Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

ORIGINAL ARTICLE

Patterned Neuropathologic Events Occurring in hyh CongenitalHydrocephalic Mutant Mice

Patricia Paez, PhD, Luis-Federico Batiz, MD, MSc, Ruth Roales-Bujan, Luis-Manuel Rodrıguez-Perez,

Sara Rodrıguez, PhD, Antonio Jesus Jimenez, PhD, Esteban Martın Rodrıguez, MD, PhD, and

Jose Manuel Perez-Fıgares, PhD

AbstractHyh mutant mice develop long-lasting hydrocephalus and

represent a good model for investigating neuropathologic events

associated with hydrocephalus. The study of their brains by use of

lectin binding, bromodeoxyuridine labeling, immunochemistry, and

scanning electron microscopy revealed that certain events related to

hydrocephalus followed a well-defined pattern. A program of

neuroepithelium/ependyma denudation was initiated at embryonic

day 12 and terminated at the end of the second postnatal week.

After the third postnatal week the denuded areas remained

permanently devoid of ependyma. In contrast, a selective group of

ependymal areas resisted denudation throughout the lifespan.

Ependymal denudation triggered neighboring astrocytes to prolif-

erate. These astrocytes expressed particular glial markers and

formed a superficial cell layer replacing the lost ependyma. The

loss of the neuroepithelium/ependyma layer at specific regions of

the ventricular walls and at specific stages of brain development

would explain the fact that only certain brain structures had

abnormal development. Therefore, commissural axons forming the

corpus callosum and the hippocampal commissure displayed

abnormalities, whereas those forming the anterior and posterior

commissures did not; and the brain cortex was not homogenously

affected, with the cingular and frontal cortices being the most

altered regions. All of these telencephalic alterations developed at

stages when hydrocephalus was not yet patent at the lateral

ventricles, indicating that abnormal neural development and hydro-

cephalus are linked at the etiologic level, rather than the former

being a consequence of the latter. All evidence collected on

hydrocephalic hyh mutant mice indicates that a primary alteration

in the neuroepithelium/ependyma cell lineage triggers both hydro-

cephalus and abnormalities in telencephalic development.

Key Words: Astroglial response, Corpus callosum agenesis,

Cortical disorders, hyh mutant, Inherited hydrocephalus, Neuro-

degenerative process, Neuroepithelium/ependyma defect.

INTRODUCTIONHydrocephalus can be originated by a variety of

developmental abnormalities and has been frequently asso-ciated with other neurologic disorders (1). In humans,alterations of certain brain regions such as the cerebralcortex and corpus callosum and ependymal denudation havebeen previously considered as consequences of fetal onsethydrocephalus (2). Therefore, it has been suggested thatincreased intracranial pressure could stretch, destroy orinterfere with development of corpus callosum and cause areduction of gray matter (2, 3).

The hyh mutant mouse (hydrocephalus with hop gait[4]) shows certain attributes that make it an appropriateanimal model for congenital hydrocephalus. Certain aspects,such as time of onset, type of abnormality of cerebrospinalfluid (CSF) dynamics, clinical evolution, and survival/deathrate (5Y8), are similar to those found in several types ofhuman congenital hydrocephalus. Several of the neuro-pathologic events traditionally related to hydrocephalus havebeen observed in hyh mutant mice, including ependymaldenudation (5Y7), astrocytosis (7), cortical alterations (9),agenesis of corpus callosum (4), axonal degeneration, brainedemas, and ventriculostomies (8). Considering that in thehyh mutant a key event is the loss of the neuroepithelium/ependyma during fetal life and early postnatal period (6), inthe present investigation we aimed to analyze both the originof these neuropathologic alterations and their direct orindirect relationship with ependymal denudation and/orhydrocephalus.

MATERIALS AND METHODS

AnimalsThe hyh mutant mice (hydrocephalus with hop gait)

(B6C3Fe-a/a-hyh/J) were obtained from The Jackson Labo-ratory (Bar Harbor, ME) and bred into 2 colonies (Uni-versidad de Malaga and Universidad Austral de Chile).Housing, handling, care, and processing of animals were

1J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007

J Neuropathol Exp NeurolCopyright � 2007 by the American Association of Neuropathologists, Inc.

Vol. 66, No. 12December 2007

pp. 1Y11

From the Departamento de Biologıa Celular (PP, RR-B, L-MR-P, AJJ,JMP-F), Genetica y Fisiologıa, Facultad de Ciencias, Universidad deMalaga, Malaga, Spain; and Instituto de Anatomıa (L-FB, SR, EMR),Histologıa y Patologıa, Facultad de Medicina, Universidad Austral deChile, Valdivia, Chile.

Send correspondence and reprint requests to: Antonio J. Jimenez, PhD,Departamento de Biologıa Celular, Genetica y Fisiologıa, Universidadde Malaga, Campus Universitario de Teatinos, E-29071-Malaga, Spain;E-mail: ajjimenez@uma.es

Drs. Paez and Batize contributed equally to this work.This work was by grants from Fondo de Investigaciones Sanitarias

(PI 030756 and PI 060243), Instituto de Salud Carlos III, and ServicioAndaluz de Salud, Spain, to J.M.P.-F., from Fondo Nacional deDesarrollo Cientıfico y Tecnologico, Chile (1030265 and 1070241),to E.M.R., and from Universidad Austral de Chile (DID D-2005-12)to L.F.B.

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

carried out in accordance with European and Spanish laws(DC 86/609/CEE and RD 1201/2005) and following theregulations approved by the Council of the AmericanPhysiologic Society. Embryonic day (E) 0.5 (E0.5) wasdesignated when the presence of a vaginal plug was evidentin the heterozygous female.

