J Neurocytol (2005) 34:435–446 DOI 10.1007/s11068-006-8729-x

Immunogold study of effects of prenatal exposure to lipopolysaccharide and/or valproic acid on the rat blood-brain barrier vessels A. W. Vorbrodt · D. H. Dobrogowska · P. B. Kozlowski · A. Rabe · M. Tarnawski∗ · M. H. Lee

Received: 25 February 2005 / Revised: 27 July 2005 / Accepted: 15 August 2005 C Springer Science + Business Media, LLC 2006 

Abstract The involvement of blood microvessels, representing the anatomic site of the blood-brain barrier (BBB), in brain damage induced by prenatal exposure to lipopolysaccharide (LPS) and/or valproic acid (VPA) was studied in four-week-old rats. The immunogold procedure was applied for localization at the ultrastructural level of endogenous albumin and glucose transporter (GLUT-1) in three brain regions: cerebral cortex, cerebellum and hippocampus. Four groups of rats were used: (1) untreated control, (2) prenatally VPA-treated, (3) prenatally LPS-treated, and (4) prenatally LPS- and VPA-treated. The functional state of the BBB was evaluated as follows: (a) by its tightness, i.e., permeability to blood-borne albumin, and (b) by the expression of GLUT-1 in the endothelial cells (ECs). Using morphometry, the labelling density for GLUT-1 was recorded over luminal and abluminal plasma membranes of the ECs, also providing information on their functional polarity. No extensive increase of vascular permeability and/or any considerable dysfunction of the BBB in experimental groups nos. 2 and 3 were observed, although in solitary vascular profiles, increased endocytosis or even transcytosis of albumin by ECs was noted. In experimental group no. 4, some vascular profiles showed scanty leakage (microleakage), manifested by the presence of immunosignals for albumin in the perivascular area. Although A. W. Vorbrodt () . D. H. Dobrogowska · P. B. Kozlowski · A. Rabe · M. Tarnawski · M. H. Lee New York State Office of Mental Retardation and Developmental Disabilities, Institute for Basic Research in Developmental Disabilities, 1050 Forest Hill Road, Staten Island, New York 10314, USA e-mail: [email protected] M. Tarnawski Present address: Lindsley F. Kimball Research Institute of the New York Blood Center, 310 East 67th Street, New York, N.Y. 10021

∗

some fluctuations in the expression of GLUT-1 occurred in all experimental groups, especially in group no. 3, a most pronounced and significant diminution of the labelling density, in all three regions of the brain, was observed in group no. 4. This finding suggests the synergistic action of prenatally applied LPS and VPA that affects specific transport functions of glucose in the microvascular endothelium. The diminished or disturbed supply of glucose to selected brain regions can be one of the factors leading to previously observed behavioral disturbances in similarly treated rats.

Introduction The mechanisms associated with or leading to developmental disabilities (DD) manifested by mental retardation still have not been fully elucidated. Among the factors responsible for developmental brain disorders, besides genetic or genetically determined (e.g., in Down syndrome, fragile X syndrome), some other agents can create unfavorable prenatal events (e.g., maternal infections, drug and substance abuse, malnutrition, hypoxia, etc.) leading to DD. The involvement of multiple factors is likely, even the precise outcome of genetic flaws can be influenced by other factors. Acquiring knowledge of these factors and their interactions causing aberrant development of the brain is slowed by the lack of valid and reliable animal models. This fact is due not only to the complexity of symptoms and multiple gene involvement (McIlvane & Cataldo, 1996), but also to emphasis on a single causal factor be it genetic or environmental. For example, all previous attempts to induce autismlike brain and/or behavioral changes in animals exposed to a single environmental agent, such as influenza or Borna disease virus (Fatemi et al., 2002; Hornig & Lipkin, 2001; Hornig et al., 1999; Pletnikov et al., 2002) or teratogens, Springer

