear. In A/P rotated ears that have reasonable extensions of the co(C,D) as in the control (A). The endolymphatic duct (ed) in A/P rotorientations. ed, endolymphatic duct; lg, lagena; u, utricle. ssc*, ppsc*, superior canal that has adopted the pattern of a posterior cacommon crus. Scale bars, 100 µm; bar in B also applies to H, bar in C

s of, theandial

arstive

sensory patches were present although in an A/P-reverpattern (data not shown). This result is consistent with thypothesis that BMP4 gene expression is associated withsensory organ formation.

Dorsal/ventral rotations The phenotypes of the D/V rotated inner ears were mo

) are rotation controls such that the right otocyst was replaced by a rightct (cd) in A did not fill completely with paint. Inner ears shown in (B,C,D) all

(B-D) and (F-H) are inner ears with their A/P axes rotated. Inner ears were-H). A, C, D, E and F are postero-lateral views, B is a lateral view and Gted dorsally to better reveal the lateral canal. F and G are from the same ear.r ear (sa and la in B,C,D) instead of one as in the control (pa in A). In (pa in B,C,D) instead of two as in the control (sa and la in A). The superiorttern on the common crus in B and C, as in A. Whereas the pattern of thetion on the common crus are not (ssc* and psc* in D). Each lateral canalrtion on the common crus, rather than a ventral and posterior insertion pointterior group of sensory organs: the superior crista (W shape, sc in F,G,H),) and saccule (ms in G,H) are in the back of the inner ear instead of the front H) and the posterior crista (pc in F,G,H) are located in the front of the innerchlear duct, the distal end still points towards the posterior part of the inner earated ears also points dorsally and posteriorly (F,G). Refer to Fig. 2 forosterior canal that has adopted the pattern of a superior canal in host embryo;nal in host embryo; lsc*, lateral canal with an anterior ventral insertion on the also applies to D-G.

16

0%uctarslear

ars18emner

ulartionheller

(psc)ointtralralederse

wasuroryof(e).s

e 2,ted2).salor ofD/Venotht the

interatedowedtinghan

tionend

tae

asuctof itsarnaars.e

edre


Fig. 4.Anatomy of D/V rotated inner ears. D/V rotated inner earswere either filled with paint (A,B) or stained with anti-HCAantibodies (C,D). A and B are lateral views, C is an anterior view,and D is a medial view of D/V rotated inner ears. The pattern of thesemicircular canals in (A) is incorrect; most strikingly the lateralcanal (lsc) is not in a horizontal plane. The lateral crista and canal aremissing in (B). Despite the fact that many D/V rotated ears arerudimentary in structure, their sensory organs are always in correctD/V and A/P orientations (C). Very often, multiple cristae (cr) areformed in the dorsal portion of the inner ear (D). M, medial; bp,basilar papilla; la, lateral ampulla; lc, lateral crista, lg, lagena; m,macula; ms, macula sacculi; mu, macula utriculi; pa, posteriorampulla; pc, posterior crista; psc, posterior semicircular canal; sa,superior ampulla; sc, superior crista; ssc, superior semicircular canal;u, utricle. Scale bars, 100 µm; bar in A also applies to B and C.

abnormal than the A/P rotated ears (Table 1). Only about 2of the D/V rotated inner ears contained an endolymphatic d(Table 1). In most cases (18 out of 24, Table 1), the inner econtained at least one semicircular canal dorsally, and a cochduct ventrally, as shown in Fig. 4A,B. The rest of the inner ewere unidentifiable in their axial orientations. Among the inner ears that had a correct D/V orientation, seven of thdifferentiated enough to indicate that the A/P axes of these inears were also correct (Table 1). The pattern of semicirccanals was usually not correct, nor were their points of inseron the common crus (compare Fig. 4A,B to Fig. 2A). In texample shown in Fig. 4A, the superior canal (ssc) was smathan usual and contained an appendage. The posterior canalwas at a right angle to the superior canal, but its insertion pon the common crus was incorrect, being posterior and vento the superior canal (compare Figs 4A and 2A). The latecanal (lsc) was frequently missing (Fig. 4B), but when formwas usually the most severely affected, located in a transvplane instead of a horizontal plane (Fig. 4A).