Investigations identifying the gene (>-SNAP) mutatedin hyh mice were published in 2004 (9, 10).AQ1 As the presentstudy started in 2002, the identification of the mutant micebefore 2004 was performed on clinical and neuropathologicgrounds (5Y8). The definitive diagnosis was made by micro-scopic analysis of brain tissue. Since 2004, the hydro-cephalic (hy) mutant mice, particularly embryos, have beenidentified by genotyping. The wild-type mice were designatedas non-hydrocephalic (non-hy). Embryos (E11.5YE17.5)were obtained from anesthetized pregnant dams.

Sampling for ImmunocytochemistryThe head of fetuses from E11.5 to E17.5 and the brain

of mice from postnatal day (PN) 1 to PN150 were processedfor conventional light microscopy, immunocytochemistry,and lectin binding as described previously (4Y6). Samplingwas as follows: i) E11.5, 3 hy and 3 non-hy; ii) E12.5,6 hy and 3 non-hy; iii) E13.5, 4 hy and 3 non-hy; iv) E14.5,5 hy and 3 non-hy; v) E15.5, 6 hy and 3 non-hy; vi) E16.5, 5hy and 4 non-hy; vii) E17.5, 6 hy and 3 non-hy; viii) firstpostnatal week, 26 hy and 13 non-hy; ix) second postnatalweek, 28 hy and 13 non-hy; x) third to fourth postnatalweek, 14 hy and 9 non-hy; xi) second month, 10 hy and 6non-hy; and xii) fourth to fifth month, 9 hy and 5 non-hy.

1,1-Dioctadecyl-3,3,3’,3’-Tetramethylindocarbocyanine PerchlorateTracing of Axons of Corpus Callosum and

Hippocampal CommissureThe growing of callosal axons toward the midline

during brain development was examined with 1,1-dioctadecyl-3,3,3¶,3¶-tetramethylindocarbocyanine perchlo-rate (DiI) tracing. The brains of 8 hy and 5 non-hy micefrom E15.5 to PN2 were fixed with 4% paraformaldehyde in0.1 M phosphate buffer for 24 hours at 4-C. DiI (largecrystals, D-3911; Invitrogen, Molecular Probes, Carlsbad,CA) was placed at the frontal neocortex of non-hy and hyanimals. Brains were immersed in fresh fixative solution andkept at 37-C for 3 weeks. Vibratome 150- to 300-Kmsections were cut consecutively and stored at 4-C.

Immunocytochemistry and Lectin BindingSections were processed for immunocytochemistry

and lectin binding as described previously (5Y7). Thefollowing primary antibodies were used: i) anti-glial fibrillaryacidic protein (GFAP) (astrocytes), raised in rabbit (Bio-genesis, Oxford, UK), 1:250 dilution; ii) anti-neurofilament(polypeptide 220 kDa; nerve fibers) raised in rabbit (labo-ratory of E. M. Rodrıguez) 1:1000 dilution; iii) anti-neuralcell adhesion molecules (nerve fibers), monoclonal, whichrecognizes polysialylated and nonsialylated forms of neuralcell adhesion molecules (5B8 clone; Developmental StudiesHybridoma Bank, Iowa City, IA), supernatant used undiluted;

iv) anti-proliferating cell nuclear antigen (cell proliferation),monoclonal (Sigma, St. Louis, MO), 1:200 dilution; v) anti-S100A protein (ependymal cells), raised in rabbit (Bio-genesis), 1:2 dilution; vi) anti-vimentin (ependymal cells),raised in goat (Sigma), 1:500 dilution; vii) anti-human myelinbasic protein (myelinated fibers) raised in rabbits (Dako,Glostrup, Denmark), 1:200 dilution; and viii) anti-LN5(Sigma), monoclonal, raised in mouse, which recognizesmacrophages, 1:400 dilution. Lycopersicon esculentum lectin(tomato lectin; affinity = N-acetylglucosamine, N-acetylga-lactosamine) was used to label microglia/macrophages.

Bromodeoxyuridine LabelingBromodeoxyuridine (BrdU) (Sigma) was used to label

nuclei of cells undergoing proliferation in denuding walls.Two different experiments were performed: i) 2 non-hy and2 hy mice were injected i.p. with BrdU at PN4 and killed 90minutes later; ii) 6 non-hy and 6 hy mice were injected i.p.with 3 daily doses of BrdU (100 mg/kg) for 3 consecutivedays (PN4YPN6) and killed at PN6, 3 hours after the lastinjection. For the first experiment, animals were perfusedwith 4% buffered paraformaldehyde and cryoprotected in30% sucrose to obtain frozen sections for immunofluores-cence. Animals from the second experiment were processedto obtain paraffin sections for immunocytochemistry andimmunofluorescence.

For immunocytochemistry, sections were sequentiallyincubated with i) a monoclonal anti-BrdU (G3G4; Devel-opmental Studies Hybridoma Bank) at a 1:50 dilution for 18hours; ii) a biotinylated secondary antibody (Vector Labo-ratories, Burlingame, CA) for 30 minutes; and iii) VectastainElite ABC reagent (Vector Laboratories) for 30 minutes.Dilutions were made according to the manufacturers_instructions. The reaction product was detected using DAB.

To examine whether proliferative cells expressedastroglial antigens corresponding to astrocytes born duringthe BrdU labeling period, double immunofluorescencestaining for BrdU and GFAP was performed. Sections wereincubated with i) anti-BrdU (see above) and ii) rabbit anti-GFAP (1:300 dilution, Sigma; or 1:250 dilution, Biogenesis,4650 AQ2) for 18 hours. Alexa Fluor 594 or Alexa Fluor 568goat anti-mouse IgG (1:500; Molecular Probes) was used asa fluorescent linking antibody for anti-BrdU. Alexa Fluor488 goat anti-rabbit IgG (1:500; Molecular Probes) was usedas a linking antibody to detect GFAP. Sections were rinsedand mounted in fluorescent mounting medium (S3023;Dako). For immunofluorescence all antibodies were dilutedas described previously (5Y7). Confocal microscopy wascarried out with FluoView FV1000 Olympus and Leica TCSNT confocal laser scanning microscopes. In some samples,projections in Z axis from a stack of consecutive confocalimages were obtained using ImageJ (public domain Javaimage-processing program). Images of negative controlsections were taken with settings identical to those used toimage experimental sections.

