OUP user menu

Deficient cortical development in the hydrocephalic Texas (H‐Tx) rat: a role for CSF

Farhad Mashayekhi, Clare E. Draper, Carys M. Bannister, Mohsen Pourghasem, P. Jane Owen‐Lynch, Jaleel A. Miyan
DOI: http://dx.doi.org/10.1093/brain/awf182 1859-1874 First published online: 1 August 2002

Summary

The objectives of this study were to demonstrate the presence and nature of abnormal cortical development in a rat model of hydrocephalus, the hydrocephalic Texas (H‐Tx) rat, and to test the hypothesis that the obstruction of CSF flow in affected animals can be linked to this effect. CSF is secreted continuously by the choroid plexus, located in the lateral, third and fourth ventricles. The fluid flows through the ventricular system, passing over all regions of germinal activity. In the H‐Tx rat, obstruction and eventual blockage of CSF flow occurs in the cerebral aqueduct, between the third and fourth ventricles, at embryonic day 18. Prior to obstruction, neurogenesis and migration occur as in normal rats. Here we show that, following obstruction of fluid flow, neurogenesis from the germinal epithelium becomes abnormal. Cell proliferation decreases and proliferating cells are not retained in the germinal layer, as they appear to be in the normal brain. Cell migration is apparently unaffected, although a decrease in the number of migrating cells does occur after CSF obstruction. These data from our study indicate that a rapid primary effect of CSF obstruction, prior to any mechanical effects of fluid accumulation, is to alter the activity of cells in the germinal epithelium of the developing cortex. Further evidence for this is gained from in vitro studies. Once removed from their in vivo environment, cortical cells from the H‐Tx rat have the ability to proliferate as normal. CSF extracted from the enlarged ventricles of affected brains is able to inhibit the proliferation of normal cells. Thus, we hypothesize that CSF has a potential role in the developmental process. The damming up and accumulation of CSF, whatever the cause, may result in abnormal cortical development through accumulation of CSF factors that are, or become, inhibitory to normal neuronal proliferation.

  • Keywords: cerebrospinal fluid; cerebral cortex; development; hydrocephalus; H‐Tx rat
  • Abbreviations: BrdU = 5‐bromo‐2′‐deoxyuridine; H‐Tx = hydrocephalic Texas rat; ICP = intracranial pressure

Received June 13, 2001. Revised October 21, 2001. Second revision January 9, 2002. Accepted February 23, 2002

Introduction

The development of the brain, particularly the cerebral cortex, has received a great deal of attention over the past few years and many of the molecular mechanisms involved are becoming clear (Lambert de Rouvroit and Goffinet, 1998; Desai and McConnell, 2000; Homayouni and Curran, 2000; Kwon and Tsai, 2000). Attention has focused on two main areas of research. The first of these is brain stem cells, found in the neural tube and ventricular zone of the cerebral cortex, and the mechanisms involved in their proliferation and differentiation into neurones and glia (Svendsen and Smith, 1999; McKay, 2000; Mansergh et al., 2000). The second area of research is the hierarchical signals involved in the migration of neurones from the germinal epithelium into the cortex, along the radial glia. In the latter area, the vital role of Cajal–Retzius cells has been elucidated (Marin‐Padilla, 1998, 1999; Super et al., 1998, 2000). These cells, located in the marginal zone, secrete the extracellular glycoprotein reelin. Reelin stimulates receptor and intracellular signalling systems in responsive neural precursors and mediates their migration and stratification into the cortex (Bar et al., 2000; Gilmore and Herrup, 2000). Clearly, normal corticogenesis involves the coordination of a tightly regulated set of processes.

An understanding of the normal processes of cortical development leads inevitably to a clearer basis for defining cases where cortical development appears to be abnormal, as in the case of early onset hydrocephalus (Rekate, 1997; Bannister et al., 1998, 2000). Even though the postnatal treatment of affected infants by the insertion of a shunt to drain CSF from the ventricular system prevents the damaging build‐up of intracranial pressure (ICP) (Hanigan et al., 1991; Miyazawa and Sato, 1991; Kriebel et al., 1993; Rekate, 1997; Mataro et al., 2000), cortical deficiencies in these infants remain (McAllister and Chovan, 1998). Similar findings have been reported for the H‐Tx rat model of hydrocephalus (Miyazawa and Sato, 1991).

The early signs of hydrocephalus in the prenatal period, including ventriculomegaly, follow obstruction of the flow and consequent fluid accumulation of CSF in the fluid spaces in and around the brain. The choroid plexus, located in the ventricles of the brain, continuously secretes fluid. The fluid flows from the lateral ventricles into the third ventricle, through the aqueduct of Sylvius into the fourth ventricle and then into the subarachnoid space, which lies between the pia and arachnoid meninges covering the surface of the brain and spinal cord. Fluid eventually arrives at the superior sagittal sinus, where it is absorbed into the venous circulation.

