Specific expression of ezrin, a cytoskeletal-membrane linker protein, in a subset of chick retinotectal and sensory projections

European Journal of Neuroscience, Vol. 11, pp. 545–558, 1999 © European Neuroscience Association

Specific expression of ezrin, a cytoskeletal-membranelinker protein, in a subset of chick retinotectal andsensory projections

Masakazu Takahashi,1 Masahito Yamagata1,2 and Masaharu Noda1,2

1Division of Molecular Neurobiology, National Institute for Basic Biology, and 2Department of Molecular Biomechanics, GraduateUniversity for Advanced Studies, Okazaki 444-8585, Japan

Keywords: dorsal root ganglion, ezrin, lamina-specific projection, retinal ganglion cell, retinotectal system, sensory neuron

Abstract

Lamina-specific neuronal connections are a fundamental feature in many parts of the vertebrate central nervous system. In thechick, the optic tectum is the primary visual centre, and it has a multilaminated structure consisting of 15 laminae, of which onlythree or four receive retinal projections. Each of the retinorecipient laminae establishes synaptic connections selectively from oneof a few subsets of retinal ganglion cells (RGCs). We have generated a series of monoclonal antibodies that appear to stain onlyone of the retinorecipient laminae. One of these, TB4, stained lamina F which receives inputs from a subpopulation of µ 10–20%of RGCs which express the presynaptic acetylcholine receptor β2-subunit. TB4 recognized a single 79-kDa protein on immunoblotting.cDNA cloning and immunochemical analysis revealed that the TB4 antigen molecule was ezrin, a cytoskeletal-membrane linkermolecule belonging to the ezrin-radixin-moesin family. Unilateral enucleation of the eye, both prior to and after the establishmentof retinotectal projections, attenuated the lamina-selective staining with TB4 in the contralateral tectum, suggesting that ezrin isanterogradely transported from RGCs to lamina F. Ezrin was thus expressed in a subset of RGCs that project to lamina F. Similarsubset-selective expression and resultant lamina-selective distribution of ezrin were also observed in the lamina-specific centralprojections from the dorsal root ganglia. The staining pattern with TB4 in the dorsal root ganglia and spinal cord indicated that highexpression of ezrin was restricted in cutaneous sensory neurons, but not in muscle sensory neurons. Since ezrin modulates cellmorphology and cell adhesion profiles by linking membrane proteins with the cytoskeleton, it was suggested that ezrin isinvolved in the formation and/or maintenance of lamina-specific connections for neuronal subpopulations in the visual andsomatosensory systems.

Introduction

Laminar arrangements of neurons and their connections are prominentfeatures in many regions of the vertebrate central nervous system,including the telencephalon, diencephalon, mesencephalon, metence-phalon and spinal cord. Each lamina contains a specific type ofneuron which receives inputs from a particular class of afferent,thereby constructing functional neural networks.

The optic tectum is the primary visual centre in birds, and it hasa multilaminated structure consisting of an alternating series of 15cell-soma-rich or neuropil laminae (Hunt & Brecha, 1984). As in thecerebral cortex in mammals, extratectal inputs terminate in discretelaminae in the tectum, and tectal neurons in different laminae projectto distinct extratectal targets (Brecha, 1978; Hunt & Brecha, 1984).The retinal afferents are the best studied of the inputs (Mey & Thanos,1992). During development of the retinotectal projection, retinalganglion cell (RGC) axons extend along the surface of the tectum,then invade the deep laminae, and finally develop arbors at theirterminals in only three to four of the 15 tectal laminae, presumablyin response to lamina-specific guidance cues (Yamagata & Sanes,

Correspondence: Dr Masaharu Noda, Division of Molecular Neurobiology,National Institute for Basic Biology, Myodaiji-cho, Okazaki 444–8585, Japan.E-mail: [email protected]

Received 17 July 1998, revised 8 September 1998, accepted 10 September 1998

1995a,b). The majority of the RGC axons project to the neuropil inthe lamina named D, whereas minor populations of axons connect todistinct upper or deeper laminae (Angaut & Repe´rant, 1976; Achesonet al., 1980). These minor RGC subsets are distinguishable by theneurochemical phenotype; a subset expressing substance P projectsto lamina B, which is positive for the substance P-receptor, while adifferent subset expressing a presynaptic acetylcholine receptor sub-unit projects to the cholinergic lamina F (Ehrlichet al., 1987; Brittoet al., 1992; Yamagata & Sanes, 1995b). These complementaryneurochemical phenotypes were determined intrinsically long beforethe specific neural connections occurred (Yamagata & Sanes, 1995b),intimating the presence of complementary cues for specific projectionsbetween RGC subsets and their target laminae.

Lamina-selective projection of neuronal subsets has also beenobserved in the primary somatosensory system, where differentpopulations of sensory neurons make distinct central projectionswithin the spinal cord (Brown, 1981). Most sensory afferents frommuscle grow ventrally to the ventral cord through the dorsal horn,whereas most cutaneous sensory terminals terminate within the dorsalhorn. Each subset of sensory neurons populates a different part ofthe dorsal root ganglion (DRG), and they show distinct neurochemicalprofiles (Snider & Wright, 1996). The lamina-specific projection ofthe DRG afferents was recently suggested to be guided by localpositional cues such as the collapsin/semaphorin family by confining

546 M. Takahashiet al.

each afferent to an appropriate region within the spinal cord (Mes-sersmith et al., 1995; Wright et al., 1995; Pu¨schel et al., 1996;Redmondet al., 1997; Sharma & Frank, 1998)

To understand how the lamina-specific projections in the tectumarize, we have searched for the lamina-specific cues. In the presentstudy, we generated and screened hybridomas to obtain monoclonalantibodies (mAbs) that stained only one of the retinorecipient laminaein the optic tectum. We obtained one such mAb for each of the threeretinorecipient laminae. Among these, TB4 stained lamina F, becausea subset of RGCs projecting to lamina F was positive. TB4 alsostained a subset of sensory neurons in the developing DRG. TB4recognized ezrin, a member of the cytoskeletal-membrane linkerprotein (Bretscheret al., 1997; Tsukitaet al., 1997). Our findingssuggested that ezrin may participate in a specific adhesion structurecrucial for subset-specific and, therefore, lamina-specific neuronalconnections in the visual and somatosensory systems.

Materials and methods

Animals

White Leghorn eggs were obtained from a local supplier, andincubated at 37.5 °C. Embryos were staged according to Hamburger& Hamilton (1951). Enucleation of the eye primordium at embryonicday 3 (E3) was carried out as described previously (Kelly & Cowan,1972) using an electrocautery device (Model 285101, Abco Inc.,Milwaukee, WI, USA). Unilateral ablation of the whole retinaof hatched chicks was performed with a heated cauterizer underpentobarbital anaesthesia (15–20 mg/kg, i.p.). Experiments wereperformed according to the guidelines of the National Institute forBasic Biology, Japan.

Production and screening of mAbs

For generation of mAbs, Balb/c mice (Japan SLC, Hamamatsu, Japan)were injected intraperitoneally with tectal tissues prepared as follows.Dissected tecta from E15 chick embryos were dissociated in 1 mLof phosphate-buffered saline (PBS) by passing them several timesthrough an 18-gauge needle. The tectal preparation from one chickembryo was injected into each mouse. Two and 4 weeks after theinitial immunization, mice received intraperitoneal booster injectionsof the same tectal preparation. Three days after the final injection,splenocytes from the immunized mice were fused with PAI myelomacells (developed by Dr T. Stahelin; gift from Dr T. Arai), usingpolyethyleneglycol by the standard method (Harlow & Lane, 1988).Cells were selected in a medium containing hypoxanthine, aminopter-ine and thymidine with MRC5 human fibroblasts as feeder cells(Harlow & Lane, 1988). Supernatants from a total ofµ 10 000 wells(each well contained two or three hybridoma clones) were screenedby staining cryosections of the E15 optic tectum. Hybridomas thatstained the retinorecipient laminae were first selected and cloned bythe limiting dilution method.

Anti-chick ezrin antiserum

Anti-chick ezrin antiserum was raised in rabbits by immunizationwith a multiple antigenic peptide (MAP) (Tam, 1988) ofN-acetylatedPVNYHVHDNL [corresponding to amino acid (AA) residues 479–488 of chick ezrin], which was synthesized on an octabranchinglysine core by F-moc chemistry (Sawady Technology, Tokyo, Japan).The raised antiserum, E388, was affinity-purified using Amino Cellul-ofine® (Seikagaku Kogyo, Tokyo, Japan) on which an unmodifiedpeptide, PVNYHVHDNL, was conjugated by the carbodiimide coup-ling method. Specific antibodies bound to the affinity column were

© 1999 European Neuroscience Association,European Journal of Neuroscience, 11, 545–558

eluted with 0.1M glycine-HCl buffer (pH 2.5), and neutralized with1 M tris aminomethane. Approximately 2 mL of affinity-purified E388was obtained from 5 mL of the antiserum and used at dilutions of1 : 10 to 1 : 20 for tissue staining.

