Hydrocephalus compacted cortex and hippocampus and altered their output neurons in association with spatial learning and memory deficits in rats

Abstract

Hydrocephalus is a common neurological disorder in children characterized by abnormal dilation of cerebral ventricles as a result of the impairment of cerebrospinal fluid flow or absorption. Clinical presentation of hydrocephalus varies with chronicity and often shows cognitive dysfunction. Here we used a kaolin‐induction method in rats and studied the effects of hydrocephalus on cerebral cortex and hippocampus, the two regions highly related to cognition. Hydrocephalus impaired rats' performance in Morris water maze task. Serial three‐dimensional reconstruction from sections of the whole brain freshly froze in situ with skull shows that the volumes of both structures were reduced. Morphologically, pyramidal neurons of the somatosensory cortex and hippocampus appear to be distorted. Intracellular dye injection and subsequent three‐dimensional reconstruction and analyses revealed that the dendritic arbors of layer III and V cortical pyramid neurons were reduced. The total dendritic length of CA1, but not CA3, pyramidal neurons was also reduced. Dendritic spine densities on both cortical and hippocampal pyramidal neurons were decreased, consistent with our concomitant findings that the expressions of both synaptophysin and postsynaptic density protein 95 were reduced. These cortical and hippocampal changes suggest reductions of excitatory connectivity, which could underlie the learning and memory deficits in hydrocephalus.

Introduction

Hydrocephalus is a pathological condition characterized by abnormal dilation of cerebral ventricles as a result of the obstruction of the flow or malabsorption of cerebrospinal fluid (CSF). The etiology of hydrocephalus can be genetic involving abnormality in ventricular zone neural stem cells during embryonic development 27, 54, congenital caused by malformation or acquired mostly by intraventricular hemorrhage or infections 48, 64. Childhood hydrocephalus is associated with various cognitive deficits 43. In the acute stage, ventricles dilate and intracranial pressure increases and causes headaches, vomiting, nausea, sleepiness or coma. In the chronic stage, CSF dynamics normalize with the enlarged cerebral ventricles, if untreated, can result in cognitive impairment in memory, attention, information processing and executive function similar to subcortical type dementia 56. Cognitive dysfunctions are attributable to destruction of neural circuits. In this regard, hydrocephalus is known to induce axonal degeneration and demyelination of periventricular white matter 12, 14, 16 and change cerebral neurotransmitter level 20. However, whether hydrocephalus affects central synaptic density and neuronal structure remains largely unknown.

The neuronal pathology of the hydrocephalic cortex is easily overlooked or reported as subtle 68, 69. Changes of the hippocampus in either human or experimental hydrocephalus have rarely been examined 41. In both cerebral cortex and hippocampus, pyramidal neurons are output neurons and connect to neural circuits controlling cognition, memory and motor functions. Alterations of cortical and hippocampal pyramidal neuronal dendrites are often studied in disorders such as Alzheimer's disease and trauma 25, 37; however, we know very little about the effects of hydrocephalus on the dendrites of these two groups of neurons. Dendritic structural changes of cortical pyramidal neurons had been studied in genetic model of congenital hydrocephalus 28, 48, 50. In the nongenetic hydrocephalus category, Golgi staining method has been applied to examine the dendrites of visual cortical pyramidal neurons in posthemorrhagic hydrocephalus infants 62, and association 42 and parieto‐occipital 49 cortical neurons in kaolin‐induced, hydrocephalus rats. However, changes of dendritic spines were quantitated in the human infant visual cortical neurons only 62. Another study of the feline experimental hydrocephalus model reported the degradation of dendrites and decrease of synaptic contacts in the hippocampal pyramidal neurons 41. A more systematic comprehensive evaluation of the changes of dendritic arbors and spines of cortical and hippocampal pyramidal neurons under different chronicities of hydrocephalus is awaited.

Here we used the intracisternal kaolin injection rat hydrocephalus model, which models post‐meningitis or post‐hemorrhage hydrocephalus 13, to study the effects of hydrocephalus on cerebral cortex and hippocampus. Kaolin was injected at postnatal day 21 to correlate with human infancy 55 when hydrocephalus most commonly develops clinically. Hydrocephalic rats were studied 1 and 5 weeks post‐induction to represent the acute and chronic hydrocephalic stages, respectively, because the ventriculomegaly reached maximum and remained stable 3 weeks following kaolin injection 18, 70. Cognitive changes and anatomic and biochemical alterations in the primary somatosensory cortex and dorsal hippocampus were explored. The spatial learning and memory of these rats were evaluated with a single‐day Morris water maze (MWM) task. The brains of the animals were freshly frozen in situ with skulls and sectioned serially for three‐dimensional reconstruction to evaluate tissue volumetric changes. On the other hand, the intracellular dye injection technique and subsequent three‐dimensional reconstruction were employed to study the dendritic arbors and spines of the layer III and layer V cortical pyramidal neurons and CA1 and CA3 hippocampal pyramidal neurons. Biochemically, the expressions of synaptophysin, a presynaptic vesicle protein as presynaptic terminal marker and postsynaptic density protein 95 (PSD95), an excitatory glutamatergic postsynaptic density marker in both the primary somatosensory cortex and dorsal hippocampus were also examined. Evaluations of the structural and biochemical changes in these two brain areas in association with the assessment of spatial learning and memory alterations were expected to further our understanding on the pathophysiology of hydrocephalus.

Materials and Methods

Animal experiments were approved by the Animal Care and Use Committee of the Tzu‐Chi University. All efforts were made to minimize both the suffering and to reduce the number of animals used.

Animal preparation

A total of 62 three‐week‐old Sprague‐Dawley rats (45–60 g body weight; Lasco, Ilan, Taiwan) from 10 litters were studied. To induce hydrocephalus, rats were first anesthetized with inhalation of isoflurane (1.5% in oxygen). A syringe with 27‐gauge needle was inserted percutaneously into the cisterna magna and 0.06 mL of sterile kaolin (Sigma–Aldrich, St. Louis, MO, USA) suspension (250 mg/mL in 0.9% saline) was slowly injected. The rats were returned to their mothers after recovery from anesthesia and allowed to survive for 1 and 5 weeks post‐operation, representing acute and chronic hydrocephalus, respectively. Age‐matched sham‐operated controls were injected with the same volume of sterile saline solution. The body weights of these rats were recorded weekly for the duration of the study. Rats experiencing severe neurologic impairment or weight loss were euthanized (n = 3) and hence not included in the result.

Morris water maze task

Hydrocephalic rats 1 and 5 weeks after kaolin injection and their age‐matched controls were tested with a single‐day MWM task of multiple trials 15, 40. Test was done in a circular pool (152 cm in diameter, 29 cm deep) located in a lit room with visual cues. A hidden transparent platform (10 cm × 10 cm × 28 cm) was fixed in the center of southwest quadrant and submerged 1.0 cm beneath water surface with water temperature at 23 ± 1°C. The test was performed in three blocks during the day with a 2 h rest between blocks. In each block, rats were subjected to four trials. For each trial, the rat was placed in the center of the pool facing a different direction and allowed to swim until they found the underwater platform. If the rat failed to complete the task in 60 s, it was guided onto the platform for a 60 s rest before the next trial. The mean of the escape times of the four trials of each block was taken as the escape time for that trial block. Rats capable of learning showed progressively shorter mean escape latency during the course of the MWM task.

