Acta Neuropathol (2002) 104 : 188–196 DOI 10.1007/s00401-002-0540-x
R E G U L A R PA P E R
Michael W. Johnson · Hajime Miyata · Harry V. Vinters
Ezrin and moesin expression within the developing human cerebrum and tuberous sclerosis-associated cortical tubers
Received: 25 October 2001 / Revised: 4 February 2002 / Accepted: 4 February 2002 / Published online: 5 June 2002 © Springer-Verlag 2002
Abstract The ERM (ezrin, radixin, and moesin) proteins belong to the band-4.1 superfamily of membrane-cytoskeleton-linking proteins which bind to the actin cytoskeleton via their C-terminal sequences and bind ERM binding membrane proteins (ERMBMPs). We investigated the immunohistochemical expression of two of the ERM proteins (ezrin and moesin) in developing human cerebral cortex and in cortical tubers from patients with tuberous sclerosis (TSC), to assess possible consequences of TSC gene product malfunction or inactivation in the developing brain in relation to ERM protein expression. Ezrin is abundantly expressed within radial glia and migrating cells in the intermediate zone in the prenatal human cerebrum, while moesin is primarily expressed in vascular endothelial cells in developing and adult human brain and scattered microglia in adult brain. In addition, both ezrin and moesin are abundantly co-expressed with hamartin and tuberin within a population of abnormal cells in TSCassociated cortical tubers. The expression of these two proteins – primarily ezrin – suggests that they are developmentally regulated and abundantly expressed in germinal matrix and/or migrating cells during cerebral cortical development. In TSC-associated cortical tubers, both proM.W. Johnson Interdepartmental Program in Neuroscience, UCLA Medical Center, Los Angeles, CA 90095-1732, USA M.W. Johnson · H. Miyata · H.V. Vinters (✉) Section of Neuropathology, Department of Pathology and Laboratory Medicine (Neuropathology), UCLA Medical Center, CHS 18–170, Los Angeles, CA 90095-1732, USA e-mail:
[email protected], Tel.: +1-310-8256191, Fax: +1-310-2068290 H.V. Vinters Brain Research Institute, Mental Retardation Research Center and Neuropsychiatric Institute, UCLA Medical Center, Los Angeles, CA 90095-1732, USA H. Miyata Department of Neuropathology, Institute of Neurological Sciences, Faculty of Medicine, Tottori University, Yonago, Tottori 683-8504, Japan
teins appeared to be up-regulated and are co-localized within a population of abnormal neuroglial cells typical of those seen in tubers. Expression of these proteins and their co-localization with tuberin and hamartin in these cells may suggest a compensatory up-regulation in response to TSC gene mutation. Keywords Tuberous sclerosis · Ezrin · Moesin · Radial glia · Immunohistochemistry
Introduction Tuberous sclerosis (TSC) is a bigenic autosomal dominant disorder caused by mutations in either of two genes, TSC1 or TSC2, and characterized histologically by the formation of benign tumors in multiple organs [8]. Hamartin and tuberin, the TSC1 and TSC2 gene products, have been suggested to regulate cell proliferation and cell cycle progression [7, 17]. These reports support the designation of the TSC genes as tumor suppressor genes and suggest that hamartin and tuberin inhibit abnormal cellular growth within human tissues [31]. However, the specific causal role(s) of TSC gene mutations in hamartomatous malformations within the viscera or central nervous system (CNS) of affected patients is not known. Recent work has shown that a portion of the hamartin protein binds to an N-terminal domain highly conserved within a group of proteins including ezrin, radixin, and moesin (ERM proteins). In addition, exogenous over-expression of an N-terminal portion of hamartin, which does not include the ERM binding domain, in umbilical endothelial cells causes increased activation of the RhoA GTPase and assembly of focal adhesions. The result of this activation was loss of cell adhesion to extracellular substrate [21]. The ERM proteins belong to the band-4.1 superfamily of membrane-cytoskeleton-linking proteins [33]. These and related proteins have been shown to bind to the actin cytoskeleton via their C-terminal sequences and bind ERM-binding membrane proteins (ERMBMPs), which are integral membrane proteins or membrane ‘adaptor’ mole-
189