436

e.g., thalidomide (Vorhees et al., 2001), valproic acid (Ingram et al., 2000; Schneider et al., 2001b), and ethanol (Schneider et al., 2000a), appeared to be unsuccessful. To date, none of them have produced anatomical and behavioral changes similar to those associated with autism spectrum disorder (ASD) in humans. Since it is being increasingly recognized that multiple factors are involved in developmental brain disorders, we are focusing our research studies on the consequences of the action of two toxic substances to which pregnant women may be exposed: bacterial lipopolysaccharides (LPS), simulating maternal infection, and valproic acid (VPA), an antiepileptic drug with teratogenic properties. In our study, these two agents that may be encountered in real life situations are given separately or together at specific developmental periods, to induce autism-like neurobehavioral changes. We chose the specific developmental periods because at each of them, the central nervous system (CNS) may be vulnerable to the particular agent: LPS will be given at E10 as an infection inducing factor (pro-inflammatory response) during early embryonic development (the period of neural folds). The neurotoxic agent (VPA) will be given at E12.5 when neural tube closure (pre-cortical development) takes place in the rat. In our preliminary experiments on rat model, both separate as well as combined application of VPA and LPS produced behavioral effects that demonstrate the substances synergistic action (El-Khodor et al., 2003; Tarnawski et al., 2004). These observations suggest that such a two-hit concept may be of great value in the search for the cause of the aberrant embryonic and fetal development of the brain underlying behavioral and mental disorders. It appears that multiple mechanisms may be engaged in the production of neurobehavioral changes in the progeny of rats treated with the noxious substances mentioned above. One possible mechanism may involve the microvascular network of the brain representing the anatomical site of the bloodbrain barrier (BBB). The function of the BBB is twofold: it restricts the movement of micro- and macro-molecular soluble substances (solutes) from blood to brain parenchyma, and concomitantly, it secures (through a specialized system of carriers) a continuous supply of important metabolites necessary for optimal functioning of the brain. The structuralfunctional correlations related to both of these functions of the BBB can be studied by using immunocytochemistry applied to electron microscopy. The first feature, i.e., the tightness of the endothelial lining of blood microvessels (mainly capillaries) can be evaluated morphologically by the localization of endogenous blood serum albumin. In normal conditions, this protein does not cross the wall of microvessels, and consequently, it is perfectly suitable for studies of the integrity of the BBB. In addition, we can avoid any unwanted and unpredictable effects Springer

J Neurocytol (2005) 34:435–446

that can be induced in the microvascular endothelium by foreign proteins (HRP, WGA-HRP) or other substances (e.g., lanthanum) used as tracers. The problem of the usefulness of endogenous albumin in experimental studies on the tightness of the BBB was discussed in details in our previous publications (Vorbrodt et al., 1994; Vorbrodt, 1995). The applied immunogold procedure appeared to be sensitive for demonstrating even small leakage (microleakage) of the BBB, manifested by the appearance of immunosignals for albumin in the perivascular space (Vorbrodt et al., 1997). The distribution of glucose transporter (GLUT-1) is considered another sensitive indicator of normal or abnormal BBB function (Dermietzel et al., 1992; Drewes, 1998). This parameter not only gives information about the ability of the microvascular network to supply glucose, the major source of energy, for the brain cells but also reveals the functional polarity of the microvascular endothelium. This polarity is manifested by a higher (2.5-3 fold) concentration of GLUT-1 molecules in abluminal than luminal plasma membranes of the ECs (Farrell & Pardridge, 1991; Dobrogowska & Vorbrodt, 1999). The changed polarity of these cells is considered a sensitive indicator of functional perturbation of the BBB that can also affect the function of neighboring neurons (Vorbrodt, 1993; Vorbrodt et al., 1999). This notion is based on the critical role of glucose as a main source of energy for all brain cells, especially neurons and their neurites and synapses (Cornford et al., 1993; Lippoldt et al., 2000; Vorbrodt et al., 2001b) The main objective of the present work is to investigate whether the microvascular network is associated with the process of perinatal brain damage leading to neurobehavioral deficits. The choice of three brain regions (cerebral cortex, cerebellum and hippocampus) and the application of immunogold procedures were based in part on our previous studies of the BBB vasculature in mice prenatally exposed to a single teratogenic dose of ethanol (Vorbrodt et al., 2001a).

Materials and methods Animals Fifteen female Long-Evans rats were mated overnight with male breeders, and vaginal plugs were checked each morning. The day the plug was found was considered day 1 (E1) of pregnancy. The dams received an intraperitoneal injection with 0.1 mg/kg of bacterial LPS (Sigma) at E10, a subcutaneous injection of 350 mg/kg of VPA (Sigma) on E12.5, or both injections, i.e., LPS at E10 followed by VPA at E12.5. Age-matched rats that received no injections served as controls. LPS was dissolved in saline, whereas VPA was dissolved in 0.1M Tris-HCl buffer, pH 7.4. The day of offspring birth was defined as P1.