We have used the H‐Tx rat model of hydrocephalus to investigate the effects of obstruction of CSF flow on cortical development and have begun to examine the underlying mechanism for the observed defective development in the prenatal stages of this condition. The H‐Tx rat was isolated from a normal colony as a spontaneous mutation (Kohn et al., 1981, 1984) and subsequently maintained through brother–sister mating. Recent breeding analysis suggests that the mutation is monogenic, the H‐Tx rat being a homozygous carrier of an autosomal recessive gene mutation with incomplete penetrance (Cai et al., 2000). However, preliminary genetic studies (using backcross progeny from normal and H‐Tx rats) implicate up to four genetic loci (Jones et al., 2001). Approximately 40% of the pups in a purebred H‐Tx litter are born with hydrocephalus (Cai et al., 2000). In these rats, hydrocephalus results from impedance of the CSF flow and, in many cases, obstruction by constriction of the cerebral aqueduct between the third and fourth ventricles of the brain (Jones and Bucknall, 1988; Oi et al., 1996). Severe obstruction occurs at day 18 of gestation at the height of neurogenesis and when mass migration of cells into the developing cortex is taking place. Controversy exists about the primary cause of hydrocephalus in this model. Early studies focused on obstruction in the aqueduct (Jones and Bucknall, 1988; Jones, 1992, 1997), but more recent studies suggest that the primary defect may be failure of CSF absorption external to the brain (Oi et al., 1996). Our own recent morphological studies support the latter view: that aqueduct obstruction is secondary to a primary defect external to the brain (Pourghasem et al., 2001). Furthermore, explanations for why some foetuses are affected while others are not in this inbred strain favour the possibility that there may be a maternal factor that is either transported or fails to be transported across the placenta of each separate foetus, which interacts with the genetic defect present in foetuses expressing the hydrocephalic phenotype. For the present study, the important factor is the damming of CSF and the failure of CSF to circulate during a critical period in cortical neurogenesis.

Most studies with the H‐Tx model have focused on postnatal events, such as the effects of shunting on behaviour, measurements of raised ICP and the alterations in growth factor expression in the brains of affected animals (Richards et al., 1989; Kaiser and Jones, 1991; Jones et al., 1991, 1995; Miyazawa and Sato, 1991; Harris et al., 1994; Cai et al., 1999). In this paper we describe our studies on prenatal cortical development in this model and show that there is abnormal development of the cerebral cortex. The observed abnormal development shares features with some forms of human early‐onset hydrocephalus. The objectives of the present study were to demonstrate the presence and nature of the abnormal development of the cerebral cortex in the H‐Tx rat and to test the hypothesis that the obstruction of CSF flow in the affected animals can be linked to them. We define a decreased cell output from the germinal epithelium along with failure to retain dividing cells within this layer. In vitro studies provide evidence for our hypothesis that it is the obstruction of CSF flow that is responsible for these effects and that the accumulated CSF is a major factor in the abnormal development.

Material and methods

Maintenance of H‐Tx colony

All experiments were sanctioned by the Home Office Animal Procedures Inspectorate. Both Wistar and H‐Tx rat colonies were maintained on a 12 h light : 12 h dark cycle beginning at 8.00 a.m. They were kept at a constant temperature in large rat boxes with unrestricted access to food and water. The H‐Tx colony was maintained through brother–sister mating between unaffected animals, whereas the Wistar colony was maintained through random pair matings. A male and female were placed in a box for a timed mating and were then checked hourly for the presence of a vaginal plug. The presence of a plug indicated a successful mating and the day of observation was recorded as gestational day 0.

Identification of affected animals

In order to investigate cortical development, the foetuses from timed matings of H‐Tx rats were harvested from pregnant dams killed by intraperitoneal injection of sodium pentobarbitone. Foetuses were collected at gestation days 14–16 (pre‐obstruction), 17–18 (time of obstruction) and 19–20 (post‐obstruction), and decapitated immediately after collection. Postnatal pups were killed on days 1–3 after birth by intraperitoneal injection of sodium pentobarbitone. Brains were removed from the heads of foetuses or pups and processed as described below. Ventricular dilatation, seen as increased transillumination of the skull, and the clinical appearance of an enlarged head were used to identify affected H‐Tx foetuses and pups. Histological analysis confirmed that ventricular dilatation and aqueduct obstruction had occurred in our affected samples and was not present in our defined population of normal, unaffected H‐Tx rats. We define our affected H‐Tx foetuses as those in which a blockage of the aqueduct is observed in histological sections, associated with severe ventricular dilatation (Fig. 1). It is possible that there is still a limited flow of fluid through the aqueduct, and we therefore refer to an obstruction of fluid flow rather than a blockage to allow for this possibility. Results from affected H‐Tx rats were compared with those from non‐hydrocephalic littermates (unaffected H‐Tx) and control Wistar rats.

Fig. 1 (A and B) Sagittal sections through the brain of day 21 H‐Tx foetuses showing the aqueduct obstruction that occurs in (B) affected H‐Tx but not (A) unaffected foetuses after day 18. (CK) Coronal sections through the cerebral cortex of Wistar (CE) and affected H‐Tx (FH) and unaffected H‐Tx (F and IK) foetuses. Note that F is common to affected and unaffected animals since it is not possible to distinguish them at this stage. Sections were prepared from day 18 (C, F), 19 (D, G, J) and 21 (E, H, K) foetuses and pups as described in Material and methods. Sections were stained with methyl green–pyronin and imaged as described. Sections for analysis were taken from comparable regions of brain. Sections were selected as representative specimens from at least six separate pups from different matings. The germinal epithelium (GE) is highlighted by this stain. CP = choroid plexus; CC = cerebral cortex; LV = lateral ventricle; 3V = third ventricle; 4V = fourth ventricle; AQ = cerebral aqueduct. Scale bar = 0.5 mm.