Brain preparation and sectioning for stereometry

Rats intended for whole brain sectioning were deeply anesthetized intraperitoneally with Zoletil (tiletamine 25 mg/kg and zolazepam 25 mg/kg; Virbac, Carros, France) and xylazine (10 mg/kg; Rompun; Bayer, Leuverkeusen, Germany). Animals were then decapitated, and the lower jaws and snouts quickly removed. The brain in situ with skull was processed with the Kawamoto's protocol 34 to minimize structural distortion. Methodologically, the head was rapidly frozen in hexane (−73°C) to minimize ice crystal formation. The frozen head was then placed in a stainless steel container filled with 4%–5% carboxymethylcellulose gel. The container with its contents was then frozen in hexane. The frozen block was then removed from the container and stored at −80°C until sectioning. The block was sectioned serially with a cryostat microtome (CM3050S, Leica, Wetzlar, Germany) with disposable tungsten carbide blade (Leica) in 10‐μm thickness. An adhesive film (Cryofilm type IIC, Section‐LAB Co., Hiroshima, Japan) was used in the sectioning and the subsequent staining and later transferring of individual section onto slide. Briefly, a piece of the adhesive film slightly larger than the cutting surface of the tissue block was first adhered to the cutting surface of the block. This film allowed the section to be handled in subsequent processes. A section, adhered on one side to the film, was then made. The section with its attached film was then treated with 100% ethanol for 10 s and followed by 4% paraformaldehyde for 1 minute for fixation. It was then stained with hematoxylin and eosin (H&E). After staining, the section was mounted, face down, on a glass slide with a customized mounting medium (super cryomounting medium type R2; Section‐LAB). The mounting medium was then polymerized with UV irradiation from a UV lamp (Section‐LAB). The mounted section was then ready for light microscopy.

To analyze primary somatosensory cortical thickness, the above‐prepared sections at approximately −0.36 mm relative to bregma from each animal were sampled (n = 5 for each of the hydrocephalic and age‐matched control groups). The linear vertical dimension of the right somatosensory cortex 2.25 mm lateral to the midline was measured with a 2.5× objective and a digital camera fitted with iSolution Lite (IMT Technology, Vancouver, BC, Canada).

3D reconstruction and stereometry of the cerebral cortex and dorsal hippocampus

To estimate cortical and hippocampal volumes, the contours of the cerebral cortex and dorsal hippocampus in one in every 25 sections were traced through the serial whole brain sections with a PC‐based reconstruction software (Neurolucida; MicroBrightField, MA, USA) fitted to an Olympus microscope. For the cerebral cortex, sections from bregma −0.36 to −4.86 mm, excluding the piriform cortex, and for the hippocampus, from bregma −1.8 mm to its caudal end, including the fimbria, were traced. The volume of this segment of the cerebral cortex and hippocampus of each brain were calculated by adding each section (10‐μm thickness) multiplied by the sum of sectional interval (250 μm) with Neuroexplorer (MicroBrightField). The numbers of rat brains reconstructed in each experimental group are shown in Table 1.

Table 1.

Changes in cerebral cortical and hippocampal volume in hydrocephalus.

	Acute stage		Chronic stage
	Control	Hydrocephalus	Control	Hydrocephalus
Cortical thickness (mm)	1.92 ± 0.02 (n = 5)	1.48 ± 0.05* (n = 5)	1.94 ± 0.05 (n = 5)	1.49 ± 0.09* (n = 5)
Cortical volume (mm3)	157.41 ± 2.26 (n = 3)	131.52 ± 4.61* (n = 4)	168.87 ± 4.49 (n = 4)	151.01 ± 2.80* (n = 5)
Hippocampal volume (mm3)	69.42 ± 3.90 (n = 3)	57.41 ± 2.29* (n = 4)	80.94 ± 2.13 (n = 4)	70.42 ± 1.90* (n = 5)

In each group, the control was the age‐matched sham‐operated animals.

*P < 0.05 between the marked and its age‐matched control.

Intracellular dye injection

Fixed tissue intracellular dye injection technique was used to reveal the dendritic arbors of layer III and layer V cortical pyramidal neurons of the somatosensory cortex and CA1 and CA3 pyramidal neurons of the dorsal hippocampus. Only the brains of male rats were analyzed to avoid complications from the cyclic dendritic spine density changes of females during estrous cycle 7. Numbers of rats studied in each group are listed in Table 2. At the end of the survival, rats intended for this part of the study were deeply anesthetized as above and perfused with 2% paraformaldehyde in 0.1 M phosphate buffer (PB) for 30 minutes. The rat brain removed was then sectioned with a vibratome (Technical Product International, St. Louis, MO, USA) into 400‐μm‐thick coronal slices. The brain slices were treated with 10⁻⁷ M 4′,6‐diamidino‐2‐phenyl‐indole (DAPI, Sigma–Aldrich) in 0.1 M PB for 30 minutes to make cell nuclei fluoresced blue under the filter set that visualized the yellow fluorescence of the intracellular dye Lucifer yellow (LY, Sigma–Aldrich) as previously reported 7. The intracellular micropipettes were fabricated with a micropipette puller (Sutter, CA, USA) with standard wall borosilicate capillary glass (1 mm outer diameter) with filament (Sutter). The micropipette had an approximately tip size of 0.5 μm 3. DAPI‐labeled nuclei of layer III and layer V cortical pyramidal neurons could be easily identified, for their larger sizes and characteristic layering densities as compared to other cortical layers. CA1 and CA3 pyramidal neuronal nuclei concentrated in the striatum pyramidale could be easily identified as well. For the primary somatosensory cortex, pyramidal neurons located approximately −0.36 to −2.92 mm posterior and −2.4 to −4.4 mm lateral to the bregma, and for the CA1 and CA3 hippocampus −2.16 to −3.96 mm posterior to the bregma in the stereotaxic coordinates of the rat brain were sampled. Briefly, the slices were placed in a dish on the stage of fixed stage Zeiss Axioskop equipped with epifluorescence. An intracellular micropipette filled with 4% LY in water driven by a three‐axial hydraulic micromanipulator (Narishige, Tokyo, Japan) was used to impale neuron. Constant negative current generated by an Axoclamp‐IIB amplifier (Axon, Foster City, CA, USA) was used to inject LY into selected neurons. Several neurons could be injected separately in the area of interest in each slice. The injected slices were postfixed with 4% paraformaldehyde in 0.1 M PB immediately after for 3 days. To analyze the dye‐filled cells, the injected intracellular dye LY was first immuno‐converted into nonfading reaction product following previous protocol 6. Briefly, each injected slice was first cryoprotected and carefully sectioned into 60‐μm‐thick serial sections with a cryostat microtome (CM3050S, Leica). Sections were treated with 1% H₂O₂ in PB for 30 minutes to remove endogenous peroxidase activity. After washing with phosphate‐buffered saline (PBS), sections were incubated in 2% bovine serum albumin and 1% Triton X‐100 in PBS for 1 h. They were then incubated with biotinylated rabbit anti‐LY in PBS (1:200; Molecular Probes, Eugene, OR, USA) at 4°C for 18 h. Following subsequent rinses in PBS, sections were incubated with avidin‐biotin horseradish peroxidase reagent (Vector, Burlingame, CA, USA) for 3 h at room temperature. They were then reacted with a solution containing 0.05% 3,3′‐diaminobenzidine tetrahydrochloride (DAB; Sigma–Aldrich) and 0.01% H₂O₂ in 0.05 M Tris buffer (pH 7.4) at room temperature. The stained sections were mounted on slides, dehydrated, and cover‐slipped with Permount (Fisher Scientific, Fair Lawn, NJ, USA).

Table 2.

Number of rats and neurons studied in the dendritic arbor and dendritic trunk diameter analyses.

Group	Number of rats	Number of neurons for dendritic arbor analyses				Number of neurons for dendritic trunk diameter analyses
		Cortical pyramidal neurons		Hippocampal pyramidal neurons		Cortical pyramidal neurons			Hippocampal pyramidal neurons
		Layer III	Layer V	CA1	CA3	Layer III	Layer V	CA1		CA3
Acute stage
Control	4	9	10	17	11	31	33	100		54
Hydrocephalus	6	17	12	19	11	133	221	284		178
Chronic stage
Control	4	12	9	18	7	48	48	108		34
Hydrocephalus	5	16	13	21	15	101	145	251		79

Neurons were filled with intracellular dye to reveal theirs dendritic arbors. The dendritic arbor of each neuron was reconstructed three‐dimensionally with a PC‐based software through the serial sections of the brain slice in which the neuron was filled with the dye.