cules, via their conserved N-terminal domains [2, 23]. The ERM proteins are thought to function in many different fashions according to their interaction with various membrane proteins, Ras family GTPases, and the actin cytoskeleton in response to extracellular contact and chemical signals [23]. Much of the current research regarding these proteins has concentrated upon the essential role of ERM proteins in the formation of intercellular focal adhesions and contacts [23]. These two types of local clustering of F-actin filaments, ERM proteins, and Rho family GTPases appear to be important in cellular adhesion and motility [18]. Evidence from several reports suggests that ERM proteins function upstream and downstream of Rho GTPases to regulate these events [25, 34, 35]. The evidence that hamartin, the TSC1 gene product, binds to ezrin (and purportedly other ERM proteins) in vivo and its demonstrated ability to modulate the activity of the Ras superfamily GTPase, RhoA, suggest that tuberin and hamartin may be attached to the membrane-cytoskeletal cortex through activated ERM proteins. They may act to facilitate local modulation of RhoA activation [21]. Despite the highly conserved amino acid structure of the ERM proteins, they are not entirely redundant molecules [10]. Although ezrin, radixin, and moesin are co-expressed in many cell lines, their cell and tissue type expressions are highly divergent in mammalian organs [4, 5, 11, 32]. Epitopes common to ERM proteins have been reported in hippocampal pyramidal neurons [37], and expression of all three proteins has been reported to peak in late embryonic and early postnatal developing rat brain homogenates, according to Western blot studies [27]. However, specific cellular expression of these proteins in the developing cerebral cortex, especially in humans, has not been reported. We hypothesized that because of the demonstrated interaction of hamartin and ERM proteins and hamartin’s ability to modulate RhoA, perturbation of these interactions and activity modulations might be a cause of cerebral cortical malformations in patients with mutations in the genes that encode hamartin or tuberin (TSC patients). The purpose of this work was to (1) demonstrate the developmental expressions of ezrin and moesin – two of the ERM proteins to which specific antibodies are commercially available – in developing human cerebral cortex, (2) evaluate the expression of ezrin and moesin in cortical tubers from patients with TSC, and (3) propose possible consequences of TSC gene product malfunction or inactivation in the developing brain in relation to ERM protein expression.
Materials and methods Immunohistochemistry Blocks of normal cerebral cortex from human autopsy cases were chosen for a pilot developmental study of the relative expressions of moesin, ezrin, tuberin, and hamartin. The ages of the patients from which blocks (appearing histologically normal) were chosen were: 20, 21, and 31 weeks gestation, 3 months (postnatal age),
8 years, 10 years, and 33 years. Fetal brains under the age of 20 gestational weeks were not available for study. In addition, tubers from five patients diagnosed with TSC and histologically normal neocortex of temporal lobe resected from a 29-year-old female patient with temporal lobe epilepsy/hippocampal sclerosis were selected for staining with the four antisera. Consecutive serial sections were cut at 3–4 µm to compare adjacent sections of tubers stained with all four antibodies. Immunohistochemistry was performed according to commercial product specification and using previously published protocols [19, 20] with appropriate positive and negative controls. The antibodies/antisera used were as follows; monoclonal antibody against ezrin (clone 18, diluted 1:70), monoclonal antibody against moesin (clone 38, diluted 1:150), (both purchased from Transduction Laboratories, Ky.), monoclonal antibody against moesin Ab-1 (clone 38/87, NeoMarkers, Calif., diluted 1:500), monoclonal antibody against vimentin (clone V9, diluted 1:50), polyclonal antibody against cow glial fibrillary acidic protein (GFAP, 1:300) (both purchased from DAKO, Glostrup, Denmark), αp1 tuberin antiserum and C-term hamartin antiserum [19, 20]. Sections were counterstained with hematoxylin. Analysis of colocalization Numbered consecutive serial sections were stained with hamartin, moesin, tuberin, ezrin, and GFAP antisera. To determine the extent of co-localization of moesin, ezrin, tuberin, and hamartin in the same cell profiles, cells were chosen in moesin (clone 38)-stained sections near easily identifiable histological landmarks. The cell profiles were located in adjacent serial sections stained with ezrin, tuberin, hamartin, GFAP, and vimentin with the assistance of a digital camera (Olympus DP10, Tokyo, Japan). Coincidental staining was noted in all four sections.