To address whether the D/V axis of the sensory organs also fixed at the time of transplantation, we first focused oattention on the anterior group of sensory organs. If sensorgans were also fixed in the D/V axis at the time transplantation, macula(e) should form dorsal to the cristaAll the D/V rotated inner ears stained with HCA antibodie(n=10) showed cristae located dorsal to the maculae (TablFig. 4C,D). 7 out of 11 of these inner ears also differentiaenough to show a correct A/P pattern (Fig. 4C, Table However, multiple cristae were often observed in the dorportion of the inner ear (Fig. 4D, Table 2). The posterisensory structures were also not specified at the timetransplantation because we never observed a switch in the alignment of the posterior crista and basilar papilla. Taktogether, these results indicated that the D/V axes for bsensory and non-sensory structures were not specified atime of transplantation.

To address whether the D/V axis is fixed later development, we extended D/V rotation experiments to lastages. Among 13 inner ears analyzed that were operbetween stages 18 (E2.5) and stage 21/22 (E3.5), none shstructures of the macula(e) dorsal to the crista(e), suggesthat the sensory organs are specified in the D/V axis later tE3.5.

The gene expression patterns 48 hours after transplantacorrelated well with the inner ear structures obtained at the of a week. Overall, the gross expression patterns of SOHo-1,a marker for all three semicircular canals and their cris(Kiernan et al., 1997), and Otx1, a marker primarily for thecochlear duct (Fig. 5E), were consistent with the ensuphenotypes of the D/V rotated ears. In the most dorsal sectiat the level of the common outpouch for the superior aposterior canals, SOHo-1 expression was identical to thecontrol (compare Fig. 5A and B). In slightly more ventrsections, at the level of presumptive cristae, SOHo-1 geneexpression in the control was restricted to the lateral portionthe inner ear (Fig. 5C). In the D/V rotated ears, SOHo-1geneexpression at this level usually expanded into the medportion of the inner ear as well (Fig. 5D, n=4). In the chiinner ear, Otx1 was mainly expressed in the lateral wall fomost of the cochlear duct (Fig. 5E) and in the entire oepithelium at the most ventral tip of the duct (data not show

ingons,nd

al

of

ialckrticn).

The expression of the Otx1did not overlap with thepresumptive sensory area, the basilar papilla, which wlocated in the posterior and medial portion of the cochlear d(bp in Fig. 5F). In the D/V rotated inner ears, the location the basilar papilla, when present, was not always located incorrect posterior and medial position (bp in Fig. 5H). A cledistinction between presumptive basilar papilla and lagesensory areas also could not be made in D/V rotated eIdentification of presumptive sensory tissues within thcochlear duct were based on BMP4hybridization signal on theventral portion of the inner ear, which usually has a restrictlumen at that level (Fig. 5G,H). Whenever there wepresumptive sensory tissues found in the cochlear duct, BMP4


rser

In

ion patterns of SOHo-1, Otx1, BMP4and p75NGFRin D/V rotated innerhorizontal sections such that the anterior part of the embryo is towardsor D/V rotated (D/V) inner ear sections are indicated on the top left, and is on the top right hand corner of each panel. (A-D) are sections from D/V rotated ears (B), SOHo-1is expressed in the primordial structure

posterior canals, as in normal (A). Note the presence of the(ed) in the control ear (A) but not in the D/V rotated ear (B). At the levelristae, SOHo-1expression in a normal ear is restricted to the lateralt as well as part of the eighth ganglion (gVIII) (C). In D/V rotated ears,ene expression often expands into the medial part of the otocyst (D).om the same embryo. (E/F) and (G/H) are two pairs of 12 µm adjacenttx1or BMP4mRNA. In normal ears, the basilar papilla (bp) is locatedl portion of the cochlear duct (F), whereas Otx1 is expressed in the

chlear duct (E), non-overlapping with the expression of BMP4. In D/Vation of the basilar papilla (bp) is not always consistent, and sometimes itral wall of the cochlear duct in (H). However, the gene expression Otx1are non-overlapping (G,H) with each other. BMP4genef D/V rotated ears at the level of presumptive cristae (J) resembles that of

m a normal ear (I), with two medial hybridization signals. These twoe on more ventral sections (K). An adjacent section to (K) probed for shows that the BMP4positive area is also positive for p75NGFR,presumptive sensory tissue is most likely to become cristae. bp, basilarphatic duct; ms, macula sacculi. Scale bar, 100 µm.

expression usually did not overlap with that of the Otx1(compare Fig. 5G,H; n=4). The D/V rotated inner ear showin Fig. 5B,D did not contain an endolymphatic duct; howevin other specimens where anendolymphatic duct was evident, it waspositive for Msx1(data not shown).