Scanning Electron MicroscopyThe brains of 32 hy and 24 non-hy mice (from PN1 to

PN135) were processed as described previously (5Y8).

Paez et al J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007

� 2007 American Association of Neuropathologists, Inc.2

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

Data AnalysisTo analyze the alteration in the thickness of the brain

cortex, 3 groups of animals were analyzed: i) non-hyanimals; ii) mice with slowly progressive or severe compen-sated hydrocephalus (SCH); and iii) animals with rapidlyprogressive or very severe hydrocephalus (VSH) (8). For thispurpose, coronal sections from 3 different regions of thebrain were used at the following levels: i) of the rostral hornof the lateral ventricles (1.14 mm caudal to bregma) and ii)of the medial lateral ventricle (0.14 mm caudal to bregma);sections at these 2 levels were used to analyze the cingular,frontal and parietal cortices; and iii) at the caudal horn ofthe lateral ventricles (j2.04 mm of bregma); these sectionswere used for the analysis of the occipital cortex. Samplingwas as follows: i) PN3, 6 non-hy and 6 hy; ii) PN8, 4 non-hy, 5 SCH, and 5 VSH; iii) PN30, 4 non-hy, 5 SCH, and 5

VSH; and iv) PN120, 4 non-hy, 3 SCH, and 3 VSH.Sections (10 Km) were photographed with a digital imagecapturer, and measurement of cortical thickness was per-formed with Visilog 5.0 software. The mean cortical thick-ness for each cortical region and each age interval wasobtained from 18 to 24 measurements. Statistical analysiswas performed with Microsoft Excel XP software and SigmaStat 2.0 software.

RESULTS

Postnatal Program of Ependymal Denudation inHy Mice

A programmed ependymal denudation was shown tooccur in hy embryos (6, 7). This program continued during thefirst 2 postnatal weeks at well-defined regions ( F1Fig. 1A, B).

FIGURE 1. (A, B) Line drawings representing a frontal A and a sagittal section B of the brain of an adult hydrocephalic (hy)mouse depicting in blue the ependymal regions resistant to denudation. The broken line in A indicates the plane of the sectionshown in B. The area framed in B that includes the medial wall of the lateral ventricle and a portion of the hippocampus is shownin CYE. (CYE) Walls of the lateral ventricle of hy mice injected daily with bromodeoxyuridine (BrdU) from postnatal day (PN) 4 toPN6. (C) In ependymal denuded areas there are numerous nuclei labeled with BrdU in the ventricular surface (arrowhead) anddeeper in the brain parenchyma (arrow), in contrast to the adjacent area where the ependyma was present (double endedarrow). Counterstaining with hematoxylin. (D) Covering denuded areas, BrdU-labeled nuclei appear to be present in cells thatcorrespond to astrocytes (arrow). (E) Detail of a subependymal astrocyte with the nucleus labeled for BrdU and a long cellprocess. (F, G) Projections in Z axis for a thickness of 8 Km of the wall of the lateral ventricle of a hy PN4 mouse injected with asingle dose of BrdU. Immunostaining for glial fibrillary acidic protein (GFAP) (green) and BrdU (red). (F) Some astrocytesimmunoreactive for GFAP covering the ventricular surface (thick arrow) and deeper in these areas (thin arrow) displayed nuclearlabeling for BrdU. (G) Detail of an astrocyte in the brain parenchyma near a denuded area immunolabeled for GFAP and BrdU.Ceb, cerebrum; CP, choroid plexus; H, habenula; Hip, hippocampus; LV, lateral ventricle; 3V, third ventricle; 4V, fourth ventricle;ME, median eminence; SA, aqueduct of Silvius; SCO, subcommissural organ; SFO, subfornical organ.AQ3

Fig

14/C

J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007 Neuropathologic Pattern of Hydrocephalus

� 2007 American Association of Neuropathologists, Inc. 3

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

By the end of the second postnatal week the program ofdenudation had been completed.

The ependymal regions resistant to denudation weresystematically and consistently the same in all animalsstudied, regardless of their age or severity of hydrocephalus.These areas included most circumventricular organs,namely, choroid plexuses, subcommissural organ, subforn-ical organ, organon vasculosum of the lamina terminalis,median eminence, and collicular recess organ (Fig. 1A). Inaddition, some other distinct ependymal patches localizedin i) the lateral and medial walls of the lateral ventricles, ii)the area of the third ventricle lining the habenula, iii)the habenular and pineal recesses, and iv) the floor ofthe aqueduct of Silvius, were also resistant to denudation(Fig. 1A, B).

Astroglial Reaction Occurring in the DenudedVentricular Areas

After ependymal loss, the underlying neuropil remaineddenuded and exposed to CSF for a few days (F2 Figs. 2,F3 3B, C).A reaction of those astrocytes lying close to the denudedarea then started to occur (Fig. 2). In non-hy PN6 mice, afew BrdU labeled nuclei were seen scattered in thesubependymal neuropil and in the ependymal lining proper.However, PN6 hy mice showed a large population of labelednuclei at the denuded area (Fig. 1C). BrdU and GFAP werecolocalized in cells lining the denuded area (Fig. 1D, F) andin cells located in the underlying neuropil (Fig. 1EYG).