Results We provide evidence that ezrin is abundantly expressed within radial glia and migrating cells in the intermediate zone in the prenatal human cerebrum. In addition, both ezrin and moesin are abundantly co-expressed with hamartin and tuberin within a population of abnormal cells in TSCassociated cortical tubers. Ezrin and moesin expression in developing human cerebral cortex Ezrin immunoreactivity was moderate to prominent in the periventricular germinal matrix of the cerebrum as early as 20 weeks of gestation (Fig. 1A). Moving outward toward the pial surface, staining was clustered in vertical columns of cells adjacent to the germinal matrix (Fig. 1B), which appeared to represent cells migrating along ezrinpositive processes extending toward the cortical plate. The appearance of these processes was comparable to radial glia in adjacent serial sections stained with an antibody to the intermediate filament, vimentin (data not shown). Scattered cells in the intermediate zone and subplate were prominently ezrin immunoreactive. Similar cells in the intermediate zone were vimentin immunoreactive in adjacent serial sections. Very few scattered cells within the developing cortical plate stained minimally, but rows of fine processes running perpendicular to the pial surface through the developing cortex were ezrin immunoreac-
190
Fig. 1A, B Ezrin immunoreactivity in the cerebrum at 20 weeks gestation. A Ezrin is abundantly expressed by cells in the periventricular germinal matrix (top right of A). Immunoreactivity extends out toward the intermediate zone in processes whose appearances are consistent with radial glia (arrows in B). Note the columnar bundles of cells apparently migrating along the processes. At this age many neurons have already migrated to the cortical plate. However, the phenotype of these migrating cells is indeterminate and they may represent neuronal precursors. Bars A 50 µm, B 10 µm
tive. We did not detect any immunostaining within cortical neurons at any of the ages studied. Moderate cellular staining was noted in the marginal and superficial granular layer in prenatal cerebrum. By 30 weeks of gestation, prominent immunostaining was also noted in what appeared to be migrating unipolar cells with long thin processes (most often leading the soma and oriented toward the pial surface) (Fig. 2A). Vimentin was abundantly expressed within similar appearing cells (Fig. 2B) in an adjacent section of the same specimen. No apparent immunoreactivity was detected in what appeared to be Cajal-Retzius cells throughout the gestational ages studied. Moesin was detected in some cells in the germinal matrix and cells in the intermediate zone at 20 weeks of gestation, but was prominently localized to vascular endothelial cells. Migrating cells in the intermediate zone, at 30 weeks
Fig. 2A, B Ezrin and vimentin expression in the cerebrum at 30 weeks of gestation. A Ezrin prominently labels the cell bodies and processes of cells within the intermediate zone in prenatal cerebral cortex. In most cases, the long thin processes of these cells point toward the pial surface (towards the top of picture), whereas some cells extend processes away from the pial surface (arrowhead). B In a similar field in an adjacent section of the same specimen, vimentin is abundantly expressed within similar-appearing cells (compare A and B). Bars A, B 10 µm
of gestation, exhibited moderate to prominent process staining similar to ezrin and vimentin immunoreactivity in serial sections (Fig. 3A). The prominent vascular endothelial expression is retained throughout the other ages studied up to and including 33 years of age (Fig. 3B). In contrast, from 3 months until adulthood, ezrin immunoreactivity was limited to astrocyte cell bodies and processes in the cortex and subcortical white matter (Fig. 4A). Cortical neurons were not detectably labeled in the prenatal or adult CNS (Fig. 4B) by either moesin antibody. Occasional glial cells, possibly astrocytes and/or oligodendrocytes in the cortex and subcortical white matter, were weakly to moderately moesin immunoreactive throughout the ages studied, but the staining was not as robust as that found with the ezrin antibody. While no apparent reaction was detected in cells of microglial morphology using anti-moesin (clone 38) immunohistochemistry, another moesin antibody (clone 38/87) recognized fine processes of scattered microglia (Fig. 4B), as confirmed by double-label immunohistochemistry with CD68 (data not shown).
191
Fig. 3A, B Moesin expression in the cerebral cortex at 30 weeks of gestation. A Moesin antibody prominently immunolabels rare cells apparently migrating in the intermediate zone. This is similar to the finding observed with ezrin and vimentin immunostaining (see Fig. 2). This cell (arrow) projects a long thin process toward the developing cortical plate. B Moesin, although expressed minimally in the germinal matrix and prominently within some scattered cells in the intermediate zone, is most abundantly expressed in vascular endothelial cells, as in this venule. In addition, weak staining is seen in scattered glial cells (arrowheads). Bars A, B 10 µm
Ezrin and moesin expression in TSC cortical tubers Ezrin is highly expressed within progenitor cells, radial glia, and migrating neuronal and/or glial precursors in the intermediate zone and developing human cerebral cortical plate. Although ezrin expression was limited to what appeared to be, using traditional morphologic criteria, astrocytes in the cortex and subcortical white matter of human cerebrum after the age of 3 months in control cases, ezrin was highly expressed within large abnormal cells in the five cortical tubers (Fig. 5A). The pattern of tuberin immunoreactivity was almost identical and it co-localized with ezrin in many cells of apparent glial morphology, including the same cluster of balloon cells (Fig. 5B). Antibodies to hamartin and moesin labeled cells of similar morphology in the five cortical tubers (Fig. 6). The majority of ezrin- and moesin-immunoreactive cells were
