Susceptibility of human glial cells to infection with human immunodeficiency virus (HIV
Expression patterns of glial fibrillary acidic protein (GFAP)-delta in epilepsy-associated lesional...
Transcript of Expression patterns of glial fibrillary acidic protein (GFAP)-delta in epilepsy-associated lesional...
Expression patterns of glial fibrillary acidicprotein (GFAP)-delta in epilepsy-associatedlesional pathologies
L. Martinian*, K. Boer†, J. Middeldorp‡, E. M. Hol‡, S. M. Sisodiya*, W. Squier§, E. Aronica†¶ and
M. Thom*
*Department of Clinical and Experimental Epilepsy, Institute of Neurology, University College, London, §Department of
Neuropathology, John Radcliffe Hospital, Oxford, UK, ‡Netherlands Institute for Neuroscience, Royal Netherlands
Academy of Arts and Sciences, Amsterdam, †Department of Neuropathology, Academic Medical Center, Amsterdam, and
¶Stichting Epilepsie Instellingen Nederland, Heemstede, the Netherlands
L. Martinian, K. Boer, J. Middeldorp, E. M. Hol, S. M. Sisodiya, W. Squier, E. Aronica and M. Thom (2009)
Neuropathology and Applied Neurobiology 35, 394–405_ 394..405
Expression patterns of glial fibrillary acidic protein (GFAP)-delta in epilepsy-associated lesional
pathologies
Aims: Glial fibrillary acidic protein (GFAP)-d is a novel
isoform that differs in its C-terminal sequence from other
GFAP isoforms. Previous studies suggest restriction of
expression to the subpial layer, subventricular zone and
the subgranular zone astrocytes, with an absence in
pathological conditions causing reactive gliosis. GFAP-d is
speculated to have roles in regulation of astrocyte size and
motility and a subpopulation of GFAP-d-positive glia may
be multipotent stem cells. The aim of this study was to
investigate its expression in common causes of lesion-
related refractory epilepsy. Methods: Hippocampal sclero-
sis (HS), focal cortical dysplasia (FCD) type IIB, cortical
tuberous sclerosis (TSC) lesions, gangliogliomas, grey
matter heterotopias and hemimegalencephaly from a
wide age range of patients using both surgical and post
mortem tissue specimens were studied. Results: GFAP-d
expression was observed in CA4 and CA1 astrocytes in HS
with less frequent labelling in the granule cell layer, even
where granule cell dispersion was present. No significant
labelling was noted in the subiculum in HS cases or in any
subfields in non-HS epilepsy cases. Balloon cells in FCDIIB
and hemimegalencephaly, giant cells in TSC and the astro-
cytic component of gangliogliomas showed immunoreac-
tivity, colocalizing with conventional GFAP. No neuronal
expression for GFAP-d was seen in any of the pathologies.
Quantitative analysis in 10 FCDIIB and five TSC cases
revealed greater numbers of GFAP-d-positive balloon cells
than conventional GFAP. There was no GFAP-d expression
within nodular heterotopia. Conclusions: GFAP-d expres-
sion patterns in HS overall appears to mirror regional
reactive gliosis. It is a useful marker for the demonstration
of balloon cells in FCD and TSC, which may be relevant to
their abnormal size and localization. The lack of GFAP-d
within heterotopia supports their composition from cells
destined for deeper cortical layers.
Keywords: balloon cells, focal cortical dysplasia, GFAP-delta, hippocampal sclerosis
Introduction
Glial fibrillary acidic protein (GFAP) is an intermediate
filament of which several isoforms have been described,
including a, b, g, d [1], D135, D164, Dexon 6 [2] and more
Correspondence: Maria Thom, Division of Neuropathology, Depart-ment of Clinical and Experimental Epilepsy, Institute of Neurology,Queen Square, London WC1N 3BG, UK. Tel: +44 020 7837 3611;Fax: +44 020 7916 9546; E-mail: [email protected]
394 © 2009 Blackwell Publishing Ltd
Neuropathology and Applied Neurobiology (2009), 35, 394–405 doi: 10.1111/j.1365-2990.2008.00996.x
recently k [3]. The predominant splice form, GFAP-a, is
expressed in the majority of astrocytes, is highly specific
for astroglial lineage, upregulated in response to injuries
and ageing and is widely used as a diagnostic indicator
of reactive gliosis. GFAP-b is predominantly found in
peripheral nervous system [4], whereas GFAP-g has been
identified in and outside the central nervous system, in
mouse bone marrow and spleen [5,6].
The GFAP-d or -delta, discovered by Condorelli [7], is
also referred to as GFAP-e [1,3]. It differs from other iso-
forms in its C-terminal tail sequence, the last two exons
being replaced with an alterative terminal exon, resulting
in a unique 41-amino-acid sequence [3]. Recent studies of
the expression patterns of GFAP-d in adult human tissues
have demonstrated specific localization in subpial astro-
cytes of the cerebral cortex, the subgranular zone of
hippocampus and subependymal layer of the cerebral
ventricles [8], the latter regions harbouring neural stem
cells in the normal adult human brain. The specific func-
tions of GFAP-d are not well characterized, but its expres-
sion influences filament stability and probably cell motility
[8–11]. Recently, GFAP-d expression has been demon-
strated in balloon cells (BC) in cortical dysplasia from
patients with epilepsy [12]. The aim of the present study
was to extend these studies and more widely investigate
the expression patterns of GFAP-d in common causes of
lesion-related refractory epilepsy, many of which are
either maldevelopmental in aetiology or elicit a reactive
gliosis. By comparison with canonical GFAP expression,
the potential of GFAP-d as a discriminative marker in the
diagnosis of these pathologies can be assessed.