Other antibodies and immunochemical reagents

Mouse mAb 13H9 (Goslinet al., 1989; Winckleret al., 1994) thatrecognizes all chick ezrin-radixin-moesin (ERM) proteins, i.e. ezrin,radixin and moesin, was a generous gift from Dr F. Solomon(Massachusetts Institute of Technology, Cambridge, MA, USA).Specific rat mAbs to mouse ezrin (M11; Takeuchiet al., 1994b),mouse radixin (R2-1; Yonemuraet al., 1998) and mouse moesin(M22; Takeuchiet al., 1994b), and a mouse mAb that recognizes allthree of the ERM proteins, with a bias toward moesin (CR22; Satoet al., 1992) were generously provided by Dr S. Tsukita (KyotoUniversity, Japan). Rabbit antiserum to choline acetyltransferase(ChAT) (Johnson & Epstein, 1986) was obtained from Dr J. R. Sanes(Washington University School of Medicine, St Louis, MO, USA).A rat mAb to the neuronal nicotinic acetylcholine receptor (AChR)β2-subunit designated as 270 (Whiting & Lindstrom, 1987), and23.4-5, a mouse mAb to chick axonin-1 (Redieset al., 1997),were obtained from the Developmental Studies Hybridoma Bank,University of Iowa (Iowa City, IA, USA). Rabbit antibody toneurofilament 200 was purchased from Sigma.

Fluorochrome-conjugated species-specific secondary antibodieswere used for immunohistochemistry. Fluorescein-conjugated anti-mouse, Texas-Red-conjugated antirabbit and antirat, and biotinylatedantimouse and antirabbit Ig antibodies, and Cy2- and Cy3-conjugatedstreptavidin were purchased from Amersham Japan (Tokyo, Japan).Fluorescein-conjugated antirat IgG and Cy3-conjuated antimouse,antirabbit and antirat IgG antibodies were from Jackson Immuno-Research Laboratories, Inc. (West Grove, PA, USA). Fluorescein-conjugated avidin DCS was from Vector Laboratories (Burlingame,CA, USA). Goat antiglutathione S-transferase (GST) antibody waspurchased from Pharmacia Biotech (Uppsala, Sweden). Horseradishperoxidase (HRP)-conjugated antimouse and antirabbit antibodieswere from Amersham Japan, and HRP-conjugated antigoat antibodywas from ICN Pharmaceuticals Inc.-Cappel Products (Durham, NC,USA).

Immunohistochemistry

Brains, eyes, trunks and other tissues of chicks and chick embryoswere fixed with 3.5% paraformaldehyde in Hanks’ solution or 4%paraformaldehyde in PBS overnight at 4 °C. After cryoprotection byimmersion in increasingly concentrated solutions of sucrose (7–18%)in PBS at 4 °C, the tissues were embedded in Tissue-Tek® OCTcompound (Miles Inc., Elkhart, IN, USA), frozen at –80 °C, and cutinto sections 20µM thick. The cryosections of the anterior tectumand retina were permeabilized with 0.1% Triton X-100 in PBS for15 min before immunostaining with mAbs 270 and 13H9. For otherprimary antibodies, cryosections were permeabilized with Triton X-100 as described above, or with methanol for 10 min previouly chilledat –20 °C. The permeabilized sections were incubated with a blockingbuffer (5% skimmed milk in PBS) for 1 h at room temperature.The sections were then incubated with primary antibody solutionsovernight at 4 °C, followed by fluorochrome-conjugated or biotinyl-ated secondary antibodies for 2 h at room temperature, and incubationwith fluorochrome-conjugated avidin DCS/streptavidin for 30 min atroom temperature when the biotin-avidin system was employed.During the final incubation, nuclei were counterstained with 49, 6-diamidino-2-phenylindole (DAPI; Molecular Probes, Eugene, OR,

Ezrin expression in a subset of neural projections 547

USA). The antibodies were diluted with the blocking buffer asrequired. For double staining, the absence of cross-reactivity ofspecies-specific secondary antibodies was confirmed by staining witha mixture of two secondary antibodies after incubation with one ofthe primary antibodies.

Nomenclature of tectal laminae

The nomenclature of developing and mature tectal laminae wasdescribed in Yamagataet al. (1995). Sublaminae of the embryonicstratum griseum et fibrosum superficiale(SGFS) are shown in uppercase letters to distinguish them from the mature laminae of posthatchedchicks shown in lower case.

cDNA cloning

Molecular cloning was carried out by standard methods (Sambrooket al., 1989). A λgt11 cDNA library of E10 chick brain (23 105

clones) was screened with the culture supernatant from TB4hybridoma, and two overlapping but independent clones, TB4-1 andTB4-2, were isolated. The EcoRI-released inserts of the clones weresubcloned in pBluescript II KS (1) (Stratagene, La Jolla, CA, USA).Sequencing of both strands of the cDNA was performed with anALF DNA sequencer (Pharmacia Biotech, Uppsala, Sweden).

For isolation of chick radixin cDNA clones, a Uni-ZAP XR cDNAlibrary of E5 chick whole embryos (Stratagene) (6.83 105 clones)was screened under low stringency conditions (washing temperaturewas 45 °C) using TB4-1 cDNA as a probe. Ten independent clones(R1-R10) encoding chick radixin were identified by restriction map-ping and nucleotide sequencing.

Production of GST-fusion proteins

A cDNA encoding the full-length chick ezrin (AAs 1–585) waspreprepared by ligating cDNA fragments from TB4-1 and TB4-2 atthe AccI site in the overlapping region. The full-length cDNA ofchick ezrin was subcloned into the pGEX-4T-2 expression vector(Pharmacia Biotech) to produce the GST-ezrin fusion protein. Forpreparation of cDNA constructs to produce GST-COOH-radixin andGST-NH2-radixin, the full insert of the R4 clone encoding AAs 247–583 and a fragment of the R10 clone encoding AAs 49–438,respectively, were cloned into the pGEX-4T-1 expression vector. TheGST-fusion proteins were generated with these constructs inE. coli(strain DH5α) according to the manufacturer’s protocol (PharmaciaBiotech).

Immunoblotting

Immunoblotting was performed as described previously (Towbinet al., 1979). Briefly, tissues or bacterial pellets were lysed in standardsample buffer containing, additionally, a mixture of protease inhibitors(0.25 mM phenylmethanesulphonyl fluoride, 10µg/mL leupeptin and10 µM pepstatin A), 1 mM EDTA and 5% mercaptoethanol (Laemmli,1970). Sodium dodecyl sulphate-polyacrylamide gel electrophoresis(SDS-PAGE) was carried out using 7.5, 8.7 or 15% gels. The proteinswere electrophoretically transferred onto polyvinylidene difluoridemembranes, and blocked with 5% skimmed milk in 150 mM NaCl,10 mM Tris-HCl (pH 7.5) containing 0.1% Tween 20. The membraneswere then incubated with primary antibodies overnight at 4 °C, andantigens were detected with HRP-conjugated secondary antibodiesusing ECL Western blotting detection reagents (Amersham Japan) ordiaminobenzidine as a substrate.


Results

Isolation of mAbs that stain one of the retinorecipient laminae

We screened. 20 000 newly generated hybridoma clones to obtainantibodies that stain specific laminae in the tectum. Among these,TB5, TB2 and TB4 labelled laminae B, D and F, respectively (Fig. 1).In contrast to the previously described markers that are widelydistributed in the retinorecipient laminae (i.e. SC1/JC7, VVA-B4lectin-binding carbohydrates,N-cadherin and neuropilin-1; Yamagataet al., 1995), the staining patterns of TB5, TB2, TB4 were differentfrom these because they stained only one of the retinorecipientlaminae. In this report, we focus on TB4, an antibody that stainedlamina F. The lamina-selective staining with TB4 was similarlyobserved when tissue cryosections were treated with absolute meth-anol, absolute acetone or 0.1% Triton X-100/PBS. Characterizationsof the other antibodies will be reported elsewhere.

Identification of TB4 antigen

We screened a chick brainλgt11 cDNA library with TB4 to identifythe antigen. Two clones (λTB4-1 andλTB4-2) thus isolated weresubjected to nucleotide sequence analysis. These two clones over-lapped each other by 462 bp and the encoded protein (585 AA;calculated molecular mass 69 366, including the initiative methionine)was highly homologous to the ERM family of proteins (Fig. 2),which are known to be cytoskeletal-membrane linker proteins. Amongthe cognate molecules characterized in mammals, our chick moleculeshowed the highest sequence similarity to mouse and human ezrin assummarized in the cladogram in Fig. 2C. The band-4.1-like domainshowed highest identity and the hydrophilic domain showed lowestidentity between our molecule and mouse ezrin (Fig. 2A). Theoverlapping sequences between the two independent TB4-immunore-active clones suggested that the TB4 epitope resides in this 153-AAregion covering part of theα-helical and hydrophilic domains ofezrin (Fig. 2A and B).