Identification of different sensoryorgans based on BMP4 gene expressionpattern in D/V rotated ears was morecomplicated than in A/P rotated ones.BMP4gene expression in the most dorsalsections of a D/V rotated ear alwaysappeared as two medial hybridizationsignals (Fig. 5J) which resembled theexpression pattern of ventral sectionsfrom a control inner ear (Fig. 5I). Thesetwo medial hybridization signals in theD/V rotated ears soon merged togetheron the medial side of the inner ear onmore ventral sections (Fig. 5K). Otherventral BMP4hybridization signals werenot always consistent or easilyidentifiable. However, the medial BMP4hybridization signal always overlappedwith that of the p75NGFR(Fig. 5L),which has been shown to be a marker forthe cristae only (Wu and Oh, 1996),suggesting this presumptive sensory areawas destined to form cristae. In addition,the medial expansion of SOHo-1geneexpression at this level (Fig. 5D) alsoidentified this area to form either cristaeor semicircular canals. These geneexpression patterns correlated well withthe phenotype of cristae forming dorsalto maculae (Fig. 4C) as well as thephenotype of ectopic cristae observed atsubsequent stages (Fig. 4D).Furthermore, the aberrant BMP4 geneexpression pattern in D/V rotated earsalso supported the finding that sensoryorgans were being modified according totheir new axial information.

Medial/lateral rotationsThe M/L rotated inner ears were by farthe most severe among the three differentrotations. The inner ears resulting fromM/L rotations were often rudimentaryand lacked good semicircular canals andampullae development (Fig. 6A, Table1). However, more than half of the innerears analyzed contained anendolymphatic duct on the medial sideand a medio-ventrally extended cochlearduct (Fig. 6B, Table 1), indicating thatsome M/L aspects of the non-sensorystructures were not fixed at the time ofrotation.

Unfortunately, most presumptivesensory organs failed to differentiate

Fig. 5.Gene expressears. All panels are the top. Control (C) the RNA probe usedthe same embryo. Infor the superior andendolymphatic duct of the presumptive cportion of the otocysat this level, SOHo-1g(E-H) are sections frsections probed for Oin the postero-medialateral wall of the corotated ears, the locis located in the latepatterns of BMP4andexpression pattern oa ventral section fromedial signals mergp75NGFRmRNA (L)suggesting that the papilla; ed, endolym

ner,

properly after M/L axial rotations. Among the 13 inner eaanalyzed for sensory organ formation (Table 2), only one innear contained a crista on the medial side (cr in Fig. 6B).

18

s

n

oak

llyhe

p

m

ise

oftnPne,r

ickh

streate/Pst

clso),heesin

ndthelts

;l

edined

eryhest/P


Fig. 6. Anatomy and gene expression of M/L rotated inner ear. Alateral view of a paint-filled M/L rotated ear is shown in (A). Theinner ear is rudimentary in structure with no semicircular canals. Tampullae (am) are bulge-like structures rather than the normal domshape (see Fig. 2). A medial view of a HCA stained inner ear isshown in (B). This inner ear contains an endolymphatic duct (ed) aa rudimentary semicircular canal (scc). There are two cristae (cr),one in the medial and one in the posterior part of the ear. Eventhough this rotated inner ear contains a cochlear duct (cd), only thlagena (lg) can be detected within the duct (B) and not the basilarpapilla. (C-E) are 12 µm cryosections probed for SOHo-1(C), Otx1(D) and BMP4(E) mRNA by in situ hybridization. D and E areadjacent sections. In C, SOHo-1gene expression is restricted to themedial side of the inner ear rather than the lateral as in the contro(Fig. 4C). Expression of Otx1(D) often expands into the BMP4positive, presumptive basilar papilla area (E). Scale bar, 100 µm

addition, despite the fact that most of the M/L rotated paifilled ears contained a cochlear duct (Table 1), most of thducts probably did not contain any sensory tissue, since bapapilla was not detected in any of the inner ears processedanti-HCA staining (Table 2).