Preparation of tissue samples

To identify cells born at particular stages of development, 5‐bromo‐2′‐deoxyuridine (BrdU) (Sigma, Poole, UK) was administered to pregnant dams by intraperitoneal injection at a dose of 60 mg/kg. Foetuses and postnatal pups were recovered at set days after BrdU injection of their mothers, and the heads were fixed in ice‐cold 4% paraformaldehyde in PBS (phosphate‐buffered saline) at pH 7.3. In most cases it proved sufficient to immerse the whole head in fixative, but in the larger foetuses and postnatal pups it was necessary to expose the brain by dissecting away the scalp and skull. Fixation was carried out for a minimum of 2 h and usually overnight. Fixed heads or brains were then washed in PBS and placed in 20% sucrose solution for 12–24 h before embedding in OCT compound (Bright Instruments, Huntingdon, UK) and freezing in isopentane cooled with dry ice. Coronal or sagittal sections were cut at 15 µm on a Slee cryostat and collected onto subbed slides for staining and analysis. Coronal sections for comparative analysis contained the anatomical landmarks of the hypothalamus and the optic chiasma (line labelled PS in Fig. 2A). Photomontages were made of strips of cortex taken from the ventricular to the pial surface (along the line AP in Fig. 2B). This location of cortex was standardized for all analyses and we estimate that most sections analysed were within 10–20 sections of each other. Sagittal sections were taken at the mid‐line.

Fig. 2 The diagrams show (A) sagittal and (B) coronal views of the foetal rat brain. The line labelled PS in A shows the coronal plane of the section and the region of cortex used in all histological analyses. The line AP in B shows the sector taken across the cortex in coronal sections for all the morphometric analyses. CC = cerebral cortex; LV = lateral ventricle; 3V = third ventricle; OB = olfactory bulb; AQ = cerebral aqueduct; 4V = fourth ventricle; Ce = cerebellum; AP = anterior pituitary; MP = medial preoptic nucleus; VM = ventromedial hypothalamus; OC = optic chiasma; SC = superior colliculus.

Staining and immunocytochemical analysis of tissue sections

Histological staining was by methyl green–pyronin for nucleic acids (Moffitt, 1994). This stain is ideal for the identification of the germinal zones and proliferating cells and is also an excellent general histological stain. For immunocytochemistry, we used a mouse anti‐BrdU antibody (Nova Castro Laboratories, Newcastle, UK). Coronal sections were incubated in 10% normal serum in PBS for 2 h prior to flooding with primary antibody diluted 1 : 1000 in 10% normal serum in PBS. The optimal dilution of the primary antibody was previously established by testing a dilution series on positive control brain sections. Overnight immersion in primary antibody solution at 4°C was followed by washing in PBS (3 × 1 h) and then flooding with secondary antibody at a dilution of 1 : 200 in 10% normal serum in PBS for 2–4 h. A universal horse anti‐rabbit anti‐mouse (Vector Laboratories, Peterborough, UK) biotinylated secondary antibody was used and its location visualized by avidin–biotin–peroxidase diaminobenzidine staining (Vector Elite ABC kits). All stained sections were mounted in glycerine–albumin (Sigma) and images were captured using a Coolsnap digital camera (Princeton Instruments, Tucson, Arizona, USA) fitted to a Leica DMLB microscope. Image capture and analysis were carried out using Metaview software (Universal Imaging Corporation, Harlow, UK). Different regions of the coronal cortical section were identified (see labelling of zone margins on figures) and the number of positive cells in each zone was counted on photomontages constructed from micrographs taken at ×400 magnification across the cortex (line AP in Fig. 2B). These gave a strip of cortex 400 µm wide from the ventricular ependymal layer to the surface of the cortex next to the pia. Morphometry was also carried out using Metaview to measure intercellular distances and the width of the cortical zones.

In all experiments, a minimum of three measurements (BrdU‐positive cell counts, cortical analysis) was taken for each individual to give an average for that individual. These averages, from at least four separate foetuses of each type from separate litters, were pooled to give a mean and standard error of the mean (n ≥ 4). Statistical analysis was performed using Student’s t‐test.

Preparation of cortical cultures

Cortical hemispheres from the brains of embryonic day 20 Wistar (normal, control) and H‐Tx (divided into two groups: unaffected H‐Tx and affected H‐Tx) foetuses were dissected in ice‐cold sterile PBS, pH 7.4. The tissue was digested in 0.25% trypsin–EDTA (Sigma) for 20 min at 37°C. Following incubation, trypsin–EDTA was inactivated with Neurobasal medium (Gibco, Paisley, UK) containing B27 medium supplement (Gibco). The suspension was subsequently centrifuged at 1700 r.p.m. for 5 min and resuspended in fresh neurobasal medium. The tissue was then dissociated through sterile tips of decreasing bore (3 mm, 2 mm, 1 mm) followed by centrifuging at 1700 r.p.m. for 5 min. The cells were resuspended in fresh medium and counted on a haemocytometer. The viability of the cells was >95% when tested by trypan blue exclusion.

Proliferation assay

The dissociated cells were plated in poly‐d‐lysine (0.05 mg/ml)‐coated 96‐well plates at a density of 1 × 104 cells/ml in neurobasal medium, which preferentially supports progenitor and neuronal cell types (not glial cells), containing B27 supplement, 2 mM glutamine and penicillin–streptomycin (0.1 mg/ml). The cultures were maintained at 37°C in 5% CO2. Proliferation of cells was determined using the fluorescence‐based Vialight high‐sensitivity cell proliferation and cytotoxicity kit (Lumitech, Nottingham, UK) according to the manufacturer’s instructions. Measurements were made on a Multilabel Counter (Wallac Victor 2 1420, Perkin Elmer, Shelton, Connecticut, USA). Results are expressed as mean luminescence ± standard error of the mean (n = 3) of three experiments involving at least three litters, each performed in triplicate. Statistical analysis was performed using Student’s t‐test.