As ERM proteins are homologous to each other, some antisera andmAbs in previous studies cross-reacted with other members of theERM family (Satoet al., 1992; Francket al., 1993; Winckleret al.,1994). Furthermore, it is difficult to distinguish chick ezrin fromchick radixin on SDS-PAGE because both proteins migrate to similarpositions (Winckleret al., 1994). To confirm that the TB4 antigenwas chick ezrin, we raised an antichick ezrin antiserum, E388, byimmunizing rabbits with a MAP of the ezrin-specific sequencePVNYHVHDNL (AAs 479–488 of chick ezrin; Fig. 2B). Thissequence was chosen because it is the most ezrin-specific regionamong ezrin, radixin, moesin and other related proteins such as merlin(also called schwannomin and neurofibromatosis-2 gene product:Trofatteret al., 1993).

The affinity-purified antiserum thus obtained, E388, recognized theMAP used as the immunogen, but TB4 did not bind to it (Fig. 3A).This suggested that the epitope for TB4 in the overlapping regionbetween clonesλTB4-1 and λTB4-2 differs from that for E388(Fig. 3B).

In the whole retinal extract of E18 embryos, E388 recognized asingle band, the size of which was identical to that of the mAb-TB4-reactive band (Fig. 3B). Moreover, TB4 (epitope lying between AAs339 and 491) and E388 (epitope consisting of AAs 479–488)recognized the fusion protein of full-length chick ezrin (AAs 1–585)and GST, but did not cross-react with radixin-GST fusion proteinscontaining either an NH2-terminal (AAs 49–438) or a COOH-terminalportion (AAs 247–583) of chick radixin (Fig. 3C). These resultsshowed that TB4 and E388 recognized the same antigen, ezrin, butnot radixin. Thus, we concluded that the TB4 antigen is chick ezrin.


FIG. 1. Monoclonal antibodies that stain only one of the retinorecipient laminae of the optic tectum. Cryosections of the tectum were stained with mAbs TB5(A), TB2 (B) and TB4 (C), which labelled laminae B, D and F, respectively. The tecta used were from E14 (A and B) and E18 embryos (C). The TB5-positivepunctate materials were associated with a subpopulation of cells in lamina B, whereas TB2 stained fine fibrous structures in lamina D. Similar, although weaker,staining patterns with mAbs TB5 and TB2 were seen at E18 (data not shown). Nomenclature of the developing tectal laminae was described in Yamagataet al.(1995). SO,stratum opticum. The pial surface (P) is marked with dashes. Scale bar, 25µm.

Immunofluorescence staining of sections revealed that mAb TB4reacted not only with lamina F of the optic tectum but also widelywith epithelial tissues such as skin and intestine (data not shown), asreported for ezrin. Immunoblotting analysis of tissue extracts wasconsistent with the results of the histological study, showing a single79-kDa band of ezrin in the brain, retina, liver and intestine (Fig. 3D).These observations also supported our conclusion that TB4 recog-nizes ezrin.

Antiserum E388 reproduced the selective staining of lamina F inthe tectum as TB4 (Fig. 3E and F). Thus, the two distinct antibodies,both of which specifically recognized ezrin, showed F-lamina-specificstaining, suggesting that ezrin is concentrated in a lamina-selectivemanner. On the other hand, mAb 13H9, which recognized all threeof the ERM proteins, stained broad regions with stronger staining inthe upper laminae of the SGFS (Fig. 3G). The epitope of 13H9resides in the sequence IAQDLEMYGINYF (AAs 194–206 in chickezrin) which is common to radixin, moesin and merlin except forone AA; the isoleucine residue at position 203 is replaced with valine(Winckler et al., 1994). Therefore, Fig. 3(G) showed cumulativestaining for all the ERM proteins.

Lamina-selective accumulation of ezrin

The major constituents of lamina F of the tectum are the terminalarbours of the cholinoceptive RGC subset and the dendrites of thecholinergic r2 neurons, the somata of which are located in lamina I(Reiner & Karten, 1982; Hunt & Brecha, 1984; Yamagata & Sanes,1995b; also see Fig. 8A), although other minor inputs also contributeto the neuropil (Brecha & Karten, 1981; Hunt & Brecha, 1984). Theterminals of the cholinoceptive RGC subset were stained positivelywith mAb 270, an antibody which recognizes theβ2 subunit of the


neuronal nicotinic AChR (Keyseret al., 1988, 1993; Brittoet al.,1992; Hamassaki-Brittoet al., 1994), and the cholinergic r2 cells inthe tectum were stained with antibodies to ChAT (Sorensonet al.,1989; Bagnoliet al., 1992; Medina & Reiner, 1994). Thus, theseantibodies also selectively stain the neuropil of lamina F similarlyto TB4.

A critical question is which cellular components or events areresponsible for the accumulation of the ezrin in specific lamina. Wenext examined when and how ezrin begins to accumulate in thelamina during development of the optic tectum in comparison withthe appearance of theβ2 subunit (Fig. 4). At early stages (E5), mAbTB4 faintly stained the mesencephalic neuroepithelium, and thestaining was associated with cell–cell contacts as observed in otherepithelial tissues (not shown). As the neuroepithelium produces alaminar structure including plexiform layers, the signals at cell-cellcontacts became faint. At E11, just before the retinal axons formsynapses in their final target laminae, overall staining of the tectumwith TB4 was not evident (Fig. 4A). At E14, the stage at whichretinal terminals begin to arborize and form synapses in the finalretinorecipient laminae, the antigen was accumulated at the base oflamina D, which corresponds to the prospective lamina F (Fig. 4C).The signal emerged cocomitantly with that for the mAb-270-positiveAChR subunit (Fig. 4B,D and F). From E15 onwards, the intensityof the staining increased, and at day 2 posthatching (P2), stainingwas clearly seen in lamina F (Fig. 4E). The clear staining patternwas retained at P8, the latest stage examined. Thus, accumulation ofezrin antigen was synchronous with the formation of lamina F itself,and with the arborization of theβ2-subunit-positive retinal inputs inthe retinorecipient lamina.

To test whether or not the presence of retinal inputs is a prerequisite


FIG. 2. Molecular cloning of the TB4 antigen. (A) Schematic representation of two independent TB4 clones (λTB4-1 andλTB4-2). The coding region anddomain structure of ezrin are also shown, underneath. The boundary of each domain is indicated by the AA residue numbers. For each domain, the percentageof AA identity with mouse ezrin is shown in parentheses. A,AccI; B, BglII; E, EcoRV; P,PstI; S, SmaI. (B) Nucleotide sequence of the cDNA and deduced AAsequence. The overlapping region betweenλTB4-1 andλTB4-2 is underlined (single underline). Two nucleotide differences were found between the two cloneswithout alteration of the encoded AAs; A at nucleotide residue 1043 and T at 1430 inλTB4-1 were substituted with G inλTB4-2. The proline cluster betweenthe α-helical domain and hydrophilic domain is boxed. The AA sequence PVNYHVHDNL (AAs 479–488) used for synthesis of the MAP for raising E388antiserum and the epitope sequence IAQDLEMYGINYF (AAs 194–206) of mAb 13H9 are double-underlined. (C) A phylogenic tree showing the evolutionaryrelationship between TB4 antigen (chick ezrin) and mammalian members of the ERM family. The dendrogram, and AA identity with chick ezrin were calculatedby the UWGCG program DISTANCE and BESTFIT, respectively.



FIG. 3. TB4 antigen is chick ezrin. (A) Anti-chick ezrin antibody E388 recognized the MAP antigen but mAb TB4 did not. The MAP ofN-acetylatedPVNYHVHDNL (0.1 µg), the immunogen peptide used to raise E388, was electrophoresed and immunoblotted with affinity-purified E388 (lane 1, 1 : 1000dilution) or TB4 (lane 2, 1 : 100 dilution). The result indicated that E388 and TB4 recognized different epitopes in chick ezrin. (B) Affinity-purified E388 (lane1, 1 : 30 dilution) and TB4 (lane 2, 1 : 100 dilution) detected a single 79 kDa band on immunoblots of a chick E18 retinal lysate. About 10µg of total proteinwas loaded in each lane. (C) Immunoblots of GST-fusion proteins of full-length ezrin (AAs 1–585; lanes 1, 4, 7), NH2-radixin (AAs 49–438; lanes 2, 5, 8),and COOH-radixin (AAs 247–583; lanes 3, 6, 9). Each lane was loaded withµ 2 µg of the total protein from bacterial lysates. TB4 (lanes 1–3) and E388(lanes 4–6) recognized GST-ezrin (lanes 1, 4), but not GST-NH2- (lanes 2, 5) or GST-COOH-radixin (lanes 3, 6). Note that the epitopes recognized by TB4and E388 reside in AAs 339–491 and 479–488 of ezrin, respectively (see Fig. 2B). All fusion proteins were detected by anti-GST antibody (lanes 7–9),indicating that the amounts of fusion proteins loaded were sufficient for immunoblotting analysis. Multiple protein bands which migrated faster than the fullsize molecule (each indicated by arrows and letters: E, GST-ezrin; NR, GST-NH2-radixin; CR, GST-COOH-radixin) seemed to be degradation products.Polyacrylamide concentrations in separation gels were 8.7% in (A) and (C), and 15% in (B). (D) Tissue distribution of TB4 antigen determined by immunoblotting.The extracts (5µg of protein) of the retina (lane 1), optic tectum (lane 2), cerebellum (lane 3), forebrain (lane 4), liver (lane 5) and intestine (lane 6) of E14embryos were subjected to SDS-PAGE using 7.5% separation gels, and immunoblotted with mAb TB4. All of the samples yielded a single protein band ofezrin with an apparent molecular weight of 79 kDa. (E) The E18 optic tectum stained with affinity-purified E388. Lamina F was specifically stained in theretinorecipient laminae, together with thestratum opticum(SO) through which the retinal axons extend. Due to the low sensitivity of the antibody, TB4 wasmainly used for other tissue staining (see Figs 1C and 3B). (F) Counterstaining of the same section as shown in (E) with DAPI. (G) The E18 optic tectumstained with mAb 13H9 that recognizes all three ERM proteins. Scale bar, 100µm.

