laminar boundaries persist in the hippocampal dentate molecular layer of the mutantshaking rat...

Laminar Boundaries Persist in theHippocampal Dentate Molecular Layerof the Mutant Shaking Rat Kawasaki

Despite Aberrant Granule Cell Migration

PETER L. WOODHAMS1* AND TOSHIO TERASHIMA2

1Division of Neurobiology, National Institute for Medical Research,London NW7 1AA, United Kingdom

2Department of Anatomy, Kobe University School of Medicine, Kobe 758, Japan

ABSTRACTThe present report provides the first detailed description of the hippocampus in the

Shaking Rat Kawasaki (SRK) mutant by using a panel of antibody markers to delineate itslaminar organization. The mutant was characterised at postnatal day 21 by severe malforma-tions of both neuronal position and orientation, the most striking of which was the presence ofa rounded central granule cell mass in the dentate gyrus rather than the normal V-shapedgranule cell layer. Despite this finding, the SRK dentate gyrus not only retained a cell-sparsemolecular layer (thinner but similar in gross appearance to that of control littermates), butthe sharp laminar boundary between its inner and outer parts was as clearly marked by IM1and OM4 antibody staining as it was in the normal dentate gyrus. These immunocytochemicaldata suggest that the entorhinal terminal field of the dentate gyrus may be relatively normalin the mutant, despite entorhinal afferents appearing to take an abnormal trajectory afterthey fail to cross the hippocampal fissure. Laminar malformations included disruption of theSRK pyramidal cell layer, with spreading of the CA3 mossy fibre projection to an ectopicinfrapyramidal position, radial displacement of CA1 pyramids, and transposition of a hithertounremarked longitudinal fibre bundle immunoreactive for calretinin from its normal positionin the stratum lacunosum-moleculare of field CA2 to an alvear position in SRK. The SRKmalformations were very like but not identical to those seen in the mouse reeler mutant,suggesting similar underlying developmental mechanisms. J. Comp. Neurol. 409:57–70, 1999.r 1999 Wiley-Liss, Inc.

Indexing terms: reeler mouse; neurological mutant rat; entorhinal cortex; immunocytochemistry

The Shaking Rat Kawasaki (SRK) is an autosomalrecessive mutant that was first described in 1988 byAikawa et al. However, a detailed description of theabnormal laminar cytoarchitecture of the cerebral cortexof this mutant has only recently been provided (Ikeda andTerashima, 1997), from which it is apparent that SRKshares strong phenotypic similarities with the murinereeler mutant. There has been a recent resurgence ofinterest in the reeler mouse and related mutants aftercloning of the gene for reelin (D’Arcangelo et al., 1995,1997), and studies on the reeler mutant are at last begin-ning to shed light on some of the possible mechanismsunderlying the specification of cortical connectivity andthe laminar organization of cortical structures (Ogawa etal., 1995; Del Rıo et al., 1997; Frotscher, 1997, 1998;Howell et al., 1997; Sheldon et al., 1997).

We have described a number of molecular markerswhose expression appears to be related to the laminarsegregation of inputs to the hippocampus and that may, asa consequence, be of use in analysing the developmentalprocesses that underlie these mutant phenotypes. Monoclo-nal antibody IM1 preferentially stains the inner one-thirdof the molecular layer of the dentate gyrus, where commis-sural and associational afferents terminate, and a family

Grant sponsor: Medical Research Council.*Correspondence to: Peter L. Woodhams, Division of Neurobiology,

National Institute for Medical Research, The Ridgeway, London NW7 1AA,U.K. E-mail: [email protected]

Received 30 June 1998; Revised 8 January 1999; Accepted 4 February1999

THE JOURNAL OF COMPARATIVE NEUROLOGY 409:57–70 (1999)

r 1999 WILEY-LISS, INC.

of at least two different antigens in the outer two-thirds ofthe molecular layer (the entorhinal zone) are defined byantibodies OM1 and OMs 2, 3, and 4 (Woodhams et al.,1992b). The characteristic staining patterns of the OMmonoclonal antibodies in the molecular layer of the den-tate gyrus are abolished by lesions to the ipsilateralentorhinal cortex (the major source of afferents to theouter molecular layer) and restored by grafts of embryonicentorhinal tissue (Woodhams et al., 1992a), indicating thatexpression of these laminar antigens is dependant on anormal pattern of connectivity. The OM antibodies havealso provided a means of studying in vitro the developmentand regeneration of laminated pathways such as theentorhinal projection by using organotypic hippocampalslices cocultured with entorhinal explants (Woodhams etal., 1993; Woodhams and Atkinson, 1996). Although workto identify and fully characterise the glycoprotein antigensrecognised by our antibodies has yet to be completed(Webb, Atkinson, and Woodhams, in preparation), thesemonoclonal reagents nevertheless provide a convenientmeans of identifying and analysing laminar patterns ofhippocampal connectivity.

Whereas antibody IM1 recognises tissue from both ratsand mice, the OM monoclonals do not recognise mousebrain, a feature that has been used to advantage inidentifying axons in rat-to-mouse cocultures (Woodhamset al., 1993). Unfortunately, this species specificity of theOM antibodies precludes their use as reagents with whichto study mouse mutants such as the reeler, where thedistribution of the OM and IM laminar markers could shedlight on the nature of the cortical malformations presentand the possible role of these antigens. However, the factthat SRK is a rat rather than a mouse mutation suggeststhat it may provide a means of overcoming the limitationspresented by the species specificity of the OM monoclonalantibodies. In the present report, data from OM and IMstaining of the SRK mutant hippocampus are correlatedwith the distribution of a number of other immunocyto-chemical markers, including antibodies to the pyramidalcell antigen, Py (Woodhams et al., 1989), to neurofilamentsto show axons, to microtubule-associated protein 2 as amarker of dendrites (Bernhardt et al., 1985; Caceres et al.,1986), and to the calcium-binding proteins calbindin andcalretinin, which each have their own characteristic,nonoverlapping labeling patterns in the hippocampus(Celio, 1990; Jacobowitz and Winsky, 1991; Baimbridge etal., 1992; Resibois and Rogers, 1992). Where possible,comparisons have also been made with the reeler mutant,in which staining can be carried out for all these markersexcept the OM monoclonals.