In experiments to determine the effect of CSF on Wistar cortical progenitor and neuronal cell cultures, CSF was removed from the cisterna magna of unaffected H‐Tx and Wistar foetuses or lateral ventricles of affected H‐Tx foetuses at embryonic day 20. The samples were spun twice at 13 000 r.p.m. in a microcentrifuge for 15 min each time. The supernatant was removed, snap‐frozen in dry ice and stored at –80°C. This CSF was included in the assays at 10% v/v in normal culture medium. Results are expressed as mean luminescence ± standard error of the mean (n = 3) of at least three experiments involving at least three litters and separate CSF collections, each performed in triplicate.

Results

Alterations in gross cortical morphology in affected H‐Tx rats

Coronal sections through the brains of the developing Wistar and unaffected and affected H‐Tx rat foetuses were analysed to determine the effects of obstruction of CSF flow on gross cortical morphology. The most apparent difference was in ventricular size in the latter part of gestation (Fig. 1). There was no difference between any of the foetal brains analysed until day 18, when there was observable ventricular dilatation in affected H‐Tx foetuses. In control Wistar rats, foetal cortical development followed the normal pattern of decreasing ventricular size correlated with growth of the brain parenchyma (Fig. 1C–E). In the H‐Tx rats the morphology of the brain up to day 18 was similar to that of the Wistar rats (Fig. 1F) and in the unaffected pups, in which aqueduct stenosis does not occur, there was a similar progression to decreased ventricular size as the brain grew (Fig. 1F, J and K). In stark contrast to these results, the ventricles of the affected H‐Tx rat pups remained enlarged through the same growth period (Fig. 1F–H); furthermore, there was progressive enlargement of the ventricles in the subset of affected foetuses that continued through to the postnatal period (Fig. 1G and H). Enlargement was observed in both the lateral and the third ventricle. In addition, the cortical mantle was conspicuously thinner in affected animals than in Wistar and unaffected H‐Tx foetuses from day 18 onwards. These observations can be accounted for by an accumulation of fluid in the ventricular system following constriction and eventual blockage of the cerebral aqueduct between the third and fourth ventricles (Fig. 1A and B).

Analysis of the changes in cortical morphology in affected H‐Tx rats

Cortical morphology was examined in more detail in order to pinpoint which aspects of development were abnormal in affected H‐Tx rats. Previously, we had also looked for evidence of any brain damage or cell death, but none was found (Miyan et al., 1998); analysis of the photomontages of comparable sections through the cortex demonstrated that there was a striking retardation of the growth of the cortex of the affected H‐Tx rat pups (Fig. 3). In the unaffected H‐Tx rats (i.e. those in which the aqueduct was not obstructed and CSF flow was apparently normal) the thickness of the cerebral cortex matched that of the normal Wistar rats. Quantitative measurements of cortical thickness in several samples were performed in order to assess the extent of this growth deficiency. Until the point of obstruction of CSF flow, there was no difference in the thickness of the cortex between the H‐Tx and Wistar brains. However, obstruction in affected H‐Tx rats was associated with significant retardation of growth, i.e. decrease in cortical thickness (by 30–60%) compared with normal Wistar and unaffected H‐Tx rats (Fig. 4A). The cortex of the affected H‐Tx rats did continue to grow over the following days but at all points (after day 18) was significantly reduced (P < 0.001) in size compared with both normal unaffected H‐Tx foetuses and Wistar controls.

Fig. 3 Photomontages of methyl green–pyronin‐stained coronal sections across the thickness of the cortical mantle from the brains of gestational day 19 (AC) and 2‐day‐old (DF) (postnatal) Wistar (A and D), unaffected H‐Tx (B and E) and affected H‐Tx (C and E) rats. Montages were selected as representative specimens from at least three separate animals from different litters. Methyl green–pyronin is a specific stain for nucleic acids and highlights areas in the section containing proliferating cells, in this case the germinal epithelium (or germinal matrix). As an alternative to Nissl staining, it also allows identification of different layers in the cortex and, at this developmental stage, highlights the following layers: GE = germinal epithelium; IZ = intermediate zone; CP = cortical plate; A = marginal zone. V = ventricular surface. Scale bar = 200 µm.

Fig. 4 The thickness of the (A) cerebral cortex (CC) and (B) germinal epithelium (GE) was measured in comparable coronal sections from Wistar rats (filled diamonds) and unaffected (open squares) and affected (open triangles) H‐Tx rats at days 14–24 days after conception. From day 18, the growth of the cortex and thickness of the germinal epithelium were significantly reduced (P < 0.001 for all time‐points) in affected H‐Tx compared with unaffected H‐Tx and Wistar controls, except for the thickness of the germinal epithelium at day 21. At this single time‐point, the H‐Tx group was significantly different (P < 0.001) from the unaffected H‐Tx but not the Wistar group. Data points are mean and standard error of the mean (n = 5) of measurements from five foetuses from separate matings. Significance values are shown for comparisons of affected with unaffected H‐Tx data points. **P ≤ 0.01; ***P ≤ 0.001.