for accumulation of the antigen in lamina F, we analysed accumulationin the tecta of embryos in which the retina had been removed beforethe RGC axons reached the tecta, and in hatched chicks in whichRGC axons had been degenerated for 1 week by retinal ablation. Inthe former, there was no accumulation of ezrin in any retinorecipientlaminae in the tectum contralateral to the enucleated side (Fig. 5Aand B), suggesting that innervation by retinal terminals is necessaryfor accumulation of ezrin in the developing tectum. In contrast, TB5and TB2 antigens as well as VVA-B4-lectin-binding carbohydratesand SC1/JC7 remained in the contralateral tectum (data not shown).In the latter experiment, when the eyes of newly hatched chicks wereunilaterally ablated, both ezrin and the AChRβ2-subunit disappearedfrom lamina f of the contralateral tectum (Fig. 5C–E). In contrast,ChAT, a neurochemical marker for the putative postsynaptic r2 cells,was still seen in lamina f (Fig. 5F and G), suggesting that thedendrites of the r2 neurons did not retract from lamina f, even in theabsence of retinal inputs. These observations suggested that the retinalinputs are crucial for accumulation of ezrin in lamina F or f.


RGC subset-selective expression of ezrinOur observations of antigen expression in the tectum suggested thatretinal afferents are TB4-positive. In support of this finding, TB4selectively stained a subset of RGCs together with some othercomponents of the retina (Fig. 6A). The AChRβ2-subunit-positivesubset of RGCs arose at E8 in a target-independent manner, andabout 10–20% of RGCs were stained with mAb 270. The ezrin-positive RGC subset became apparent from E11, and the number ofpositive cells appeared to increase thereafter. Ezrin-positive RGCswere also seen in the E15 retina of experimentally produced tectalessembryos (data not shown) asβ2-subunit-positive cells (Yamagata &Sanes, 1995b). Double immunofluorescence staining of E18 retinalsections showed that a subset of RGCs were stained with mAbs TB4and 270, and all of the ezrin-positive RGCs wereβ2-subunit-positive(indicated by arrowheads in Fig. 6A and B). This RGC subset-specificstaining pattern was also observed in the retinal sections at P2.

In the retina, TB4 also stained the outer segment significantly andalso the outer plexiform layer and inner nuclear layer to a lesser


FIG. 4. Accumulation of ezrin in lamina Fduring development of retinotectal laminarconnections. Cryosections of E11 (A, B), E14(C, D) and P2 (E, F) optic tecta were double-stained with mouse mAb TB4 (A, C, E) andrat mAb 270 (B, D, F). At E14, the lamina-selective accumulation of ezrin began at thebase of lamina D (arrows), and as developmentproceeded, the lamina-specific staining becamemore evident. For double staining, the absenceof cross-reactivity of the species-specificsecondary antibodies was confirmed. Scale bar,25 µm.



FIG. 5. Eye enucleation and the lamina-selective accumulation of ezrin. In (A) and(B), the eye primordium of E3 chicks wasunilaterally removed, and the embryos wereanalysed at E18. (A) In the tectumcontralateral to the enucleated side, DAPIstaining revealed that laminar development wasperturbed at stages after E14, and that laminaF was not formed, because even lamina E wasnot established. (B) In such altered tecta, theaccumulation of ezrin was not observedsimilarly to the mAb-270-positive AChRsubunit (see Yamagata & Sanes, 1995b). In(C)–(G), the retina was unilaterally ablated atP2, and the chicks kept alive for 1 week.(C) In the operated chicks, the retinal axonsdegenerated, but lamina f was apparentlypreserved in the contralateral tectum. Whencryosections from such tecta were stained withmAb TB4 (D) or mAb 270 (E), both antigenswere absent in lamina f (asterisk) contralateralto the ablated side, whereas the staining in theipsilateral tectum was intact. In contrast, anantibody to ChAT stained lamina f on theaffected (F) as well as unaffected (G) sides ofthe tectum equally, indicating that thepostsynaptic neurons (r2 neurons) for theretinal inputs retained their processes in laminaf even in the absence of retinal inputs. The cellbodies of r2 neurons were stained with theantibody to ChAT in lamina I (arrows; see alsoFig. 8A). Scale bar, 25µm.



FIG. 6. RGC subset-specific expression of ezrin. Cryosections from the E18 retina were triple-stained with mAb TB4 (A), mAb 270 (B) and DAPI (C). All ofthe mAb-TB4-positive RGCs (arrowheads) were also mAb-270-positive. Strong expression of ezrin was also observed in OS. OS, outer segment; ONL, outernuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. Scale bar, 50µm.

extent (Fig. 6A). The staining pattern was similar to that of mAb 270except for the outer part of the inner nuclear layer. This overallstaining hampered whole-mount analysis of the retina with TB4.Strongly TB4-positive cells were also occasionally found on thevitreal side of the inner nuclear layer (data not shown). Theirincidence, morphology and laminar position suggested that theseTB4-positive cells corresponded to displaced RGCs that project ontothe accessory optic centres such as the nucleus of the basal optic root(Brecha & Karten, 1981). In contrast to the TB4-positive cells in theganglion cell layer, these TB4-positive cells in the inner nuclear layerwere mAb-270-antigen-negative, showing that theβ2-subunit-positiveneurons and ezrin-positive neurons do not always overlap. This wasalso the case with other regions of the central and peripheral nervoussystems (data not shown, but see below).

Distribution of ezrin in other neural tissues

Although expression of ezrin was first characterized in epithelial tissues(Bretscher, 1983), several studies have since indicated the expressionof ezrin and related molecules in the nervous system of mammals andbirds (Goslinetal., 1989;Everett&Nichol,1990;Birgbaueretal., 1991;Bohling et al., 1996; Gonzalez-Agosti & Solomon, 1996; Stemmer-Rachamimovet al., 1997). However, some of the antibodies used inthese studies were not specific for ezrin, and immunohistochemicalanalyses in neural tissues have not been performed sufficiently.

The availability of mAb TB4 allowed us to observe the distributionof ezrin in avian embryos. In addition to the visual system, we foundthat TB4 stained other nervous system tissues, together with variousepithelial tissues. In various parts of the nervous system, including thebrainstem, some neuronal populations showed strong expression ofTB4 antigen (data not shown). Particularly interesting staining wasobserved in the DRG and their central projections to the spinal cord(Fig. 7).

The DRG contains several subsets of somatosensory neurons, whichhave distinct peripheral target tissues (e.g. skin, muscle and viscera).They include distinct populations, which transmit different modalitiesof sense such as temperature, pain or pressure. The subpopulations ofDRG neurons are also distinguishable by their size and the position oftheir cell somata in the DRG, their target laminae in the spinal cord,andalsobyotherproperties suchassensitivity todifferentneurotrophins(Fig. 8B; Snider & Wright, 1996; Eide & Glover, 1997; Ozaki & Snider,1997; Shigaet al., 1997; Sharma & Frank, 1998). In E10 chick embryos,the somatosensory neurons located in the dorsomedial and ventrolateralpart of the DRG give rise to lateral and medial afferents in the spinal


cord, respectively (Fig. 8B). The lateral and medial cutaneous afferentsproject into the lateral and medial dorsal horn, respectively, whereasmuscle afferents predominantly project to the lateral motor column inthe ventral region of the spinal cord, and also to the intermediate zone(Eide & Glover, 1997; Ozaki & Snider, 1997; Shigaet al., 1997).

The DRG and associated areas of E10 embryos were stained withE388 (Fig. 7). As shown in Fig. 7A, E388 strongly stained cell somataas well as axons of a subpopulation of somatosensory neurons locatedin the dorsomedial portion of the DRG. The strongly stained afferentaxons projected into the spinal cord predominantly via lateral dorsalfuniculus, and appeared to terminate within the lateral dorsal horn(Fig. 7A and E). The staining pattern was in marked contrast with thatwith antineurofilament 200 antibody where the whole white matter inthe spinal cord was positive (Fig. 7D). Motoneurons and their axonswere only weakly stained with E388 and thereby clearly displayedsegregation of the fascicles of the sensory and motor axons (Fig. 7Aand D). The mAb TB4 displayed the same staining pattern as that withE388 (data not shown).