Material and method
Case selection
Cases were selected from the neuropathology archives at
the Institute of Neurology and the National Hospital for
Neurology and Neurosurgery, London (UK), the John
Radcliffe Hospital, Oxford (UK), Academic Medical Center
(University of Amsterdam) in Amsterdam and University
Medical Center in Utrecht (Netherlands). The study was
approved by the Joint Research Ethics Committee of the
Institute of Neurology and the National Hospital for Neu-
rology and in Amsterdam where informed consent was
obtained for the use of brain tissue and for access to
medical records for research purposes. For all surgical
cases, patients underwent therapeutic surgical resection
for refractory epilepsy and the pathological tissue was
surplus to diagnostic requirement. The post mortem tissues
were from patients with long histories of focal epilepsy and
era-appropriate consent for retention of post mortem tissue
was granted by the next of kin.
In all, 74 epilepsy lesions were studied and included
surgical and post mortem tissues from a wide age range
as detailed in Table 1. Pathologies included focal cortical
dysplasia (FCD) type IIB (13 cases), hippocampal scle-
rosis (HS) [seven with granule cell dispersion (GCD),
seven without GCD, eight with bilateral or asymmetrical
damage at post mortem and six paediatric cases with HS
associated with cortical infarct), hemimegalencepahly
(HMEG) (five cases), cortical tubers from patients with
tuberous sclerosis (TSC) (six cases), ganglioglioma (GG)
(six cases) and laminar and nodular grey matter heteroto-
pia (five cases). In addition, temporal lobes from 10 epi-
lepsy surgical patients with no HS were studied as well as
control cases from 15 patients without history of epilepsy
and no significant neuropathology at surgery or autopsy,
from a wide patient age range.
Immunohistochemistry
Seven-micron-thick paraffin wax-embedded sections were
dewaxed and rehydrated through graded alcohols and
taken to water. Endogenous peroxidase was blocked with
3% hydrogen peroxide in deionized water for 10 min.
Sections were microwaved in Vector Antigen Retrieval
Solution (Vector, Burlington, CA, USA) for 12 min and
cooled for 20 min. For the study of FCD, HS and heteroto-
pias cases, sections were incubated with polyclonal
primary antibody against GFAP-d (1:5000, Chemicon
International, Temecula, CA USA) for 40 h at 4°C. For
TS, HMEG and GG cases, we used the identical rabbit
polyclonal GFAP-d antibody (first raised at the Netherlands
Institute for Neuroscience [8]; immunocytochemistry was
carried out as previously described [13]).
In addition, staining in each case was carried out with
a conventional GFAP antibody (which also recognizes
GFAP-d in addition to more abundant GFAP-a and
henceforth referred to as GFAP). Sections were digested
with Proteinase K enzyme and incubated with primary
rabbit polyclonal anti-GFAP (1:1500, DAKO, Glostrup,
Denmark) for 1 h. Labelling was detected with DAKO
ChemMate Envision (DAKO) and staining was visualized
with DAKO DAB + Chromogen. Tissue from a glioblas-
toma multiforme was used as a positive control. Negative
GFAP-delta expression in epilepsy lesional pathologies 395
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
controls were treated identically except that the primary
antibody was replaced with normal rabbit immunoglobu-
lin fraction from DAKO. Between all steps, sections were
washed with PBS and 0.05% Tween 20. Antibodies were
diluted in DAKO ChemMate Diluent.
Confocal immunofluorescence
For the study of FCD and HS pathologies, 7-mm sections
were dewaxed and rehydrated and washed in water.
Endogenous peroxidase was quenched with 3% hydrogen
peroxide and deionized water. Sections were microwaved
in antigen retrieval buffer (Vector). Protein blocking was
done with normal horse serum (Vector) followed by incu-
bation of primary antibodies overnight at 4°C. The com-
binations were rabbit polyclonal anti-GFAP-d (1:5000
dilution) with monoclonal anti-GFAP (1:30 dilution,
DAKO), CD34 (1:30 dilution, DAKO), anti-NeuN (1:1500
dilution, Chemicon), monoclonal anti-human nestin
(1:500, R&D systems) and polyclonal doublecortin
(1:200, Santa Cruz Biotechnology, Santa Cruz, CA, USA).
Sections were washed and incubated with secondary anti-
rabbit ImmPRESS (Vector) for polyclonal GFAP-d followed
by Cy3 tyramide signal amplification (PerkinElmer Life
and Analytical Sciences, Boston, MA, USA). Sections then
were washed and quenched in 1% hydrogen peroxide in
PBS for 20 min in order to prevent any deposited tyramide
combining with the second tyramide signal that followed.