Fig. 4A, B Ezrin and moesin expression in surgically resected temporal neocortex of a 29-year-old female with hippocampal sclerosis. Histologically normal cortex is stained with antibodies to ezrin (A) and moesin (B). A Ezrin immunoreactivity is characterized by punctate staining of astrocytic process and cell bodies (arrows). Neurons do not show detectable immunoreactivity. B Moesin (clone 38/87) is not detectable in cortical neurons. Note the moderate to strong staining of capillary endothelial cells (arrow) and scattered microglia (arrowhead). Bars A, B 10 µm
consistent with large gemistocytic astrocytes or ‘balloon cells’ typical of the type frequently encountered in TSC tubers [36]. However, stained cells were of variable morphology. Large and/or abnormally oriented pyramidalshaped cells with prominent nucleoli – which appeared to be more neuronal in character – were usually not immunoreactive (Fig. 6B). Double-label immunohistochemistry (data not shown) and comparison of adjacent sections stained with ezrin, moesin, or glial fibrillary acidic protein (GFAP) demonstrated co-localization of ezrin and moesin with the glial marker in many cells (Fig. 7). However, conspicuous lack of GFAP expression was also noted in adjacent profiles of other moesin- and ezrin-positive cells in serial sections (Fig. 8). Staining of adjacent tuber sections with hamartin, moesin, tuberin and ezrin allowed us to demonstrate colocalized expression of the TSC gene products with the two ERM proteins in a subpopulation of abnormal neuroglial cells. Hamartin antisera immunolabeled more ab-
192
Fig. 5A, B Ezrin and tuberin immunostaining of adjacent sections of a surgically resected tuber from a 6-month-old child. Micrographs are from similar fields of adjacent serial sections. A Ezrin is prominently expressed in several cells in this region, including a cluster of balloon cells (arrow) to the left of a small vessel cut in crosssection. B Pattern of tuberin immunoreactivity is almost identical (though of weaker intensity) and it co-localizes with ezrin in many cells of apparent glial morphology, including the same cluster of balloon cells (arrow). Notice the conspicuous lack of ezrin immunoreactivity and very weak tuberin immunoreactivity within some cells in this region (arrowheads, both panels). Bars A, B 50 µm
Fig. 6 Adjacent serial sections of a resected tuber from a 6-monthold child immunostained with antibodies to hamartin (A) and moesin (B). Hamartin (A) and moesin (B) immunoreactivity co-localize to several cells within this region including three clustered (arrowheads in both panels) to the left of a small vessel cut in cross-section (asterisk). However, moesin immunoreactivity (arrow in B) is not seen within a large abnormal cell of neuronal morphology that is stained with hamartin antiserum (arrow in A). Bars A, B 50 µm
Discussion normal cells including dysmorphic and cytomegalic neurons and ‘balloon cells’ within the five cortical tubers than did our tuberin antisera. A semiquantitative observation revealed that the number of cells detected by hamartin antisera was on average 250–282/mm2 in a cortical tuber, whereas those detected by tuberin antisera in corresponding areas numbered 218–234/mm2. The pattern of immunolabeling of abnormal neuroglial cells with our tuberin antisera was very similar to moesin and ezrin immunolabeling within all five tubers, whereas hamartin antiserum often prominently labeled cells not immunoreactive for moesin and ezrin. In all the cases in which the abnormal cell profiles in question were found in all four adjacent sections, moesin and ezrin expression were always co-localized with each other and with hamartin and tuberin expression (see Figs. 5, 6, 7).
TSC gene products have been implicated in events that regulate cell adhesion through hamartin’s interaction with ERM proteins and through its activation of RhoA [21]. Mutations in one of the two TSC genes might lead to cerebral cortical malformation within patients with TSC through mechanisms involving ERM proteins and RhoA. However, the current paucity of evidence regarding ERM cellular expression in the developing human cerebrum limits one’s ability to speculate as to their specific functions. In response to this limitation, we determined the distribution of two of the ERM proteins, ezrin and moesin, in the developing normal human cerebrum and within tubers from patients with TSC. We chose these two proteins because ezrin, unlike the other ERM proteins, has been reported to interact with endogenous hamartin [21], antibodies are
193
Fig. 7A–C Tuberin, ezrin, and GFAP expression in adjacent serial sections of a surgically resected tuber from a 14-year-old child. Note the co-localized immunolabeling of tuberin (A) and ezrin (B) in large abnormal cells – some with irregular processes (arrows in both panels). C The same cells are also immunolabeled with an antibody to GFAP, suggesting that they are large abnormal astrocytes (arrows; compare to cells in A and B). Both ezrin and GFAP antibodies label clusters of astrocytes in the top right corner of both panels (B and C) (GFAP glial fibrillary acidic protein). Bars A–C 50 µm
commercially available for these two ERM proteins, and some information on their specific expression in tissues has already been published. We performed an immunohistochemical pilot study to determine the specific developmental expressions of two ERM proteins, ezrin and moesin, in human cerebral cortex. Ezrin and moesin expression have been reported in multiple cell lines and various tissues. Although ezrin immunoreactivity has been reported in sensory and motor neurons, and some CNS tumors [4, 13], the developmental expression of ezrin and moesin in the prenatal human cerebral cortex has not previously been described [4, 5, 11, 32]. Moesin immunoreactivity is abundant in cerebral vascular endothelium during cortical development extend-
Fig. 8A–C Moesin, ezrin, and GFAP expression in adjacent sections of a resected tuber from a 3-year-old child. Note the immunolabeling of moesin (A) and ezrin (B) in a large ‘balloon cell’ (arrows in all three panels) in addition to other scattered cells. Unlike in Figs. 5 and 7, many of these cells, including the large balloon cell, do not express detectable GFAP (C) and thus appear clear against the homogeneous chromogen deposition, suggesting that moesin and ezrin are not expressed in all GFAP-positive cells. Bars A–C 50 µm