The expression pattern of ezrin was reminiscent of that of axonin-1/TAG-1, a cell adhesion molecule of the immunoglobulin superfamily; inchick DRG neurons and associated areas, axonin-1/TAG-1 is transientlyexpressed in cutaneous and visceral sensory neurons during their axonaloutgrowth to the spinal cord, but not in the muscle afferent neurons(Halfter et al., 1994; Shigaet al., 1997). Double staining with mAb23.4-5 (an antiaxonin-1/TAG-1 antibody) and E388 indicated thataxonin-1/TAG-1-positive neurons populated a dorsomedial populationof DRG neurons similar to ezrin (Fig. 7B). Like the ezrin-positiveafferents, the axonin-1/TAG-1-expressing afferents terminated in thesuperficial laminae in the dorsal horn (Fig. 7F). However, the localiz-ation of ezrin appeared to be more confined to the lateral dorsal funiculusthan that of axonin-1/TAG-1 (Fig. 7E and F). Thus, the ezrin-positivesubset appeared to be included in the axonin-1/TAG-1-positive popula-tion of DRG neurons.

The regionally-specific distribution of the ezrin-positive neurons inthe DRG, and specificity in the projection in the spinal cord, suggestedthat strong expression of ezrin was restricted to a subpopulation ofcutaneous sensory neurons, and was absent in sensory neurons frommuscle in E10 chick embryos (Eide & Glover, 1997; Sharma & Frank,1998). This possibility was further supported by the observation thatperipheral nerves in the skin were stained with TB4 but no staining wasobserved in nerves in muscle tissue (Fig. 7G and H). Expression ofezrin was also observed in a subset of DRG neurons and their axons atE4.5–5 (stage 26; Hamburger & Hamilton, 1951) (data not shown), at


FIG. 7. Expression of ezrin in a subset ofdeveloping DRG neurons. (A–C) Transversesections at the thoracic level of an E10 embryowere triple-stained with affinity-purified E388(rabbit antiserum to the chick ezrin peptide)(A), mAb 23.4-5 (mouse mAb to axonin-1/TAG-1) (B) and DAPI (C). Boxed areas in (A)and (B) are enlarged in (E) and (F),respectively. The DRG is indicated as ‘drg’and is surrounded by small arrowheads in (A).Highly E388-positive DRG neurons populateda dorsomedial portion of the DRG. DRGneurons ventrolaterally located were negativefor ezrin (asterisk). DRG neurons stained withmAb 23.4-5 also populated a dorsomedialregion of the DRG. This was correlated withthe weak staining of medial dorsal funiculus(mdf in E) compared with lateral dorsalfuniculus (ldf). The DRG cell bodies wereuniformly distributed, as revealed by stainingwith DAPI (arrowheads in C). (D) Othercryosections were stained with a rabbitantineurofilament 200 antibody. Note that thestaining of the dorsal funiculus withantineurofilament antibody was more uniformthan those with E388 and mAb 23.4-5. Thecentral and peripheral axons extending fromthe ezrin-expressing neurons were stronglystained with E388 (large arrowheads in A), butaxons of motoneurons (mt) and most of thespinal cord were not strongly stained with theexception of the lateral dorsal funiculus (df).(E) Ezrin-expressing afferents projected intothe spinal cord, predominantly via the lateraldorsal funiculus (ldf) and terminated in thelateral dorsal horn (asterisk). The medial dorsalfuniculus (mdf) was only weakly stained. (F)The axonin-1/TAG-1-positive afferentsappeared to terminate predominantly in thedorsal horn (dh), and did not extend ventrallyto the lateral motor column (lmc), confirmingthe absence of axonin-1/TAG-1 in musclesensory afferents. (G, H) Photographs of asection of an E10 embryo double-stained withmouse mAb TB4 (G) and rabbit antiserum toneurofilament 200 (H). Although peripheralnerves in the skin (sk) were stained with TB4(arrows), the neurofilament-positive nerves inmuscle (ms) were not stained (arrowheads).The staining patterns with mAb TB4 wereessentially identical to those with E388 (datanot shown). Scale bars, 100µm.

the stage before each subpopulation was spatially sorted out in the DRGand innervated the spinal cord (Sharmaet al., 1994).

Discussion

Lamina-specific molecules in the optic tectum

A series of cell-surface markers were reported to be expressed in alamina-selective manner in the developing chick optic tectum(Yamagataet al., 1995). However, all the molecules characterized,i.e. N-cadherin, SC1/JC7, neuropilin-1 (A5), VVA-B4-bound terminalN-acetylgalactosamine and ephrin-B2, were widely distributed in the


retinorecipient laminae of the SGFS (Takagiet al., 1995; Yamagataet al., 1995; Braistedet al., 1997) and, apart from some neurotransmit-ters and their receptors, no markers have been identified which areexpressed in only one lamina. In this study, we screened for newmarkers capable of specifically labelling one of the retinorecipientlaminae using mAbs. We obtained three antibodies, named TB2, TB4and TB5, each of which stained only one of the retinorecipientlaminae at the developmental stage when retinal afferents establishtheir lamina-specific connections (Fig. 1).

The occurrence of such lamina-selective antigens indicates thateach lamina contains distinct molecules that are either produced by


FIG. 8. Expression of ezrin in a subset of retinotectal and sensory projections. (A) Lamina-specific retinotectal projections. RGC axons extended along thestratum opticum(SO) in the optic tectum, and branched into the deeper retinorecipient laminae B, D and F. Ezrin was expressed in a subset of RGCs bearingthe mAb-270-positive AChR subunit (hatched circles). This subset projected to lamina F, whereas other subsets projected to the other retinorecipient laminaesuch as laminae B and D (closed and open circles, respectively). The r2 cells, putative postsynaptic neurons for the RGC subset projecting to lamina F, hadprocesses in lamina F. The somata of r2 cells resided in lamina I, and projected to the ventral lateral geniculate nucleus (LGNv) in the thalamus. SGC,stratumgriseum centrale. (B) Lamina-specific sensory afferent projections. Somatosensory neurons with large and small cell somata were localized to ventrolateral anddorsomedial regions of the DRG, respectively. Major small cutaneous sensory neurons (small open and hatched circles) supplied lateral afferents that terminatedin the lateral dorsal horn. Large cutaneous sensory neurons (large open circles) terminated predominantly in the medial dorsal horn via a medial pathway. Majorlarge sensory neurons from muscle (large filled circles) projected medial afferents into either the lateral motor column (lmc) to innervate on motoneurons or theintermediate zone (int). Our observations suggested that a part of cutaneous sensory neurons (small hatched circle) expressed ezrin.

the cells located in the lamina itself or transported from cells locatedin the other laminae (or even in extratectal regions). The lamina-specific accumulation of TB5 and TB2 antigens was preserved in thetectum of the eyeless embryos, suggesting that both antigens weresynthesized in tectal cells and/or nonretinal afferents. Recent studiesrevealed that TB5 antigen is located in the cytoplasm of horizontallyorientated cells situated in lamina B, and TB2 antigen is associatedwith fine processes nested in lamina D (our unpublished observations).

In contrast, the TB4 antigen was absent in the tectum contralateralto the enucleated side (Fig. 5). During the development of retinotectalprojections, the lamina-F-selective accumulation of TB4 antigen wassynchronous with the innervation by retinal afferents (Fig. 4). TB4antigen expression was attenuated in lamina f in the retina-ablatedchicks, although ChAT-positive processes of the putative postsynapticneurons (r2 cells) were preserved (Figs 5D,F and G, and 8A). Finally,mAb TB4 stained a subset of RGCs that innervated lamina F (Fig. 6).All of these observations support the idea that presynaptic retinalarbors are responsible for the lamina-selective staining observedwith TB4.

However, we cannot exclude the possibility that postsynapticneurons, and/or nonretinal afferents, also contribute to the accumula-tion in lamina F. In this case, it is necessary to speculate that theretinal inputs evoked transportation of the TB4 antigen to lamina Ffrom the somata of r2 tectal neurons located in lamina I (see Fig. 8A),and/or that they guided minor nonretinal afferents to lamina F (Brecha& Karten, 1981; Hunt & Brecha, 1984).

RGC subset-selective expression of ezrin

In birds as well as mammals, heterogeneity in RGCs is discernibleby their neuronal morphology, electrophysiology and specificity ofprojection to central targets (Stone, 1983; Shapley & Perry, 1986;


Wassle & Boycott, 1991). Moreover, neurochemical phenotypes andtranscriptional factors were found to be divergent among the subsetsof RGCs (Kartenet al., 1990; Xianget al., 1996). Our study revealedthat TB4 antigen is expressed in a subset of RGCs projecting tolamina F (Figs 6 and 8A). The mAb 270, which is an antibody to aneuronal AChRβ2-subunit, also labels 10–20% of the RGCs (Yama-gata & Sanes, 1995b). Although the TB4-positive RGCs overlappedwith the mAb-270-positive cells, the former population appeared tobe included in the latter. This might be due to differences, betweenmAbs TB4 and 270, in staining sensitivity. Both proteins were alsoexpressed in the RGC subset in tectum-ablated embryos, suggestingthat the phenotype of the RGC subset is determined autonomouslyin the retina.