Another feature of the affected H‐Tx cortex, assessed morphologically 2 days after birth, is that the thickness of the germinal epithelium also appeared to be reduced compared with the normal control animals and unaffected littermates. Quantitative assessment of this feature was also performed over the 14‐ to 24‐day period after conception. In normal development there is a steady reduction in the size of the germinal epithelium following a peak at day 16, but in the H‐Tx rats the thickness of the germinal epithelium was significantly reduced following obstruction of CSF flow on day 18. This reduction was most evident at gestational days 19–20 and postnatal days 1–2, time‐points corresponding to a burst of germinal matrix stem‐cell activity producing neuronal and glial precursors, respectively, in normal animals (Fig. 4B).

The reduction in the overall size of the cortex and in the thickness of the germinal epithelium demonstrates what appears to be a significant effect of CSF obstruction on the growth of the cerebral cortex in affected H‐Tx foetal brains. One possible explanation of the defective cortical development in H‐Tx rats could be malfunction of the germinal matrix function, and this will be discussed later.

Is the defective cortical development due to compression effects?

CSF obstruction and accumulation of fluid in the ventricular system may lead to raised ICP, which is associated with compression and stretching of the cerebral cortex. No rise in ICP has been reported in affected H‐Tx rats until 10 days after birth, when the skull plates are thought to fuse (Jones and Bucknall, 1987; Kaiser and Jones, 1991; Jones and Lopman, 1998). However, in order to establish whether compression and/or stretching were occurring in the foetal brain after CSF obstruction, we measured intercellular distances between neurones in different regions of the cortex in the dorsoventral (compression) and lateral (stretching) planes. As expected, there were no significant differences between the experimental groups (Wistar, unaffected H‐Tx and affected H‐Tx) in either the dorsoventral or the lateral plane in any region of the cortex when measurements were taken at day 17 (n = 5 for each experimental group). Similarly, we found no evidence for any compression or stretching of the cerebral cortex in affected H‐Tx compared with unaffected H‐Tx or Wistar brains at day 21 (3 days after obstruction). There were no significant differences between intercellular distances in any region of the cortex (frontal, parietal or occipital) measured at this age (n = 5 for each experimental group).

Role of germinal matrix cell proliferation and differentiation in defective cortical development in H‐Tx rats

One explanation for the defective cortical development observed in the H‐Tx rat cortex is that the processes of proliferation and differentiation of multipotential stem and progenitor cells in the germinal matrix are abnormal. Alternatively, the migration of the newly formed neurones and glia from the germinal matrix into the intermediate zone and cortical plate may be defective.

In this study, in order to assess in vivo germinal matrix function, pregnant females were injected with BrdU. Time‐mated H‐Tx and Wistar females were injected on day 17 of gestation to analyse cells generated before CSF obstruction, and on day 19 of gestation to analyse cells born after CSF obstruction. No significant differences were observed in the pattern of staining, the number of labelled cells or the pattern of migration of labelled cells between Wistar and unaffected H‐Tx foetuses labelled before CSF obstruction (Fig. 5A and B). However, in affected H‐Tx animals there were significant differences in the number and distribution of labelled cells generated after CSF obstruction (Fig. 5C–E). In cortical sections of affected H‐Tx rats there was a decrease in the total number of labelled cells and a decrease in the number of cells in the germinal matrix. Both Wistar (Fig. 5C) and unaffected H‐Tx (Fig. 5D) brains had BrdU‐labelled cells in the germinal epithelium, cells scattered through the intermediate zone and cortical plate and a concentration of labelled cells at the dorsum of the cortex. In affected H‐Tx rat brains (Fig. 5E), very few labelled cells remained in the germinal epithelium or intermediate zone. However, a band of labelled cells was present in the dorsum of the cortex, similar to that seen in the Wistar and unaffected H‐Tx rats. Thus, the most significant effect appears to be a loss of labelled cells in the germinal matrix of the brains of affected H‐Tx rats.

Fig. 5 Photomontage of BrdU antibody‐stained coronal sections through the cerebral cortex of Wistar (A and C), unaffected H‐Tx (B and D) and affected H‐Tx (E) day 18 foetuses (A and B) and pups 1 day after birth (CE), whose mothers were injected with BrdU on day 17 (A and B) or day 19 of gestation (CE). Twenty‐four hours after injection, all labelled cells were located in the germinal epithelium and intermediate zone (A, B). No difference was observed between Wistar and H‐Tx rats at the earlier time‐point (A and B). At the later, post‐obstruction time‐point, the affected H‐Tx foetuses showed a similar band of labelled cells within the cortical plate but very few labelled cells were present in the intermediate zone and germinal matrix compared with the Wistar and unaffected H‐Tx rats. Montages were selected as representative specimens from at least three pups from separate litters. GE = germinal epithelium; IZ = intermediate zone; CP = cortical plate; MZ = marginal zone. The arrows indicate the borders between the different zones of the cortex. Scale bar = 200 µm.

A more detailed analysis of BrdU‐labelled cells was performed by counting the number of positive cells in the germinal epithelium, the intermediate zone and the cortical plate at specific time‐points after injection of BrdU. Cells generated on day 17 appeared initially in the germinal matrix and intermediate zone and then spread into the cortical plate region (Fig. 6 and Table 1). At all time‐points, significant numbers of labelled cells were present in the germinal epithelium. In H‐Tx brains there was no initial difference up to day 18 in the number and location of labelled cells compared with Wistar brains. However, at later time‐points after the obstruction of CSF flow had occurred, there was a decrease in the number of labelled cells present in the affected H‐Tx germinal epithelium. Another feature of the BrdU staining patterns in the affected H‐Tx animals was that all the labelled cells appeared to migrate out of the germinal epithelium into the intermediate zone and the cortical plate. This abnormality in the function of the germinal matrix can be more clearly seen in the data from the brains of foetuses and pups whose dams were injected with BrdU at day 19 of gestation, when cells generated after the obstruction has occurred are labelled (Fig. 7 and Table 2). The pattern of staining and cell migration in the Wistar and unaffected H‐Tx cortex was similar to that of cells generated before the obstruction (Fig. 7). However, in affected H‐Tx animals there was again a significant overall decrease in the number of labelled cells and the germinal epithelium was cleared of these labelled cells more rapidly, suggesting that they may migrate earlier than normal into the cortex.