ing from 20 weeks of gestation until adulthood; although this pattern of moesin expression is consistent with previous reports of its expression in endothelial cells and glia, immunohistochemical detection of diffuse cytoplasmic moesin immunoreactivity in human CNS neurons has also been reported [32]. As stated above, we did not detect neuronal expression in either immunostaining, corroborated using two different moesin antibodies. Furthermore, a population of scattered immunoreactive microglia was definitely detected using one of the moesin antibodies (clone 38/87), a finding not previously reported. Since moesin has been reported to be expressed in macrophages [3], this observation is suggestive of specific expression of moesin among ERM proteins in brain microglia, possibly reflecting the similarities between brain microglia and macrophages. Moesin immunoreactivity was not detected in cortical neurons but was rarely seen in astrocytes and/or oligo-
194
dendroglia in the normal adult cerebral cortex, and it was only weakly expressed in the periventricular zone and within few cells in the intermediate zone in prenatal cerebrum. Ezrin expression was abundantly detected within cells of the germinal matrix, in radial glia and migrating cells in the intermediate zone. These distributions suggest, at least indirectly, that ezrin and moesin expression are important in neuroglial precursor cell proliferation, formation of radial glia, and migration of neuronal and/or glial cells to the cortical mantle. The localization of ezrin and moesin within populations of cells that also express the intermediate filament, vimentin, suggests that these two proteins are primarily expressed within glial cells in developing cerebral cortex. Although we were unable to consistently detect tuberin and hamartin within radial glia and astrocytes in prenatal or postnatal human brain, tuberin and hamartin are reported to be expressed in mouse and rat primary astrocytic cultures [6, 16]. It has furthermore been reported that both tuberin and hamartin expression in astrocytes declines until they reach their growth arrest in vitro [16]. Thus, there may be distinct developmental ‘temporal windows’ in which ezrin, moesin, tuberin, and hamartin co-localize and function together to affect proliferation and/or migration of neuronal precursor cells. However, the specifics of the times and locations of ERM and TSC gene product expression need to be evaluated in greater detail. The cerebral cortex of the youngest age we have studied originated from a human fetus of 20 weeks gestation, so it is possible that such collaborative functions occur earlier in association with the initial migration of neuroblasts and/or formation of radial glia. TSC-associated cortical tubers have been thought to arise because of abnormal neuronal precursor migration or development or both [14, 24]. Thus, we hypothesized that malfunction of a TSC gene product, caused by a gene mutation, might cause abnormal cortical development through abnormal modulation of ERM protein function. To determine the expression of ERM proteins as compared to hamartin and tuberin in abnormal neuroglial cells, we performed immunohistochemical staining of five surgically resected TSC cortical tubers with antibodies to ezrin, moesin, tuberin, and hamartin on adjacent serial sections. Both ezrin and moesin were abundantly expressed within some, but not all, abnormal neuroglial cells within five TSC-associated cortical tubers. Although this is a qualitative observation, ezrin and moesin immunoreactivity within these hamartomas appears to be much greater than immunoreactivity of these proteins in normal adult cerebral cortex. This suggests that both proteins are up-regulated within cortical tubers. The morphology of these immunoreactive cells furthermore suggests that a majority of them are large gemistocytic astrocytes or ‘balloon cells’ typical of the type frequently encountered in TSC tubers [36]. The observation of abundant expression of ezrin in abnormal neuroglial cells in cortical tubers is, by itself, somewhat difficult to interpret given the fact that ezrin is abundantly expressed in astrocytes in normal human cerebral cortex (data presented here; [13]). However, the co-local-
ization of moesin immunoreactivity in the same cells is most surprising and provides evidence that the expression of another ERM protein, whose detectable expression is primarily limited to endothelial cells and microglia in normal cerebral cortex, is up-regulated in abnormal cells within cortical tubers. Although many of the cells immunolabeled with ezrin and moesin antisera may be of glial origin or phenotype, the presence of ezrin- and moesin-immunoreactive cells lacking GFAP expression suggests that the expression of ezrin and moesin is not limited to cells expressing this glial marker. However, this observation is not entirely informative because of the mixed neuroglial nature and morphological heterogeneity of cells – including ‘undifferentiated’ or ‘dedifferentiated’ forms – within TSC-associated cortical tubers [36]. Increased ezrin expression has been reported to be associated with malignancy and metastatic potential in endocrine and glial tumors, suggesting that overexpression of the protein may stimulate the transformation and/or motile behavior of cancerous cells in multiple tissues [1, 13]. Although we have demonstrated abundant levels of ezrin and moesin in abnormal and dysmorphic cells in cortical tubers, it is unlikely that these apparent qualitative increases are associated with malignancy or invasiveness, as hamartomas themselves are histologically benign and show restricted growth potential. The cell biology of malignant transformation is complex, and increased ezrin expression associated with malignancy in astrocytomas may, according to previous suggestions, only reflect the increasingly altered shape of cancer cells in association with their malignant transformation [13], i.e., up-regulation of ERM proteins within TSC cortical tubers may be a consequence of the abnormal morphology of neuroglial cells in these hamartomas. Another possible explanation for the apparent up-regulation of the ERM proteins within TSC cortical tubers is that they are up-regulated to compensate for TSC gene product dysfunction within