The sections were incubated with anti-mouse ImmPRESS
for monoclonals GFAP and NeuN; for doublecortin, the
detection system was Avidin Biotin Kit from Santa Cruz
followed by fluorescein tyramide (Perkin Elmer, Boston,
MA, USA). Sections were mounted on Vectashield with
Dapi (Vector) and visualized with a Zeiss (Gena, Germany)
LSM 510 Meta confocal laser microscope. For TS cases
sections, after incubation with the primary antibodies,
were incubated for 2 h at room temperature with Alexa
Fluor® 568-conjugated anti-rabbit or Alexa Fluor® 488
anti-mouse IgG (1:1000, Molecular Probes, Eugene, OR,
USA). Sections were then analysed by means of a laser
scanning confocal microscope (Bio-Rad, Hercules, CA,
USA; MRC1024) equipped with an argon-ion laser.
Table 1. Details of cases studied according to pathology group
Pathology type (PM or S)
Number of
cases in group
Mean age at surgery or
death (range) (years) Localization Mean duration of epilepsy (range)* (years)
FCD IIB (S) 12 33 (14–52) Temporal (6)
Frontal (6)
20.5 (9–41)
FCD IIB (PM) 1 59 Frontal
HS with GCD (S) 7 24 (10–40) Temporal 19.6 (9–31)*
HS without GCD (S) 7 33.2 (18–55) Temporal 21 (7–36)*
Bilateral HS (PM) 8 69.75 (38–90) Bilateral
Temporal
62.8 (31–74)
Paediatric HS with neonatal infarct (S) 6 14 (9–20) Temporal All childhood onset of seizures*
No HS (S) 10 27.2 (14–36) Temporal 18 (14–25)*
HMEG (S) 6 18 (2–24) months CH 4.5 (2–7) months
Cortical tubers (TSC) (S) 6 20.3 (6–35) Temporal (2)
Frontal (4)
15.1 (4–34)
Ganglioglioma (S) 6 30 (16–60) Temporal 15.8 (9–25)
Grey matter heterotopias (PM) 5 48.25 (25–69) Bilateral (3)
Temporal (1)
Localized (1)
37.25 (23–69)
Controls adult (S) 4 37 (27–57) Temporal (4) NA
Controls adult (PM) 6 51.6 (26–79) Temporal (6) NA
Controls paediatric (PM) 5 4.2 (1–15) months Temporal (5) NA
Clinical details of the groups of pathologies studied. *In some cases, accurate data on duration of seizures in all cases in group were not available.
CH, cerebral hemisphere; FCD, focal cortical dysplasia; GCD, granule cell dispersion; HMEG, hemimegalencephaly; HS, hippocampal sclerosis;
NA, not applicable; PM, Post mortem tissues; S, surgical tissues; TSC, tuberous sclerosis.
396 L. Martinian et al.
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
Quantitative analysis in FCD cases
In 10 cases of FCDIIB and five TS cases, comparative
quantitative analysis was carried out for the numbers of
immunoreactive (IR) BC in the white matter on adjacent
sections stained with GFAP-d and GFAP. Using Histometrix
software (Kinetic Imaging, Liverpool, UK) and image
acquisition and analysis software (Image Pro Plus vs 5.02;
Media Cybernetics, Silver Spring, MD, USA), an identical
region of interest in the white matter, beneath the region
of dysplasia or tuber, was outlined at low magnification
(¥2.5 objective). One slide per case was selected and the
mean area for quantitative analysis was 5.95 mm2. All BC
within this region were counted systematically at high
magnification (¥40 objective) as positive IR (including
strong or intermediate intensity of labelling) or negative.
The density of BC (or giant cells) per mm2 and percentage
of labelled BC (or giant cells) compared with the total
number was compared with both antibodies.
Results
In the control tissues, GFAP-d IR-glial cells were present in
the subpial region of layer I. Small cells, with short bi- or
multipolar processes, sometimes extending horizontally
along the limiting margin of the brain, were identified
discontinuously along the subpial region with occasional
cells deeper in layer I (Figure 1a). Colocalization of GFAP-d
was seen with GFAP in a proportion of small glial cells
in cortical layer I, including the subpial region. GFAP-d
reactivity was confined to the cell bodies whereas GFAP
showed formation of extensive fibre networks and was
present in greater numbers of cells (Figure 1a). GFAP-d
IR-glia was not present in the deeper cortex, but small
numbers of GFAP-d IR astrocytic-appearing cells were
present in the white matter.
Focal cortical dysplasia
Immunostaining patterns with GFAP-d were consistent
in all the cases studied. The majority of the BC showed dif-
fuse cytoplasmic positivity with variable intensity of label-
ling, many of the cells showing multipolar processes
(Figure 1b,c). IR-BC were present both in the cortex and
white matter. Labelling of small astrocytic cells was also
noted, some with multiple nuclei, present in the region of
dysplasia as well as in cortical layer I in the subpial region.
Dysplastic neurones were not labelled with GFAP-d
(Figure 1d), which was confirmed by double labelling with
NeuN (Figure 1h). In contrast to staining with conven-
tional GFAP, where BC were often obscured by dense mesh-
works of IR-glial fibres, GFAP-d distinctly demonstrated
BC (Figure 1e–g). Quantitative analysis revealed that
lower numbers per mm2 of IR-BC were consistently
present in the white matter with conventional GFAP com-
pared with GFAP-d. GFAP-d also showed a higher mean
percentage positivity of BC of 82% compared with 64%
with GFAP (Table 2). There was no correlation between
the density of GFAP-d IR-BC and patient age, sex, age of
onset or duration of seizures.