J Neurocytol (2005) 34:435–446

At P21, i.e., at 3 weeks of age, the animals were tested on EPM (elevated plus maze 5-min session) for behavioral examination, the results of which were presented separately (El-Khodor et al., 2003). At 4 weeks of age (P28), three or four animals from each group of dams were used for our study, giving four groups altogether. Group no.1: Control animals (three rats—offspring of untreated dams); group no. 2: four rats—offspring of VPA-treated (at E12.5) dams; group no 3: three rats—offspring of LPS-treated (at E10) dams; group no 4: four rats—offspring of dams that received both LPS (at E10) and VPA (at E12.5). Each animal was anesthetized with Nembutal (sodium pentobarbital, 50 mg/kg) and then perfused through the left ventricle of the heart with 0.9% saline (2–3 min to wash out the blood), followed by freshly prepared fixative containing 3% paraformaldehyde, 0.05% glutaraldehyde, 0.15 M sucrose and 0.002M calcium chloride in 0.08 M cacodylate buffer, pH 7.4. After 15 min of perfusion, the brain was removed, and the cerebral cortex (gray matter of the parietal cortex), cerebellum (lobules 5–7) and hippocampus (dorsal area inclusive of the Ammon horn and dentate gyrus) were dissected, cut into small blocks (1–2 mm) and immersion fixed for up to 3 h in ice-cold fixative. For comparative purposes, samples of tissues with non-barrier vasculature, such as heart and skeletal muscles, were taken from the control rat. After fixation, all samples were immersed for 3 h in 0.1M solution of glycine in the same buffer to quench free aldehyde groups and were washed overnight in ice-cold 0.05M cacodylate buffer, pH 7.4, containing 0.15 M sucrose. This was followed by dehydration of small tissue blocks in ethanol with concomitant lowering of temperature to −35˚C and, finally, embedding in hydrophilic resin Lowicryl K4M. After polymerization under UV lamp in low-temperature embedding TTP 010 apparatus (Balzers Union, Liechtenstein), the samples were cut with a diamond knife on an ultramicrotome (Sorvall MT-5000, DuPont). Immunocytochemical procedure

437

sections were examined in a Hitachi 7000 electron microscope. Controls for specificity of the immunostaining consisted of sections incubated with protein A-gold solution only or the anti-albumin antiserum preabsorbed with 0.1% albumin. Glucose transporter (GLUT-1) Ultrathin sections attached to nickel grids were incubated in a refrigerator overnight on a drop of solution of rabbit antiGLUT-1 antiserum (Chemicon, Inc., Temecula, CA, USA) diluted 1:200 with PBS containing 0.5% ovalbumin. After washing with PBS, the sections were exposed for 1 h at room temperature to goat-anti-rabbit antiserum complexed with colloidal gold particles of 15 nm (GAR-G15), diluted 1:30 (AuroProbe EM, Amersham, UK). Controls for the specificity of the immunostaining consisted of sections incubated with normal rabbit preimmune serum or with secondary antibodies only. The rest of the procedure was the same as that described above for albumin detection. Label quantification Randomly sampled capillary profiles were examined in representative sections of brain regions chosen for study in all animals in each experimental group. Density of labelling for GLUT-1 associated with luminal and abluminal plasma membranes of the microvascular ECs was expressed as the number of colloidal gold particles (GPs) per μm of the EC plasmalemma length. Morphometry was performed with Sigma Scan software packages on a PC 486 DX equipped with Jandel JS-2 tablet. General linear model (GLM) and post-hoc tests (Bonferroni procedure and Dunnet T3) were performed for statistical evaluation of the results (SPSS program). Significance levels of differences between the control and experimental groups for each of three brain regions were accepted for p < 0.05.

Albumin Results Ultrathin sections attached to Formvar-carbon-coated nickel grids were incubated overnight (16 h) in a refrigerator on a drop of a solution of rabbit anti-rat albumin antiserum (Cappel) diluted 1:200 with phosphate-buffered saline (PBS) containing 0.5% ovalbumin. After washing with PBS, the sections were exposed to protein A conjugated with colloidal gold of 15-nm particle diameter (AuroProbe EM, Amersham, UK) diluted with PBS (1:25) for 1 h at room temperature. It was followed by washings with PBS and glass distilled water, staining with 4% uranyl acetate (3 min), again washing with distilled water and staining with Reynold’s lead citrate (1 min). After final washing, the

Albumin The heart and skeletal muscle capillaries were examined as a positive control for evaluation of the method we used; the results obtained have reference to those presented previously (Vorbrodt & Dobrogowska, 1994; Vorbrodt et al., 1994). During perfusion fixation, the blood serum albumin was washed out from the vascular lumen, which in effect remained unlabelled. In contrast, the albumin molecules taken up and transported across the EC body by cytoplasmic vesicles and finally extravasated into the perivascular space were Springer

438

J Neurocytol (2005) 34:435–446

Fig. 1 Immunogold detection of endogenous albumin in the capillary profile of control rat (group no. 1) skeletal muscle. After perfusion fixation, the blood serum containing albumin is washed out from the vascular lumen (L), which in effect remains unlabeled. In contrast, the distribution of immunosignals for albumin, represented by colloidal gold particles, shows the vesicular transport of this protein in numerous plasmalemmal vesicles (arrows) located in the EC cytoplasm. The extravasated (transcytosed) albumin remains in the perivascular space (arrowheads). The following abbreviations are used in this and the other figures: B, basement membrane; E, endothelial cell; L, vessel lumen; M, muscle fibers; N, cell nucleus; P, pericyte. Scale bars = 0.5 μm. Magnification: X 45,000 Fig. 2

of control rat showing only one immunosignal for albumin located inside the EC cytoplasm (arrow) ×30,000 Fig. 3 In this capillary from the hippocampus of a rat from group no. 2 (VPA-treated), a few immunosignals for albumin are present inside the EC cytoplasm (arrows) and also in the subendothelial basement membrane and in perivascular neuropil (arrowheads). ×30,000 Fig. 4 A portion of the capillary wall in the cerebral cortex of the rat from group no. 3 (LPS-treated) showing increased endocytosis of the albumin manifested by the presence of several immunosignals inside the vacuole, presumably endosome (arrow), and in the EC cytoplsm (arrowheads). X 34,000