Fig. 6 The numbers of BrdU‐positive cells in the germinal epithelium (GE), intermediate zone (IZ) and cortical plate (CP) of the cerebral cortical sections were counted. Sections were taken from gestational days 18, 19 and 21 and postnatal days 2 and 3 (AE, respectively) Wistar (black bars), unaffected H‐Tx (white bars) and affected H‐Tx (hatched bars) rats whose pregnant dams had been injected with BrdU on day 17 after mating. Results are mean ± standard error of the mean for four separate injections.

Fig. 7 The numbers of BrdU‐positive cells in the germinal epithelium (GE), intermediate zone (IZ) and cortical plate (CP) were counted in comparable sections of the cerebral cortex from Wistar (black bars), unaffected H‐Tx (white bars) and affected H‐Tx (hatched bars) rat foetuses and pups on days 20–23 after mating (AD, respectively) after the pregnant dams had been injected with BrdU on day 19 after mating. Results shown are mean ± standard error of the mean for four separate injections.

View this table:
Table 1

Probability values from t‐tests carried out on data comparing labelled cortical cells generated on day E17 in affected and unaffected H‐Tx and Wistar foetuses

DayRegionP values to four decimal places
W vs N‐HTxN‐HTx vs H‐HTxW vs H‐Htx
18GE**NANA
IZNSNANA
CP**NANA
19GE*********
IZ*******
CP************
21GE***********
IZ********
CP*******
22GENS********
IZ*********
CP*********
23GENS********
IZ********
CP**********

No differences were evident between foetuses collected on day 18 in the litters examined. NA = affected and non‐affected H‐Tx could not be distinguished at these time‐points; NS = no significant difference; W = Wistar rats; N‐HTx = unaffected H‐Tx rats; H‐Htx = affected H‐Tx rats. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

View this table:
Table 2

Probability values from t‐tests carried out on data comparing labelled cortical cells generated on day 19 in affected and unaffected H‐Tx and Wistar foetuses

DayRegionP values
W vs N‐HTxN‐HTx vs H‐HTxW vs H‐Htx
19GE**********
IZ************
CP*****NS
21GE**********
IZ*****NS
CPNSNSNS
22GENS********
IZNS********
CP******
23GE*********
IZ********
CP**********

NS = no significant difference; W = Wistar rats; N‐HTx = unaffected H‐Tx rats; H‐Htx = affected H‐Tx rats. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Growth of cultured germinal matrix cells

On the basis of the data in Figs 5–7, it appears that there is a change in the activity of the germinal epithelium of affected H‐Tx rats that results in the generation of fewer neuronal progenitors. One explanation of this abnormal development is that the obstruction of CSF flow itself in affected H‐Tx foetuses is a critical factor. CSF may contain growth regulators or other signalling molecules that are required for germinal matrix stem and progenitor cells to behave normally. Obstruction of the normal flow of CSF would thus perturb normal development by altering the normal concentrations of these molecules.

In order to test this hypothesis, we examined the behaviour of the cortical cells in in vitro tissue culture. Initially, day 20 cortical cells were extracted from all experimental groups and placed into culture in artificial medium with serum but in the absence of any added growth factors. Proliferation was analysed 24, 48 and 96 h after extraction (Fig. 8A). Proliferation was significantly greater in cultures of affected H‐Tx cells than in either the unaffected H‐Tx or Wistar cultures at both 48 and 96 h. There was no difference between the unaffected H‐Tx and Wistar cultures at 48 h, but at 96 h proliferation was significantly increased in unaffected H‐Tx compared with Wistar controls.

Fig. 8 (A) Cortical cells from day 20 Wistar (n = 10), unaffected H‐Tx (n = 5) and affected H‐Tx (n = 6) foetuses were plated in poly‐d‐lysine‐coated 96‐well plates at a starting density of 1 × 104 cells/ml in supplemented neurobasal medium. The cultures were maintained at 37°C in 5% CO2. The extent of proliferation at the time‐points shown was measured using a Lumitech Vialight high‐sensitivity cell proliferation and cytotoxicity kit according to the manufacturer’s instructions. The affected H‐Tx rats differed significantly from the Wistar and unaffected H‐Tx rats at both 48 (P < 0.001 and P < 0.05, respectively) and 96 h (P < 0.001 and P < 0.005, respectively). (B) Day 20 cortical cells from Wistar rats were isolated and plated as described above except that additions as shown were made to the culture media. The extent of proliferation was determined after 96 h of incubation. All results shown are mean ± standard error of the mean (n = 5–10) for experiments from separate matings, each performed in triplicate. **P < 0.01; ***P < 0.001. N‐HTx = unaffected H‐Tx rats; H‐HTx = affected H‐Tx rats.