Immunolabeling for GFAP, S100A protein, andvimentin showed differences between i) nonreactive astro-cytes distributed throughout the CNS of both, non-hy and hymice, ii) periventricular reactive astrocytes appearing atdenuded areas, and iii) reactive astrocytes found in injuredareas of the CNS, such as the degenerating Probst bundles

occurring in hy mice (Figs. 2, 3). Astrocytes expressedGFAP (Fig. 3D, G, J). In addition to GFAP, all reactiveastrocytes expressed vimentin (Fig. 3E, H, K), and onlyperiventricular reactive astrocytes were labeled with anti-S100A antibody (Fig. 3I, L). Reactive astrocytes expressedproliferating cell nuclear antigen (PCNA) (Fig. 3F), furthersupporting the findings obtained with BrdU (see above).This probe also showed differences between the 2 popula-tions of reactive astrocytes. Thus, the PCNA-positive cellsfound in injured areas appeared dispersed, whereas theperiventricular glial cells expressing PCNA were groupedinto small clusters (Fig. 3F). Similar clusters of cells locatedin the vicinity of denuded areas were seen to display BrdU-labeled nuclei (Fig. 1C).

The proliferation of the periventricular reactive astro-cytes at the denuded areas followed a certain pattern withwell-defined phases (Figs. 2, 3). Periventricular reactive as-trocytes were detected 2 to 6 days after denudation; thesefirst appeared loosely arranged and expressed vimentin andGFAP (Figs. 2 Phase C, 3D, E, G, H, J, K). At a later phase,some of these astrocytes displayed a PCNA-reactive nucleusand formed small cell clusters (Figs. 2 Phase D, 3F).Approximately 1 week after denudation an almost continu-ous layer of GFAP and vimentin-reactive astrocytes, whichincluded those with a PCNA-positive nucleus, covered thenaked neuropil (Figs. 2 Phase E, 3DYF). Some days latermany of these astrocytes expressed S100A protein (Figs. 2Phases F1 and F2, 3I, L). Approximately 2 weeks afterdenudation the astroglial layer lining the denuded area wasquite compact. At the age of 1 month the arrangement of theglial layer had reached its definitive profile, which variedaccording to the location of the affected area (Figs. 2 PhasesG1 and G2, 3M, N). In this way, at PN30, in hy mice thefloor of the fourth ventricle was lined by a thick and compact

FIGURE 2. Linedrawing depicting the time progress of the astroglial reaction at the ventricular denuded areas of hydrocephalic(hy) mice. The different stages of the astrocytic reaction and their immunocytochemical characteristics are represented. PhasesAYE are common to all surfaces with an astroglial reaction; phases F1 and G1 are seen in surfaces with a mild astroglial reaction;phases F2 and G2 are found in surfaces with a severe astroglial reaction. Phase A: ventricular surface with neuroepithelium beforedenudation. Ventricular surface after denudation is present in Phase B. In Phase C reactive astrocytes are detected in the denudedventricular surface. Phase G1 arose from F1. Phase G2 arose from F2. PCNA(+), cells positive for proliferating cell nuclear antigen;S100A(+), cells positive for S100A; vimentin(+), astrocytes positive for vimentin.

Fig

24/C

Paez et al J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007

� 2007 American Association of Neuropathologists, Inc.4

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

FIGURE 3. (A, B) Details of the neuroepithelium of the lateral ventricle of a postnatal day (PN) 1 hydrocephalic (hy) mousebefore denudation in A and of a PN3 hy mouse after denudation of the ventricular layer in B. Hematoxylin and eosin staining.Neuroepithelium is seen in A and is missing in B. (B, inset) Vimentin immunocytochemistry of a section adjacent to that shown inB. (C) Scanning electron microscopy showing the denuded ventricular surface of a PN3 hy mouse. Asterisk labels a neuroblast-like cell. (D, E, G, H, J, K) glial fibrillary acidic protein (GFAP) and vimentin (Vim) immunostaining in consecutive sectionsshowing the astroglial reaction occurring at denuded areas of the ventricular walls in hy mice at phases E (D, E), G1 (G, H), andG2 (J, K). Reactive astrocytes coexpress GFAP and vimentin (arrows). (F) Glial scar in formation immunostained with anti-proliferating cell nuclear antigen in phase E (arrows). (I, L) Immunolabeling with anti-S100A of PN30 hy mice. Phases G1 (I) andG2 (L) present reactive astrocytes expressing S100A. (M, N) Scanning electron microscopy of denuded ventricular surfacesshowing 2 different astrocyte arrangements: namely, a surface with an astroglial reaction characterized by numerous astrocyticprocesses partially covering the denuded region (M) and a surface almost fully covered by thin cytoplasmic sheaths of astrocytes(N). Asterisks label cell bodies of astrocytes. cp, ; ds, .AQ4

Fig

34/C

J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007 Neuropathologic Pattern of Hydrocephalus

� 2007 American Association of Neuropathologists, Inc. 5

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

layer of phase G2 type astrocytes (Fig. 3JYL, N). The floorof the aqueduct of Sylvius and the lateral wall of the lateralventricles had a thinner, although compact, layer ofastrocytes (phase G1 type).

The glial reaction followed a spatial and temporalpattern that paralleled that of neuroepithelium/ependymadenudation. No glial reaction was seen in those areasresistant to denudation and in the denuded medial rostralwalls of the lateral ventricles (Fig. 1C).

Malformation of the Corpus Callosum andHippocampal Commissure

In non-hy mice, callosal and hippocampal axonscrossed the midline and established connections with thecontralateral hemisphere (F4 Fig. 4A, C, E, G). The corpuscallosum was already evident at E16.5 (Fig. 4G).