Hippocampal sclerosis (surgical cases)
A similar regional staining pattern was present in all
cases of classical HS. Prominent numbers of distinct,
single IR-glial cells were present in the CA1 subfield in
regions of neuronal loss in the stratum pyramidale and
adjacent white matter (stratum radiatum) (Figure 2a).
The junction between CA1 and subiculum showed an
abrupt transition, with the subiculum being devoid of
GFAP-d-positive cells (Figure 2b). The IR-cells were multi-
polar in morphology with short cytoplasmic processes and
frequent binucleate and multinucleate cells were identified
(Figure 2c). The dense fibrillary meshworks of fibres
typically present in chronic HS with conventional GFAP
were not visualized with GFAP-d staining. Frequent IR-
cells with similar morphology were present in the hilus
(Figure 2c), in some cases particularly in the subgranular
zone (Figure 2d). The characteristic radial glial fibre
pattern in the dentate gyrus, as observed with conven-
tional GFAP, was not prominent with GFAP-d (Figure 2h).
Furthermore, HS cases with GCD did not show qua-
litatively different patterns to those without GCD
(Figure 2d,e). Cases with no HS showed small numbers of
IR-cells in the hilus, some in the subgranular zone, but
little positivity in CA1 sector. The temporal lobe adjacent
to HS showed varying numbers of GFAP-d-positive in the
white matter (Figure 2f). In all cases, IR-cells were present
along the ventricular border of the temporal horn.
Hippocampal sclerosis (post mortem cases)
In cases with unilateral HS, the morphology and distribu-
tion of IR-cells was similar to that observed in the surgical
cases of classical HS. Small IR-glial cells were noted in the
hilus, subgranular zone and CA1 region, but they were
GFAP-delta expression in epilepsy lesional pathologies 397
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
less conspicuous in the dentate gyrus. However, in many
of the post mortem cases, the overall staining reaction in
the hippocampus was weaker compared with surgical
cases of HS, although subventricular zone astrocytes often
retained strong labelling intensity. In the contralateral
preserved hippocampus, occasional IR-hilar cells and
immunopositive fibres were observed.
In post mortem cases with bilateral HS, GFAP-d expres-
sion was seen in both sides in the hilar region and CA1.
Where the severity of HS appeared asymmetrical, GFAP-d
IR-cells were more frequent on the side, showing the more
severe cell loss.
Paediatric temporal lobectomies
In two patients, classical patterns of HS were present and
GFAP-d IR patterns were similar to adult cases. In three
cases with end folium HS, GFAP-d IR-cells were present in
the CA4 region only and in a single case without HS, no
GFAP-d-positive cells were observed in the hippocampus,
even in the subgranular region. In all cases, GFAP-d-
positive cells were present in the subventricular zone,
around the hippocampal fissure (Figure 2g) and also
extensively in the region of cortical scarring at the site of
the old infarcts.
Figure 1. Focal cortical dysplasia (FCD) type IIB. (a) Glial fibrillary acidic protein (GFAP)-d-positive astrocytes in the marginal subpial zone
of the cortex (arrow) in a control case with expression relatively limited to the cell body. Inset shows double labelling of GFAP-d (cy3: red)
with conventional GFAP (FitC: green) in the same region which demonstrates colocalization in a few glial cells, but more extensive processes
and fibre formation are visualized with GFAP (FitC: green). (b) GFAP-d showed intense labelling of balloon cells in FCD type IIB cases, many
showing striking multipolar cytoplasmic processes. (c) A variability of intensity of immunolabelling with GFAP-d was noted, with some
balloon cells appearing weak or negatively labelled (arrow). (d) Dysplastic and hypertrophic neurones appeared negative with GFAP-d;
however, prominent numbers of small, bi- and multinucleate cells were present throughout the region of the dysplasia (inset). (e) White
matter beneath region of dysplasia clearly delineates the balloon cell populations whereas labelling with conventional GFAP (FitC: green)
(f) in the same region demonstrates that the balloon cells are partially obscured by the dense glial fibres. (g) Combination of GFAP-d with
conventional GFAP (FitC: green) confirms colocalization of labelling in balloon cells with delta isoform being largely restricted to the cell
body and absent in the fibres of the gliosis. (h) GFAP-d (cy3: red) with NeuN (cy3: red) in FCD IIB showed no colocalization of staining (scale
bars in a,c,d,e,f = 100 microns, b = 10 microns, g = 20 microns, h = 30 microns).
398 L. Martinian et al.
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
TSC lesions, HMEG and GG
In cortical tubers, moderate to strong intensity of labelling
of a proportion of giant cells (Figure 3a,c) was seen in all
cortical layers for GFAP-d. Many of these cells exhibited
multipolar processes and in addition, multinucleate cells
and smaller reactive astrocytic cells were seen. In HMEG,
similar labelling of balloon-like cells was noted, although
a proportion was negative or weakly labelled (Figure 3d).
Larger GFAP-d IR astrocytes were noted in GG, but
were infrequent. In all these pathologies, no labelling of
dysplastic or abnormal neuronal cell types was seen
(Figure 3b,e,f). Quantification of giant cells in TSC lesions
also showed a higher proportion and number per mm2
with the GFAP-d compared with conventional GFAP.
Grey matter heterotopia
The GFAP-d showed labelling of small astrocytes with short
processes in the subpial region of the overlying cortex and
in the periventricular tissue beneath the ependymal lining.