Interestingly, TB4 antigen was identified as ezrin. To our knowledge,ezrin is the first cytoskeleton-associated molecule that has been shownto be selectively expressed in a specific subset of RGCs. To test thepossibility that ezrin and other ERM members can classify RGCsubsets in other species, we stained mouse retina and brain with apanel of antibodies reactive with mouse ezrin, radixin or moesin(M11, R2-1, M22, and CR22). However, in our preliminary experi-ments, none of their staining patterns were associated with RGCsubsets or laminar structures in the brain such as the superiorcolliculus, lateral geniculate nucleus, hippocampus, and neocortex(data not shown).

Expression of ezrin in a subset of DRG neurons

The lamina-selective projections of different subsets in DRG neuronshave been well characterized. Recent studies suggested that region-specific expression of the collapsin/semaphorin family guide eachsubset of sensory axons to a particular region in the spinal cord(Messersmithet al., 1995; Wrightet al., 1995; Pu¨schelet al., 1996;


Redmondet al., 1997). Interestingly, collapsin-2 is expressed also inthe optic tectum in a lamina-specific pattern (Luoet al., 1995). Inthe DRG, the neuronal subsets are divergent with respect to expressionprofiles of cell adhesion molecules (Shimamuraet al., 1992; Halfteret al., 1994) and neurotrophin receptors (Snider & Wright, 1996).Our findings indicate that membrane-cytoskeletal linker proteins alsoshow diversity among the neuronal subsets in the DRG.

Ezrin-positive cells were localized predominantly in the dorsomed-ial part of the DRG in E10 chick embryos (Figs 7A and 8B), andtheir afferents appeared, predominantly, to pass through the lateraldorsal funiculus and terminate in the lateral dorsal horn (Fig. 7A andE). This projection pattern suggested that ezrin-positive neurons werea cutaneous subset terminating in the superficial layer of the dorsalhorn, because muscle afferents predominantly pass through the medialdorsal funiculus and terminate in the ventral horn and intermediatezone (Fig. 8B; Eide & Glover, 1997). Consistent with this observation,in the peripheral part, ezrin-positive nerve fibers were observed inthe skin but not in the muscle (Fig. 7G and H). A small number ofezrin-positive axonal fragments were also seen in the medial dorsalfuniculus, indicating that another minor subpopulation of DRGsensory neurons may also express ezrin (Fig. 7E).

It was reported that cultured DRG neurons prepared from E10chick embryos were stained with antiradixin antibody but not withantiezrin or antimoesin antibodies (Gonzalez-Agosti & Solomon,1996). The antibodies used were raised against mouse ERM sequences.However, our results, obtained using two independent antibodies,E388 and TB4, which specifically recognize chick ezrin but notradixin, indicated that ezrin is expressed at high levels in a subpopul-ation of DRG neurons in E10 chick embryos. This discrepancy mayhave been due to differences between tissue sections and culturedcells, or in the sensitivities of the antibodies.

The expression pattern of ezrin suggested a common mechanismthat underlies the development of neuronal subsets in both theretinotectal and somatosensory systems. This notion was reminiscentof the subset-specific expression of a member of the Brn-3 family oftranscription factors with a POU-domain in the RGC and sensoryneurons in mammals (Xianget al., 1995). However, the relationshipbetween the expression of members of the ERM and Brn-3 familiesis not clear at present.

Possible function of ezrin in the lamina-specific projection

The ERM family is a subgroup of the band 4.1 superfamily ofcytoskeleton-associated proteins that interact with various proteins atthe interface between the plasma membrane and cytoskeleton. Mostof the members which harbour a conservedN-terminal domain, i.e.a band-4.1-like domain, have been suggested to be involved in thecontrol of cell shape and adhesion (e.g. band 4.1, ezrin, radixin,moesin, merlin, talin and some protein-tyrosine phosphatases) (Takeu-chi et al., 1994a). More specifically, ezrin is involved in microvillusformation (Berrymanet al., 1993; Takeuchiet al., 1994b) and target-cell recognition during immune responses (Helanderet al., 1996).Several factors modulate the function of ezrin by phosphorylation,which is correlated with the dynamics of morphogenesis and adhesionof cells (Bretscher, 1989; Hanzelet al., 1991; Egertonet al., 1992;Krieg & Hunter, 1992; Berrymanet al., 1995; Chenet al., 1995).

Ezrin-related epitopes were found in the lamellipodia of growthcones in cultured neurons (Goslinet al., 1989; Birgbaueret al.,1991), although these previous studies could not detect ezrin expressedin neuronal subsets due to the low specificity of the antibodies used.Our finding, that ezrin is expressed in neuronal subsets, suggested aspecific function of ezrin in the nervous system. Several reportssuggesting such possibilities have recently appeared. Some of the


band-4.1 superfamily members are likely to be associated with ahomologue of theDrosophila discs-largetumour suppressor protein,which is a member of the PSD-95, DIg and Z0-1 (PDZ) proteinfamily and is involved in development and regulation of synapticarchitecture (Choet al., 1992; Marfatiaet al., 1995; Budniket al.,1996; Broadie, 1998). The PDZ-containing proteins are mostly locatednear the synaptic membrane and, therefore, are thought to transportsynaptic proteins to the synapse and/or anchor them at synaptic sites(Craven & Bredt, 1998). Ezrin anchors a series of cell adhesionmolecules such as CD44, CD43, ICAM2 and ICAM1/CD54 (Tsukitaet al., 1994; Helanderet al., 1996; Yonemuraet al., 1998), whichmay play important roles in a variety of neuronal connections (Tessier-Lavigne & Goodman, 1996). Intracellularly, ERM proteins have beenreported to bind other membrane-associated intracellular protein suchas EBP50/NHE-RF (Reczeket al., 1997; Murthyet al., 1998) andE3KARP/TKA-1 (Yun et al., 1997). Both proteins contain two PDZdomains and have been identified as cofactors necessary for theregulation of Na1/H1 ion exchangers (Weinmanet al., 1995; Reczeket al., 1997; Yunet al., 1997; Murthyet al., 1998). Thus, ezrin (orall the ERM family) is expected to be involved in the formation ofmultiprotein complexes consisting of membrane proteins (e.g. celladhesion molecules, ion channels and ion exchangers), PDZ proteinsand cytoskeletal proteins beneath the neuronal plasma membraneincluding those at synaptic sites (Craven & Bredt, 1998).

We surmise that the lamina-specific projection of ezrin-expressingaxonal terminals in the visual and somatosensory systems may reflecta role for ezrin in the maintenance and/or formation of specificneuronal connections in specific laminae, and it may indicate a novelfunction for ezrin and its related molecules in the nervous system.

AcknowledgementsWe are grateful to Dr F. Solomon (MIT) for monoclonal antibody 13H9, DrS. Tsukita (Kyoto University) for monoclonal antibodies E1, R2-1, M22, andCR22, and Dr T. Arai (Tsukuba University) for PAI myeloma cells. We alsothank Ms A. Kawai and Ms M. Mizoguchi for technical assistance, and MsA. Kodama for secretarial assistance. This work was supported by grants fromthe Ministry of Education, Science, Sports and Culture of Japan, CREST of theJapan Science and Technology Corporation, and the Yamanouchi, Sumitomo,Uehara, Ciba-Geigy and Sasagawa Foundation.

AbbreviationsAA, amino acid; AChR, acetylcholine receptor; ChAT, cholineacetyltrans-ferase; DAPI, 49, 6-diamidino-2-phenylindole; DRG, dorsal root gangion; E,embryonic day; ERM, ezrin-radixin-moesin; GST, glutathione S-transferase;mAb, monoclonal antibody; MAP, multiple antigenic peptide; P, posthatchingday; PBS, phosphate-buffered saline; RGC, retinal ganglion cell; SDS-PAGE,sodium dodecyl sulphate-polyacrylamide gel electrophoresis; SGFS,stratumgriseum et fibrosum superficiale.

ReferencesAcheson, D.W.K., Kemplay, S.K. & Webster, K.E. (1980) Quantitative analysis

of optic terminal profile distribution within the pigeon optic tectum.Neuroscience, 5, 1067–1084.

Angaut, P. & Repe´rant, J. (1976) Fine structure of the optic fibre terminationlayers in the pigeon optic tectum: a golgi and electron microscope study.Neuroscience, 1, 93–105.

Bagnoli, P., Fontanesi, G., Alesci, R. & Erichsen, J.T. (1992) Distribution ofneuropeptide Y, substance P, and choline acetyltransferase in the developingvisual system of the pigeon and effects of unilateral retina removal.J. Comp.Neurol., 318, 392–414.

Berryman, M., Franck, Z. & Bretscher, A. (1993) Ezrin is concentrated in theapical microvilli of a wide variety of epithelial cells whereas moesin isfound primarily in endothelial cells.J. Cell Sci., 105, 1025–1043.

Berryman, M., Gary, R. & Bretscher, A. (1995) Ezrin oligomers are major


cytoskeletal components of placental microvilli: a proposal for theirinvolvement in cortical morphogenesis.J. Cell Biol., 131, 1231–1242.

Birgbauer, E., Dinsmore, J.H., Winckler, B., Lander, A.D. & Solomon, F.(1991) Association of ezrin isoforms with the neuronal cytoskeleton.J. Neurosci. Res., 30, 232–241.