A portion of cross-sectioned capillary from the cerebral cortex

sufficiently sticky to become firmly attached to the cellular and tissue components (Fig. 1). This finding indicates that after fixation, even the complex procedure of dehydration, embedding, cutting, incubation and multiple washing do not remove this protein from the ultrathin section of the tissue samples. In BBB-type microvessels, irrespective of their localization in different brain regions (e.g., in cerebral cortex, cerebellum or hippocampus) of untreated, control rats (group no.1), the immunosignals for albumin were scanty. Occasionally, single colloidal gold particles were present inside the EC cytoplasm (Fig. 2) or were attached to the luminal plasmalemma of these cells. In a few sections, solitary gold particles were present in the cytoplasm of pericytes or inside the perivascular astrocytic endfoot.

Springer

In brain samples of rats from experimental group no. 2 (VPA), the vast majority of examined capillary profiles revealed scanty labelling, similar to that observed in the control group. Labelling was restricted to solitary immunosignals for albumin, represented by colloidal gold particles located on the luminal surface or inside the EC cytoplasm. Such a pattern of albumin localization was observed in all brain areas examined, i.e., in the cerebral cortex, cerebellum and hippocampus. However, in a few microvascular profiles, signs of increased permeability to albumin were observed. It was manifested not only by the presence of solitary immunosignals for albumin on the luminal surface and inside the EC cytoplasm, but also by the presence of several gold particles in the cytoplasmic vacuoles resembling

J Neurocytol (2005) 34:435–446

Fig. 5 A portion of the capillary wall in the cerebellum of the rat from group no. 3 (LPS-treated), showing several immunosignals for albumin in subendothelial basement membrane (arrows), and in perivascular neuropil, i.e., in the astrocytic endfoot cytoplasm (arrowheads). ×30,000 Fig. 6 Demonstration of microleakage in the capillary profile in the cerebral cortex of the rat from the group no. 4 (LPS + VPA) showing endocytosed and/or transcytosed albumin in the EC cytoplasm (arrows), and extravasated albumin in the subendothelial and perivascular area (arrowheads). ×30,000 Fig. 7

Another, apparently “leaking” segment of the rat cerebellar

endosomes, in subendothelial basement membrane and in the perivascular neuropil. These few (2–5%) capillary segments showing slightly increased permeability (focal microleakage) to albumin were found in a few sections of the cerebral cortex, cerebellum and hippocampus (Fig. 3). In brain samples of rats from group no. 3 (LPS), the majority of the capillary profiles were similar to those observed in control rats and in group no. 2, i.e., they did not reveal any symptoms of structural abnormality or increased permeability. There were, however, microvascular profiles showing increased endocytosis of albumin manifested by the appearance of small or larger, frequently elongated (oval) vacuoles in the EC cytoplasm, resembling endosomes. These vacuoles, containing single or few immunosignals for al-

439

capillary from group no. 4 (as above) showing increased endocytosis and transcytosis of the albumin. The endocytic vacuole (endosome) containing immunosignals for albumin represented by several gold particles (arrow) is located in a relatively thick portion of the EC body. A few immunosignals for extravasated albumin are also present in the subendothelial space (arrowheads). ×30,000 Fig. 8 In this capillary profile from the hippocampus of the rat from group no. 4 (LPS + VPA), a single immunosignal for albumin is present inside the EC cytoplasm near to the interendothelial junction (arrow), whereas extravasated albumin is present in the basement membrane and in the adjacent perivascular neuropil (arrowheads). ×30,000

bumin, appeared in a few capillary profiles in the cerebral cortex (Fig. 4), cerebellum and hippocampus. A few capillaries showed leakage of albumin manifested by the presence of several gold particles in the basement membrane, in the cytoplasm of pericytes and in the perivascular area of neuropil. The leaking microvascular segments appeared in all brain areas under study, although they were most conspicuous in the cerebellum (Fig. 5). The immunogold reaction for albumin in the brain samples of rats from group no. 4 (offspring of LPS- and VPA-treated dams) was in the vast majority of capillary profiles typical, i.e., similar to that observed in control rats. The immunosignals represented by solitary gold particles were associated with the luminal plasmalemma or with the EC cytoplasm.