Gene expression studies of M/L rotated inner ears ashowed variable results. Most M/L rotated inner ears that wprobed for SOHo-1 (n=7), did not show any hybridizationsignals in their dorsal sections. This is correlated with tabsence of semicircular canal formation at E9 (data shown). At the level of the presumptive cristae, where SOHo-1 expression was normally restricted to the lateral portionthe inner ear, SOHo-1expression was often found restricted tthe medial side of the rotated ears instead (Fig. 6C). BMP4gene expression patterns among M/L rotated inner ears winconsistent. Instead of the normal five promine

nt-eseilar for

lsoere

heot

ofo

erent

hybridization signals by this age (Wu and Oh, 1996), only twor three foci were present and the rest of the signals were weand dispersed in the rotated ears. Gene expression of Otx1 inM/L rotated ears also showed a trend that was not usuaobserved in the D/V rotated ears. In normal ears, most of tcochlear duct had restricted Otx1expression on its lateral wall(Fig. 5E) and only the otic epithelium at the most ventral tiof the cochlear duct was entirely positive for Otx1 (data notshown). In contrast, in M/L rotated ears, Otx1expression wasoften found to spread around most of the cochlear epitheliu(Fig. 6D, n=5) and overlapped with that of the BMP4(Fig. 6E).It is not clear if the expansion of Otx1expression into a BMP4positive area affected sensory organ development, but thoverlapping pattern was observed only partially in one of thD/V rotated ears.

DISCUSSION

Antero-posterior axial specification before dorso-ventralTransplantation studies of axolotls showed that the A/P axis the ear rudiment is fixed shortly after neural tube closure. Athis stage, the otic placode has not formed. Yet grafting of aear rudiment from the contralateral side, so that only the A/axis is changed, resulted in an A/P axial reversal, but aotherwise normal inner ear (Harrison, 1936; Hall, 1939). ThD/V axis of an axolotl inner ear is specified at a later timebefore the otocyst is completely closed. Similar to axolotls, ouresults presented here show that the patterning of the chinner ear is determined by sequential axial information sucthat the A/P axis is specified before the D/V axis. In contrato axolotls, the developmental times when these axes wefixed in the chick seem to be much later. Our results show ththe A/P axis for the sensory organs is already fixed at thotocyst stage (stage 16, E2.5). We have extended the Arotations to earlier stages and our preliminary results suggethat the A/P axis of the chick inner ear is fixed during the oticup stage (stage 12, E2) (data not shown). We have aextended our D/V rotation experiments to stage 21/22 (E3.5an age when the endolymphatic duct is already formed on tmedial and dorsal part of the otocyst, yet the D/V axis of thinner ear could still be respecified after D/V rotations. This ian unexpected result given that many genes implicated pattern formation, such as Msx1, Otx1and Pax2(Nornes et al.,1990; Torres et al., 1996), are expressed in appropriate arestricted domains at the otic cup stage, suggesting that inner ear may already be patterned. However, our resucorrelate with previous reports that the Pax2gene expressionwas altered after otocyst rotations (Lindberg et. al., 1995Barald et al., 1997). Furthermore, the sequential axiapatterning in the inner ear is similar to what has been reportin amphibian neural plate (Roach, 1945) and chick hindbra(Simon et al., 1995). In both systems, A/P axes are determinbefore the D/V axes.