There were no IR-cells within the heterotopia, even where
rudimentary laminar architecture was present. In only one
case, small numbers of IR-glia were noted around the
margin of some of the nodules (Figure 3g). Comparison
with conventional GFAP staining patterns in this case
showed marked gliosis surrounding nodules with glial
fibres extending into the nodules (Figure 3h), and focal
variable astrocytic gliosis was also confirmed within the
heterotopia in all other cases.
Co-expression of GFAP-d with
developmental markers
The GFAP-d IR-BC in FCD cases showed evidence of
co-expression with developmentally regulated protein
nestin (Figure 4a–c); co-expression with CD34 and dou-
blecortin in BC was not identified. Giant cells in TS and
HMEG also show co-expression of GFAP-dand nestin. In HS
cases, co-expression of GFAP-dwith nestin was noted in the
small and binucleate astrocytes in the region of sclerosis
(Figure 4d–f), but not with doublecortin or CD34 in these
cell types (Figure 4j–l). In control tissues, a lack of colocal-
ization between GFAP-d and doublecortin-expressing cells
in the subpial region was noted (Figure 4g–i).
Discussion
We have described consistent expression patterns of
GFAP-d, an isoform of GFAP, in a large series of
Table 2. Quantification of BC in FCD IIB and GC in TSC with GFAP and GFAP-d immunohistochemistry
Case
(remove numbers)
Percentage of GFAP-d
BC/GC of total number (n)
Percentage of GFAP
BC/GC of total number (n)
Density of GFAP-d
positive BC/GC (mm2)
Density of GFAP-positive
BC/GC (mm2)
FCD
1 95 (110) 87 (41) 10.53 1.8
2 75 (62) 77 (54) 2.53 2.27
3 97 (47) 86 (22) 2.09 0.86
4 93 (322) 86 (45) 14.7 1.88
5 98 (397) 87 (245) 19.45 11.23
6 69 (74) 45 (45) 4.56 0.95
7 74 (69) 52 (62) 8.23 2.32
8 67 (83) 45 (34) 7.56 1.67
9 75 (121) 41 (38) 9.32 2.89
10 76 (115) 38 (48) 10.65 2.54
TSC
1 74 (97) 49 (66) 9.87 2.76
2 71 (78) 38 (75) 8.10 2.56
3 78 (103) 41 (32) 11.67 1.78
4 63 (65) 35 (34) 6.94 2.67
5 67 (74) 26 (24) 7.89 1.82
The number in parentheses in columns 2 and 3 is the total number of cells counted in region of interest. BC, balloon cells in focal cortical
dysplasia (FCD) (cells with both intense and intermediate staining were included and expressed as a percentage of all balloon cells in the white
matter); GC, giant cells in tuberous sclerosis (TSC) lesion (similar analysis was carried out on white matter); GFAP-d, glial fibrillary acidic
protein-delta compared with conventional GFAP (DAKO), which recognizes both a and d isoforms.
GFAP-delta expression in epilepsy lesional pathologies 399
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
pathological lesions associated with focal epilepsy from a
wide age range of patients. GFAP intermediate filaments
are part of the cytoskeleton of glial cells and all share a
common structure of a highly conserved a-helical rod-
like domain with nonhelical head and tail domains. The
delta isoform of GFAP (which has also been termed
GFAP-e [3]) with its unique tail domain has been little
studied in human disease processes to date. It is consid-
ered to have roles in intermediate filament assembly and
stability [3], cell motility and migration [8] and may
interact with other cytoplasmic and nuclear proteins,
including histones [7].
Hippocampal sclerosis is the commonest pathology
associated with temporal lobe epilepsy [14]. In HS, a
subfield-specific dense fibrillary gliosis, as highlighted with
conventional GFAP labelling, accompanies the neuronal
loss in CA1 and CA4 in the classical pattern. In our series,
GFAP-d expression consistently showed a similar distribu-
tion mirroring this pattern of gliosis. The morphology of
the GFAP-d IR astrocytes differed from that shown by con-
ventional GFAP, with predominantly a perinuclear stain-
ing pattern and frequently in binucleate cells and showed
colocalization with nestin. This later novel finding could
suggest a role of GFAP-d in glial cell proliferation and a
Figure 2. Hippocampal sclerosis. (a) Surgical case of hippocampal sclerosis with glial fibrillary acidic protein (GFAP)-d expression
highlighting the regions of gliosis and neuronal loss, in particular CA1 and the hilus with an abrupt transition of staining appreciated
between CA1 and the subiculum (S) and shown at higher magnification in (b). (c) Hilus showing small, distinct glial cell populations with
short cellular processes labelling with GFAP-d; inset showing binucleate GFAP-d-positive cells, which were frequently observed. (d) GFAP-d
cells were more frequent in the hilus than in the granule cell layer (GCL) in case without granule cell dispersion (GCD). (e) In cases with HS
and GCD, radial glial fibres were also not highlighted and (e) illustrates the case with maximal staining in the GCL, which was always less
than the hilus. (f) The white matter of the temporal lobe adjacent to hippocampal sclerosis showed GFAP-d-positive cells. (g) In paediatric
hippocampal sclerosis cases, similar regional staining patterns with GFAP-d were observed and, as in controls, prominent glial labelling
around the ventricle and along the hippocampal fissure (arrowed) was noted. (h) Double labelling with GFAP (FITC: green) and GFAP-d (cy3:
red) in the GCL in hippocampal sclerosis confirmed that the numerous radial glial fibres did not label with delta that colocalized only to a
small number of glial cells in this region (scale bars in a,g = 1000 microns, b = 100 microns, c–f = 10 microns, h = 20 microns).
