Seizures Decrease Postnatal Neurogenesis and Granule Cell Development in the Human Fascia tDentata


  • Gary W. Mathern,

    1. Division of Neurosurgery, The Mental Retardation Research Center, and The Brain Research Institute, University of California, Los Angeles, Los Angeles, California, and
    Search for more papers by this author
  • James L. Leiphart,

    1. Division of Neurosurgery, The Mental Retardation Research Center, and The Brain Research Institute, University of California, Los Angeles, Los Angeles, California, and
    Search for more papers by this author
  • Adelaine De Vera,

    1. Division of Neurosurgery, The Mental Retardation Research Center, and The Brain Research Institute, University of California, Los Angeles, Los Angeles, California, and
    Search for more papers by this author
  • P. David Adelson,

    1. Department of Neurosurgery, University of Pittsburgh, Pittsburgh, Pennsylvania, U.S.A.;
    Search for more papers by this author
  • Tatsunori Seki,

    1. Department of Anatomy, Juntendo University School of Medicine, Tokyo, Japan; and
    Search for more papers by this author
  • Luciano Neder,

    1. Departments of Neurology and Pathology, Ribeirão Preto School of Medicine, University of São Paulo, Ribeirão Preto, Brazil
    Search for more papers by this author
  • Joao P. Leite

    1. Departments of Neurology and Pathology, Ribeirão Preto School of Medicine, University of São Paulo, Ribeirão Preto, Brazil
    Search for more papers by this author

Address correspondence and reprint requests to Dr. G.W. Mathern at Division of Neurosurgery, Reed Neurological Research Center, 710 Westwood Plaza, Room 2123, Los Angeles, CA 90095-1769, U.S.A. E-mail:


Summary:  Purpose: There is considerable controversy whether childhood seizures damage existing neurons and/or adversely affect neurogenesis and synaptogenesis. This study addressed this question by examining fascia dentata neurogenesis, cell death, and aberrant axon connections in hippocampi from children with extratemporal seizure foci.

Methods: Surgically resected (n = 53) and age-comparable autopsy (n = 22) hippocampi were studied for neuronal densities, polysialic acid (PSA) neural cell adhesion molecule (NCAM) immunoreactivity (IR), TUNEL, and neo-Timm's histochemistry.

Results: Compared with autopsy cases, hippocampi from children with frequent seizures showed (a) decreased fascia dentata granule cell densities; (b) decreased PSA NCAM IR cell densities in the stratum granulosum, infragranular, and hilar regions; (c) no positive TUNEL-stained cells; and (d) aberrant supragranular mossy fiber axon connections.

Conclusions: These results indicate that severe seizures during early childhood are associated with anatomic signs of decreased postnatal granule cell neurogenesis (PSA NCAM IR) and aberrant mossy fiber axon connections (neo-Timm's) without evidence of seizure-induced cell death (TUNEL). In humans, these results support the concept that seizures do not damage existing neurons, but adversely affect processes involved with normal postnatal neuronal development such as neurogenesis and axon formation. Such alterations probably negatively affect normal brain development, and/or promote epileptogenesis.

The effects of seizures in the developing brain differ compared with those in the mature brain. In adult animals, frequent seizures and/or status epilepticus result in neuronal loss, aberrant axon sprouting, and long-term deficits in learning, memory, and behavior (1,2). Similar experiments in immature rats demonstrate minimal or no pathology unless the brain insults are severe, and even then, the amount of damage is generally less than with the same insult in adult rats (3–5). Some authors have concluded, therefore, that the immature brain is resistant to seizure-induced neuronal damage. However, these experiments do not address whether seizures adversely affect other developmental processes, such as postnatal neurogenesis, axogenesis, and synaptogenesis, which are equally important components of early brain development.