Springer

440

J Neurocytol (2005) 34:435–446

Fig. 9 Typical distribution of GLUT-1 in the capillary wall from the cerebral cortex of control (untreated) rat. The immunosignals for GLUT1 in the abluminal plasmalemma of the EC are more numerous and more regularly distributed (arrowheads) than in the luminal plasmalemma (arrows). X 36,000

Fig. 11 A cortical capillary from rat treated prenatally with LPS (group 3). In the cytoplasm of the EC, unusual vacuoles with goldlabeled membranous coils (asterisks) are present. Immunosignals for GLUT-1 associated with both luminal (arrows) and abluminal (arrowheads) plasma membranes are numerous. ×30,000

Fig. 10 In this capillary from the cerebellum of the rat treated prenatally with VPA (group 2), in spite of the presence of vacuoles in the EC cytoplasm (asterisks), the density of immunosignals for GLUT-1 in both luminal (arrows) and abluminal (arrowheads) plasma membranes remains unaffected. Inside some vacuoles a few gold particles are also present. ×30,000

Fig. 12 A portion of capillary wall in the rat cerebellum from group 3 (LPS-treated) shows the EC cytoplasm containing a few vacuoles (asterisks) and scanty and irregular labeling for GLUT-1 associated with both luminal (arrows) and abluminal (arrowheads) plasma membranes. ×30,000

In this group of animals, however, several segments of microvascular endothelial lining revealed increased endocytosis with concomitant presence of cytoplasmic vesicles or vacuoles labelled with single, few, or several gold particles. There were also a few microvascular profiles, especially in the cerebral cortex, showing slight extravasation of albumin without formation of distinct plasmalemmal vesicles or endosome-like vacuoles containing albumin (Fig. 6). Albumin-containing vacuoles were more frequently present in the EC cytoplasm of the cerebellar capillary profiles with concomitant inconsiderable leakage of albumin (Fig. 7). In some capillary segments of the hippocampus, a few immunosignals for albumin were present inside the EC cytoplasm, in subendothelial basement membrane and in perivascular neuropil (Fig. 8). Generally, although the increased endocytosis and extravasation of albumin was most pronounced in this ex-

perimental group, it was observed only in a few (4– 8%) of the examined vascular profiles (most frequently in the cerebellum), and their intensity was low, as evidenced by only a few to several gold particles located in the perivascular area. Thus, the term “focal microleakage” seems to be the most adequate for this type of BBB inefficiency.

Springer

Glucose transporter (GLUT-1) In BBB-type blood microvessels of untreated, control rats (group no.1), irrespective of their localization in the different brain regions (e.g., in the cerebral cortex, cerebellum or hippocampus), the distribution of immunosignals for GLUT-1 was similar, although some variability in their localization and density was frequently noted. The most reproducible feature of the endothelial lining of brain microvessels was

J Neurocytol (2005) 34:435–446

441

Table 1 Cumulative results of morphometric analysis of labelling density distribution for GLUT-1 in three brain regions of control and VPAand/or LPS-treated rates Cortex

Cerebellum

Experimental group

LPM

APM

Ratio LPM:APM LPM

(1) Control

1.05 ± 0.05

2.82 ± 0.18

1:2.7

1.20 ± 0.11

(2) VPA

0.97 ± 0.04

1:2.3

1.09 ± 0.22

n = 44 2.24 ± 0.08

1.11 ± 0.05

2.58 ± 0.09

APM

Ratio LPM:APM LPM

3.09 ± 0.07

1:2.5

1.04 ± 0.09 2.72 ± 0.15

1:2.4

1.06 ± 0.18 2.53 ± 0.14

n = 35

n = 33 (3) LPS

Hippocampus

2.65 ± 0.58

0.92 ± 0.10

n = 34 (4) VPA + LPS 0.81** ± 0.04 1.80** ± 0.34 1:2.2

2.15* ± 0.10 1:2.3

n = 34 0.78* ± 0.08 1.97*± 0.25 1:2.5

n = 40

n = 40

Ratio LPM:APM 1:2.6

n = 37

n = 40 1:2.3

APM

1:2.4

n = 34 0.95 ± 0.01 2.06* ± 0.28 1:2.2 n = 41 0.88 ± 0.21 2.08* ± 0.32 1:2.3 n = 40

n = number of microvascular profiles examined; numbers represents gold particles (GPs) per μm of the luminal (LPM) and abluminal (APM) plasma membranes of microvascular endothelial cells; mean values and standard deviations are shown; mean values differing significantly between control and experimental groups are marked as follows: *p < 0.05; **p < 0.001.