a subpopulation of abnormal cells. To determine qualitatively if ezrin and moesin expressions are correlated with hamartin and tuberin expression, we chose moesin-positive cell profiles near obvious histological landmarks (usually blood vessels) from within all five immunostained tubers. Subsequently, we located this region of interest in adjacent serial sections stained with antisera to tuberin, hamartin, and ezrin. These comparisons of moesin-immunoreactive, abnormal cell profiles to adjacent profiles identifiable in serial sections suggest that moesin expression is always co-localized with ezrin, tuberin, and hamartin immunoreactivity – but not vice versa, as tuberin and hamartin appear to be more widely expressed. This method of determining co-localization is effective, but preliminary. A more unbiased cell profile selection method may be warranted. However, we believe that our observation is strongly suggestive evidence that ezrin and moesin and TSC gene products (both tuberin and hamartin) are co-expressed within a subpopulation of abnormal cells within TSC-associated tubers. Specific localization to vimentin-positive radial glia and migrating cells in the intermediate zone in the prenatal cerebrum
195
may suggest that ERM proteins function in glia within a period defined by radial glial-assisted cortical migration and radial glial transformation into astrocytes. Thus, this subpopulation of abnormal cells expressing ezrin, moesin, tuberin, and hamartin may represent abnormally developed astrocytes resulting from abnormal TSC gene product regulation of ERM proteins during brain development. An important question that arises from this work and a previous report [21] is: how might the purported interactions of tuberin, hamartin, ERM proteins, and RhoA be involved in normal cortical development and abnormal TSC-associated development leading to formation of tubers? A plausible hypothesis is that ERM proteins and TSC gene products interact to regulate cell shape and polarity, cell adhesion, cell motility, and cell fate simultaneously or in distinct temporal ‘windows’ during the development and migration of neuronal and glial precursor cells in the cerebral cortex. The interaction of hamartin with an N-terminal domain common to ERM proteins and its ability to activate RhoA GTPase to affect cell adhesion [21] place hamartin and tuberin in the middle of complex protein cascades linking extracellular proteins to localized cortical cytoskeletal-associated intracellular proteins and ultimately nuclear transcription factors [21, 23, 30]. Definitive evidence for the functions of ERM and Rho in cortical development is not available. However, several reports provide data that theoretically implicate these proteins in neuronal and glial migration and morphological determinance. Antisense inhibition of both radixin and moesin expression have been reported to alter growth cone morphology, decrease cone motility, and decrease process formation and length in cultured primary hippocampal neurons [12]. Radixin expression and its subcellular localization have been shown to change in growth cones of cultured chick sympathetic neurons in response to several extracellular cues – including withdrawal of growth factor or exposure to brain homogenate factors and tactile stimuli. [15]. Constitutive activation of RhoA inhibits astrocyte stellation and process length, and toxic inactivation of RhoA disrupts lysophosphatidic acid-triggered glial cytokinesis by inhibiting the action of rho-specific kinases on GFAP [9, 29, 38]. Thus, ERM proteins and RhoA might function in the regulation of neuronal and/or glial cytokinesis, cell polarity, process growth, and cell migration during cerebral cortical development. Because hamartin binds to ERM proteins and can modulate RhoA activation, tuberin and hamartin may also be important regulators of these cellular events. We have demonstrated that ezrin and moesin are expressed by cells in the germinal matrix and in apparent migrating cells in the intermediate zone after 20 weeks of gestation and prior to birth. This specific expression correlates with the expression of RhoA GTPase in developing rat cerebral cortex. Immunohistochemical detection of RhoA in the rat cerebrum has shown that its expression is found in proliferating and migrating cells but not in cells already positioned in the cortical plate [26]. Dysfunction of tuberin or hamartin in response to gene mutation may possibly interrupt communication between
RhoA and cytoskeletal-associated ERM proteins to cause abnormal migration and morphology of neuroglial precursors. Such a possibility may be supported by the report that neuronal heterotopia results from mutations in filamin, another F-actin binding protein whose activity is regulated by RhoA phosphorylation [22]. We noted prominent ezrin immunoreactivity in radial glial fibers in normal human cerebrum at 20 weeks of gestation. Loss of radial glia in the brain of a 20-week-gestational human with TSC has been reported [28]. Thus, disruption of TSC product/ERM interaction and abnormal modulation of RhoA activation within a population of precursor cells could result in focal migrational abnormalities through the loss of radial gliaassisted migration. In conclusion, we have demonstrated the expression of two of the ERM proteins, ezrin and moesin, in the developing human cerebral cortex. The expression of these two proteins – primarily ezrin – suggests that they are developmentally regulated and abundantly expressed in proliferative and/or migrating cells during cerebral cortical development. In TSC-associated cortical tubers, both proteins appeared to be up-regulated and are co-localized within a population of abnormal neuroglial cells. The expression of these proteins and their co-localization with tuberin and hamartin in these cells may suggest a compensatory up-regulation in response to TSC gene mutation. Acknowledgements M.W.J. is supported by a USPHS/NIH Molecular and Cellular Neurobiology Training Grant. H.M. is supported by a grant from the Ministry of Education, Culture, Sports, Science, and Technology of Japan as a visiting research scholar at UCLA Medical Center. The authors wish to thank Ms. Justine Garakian for invaluable technical assistance.