MATERIALS AND METHODS

Animals

Wistar rats carrying the SRK mutation and C57BL/6Jreeler mice were raised at the Tokyo Metropolitan Insti-tute for Neuroscience (Ikeda and Terashima, 1997). Allanimals were housed in a temperature-controlled (22°C)colony room with a 12-hour light/dark cycle, in groups inacrylic cages with woodchip bedding and unlimited accessto normal laboratory chow and water. All procedures wereapproved by the Committee on Animal Care and Welfare,Kobe University Medical School.

Analysis was carried out on 10 rats homozygous for theautosomal recessive SRK mutation and on seven normal

littermates. Because homozygous SRK rats are infertile,mutants were obtained by crosses from heterozygotes. Asexpected, 25% of the offspring of such crosses proved to beSRK. Heterozygous 1/SRK control animals (50% of eachlitter) were not separated from the 25% 1/1 wild-typelittermates: they can only be distinguished by progenytesting, and no phenotypic distinctions between these twogenotypes have been described because their central ner-vous system (CNS) neuroanatomy appears to be indistin-guishable. The SRK mutant phenotype was, however,easily recognised from approximately postnatal day 10 byabnormal locomotor behaviour (tremor, dystonia, andataxia), poor suckling, and emaciation (Aikawa et al.,1988).

Immunocytochemistry

Because of the reduced viability of the SRK homozy-gotes, all animals were killed for immunohistochemicalexamination on postnatal day 21 (P21). They were deeplyanaesthetised with intraperitoneal chloral hydrate (3.5%,0.8 ml/100 g body weight) and cannulated through the leftventricle. After transcardiac flushing of the blood withsaline, the dissected brains were immersion fixed over-night in cold 5% glacial acetic acid in 96% ethanol and thenwashed several times in 96% ethanol. For transit betweeninstitutions, fixed samples were stored for 5–7 days in 70%ethanol with no adverse effects. The rostral forebrain ofeach brain was discarded after a coronal cut at the level ofthe fimbria, and the caudal portion was divided in two by amid-horizontal cut: the upper half of the caudal forebrainwas orientated to provide transverse sections of the septalpole of the hippocampus and the lower half horizontally toprovide transverse sections of the temporal pole. Tissueblocks were embedded in polyester wax (Merck, Darm-stadt, Germany; Woodhams et al., 1989, 1992b) and sec-tioned at 7 µm before staining by either the ABC techniquewith silver enhancement and thionin counterstaining(Woodhams et al., 1992b) or the glucose oxidase-nickeldiaminobenzidine method (Shu et al., 1988).

The following primary antibodies were used (all weremouse monoclonal antibodies unless otherwise stated):ascitic fluid for IM1, OM1, and OM4 (Woodhams et al.,1992b) diluted to 1:4,000, 1:2,000, and 1:1,000 respec-tively; Py marker for large neurons (Woodhams et al.,1989), 1:2,000; anti-microtubule-associated protein 2 a 1 b(MAP2; 1:500; cat. no. M-1406, Sigma, Poole, UK); axonalmarker anti-neurofilament SMI-31 (1:5,000; NA 1219,Affiniti Research, Nottingham, UK); rabbit anti-calretinin(1:5,000; code 7696, SWant, Bellinzona, Switzerland); rab-bit anti-calbindin (1:5,000; code CB-38, SWant); and rabbitanti-glial fibrillary acidic protein (1:2,000; code Z0334;DAKO, Cambridge, UK). Control incubations used normalmouse or rabbit serum diluted 1:1,000 in place of theprimary antibody.

Choice of laminar antibodies

Our original report of the OM monoclonals (Woodhamset al., 1992b) described at least two separate families ofantigens. The OM1 antigen was equally enriched in glyco-proteins eluted from either lentil lectin or wheatgermagglutinin affinity columns and had a molecular weight ofabout 93 kDa, whereas monoclonals OM2, 3, and 4 allrecognised a second antigen with a molecular weight ofabout 36 kDa and a much higher affinity for wheatgermagglutinin than for lentil lectin. However, for reasons that

58 P.L. WOODHAMS AND T. TERASHIMA

remain unclear, no brain samples sent from the SRKcolony in Tokyo to London for analysis showed any specificOM1 staining, regardless of whether they were fromhomozygous mutant animals or normal littermates. Con-trol tissue from the local AS rat colony at the NationalInstitute for Medical Research, London, which was pro-cessed in an apparently identical fashion and stained atthe same time as the Japanese samples, showed thecharacteristic pattern of OM1 immunoreactivity as previ-ously described (Woodhams et al., 1992b). Thus, monoclo-nal antibody OM4 was used in preference to OM1 as amarker for the outer molecular layer and its stainingpattern was compared with that of the more robustcomplementary marker of the inner molecular layer, IM1.

RESULTS

Gross hippocampal morphology in SRK

Previous data on the neuroanatomical malformationspresent in the SRK mutant have been confined to a generaloverview of brain organisation (Aikawa et al., 1988) and adetailed description of the radial mispositioning of cortico-spinal neurons in the sensorimotor cortex (Ikeda andTerashima, 1997). As far as the hippocampus is concerned,Aikawa et al. (1988) merely noted the disruption of thedentate gyrus and ‘‘mixing up’’ of the granule and pyrami-dal cells. Our present findings largely confirm but consider-ably extend these data, with striking differences betweennormal and SRK hippocampi becoming immediately appar-ent when the gross appearance of the two phenotypes iscompared at low magnification (Fig. 1).

Nissl preparations and immunostaining for the calcium-binding protein calbindin showed that the granule cells ofthe SRK dentate gyrus constitute a single cellular massrather than the clearly defined V-shaped cell layer seen innormal cross sections (Fig. 1a,b,e). Under closer examina-tion, the granule cells appear to be more loosely packed inthe centre of the SRK cell mass than at its outer edges (Fig.1e; see also Fig. 2c), although never to the extent offorming the cell-sparse hilar zone that is so readily appar-ent in the normal dentate gyrus (e.g., Figs. 1a, 2f). Adiscrete molecular layer was still present, although itappeared to be reduced in thickness. Presumably as aconsequence of disrupted granule cell migration, the over-all shape of the SRK dentate gyrus in cross section wasusually much more rounded and bulbous than that innormal animals. Calbindin staining also served to clearlydemarcate the mossy fibre projection from granule cells inthe dentate gyrus to the stratum lucidum (Fig. 1a,b).These axons were radially dispersed across the pyramidalcell layer in SRK, concomitant with the abnormal radialposition of the pyramids themselves (Fig. 1d,e). Abnormalpyramidal cell organization was also apparent in fieldCA1, where calbindin immunostaining normally delin-eates a narrow band of small neurons in the upperone-third of the pyramidal cell layer, with radially orien-tated dendrites (Fig. 1a). These pyramidal neurons werescattered and dispersed into irregular clumps in SRK (Fig.1b), with some loss of their radial orientation.