higher labelling of the abluminal than of the luminal plasma membrane (Fig. 9). Morphometric analysis revealed that the labelling density of theEC abluminal plasmalemma (AP) was approximately 2.5–3.5 times higher than that of the luminal plasmalemma (LP), giving a ratio of mean values (LP:AP), depending on the brain area examined, from 1:2.5 to 1:2.7 (Table 1). In the rat brains from group no. 2 (VPA), some heterogeneity in the distribution of immunosignals for GLUT-1 was noted within single capillary profiles. This consisted of uneven distribution of immunosignals, located mainly on the abluminal side of the endothelial lining, leaving shorter or longer segments of the plasmalemma unlabelled. These occasionally observed irregularities in the intensity of labelling were rather slight and statistically not significant (Table 1). Nevertheless, they indicated some tendency to diminution (statistically insignificant) of the labelling density ratio (LP:AP), mostly in microvessels of the cerebral cortex. The presence of intracytoplasmic endothelial vacuoles in a few microvascular profiles did not appear to affect the distribution of the immunosignals, although several gold particles appeared in the periphery of vacuoles and/or in the surrounding cytoplasm (Fig. 10). In the offspring of LPS-treated dams (group no. 3), in a few vascular profiles, some heterogeneity of moderate intensity in the distribution and density of immunosig nals were observed. In the cerebral cortex, several capillary profiles were lined with endothelium of uneven thickness. The thickened portions of the EC body usually contained one or two spherical or oval vacuoles, occasionally containing membranous

coilings labelled with a few gold particles (Fig. 11). The immunosignals in these ECs, associated with luminal and abluminal plasma membranes, were numerous and typically distributed. The microvascular profiles in the cerebellum and hippocampus, lined with unchanged or slightly attenuated endothelium with a few small oval-shaped vacuoles, showed a diminished number of plasma membrane-bound immunosignals for GLUT-1. Such a diminution and irregularity of the immunolabelling were evident in the abluminal plasmalemma (Fig. 12). The reduction of the labelling density of the abluminal plasmalemma in the vasculature of cerebellum and hippocampus was statistically significant (Table 1). In the rats of group no. 4 (LPS- and VPA-treated), the disturbances in the distribution, paralleled by the reduction of the labelling density, were manifested in several vascular profiles from all three brain regions. In some microvessels from cerebral cortex, the endothelium showed segmental attenuation with concomitant reduction of the labelling (Fig.13). In some vascular profiles, not only segmental attenuation of the endothelium but also the presence of vacuoles in the thicker portion of the EC body was observed (Fig. 14). The morphometry revealed significant lowering of the labelling density of both luminal and abluminal plasma membranes of the ECs (Table 1). In the cerebellum, some capillary profiles also were lined with segmentally attenuated or thickened endothelium. Thick portions of the lining frequently contained cytoplasmic vacuoles and were labelled with a reduced number of immunosignals associated with the luminal and abluminal plasma membranes. The labelling of the adjacent thin

Springer

442

J Neurocytol (2005) 34:435–446

Fig. 13 The segmentally attenuated endothelial lining (upper portion of the picture) of a rat cerebral cortex capillary from group no. 4 (LPS- and VPA-treated rats) showing irregular and slightly lowered immunogold labelling for GLUT-1 of the luminal (arrows) and abluminal (arrowheads) plasma membranes. X 36,000 Fig. 14 A portion of the capillary wall in the cerebral cortex of a rat from group no.4. The endothelial lining is of uneven thickness and shows irregular and segmentally scanty labeling for GLUT-1 on the luminal (arrows) and abluminal (arrowheads) plasma membranes. In the lower portion of the electron micrograph, a vacuole (asterisk) is present in the EC cytoplasm. ×30,000 Fig. 15

A cerebellar capillary of the rat from group no. 4 (similar to

segments of the endothelium was similar to that observed in normal, unaffected microvessels (Fig. 15). Morphometric analysis showed significant lowering of the labelling density of both luminal and abluminal plasma membranes (Table 1). The majority of capillary profiles in the hippocampus were unchanged. There was a tendency, however, to attenuation of the endothelial lining, with concomitant lowering of the labelling density of the abluminal plasmalemma, similar to that oberved in the cortical microvessels, as shown in Fig. 13. Solitary microvascular segments showed slight thickening and vacuolization of the EC cytoplasm. The labelling density for GLUT-1 was in some of these segments unchanged and relatively intense (Fig. 16), whereas in others it was irregular and diminished, especially in the abluminal plasma membrane. Springer

that shown in Fig. 14) is lined with endothelium of uneven thickness: the lower, thinner portion is labeled with numerous immunosignals for GLUT-1 on luminal (arrow) as well as abluminal (arrowheads) plasma membranes. The upper portion with large vacuoles (asterisks) shows scanty and irregular labelling. In the adjacent cell (presumably pericyte), a large cytoplasmic vacuole (dense body?) is labelled with a few gold particles (curved arrow). ×30,000 Fig. 16 A portion of capillary wall in the rat hippocampus from experimental group no. 4 is shown. Although the EC body contains a few cytoplasmic vacuoles, the immunogold labelling for GLUT-1 of the luminal (arrows) and abluminal (arrowheads) plasma membranes is relatively intense. Numerous gold particles are also present in the EC cytoplasm and inside cytoplasmic vacuoles (asterisks). ×30,000