Measurement of apical dendritic trunk diameter

The apical dendritic trunks of the dye‐filled neurons were visualized with 100× oil immersion objective lens. The diameter of the apical dendritic trunk at where it originated from the cell body, where the convex contour of the soma turned concave into the dendritic trunk, was measured with a digital camera fitted with iSolution Lite (IMT Technology). Number of neurons analyzed in each group is shown in Table 2.

3D reconstruction of dendritic arbors

The somatodendritic tree of the dye‐filled neuron was reconstructed through the serial sections of the injected brain slice with the three‐dimensional reconstruction software Neurolucida (MicroBrightField) configured to an Olympus microscope with motorized stage and video camera. The software yields data including dendritic length and end for subsequent analyses. Numbers of neurons reconstructed in each group are listed in Table 2.

Analyses of dendritic spines

To find out whether hydrocephalus affects the number of dendritic spines on dendrites, representative proximal and distal segments of the apical and basal dendrites of each category of cortical pyramidal neurons were counted under a 100× oil immersion objective. For layer III and layer V cortical pyramidal neurons, the sampling method and analysis were as previously described 7. For hippocampal CA1 and CA3 pyramidal neurons, distal basal dendrites in the stratum oriens and proximal and distal apical dendrites in the stratum radiatum and stratum lacunosum‐moleculare, were chosen and studied as described previously 5. Spine densities on three to five pieces of each category of the dendritic segment were measured in each neuron. The mean was the spine density of the category of dendritic segment of the neuron. Three to six neurons were sampled in each animals and the mean of these neurons represent the spine density of the particular type of neuron of the animal. Dendritic spine densities of each group were the mean ± SEM of the values of all animals studied in that group.

Western blot analysis

Like in the analyses of dendritic spine densities, only male rats were used to examine the level of synapse‐related proteins for estrogen is known to affect the expressions of synapse‐related proteins during the different phases of the estrous cycle in females 7. Methodologically, rats were deeply anesthetized with Zoletil and xylazine intraperitoneally. After decapitation, the brain was quickly removed and the grey matter of the somatosensory cortex and dorsal hippocampus were quickly dissected and frozen in liquid nitrogen. A fixed size biopsy punch (approximately 3 mm in diameter) was used to collect a tissue column containing the grey and white matter of the somatosensory cortical region. The white mater was then removed with a scalpel blade based on the color difference from the grey matter (naked eye). The remaining grey matter had a cylinder shape of approximately 2 mm in height and 3 mm in its largest diameter. Tissue samples were then homogenized in 4°C lysis buffer containing tissue protein extraction buffer (#78510) with protease (#78430) and phosphatase (#78420) inhibitors from Pierce (Rockford, IL, USA). The homogenates were centrifuged at 10,000 × g for 10 minutes at 4°C. The supernatant was taken and the protein concentration determined with BCA protein assay kit (Pierce). Equal amounts of proteins were separated on 10% polyacrylamide gels and transferred onto polyvinylidene difluoride membranes (Bio‐Rad, Hercules, CA, USA). Primary antibodies including mouse anti‐PSD95 (MAB1598; Millipore, Billerica, MA, USA), mouse anti‐synaptophysin (MAB5258; Millipore) and rabbit anti‐glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) (sc‐25778; Santa Cruz Biotechnology, Santa Cruz, CA, USA) were used. Goat anti‐mouse (AP124P; Millipore) and goat anti‐rabbit secondary antibodies (AP132P; Millipore) conjugated with horseradish peroxidase were used for visualization with enhanced chemiluminescence system (Amersham, Piscataway, NJ, USA). The densities of immunoreactive bands were analyzed with Gel‐Pro (Media Cybernetics, Silver Spring, MD, USA). Protein levels of PSD95 and synaptophysin were standardized against GAPDH as loading control. Data were expressed as the relative ratio of the value to the corresponding control animals.

Statistical analysis

Data were expressed as mean ± SEM, and analyzed with SigmaStat (Jandel Scientific, San Rafel, CA, USA). Group data were tested for normal distribution with Kolmogorov‐Simirnov test. Data from hydrocephalic rats 1 and 5 weeks after kaolin injection were compared with those of the corresponding sham‐operated age‐matched controls with unpaired two‐tailed Student's t‐test when normality was verified, or with the Mann–Whitney U‐test if the normality test failed. To compare dendritic spine densities and levels of PSD95 and synaptophysin, the data were analyzed with ANOVA followed by the Holm‐Sidak post hoc test when normality was verified or by ANOVA on ranks and Dunn's post hoc test if the normality test failed. Statistical significance was taken at P < 0.05.

Results

Cisternal kaolin injection in juvenile rats induced hydrocephalus

Rats developed varying degrees of unsteady gait and lethargic pace following kaolin injection. In addition, the animals had scruffy fur indicating impaired grooming. The heads enlarged and became dome‐shaped within a week. These are all apparent signs of hydrocephalus. These rats showed significantly slower body weight gain than their age‐matched controls (Figure 1A).

Figure 1.

The effect of hydrocephalus on body weight and MWM performance of rats. A. Shows the body weight gains of control and hydrocephalic rats over time. B–D. Show animals' performance in the single‐day MWM task. (B,C) plotted the mean escape latencies of the three trial block (I,II and III) of the one‐day MWM tasks of the acute and chronic hydrocephalic rats (n = 10 each) and their corresponding age‐matched controls (n = 8 each) as labeled. D. Shows the mean swimming speeds of the acute and chronic hydrocephalic and age‐matched control groups (n = 6 each). Data are mean ± SEM. *, P < 0.05 and **, P < 0.001 between the marked and its age‐matched controls; $, P < 0.05 between the marked and its age‐matched controls of the same trial block; #, P < 0.05 between the marked and the first trial block of each experimental group.

We then prepared the whole brain sections of these animals in situ with skull for analyses. With a special supporting film technique, serial coronal sections of the whole brain could be consistently prepared as shown in Figure 2. Figure 2A1,B1 are representative low‐powered micrographs of the coronal sections of 4 and 8‐week old control rats, that is, age‐matched controls of the acute and chronic hydrocephalus, respectively. There was no apparent distortion or tissue damage and the relative position of all brain structures within the skull such as the cerebrum and hippocampus was well preserved. Kaolin injection induced apparent ventriculomegaly, enlargement of both lateral and third ventricles (3V in Figure 2A2,B2), and in addition reductions in the size of the dorsal hippocampus and the thickness of the cerebral cortex (double headed arrows) under both acute (Figure 2A2) and chronic (Figure 2B2) hydrocephalic conditions as compared to respective age‐matched controls. Chipping or peeling of tissue and edema were often observed in the corpus callosum indicating white matter damages. Thus, cisternal kaolin injection in juvenile rats is a reasonable model for studying infantile hydrocephalus.

Figure 2.

Representative cross‐sectional micrographs of the hydrocephalic and control brains. Representative H&E‐stained images of the hydrocephalic and control brains in the acute (upper row) and chronic (lower row) stages are illustrated. The brain was freshly frozen in situ with skull (arrow) and then sectioned into 10‐μm‐thick coronal sections. A1,B1. From age‐matched controls of the acute and chronic rats, respectively. A2,B2. From acute and chronic hydrocephalic rats. Note the dilation of the ventricles, thinning of the cortex (double‐headed arrows) and compaction of the dorsal hippocampus in the hydrocephalic brain. 3V = third ventricle. Scale bar in A1 = 5 mm, also for A2; in B1 = 5 mm, also for B2.