An explanation for this increase in proliferation observed in the affected H‐Tx cultures after removal from their in vivo environment is that the cortical cells are under an inhibitory influence in the cortex and that once this is lifted enhanced proliferation occurs. This inhibitor could be contained within the CSF. In order to test this, cortical cells from normal Wistar rats were isolated and placed into culture as described above, and 24 h later the culture medium was replaced with medium containing 10% CSF extracted from the brains of each of the experimental groups. This was taken as time zero. In control cultures, BSA (bovine serum albumin; final concentration 1%) was added. Following a further 96 h of incubation in the presence of CSF, the extent of proliferation in the cultures was assessed. In agreement with the data in Fig. 8A, under control conditions (BSA or PBS addition) there was an ∼3‐fold increase in cell number (Fig. 8B). There was a similar increase in cell number when CSF from either Wistar or unaffected H‐Tx rats was included in the cultures. However, when cultures were treated with CSF from affected H‐Tx rats, no such increase was observed. In fact, the number of cells in the culture was not significantly different from the starting cell number, suggesting that proliferation had been blocked.

Discussion

In this study we defined defects in cortical development that occur in the hydrocephalic H‐Tx rat foetus. These defects develop in affected foetuses after the flow of CSF is obstructed on day 18 of gestation. Prior to this time there are no distinguishing features that allow identification of a pup as being affected. We attempted to determine the origins of these developmental defects by following the generation and movement of neuronal progenitor cells over the time spanning the stage of defective development. It is unlikely that the defective cortical development we observe is a direct consequence of the genetic defect in these inbred rats, because they are homozygous. Until the genetic defect is fully characterized, we can consider a number of possibilities, one being that there is an interaction between a genetic defect present in all of the foetuses and a maternal factor, which may cross the placental barrier into the foetal blood. Since it has been shown that an increased level of normal growth factors in the CSF results in hydrocephalus in normal rats (Johanson et al., 1999), it can be hypothesized that the maternal factor might act by stimulating abnormal levels of growth factors in the foetal CSF. Whilst it is possible that the genetic defect alone may be directly responsible for the abnormal cortical development in the brain of affected H‐Tx rats, it is unlikely because our data strongly support the view that the effects we observe are due to a factor or factors (either abnormal or normal, but present at abnormal levels) in the CSF and result from obstruction of CSF flow. In all of our experiments we used both unaffected littermates and foetuses from Wistar rats as controls.

The BrdU labelling experiments demonstrate that, in normal development, some of the labelled proliferating neuronal progenitor cells remain in the germinal epithelium whilst other labelled cells migrate into the cortex. In affected H‐Tx brains, although migration does occur, labelled cells do not remain in the germinal epithelium and fewer labelled cells are produced. It seems likely that the progenitor cells are migrating prematurely from the germinal epithelium at an earlier stage of differentiation. The consequence of this effect is that, overall, fewer cells are produced from the germinal epithelium, resulting in the developmental retardation of the cortex. There is thus evidence that the activity of the cells in the germinal matrix is in some way altered in affected H‐Tx rats. One possible reason for the thinning of the cortex is that it is due to altered compression or stretching of the brain. However, in this rat model there is no rise in ICP until day 10 after birth, when the skull plates are thought to fuse (Jones, 1985; Jones and Bucknall, 1987; Kaiser and Jones, 1991; Jones and Lopman, 1998). Our data support the lack of a pressure effect since we were unable to find any changes in intercellular distances in any region of the cortex in either the compression or the stretch plane. Before the bony plates of the skull fuse, distension of the vault is possible, and this seems to prevent a significant rise in ICP even though there is a fluid build‐up of CSF in the lateral and third ventricles. It is also well established that fluid accumulation is not immediately associated with a rise in ICP. There is a range of ventricular enlargement over which ICP does not increase significantly from normal (Guthkelch, 1972).

Another possible reason for the thinning of the cortex of affected H‐Tx rats is that there is excessive cell death in the affected H‐Tx foetal cortex. Using an antibody to single‐stranded DNA, we have previously looked for evidence of apoptosis in the postnatal hydrocephalic brain (Miyan et al., 1998). Using this antibody, we examined coronal sections taken through the cortex of prenatal Wistar and unaffected and affected H‐Tx rat foetuses from gestational days 16–21 and also looked at postnatal pups up to age 15 days. We found that there was no evidence for increased apoptosis in the affected H‐Tx rat brain compared with either unaffected H‐Tx or Wistar rats until postnatal day 10. There is thus no evidence to show that accumulation of CSF in the affected H‐Tx brain elicits increased cell death by apoptosis.

A third explanation for the abnormal development is that it is a direct consequence of the obstruction of CSF flow. There are several possible reasons. The CSF may contain growth factors, inhibitors or other signalling molecules that are required for the normal activity of the stem and progenitor cells of the germinal matrix. Accumulation of these growth factors, some or all of which may be released from the choroid plexus, might become inhibitory to germinal matrix activity. The factors may additionally be involved in physiological changes leading to CSF obstruction and hydrocephalus, as it has been shown that changes in the concentration of some of these growth factors can lead to serious consequences of imbalance in CSF production and absorption. Both fibroblast growth factor 2 (FGF‐2) (Johanson et al., 1999) and transforming growth factor β (TGF‐β) (Moinuddin and Tada, 2000) are known to induce hydrocephalus when present at an elevated level in the CSF. Alternatively, there could be a build‐up of waste products from the brain parenchyma in the CSF, resulting in inhibition of germinal matrix function. The CSF may also be an important signalling pathway linking the germinal epithelium with the dorsal surface of the cortex, which is involved in coordinating the activity of the germinal epithelium and the migration of cells into the cortex (Johnson et al., 1992; Marin‐Padilla, 1998; Nicholson, 1999).