Phenotypes of rotated inner earsOur results show that at the time of transplantation, when thotocyst was almost formed, only the A/P axis of the sensoorgans, and maybe their M/L axes, were specified. Among tthree axial rotations, A/P rotated ears contained the highepercentage of normal inner ear structures, even though their A

hee-

nd

e

l


s, ellal cell6).for beon.

theld

on- inveory

ofusnye at

ofgle

ofnt/L

nadynot

/Vganct

ingesusy amsueory./V

s

xisthats.hats.ans is

ltser,

earese

orientations were not correct. In axolotls, when grafts wemade during A/P axial specification, enantiomorphic eacontaining two anterior or two posterior halves often resulteThese enantiomorphic phenotype was attributed to paradjustment to the new environment. In addition, since the rudiment was not a distinct entity at the time of operation, graof partial ear rudiments or incomplete removal of host etissues could have also resulted in these duplicated labyrinThis latter possibility does not apply to the chick transplantatstudies, since the inner ear was already a well defined cysthe time of operation. Of all the A/P rotated ears performed, have not found any ears with two anterior or posterior halvThis may be because our operations were carried out at a swhen the A/P axis was already fixed rather than during the aspecification period. Unfortunately, stages slightly younger ththe age of operation were difficult to study because the otic was half-open, and it was difficult to maintain the integrity the donor otic cup in the host embryo.

The first structure that forms after closure of an otocyst is endolymphatic duct, which originates as a dorsal protrusaround stage 21 (E3.5), and the appendage is gradually displmedially. In our studies, 80% of D/V rotated ears failed develop an endolymphatic duct. The duct normally startedform within 12 hours from the time of transplantation, whicmay not allow sufficient time for the inner ear to be respecifito form a duct in the new dorsal position. Many morphogenemutants, such as the FGF3 knock out (Mansour et al., 1993)the kreisler (Ruben, 1973) mice, fail to form endolymphatiducts and proper inner ear chambers. It is not clear whetherformation of an endolymphatic duct is a prerequisite fsubsequent morphogenetic events since ablation endolymphatic duct experiments yielded conflicting resu(Hendricks and Toerien, 1973; Van De Water, 1977). In the Drotated ears, however, the presence of an endolymphatic was not always associated with more highly differentiated innears.

Specification of sensory organs may be independent ofnon-sensory organsOur results showed that in A/P rotated ears, the sensory orgwere patterned according to the A/P axis before transplantatwhile most of the non-sensory structures such as the supeand posterior canals, the distal portion of the cochlear duct the endolymphatic duct, were able to be patterned accordinnew axial information from the host. However, the insertion the lateral canal on the common crus and the proximal portof the cochlear duct followed the axis of the donor, suggestthat certain patterns of nonsensory structures were also fixethe time of transplantation. It is interesting that while the latecanal is the last canal to be formed among the three, soaspects of its structure are already fixed. It is also intriguing twithin the single structure of the cochlear duct, there wdifferential specification between its proximal and distportions. The orientation of the proximal portion of the cochleduct may be dictated by the location of the presumptive baspapilla, which is already fixed at the time of transplant, whthe remaining shape of the duct may be governed by surrounding tissue as the duct extends. It is not clear whatsignaling factors are that confer axial patterning in the chinner ear, but tissues such as the hindbrain, mesenchyme

rersd.

tialearftsarths.

iont atwees.tagexialan

cupof

theionacedto tohedtic

andc theorof

lts/Vducter

ansion,rior

andg toofioningd atralme

hatasalarilarilethe theick and

the notochord are likely sources of such factors (for reviewsee Fritzsch et. al., 1997; Noden and Van De Water, 1992).

Previous studies have shown that sensory hair cdifferentiation can proceed in the absence of normmorphogenesis, suggesting an independent pathway of hairdevelopment (Swanson et al., 1990; Whitfield et al., 199Here, we provide additional evidence that the pathway(s) forming sensory organs and non-sensory structures mayunlinked, based on their asynchronous axial specificatiHowever, studies from zebrafish mutants such as dog-eared,van gogh and colourless, displayed abnormalities in bothcristae and semicircular canals, suggesting that in fact development of these two structures may be linked (Whitfieet al., 1996). The inductive events specifying sensory or nsensory structures may be linked initially and diverge laterdevelopment. An equally likely scenario is that the inductievents are molecularly linked for both sensory and non-sensstructures, but there is a temporal difference in the abilitythe two type of tissues to respond to axial information, thresulting in different axial patterns for these structures. Aone or a combination of the proposed scenarios is plausiblthe moment.