Hydrocephalus compromised rats' MWM learning

With this, we first asked whether hydrocephalus impaired learning and memory. Unlike adults, the steady body growth of juvenile rats and the progression of the induced hydrocephalus over time could complicate the outcome of a conventional MWM task conducted over several days. To avoid this, we shortened the test to a day consisting of three blocks of four trials each. Rats 1 and 5 weeks after kaolin injection, representing acute and chronic hydrocephalic stages, respectively (n = 10 each), were tested.

In the acute hydrocephalic rats, the escape latency of the first block of the task was no different from that of the age‐matched controls (n = 8, Figure 1B). Hydrocephalic animals showed no significant improvement in escape latency in the second and third trial blocks of the task (Figure 1B). On the contrary, age‐matched controls showed significantly shorter escape latencies during the second and third trial blocks (#, P < 0.05 vs. the first trial block, Figure 1B). Between acute hydrocephalic and age‐matched control rats, the former performed significantly poorer than the later on the second and third trial blocks (asterisks, P < 0.05, Figure 1B).

At the chronic stage, hydrocephalic rats showed significantly improved escape latency by the third trial block (#, P < 0.05 vs. the first trial block of the task, Figure 1C). Their age‐matched controls started to show improved escape latencies on the second trial block of the task (n = 8, #, P < 0.05, Figure 1C). Chronic hydrocephalic rats performed poorer than their age‐matched controls in the first as well as the second and third trial blocks (asterisks, P < 0.05, Figure 1C). There was no significant difference in swimming speed between the hydrocephalic and the age‐matched control rats in either the acute or chronic stage (n = 6 in each group, Figure 1D).

A comparison between the performance in the first trial block of the 4‐week and 8‐week‐old control rats showed that the latter performed better than the former (compared between Figure 1B,C, P < 0.001). However, no difference in swimming speed was detected between the control rats at these two ages either (Figure 1D). These support our concern that there could be an influence of age and/or other growth‐associated factors on rats' MWM performance and justified our choice of adopting a single‐day MWM task in the present study.

Hydrocephalic rats had reduced cortical and hippocampal volume

We then asked whether learning and memory impairments in acute and chronic hydrocephalus were accompanied by brain structural changes. We first looked at the thickness of the primary somatosensory cortex. The 4‐week‐old, acute control rats had a thickness of 1.92 ± 0.02 mm while that of the 8‐week‐old, chronic control rats was 1.94 ± 0.05 mm (n = 5 each, Table 1). Hydrocephalic rats had thinner cortex, 1.48 ± 0.05 mm thickness in the acute and 1.49 ± 0.09 mm in the chronic condition (n = 5 each, P < 0.05 vs. corresponding control, Table 1).

We then investigated whether hydrocephalus changed the volume of the cerebral cortex and hippocampus. The contours of the cerebral cortex (bregma −0.36 to −4.86 mm, excluding the piriform cortex) and hippocampus (bregma −1.8 mm to the end of the hippocampus, including the fimbria) in, one in every 25, 10‐μm serial sections were traced with a PC‐based three‐dimensional reconstruction software. The reconstructed stereo images of the selected cortical and hippocampal areas are illustrated in Figure 3. The control cerebral cortex had a volume of 157.41 ± 2.26 mm³ in the 4‐week old acute control rats (n = 3) and increased to 168.87 ± 4.49 mm³ in the 8‐week‐old chronic control animals (n = 4). Hydrocephalic rats had significantly reduced cortical volume, 131.52 ± 4.61 mm³ in acute (n = 4) and 151.01 ± 2.8 mm³ in chronic (n = 5) stages, as compared to its corresponding age‐matched controls (P < 0.05, Table 1). The volume of the segment of the hippocampus reconstructed was also significantly reduced from 69.42 ± 4.61 mm³ in acute controls (n = 3) to 57.41 ± 2.29 mm³ in acute hydrocephalic rats (n = 4), and from 80.94 ± 2.13 mm³ in chronic controls (n = 4) to 70.42 ± 1.9 mm³ in chronic hydrocephalic rats (n = 5) (P < 0.05, Table 1). The reduction in the volume of both structures under hydrocephalic conditions prompted us to investigate whether their principal output neurons, namely pyramidal neurons were altered.

Figure 3.

Representative stereological views of the cerebral cortex and dorsal hippocampus reconstructed from serial coronal sections. The cerebral cortex, except the piriform cortex, −0.36 to −4.86 mm from bregma and the dorsal hippocampus, including the fimbria, −1.8 mm from bregma to its caudal end, were reconstructed. Control and hydrocephalic rats of the acute (upper two rows) and chronic (lower two rows) stages are demonstrated. Anterolateral view of the cerebral cortex of a representative animal from each group is shown in the left column. A right (middle column) and left (right column) antero‐inferior views of the dorsal hippocampus of a representative animal of each group are shown in the right.

Hydrocephalus altered cortical and hippocampal pyramidal neuronal morphology

Although the quality of the detail morphology of the H&E‐stained sections that we obtained from the in situ whole brain freezing and sectioning specimen was somewhat compromised as compared to that of the transcardially perfused tissue, closer examination revealed that the shapes of the cortical and hippocampal pyramidal neurons were altered (Figure 4). Under either acute and chronic hydrocephalic condition, the normal pyramidal shaped cell bodies in the layer V of the control somatosensory cortex (Figure 4A1,B1) were squeezed into various forms (Figure 4A2,B2). In CA1 hippocampus, the compact 4–8 rows of pyramidal cell bodies typical of control rats (Figure 4C1,D1) became dispersed under both acute and chronic hydrocephalus (Figure 4C2,D2). There were no apparent hyperchromatic or pyknotic cells in either the cortex or the hippocampus. In addition, no sign of apparent decrease in cell densities was observed. These suggest the lack of an ongoing neuronal death in the primary somatosensory cortex or hippocampus during the two time points that we examined.

Figure 4.

Cortical and hippocampal pyramidal neuronal features in the H&E stained freshly frozen coronal sections of the whole brain. Sections from control and hydrocephalic brains of the acute (left column) and chronic (right column) animals are illustrated. A1,B1. Cell bodies of representative layer V pyramidal neurons in the primary somatosensory cortex of the acute and chronic control rats, respectively. A2,B2. Corresponding representative neurons of the acute and chronic hydrocephalic brains. The CA1 pyramidal layer of the acute (C2) and its corresponding age matched control (C1) are shown in the left column. While those of the chronic (D2) and the corresponding control (D1) are shown in the right column. Scale bar in B2 = 20 μm, also for A1, A2 and B1; that in D2 = 20 μm, also for C1, C2 and D1.

To further analyze hydrocephalus‐induced neuronal morphology changes, the dendritic arbors of the primary somatosensory cortical and hippocampal pyramidal neuron were unveiled with intracellular dye injection in brain slices. Figure 5 shows a representative dye‐filled layer V pyramidal neuron of the primary somatosensory cortex of a chronic control rat (A) and a CA1 hippocampal pyramidal neuron of an acute control rat (B) in one of the 60‐μm‐thick section of the injected brain slice. With proper orientation of the brain while preparing the brain slices and the selection of deep‐lying neuron for dye injection, the complete dendritic arbor of individual cortical pyramidal neuron could be revealed in clear isolation (Figure 5A). Notice that there is another dye‐filled layer III pyramidal neuron in the section shown in Figure 5A. Higher magnifications of the proximal and distal parts of the apical and basal dendrites (Figure 5A, insets a1–a4) of the layer V neuron show that dendritic spines were clearly revealed and could be readily analyzed. Likewise, the dendritic arbors and spines of hippocampal pyramidal neurons were also adequately revealed for subsequent analyses (Figure 5B, insets b1–b3).

Figure 5.