Previous animal studies demonstrated that neonatal seizures may adversely affect some aspects of brain development, and in adult rats, reported increased postseizure neurogenesis of fascia dentata granule cells (6–10). Another study, however, showed that granule cell neurogenesis was reduced in postnatal (PN) day 1–4 rats with repeated flurothyl-induced seizures (11). Hence, it is unclear if seizures during early hippocampal development might alter postnatal granule cell neurogenesis, and whether this is associated with aberrant mossy fiber sprouting. Previous human studies from our laboratory found decreased granule cell densities in young patients with extratemporal seizures (12,13). Whether decreased granule cell densities began early in life or were the consequence of repeated seizures over time was unclear because of the small sample size. Likewise, it was unknown if decreased granule cell numbers were from reduced neurogenesis or seizure-induced cell death of already formed granule cells. The current human study addresses these questions by performing immunocytochemistry (ICC) for polysialic acid (PSA) neural cell adhesion molecule (NCAM), which identifies newly formed and migrating granule cells, and TUNEL to detect cell death (14,15). We hypothesized that human hippocampi would show decreased neurogenesis and aberrant axon connections in association with repeated early-onset seizures.


Clinical material

Patients with seizures from extratemporal cortical lesions in which the hippocampus was removed as part of the planned resection were used (n = 53). Most patients were treated at the University of California, Los Angeles (UCLA) Pediatric Epilepsy Surgery Program (n = 46). Other cases were from the Ribeirão Preto School of Medicine and the University of Pittsburgh. The clinical-evaluation protocols were previously published (13,16). Surgical cases were subclassified based on the pathologic substrate into patients with cortical dysplasia (CD; n = 32) or without dysplasia (non-CD; n = 21). Hippocampi from autopsy cases (n = 22) of similar ages were used for comparison purposes. Clinical data were collected from the medical record, and informed consent was obtained for medical treatment and use of clinicopathologic data for research purposes.

Histologic procedures and data analysis

At surgery or autopsy, 0.3- to 1.0-cm specimens cut transverse to the hippocampal axis were placed in freshly prepared phosphate-buffered 4% paraformaldehyde (ICC and TUNEL), and neo-Timm's fixative. Specimens were processed for cell counts (cresylecht violet; 10-μm sections), neo-Timm's histochemistry, PSA-NCAM ICC, and TUNEL staining (30 μm), as previously published (12,13,17). Densities of hippocampal neurons and/or PSA NCAM IR–positive cells were determined using cell-counting techniques. Data were entered into a database on a personal computer and analyzed using a statistical program (Statview 5; SAS Institute Inc., Cary, NC, U.S.A.).


Clinical summary of patients and autopsy cases

Autopsy (n = 22) and surgical patients (n = 53) showed similar demographic features. The age at collection for autopsy cases (mean ± SEM; 9.9 ± 2.2 years) and surgical patients (7.2 ± 1.3 years) were similar (t test; p = 0.28). The average postmortem delay for autopsy cases was 7.7 ± 0.5 h. For the surgical group, the average age at seizure onset was 1.52 ± 0.3 years, and the seizure duration before surgery was 5.67 ± 1.18 years. There were no differences in the pathological findings for CD and non-CD cases, and the two groups were combined into a single group for this study.

Hippocampal cell counts in extratemporal epilepsy

Surgical cases showed decreased granule cell densities, and no differences in Ammon's horn neuron densities. Figure 1 shows cell densities for surgical cases with chronic seizures compared with autopsy cases. Granule cell counts were decreased compared with Ammon's horn neuronal densities (Fig. 1, top row; asterisk). In other words, hippocampal sclerosis was not detected in this patient population. Instead, decreased granule cell densities were the most consistent pathologic finding. To determine if decreased granule cell densities were related to age at surgery, the analysis was repeated for cases operated on before or after age 4 years, and this showed that decreased granule cell counts were mostly from younger cases (Fig. 1, bottom row; asterisk). In the surgical group, longer seizure durations were not associated with decreased granule cell neuronal densities (r = 0.03; p = 0.86). Thus, surgical cases showed decreased granule cell densities, beginning in younger patients.

Figure 1.

Histograms showing fascia dentata (FD) granule cell and Ammon's horn (CA4 to prosubiculum) neuronal densities from patients with extratemporal seizures (Peds Epilepsy) and autopsy cases. The top row shows cell densities comparing autopsy cases and all surgical cases, and the second row compares autopsy cases with patients with seizure onset and surgery was before or after age 4 years. Compared with autopsy cases (top row), FD cell densities were decreased [analysis of variance, ANOVA; F = 8.69; p = 0.0043; upper left asterisk) and Ammon's horn densities were not (p = 0.63). Analysis for early versus later surgery (bottom row) showed that compared with autopsy cases, surgeries before age 4 years had decreased FD cell densities (p = 0.0022; lower left histogram; asterisk), whereas surgeries after age 4 years trended toward decreased FD densities (p = 0.060). There were no differences in averaged Ammon's horn neuronal densities between autopsy cases and surgical cases operated on before or after age 4 years (ANOVA; p = 0.075; lower right histogram).