In all hy mice, callosal and hippocampal axons grewtoward the midline but never crossed it, resulting in acomplete lack of corpus callosum and hippocampal com-missure (Fig. 4B, D, F, H). When approaching the midline,axons twisted and formed tangles, a pathologic structureknown as Probst bundles (Fig. 4D, H); these had alreadybeen seen in E16.5 mice (Fig. 4H). It is worth mentioningthat at this embryonic age the lateral ventricles were notdilated (Fig. 4H). Interestingly, the habenular, posterior, andanterior commissures did form in hy mice (Fig. 3B).

Callosal and Hippocampal Fibers DegenerateAfter Failing to Cross the Midline

At PN3, the Probst bundles and the external capsuleof hy mice started to show microscopic signs of edema.At this age, cells labeled with tomato lectin and anti-LN5,most likely corresponding to phagocytic cells, started toappear in these structures. A small cavity appeared at thedistal portion of the external capsule, starting a cavitationprocess that led to a disarrangement of corticostriatalfibers and their exposure to the CSF ( F5Fig. 5F). From PN5to PN8 these nerve fibers showed evident signs of degener-ation; the number of phagocytic cells increased and thecavity became larger (Fig. 5C, D, E). Reactive astrocytes,immunopositive to vimentin and GFAP, were detected in theedematous neuropil. Some of these displayed a PCNA-positive nucleus but, unlike the periventricular reactiveastrocytes, they neither formed clusters nor expressedS100A protein.

During the second postnatal week the density ofphagocytic cells and reactive astrocytes reached the max-imum degree (Fig. 5GYJ), whereas the density ofnerve fibers in the external capsule was drastically reduced(Fig. 5E, F). The lateral ventricles underwent a largeexpansion and fused with the expanded cavity, resulting ina large fistula located between the external capsule and thestriatum (Fig. 1F).

FIGURE 4. Agenesis of corpus callosum and hippocampal commissure. Frontal sections of nonhydrocephalic (non-hy) (A, C, E)and hydrocephalic (hy) mice (B, D, F) at postnatal day (PN) 1. (A, B) Immunostaining with anti-neural cell adhesion molecules(NCAM). The lateral ventricles appear collapsed in the hy mouse. (C, D) 1,1-Dioctadecyl-3,3,3¶,3¶-tetramethylindocarbocyanineperchlorate tracing of callosal fibers shows their absence in the midline in the hy mouse. (E, F) Hematoxylin and eosin (H-E)staining showing the absence of the hippocampal commissure in the hy mouse. (G, H) Frontal sections of the brain of non-hy(G) and hy (H) mice at embryonic day 16.5 (E16.5) immunostained with anti-NCAM. The lateral ventricles appear collapsed inthe hy embryo. AC, anterior commissure; CC, corpus callosum; Ct-St, corticostriatal fibers; Ctx, cortex; EC, external capsule; HC,habenular commissure; Hip, hippocampus; if, ; LV, lateral ventricle; PB, Probst bundles; 3V, third ventricle.AQ5

Fig

44/C

Paez et al J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007

� 2007 American Association of Neuropathologists, Inc.6

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

FIGURE 5. Degeneration of corpus callosum and external capsule. (A, B) Corpus callosum and external capsule ofnonhydrocephalic (non-hy) (A) and hydrocephalic (hy) (B) mice at postnatal day (PN) 5 immunostained with anti-neural celladhesion molecules (NCAM). (C) Hematoxylin and eosin (H-E) staining of a PN5 hy mouse. Edema is present in the capsuleregion (arrows), and a cavity appears in its terminal portion (asterisk in inset). (DYF) Immunostaining with anti-neurofilament(NF) of frontal sections through the external capsule of PN5 (D), PN8 (E), and PN14 (F) hy mice. The progress of thedegeneration of the external capsule (arrowhead), the disarrangement of the corticostriatal connections, and the expansion ofthe lateral ventricle between the external capsule and striatum originating a cavitation (asterisk) are shown. (G, I) Frontalsections through the brain, at the level of the rostral horn of the lateral ventricle of a PN11 hy mouse. Tomato lectin (TL) binding.A large population of macrophages is seen among the Probst bundles (G) and in the external capsule (I). The cavity (asterisk in I)is now enlarged. (H, J) Detailed magnifications of macrophages concentrated in the Probst bundles (H) and the external capsule(J). (KYM) Scanning electron microscopy (SEM) of a PN30 hy mouse showing the dilation of the lateral ventricle (K), thedegenerated external capsule (KYM), and the presence of macrophages on the corticostriatal connections exposed tocerebrospinal fluid (CSF) (L, M). (N) Immunostaining of corticostriatal fibers with anti-myelin basic protein (MBP). (N, inset)Detail of a macrophage associated with CSF-exposed corticostriatal fibers. AC, anterior commissure; CC, corpus callosum; Ct-St,corticostriatal; Ctx, cortex; EC, external capsule; HC, habenular commisure; Hip, hippocampus; if, ; LV, lateral ventricle; PB,Probst bundles; St, striatum; 3V, third ventricle.

Fig

54/C

J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007 Neuropathologic Pattern of Hydrocephalus