Representative intracellular dye‐filled somatosensory cortical and CA1 hippocampal pyramidal neuron. A. Shows a layer V somatosensory cortical pyramidal neurons of an age‐matched chronic control rat. Notice that there is another dye‐filled pyramidal neuron in the layer III of the same section. B. Shows a CA1 dorsal hippocampal pyramidal neuron of an age‐matched acute control rat. The photographs were taken from a 60‐μm‐thick section of the 400‐μm‐thick brain slice in which the neuron was filled with intracellular dye. This demonstrated that with proper orientation of the brain while preparing the brain slices, most of the dendritic arbor of a large pyramidal neuron could be seen in a single section of this thickness. Insets a1–a4 show magnified views of the proximal and distal apical and basal dendritic segments of the layer V cortical pyramidal neuron in A as labeled. Insets b1–b3 as labeled, show magnified views of representative proximal and distal apical, and distal basal dendritic segments of the CA1 pyramidal neuron in B, respectively. Scale bar = 100 μm for A and B, and 10 μm for the insets.

In the hydrocephalic brains, the diameters of the apical dendritic trunks of cortical pyramidal neurons appeared to be altered as well. Unlike the relatively straight and uniform appearance in control cortex (Figure 6A), apical dendritic trunks of the layer V pyramidal neurons in the chronic hydrocephalic cortex were often twisted and variable in widths (Figure 6B,C). To quantitate this, we measured the diameters of the apical dendritic trunks of cortical and hippocampal pyramids at where they exited the cell bodies. Numbers of neurons and animals studied in each group were shown in Table 2. In control animals, the apical dendritic trunks of the layer III and layer V cortical and also CA3 hippocampal pyramidal neurons had relatively uniform diameters while those of the hydrocephalic brains were more variable as revealed by the large standard deviations of the means in the box and whisker plots of the hydrocephalic rats (Figure 6D,E,G, compared between hydrocephalic and age‐matched control). Those of the CA1 hippocampal pyramidal neurons however did not show such a tendency of changes (Figure 6F).

Figure 6.

The effect of hydrocephalus on the apical dendritic trunks of the somatosensory cortical pyramidal neurons. The diameter of the apical dendritic trunk of the studied pyramidal neuron at where it exited the cell body was measured and analyzed. A. Representative image of the proximal apical trunk of a layer V cortical pyramidal neuron of the chronic control rat. B,C. Show examples of a thick and thin proximal apical dendritic trunks of the layer V cortical pyramidal neurons in the chronic hydrocephalic brains, respectively. Box and whisker plots of the apical trunk diameters of layer III (D) and layer V cortical (E) and CA1 (F) and CA3 hippocampal (G) pyramidal neurons from the acute and chronic hydrocephalic and corresponding control brains are show in the bottom two rows. In the plot, box gives median and lower and upper quartiles; whiskers show the 10th and 90th percentiles. Outlying data points are plotted individually. Number of neurons studied in each group is listed in Table 2. The standard deviation of each group is shown right above the abscissa. Abbreviations: con = control; hydro = hydrocephalus. Scale bar = 10 μm in A–C.

Hydrocephalus trimmed the total dendritic arbors and lengths of cortical pyramidal neurons

To analyze the effects on dendrites, the dendritic processes of dye‐filled layer III and layer V pyramidal neurons of the primary somatosensory cortex were traced three‐dimensionally through serial sections to reconstruct their complete arbors. Figure 7A–D shows the dendritic arbors of representative reconstructed layer V cortical pyramidal neurons of the acute and chronic hydrocephalic and corresponding age‐matched control groups. Numbers of neurons and animals studied in each group are shown in Table 2. The studied neurons in the apparently thinned primary somatosensory cortex retained the configuration with clear apical and basal dendrites. Analyses show that the total dendritic lengths of layer III as well as layer V cortical pyramids were both significantly reduced in the acute hydrocephalic brains (Figure 7E1,F1, layer III: 4.12 ± 0.21 mm in controls vs. 3.51 ± 0.19 mm in hydrocephalus; layer V: 11.78 ± 0.86 mm in controls vs. 7.09 ± 0.46 mm in hydrocephalus) and in the chronic hydrocephalic brains (Figure 7E1,F1, layer III: 4.71 ± 0.20 mm in controls vs. 3.65 ± 0.18 mm in hydrocephalus; layer V: 12.93 ± 1.02 mm in controls vs. 6.91 ± 0.49 mm in hydrocephalus) as compared to their age‐matched controls. The total dendritic lengths of both cortical pyramids of the 8‐week‐old controls were no different from that of the 4‐week‐old controls (P = 0.061) suggesting that the dendritic arbors of normal cortical neurons are already matured around a month after birth. Likewise, the reduction of the dendritic lengths of both layer III and layer V pyramidal neurons during the acute stage maintained to the chronic stage so that the dendritic lengths remain almost identical.

Figure 7.

The effect of hydrocephalus on the dendritic arbors of somatosensory cortical pyramidal neurons. The dendritic arbors of a representative layer V cortical pyramidal neurons from acute (B) and chronic (D) hydrocephalic animals and their age‐matched controls (A and C, respectively) were reconstructed three‐dimensionally through the serial sections of the brain slice in which the neuron was injected with intracellular dye. The dendritic branches of each trunk are depicted in the same color. The dendritic lengths of the layer III and layer V pyramids of the control and hydrocephalic groups were derived from the three‐dimensional reconstruction digital data and analyzed in E1 and F1, respectively. The numbers of dendritic ends of these neurons were also quantitated and illustrated in E2 and F2, respectively. Data presented are mean ± SEM. Numbers of neurons and rats analyzed in each group are shown in Table 2. Hydro is the abbreviation for hydrocephalus. Scale bar = 200 μm in A–D; *, P < 0.05 and **, P < 0.001 between the marked and its corresponding age‐matched control.

When dendrites were subdivided into apical and basal categories, the apical dendritic length of both layer III and V cortical pyramids, and the basal dendritic length of layer V cortical pyramids of the hydrocephalic rats were shorter than their age‐matched counterparts. However, the basal dendritic length of the layer III cortical pyramids became significantly shorter by the chronic stage.

To find out whether shorter dendritic length was accompanied by trimming of dendritic branches, we compared the number of dendritic ends between pyramidal neurons of the control and hydrocephalic cortices. The numbers of the apical dendritic ends of both layer III and layer V cortical pyramids and that of basal dendritic ends of the layer V cortical pyramids of the acute hydrocephalic cortices were fewer than their corresponding age‐matched controls, and remained reduced in the chronic stage (Figure 7E2,F2). These suggest that the reduction of dendritic length following hydrocephalus could be resulted from a reduction in branching. However, the number of basal dendritic ends in layer III cortical pyramids did not change in control rats over time or under hydrocephalic conditions (Figure 7E2,F2).

Hydrocephalus reduced the total dendritic length of CA1 hippocampal pyramidal neurons

The dendritic arbors of CA1 and CA3 pyramidal neurons of the dorsal hippocampus were also three‐dimensionally reconstructed and analyzed. Numbers of neurons and animals analyzed are shown in Table 2. The dendritic arbors of representative CA1 pyramidal neurons of the hydrocephalic and age‐matched control groups were demonstrated in Figure 8A–D. CA1 pyramidal neurons of the hydrocephalic brain had significantly shorter total dendritic length than their age‐matched controls (5.86 ± 0.18 mm in acute hydrocephalic rats vs. 6.5 ± 0.22 mm in age‐matched controls; 6.49 ± 0.18 mm in chronic hydrocephalic rats vs. 7.30 ± 0.30 mm in age‐matched controls) (Figure 8E1). The apical dendritic lengths of the hydrocephalic CA1 pyramidal neurons were shorter than their age‐matched controls 1 week after kaolin injection and remained shorter than corresponding controls by the chronic stage. Total basal dendritic lengths however did not change during the acute hydrocephalus and were shorter than their age‐matched controls by the chronic stage. The number of apical dendritic ends in CA1 pyramidal neurons of the age‐matched controls and also those of the hydrocephalic brains remained stable during the time frame studied. However, slightly fewer basal dendritic ends was found in the chronic hydrocephalic CA1 pyramidal neurons studied (Figure 8E2). On the contrary, CA3 pyramidal neurons showed no consistent difference in dendritic lengths or the numbers of dendritic ends from their corresponding age‐matched controls at either the acute or chronic hydrocephalic stage (Figure 8F1,F2).