Immunostaining for the large pyramidal cell marker Py(Fig. 1c,d) confirmed the disruption of the organization ofthe large pyramids in field CA3, as suggested by thepattern of calbindin staining of mossy fibres describedabove. In addition, although the Py antigen is present ininterneurons of the hilar region of the dentate gyrus and is

a distinguishing marker between CA3 and CA1 pyramidalneurons (Fig. 1c; Woodhams et al., 1989), comparison ofthe serial sections shown in Figure 1b,d,e indicates thatCA3 pyramids do not extend into and mix with the centralgranule cell mass of the SRK dentate gyrus to any greatextent. The latter finding contradicts the conclusion ofAikawa et al. (1988), based on Nissl staining alone, thatpyramidal and granule cell layers are mixed together inSRK and illustrates the importance of using cell-type-specific markers for cellular identification.

No marked effects of the SRK mutation were seenfollowing staining for the astrocytic marker glial fibrillaryacidic protein; these results were much as expected fromthe generally disrupted hippocampal organization of themutant (data not shown). Reminiscent of the situationdescribed for the reeler mouse (Stanfield and Cowan,1979), astrocytes in the SRK dentate gyrus were morestellate in appearance than were those of controls, whichnormally bear processes having a predominantly radialorganization, particularly where they are interdigitatedamongst the neurons of the granule cell layer. This alteredpattern of glial staining in the mutant correlates with themisorientation of granule cell processes in SRK, as shownby the results of staining for the dendritic marker MAP2(see below).

Distribution of laminar markers in thedentate gyrus

Despite severe disruption to the organisation of granulecells in the SRK dentate gyrus, a relatively sharp borderbetween the outer and inner parts of the molecular layerwas nevertheless clearly recognisable when sections werestained with either of the two laminar antibodies. Semise-rial sections viewed at low magnification showed that thedistribution of IM1 staining of the inner commissural/associational zone (Fig. 2a,c) was complementary to thepattern of high levels of OM4 immunoreactivity in theouter entorhinal field of the molecular layer (Fig. 2b,d).Although published data have indicated that this fine,granular staining represents axonal glycoprotein antigenspresent on the neuronal cell surface (Woodhams et al.,1992a,b; Woodhams and Atkinson, 1996), individual nervefibres could not generally be distinguished, and low levelsof background immunoreactivity were also seen in the‘‘inappropriate’’ terminal fields of the dentate gyrus. Athigher magnification, the straight border between theinner and outer terminal fields of the molecular layer inSRK contrasts with the irregular outline of the outer edgeof the central mass of granule cells, as shown in negativeoutline by background deposits after silver enhancement(Fig. 2e). The IM1-positive inner molecular layer band inSRK usually appeared slightly narrower than in normalrats, occupying about one-fourth of the thickness of themolecular layer in the mutant as opposed to approximatelyone-third in control littermates (cf. Fig. 2c and f). Aspreviously reported (Woodhams et al., 1992b), intense IM1immunoreactivity was also present in the neuropil of thehilus of the dentate gyrus, not only in normal controls (Fig.2f) but also in SRK, although the latter could readily bedistinguished by the presence of many more unstainedneuronal perikarya (Fig. 2d). The general pattern of OMand IM staining was similar in both septal and temporallevels of section, although the molecular layer appeared tobe thinner at the extreme temporal pole of the mutanthippocampus.

HIPPOCAMPAL LAMINATION IN SHAKING RAT KAWASAKI 59

Fig. 1. Low power photomicrographs of adjacent horizontal sections of normal (a,c) and Shaking RatKawasaki (SRK; b,d) temporal hippocampi immunostained for calbindin (a,b) or Py (c,d), followed bysilver enhancement. An adjacent Nissl-stained SRK section is shown in e. DG, dentate gyrus; Pyr,pyramidal cell layer; SL, stratum lucidum. Scale bar 5 500 µm.


Fig. 2. Laminar markers in the dentate molecular layer.a,b: Adjacent serial sections of the Shaking Rat Kawasaki dentategyrus (DG) immunostained (with glucose oxidase-nickel development)for IM1 and OM4, respectively, with details of the complementary IM1and OM4 patterns in c and d. e: Greater detail of the OML/IML

boundary (arrowheads) after silver enhancement of OM4 staining.f: IM1 staining of a section of normal septal hippocampus. G, granulecells; H, hilus; IML, inner molecular layer; OML, outer molecularlayer. Scale bars 5 500 µm in a,b; 100 µm in c–f.


This relatively normal pattern of laminar staining in theSRK mutant appeared despite severe disturbance of notonly the position of most of the granule neurons but alsotheir orientation. Staining for MAP2, a cytoskeletal markerof dendritic processes (Bernhardt et al., 1985; Caceres etal., 1986), shows that, in contrast to the radial orientationof dendrites in the normal granule cell layer (Fig. 3a), thealignment of granule cells in SRK was markedly dis-turbed. Instead of being organised into a dense layer some6–10 cells thick, the SRK dentate gyrus comprised adisorganised mass of granule neurons dispersed across thehilus, many of which bore dendrites that showed a horizon-tal or even inverted orientation in comparison with theirnormal counterparts (Fig. 3b). Evenly interspersed amongstthese displaced granule cells were larger multipolar neu-rons that were strongly MAP2-positive and mostly calbin-din negative, and presumed to be hilar interneurons.

Although high levels of expression of the OM and IMantigens characterise the dentate molecular layer of theentorhinal and commissural/associational terminal fields,respectively, and immunostaining thus serves to distin-guish the latter, these antigens are not unique to themolecular layer of the dentate gyrus, and immunoreactiv-ity is present throughout the CNS, including the rest of thehippocampus proper (Woodhams et al., 1992a,b). However,OM4 staining of the entorhinal terminal field in the

stratum lacunosum-moleculare of hippocampal fieldsCA1–3 was diffuse in both controls and the mutant and didnot provide any useful information about the entorhinalprojection to these areas.