� 2007 American Association of Neuropathologists, Inc. 7

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

FIGURE 6. Alterations of the cerebral cortex. (A) Table of statistical analyses of differences for thicknesses of the cingular cortex(CgC), frontal cortex (FC), parietal cortex (PC), and occipital cortex (OC) compared among nonhydrocephalic (non-hy) mice,hydrocephalic (hy) mice with slowly progressive or compensated hydrocephalus (SCH), and hy mice with rapidly progressive andvery severe hydrocephalus (VSH). Before postnatal day (PN) 7, the hy mice do not show all of the characteristics to be directlyclassified into animals with VSH or with SCH (8); therefore, hy mice were classified into 2 groups based on early externalsymptoms: i) Ha, probably corresponding to VSH mice and (ii) Hb, probably corresponding to SCH mice. Differences aredetected between most of the groups of animals studied except between the FC and the PC of PN3-Ha and PN3-Hb mice.Curiously, at PN8 significant differences are found in these cortical areas, indicating that this interval of age may be veryimportant in their different evolution. (BYE) Rates of increase of the thickness at different ages of CgC, FC, PC, and OC from non-hy mice (black circles), SCH mice (grey circles), and VSH mice (open circles). Slopes (m) indicate the rate of increase of corticalthickness for each interval of age and for each cortical region. (FYI) Frontal sections through the lateral ventricles of PN8 (F),PN14 (G), and PN30 (H) hy mice and of a PN30 non-hy mouse (I). Hematoxylin and eosin. Increasing edema in Probst bundlesis observed from PN8 to PN14 hy mice (asterisks in F and G), and ventriculostomies appear in these edematous areas at PN30(arrows in H). Simultaneously, the cortical thickness decreases in hy mice (brackets in F and G). Compare the difference incortical thickness between hy and non-hy animals (double-ended arrow in H and I). EC, external capsule; Hip, hippocampus; LV,lateral ventricle; PB, Probst bundles.

Paez et al J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007

� 2007 American Association of Neuropathologists, Inc.8

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

At the third postnatal week, the Probst bundles and theexternal capsule had disappeared and the number ofphagocytic cells in these areas was drastically reduced. AtPN30, corticostriatal fibers were fully exposed to the CSFand associated with phagocytic cells (Fig. 5KYN).

Postnatal Alterations in the Brain Cortex ofHy Mice

Essentially, 2 types of clinical evolution were distin-guished (8): i) mice with a VSH, many of which died duringthe first 2 months of life; and (ii) mice with SCH, which sur-vived for several months, with some reaching 2 years of age.

In non-hy mice, the thickness of the cingular, frontal,parietal, and occipital cortical regions steadily increasedduring the first month of life. In fact, at PN30 the braincortex had reached a thickness no different from that ofPN120 mice (F6 Fig. 6AYE).

The cerebral cortex of hy mice was thinner than that ofnon-hy animals, even at early ages when the lateralventricles were not yet dilated (Fig. 6AYE). Furthermore,from PN8 on the cerebral cortex of VSH mice was signifi-cantly thinner than that of animals with SCH (Fig. 6AYE).During the first postnatal month, the increase in thickness ofthe cerebral cortex of SCH mice paralleled that of non-hymice, although it was significantly lower (Fig. 6AYE). Thisdifference was more pronounced in the parietal and occipitalcortical regions (Fig. 6D, E).

Interestingly, in VSH mice we observed not only areduction in the increase of cortical thickness (lower positiveslope values compared to controls), but an actual decrease inthickness (negative slope values) of specific cortical regions(Fig. 6B, C, E). A drastic reduction in thickness of thecingular and frontal cortices was observed in PN120 VSHanimals as compared to VSH mice at PN30 (Fig. 6B, C). Inthe occipital cortex a reduction in thickness was alreadypatent shortly after birth (from PN3 to PN8) (Fig. 6E).

The histologic study of the cingular, frontal andparietal areas revealed the appearance of ventriculostomiesin the edematous area where degeneration of Probst bundleswas occurring, communicating lateral ventricles with thesubarachnoid space (Fig. 6FYI).

DISCUSSIONThe outcome of the present study in an animal model

of hydrocephalus was that virtually none of the pathologicevents could be regarded as the result of mechanicalphenomena triggered by ventricular expansion or increasedintracranial pressure. Rather, abnormalities occurring atcellular levels, such as ependymal denudation, astrocytosis,cortical alterations, agenesis of corpus callosum, axonaldegeneration, and brain edemas, precede the onset ofhydrocephalus.

Neuroepithelium/Ependyma DenudationInitiated at Embryonic Day 12 ContinuesAfter Birth

In previous reports we have shown that ependymaldenudation is a key event in the pathogenesis of hydro-cephalus in hyh mutant mice (6, 7) and also probably in

humans (10). In hyh mutant mice, the neuroepithelium/ependyma denudation follows a program initiated at E12,before the onset of hydrocephalus (6). The present study hasshown that the program of ependymal denudation continuesduring the second postnatal week in the walls of the lateralventricles. The mechanism underlying this denudation isunknown. It has recently been determined that the mutatedgene in the hyh mouse corresponds to that encoding forsoluble N-ethylmaleimide-sensitive factor attachment pro-tein >, resulting in an altered localization of apical adhesionproteins in neuroepithelial cells (9, 10) and certain ependy-mal cells (12). Therefore, an alteration of the cellularadhesion may be suggested to trigger their denudation.

Those ependymal areas that are consistently resistantto denudation, mostly corresponding to the circumventricu-lar organs, represent the counterpart of the program ofependymal denudation. Two features common to all thesestructures are that their ependymal cells are joined togetherby junction complexes that include tight junctions and thatthey lie on a basal lamina (13Y15). At variance, the ciliatedependyma undergoing detachment is devoid of these 2properties (13Y15), suggesting that differences in the cellularadhesion may be the origin of the resistance to denudationand probably reflecting different cellular lineages.

Astrocytosis: Ependyma Denudation Triggers aResponse of Periventricular Astroglia

The periventricular astroglial response detected inhyh mutant mice involves periventricular astrocytes thatproliferate and form a glial layer Bresembling[ the ependy-mal layer. This glial response follows the same temporal andspatial pattern of neuroepithelium/ependyma denudation(present work), suggesting that this denudation triggers thesignals responsible for the astrocytosis. A control mecha-nism of proliferation of subependymal astroglial by theependyma has been described in specific areas of ventricularcavities (16). The degree of astroglial reaction to denudationvaries throughout the different regions of ventricular wallsand would reflect differences in the structure and in thecapacity of response of the subventricular layer (17).Discontinuities in the ependyma of hydrocephalic humanfetuses, associated with proliferation of astrocytes, have alsobeen reported (11, 18). The actual mechanism(s) underlyingthese distinct responses to ependymal denudation, as well asthe functional significance of this phenomenon, remain to beinvestigated.