Figure 8.

The effects of hydrocephalus on the dendritic arbors of dorsal hippocampal pyramidal neurons. Representative three‐dimensionally reconstructed CA1 hippocampal pyramids of the dorsal hippocampus of the acute and chronic hydrocephalic and corresponding age‐matched controls are shown in A–D as labeled. Dendritic branches of each trunk are shown in one color. Analyses of the dendritic lengths of these CA1 and CA3 hippocampal pyramidal neurons of the control and hydrocephalic groups are shown in E1 and F1, respectively. Analyses of the numbers of dendritic ends of CA1 and CA3 pyramids of the studied groups are shown in E2 and F2, respectively. Numbers of reconstructed neurons and rats studied are shown in Table 2. Hydro stands for hydrocephalus. Scale bar = 200 μm in A–D; *, P < 0.05 between the marked and its corresponding age‐matched control.

Hydrocephalus downregulated dendritic spines on cortical and hippocampal pyramidal neurons

We then asked whether hydrocephalus altered the densities of dendritic spines, protrusions known to form synapses with excitatory presynaptic terminals on primary somatosensory cortical and hippocampal pyramidal neurons. Because spine densities on these relatively large neurons varied depending on their locations on dendrites therefore, proximal and distal segments of the apical and basal dendrites were distinguished. In the hydrocephalic cortex, fewer dendritic spines were found on all 4 dendritic segments of both layer III and layer V pyramidal neurons than their age‐matched controls during the acute and chronic hydrocephalic stages (Figure 9A,B). Dendritic spines on layer III and layer V pyramidal neurons of the acute hydrocephalic cortex were approximately 62%–70% and 56%–65% of that of the age‐matched controls, while those of the chronic hydrocephalic rats were further less, reduced to 33%–61% and 20%–47% of that of the age‐matched controls, respectively (Figure 9C,D).

Figure 9.

The effects of hydrocephalus on the densities of dendritic spines on somatosensory cortical pyramidal neuron. Representative images of the distal apical and proximal basal dendritic segments of the layer III (A) and layer V (B) pyramidal neurons of the hydrocephalic and age‐matched control rats in the acute and chronic stages are illustrated as labeled. The densities of dendritic spines on the proximal and distal apical and basal dendrites of these neurons are analyzed in C and D as labeled. Data are mean ± SEM. Scale = 10 μm in A and B. E,F. Representative immunoblots and analyses of the expressions of PSD95 and synaptophysin in the somatosensory cortex of the studied animal groups as labeled (n = 5 each). The amounts of PSD95 and synaptophysin were normalized to that of the GAPDH (loading control). Data are mean ± SEM of the ratio of the hydrocephalic to the corresponding control. *, P < 0.05 and **, P < 0.001 between the marked and its corresponding age‐matched controls; #, P < 0.05 between the acute and chronic hydrocephalic rats. Abbreviations: con = control; hydro = hydrocephalus.

In the hippocampus, dendritic spines on all dendritic segments of both CA1 and CA3 pyramidal neurons analyzed were significantly reduced in acute hydrocephalus too, 67%–70% and 70%–76% of that of the corresponding age‐matched controls, respectively (Figure 10A,B, *, P < 0.05). Spine densities on these dendritic segments remained at approximately the same level during the chronic hydrocephalic stage as well, 68%–74% and 63%–69% of that of the corresponding controls, respectively (Figure 10C,D), except that the distal apical dendrites of the CA3 pyramids showed a small but significant further reduction (Figure 10D, #, P < 0.05).

Figure 10.

The effects of hydrocephalus on the densities of dendritic spines on CA1 and CA3 dorsal hippocampal pyramidal neurons. Representative micrographs of the distal apical and basal dendritic segments of the CA1 (A) and CA3 (B) pyramidal neurons of control and hydrocephalic rats in the acute and chronic stages are illustrated as labeled. The densities of the dendritic spines on the distal and proximal apical and distal basal dendrites of these neurons are analyzed in C and D as labeled. Representative immunoblots and analyses of the expressions of PSD95 (E) and synaptophysin (F) in the dorsal hippocampus are shown as labeled. Data were normalized to that of the GAPDH and shown as mean ± SEM of the ratio of the hydrocephalic to that of the control (n = 5 for each group). Scale = 10 μm in A and B. *, P < 0.05 between the marked and its corresponding control; #, P < 0.05 between the acute and chronic hydrocephalic rats.

Quantitative analyses showed that the levels of PSD95 in the primary somatosensory cortex of the acute and chronic hydrocephalic brains were significantly less, decreased to 58% ± 10% and 77% ± 6% of that of the corresponding age‐matched controls, respectively (n = 5 each, Figure 9E). In the dorsal hippocampus, PSD95 expressions in the acute and chronic hydrocephalic brains were reduced to 62% ± 14% and 67% ± 15% of the corresponding age‐matched controls, respectively (n = 5 each, Figure 10E). Thus, reduction of dendritic spines on primary somatosensory cortical and CA1 and CA3 hippocampal pyramidal neurons under hydrocephalus was accompanied by downregulation of glutamatergic postsynaptic marker.

To explore whether hydrocephalus also altered presynaptic inputs, the expressions of synaptophysin were investigated. In the primary somatosensory cortex, the amount of synaptophysin was significantly decreased to 52% ± 11% and 33% ± 4% of that of the corresponding age‐matched controls during the acute and chronic stages, respectively (n = 5 each, Figure 9F). In the dorsal hippocampus, the levels of synaptophysin were reduced to 68% ± 4% and 62% ± 8% of the corresponding age‐matched controls during the acute and chronic hydrocephalic stages, respectively (n = 5 each, Figure 10F). Thus, hydrocephalus resulted in simultaneous reductions of pre and postsynaptic components on the main output neurons of the primary somatosensory cortex and hippocampus.

Discussion

Using an experimental infantile rat hydrocephalus model, we found hydrocephalus resulted in spatial learning and memory deficits. In addition, hydrocephalus reduced cortical and hippocampal volumes and altered their cytoarchitectures. At the cellular level, cortical and hippocampal pyramidal neurons trimmed their dendritic arbors, lengths, and spines, and at the same time reduced the expressions of the glutamatergic postsynaptic and presynaptic vesicle markers. Reduction of dendritic spines exacerbated with chronicity in primary somatosensory cortical pyramidal neurons, however those in CA1 and CA3 hippocampal pyramidal neurons remained at the same level throughout the acute and chronic hydrocephalus.

Hydrocephalus induced the atrophy of cerebral cortex and hippocampus

Several factors appeared to be associated with or contributed to the reduction of the volume or atrophy of the cerebral cortex and hippocampus in hydrocephalus. The brain behaves as a poroviscoelastic material and bulk deformation or consolidation of brain tissue occurs mainly by fluid movement through the extracellular porous medium 10. Increase of intracranial pulsatility during hydrocephalus could displace intraparenchymal fluid 60 and consequently reduced brain tissue volume. This proposition is consistent with earlier findings that the extracellular flow and volume fraction in the cerebral cortex of kaolin‐induced hydrocephalic rats are reduced 17, 59. Alternatively, neuronal cell death and/or damage of connections between neurons could also lead to the atrophy of brain tissue 45, 58. However, neuronal death might not be essential to hydrocephalus as no apparent neuronal death was reported in congenital hydrocephalus 33, previous kaolin‐induced obstructive hydrocephalus studies 20, 23 and in the present study. In fact, the cell density of the hydrocephalic cerebral cortex might have increased as has been demonstrated in the adult rabbit silicone oil‐induced hydrocephalus model 12. Despite this, hydrocephalus appeared to reduce cell connectivity. In the present study, the dendritic fields and spines of the layer III and layer V cortical (Figures 7 and 9) and CA1 and CA3 hippocampal pyramidal neurons were all significantly reduced following hydrocephalus (Figures 8 and 10). In addition, the hydrocephalic cortex and dorsal hippocampus showed reduced expressions of the presynaptic terminal marker synaptophysin and the glutamatergic postsynaptic maker PSD95 expressions. These support the notion that hydrocephalus is associated with substantial loss of cerebral and hippocampal cell connectivities. Taken together, hydrocephalus could have reduced extracellular space volume and in addition neuronal connectivity to contribute to the reduction of cortical and hippocampal volumes.