Sensory organ formation requires multiple stepsWe raised the question earlier as to whether different typessensory organs in the chick inner ear are formed by a sininductive event or by multiple steps of gradual specificationposition and type. From our results, we conclude that differetypes of sensory organs are formed in response to A/P, Mand D/V positional information. At the time of transplantatio(E2.5), the A/P positions of the sensory organs were alrefixed and the types of organs formed in the host were affected. Yet, the types of sensory organs formed along the Daxis remained plastic, at least until E3.5. If each sensory oris formed by a single inductive event, one would not expethis lag-time between the two axial specifications, suggestthat formation of different types of sensory organs requirmultiple steps. Furthermore, it is not clear whether the variopatterns of sensory organs in D/V rotated ears are mainlresult of simple conversion of sensory organ type fropreviously specified sensory regions or of new sensory tisinduction in areas not normally destined to become sensMost likely, the ectopic cristae obtained in some of the Drotated ears are a result of de novo inductions.

The early onset of BMP4 expression in sensory organs haled to the suggestion that BMP4may specify sensory organformation (Wu and Oh, 1996). Our results indicate that BMP4expression is not well established at a time when the A/P aof the sensory organs appears to be fixed, suggesting BMP4 is not involved in the specification of sensory organAn alternative interpretation of these results, however, is tBMP4 is required for the specification of the sensory organIn this case, at the time of transplantation, the sensory orgthemselves are not specified, although their future A/P axisalready determined by positional information. Our resucannot distinguish between these two possibilities. Howevwe note that the expression patterns of BMP4 in rotated earsare consistent with a role for BMP4 in the specification ofsensory organs. Similarly, Otx1and SOHo-1gene expressionsare always found associated with the formation of the cochland semicircular canals, respectively, suggesting that th

20

ick

stf

ve

e

al

)

ates

ing

el

t’: A

r


gene products may play a role in specification of these inear structures. Future experiments will focus on identifyinmolecular pathways in sensory organ formation as well genes specifying axial information for the inner ear.

The authors thank Dr Guy Richardson for the anti-HCA antibodieDrs James Battey and Susan Sullivan for critical reading of manuscript and Ms Mirene Boerner for editing.

REFERENCES

Acampora, D., Mazan, S., Avvantaggiato, V., Barone, P., Tuorto, F.,Lallemand, Y., Brulet, P. and Simeone, A.(1996). Epilepsy and brainabnormalities in mice lacking the Otx1gene. Nature Genet.14, 218-222.

Barald, K. F., Lindberg, K. H., Lewis, J. E., Victor, J. C., van Nguyen, D.,Ng, A., Germiller, J., Siram, S. and Kudelko, K.(1997). Plasticity of geneexpression in the developing chick inner ear. 20th Annual Meeting of TheAssociation for Research in Otolaryngology. Abstract #16.

Bartolami, S., Goodyear, R. and Richardson, G.(1991). Appearance anddistribution of the 275 kD hair-cell antigen during development of the aviinner ear. J. Comp. Neurol.314, 777-788.

Bissonnette, J. P. and Fekete, D. M.(1996). Standard atlas of the grossanatomy of the developing inner ear of the chicken. J. Comp. Neurol.174, 1-11.

Deitcher, D. L., Fekete, D. M. and Cepko, C. L.(1994). Asymmetricexpression of a novel homeobox gne in vertebrate sensory organsJ.Neurosci.14, 486-498.

Fekete, D. M. (1996). Cell fate specification in the inner ear. Curr. Opin.Neurobiol.6, 533-541.

Fritzsch, B., Barald, K. and Lomax, M. (1997). Early embryology of thevertebrate ear. In Springer Handbook of Auditory Research, Vol XII,Development of the Auditory System (ed. E. Rubel, A. Popper and R. Fay)Springer, New York, in press.

Hall, E. K. (1939). On the duration of the polarization process in the eprimordium of embryos of Amblystoma punctatum. J. Exp. Zool.82, 173-192.

Hamburger, V. and Hamilton H. (1951). A series of normal stages in thedevelopment of the chick embryo. J. Morph.88, 49-52.

Harrison, R. G. (1936). Relations of symmetry in the developing ear oAmblystoma punctatum. Proc. Nat. Acad. Sci. USA22, 238-247.

Hendricks, D. M. and Toerien, M. J. (1973). Experimental endolymphatichydrops. S. Afr. Med. J.47, 2294-2298.