PSA NCAM IR labels immature granule cells, and seizures decrease IR

Human neurogenesis is difficult to study because techniques that mark new DNA synthesis are not ethical to use in humans. However, one can indirectly detect new cell differentiation and migration using protein markers such as PSA NCAM, which identifies newly formed neurons. In autopsy hippocampi, PSA NCAM IR changed during early human postnatal development, consistent with marking immature granule cells (Fig. 2, top row). In the stratum granulosum and infragranular region, PSA NCAM IR labeled small immature-appearing cells and fibers. The number of infragranular PSA NCAM IR cells exponentially decreased over the first 24 months, and by age 3–4 years was like the adult pattern. In the adult autopsy hippocampus, a subpopulation of hilar neurons maintained PSA NCAM IR, but there were minimal IR cells in the granule cell layers.

Figure 2.

Examples of polysialic acid neural cell adhesion molecule immunoreactivity (PSA-NCAM IR) in autopsy cases (top row) and surgical patients (bottom row) over different ages (age noted at bottom of each panel). Young autopsy cases (younger than 1 year; A, B ) showed PSA-NCAM IR–positive cells and fibers in the infragranular zone and stratum granulosum that formed a nearly continuous band. These cells were small and immature appearing, consistent with newly formed and migrating granule cells. In the hilus, notice a few slightly larger IR-positive cells in the younger autopsy specimens (A, B). In autopsy cases, the number of infragranular cells and fibers decreased with age and reached an adult pattern by age 3–4 years. The adult pattern consisted of IR positive hilar cells with IR fibers coursing through the stratum granulosum and molecular layer, and very few IR-positive cells in the stratum granulosum (C). Surgical cases from children with intractable epilepsy show decreased PSA-NCAM IR at all ages compared with autopsy cases. At younger ages (D, E), PSA-NCAM IR was restricted to small groups of infragranular cells and not the continuous band as noted in the nonseizure autopsy cases (A, B). Even older aged surgical cases showed fewer hilar IR cells and fibers compared with autopsy cases (compare F and C ).

Compared with autopsy cases, the surgical cases with seizures showed decreased PSA NCAM IR in the infragranular and granule cell layers. The number of PSA NCMA IR cells was decreased, and instead of a nearly continuous band of IR cells throughout the fascia dentata, there were small nests of IR cells and fibers (Fig. 2D and E). Even older children with chronic seizures showed fewer hilar PSA NCAM IR cells (Fig. 2F). Quantification of PSA NCAM IR showed that, compared with autopsy cases, surgical patients with seizures showed decreased IR cell densities in the infragranular region (p = 0.037) and hilus (p = 0.041). Thus, the PSA NCAM findings show that seizures during early human development were associated with decreased granule cell formation and migration.

No TUNEL-positive cells in surgical cases with epilepsy

Another mechanism that could explain decreased granule cell densities could be cell death. To determine whether this was an important mechanism in our surgical cases, we performed TUNEL staining. None of the surgical cases showed TUNEL-positive cells (Fig. 3B). Because TUNEL-positive cells may be present for only a few days after lethal cell injury (Fig. 3A), our results should be interpreted cautiously. However, on average each histologic slide contained >2,000 granule cells, and none showed TUNEL-positive features. Therefore, our data indicate that if seizures induce granule cell death, it is at a very slow and nearly undetectable rate.

Figure 3.

Examples of TUNEL (top) and neo-Timm's stains (bottom) in a rat brain ( A; positive control) and human hippocampi from children with seizures (B–D). TUNEL in a rat brain 48 h after status epilepticus (A) showed numerous positive cells in the entorhinal and adjoining hippocampal cortex. A human hippocampus from a 12-month-old with nearly continuous seizures since age 2 months from multilobar cortical dysplasia (CD) failed to show any TUNEL-positive cells, even though granule cell densities were 25% less than those in an age-matched autopsy case (B). Neo-Timm's in the surgical specimens showed variable staining in the supragranular inner molecular layer (IML) as a sign of aberrant axon connections. For example, C was from a 5-year-old with seizures since birth from CD and showed a normal neo-Timm's pattern without IML staining. D: A 15-month-old with seizures since birth from CD, and it shows aberrant IML staining (arrows). Both specimens showed similar neuronal densities in all hippocampal subfields.