Periventricular astrocytes show an immunocytochem-ical phenotype that is different than that of astrocytes presentin the brain parenchyma (Reference 19 and present report),suggesting that they represent a subpopulation of astroglia.They also appear to differ from nonperiventricular reactiveastrocytes in injured areas (20), such as those associated withthe degenerating Probst bundles (present report). Althoughthe functional significance of the newly formed cell layercovering the denuded areas is unknown, the possibility thatit performs a repair role of the lost ependyma should beconsidered. This recovery of the postnatal neurogenesis inNumb/numb-like mutant mice with ependymal denudationhas been reported after glial scar formation (21).

J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007 Neuropathologic Pattern of Hydrocephalus

� 2007 American Association of Neuropathologists, Inc. 9

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

Denudation of the Neuroepithelium/Ependyma, but Not Hydrocephalus, MayExplain Perinatal Cortical Alterations and

Agenesis of Corpus CallosumNeuroepithelium and immature ependymal layers

contain progenitor cells during development of the CNS(20, 22). Therefore, denudation of neuroepithelium in hyhmutant embryos would imply a loss of the germinal zoneand a disorganization of the subventricular zone. Becausedenudation of neuroepithelium and ependyma occurs atwell-defined areas (6, 7), abnormal neurogenesis would beexpected to occur at these particular points rather than beinga widespread abnormality caused by hydrocephalus. Thepresent findings support these views. i) The brain cortex isaffected before the ventricular dilatation is evident. Thereduced cortical thickness detected at PN3, when dilatationof the lateral ventricles has not yet occurred, could beexplained by a loss of neural progenitor cells (9); and ii) thebrain cortex is not affected homogenously, as has also beenreported in other models (23Y25). Because detachment ofneuroepithelium occurs only at certain regions and at certainstages of development (6, 26), differences in the nature anddegree of abnormalities of the different cortical regionsshould be expected.

Moreover, only certain commissural structures, such asthe corpus callosum and hippocampal commissure, displayabnormalities, whereas others forming the anterior andposterior commissures do not. Agenesis of the corpuscallosum could be explained by: i) an alteration of thecallosal neurons or ii) a deficient development of the cellpopulations that have been reported to function in midlineguidance of callosal fibers (27). In this mutant, alterations ofcallosal neurons could be due to i) an abnormal neurogenesisor ii) an abnormal pathfinding of the cingulated pioneeringaxons caused by an alteration in the vesicular traffic to theaxonal plasma membrane (8, 28, 29). These alternatives arein contrast with the classic view of Bronson and Lane (4),who suggested that agenesis of corpus callosum in hyhmutant mice is due to the large dorsal expansion of the thirdventricle. This suggestion certainly cannot be the casebecause crossing of callosal fibers occurs during theembryonic life (References 27, 28 and present report), anddorsal expansion of the third ventricle is a postnatal event(5Y7).

Maldevelopment and Degeneration of CallosalAxons and External Capsule Fibers Leads toPostnatal Cortical Degeneration

Hydrocephalus has been frequently associated withother neurologic disorders such as cortical defects (1, 3). Inanimal models and human cases of hydrocephalus, agenesisof the corpus callosum has also been related to corticaldefects (30).

The callosal neurons are located in the neocortex (31).The external capsule is formed by corticofugal axonsoriginating in neurons of the frontal and parietal cortices(32). In hyh mutant mice it seems likely that postnataldegeneration of axons of the corpus callosum and external

capsule and the postnatal marked decrease in thickness ofthe frontal and cingular cortices (see above) are interrelatedphenomena. Presumably, the cell body of those neuronsforming the corpus callosum and the external capsuleundergo degeneration after their axons have degenerated,thereby contributing to the cortical alteration.

Analysis of the temporospatial evolution of differentneuropathologic events traditionally related to hydrocepha-lus suggests that i) hydrocephalus and abnormalities ofcertain brain structures are linked to each other at theetiologic level, rather than the latter being a consequence ofthe former and ii) denudation of neuroepithelium/ependymamay be directly or indirectly related to the origin of theneuropathologic events described above.

ACKNOWLEDGMENTSThe authors are grateful to Mr. Genaro Alvial

(Universidad Austral de Chile) for his valuable technicalsupport. The monoclonal antibody against NCAM developedby J. M. Hessel and J. Dodd was obtained from theDevelopmental Studies Hybridoma Bank. The antibody wasdeveloped under the auspices of the National Institute ofChild Health and Human Development and is maintained byThe University of Iowa, Department of Biological Sciences,Iowa City, IA.

REFERENCES1. Bruni JE, del Bigio MR. Hydrocephalus. In: Dulbecco R, ed.

Encyclopedia of Human Biology, 2nd ed. San Diego, CA: AcademicPress, 1997:645Y51

2. Del Bigio MR. Neuropathological changes caused by hydrocephalus.Acta Neuropathol 1993;85:573Y85

3. Erickson K, Baron IS, Fantie BD. Neuropsychological functioning inearly hydrocephalus: Review from a developmental perspective. ChildNeuropsychol 2001;7:199Y229

4. Bronson RT, Lane PW. Hydrocephalus with hop gait (hyh): A newmutation on chromosome 7 in the mouse. Dev Brain Res 1990;54:131Y36

5. Perez-Fıgares JM, Jimenez AJ, Perez-Martın M, et al. Spontaneouscongenital hydrocephalus in the mutant mouse hyh: Changes in theventricular system and the subcommissural organ. J Neuropathol ExpNeurol 1998;57:188Y202