References 1. Akisawa N, Nishimori I, Iwamura T, Onishi S, Hollingsworth MA (1999) High levels of ezrin expressed by human pancreatic adenocarcinoma cell lines with high metastatic potential. Biochem Biophys Res Commun 258:395–400 2. Algrain M., Turunen O, Vaheri A, Louvard D, Arpin M (1993) Ezrin contains cytoskeleton and membrane binding domains accounting for its proposed role as a membrane-cytoskeletal linker. J Cell Biol 120:129–239 3. Amieva MR, Furthmayr H (1995) Subcellular localization of moesin in dynamic filopodia, retraction fibers, and other structures involved in substrate exploration, attachment, and cellcell contacts. Exp Cell Res 219:180–196 4. Bohling T, Turunen O, Jaaskelainen J, Carpen O, Sainio M, Wahlstrom T, Vaheri A, Haltia M (1996) Ezrin expression in stromal cells of capillary hemangioblastoma. An immunohistochemical survey of brain tumors. Am J Pathol 143:367–373 5. Bretscher A, Reczek D, Berryman M (1997) Ezrin: a protein requiring conformational activition to link microfilaments to the plasma membrane in the assembly of cell surface structures. J Cell Sci 110:3011–3018 6. Catania MG, Johnson MW, Liau LM, Kremen TJ, deVellis JS, Vinters HV (2001) Hamartin expression and interaction with tuberin in tumor cell lines and primary cultures. J Neurosci Res 63:276–283 7. Catania MG, Mischel PS, Vinters HV (2001) Hamartin and tuberin interaction with the G2/M cyclin-dependent kinase CDK1 and its regulatory cyclins A and B. J Neuropathol Exp Neurol 60:711–723
196 8. Cheadle JP, Reeve MP, Sampson JR, Kwiatkowski DJ (2000) Molecular genetic advances in tuberous sclerosis. Hum Genet 107:97–114 9. Daniel K, Gomperts SN, Hardy S, Marina K, Bishop JM (1999) Ras family GTPases control growth of astrocyte processes. Mol Biol Cell 10:1665–1683 10. Doi Y, Itoh M, Yonemura S, Ishihara S, Takano H, Noda T, Tsukita S (1999) Normal development of mice and unimpaired cell adhesion/cell motility/actin-based cytoskeleton without compensatory up-regulation of ezrin or radixin in moesin gene knockout. J Biol Chem 247:2315–2321 11. Everett AW, Nichol KA (1990) Ezrin immunoreactivity in neuron subpopulations: cellular distribution in relation to cytoskeletal proteins in sensory neurons. J Histochem Cytochem 38: 1137–1144 12. Gabriela P, Kunda P, Quiroga S, Kosik K, Caceres A (1998) Suppression of radixin and moesin alters growth cone morphology, motility, and process formation in primary cultured neurons. J Cell Biol 143:443–455 13. Geiger KD, Stoldt P, Schlote W, Derouiche A (2000) Ezrin immunoreactivity is associated with increasing malignancy of astrocytic tumors but is absent in oligodendroglioms. Am J Pathol 157:1785–1793 14. Goldman SA, Luskin MB (1998) Strategies utilized by migrating neurons of the postnatal vertebrate forebrain. Trends Neurosci 21:107–114 15. Gonzalez-Agosti C, Solomon F (1996) Response of radixin to perturbations of growth cone morphology and motility in chich sympathetic neurons in vitro. Cell Motil Cytoskeleton 34:122– 136 16. Gutmann DH, Zhang Y, Hasbani MJ, Goldberg MP, Plank TL, Petri Henske E (2000) Expression of the tuberous sclerosis complex gene products, hamartin and tuberin, in central nervous system tissues. Acta Neuropathol 99:223–230 17. Hengstschlager M, Rodman DM, Miloloza A, HengstschlagerOttnad E, Rosner M, Kubista M (2001) Tuberous sclerosis gene products in proliferation control. Mutat Res 488:233–239 18. Hiscox S, Jiang WG (1999) Ezrin regulates cell-cell and cellmatrix adhesion, a possible role with E cadherin/B-catenin. J