Aberrant mossy fiber sprouting

Of the surgical cases with extratemporal seizures, 80% showed neo-Timm's staining in the fascia dentata inner molecular layer (IML; Figs. 3C and D). Aberrant mossy fiber sprouting was observed in surgical cases as young as 2 months (18). The amount of IML neo-Timm's staining was less than that in adult cases with hippocampal sclerosis. Unlike hippocampal sclerosis cases, however, mossy fiber sprouting in this clinical group was not associated with significant hilar cell loss. Thus, seizures during early postnatal human development were associated with aberrant mossy fiber sprouting, but it does not appear to be from loss of hilar or CA3 neurons.


This study found that seizures during early human life were associated with anatomic signs of improper fascia dentata development. In surgical cases with early-onset seizures from extratemporal substrates, we found that granule cell densities were decreased beginning at an early age, PSA NCAM IR of immature granule cells and hilar neurons were decreased, TUNEL staining showed no signs of cell death, and 80% of cases showed aberrant IML mossy fiber sprouting. Taken together, these results support the concept that early-onset seizures adversely impair normal postnatal fascia dentata development, leading to alterations in granule cell number and axon circuitry. These anatomic changes are likely to be associated with improper granule cell and/or hippocampal functions, which might affect memory and learning and promote epileptogenesis as a consequence of early-onset childhood seizures.

Fascia dentata neurogenesis and seizures

Neurogenesis is a normal postnatal feature in the mammalian fascia dentata, and the results of our study and experimental literature indicate that seizures affect neurogenesis in the immature and mature brain differently. In mammals, granule cell neurogenesis continues postnatally, but differs depending on the species. In rats, ≥80% of granule cell development occurs between birth and PN 21, and the process continues throughout the animal's life (19,20). In the rhesus monkey and human, only 20% of the granule cell population is formed after birth, and neurogenesis is significantly reduced by early infant and toddler ages (17,21).

The findings from the current human study are consistent with previous rat and primate experimental studies, and indicate that early-onset seizures adversely affect fascia dentata neurogenesis. We found that recurrent seizures during early human development were associated with decreased granule cell numbers, and this was most likely from decreased neurogenesis rather than from cell death. It should be noted that we could sample hippocampal specimens at only one time point (surgery) in children with active epilepsy, and all patients were taking antiepileptic drugs (AEDs). We do not know if seizures and/or medical treatment were independent or cofactors associated with decreased fascia dentata neurogenesis in humans, nor whether neurogenesis would have increased and granule cell numbers recovered, had the seizures stopped. Thus, additional analysis of the human data will be necessary when the surgical patient population is larger to determine if age at seizure onset, AEDs, and so on alter PSA NCAM IR fascia dentata cell densities. The mechanism(s) by which seizures may alter human fascia dentata neurogenesis during development are unknown, but experimental studies suggest that it could be due to age, stress, and glutamate receptor activation (22,23).

Consequence of altered postnatal development from repeated seizures

Previous experimental studies indicated that recurrent seizures have pronounced negative effects on early rodent brain development, and our human anatomic study of surgically resected hippocampi is in agreement. The short- and long-term functional consequences of seizure-induced impaired fascia dentata neurogenesis and aberrant mossy fiber axon formation remain to be determined (24). However, our anatomic findings strongly support the notion that early-onset seizures are likely to affect hippocampal functions such as learning and memory. In addition, it is unclear what effects these anatomic changes might have on hippocampal epileptogenesis later in life. In other words, changes that occur in the fascia dentata at an early age could prevent normal brain functions or promote seizures later in life. Additional studies of the cognitive affects in this patient group will be necessary to determine if alterations in hippocampal pathology are associated with poor cerebral development. However, our data support the concept that early-onset seizures are detrimental to postnatal brain development, even in the absence of neuronal destruction of existing cells.

Acknowledgment: This work was supported at the UCLA Epilepsy Laboratory by NIH grants P01 NS02808 and RO1 NS38992, and at Ribeirão Preto by FAPESP 99/11729-2 and CNPq. Special thanks to James K. Pretorius for tissue processing and to Harry V. Vinters for pathological material.