Bohling, T., Turunen, O., Ja¨askelainen, J., Carpen, O., Sainio, M., Wahlstro¨m,T., Vaheri, A. & Haltia, M. (1996) Ezrin expression in stromal cells ofcapillary hemangioblastoma. An immunohistochemical survey of braintumors.Am. J. Pathol., 148, 367–373.

Braisted, J.E., McLaughlin, T., Wang, H.U., Friedman, G.C., Anderson, D.J.& O’Leary, D.D.M. (1997) Graded and lamina-specific distributions ofligands of EphB receptor tyrosine kinases in the developing retinotectalsystem.Dev. Biol., 191, 14–28.

Brecha, N.C. (1978) Some observations on the organization of the avian optictectum: Afferent nuclei and their tectal projections. PhD Thesis, Departmentof Psychology, State University of New York, Stony Brook.

Brecha, N.C. & Karten, H.J. (1981) Organization of the avian accessory opticsystem.Ann. NY Acad. Sci., 374, 215–229.

Bretscher, A. (1983) Purification of an 80,000-dalton protein that is acomponent of the isolated microvillus cytoskeleton, and its localization innonmuscle cells.J. Cell Biol., 97, 425–432.

Bretscher, A. (1989) Rapid phosphorylation and reorganization of ezrin andspectrin accompany morphological changes induced in A-431 cells byepidermal growth factor.J. Cell Biol., 108, 921–930.

Bretscher, A., Reczek, D. & Berryman, M. (1997) Ezrin: a protein requiringconformational activation to link microfilaments to the plasma membranein the assembly of cell surface structures.J. Cell Sci., 110, 3011–3018.

Britto, L.R.G., Keyser, K.T., Lindstrom, J.M. & Karten, H.J. (1992)Immunohistochemical localization of nicotinic acetylcholine receptorsubunits in the mesencephalon and diencephalon of the chick (Gallusgallus). J. Comp. Neurol., 317, 325–340.

Broadie, K. (1998) Forward and reverse genetic approaches to synaptogenesis.Curr. Opin. Neurobiol., 8, 128–138.

Brown, A.G. (1981)Organization in the Spinal Cord. The Anatomy andPhysiology of Identified Neurons.Springer-Verlag, New York.

Budnik, V., Koh, Y.-H., Guan, B., Hartmann, B., Hough, C., Woods, D. &Gorczyca, M. (1996) Regulation of synapse structure and function by theDrosophila tumor suppressor genedlg. Neuron, 17, 627–640.

Chen, J., Cohn, J.A. & Mandel, L.J. (1995) Dephosphorylation of ezrin as anearly event in renal microvillar breakdown and anoxic injury.Proc. NatlAcad. Sci. USA, 92, 7495–7499.

Cho, K.-O., Hunt, C.A. & Kennedy, M.B. (1992) The rat brain postsynapticdensity fraction contains a homolog of theDrosophila discs-largetumorsuppressor protein.Neuron, 9, 929–942.

Craven, S.E. & Bredt, D.S. (1998) PDZ proteins organize synaptic signalingpathways.Cell, 93, 495–498.

Egerton, M., Burgess, W.H., Chen, D., Druker, B.J., Bretscher, A. & Samelson,L.E. (1992) Identification of ezrin as an 81 kDa tyrosine phosphorylatedprotein in T cells.J. Immunol., 149, 1847–1852.

Ehrlich, D., Keyser, K.T. & Karten, H.J. (1987) Distribution of substance P-like immunoreactive retinal ganglion cells and their pattern of terminationin the optic tectum of chick (Gallus gallus). J. Comp. Neurol., 266, 220–233.

Eide, A.L. & Glover, J.C. (1997) Developmental dynamics of functionallyspecific primary sensory afferent projections in the chicken embryo.Anat.Embryol. (Berl.), 195, 237–250.

Everett, A.W. & Nichol, K.A. (1990) Ezrin immunoreactivity in neuronsubpopulations: Cellular distribution in relation to cytoskeletal proteins insensory neurons.J. Histochem. Cytochem., 38, 1137–1144.

Franck, Z., Gary, R. & Bretscher, A. (1993) Moesin, like ezrin, colocalizeswith actin in the cortical cytoskeleton in cultured cells, but its expressionis more variable.J. Cell Sci., 105, 219–231.

Gonzalez-Agosti, C. & Solomon, F. (1996) Response of radixin to perturbationsof growth cone morphology and motility in chick sympathetic neuronsin vitro. Cell Motil. Cytoskeleton, 34, 122–136.

Goslin, K., Birgbauer, E., Banker, G. & Solomon, F. (1989) The role ofcytoskeleton in organizing growth cones: a microfilament-associated growthcone component depends upon microtubules for its localization.J. CellBiol., 109, 1621–1631.

Halfter, W., Yip, Y.P.L. & Yip, J.W. (1994) Axonin 1 is expressed primarilyin subclasses of avian sensory neurons during outgrowth.Dev. Brain Res.,78, 87–101.

Hamassaki-Britto, D.E., Gardino, P.F., Hokoc, J.N., Keyser, K.T., Karten, H.J.,Lindstrom, J.M. & Britto, L.R. (1994) Differential development ofα-bungarotoxin-sensitive andα-bungarotoxin-insensitive nicotinic acetyl-choline receptors in the chick retina.J. Comp. Neurol., 347, 161–170.


Hamburger, V. & Hamilton, H.L. (1951) A series of normal stages in thedevelopment of the chick embryo.J. Morphol., 88, 49–92.

Hanzel, D., Reggio, H., Bretscher, A., Forte, J.G. & Mangeat, P. (1991) Thesecretion-stimulated 80 K phosphoprotein of parietal cells is ezrin, and hasthe properties of a membrane cytoskeletal linker in the induced apicalmicrovilli. EMBO J., 10, 2363–2373.

Harlow, E. & Lane, D. (1988)Antibodies. A Laboratory Manual.Cold SpringHarbor, Cold Spring, Harbor Laboratory Press, New York.

Helander, T.S., Carpe´n, O., Turunen, O., Kovanen, P.E., Vaheri, A. & Timonen,T. (1996) ICAM-2 redistributed by ezrin as a target for killer cells.Nature,382, 265–268.

Hunt, S.P. & Brecha, N.C. (1984) The avian optic tectum: A synthesis ofmorphology and biochemistry. In: Vanegas, H. (ed.),Comparative Neurologyof the Optic Tectum.Plenum Press, New York, pp. 619–648.

Johnson, C.D. & Epstein, M.L. (1986) Monoclonal antibodies and polyvalentantiserum to chicken choline acetyltransferase.J. Neurochem., 46, 968–976.

Karten, H.J., Keyser, K.T. & Brecha, N.C. (1990) Biochemical andmorphological heterogeneity of retinal ganglion cells. In Cohen, B. &Bodis-Wollner, I. (eds),Vision and the Brain.Raven Press, New York, pp.19–33.

Kelly, J.P. & Cowan, W.M. (1972) Studies on the development of the chickoptic tectum. III. Effects of early eye removal.Brain Res., 42, 263–288.

Keyser, K.T., Britto, L.R.G., Schoepfer, R., Whiting, P., Cooper, J., Conroy,W., Brozozowska-Prechtl, A., Karten, H.J. & Lindstrom, J. (1993) Threesubtypes ofα-bungarotoxin-sensitive nicotinic acetylcholine receptors areexpressed in chick retina.J. Neurosci., 13, 442–454.

Keyser, K.T., Hughes, T.E., Whiting, P.J., Lindstrom, J.M. & Karten, H.J.(1988) Cholinoceptive neurons in the retina of the chick: animmunohistochemical study of the nicotinic acetylcholine receptors.Vis.Neurosci., 1, 349–366.

Krieg, J. & Hunter, T. (1992) Identification of the two major epidermal growthfactor-induced tyrosine phosphorylation sites in the microvillar core proteinezrin.J. Biol. Chem., 267, 19258–19265.

Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly ofthe head of bacteriophage T4.Nature, 227, 680–685.

Luo, Y., Shepherd, I., Li, J., Renzi, M.J., Chang, S. & Raper, J.A. (1995) Afamily of molecules related to collapsin in the embryonic chick nervoussystem.Neuron, 14, 1131–1140.

Marfatia, S.M., Lue, R.A., Branton, D. & Chishti, A.H. (1995) Identificationof the protein 4.1 binding interface on glycophorin C and p55, a homologueof the Drosophila discs-largetumor suppressor protein.J. Biol. Chem.,270, 715–719.

Medina, L. & Reiner, A. (1994) Distribution of choline acetyltransferaseimmunoreactivity in the pigeon brain.J. Comp. Neurol., 342, 497–537.

Messersmith, E.K., Leonardo, E.D., Shatz, C.J., Tessier-Lavigne, M.,Goodman, C.S. & Kolodkin, A.L. (1995) Semaphorin III can function as aselective chemorepellent to pattern sensory projections in the spinal cord.Neuron, 14, 949–959.

Mey, J. & Thanos, S. (1992) Development of the visual system of the chick –a review.J. Hirnforsch., 33, 673–702.

Murthy, A., Gonzalez-Agosti, C., Cordero, E., Pinney, D., Candia, C., Solomon,F., Gusella, J. & Ramesh, V. (1998) NEH-RF, a regulatory cofactor forNa1-H1 exchange, is a common interactor for merlin and ERM (MERM)proteins.J. Biol. Chem., 273, 1273–1276.