Trajectory of entorhinodentate axons

The general pattern of distribution of axonal fibre tractswas delineated by immunostaining with monoclonal anti-body SMI-31, which recognises a phosphorylated epitopeon the 210-kDa heavy neurofilament polypeptide chain,characteristic of mature axons. This marker does notdistinguish between the different types of fibre innervat-ing the hippocampus, and at septal hippocampal levelsneurofilament staining in SRK provided little informationabout connectivity beyond reflecting the general disrup-tion of tissue morphology and loss of lamination in themutant. However, in cross sections of normal controlhippocampi cut at temporal levels, neurofilament stainingprovided a ready means of following the input from theentorhinal cortex to the dentate gyrus (Fig. 4a). Fasciclesof axons could be traced in the perforant path, traversingthe subiculum, ascending towards the pial surface, andthen turning to course parallel to the hippocampal fissureas a broad band of fibres in the stratum lacunosum-moleculare. At all points where the fissure was not occu-pied by blood vessels, numerous individual axons could be

Fig. 3. a: Microtubule-associated protein 2 staining illustratingthe laminar organisation of granule cells in the normal control dentategyrus. b: Detail of the Shaking Rat Kawasaki dentate gyrus, showingthe aberrant orientation of many granule cell dendrites, some of which

appear horizontal (arrows) rather than radial. G, granule cell layer; H,hilus; HF, hippocampal fissure; ML, molecular layer; Pyr, pyramidalcells. Scale bars 5 100 µm.


distinguished leaving this band and crossing the fissure toinnervate the outer part of the molecular layer of thedentate gyrus (inset to Fig. 4a).

By contrast, in the SRK mutant, such crossing fibreswere rarely encountered, and the hippocampal fissure wasinstead marked by a prominent but narrow fascicle ofaxons confined to the margin of the fissure on the hippocam-pal side (Fig. 4b). For most of the length of the fissure andespecially at its outer (subicular/CA1) end, it was difficultto distinguish any axons crossing from this fascicle into thedentate gyrus (inset to Fig 4b). Below this narrow bundleof axons, the disorganised neuropil in fields CA1–3 of themutant made the stratum lacunosum-moleculare difficultto recognise, and the entorhinal projection to the hippocam-pus proper could not be distinguished with neurofilamentstaining (unlike its counterpart in normal tissue: Fig. 4a).

Calretinin immunoreactivity

Immunostaining for calretinin, a calcium-binding pro-tein present in a variety of neurons throughout the CNS

and hippocampus (Jacobowitz and Winsky, 1991; Resiboisand Rogers, 1992), was used to further delineate thedisrupted laminar organization of the SRK mutant. Differ-ences from normal littermates were particularly notice-able in the dentate gyrus and adjacent stratum lacunosum-moleculare. Although calretinin-positive nonpyramidalneurons were fairly evenly distributed throughout thehippocampus in both control and SRK hippocampi (Fig.5a,b), in the former they showed a preferential localisationclose to the principal cell layers, especially at the innermargin of the dentate granule cell layer (Fig. 5a,c). How-ever, in the mutant dentate gyrus, this preferential local-isation was no longer apparent, and calretinin neuronswere evenly dispersed amongst the granule cells (Fig.5b,d). The prominent supragranular plexus of calretinin-positive axons and perisomatic puncta normally evident atthe inner border of the molecular layer in control litter-mates (Fig. 5a,c arrows; see Nitsch and Leranth, 1996) wasno longer apparent in the SRK mutant, where calretinin

Fig. 4. Neurofilament staining of axons at the hippocampal fissure(HF), temporal level. a: In control rats, ascending bundles of perforantaxons (arrows) coalesce and turn in the stratum lacunosum-moleculare (SLM), from whence they can be easily traced as they cross

the fissure (detailed in inset) into the molecular layer of the dentategyrus (ML). b: Crossing axons are difficult to trace in Shaking RatKawasaki because there is a narrow fascicle of axons at the hippocampalpial surface (inset, small arrows). Scale bars 5 200 µm, insets 5 50 µm.


Fig. 5. Calretinin staining in control (a,c,e) and Shaking RatKawasaki (SRK; b,d,f) hippocampus. a,b: Low magnification view.Small arrows in a indicate the supragranular plexus of the dentategyrus (DG). This is shown in greater detail in c (arrows), taken from anarea at the mouth of the dentate gyrus. d: In the corresponding regionin SRK mutants, the border of the granule cell mass and molecularlayer is irregular (arrows), and immunoreactive fibres and terminals

are diffusely mixed with the granule cells (G). Curved arrows indicatetangentially orientated calretinin-positive neurons in the adjacentstratum lacunosum-moleculare (SLM). A prominent calretinin-positive fascicle of axons (arrowheads in a,b,e,f) is normally situated inthe SLM of hipppocampal field CA2 (a,e) but has an inverted positionat the alveus (A) in SRK (b,f). Scale bars 5 500 µm in a,b; 100 µmin c–f.

fibres and terminals were relatively uniformly dispersedthroughout the granule cell mass (Fig. 5b,d).

The transient population of calretinin-positive Cajal-Retzius cells, which are prominent on either side of thehippocampal fissure in the outer layers of both dentategyrus and hippocampus in neonatal animals (Soriano etal., 1994), was much reduced at the age of P21 when thepresent analysis was carried out, and immunoreactive cellbodies were relatively infrequently seen. However, occa-sional neurons with horizontally orientated dendrites werenoted in the stratum lacunosum-moleculare (but veryrarely in the dentate molecular layer) of both SRK mu-tants and controls, especially towards the mouth of thedentate gyrus (e.g., Fig. 5c,d, curved arrows). Their num-bers were too few at P21 to allow conclusions to be drawnabout any differences between SRK and normal controlanimals.