400 L. Martinian et al.
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
presence in more immature astrocytes, which requires
further study. In addition, the lack of staining of fibrillary
meshworks of fibres with GFAP-d allowed clearer delinea-
tion of the astrocyte cell bodies, which could potentially
facilitate quantitative analysis. Colocalization with con-
ventional GFAP confirmed that GFAP-d expression was
present in a subset of glial cells in the region of gliosis. As
the conventional GFAP antibody we use recognizes both
GFAP-a and -d, it was not possible to determine if expres-
sion of GFAP-d isoform in HS occurs in a distinct glial
subpopulation. Previous cell transfection studies have
shown that GFAP-a has a better intrinsic filament-
forming capacity than GFAP-d, which is less able to form
multimers [3,8,15]. This could explain the lack of demon-
stration of GFAP-d in fibrillary processes in HS, including
in the radial glial fibres in the dentate gyrus.
It is known that GFAP-d is expressed at 100-fold lower
levels than conventional GFAP isoforms in normal brain
[3]. In astrocytic gliosis associated with multiple sclerosis
or Alzheimer’s disease, GFAP-d was barely detectable
using immunohistochemistry, using the same antibody
as used in our study [8]. The ratio of GFAP-a : -d tran-
scripts is highly correlated in the subventricular astro-
cytes where GFAP-d transcripts are disproportionally
more abundant [8]. Whether a strict ratio is maintained
in other anatomical regions and in conditions of patho-
logical gliosis, or if selective splicing of GFAP-d isoforms
occurs in hippocampal astrocytes in epilepsy remains to
be studied. Furthermore, expression patterns of GFAP-d
may alter in the developing brain and with cell matura-
tion [3]. In paediatric cases of HS, as well as in some
adults, it is likely that the gliotic process commences
Figure 3. Tuberous slerosis and other pathologies. (a) Giant cells (top) and multinucleate cells (bottom) were labelled with glial fibrillary
acidic protein (GFAP)-d in tuberous sclerosis cortical lesions, but (b) dysmorphic neurones were negative. (c) GFAP and GFAP-d colocalized in
the giant cells. (d) Abnormal balloon-like glia cells in hemimegalencepahly cases also showed variable intensity of labelling with GFAP-d, but
abnormal neurones were negative (e) and similarly in ganglioglioma (f). (g) Heterotopia showed mild marginal staining at the edge of
nodules in one case with GFAP-d (arrowed), although no positive labelling was seen within the nodule whereas GFAP immunolabelling on
adjacent section (h) confirmed gliosis at the margins of the nodules (arrow) as well as within the nodules in all cases (inset) (scale bars in
a = 27 microns, b,c,e = 40 microns, d,f = 80 microns and g,h = 100 microns).
GFAP-delta expression in epilepsy lesional pathologies 401
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
Figure 4. Balloon cell in focal cortical dysplasia labelled with glial fibrillary acidic protein (GFAP)-d (a) and nestin (b), showing
colocalization within the cytoplasm (c). Similarly in HS cases, small and binucleate GFAP-d-positive astrocytic cells colocalized with nestin
(e; merged in f ). In control tissues, GFAP-d cells in the supial region (g) did not show colocalization with the small number of
doublecortin-positive cells identified (h; merged in i). In the hilus in HS cases, similarly GFAP-d-positive astrocytes (j) did not label with
doublecortin that labelled occasional hilar neurones (k; merged in l). Bars in a–c = 23 microns, in d–f = 40 microns and in g–l = 20 microns.
402 L. Martinian et al.
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
early in life, possibly following an initial injury. Therefore,
the time of onset of gliosis could be a factor explaining
the discrepancy between our finding of GFAP-d in gliosis
associated with HS, but not in Alzheimer’s disease
and multiple sclerosis, acquired during adulthood [8].
Examination of post mortem cases with long histories of
epilepsy also allowed us to assess GFAP-d expression
patterns over time in comparison with surgical cases
with shorter histories. We identified strongly IR-glial cells
in CA1 in patients with HS with over 70 years of dura-
tion of epilepsy, although in other cases subventricular
astrocytes were labelled more intensely than in the scle-
rotic hippocampus.
Regional GFAP-d expression in subventricular, sub-
granular and subpial zone astrocytes in normal brains [8]
persists in disease states as shown in our present surgical
and post mortem epilepsy series. This could support an
important functional or regulatory role for specific glial
populations in these regions. For example, it has been
proposed that a subpopulation of GFAP-d-positive cells
represent multipotent stem cell [8]. In human adult
brain, a subventricular astrocyte ribbon has been identi-
fied that acts as a source of multipotent progenitor cells
[16]. In epilepsy, one possibility regarding the cause of
GCD, a common finding in HS [14], is excess proliferation
of subgranular zone progenitor cells as a result of sei-
zures [17,18]. We examined cases of HS with and
without GCD, but found no difference in GFAP-d expres-
sion patterns in the subgranular zone. An alternative
theory for GCD favours the formation of the radial glial
scaffold in the dentate gyrus as a key factor in inducing
the cell migration [19]. It has been shown that when
GFAP-d expression reaches a critical level, collapse of the
intermediate filament cytoskeleton occurs [8,15]. The
lack of GFAP-d that we noted in radial fibres in HS could
suggest that this relative reduction in these glial cells,
compared with other GFAP isoforms, is a factor confer-
ring stability of the glial network and thus promoting
GCD [20].