The effects of hydrocephalus on primary somatosensory cortex

In addition to reducing cortical volume and thinning cerebral cortex, hydrocephalus altered the principal output neurons, namely layer III and layer V pyramidal neurons, of the primary somatosensory cortex. The diameters of the proximal apical dendritic trunks of these cortical neurons became more variable. This could affect the weighting of synaptic inputs 31 and the input impedance of the neuron directly 61. Thus, hydrocephalus may induce heterogeneity of dendritic conductance and changed the properties of layer III and layer V pyramidal neurons in the somatosensory cortex.

Closer examination of the three‐dimensionally reconstructed dendritic arbors of these cortical pyramidal neurons showed that hydrocephalus reduced their total and apical and basal dendritic lengths (Figure 7). Changes were more dramatic in layer V (approximately 40% reduction of total dendritic length as compared to controls) than layer III pyramidal neurons (around 23% reduction). In layer V pyramidal neurons, the reductions of apical and basal dendritic lengths were paralleled by reductions of dendritic ends. Trimming of dendrites started in the acute hydrocephalic stage, and maintained roughly the same number of dendritic ends during chronic hydrocephalus. Similar trend of changes was found in the apical dendrites of the layer III cortical pyramidal neurons but the reduction in the basal dendritic length appeared to be delayed and was statistically significant until chronic hydrocephalus, which was not accompanied by a statistically identifiable trimming of dendritic terminals. Thus, under hydrocephalic condition both layer III and layer V somatosensory cortical pyramidal neurons appeared to trim their dendritic ends to reduce dendritic lengths.

The early onset and the large scale trimming of, especially the apical, dendritic ends in both layer III and layer V cortical pyramidal neurons are consistent with the notion that in hydrocephalus, mechanical disturbance is an important factor leading to the shrinkage of dendritic arbors for mechanical force has been shown to rapidly induce dramatic dendritic remodeling in cortical pyramidal neurons under epidural compression 9. Thinning of cerebral cortex and corpus callosum 52 in hydrocephalus suggests that the expanding ventricles also compress cortical pyramidal neurons. The trunk and terminal tuft of the apical dendrites run more perpendicular to the brain surface than basal dendrites, so that they were vertically squeezed by the expanded ventricles against the skull. In epidural cortical compression, the mechanical force downregulated protein phosphatase 2A instantly. This phosphorylated and hence detached microtubule‐associated proteins from microtubules. The exposed microtubules went on reconfiguration and resulted in the shrinkage of dendrites 9. The roles of the mechanical force generated by the expanding ventricles and factors such as the accumulation of metabolites, ions and inflammatory substances as well as interstitial fluid in the neuronal dendritic remodeling remain to be explored.

When comparing the extents of the reductions of dendritic length or ends, the deep‐lying layer V pyramidal neurons were more seriously affected by hydrocephalus than the superficially located layer III pyramidal neurons. This is consistent with previous reports of rapid Golgi stained parietal cortical pyramidal neurons in kaolin‐induced hydrocephalic rats 49. Nevertheless, the authors 49 reported that only the basal dendrites were stunted and the number and length of the apical dendrites were unaltered. Although the authors selected only the best impregnated cells for analysis, the capriciousness of the labeling and the unlikelihood to reconstruct the complete dendritic arbors of these relatively large neurons in Golgi‐impregnated sections could have marred the results. In the present study, we filled individual neuron with intracellular dye in brain slices, which were much thicker than Golgi‐impregnated sections, under the visual guidance of a fluorescence microscope. Selected neurons with apparently complete and no overlapping dendritic arbors could be consistently revealed 6. In addition, we used PC‐based software to reconstruct the complete dendritic arbor of each neuron three‐dimensionally through the serial sections of the brain slice in which the neuron was injected. These together yielded more accurate dendritic measurements for subsequent analyses 6, 7. The dendritic arbor and length data of the normal control rats that we collected in the present study are equivalent to those of our earlier studies of the same pyramidal neurons in the rat primary somatosensory cortex 5, 6, 7. With this, we are confident that we have properly revealed the pathological changes of hydrocephalic cortical pyramidal neurons. Our findings on the effect of hydrocephalus on primary somatosensory cortical pyramidal neurons are in agreement with earlier report in congenital hydrocephalic rats that the apical and basal dendritic lengths of the auditory cortical layer V pyramidal neurons were reduced 28.

Unlike the trimming of dendritic length, reduction of dendritic spines intensified with chronicity: 30%–38% reduction in acute to 39%–67% reduction in chronic in layer III pyramids, and 35%–44% reduction in acute to 53%–80% reduction in chronic in layer V pyramids (Figure 9) in primary somatosensory cortex. The reduction of dendritic spine densities on top of the reduction of dendritic length means that each neuron lost more dendritic spines than the spine density figures showed. In this study, loss of total spines per neuron will be more dramatic in layer V than layer III pyramidal neurons for the former show greater reduction of dendritic length (Figure 7). These suggest that the functions of layer V pyramidal neurons are more seriously compromised than those of the layer III pyramids under hydrocephalus.

Cerebral ischemia 29, brain inflammation 21, loss of presynaptic elements caused by deafferentation 26 and cerebral compression 8 are all known to affect presynaptic and/or postsynaptic activities and lead to pruning of neuronal dendritic spines. The reduction of afferents in the hydrocephalus reported earlier 14 and indirectly in the present study (the reduction of synaptophysin level) could contribute to the pruning of dendritic spines that we observed. The mass compression effect of ventriculomegaly, brain ischemia 2 and inflammation 21, 22 are likely additional factors to intensify the dendritic spine loss under hydrocephalus. In the present study, there was no apparent difference in the extent of the reductions of synaptophysin and PSD95 between the acute and chronic hydrocephalic cortices. Loss of dendritic spines however was more severe in the chronic than acute hydrocephalic stage. Decreases of pre‐ and postsynaptic markers in acute hydrocephalus could have reflected or started the trimming of dendritic spines. Other factors could add to contribute to the late dendritic spine loss during chronic hydrocephalus as well. In the somatosensory cortex, mechanical compression appeared to mobilize and activate NMDA receptors to increase postsynaptic calcium, which in turn activated the phosphatase calcineurin to dephosphorylate and activate the actin‐severing protein cofilin to disintegrate actin in dendritic spines for retraction 8. This mechanism could have played a role in the retraction of dendritic spines in hydrocephalus as the associated ventriculomegaly also compressed the cerebral cortex. The causal relationship between loss of afferents and pruning of dendritic spines remain to be explored.

The effects of hydrocephalus on hippocampus

To our knowledge, fewer studies had addressed the pathophysiology of hippocampus in either human or experimental hydrocephalus. In this study, we concentrated on dorsal, rather than ventral, hippocampus for the former is more related to spatial learning and memory 51. Our results show kaolin‐induced hydrocephalus caused hippocampal atrophy within a week (Table 1). Cytoarchitecturally, both CA1 and CA3 pyramidal cell body layers became more dispersed (Figure 4), suggesting the distortion of the somatodendritic arbors of CA1 and CA3 pyramidal neurons.