Kiernan, A. E., Nunes, F., Wu, D. K. and Fekete, D. M.(1997). Uniqueexpression domains of two members of a novel homeobox containing gfamily suggest a role in semicircular canal formation during inner edevelopment. Dev. Biol., in press.

Knowlton, V. Y. (1967). Correlation of the development of membranous a

nergas

s,the

an

.

.

ar

f

enear

nd

bony labyrinths, acoustics ganglia, nerves, and brain centers of the chembryos. J. Morphol.121, 179-208.

Lindberg, K. H., Lomax, M. I., Hegeman, A. D. and Barald, K. F. (1995).Pax-2: An early marker of otic placode induction that defines otocydevelopment in the chick. Eighteenth Annual Midwinter Research Meeting othe Association for Research in Otolaryngology. Abstract # 418.

Mansour, S. L., Goddard, J. M. and Capecchi, M. R.(1993). Micehomozygous for a targeted disruption of the proto-oncogene int-2 hadevelopmental defects in the tail and inner ear. Development117, 13-28.

Martin, P. and Swanson, G. J.(1993). Descriptive and experimental analysisof the epithelial remodellings that control semicircular canal formation in thdeveloping mouse inner ear. Dev. Biol.159, 549-558.

Noden, D. M. and Van De Water, T. R.(1992) Genetic analyses of mammalianear development. Trends Neurosci.15, 235-237.

Nornes, H. O., Dressler, G. R., Knapik, E. W., Deutsch, U. and Gruss, P.(1990). Spatially and temporally restricted expression of Pax2during murineneurogenesis. Development109, 797-809.

Roach, F. C.(1945). Differentiation of the central nervous system after axireversals of the medullary plate of Amblystoma. J. Exp Zool.99, 53-77.

Ruben R. J. (1973). Development and cell kinetics of the Kreisler (kr/krmouse. Laryngoscope83, 1440-68.

Simeone, A., Acampora, D., Mallamaci, A., Stornaiuolo, A., RosariaD’Apice, M., Nigro, V. and Boncinelli, E.(1993). A vertebrate gene relatedto orthodenticle cntains a homeodomain of the bicoid class and demarcanterior neuroectodem in the gastrulating mouse embryo. EMBO J.12, 2735-2747.

Simon, H., Hornbruch, A. and Lumsden, A.(1995). Independent assignmentof antero-posterior and dorso-ventral positional values in the developchick hindbrain. Curr. Biol.5, 205-214.

Stoker, A. W. and Bissell, M. J.(1987). Quantitative immunocytochemicalassay for infectious avian retroviruses. J. Gen. Virol.1987, 2481-2485.

Suzuki, H. R., Padanilam, B. J., Vitale, E., Ramirez, F. and Solursh, M.(1991). Repeating developmental expression of G-Hox 7, a novHomeobox-containing gene in the chicken. Dev. Biol.148, 375-388.

Swanson, G. J., Howard, M. and Lewis, J. (1990). Epithelial autonomy in thedevelopment of the inner ear of a bird embryo. Dev. Biol.137, 243-257.

Torres, M., Gomez-Pardo, E. and Gruss, P.(1996). Pax2contributes to innerear patterning and optic nerve trajectory. Development122, 3381-3391.

Van De Water, T.(1977). The effect of the removal of the endolymphatic ducand sac anlage upon organogenesis of the mammalian inner ear ‘in vitropreliminary report. Arch. Oto-Rhino-Laryng.217, 297-311.

Whitfield, T. T., Granato, M., van Eeden, F. J. M., Schach, U., Brand, M.,Furutani-Seiki, M., Haffter, P., Hammerschmidt, M., Heisenberg, C.,Jiang, Y., Kane, D. A., Kelsh, R. N., Mullins, M. C., Odenthal, J. andNusslein-Volhard, C. (1996). Mutations affecting development of thezebrafish inner ear and lateral line. Development123, 241-254.

Wu, D. K. and Oh, S. H.(1996). Sensory organ generation in the chick inneear. J. Neurosci.16, 6454-6462.

axial speciﬁcation for sensory organs versus non-sensory … · sensory organs are largely...

Documents