Our finding of GFAP-d expression in a large proportion
of BC in FCD IIB confirms the findings of a recent study
[12]. We have also confirmed similar expression in the
giant cells and balloon-like cells associated with TSC and
HMEG. As GFAP-d is largely absent in the fibrillary gliosis
associated with many FCD cases, it appears a superior
marker to conventional GFAP antibodies that also recog-
nize the more abundant GFAP-a-positive fibres, obscur-
ing many BC. This was supported by our quantitative
analysis of BC using the different GFAP antibodies and
may also reflect greater affinity or binding of the specific
GFAP-d antibody for this isoform, although there are no
data yet available to support this. This clearer identifica-
tion and delineation of BC with GFAP-d could be of diag-
nostic value in the classification of FCD subtypes [21]. It
has been suggested that GFAP-d expression in BC may
reflect their progenitor cell phenotype from ventricular
zone astrocytes [12]. As it has been proposed that
GFAP-d has roles in moulding the shape and size of the
cell through its effects on the cytoskeleton and in influ-
encing cell motility [8], any abnormal expression levels of
GFAP-d, or alteration in ratio of GFAP-a : -d could be a
factor determining the abnormal localization (white
matter rather than cortical plate) and morphology of BC
and giant cells. The expression of GFAP isoforms has
been shown to be tightly regulated [22], and as well as
developmental regulation, cell stress could influence the
relative ratio of GFAP transcripts. These possibilities, in
the context of any deregulation of GFAP-d expression in
BC in patients with refractory epilepsy require further
investigation. In addition, neuronal expression of some
GFAP isoforms has been shown in patients with Alzhe-
imer’s disease [2] and binding of GFAP-d with transmem-
brane proteins presenilin I and II has been shown [1]. In
the present series, which included FCD cases with dys-
morphic neurones bearing neurofibrillary tangles (as
previously reported [23]), no neuronal expression for
GFAP-d was observed as was confirmed on double label-
ling studies.
In patients with grey matter heterotopia, we demon-
strate an absence of GFAP-d glial cells within the hetero-
topia in all cases, in contrast to conventional GFAP that
showed variable astrocytic gliosis. This would support the
lack of contribution of subpial-layer or marginal-zone cell
types to the heterotopia and the proposal that heterotopia
arise from cells destined for deeper cortical layers (II–VI).
We have previously reported the lack of reelin-positive
Cajal–Retzius-like cells, also normally present in cortical
layer I, within heterotopia [24],which reinforces this
suggestion.
In conclusion, GFAP-d is expressed in a variety of
pathologies associated with epilepsy. Increased expression
is shown in regions of astrocytic gliosis in HS and in BC in
FCD IIB and allied lesions. Whether there is selective tran-
scription for GFAP-d isoform in epilepsy will be important
to establish in future studies due to the important role
this protein has in influencing the cell cytoskeleton and
GFAP-delta expression in epilepsy lesional pathologies 403
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
cell motility, which may be relevant to many of the cyto-
architectural and developmental aberrations observed in
patient with focal epilepsy.
Acknowledgements
This work has been supported by a grant from the MRC
(grant number G79059) and the UCLH/UCL Biomed
Centre (LM, SS, MT). Also the National Epilepsy Fund,
NEF 05-11 (EA,KB); EU-FP7-project 202167, Stichting
Michelle M06.011, M07.016 (EA); Hersenstichting Ned-
erland 13F05.08, 15F07.40 (EH, JM); and ISAO 04511
(EH, JM). We would like to thank Dr. W.G.M. Spliet (neu-
ropathologist; Department of Pathology, UMCU) for the
collaboration in the collection of the cases.
References
1 Nielsen AL, Holm IE, Johansen M, Bonven B, Jorgensen P,
Jorgensen AL. A new splice variant of glial fibrillary acidic
protein, GFAP epsilon, interacts with the presenilin pro-
teins. J Biol Chem 2002; 277: 29983–91
2 Hol EM, Roelofs RF, Moraal E, Sonnemans MA, Sluijs JA,
Proper EA, de Graan PN, Fischer DF, van Leeuwen FW.
Neuronal expression of GFAP in patients with Alzheimer
pathology and identification of novel GFAP splice forms.