Dense calretinin immunoreactivity was seen in thestratum lacunosum-moleculare, mostly in the form of finevaricose axons intermingled with the coarser processes(both dendrites and axons) of local calretinin interneurons.Immunoreactivity was typically very fine and granular,and sites of synaptic contact were difficult to identify inthis layer by light microscopy. In coronal sections taken atthe septal pole of the normal hippocampus, these fibrescould often be followed from the alveus across the subicu-lum and field CA1, becoming progressively more denselyconcentrated in the distal parts of the stratum lacunosum-moleculare towards CA3. At more caudal levels in horizon-tal cross-sections of the temporal hippocampus, they formeda striking compact bundle situated at the level of fieldCA2, where the depth of the hippocampal fissure ap-proaches field CA3 (Fig. 5a,e). This fascicle had an almostrectangular profile in cross section and appeared to com-prise fine fibres running longitudinally along the hippocam-pal axis; as elsewhere in the stratum lacunosum-molecu-lare, synaptic terminals were difficult to recognise andmost of the immunoreactivity appeared to be axonal. InSRK this bundle of calretinin-immunoreactive axons wasspread diffusely in a radial direction across the stratumradiatum, coalescing into an ectopically displaced irregu-lar bundle at the opposite, alvear surface of field CA2 (Fig.5b,f). Thus, this calretinin-immunoreactive fibre projec-tion seems to substantially inverted in position in the SRKmutant.

Lamination in the SRK CA3pyramidal cell layer

The principal cell layer of hippocampal field CA3 innormal rats comprises large pyramidal neurons aligned ina single row some 3–5 cells thick, with radially orientatedprincipal dendrites that receive a granule cell mossy fibreinput to their most proximal segments; this latter zoneconstitutes the stratum lucidum (Figs. 1c, 6b). In SRKanimals, as indicated in the general description of hippo-campal organization (Fig. 1d), CA3 pyramidal neuronswere misorientated and often ectopically positioned. Theywere frequently dispersed towards the pial surface, form-ing an irregular broad band rather the usual uniform row,and many showed oblique, horizontal, or even radiallyinverted dendritic orientations (Fig. 6a). No quantitativeanalysis of their number was performed in the presentstudy, but they were clearly more loosely packed in SRKthan in normal littermates. This disorganisation of theSRK pyramidal cell layer in CA3 was reflected in the

pattern of staining for its major afferent input from thedentate gyrus. In contrast to the well-demarcated band ofcalbindin-immunoreactive mossy fibre axons characteris-tic of the normal stratum lucidum (Fig. 6d), in SRK thislayer had an uneven shape, with many fibres interdigitat-ing between the CA3 pyramids to form an irregularinfrapyramidal layer (Fig. 6c).

Comparison of SRK malformationswith the mouse reeler mutant

Although a systematic and detailed comparison of SRKwith the murine reeler mutant phenotype has not beenattempted in the present study, they do show strikingsimilarities which suggest that similar disruptions ofdevelopmental mechanisms may underly the observedhippocampal abnormalities. The most noticeable commonfeature of both mutants was the presence of a centralclumped mass of dentate granule cells rather than theV-shaped profile seen in cross sections of the normalgranule cell layer. A relatively cell-free molecular layerwas nevertheless still present, and like SRK, sections ofthe reeler dentate gyrus retained the normal pattern ofexpression of the only one of our laminar markers to stainmouse brain, IM1 (Fig. 7d,e). This antigen was highlyexpressed not only in the neuropil between the granulecells but also in a characteristic narrow but distinct bandat the innermost part of the molecular layer. Both mutantsalso had an abnormal mossy fibre projection to CA3,reflecting the disrupted position of the large pyramidalcells. However, calbindin staining indicated an extensiveectopic infrapyramidal component in the case of SRK,whereas the mossy fibre layer was disrupted into irregularclumps but the granule cell axons were still predominantlysuprapyramidal in reeler, suggesting relatively little misori-entation of reeler CA3 pyramidal dendrites (data notshown).

Calbindin staining showed that some other features ofhippocampal disorganisation were clearly not equivalentin the two mutants. Pyramidal neurons of field CA1 inSRK were diffusely spread and disorientated, with thenormal compact pyramidal cell layer being broken up intoscattered small clumps (Figs. 1a–b, 7a). By contrast,MAP2 staining of the reeler showed this layer to be splitinto two major components, the uppermost of which largelyretained the normal radial disposition of the pyramidalapical dendrites (Fig. 7b,c). The latter finding confirmsprevious descriptions of the reeler phenotype publishedelsewhere (Stanfield and Cowan, 1979; Del Rıo et al.,1997).

DISCUSSION

The principal finding of the present study is that,despite aberrant cell migration and severe disruption of allcell layers in the SRK mutant hippocampus extending tothe apparent inversion of some fibre projections, themutant dentate gyrus develops with not only a clearlydefined molecular layer but one that exhibits a sharplaminar boundary between the inner and outer terminalfields, as defined by our layer-specific molecular markers.This raises a number of issues of connectivity that clearlymerit further investigation by fibre-tracing techniquesoutside the scope of the present report (the limited viabil-ity of SRK mutants suggests that such experiments willnot be easy to perform). The primary question at issue is:


Fig. 6. Disorganised laminar distribution of CA3 pyramidal neu-rons in Shaking Rat Kawasaki (SRK; a,c) versus that of control rats(b,d). a,b: Staining with the monoclonal antibody Py shows thepyramids to be radially dispersed and misorientated in SRK; some arebeing completely inverted (arrows). c,d: Calbindin staining for granule

cell mossy afferents shows a corresponding radially spread stratumlucidum (SL) in SRK, with a substantial number of the mossy fibresoccupying an ectopic infrapyramidal position (arrowheads in c). Pyr,pyramidal cell layer. Scale bars 5 100 µm.

Fig. 7. Comparison of Shaking Rat Kawasaki (SRK) malforma-tions with the reeler mouse mutation. Microtubule-associated protein2 staining shows a radial dispersion of CA1 pyramids in SRK (a),which contrasts with the bilaminar CA1 pyramidal cell layer seen inreeler (b, arrows). c: Field CA1 of the normal mouse contains a singlenarrow layer of radially orientated small pyramidal neurons. The

reeler dentate gyrus (d) shows a narrow band of IM1 staining in theinner molecular layer (IML) comparable to that present in controls (e)and is very similar to the IM1 pattern characteristic of SRK (cf. Fig. 2).A, alveus; G, granule cells; HF, hippocampal fissure; OML, outermolecular layer. Scale bars 5 100 µm.