6. Jimenez AJ, Tome M, Paez P, et al. Programmed ependymal denudationoccurring during the embryonic life precedes the development ofcongenital hydrocephalus in the mutant mouse hyh. J Neuropathol ExpNeurol 2001;60:1105Y19

7. Wagner C, Batiz LF, Rodrıguez S, et al. Cellular mechanisms involvedin the stenosis and obliteration of the cerebral aqueduct of hyh mutantmice developing congenital hydrocephalus. J Neuropathol Exp Neurol2003;62:1019Y40

8. Batiz LF, Paez P, Jimenez AJ, Rodrıguez S, et al. Heterogeneousexpression of hydrocephalic phenotype in the hyh mice carrying a pointmutation in >-SNAP. Neurobiol Dis 2006;23:152Y68

9. Chae TA, Kim S, Marz KE, et al. The hyh mutation uncovers roles for >Snap (in apical protein localization and control of neural cell fate. NatGenet 2004;36:264Y70

10. Hong HK, Chakravarti A, Takahashi S. The gene for soluble N-ethylmaleimide sensitive factor attachment protein > is mutated inhydrocephaly with hop gait (hyh) mice. Proc Natl Acad Sci USA 2004;101:1748Y53

11. Domınguez-Pinos MD, Paez P, Jimenez AJ, et al. Ependymaldenudation and alterations of the subventricular zone occur in humanfetuses with a moderate communicating hydrocephalus. J NeuropatholExp Neurol 2005;64:595Y604

12. Batiz LF, Oliver C, Alvarez M, et al. Molecular mechanisms underlying

Paez et al J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007

� 2007 American Association of Neuropathologists, Inc.10

Copyright @ 2007 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

neuroepithelial/ependymal denudation in the hydrocephalic hyh mutant:Spatial and temporal expression of >-SNAP and N-cadherin (Abstract).Cerebrospinal Fluid Res 2006;21( 3 Suppl 1):S16

13. Rodrıguez EM. The cerebrospinal fluid as a pathway in neuroendocrineintegration (Review). J Endocrinol 1976;71:407Y43

14. Rodrıguez EM, Oksche A, Hein S, Yulis CR. Cell biology of thesubcommissural organ. Int Rev Cytol 1992;135:39Y121

15. Bruni JE, del Bigio MR, Clattenburg RE. Ependyma: Normal andpathological: A review of the literature. Brain Res 1985;356:1Y19

16. Lim DA, Tramontin AD, Trevejo JM, Herrera DG,, Garcia-VerdugoJM, Alvarez-Buylla A. Noggin antagonizes BMP signaling to create aniche for adult neurogenesis. Neuron 2000;28:713Y26

17. Temple S, Alvarez-Buylla A. Stem cells in the adult mammalian centralnervous system. Curr Opin Neurobiol 1999;9:135Y41

18. Sarnat HB. Role of human fetal ependyma. Pediatric Neurol 1992;8:163Y78

19. Yamada T, Kawamata T, Walker DG, et al. Vimentin immunoreactivityin normal and pathological human brain tissue Acta Neuropathol (Berl)1992;84:157Y6

20. Fitch MT, Doller C, Combs CK, et al. Cellular and molecularmechanisms of glial scarring and progressive cavitation: In vivo andin vitro analysis of inflammation-induced secondary injury after CNStrauma. J Neurosci 1999;19:8182Y98

21. Kuo CT, Mirzadeh Z, Soriano-Navarro M, et al. Postnatal deletion ofNumb/Numblike reveals repair and remodeling capacity in thesubventricular neurogenic niche. Cell 2006;127:1253Y64

22. Tramontin AD, Garcıa-Verdugo JM, Lim DA, Alvarez-Buylla A.Postnatal development of radial glia and the ventricular zone (VZ): A

continuum of the neural stem cell compartment. Cereb Cortex 2003;13:580Y87

23. Bruni JE, del Bigio MR, Cardoso ER, et al. Neuropathology ofcongenital hydrocephalus in the SUMS/NP mouse. Acta Neurochir(Wien) 1988;92:118Y22

24. Jones HC, Dack S, Ellis C. Morphological aspects of the developmentof hydrocephalus in a mouse mutant (SUMS/NP). Acta Neuropathol(Berl) 1987;72:268Y76

25. Jones HC, Bucknall RM, Harris NG. The cerebral cortex in congenitalhydrocephalus in the H-Tx rat: A quantitative light microscopic study.Acta Neuropathol 1991;82:217Y24

26. Overholser MD, Whitley JR, O’Dell BL, et al. The ventricular systemin hydrocephalic rat brains produced by a deficiency of vitamin B12 orof folic acid in the maternal diet. Anat Rec 1954;120:917Y33

27. Shu T, Puche AC, Richards LJ. Development of midline glialpopulations at the corticoseptal boundary. J Neurobiol 2003;57:81Y94

28. Rash BG, Richards LJ. A role for cingulate pioneering axons in thedevelopment of the corpus callosum. J Comp Neurol 2001;434:147Y57

29. Tayeb MA, Skalski M, Cha MC, Kean MJ, Scaife M, Coppolino MG.Inhibition of SNARE-mediated membrane traffic impairs cell migra-tion. Exp Cell Res 2005;305:63Y73

30. Kamnasaran D. Agenesis of the corpus callosum: Lessons from humansand mice. Clin Invest Med 2005;28:267Y82

31. Richards LJ, Plachez C, Ren T. Mechanisms regulating the develop-ment of the corpus callosum and its agenesis in mouse and human. ClinGenet 2004;66:276Y89

32. Metin C, Deleglise D, Serafini T, et al. A role for netrin-1 in theguidance of cortical efferents. Development 1997;124:5063Y74

J Neuropathol Exp Neurol � Volume 66, Number 12, December 2007 Neuropathologic Pattern of Hydrocephalus

� 2007 American Association of Neuropathologists, Inc. 11