Mol Psychiatry 2003; 8: 786–96
3 Blechingberg J, Holm IE, Nielsen KB, Jensen TH,
Jorgensen AL, Nielsen AL. Identification and character-
ization of GFAPkappa, a novel glial fibrillary acidic
protein isoform. Glia 2007; 55: 497–507
4 Hagiwara N, Imada S, Sueoka N. Cell-type specific segre-
gation of transcriptional expression of glial genes in the
rat peripheral neurotumor RT4 cell lines. J Neurosci Res
1993; 36: 646–56
5 Zelenika D, Grima B, Brenner M, Pessac B. A novel glial
fibrillary acidic protein mRNA lacking exon 1. Brain Res
Mol Brain Res 1995; 30: 251–8
6 Brenner M. Structure and transcriptional regulation of
the GFAP gene. Brain Pathol 1994; 4: 245–57
7 Condorelli DF, Nicoletti VG, Barresi V, Conticello SG,
Caruso A, Tendi EA, Giuffrida Stella AM. Structural
features of the rat GFAP gene and identification of a
novel alternative transcript. J Neurosci Res 1999; 56:
219–28
8 Roelofs RF, Fischer DF, Houtman SH, Sluijs JA, Van Haren
W, Van Leeuwen FW, Hol EM. Adult human subventricu-
lar, subgranular, and subpial zones contain astrocytes
with a specialized intermediate filament cytoskeleton.
Glia 2005; 52: 289–300
9 Elobeid A, Bongcam-Rudloff E, Westermark B, Nister M.
Effects of inducible glial fibrillary acidic protein on glioma
cell motility and proliferation. J Neurosci Res 2000; 60:
245–56
10 Goldman RD, Khuon S, Chou YH, Opal P, Steinert PM.
The function of intermediate filaments in cell shape
and cytoskeletal integrity. J Cell Biol 1996; 134: 971–
83
11 Lepekhin EA, Eliasson C, Berthold CH, Berezin V, Bock E,
Pekny M. Intermediate filaments regulate astrocyte
motility. J Neurochem 2001; 79: 617–25
12 Lamparello P, Baybis M, Pollard J, Hol EM, Eisenstat DD,
Aronica E, Crino PB. Developmental lineage of cell types
in cortical dysplasia with balloon cells. Brain 2007; 130:
2267–76
13 Aronica E, Yankaya B, Jansen GH, Leenstra S, van
Veelen CW, Gorter JA, Troost D. Ionotropic and
metabotropic glutamate receptor protein expression in
glioneuronal tumours from patients with intract-
able epilepsy. Neuropathol Appl Neurobiol 2001; 27:
223–37
14 Wieser HG. ILAE Commission Report. Mesial temporal
lobe epilepsy with hippocampal sclerosis. Epilepsia 2004;
45: 695–714
15 Nielsen AL, Jorgensen AL. Self-assembly of the cytoskel-
etal glial fibrillary acidic protein is inhibited by an
isoform-specific C terminus. J Biol Chem 2004; 279:
41537–45
16 Sanai N, Tramontin AD, Quinones-Hinojosa A, Barbaro
NM, Gupta N, Kunwar S, Lawton MT, McDermott MW,
Parsa AT, Manuel-García Verdugo J, Berger MS, Alvarz-
Buylla A. Unique astrocyte ribbon in adult human brain
contains neural stem cells but lacks chain migration.
Nature 2004; 427: 740–4
17 Parent JM, Yu TW, Leibowitz RT, Geschwind DH, Sloviter
RS, Lowenstein DH. Dentate granule cell neurogenesis
is increased by seizures and contributes to aberrant
network reorganization in the adult rat hippocampus. J
Neurosci 1997; 17: 3727–38
18 Thom M, Martinian L, Williams G, Stoeber K, Sisodiya
SM. Cell proliferation and granule cell dispersion in
human hippocampal sclerosis. J Neuropathol Exp Neurol
2005; 64: 194–201
19 Fahrner A, Kann G, Flubacher A, Heinrich C, Freiman
TM, Zentner J, Frotscher M, Haas CA. Granule cell disper-
sion is not accompanied by enhanced neurogenesis in
temporal lobe epilepsy patients. Exp Neurol 2007; 203:
320–32
20 Perng MD, Wen SF, Gibbon T, Middeldorp J, Sluijs JA,
Hol EM, Quinlan RA. GFAP filaments can tolerate the
incorporation of assembly-compromised GFAP-{delta},
but with consequences for filament organization and
{alpha}B-crystallin association. Mol Biol Cell 2008; 19:
4521–33
21 Palmini A, Najm I, Avanzini G, Babb T, Guerrini R,
Foldvary-Schaefer N, Jackson G, Lüders HO, Prayson R,
Spreafico R, Vinters HV. Terminology and classification
of the cortical dysplasias. Neurology 2004; 62: S2–8
404 L. Martinian et al.
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405
22 Blechingberg J, Lykke-Andersen S, Jensen TH, Jorgensen
AL, Nielsen AL. Regulatory mechanisms for 3′-end alter-
native splicing and polyadenylation of the Glial Fibrillary
Acidic Protein, GFAP, transcript. Nucleic Acids Res 2007;
35: 7636–50
23 Sen A, Thom M, Martinian L, Harding B, Cross JH, Nikolic
M, Sisodiya SM. Pathological tau tangles localize to focal
cortical dysplasia in older patients. Epilepsia 2007; 48:
1447–54
24 Thom M, Martinian L, Parnavelas JG, Sisodiya SM.
Distribution of cortical interneurons in grey matter het-
erotopia in patients with epilepsy. Epilepsia 2004; 45:
916–23
Received 11 June 2008
Accepted after revision 3 October 2008
Published online Article Accepted on 29 October 2008
GFAP-delta expression in epilepsy lesional pathologies 405
© 2009 Blackwell Publishing Ltd, Neuropathology and Applied Neurobiology, 35, 394–405