Why does there appear to be an almost normal andrelatively wide entorhinal zone in the outer molecularlayer in the SRK mutant? A relatively straight boundarybetween it and the inner zone was still present, despiteectopic positioning and orientation of the target granulecell dendrites; a similar situation was evident in thesamples of reeler tissue that we examined. Although thelimited amount of published data suggests that there is areduced entorhinal projection in the reeler hippocampus atP5 (Del Rıo et al., 1997; Fig. 4j). the results of anterogradetracing with Phaseolus vulgaris–leucoagglutinin indicatea sharply defined entorhinal zone in the reeler dentatemolecular layer that is similar to that of normal controls(Deller et al., 1997). Thus, these fibre-tracing data are ingood agreement with our present findings. By way ofexplanation, it has been suggested (Frotscher, 1997, 1998)that entorhinal axons find their way to the hippocampusby using the early axons of Cajal-Retzius cells as guides,and preliminary tracer studies have indicated that such anearly hippocampo-entorhinal projection is indeed present(Ceranik et al., 1998). This constitutes a reelin-indepen-dent mechanism that may be relatively unaffected by thereeler mutation, and it is likely that a similar situation alsoholds true in SRK.

The results of our IM1 staining in the mouse contradictprevious data from Timm staining of the reeler dentategyrus (Stanfield and Cowan, 1979), which indicated achange from a trilaminar to a bilaminar appearance of themolecular layer, with apparent loss of Timm staining inthe innermost (commissural-associational) band. How-ever, in that study, details of the immediate supragranularregion were not clearly shown in the reeler dentate gyrus(their Fig. 12), and this issue will only be resolved byappropriate fibre-tracing experiments. Data published thusfar (Deller et al., 1997) have indicated that the commis-sural projection to the dentate gyrus is more diffuselyspread than the entorhinal field in the reeler mutant.However, it is noteworthy that in Timm preparations thesharp boundary between the outermost lateral entorhinalterminal field and the adjacent medial entorhinal terminalfield was preserved in the molecular layer of the mutantdentate gyrus, providing further evidence that, as in SRK,reeler granule cell dendrites retain major aspects of theirnormal lamination despite abnormal granule cell position-ing.

The foregoing findings make it relatively unlikely thatthe expression of our molecular markers fails to ad-equately reflect the disposition of hippocampal versusentorhinal axons, i.e., that the observed staining patternsdo not correlate with actual patterns of connectivity.Moreover, although changes in antigen expression unre-lated to alterations in fibre pathways cannot be discountedin either mutant, all the evidence available thus far pointsto the OM antigens being expressed on entorhinal axons.Thus, OM staining in the outer molecular layer is abol-ished by entorhinal lesions and restored by grafts ofembryonic entorhinal tissue (Woodhams et al., 1992a), andtissue culture experiments have clearly demonstratedOM-positive fibres growing into the entorhinal terminalfield in vitro (Woodhams et al., 1993; Woodhams andAtkinson, 1996). The results of ongoing studies on thegenetic basis of the SRK mutation and the molecularcharacterisation of the OM and IM antigens may beextremely valuable in settling this question. Althoughearly contacts with pioneer neurons suggest one mecha-

nism whereby entorhinal afferents are targetted to theirappropriate terminal field in the hippocampus (Super etal., 1998), the molecular basis of this interaction remainsuncharacterised; in vitro studies have indicated thatlamina-specific adhesive cues for entorhinal cells are pres-ent in the hippocampus, which do not require the presenceof reelin, neural cell adhesion molecule, or divalent cations(Forster et al., 1998).

Although the reeler mutant shows some hippocampalabnormalities that are different from those of SRK (e.g.,the splitting rather than clumping of the pyramidal celllayer in field CA1), many anatomical features are strik-ingly similar. Whilst available data are limited, silverstaining appears to show a narrow plexus of axons in thestratum lacunosum-moleculare of reeler (Stanfield andCowan, 1979), and immunocytochemistry for the F3/F11cell adhesion molecule shows a prominent axon bundle atthe hippocampal fissure (Ishida et al., 1994). These tractsmay represent the equivalent of the dense fascicle ofneurofilament-positive axons that we observed in thisposition in SRK. Therefore, the possibility arises that inboth mutants, rather than directly crossing the hippocam-pal fissure as in the wild-type strain, the entorhinalperforant projection reaches the molecular layer of thedentate gyrus by the alternative route of coursing alongthe stratum lacunosum-moleculare for the whole length ofthe fissure and curving around its deepest end in field CA3into the dentate gyrus. This would be consistent with theobservation of straight and unbranched entorhinohippo-campal fibres in reeler (Del Rıo et al., 1997). A detailedanalysis of the entorhinodentate projection in SRK iscurrently underway by tracing these axons with crystals ofcarbocyanine dye placed into the entorhinal cortex(Woodhams and Terashima, in preparation). Preliminarydata indicate that, unlike in normal rats, in SRK dye-labelled fibres crossing the fissure are almost entirelyabsent from the proximal regions of the subiculum andfield CA1, despite intense labelling of the projection to thestratum lacunosum moleculare. Labelled axons are, how-ever, plentiful at its depth in field CA3, where they can betraced coursing round into the molecular layer of thedentate gyrus.

The existence of aberrant axon trajectories in reeler hasbeen most clearly shown for the retinotectal projection(Frost et al., 1986), where in the rl/rl superior colliculusretinal axons spread beyond their usual path in thestratum opticum up to the overlying superficial gray, fromwhich they are normally absent. Despite this abnormalcourse, they nevertheless mostly arborise and terminate inthe appropriate laminae of the superior colliculus. Ageneral ability of reeler axons to reach their appropriatetargets has also been noted for olfactory (Devor et al.,1975), olivocerebellar (Blatt and Eisenman, 1988), thalamo-cortical (Yuasa et al., 1994; Molnar and Blakemore, 1995),and corticothalamic (Terashima et al., 1987) pathways,although minor aberrations in the finer details of theseprojections may be present (Devor et al., 1975; Wilson etal., 1981; Blatt and Eisenman, 1988). Corticospinal axonsarising from malpositioned sensorimotor neurons appearto develop both normal trajectories and collateral branch-ing patterns (Terashima, 1995). These data suggest that,despite severe disturbances of cellular positioning in thereeler mutant, powerful adaptive mechanisms are presentthat enable axons to reach and then arborise and termi-nate on their appropriate targets. It remains to be seen


whether or not this is also the case for fibre pathways inthe SRK rat.