The CSF has the potential to act as a signalling pathway for physiological control systems, as it has been demonstrated to contain molecules such as corticotropin‐releasing factor, adrenocorticotropin, leptin, interleukins and leukotrienes. Moreover, CSF contains concentrations of neurotransmitters and neuropeptides that change with the physiological activity of the animal. A number of studies have identified CSF as a carrier of important cytokines, such as TGF‐β, nerve growth factor (NGF), brain‐derived neurotrophic factor (BDNF), neurotrophin 3 (NT‐3), insulin‐like growth factor (IGF) and glial‐derived neurotrophic factor (GDNF), which are present at specific times during development under specific physiological conditions or challenges (Kasaian and Neet, 1989; Johnson et al., 1992; Kitazawa and Tada, 1994; Mogi et al., 1996; Suzaki et al., 1997; Heinze et al., 1998; Korhonen et al., 1998; Arnold et al., 1999; Ikeda et al., 1999; Riikonen et al., 1999; Van Setten et al., 1999; Grouzmann et al., 2000; Whalen et al., 2000). Thus, it is probable that the CSF contains growth regulators that have the potential to affect the function of the foetal germinal matrix.

A recent study (Proescholdt et al., 2000) demonstrates that compounds carried in the CSF not only circulate rapidly (within seconds to minutes) through the CSF pathway but also have fast (within minutes to hours) access to most regions of the brain itself, gaining entry across the pia and ependymal layers through gap junctions, other channels and modulated junctions and by active transport or diffusion. Morphological and electron microscopic studies also show that some of the cells underlying the ependyma (presumably neurones or glia) have processes passing between ependymal cells and in contact with the CSF (J. Parvavelas, personal communication). This view of the CSF is supported by experiments demonstrating that intraventricular injections of growth factor can modulate the development of the cortex (Fukumitsu et al., 1998) and can result in defective development or abnormalities (Johanson et al., 1999; Moinuddin and Tada, 2000). Together, the results of these studies indicate that the CSF is an important and rapid route for molecular signal transfer between different areas of the brain, including molecules delivered into the CSF by the choroid plexus.

Our in vitro analysis of the growth of cultured germinal matrix stem cells supports the hypothesis that the CSF is responsible for the defective development of the cortex of affected H‐Tx rats that is seen after the obstruction of CSF flow. Cells removed from the affected H‐Tx cortex are able to proliferate when in culture, a feature they share with control cells removed from normal rats when not exposed to hydrocephalic CSF. Thus the cells from affected H‐Tx foetuses do not have a defective ability to proliferate; in fact, once removed from their in vivo environment they display an enhanced ability to proliferate in culture. One explanation for this is that, in vivo, these cells are under an inhibitory influence. Evidence for CSF inhibiting proliferation in vivo is provided by the demonstration that CSF extracted from affected H‐Tx foetal brains has the ability to inhibit the proliferation of progenitor cells of normal animals. In contrast, CSF extracted from unaffected H‐Tx or normal animals has no effect on the growth of cells extracted from any of the rats, including affected rats. The inhibition is not a cytotoxic effect of the CSF because the cells remain viable when cultured with the extracted CSF from affected animals; it is simply that they do not proliferate.

Our studies strongly support the view that obstruction of CSF pathways leads to the accumulation of a substance proximal to the obstruction that is inhibitory or becomes inhibitory as its concentration rises. It is common clinical knowledge that early‐onset hydrocephalus results in very similar neurological deficits in older patients irrespective of its cause, and it can therefore be argued that the common factor is some substance in the accumulating CSF. If this argument is true, it follows that the genetic defect in the H‐Tx rat, which in all probability is responsible for the obstruction, is not responsible for the resulting effect on cortical development. Intrauterine intervention to drain CSF from the ventricular system in human foetuses aged ≥24 weeks has not been conspicuously successful (Manning, 1985; Bannister, 1986). Our studies suggest that a more successful approach would be to reduce the amount of or eliminate the inhibitory substance that is present in the CSF. Before considering how this might be done, it is necessary to identify the substance, and the first stages of this work are already being undertaken in our laboratories.

In conclusion, the data from this study support our hypothesis that it is obstruction of CSF flow in the foetal stages of early‐onset hydrocephalus which is responsible for the defective cortical development seen in our morphological and histological studies, and this in turn results in the abnormal germinal matrix stem and progenitor cell function demonstrated in our in vitro experiments. The body of evidence presented above suggests that there is an important role for CSF flow in the development of the cerebral cortex and that disruption of this flow alone is sufficient to cause the cortical deficiencies observed in cases of early‐onset hydrocephalus. Current treatment for CSF obstruction and hydrocephalus involves diversion of fluid from the lateral ventricles. This treatment addresses the later problems with raised ICP, which causes compression and stretching and consequent damage to the brain, but does not address the loss of the CSF’s vital involvement in brain development. The lack of fluid flow may underlie the retention of 80–90% neurological deficits seen in patients suffering from early‐onset hydrocephalus, even though fluid diversion was carried out soon after birth (McAllister and Chovan, 1998) or in utero. Future work will investigate the factors in the CSF controlling the activity of the cells in the germinal epithelium during cortical development.

Acknowledgements

This work was funded by grants from The Wellcome Trust and The Royal Society. F.M. and M.P. are Iranian National Scholars. We thank Nick Ritchie and Janet Wilson‐Walsh for technical assistance.

References

View Abstract