Discussion The main findings of these studies were as follows. (a) Both substances (VPA and LPS), used singly or in combination, affected slightly the tightness of some segments of the brain microvascular network, manifested by focal extravasation of endogenous blood serum albumin. (b) Both these agents also affected the expression and distribution of the endothelial plasma membrane-associated GLUT-1 in the vasculature of three different brain regions. (c) The most pronounced and consistent changes were observed in the blood microvessels of rats treated with both noxious substances (group no. 4), suggesting the synergistic action of LPS and VPA. Although some of the above-mentioned changes were of low intensity and of limited scope, they nevertheless warrant additional comments and discussion.

J Neurocytol (2005) 34:435–446

(a) The applied immunogold procedure appeared to be sensitive enough to reveal even slight extravasation of albumin that can be defined as a focal microleakage. Originally, we were planning to evaluate quantitatively the density of immunosignals for albumin by using morphometry, similar to that described in our previous publications (Vorbrodt & Dobrogowska, 1994; Vorbrodt, 1995). The low intensity of the leakage, however, as well as its scarcity, i.e., its appearance in a few capillary profiles only, made the quantitative measurements unfeasible and rather futile. The extent and intensity of the microleakage manifested by a few immunosignals for extravasated albumin in perivascular areas appeared to be much lower than that observed after osmotic opening of the BBB (Vorbrodt et al., 1994) or after intravascular injection of protamine (Vorbrodt et al., 1995). The cellular mechanisms associated with or responsible for the observed microleakage of albumin across the endothelial lining in selected microvascular segments remain unclear. Presumably, this phenomenon is associated with enhanced endocytosis of albumin by ECs, observed occasionally in rats from group no.2, and more frequently in rats from other experimental groups (nos. 3 and 4). The appearance of extravasated albumin was usually paralleled by the presence of albumin in the EC cytoplasm, frequently inside cytoplasmic vesicles or endosome-like vacuoles. No signs of paracellular albumin leakage through modified interendothelial junctions, i.e., through opened junctional intercellular clefts, were noted. One cannot exclude the possibility that this transvascular leakage resulted from an unknown, noxious action of the applied substances on some metabolic processes occurring in the ECs. Such a supposition is based on the data presented by some authors indicating that LPS enhances endocytosis in the vascular endothelium (Xaio et al., 2001; Banks et al., 2003) or even increases BBB permeability (Temesv´ari et al., 1993; Veszelka et al., 2003). There are, however, no reports indicating a similar action of VPA, although this substance can affect the metabolism and structural integrity of several cell types including microvascular endothelium (Gibbs et al., 2004; Ingram et al., 2000; Omtzigt et al., 1992; Schneider et al., 2001b; Sobaniec-Lotowska & Sobaniec, 1996). Our observation of albumin extravasation indicates that both agents applied prenatally, singly or combined, only insignificantly affect the integrity of the BBB in a very limited number of microvascular profiles. One can conclude that their influence on BBB function in young rats is negligible, at least when low microvascular permeability to protein molecules such as albumin is concerned. (b) The results of our present study indicate that both substances given prenatally induce more evident disturbances in the expression of GLUT-1. Although morphological ex-