Although we did not analyze these neurons' somatodendritic topology, our analyses show that hydrocephalus significantly reduced the total and apical dendritic lengths of CA1 pyramidal neurons starting in acute hydrocephalus. The effect on basal dendritic length was delayed and evident by the chronic stage (Figure 8). Changes in the dendritic lengths of these neurons were not accompanied by a reduction of dendritic ends. Unlike CA1 cells, CA3 hippocampal pyramidal neurons failed to show any apparent difference in their dendritic lengths after hydrocephalus (Figure 8). This could be a consequence of the heterogeneity of CA3 hippocampal pyramidal neurons 24 in having different dendritic lengths 24, 32. With this, a much larger population of CA3 pyramidal neurons that allows the division of subtypes 24 might be required to effectively analyze the effect of hydrocephalus on CA3 neurons. On the other hand, CA1 pyramidal neurons could be more responsive to hydrocephalus than the CA3s. In this regard, earlier study has shown that experimental hydrocephalus caused significantly higher pyknotic index in CA1 than CA3 pyramids 32.

Despite the difference in dendritic length effects, hydrocephalus trimmed the dendritic spines on the apical and basal dendrites of both CA1 and CA3 pyramidal neurons to roughly similar extent (Figure 10). Spine reduction started in acute hydrocephalus and maintained at the same level to the chronic stage: CA1 neurons, 30%–33% reduction in acute and 28%–32% in chronic stage; CA3 neurons, 24%–30% in acute and 31%–37% in chronic. The reduction of dendritic spines was paralleled by decreases of PSD95 and synaptophysin in both regions. These findings are in line with earlier report that the postsynaptic integration in CA1 hippocampus of the kaolin‐induced hydrocephalic rats was disturbed 63, and EM study that the number of synaptic contacts in CA3 hippocampus was reduced 41.

The reduction of dendritic spines in hippocampal pyramidal neurons was however different in time course from that in the primary somatosensory cortical pyramidal neurons. The decrease grew with time in the primary somatosensory cortex but remained steady in the hippocampus from the acute to the chronic stage of the hydrocephalus. The functional significance of this difference remains to be explored.

Functional implication of hydrocephalus‐induced cortical and hippocampal dendritic changes

Like earlier studies, kaolin‐induced hydrocephalic rats showed slower body weight gain 15, 44, 70. Acute hydrocephalic rats showed increased escape latency starting from the second trial block of the single‐day MWM task while chronic animals (a different batch of animals, first exposure to the task) took longer to find the underwater platform than their age‐matched controls starting in the first trail block of the day. These cannot be attributed to difference in swimming ability between animals (Figure 1B) 35, 70. Neither can it be a consequence of the difference in gross visual ability 70 as our acute hydrocephalic rats performed as good as the age‐matched controls during the first block of the trials (Figure 1C). Our findings that hydrocephalus impaired rats' spatial learning and memory are consistent with earlier study testing the animals' performance with multiple day MWM task 15, 70 as well.

In fetal‐onset hydrocephalus, the periventricular heterotopia and abnormal neurogenesis during embryonic development are likely to underlie cognitive deficits 27, 54. After the completion of neurogenesis, spatial learning, memory and cognitive formation depend upon the coordinated actions of diverse brain regions constituting a functionally integrated neural network. Among various brain structures, the role of hippocampus in spatial learning and memory has been repetitively explored 11, 53. Dorsal hippocampus is likely more functionally related to spatial learning as its lesion compromised MWM task more than lesion of the ventral hippocampus 51. Lesion of the somatosensory cortex yielded variable effects on rats' spatial memory acquisition in the typical multiple day MWM task 30, 38 but impaired animals' performance in single‐day MWM task 40. In fact, somatosensory cortex has been proposed to participate more specifically in the transformation process of egocentrically acquired information into allocentrically coded information in spatial cognition 57. Thus, somatosensory cortex is functionally related to hippocampus in spatial learning. A number of modeling studies have shown that the dendritic architecture determines the electrotonic properties of neurons, which are fundamental determinants of synaptic integration and neural firing pattern 46, 47, 67. In this study, hydrocephalus altered the morphology and reduced the dendritic fields and spine densities of layer III and layer V cortical pyramids of the primary somatosensory cortex and CA1 pyramids of the dorsal hippocampus. This would alter neurons' firing and affect the information processing in both areas 65 In addition to morphological evidences, our biochemical findings of a concomitant decrease of glutamatergic postsynaptic density marker PSD95 and presynaptic vesicle marker synaptophysin expressions also support the proposition that hydrocephalus reduced excitatory drives to cortical and hippocampal pyramidal neurons, consequently cortical outputs. Connection‐wise, somatosensory cortex connects via layer III cortical pyramids to the prefrontal cortex, brain region involved in goal‐directed behaviors, limbic system 1, and contralateral cortex through axons in the corpus callosum; via layer V cortical pyramids to subcortical structures mainly 36. Damage of axons of layer III cortical pyramids in the corpus callosum has also been confirmed in hydrocephalus 14. On the other hand, CA3 pyramids provide mainly intrahippocampal projections 71 while CA1 projects diversely outside the hippocampus 66. Thus, the effects of hydrocephalus‐induced cortical and hippocampal neuronal changes are likely to spread through these connections to other brain regions to consequently affect cognition. Because hydrocephalus's effects on cortical pyramids progressed continuously from the acute to the chronic stage, cortical alterations could have more weight than hippocampal changes on the defective searching strategy of chronic hydrocephalic rats in the MWM task evaluated in the present experiments. The role of these central changes in other hydrocephalus‐associated impairments such as attention and executive function deficits 43 remains to be explored.

Technical limitations of our experimental protocols

In the present study, hydrocephalus damaged the corpus callosum and caused peeling off of tissue around ventricles, especially at rostral end of the ventricles, to preclude accurate tracing of the ventricular contour, hence hampered accurate 3D reconstruction and subsequent analyses of ventricular sizes and corpus callosum thicknesses. This prevented us from checking whether changes in ventricle sizes and thicknesses of the cerebral cortex and corpus callosum were correlated with the animals' performance in Morris water maze task. Previous studies in the same animal model found a positive correlation only in the most severe cases of ventriculomegaly 52, 70. This also prevented us from finding out whether the degree of ventricular enlargement was correlated with the magnitude of dendritic changes of studied pyramidal neurons. However, our data show that the standard errors of the means of the total dendritic length, number of dendritic ends (Figures 7 and 8) and dendritic spine densities (Figures 9 and 10) of the hydrocephalic pyramidal neurons were relatively small. And the standard errors of the PSD95 and synaptophysin levels in the somatosensory cortex and dorsal hippocampus were also small (Figures 9 and 10). These argue against a positive correlation between the extent of ventriculomegaly with the severities of the morphological and biochemical changes. A positive standard correlation is expected to generate large standard errors of the means of the parameters studied. Nevertheless, further studies are needed to confirm this.

Another shortcoming of the present study is that we were unable to delineate the role of the inflammatory responses in the kaolin‐injection model. Immune responses are believed to be part of the kaolin‐induced hydrocephalus in animals 14 and childhood hydrocephalus 19. Inflammation is known to affect dendritic spine dynamics 39, however a definitive effect on dendritic length remain controversial 4, 21. The role of immune responses on the pathophysiology of neuronal dendrites in association with learning and memory deficits in hydrocephalus remains to be explored.

Conclusions

Using an experimental infantile hydrocephalic rat model, we found hydrocephalus reduced cerebral cortical and hippocampal volumes. At the cellular level, it shrank the dendritic arbors and reduced the dendritic lengths and spines of layer III and layer V pyramidal neurons of the primary somatosensory cortex and CA1 and CA3 dorsal hippocampal pyramidal neurons. These morphological changes were accompanied by reductions of glutamatergic postsynaptic density marker PSD95 and presynaptic terminal marker synaptophysin and could underlie the accompanied spatial learning and memory deficits.