Ozaki, S. & Snider, W.D. (1997) Initial trajectories of sensory axons towardlaminar targets in the developing mouse spinal cord.J. Comp. Neurol., 380,215–229.

Puschel, A.W., Adams, R.H. & Betz, H. (1996) The sensory innervation ofthe mouse spinal cord may be patterned by differential expression of anddifferential responsiveness to semaphorins.Mol. Cell Neurosci., 7, 419–431.

Reczek, D., Berryman, M. & Bretscher, A. (1997) Identification of EBP50: aPDZ-containing phosphoprotein that associates with members of the ezrin-radixin-moesin family.J. Cell Biol., 139, 169–179.

Redies, C., Arndt, K. & Ast, M. (1997) Expression of the cell adhesionmolecule axonin-1 in neuromeres of the chicken diencephalon.J. Comp.Neurol., 381, 230–252.

Redmond, L., Xie, H., Ziskind-Conhaim, L. & Hockfield, S. (1997) Cuesintrinsic to the spinal cord determine the pattern and timing of primaryafferent growth.Dev. Biol., 182, 205–218.

Reiner, A. & Karten, H.J. (1982) Laminar distribution of the cells of originof the descending tectofugal pathways in the pigeon (Columba livia). J.Comp. Neurol., 204, 165–187.

Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989)Molecular Cloning: aLaboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press,New York.


Sato, N., Funayama, N., Nagafuchi, A., Yonemura, S., Tsukita, S. & Tsukita,S. (1992) A gene family consisting of ezrin, radixin and moesin. Its specificlocalization at actin filament/plasma membrane association sites.J. CellSci., 103, 131–143.

Shapley, R. & Perry, V.H. (1986) Cat and monkey retinal ganglion cells andtheir visual functional roles.Trends Neurosci., 9, 229–235.

Sharma, K. & Frank, E. (1998) Sensory axons are guided by local cues inthe developing dorsal spinal cord.Development, 125, 635–643.

Sharma, K., Korade, Z. & Frank, E. (1994) Development of specific muscleand cutaneous sensory projections in cultured segments of spinal cord.Development, 120, 1315–1323.

Shiga, T., Lustig, M., Grumet, M. & Shirai, T. (1997) Cell adhesion moleculesregulate guidance of dorsal root ganglion axons in the marginal zone andtheir invasion into the mantle layer of embryonic spinal cord.Dev. Biol.,192, 136–148.

Shimamura, K., Takahashi, T. & Takeichi, M. (1992). E-cadherin expressionin a particular subset of sensory neurons.Dev. Biol., 152, 242–254.

Snider, W.D. & Wright, D.E. (1996). Neurotrophins cause a new sensation.Neuron, 16, 229–232.

Sorenson, E.M., Parkinson, D., Dahl, J.L. & Chiappinelli, V.A. (1989)Immunohistochemical localization of choline acetyltransferase in the chickenmesencephalon.J. Comp. Neurol., 281, 641–657.

Stemmer-Rachamimov, A.O., Gonzalez-Agosti, C., Xu, L., Burwick, J.A.,Beauchamp, R., Pinney, D., Louis, D.N. & Ramesh, V. (1997) Expressionof NF2-encoded merlin and related ERM family proteins in the humancentral nervous system.J. Neuropathol. Exp. Neurol., 56, 735–742.

Stone, J. (1983)Parallel Processing in the Visual System. The Classificationof Retinal Ganglion Cells and its Impact on the Neurobiology of Vision.Plenum Press, New York.

Takagi, S., Kasuya, Y., Shimizu, M., Matsuura, T., Tsuboi, M., Kawakami,A. & Fujisawa, H. (1995) Expression of a cell adhesion molecule, neuropilin,in the developing chick nervous system.Dev. Biol., 170, 207–222.

Takeuchi, K., Kawashima, A., Nagafuchi, A. & Tsukita, S. (1994a) Structuraldiversity of band 4.1 superfamily members.J. Cell Sci., 107, 1921–1928.

Takeuchi, K., Sato, N., Kasahara, H., Funayama, N., Nagafuchi, A., Yonemura,S., Tsukita, S. & Tsukita, S. (1994b) Perturbation of cell adhesion andmicrovilli formation by antisense oligonucleotides to ERM family members.J. Cell Biol., 125, 1371–1384.

Tam, J.P. (1988) Synthetic peptide vaccine design: Synthetic and propertiesof a high density multiple antigenic peptide system.Proc. Natl Acad. Sci.USA, 85, 5409–5413.

Tessier-Lavigne, M. & Goodman, C.S. (1996) The molecular biology of axonguidance.Science, 274, 1123–1133.

Towbin, H., Staehelin, T. & Gordon, J. (1979) Electrophoretic transfer ofproteins from polyacrylamide gels to nitrocellulose sheets: Procedure andsome applications.Proc. Natl Acad. Sci. USA, 76, 4350–4354.

Trofatter, J.A., MacCollin, M.M., Rutter, J.L., Murrell, J.R., Duyao, M.P.,Parry, D.M., Eldridge, R., Kley, N., Menon, A.G., Pulaski, K., Haase, V.H.,Ambrose, C.M., Munroe, D., Bove, C., Haines, J.L., Martuza, R.L.,


MacDonald, M.E., Seizinger, B.R., Short, M.P., Buckler, A.J. & Gusella,J.F. (1993) A novel moesin-, ezrin-, radixin-like gene is a candidate for theneurofibromatosis 2 tumor suppresor.Cell, 72, 791–800.

Tsukita, S., Oishi, K., Sato, N., Sagara, J., Kawai, A. & Tsukita, S. (1994)ERM family members as molecular linkers between the cell surfaceglycoprotein CD44 and actin-based cytoskeletons.J. Cell Biol., 126,391–401.

Tsukita, S., Yonemura, S. & Tsukita, S. (1997) ERM (ezrin/radixin/moesin)family: from cytoskeleton to signal transduction.Curr. Opin. Cell Biol., 9,70–75.

Wassle, H. & Boycott, B.B. (1991) Functional architecture of the mammalianretina.Physiol. Rev., 71, 447–480.

Weinman, E.J., Steplock, D., Wang, Y. & Shenolikar, S. (1995) Characterizationof a protein cofactor that mediates protein kinase A regulation of the renalbrush border membrane Na1-H1 exchanger.J. Clin. Invest., 95, 2143–2149.

Whiting, P. & Lindstrom, J. (1987) Purification and characterization of anicotinic acetylcholine receptor from rat brain.Proc. Natl Acad. Sci. USA,84, 595–599.

Winckler, B., Gonzalez-Agosti, C., Magendantz, M. & Solomon, F. (1994)Analysis of a cortical cytoskeletal structure: a role for ezrin-radixin-moesin(ERM proteins) in the marginal band of chicken erythrocytes.J. Cell Sci.,107, 2523–2534.

Wright, D.E., White, F.A., Gerfen, R.W., Silos-Santiago, I. & Snider, W.D.(1995) The guidance molecule semaphorin III is expresed in regions ofspinal cord and periphery avoided by growing sensory axons.J. Comp.Neurol., 361, 321–333.

Xiang, M., Zhou, L., Macke, J.P., Yoshioka, T., Hendry, S.H.C., Eddy, R.L.,Shows, T.B. & Nathans, J. (1995) The Brn-3 family of POU-domain factors:Primary structure, binding specificity, and expression in subsets of retinalganglion cells and somatosensory neurons.J. Neurosci., 15, 4762–4785.

Xiang, M., Zhou, H. & Nathans, J. (1996) Molecular biology of retinalganglion cells.Proc. Natl Acad. Sci. USA, 93, 596–601.

Yamagata, M., Herman, J.-P. & Sanes, J.R. (1995) Lamina-specific expressionof adhesion molecules in developing chick optic tectum.J. Neurosci., 15,4556–4571.

Yamagata, M. & Sanes, J.R. (1995a) Lamina-specific cues guide outgrowthand arborization of retinal axons in the optic tectum.Development, 121,189–200.

Yamagata, M. & Sanes, J.R. (1995b) Target-independent diversification andtarget-specific projection of chemically defined retinal ganglion cell subsets.Development, 121, 3763–3776.

Yonemura, S., Hirao, M., Doi, Y., Takahashi, N., Kondo, T., Tsukita, S. &Tsukita, S. (1998) Ezrin/radixin/moesin (ERM) proteins bind to a positivelycharged amino acid cluster in the juxta-membrane cytoplasmic domain ofCD44, CD43, and ICAM-2.J. Cell Biol., 140, 885–895.

Yun, C.H.C., Oh, S., Zizak, M., Steplock, D., Tsao, S., Tse, C.-M., Weinman,E.J. & Donowitz, M. (1997) cAMP-mediated inhibition of the epithelialbrush border Na1/H1 exchanger, NHE3, requires an associated regulatoryprotein.Proc. Natl Acad. Sci. USA, 94, 3010–3015.

Documents

Specific expression of ezrin, a cytoskeletal-membrane linker protein, in a subset of chick retinotectal and sensory projections