It is somewhat surprising that the dense longitudinalbundle of calretinin-immunoreactive axons seen in thestratum lacunosum-moleculare of field CA2, which isdisplaced to an alvear position in SRK, has hitherto notbeen noted. This bundle was also evident in horizontalsections of normal temporal hippocampus in the mouse(data not shown). However, a similar fascicle can bediscerned in some descriptions of calretinin staining (e.g.,Nitsch and Leranth, 1996; Fig. 3c,e) and it may have beenmissed in other studies due to examination in a sub-optimal plane of section. Alternatively, there may be realdifferences in calretinin expression between rat strains, ashave been reported to exist for other calcium-bindingproteins in projections such as calbindin in the mossy fibrepathway (Celio, 1990). The calretinin bundle appears torepresent the projection from the nucleus reuniens thalami(NRT) to the stratum lacunosum-moleculare, which runsforwards from cells in the NRT (Resibois and Rogers,1992), through the rostral striatum, around the genu of thecorpus callosum, and back caudally in the cingulum bundle(Wouterlood et al., 1990). Following this fibre projection byimmunochemistry alone is made difficult by the presenceof many scattered calretinin-immunoreactive fibres inadjacent structures, and it will be necessary to performdouble-labelling studies combining immunocytochemistrywith antero- or retrograde axonal tracers to confirm itsprecise disposition.

Our understanding of the nature of the reeler mutationhas been greatly advanced after cloning of the gene forreelin (D’Arcangelo et al., 1995), a large extracellularmatrix glycoprotein that is secreted by the transientCajal-Retzius neurons present in the marginal zone of thedeveloping cortex (D’Arcangelo et al., 1997; Drakew et al.,1998). Studies in vivo (Ogawa et al., 1995; Nakajima et al.,1997) and in vitro (Del Rıo et al., 1997) have clearlyimplicated these cells as crucial players in setting upnormal cortical architectonics (for reviews, see Frotscher,1997, 1998). It remains to be determined to what degreethe malformations described in the present study in theSRK mutant relate to these findings and to similar abnor-malities seen in mutations such as the scrambler (Gonza-lez et al., 1997; Howell et al., 1997; Sheldon et al., 1997)and yotari (Sheldon et al., 1997; Yoneshima et al., 1997)mice. Work on characterizing the genetic basis of the SRKmutation is still in progress, but preliminary data indicatethat reelin mRNA expression is markedly reduced but notcompletely absent in SRK, and Southern analysis of SRKgenomic DNA has failed to show any large deletions in thereelin gene (Kikkawa et al., 1998). Taken together, thesefindings imply that the mutation in SRK may be upstreamof reelin expression. The present morphological results,which show strong similarities with but not an identicalphenotype to the reeler, suggest that the SRK mutant maybe a fruitful subject for further analysis in experimentsaimed at understanding the developmental mechanismsunderlying the specification of cortical connectivity.

ACKNOWLEDGMENTS

We are very grateful to Dr. Geoffrey Raisman, Prof.Michael Frotscher, and Dr. Thomas Deller for their con-structive comments on the manuscript.

LITERATURE CITED

Aikawa H, Nonaka I, Woo M, Tsugani T, Esaki K. 1988. Shaking ratKawasaki (SRK); a new neurological mutant in the Wistar strain. ActaNeuropathol 76:366–372.

Baimbridge KG, Celio MR, Rogers JH. 1992. Calcium-binding proteins inthe central nervous system. Trends Neurosci 15:303–308.

Bernhardt R, Huber G, Matus A. 1985. Differences in the developmentalpatterns of three microtubule-associated proteins in the rat cerebellum.J Neurosci 5:977–991.

Blatt GJ, Eisenman LM. 1988. Topographic and zonal organization of theolicocerebellar projection in normal and reeler mice. J Comp Neurol267:603–615.

Caceres A, Banker GA, Binder L. 1986. Immunocytochemical localization oftubulin and microtubule-associated protein 2 during the developmentof hippocampal neurons in culture. J Neurosci 6:714–722.

Celio MR. 1990. Calbindin D-28k and parvalbumin in the rat nervoussystem. Neuroscience 35:375–475.

Ceranik K, Deng J, Heimrich B, Zhao S, Lubke J, Frotscher M. 1998.Neurons in the outer molecular layer of the rat dentate gyrus form anearly hippocampal–entorhinal projection. Soc Neurosci Abstr 24:60.

D’Arcangelo G, Miao GG, Chen S-G, Soares HD, Morgan JI, Curran T. 1995.A protein related to extracellular matrix proteins deleted in the mousemutant reeler. Nature 374:719–723.

D’Arcangelo G, Nakajima K, Miyata T, Ogawa M, Mikoshiba K, Curran T.1997. Reelin is a secreted glycoprotein recognized by the CR-50monoclonal antibody. J Neurosci 17:23–31.

Deller T, Drakew A, Heimrich B, Frotscher M. 1997. The fascia dentata ofthe reeler mutant: anterograde tracing of afferents with PhAL. SocNeurosci Abstr 23:2103.

Del Rıo JA, Heimrich B, Borrell V, Forster E, Drakew A, Alcantara S,Nakajima K, Miyata T, Ogawa M, Mikoshiba K, Derer P, Soriano E.1997. A role for Cajal-Retzius cells and reelin in the development ofhippocampal connections. Nature 385:70–74.

Devor M, Caviness Jr. VS, Derer P. 1975. A normally laminated afferentprojection to an abnormally laminated cortex: some olfactory connec-tions in the reeler mouse. J Comp Neurol 164:471–482.

Drakew A, Frotscher M, Deller T, Ogawa M, Heimrich B. 1998. Developmen-tal distribution of a reeler gene-related antigen in the rat hippocampalformation visualized by CR-50 immunocytochemistry. Neuroscience82:1079–1086.

Forster E, Kaltschmidt C, Deng J, Cremer H, Deller T, Frotscher M. 1998.Lamina-specific cell adhesion on living slices of hippocampus. Develop-ment 125:3399–3410.

Frost DO, Edwards MA, Sachs GM, Caviness Jr. VS. 1986. Retinotectalprojection in reeler mutant mice: relationships among axon trajectories,arborization patterns and cytoarchitecture. Dev Brain Res 28:109–120.