Cell Sci 112:3081–3090 19. Johnson MW, Emelin JK, Park SH, Vinters HV (1999) Co-localization of TSC1 and TSC2 gene products in tubers of patients with tuberous sclerosis. Brain Pathol 9:45–54 20. Johnson MW, Kerfoot C, Bushnell T, Li M, Vinters HV (2001) Hamartin and tuberin expression in human tissues. Mod Pathol 14:202–210 21. Lamb RF, Roy C, Diefenbach TJ, Vinters HV, Johnson MW, Jay DG, Hall A (2000) The TSC1 tumour suppressor hamartin regulates cell adhesion through ERM proteins and the GTPase Rho. Nat Cell Biol 2:281–287 22. Lambert de Rouvroit C, Goffinet AM (2001) Neuronal migration. Mech Dev 105:47–56 23. Louvet-Vallée S (2000) ERM proteins: from cellular architecture to cell signaling. Biol Cell 92:305–316
24. Luskin MB (1994) Neuronal cell lineage in the vertebrate central nervous system. FASEB J 8:722–730 25. McKay DJG, Esch F, Furthmayr H, Hall A (1997) Rho- and rac-dependent assembly of focal adhesion complexes and actin filaments in permeabilized fibroblasts: an essential role for ezrin/radixin/moesin proteins. J Cell Biol 138:927–938 26. Olenik C, Aktories K, Meyer DK (1999) Differential expression of the small GTP-binding proteins RhoA, RhoB, Cdc42u and Cdc42b in developing rat neocortex. Brain Res Mol Brain Res 18:9–17 27. Paglini G, Kunda P, Quiroga S, Kosik K, Caceres A (1998) Suppression of radixin and moesin alters growth cone morphology, motility, and process formation in primary cultured neurons. J Cell Biol 143:443–455 28. Park SH, Pepkowitz SH, Kerfoot C, De Rosa MJ, Poukens V, Wienecke R, DeClue JE, Vinters HV (1997) Tuberous sclerosis in a 20-week gestation fetus: an immunohistochemical study. Acta Neuropathol 94:180–186 29. Ramakers GJ, Moolenaar WH (1998) Regulation of astrocyte morphology by RhoA and lysophosphatidic acid. Exp Cell Res 245:252–262 30. Salvador A, Lacal JC (2001) Rho signals to cell growth and apoptosis. Cancer Lett 165:1–10 31. Sampson JR, Harris PC (1994) The molecular genetics of tuberous sclerosis. Hum Mol Genet 3:1477–1480 32. Stemmer-Rachamimov AO, Gonzalez-Agosti C, Xu L, Burwick JA, Beauchamp R, Pinney D, Louis DN, Ramesh V (1997) Expression of NF2-encoded merlin and related ERM family proteins in the human central nervous system. J Neuropathol Exp Neurol 56:735–742 33. Takeuchi K, Kawashima A, Nagafuchi A, Tsukita S (1994) Structural diversity of band 4.1 superfamily members. J Cell Sci 107:1921–1928 34. Takahashi K, Sasaki T, Mammoto A, Takaishi K, Kameyama T, Tsukita S, Takai Y (1997) Direct interaction of the Rho GDP dissociation inhibitor with ezrin/radixin/moesin initiates the activation of the Rho small G protein. J Biol Chem 272: 23371–23375 35. Takahashi K, Sasaki T, Mammoto A, Hotta I, Takaishi K, Imamura H, Nakano K, Kodama A, Takai Y (1998) Interaction of radixin with Rho small G protein GDP/GTP exchange protein Dbl. Oncogene 16:3279–3284 36. Vinters HV, Park SH, Johnson MW, Mischel PS, Catania M, Kerfoot C (1999) Cortical dysplasia, genetic abnormalities and neurocutaneous syndromes. Dev Neurosci 21:248–-259 37. Winckler B, Gonzalez-Agosti Ch, Magendantz M, Solomon F (1994) Analysis of a cortical cytoskeleton structure: a role for ezrin-radixin-moesin (ERM proteins) in the marginal band of chicken erythrocytes. J Cell Sci 107:2523–2534 38. Yasui Y, Amano M, Nagata K, Inagaki N, Nakamura H, Saya H, Kaibuchi K, Inagaki M (1998) Roles of Rho-associated kinase in cytokinesis; mutations in Rho-associated kinase phosphorylation sites impair cytokinetic segregation of glial filaments. J Cell Biol 143:1249–1258