Frotscher M. 1997. Dual role of Cajal-Retzius cells and reelin in corticaldevelopment. Cell Tissue Res 290:315–322.

Frotscher M. 1998. Cajal-Retzius cells, Reelin, and the formation of layers.Curr Opin Neurobiol 8:570–575.

Gonzalez JL, Russo CJ, Goldowitz D, Sweet HO, Davisson MT, Walsh CA.1997. Birthdate and cell marker analysis of scrambler: a novel mutationaffecting cortical development with a reeler-like phenotype. J Neurosci17:9204–9211.

Howell BW, Hawkes R, Soriano P, Cooper JA. 1997. Neuronal positioning inthe developing brain is regulated by mouse disabled-1. Nature 389:733–737.

Ikeda Y, Terashima T. 1997. Corticospinal tract neurons are radiallymalpositioned in the sensory-motor cortex of the Shaking Rat Ka-wasaki. J Comp Neurol 382:1–11.

Ishida A, Shimazaki K, Terashima T, Kawai N. 1994. An electrophysiologi-cal and immunohistochemical study of the hippocampus of the reelermutant mouse. Brain Res 662:60–68.

Jacobowitz DM, Winsky L. 1991. Immunocytochemical localization ofcalretinin in the forebrain of the rat. J Comp Neurol 304:198–218.

Kikkawa S, Ikeda Y, Okado H, Ogawa M, Terashima T. 1998. Reducedexpression of reelin in Shaking Rat Kawasaki (SRK), a novel mutantrat with a reeler-like phenotype. Neurosci Res 22(suppl):S304.

Molnar Z, Blakemore C. 1995. How do thalamic axons find their way to thecortex? Trends Neurosci 18:389–397.

Nakajima K, Mikoshiba K, Miyata T, Kudo C, Ogawa M. 1997. Disruptionof hippocampal development in vivo by CR-50 mAb against Reelin. ProcNatl Acad Sci USA 94:8196–8201.


Nitsch R, Leranth C. 1996. GABAergic neurons in the rat dentate gyrus areinnervated by subcortical calretinin-containing afferents. J Comp Neu-rol 364:425–438.

Ogawa M, Miyata T, Nakajima K, Yagyu K, Seike M, Ikenaka K, YamamotoH, Mikoshiba K. 1995. The reeler gene-associated antigen on Cajal-Retzius neurons is a crucial molecule for laminar organization ofcortical neurons. Neuron 14:899–912.

Resibois A, Rogers JH. 1992. Calretinin in rat brain: an immunohistochemi-cal study. Neuroscience 46:101–134.

Sheldon M, Rice DS, D’Arcangelo G, Yoneshima H, Nakajima K, MikoshibaK, Howell BW, Cooper JA, Goldowitz D, Curran T. 1997. Scrambler andyotari disrupt the disabled gene and produce a reeler-like phenotype inmice. Nature 389:730–733.

Shu S, Ju G, Fan L. 1988. The glucose oxidase-DAB-nickel method inperoxidase histochemistry of the nervous system. Neurosci Lett 85:169–171.

Soriano E, Del Rio JA, Martınez A, Super H. 1994. Organization of theembryonic and early postnatal murine hippocampus. I. Immunocyto-chemical characterization of neuronal populations in the subplate andmarginal zone. J Comp Neurol 342:571–595.

Stanfield BB, Cowan WM. 1979. The morphology of the hippocampus anddentate gyrus in normal and reeler mice. J Comp Neurol 185:393–422.

Super H, Martinez A, Del Rio JA, Soriano E. 1998. Involvement of distinctpioneer neurons in the formation of layer-specific connections in thehippocampus. J Neurosci 18:4616–4626.

Terashima T. 1995. Course and collaterals of corticospinal fibres arisingfrom the sensorimotor cortex of the reeler mouse. Dev Neurosci 17:8–19.

Terashima T, Inoue K, Inuoe Y, Mikoshiba K. 1987. Thalamic connectivityof the primary motor cortex of normal and reeler mutant mice. J CompNeurol 257:405–421.

Wilson L, Sotelo C, Caviness Jr. VS. 1981. Heterologous synapses uponPurkinje cells in the cerebellum of the reeler mutant mouse: anexperimental and light microscopic study. Brain Res 213:63–82.

Woodhams PL, Atkinson DJ. 1996. Regeneration of entorhino-dentateprojections in organotypic slices cultures: mode of axonal regrowth andeffects of growth factors. Exp Neurol 140:68–78.

Woodhams PL, Webb M, Atkinson DJ, Seeley PJ. 1989. A monoclonalantibody, Py, distinguishes different classes of hippocampal neurons. JNeurosci 9:2170–2181.

Woodhams PL, Kawano H, Raisman G. 1992a. The OM series of terminalfield-specific monoclonal antibodies demonstrate reinnervation of theadult rat dentate gyrus by embryonic entorhinal transplants. Neurosci-ence 46:71–82.

Woodhams PL, Kawano H, Seeley PJ, Atkinson DJ, Field PM, Webb M.1992b. Monoclonal antibodies reveal molecular differences betweenterminal fields in the rat dentate gyrus. Neuroscience 46:57–69.

Woodhams PL, Atkinson DJ, Raisman G. 1993. Rapid decline in the abilityof entorhinal axons to innervate the dentate gyrus with increasing timein organotypic culture. Eur J Neurosci 5:1596–1609.

Wouterlood FG, Saldana E, Witter MP. 1990. Projection from the nucleusreuniens thalami to the hippocampal region: light and elctron micro-scopic tracing study in the rat with the anterograde tracer Phaseolusvulgaris–leucoagglutinin. J Comp Neurol 296:179–203.

Yoneshima H, Nagata E, Matsumoto M, Yamada M, Nakajima K, Miyata T,Ogawa M, Mikoshiba K. 1997. A novel neurological mutant mouse,yotari, which exhibits reeler-like phenotype but expresses CR-50 antigen/reelin. Neurosci Res 29:217–223.

Yuasa S, Kitoh J, Kawamura K. 1994. Interactions between growingthalamocortical afferent axons and the neocortical primordium innormal and reeler mutant mice. Anat Embryol 190:137–154.


laminar boundaries persist in the hippocampal dentate molecular layer of the mutantshaking rat...

Documents