443

amination with electron microscope of immunosignals for GLUT-1 revealed several changes and abnormalities in their localization, only with the use of morphometry and quantitative analysis did the differences between experimental groups in the labelling density of microvessels supplying three brain regions became discernible. In experimental group no. 2 (VPA), non-significant variations in immunostaining occurred, whereas in group no. 3 (LPS), immunolabelling of the endothelial abluminal plasma membrane of microvessels supplying the cerebellum and hippocampus was significantly lowered. The most pronounced lowering of GLUT-1 immunoreactivity in all examined brain regions was found in rats from group no. 4 (VPA + LPS). In this experimental group, the microvasculature of cerebral cortex and cerebellum showed the most drastic diminution of the labelling of both endothelial luminal and abluminal plasma membranes, whereas in the hippocampus, the abluminal plasmalemma was most affected. It is interesting that the microvasculature of the cerebellum appears to be most sensitive to the action of both noxious substances, as evidenced by the lowered density of immunolabelling, especially of the EC abluminal plasma membrane, in three groups of rats (nos. 2, 3 and 4). With reference to this observation, it should be mentioned that after chronic administration of VPA to adult rats, several abnormalities were observed in the capillaries of the cerebellum, including swelling and necrosis of ECs, as well as disruption of interendothelial junctions (Sobaniec-Lotowska & Sobaniec, 1996). In our previous studies, the microvasculature of the cerebellum also appeared to be most sensitive or vulnerable to pathogenic or toxic factors. This sensitivity was evidenced by significant diminution of GLUT-1 expression in scrapie-infected mice (Vorbrodt et al., 1999; 2001b), and also in mice prenatally exposed to ethanol (Vorbrodt et al., 2001a). Of significance, several reports indicate that the cerebellum is one of the brain regions showing substantial neuropathological changes associated with autism (Muratori et al., 2001; Allen & Courchesne, 2003; Kaufmann et al., 2003). Because the most affected groups of rats (nos. 3 and 4) were treated with LPS, it appears that this agent exerts detrimental action on endothelial GLUT-1. Whether such action is direct or indirect remains unclear because no data are available in the relevant literature. However, previously mentioned reports on the action of LPS on BBB permeability (Temesv´ari et al., 1993; Banks et al., 2003) suggest that this substance does affect the brain microvascular endothelium, and one of its noxious effects can lead to disturbances in the glucose-transporting mechanism. The eventual influence of VPA should also be taken into consideration, because this factor can sensitize an immature brain to subsequent injuries (Christianson et al., 1994; Williams et al., 2001) and also can affect the structure and function of the Springer

444

BBB vessels (Sobaniec-Lotowska & Sobaniec, 1996; Gibbs et al., 2004). In effect, the simultaneous action of LPS and VPA has a synergistic character leading to impairment of some barrier functions in the brain microvascular network, manifested by focally increased permeability to albumin and decreased supply of glucose by the affected GLUT-1. This impairment appears to be widespread, affecting all three distant brain regions examined (cortex, hippocampus and cerebellum). The disturbed glucose transport across the capillary wall would be expected to have significant impact on developing and mature brain function, because glucose is the main energy source for neurons and other cells in the brain. It was shown by several authors that GLUT-1 is upregulated during development of the brain, suggesting an early functioning of the BBB-associated glucose- transporting system in the embryo (Dermietzel et al., 1992; Cornford et al., 1993; Bolz et al., 1996). The lower density of GLUT-1, observed in young rats exposed to both agents, may reduce the availability and utilization of glucose by neurons located in the area supplied by the affected microvascular segments, resulting in focal brain malfunction manifested by memory deficits and/or behavioral disturbances. The importance of glucose supply to the brain is well illustrated by the existence of a rare disease known as glucose transporter deficiency syndrome, which is characterized by seizures, developmental delay, microcephaly and hypoglycorrhachia (De Vivo et al., 1990; Pascual et al., 2002; Gordon & Newton, 2003; Xia et al., 2004). (c) The observed synergistic action of prenatally applied LPS and VPA on brain microvasculature and question of underlying mechanisms has already been mentioned while discussing the results of our study (see above). Although the applied ultrastructural immunocytochemical technique is sensitive and makes it possible to detect even discrete changes in the localization of albumin and GLUT-1 molecules, it is not suitable for full elucidation of this important and interesting issue. Multiple environmental factors are involved in induction of neurobehavioral changes. These factors can exert a negative impact on living organisms, especially during the early stages of development, when newly established biological systems are most vulnerable. Taking into consideration the probability of the concomitant action of more than one environmental noxious factor in real life (multi-hit hypothesis), we established the appropriate experimental model, utilizing two selected agents (LPS and VPA) encountered in real life situations. It is advisable to recall that according to Engelhardt and Risau (1995), the vascular system of the brain develops in three phases: (1) vasculogenesis, the formation of primary vascular plexus on the surface of the neural tube, (2) angiogenesis, i.e., sprouting of new vessels from the plexus into Springer

J Neurocytol (2005) 34:435–446

neural tube, and (3) barriergenesis, the formation of interendothelial tight junctions. The onset of angiogenesis occurs in rat at precisely reproducible stages of embryonic development on days E10-E11, whereas barriergenesis and the appearance of GLUT-1 in the EC plasma membranes start at E17 (Bolz et al., 1996). Taking these data into consideration, it is evident that the action of LPS (given at E10) occurs exactly at the onset of angiogenesis, but before the extensive proliferation of primary neuroectodermal cells. Injection of VPA (at E12.5) also occurred during the later stage of angiogenesis but before barriergenesis. Thus, the action of both agents could directly affect the proliferating primary angiogenic cells. In effect, however, the tightness of the vessels, conditioned by the newly formed interendothelial junctions, was not significantly affected after birth, as evidenced by scanty microleakage of albumin in a limited number of microvascular profiles. By contrast, the action of LPS and VPA on the expression of GLUT-1 was more pronounced. It is likely that both agents exert unknown influences on metabolic pathways occurring in several cell types, leading to developmental disturbance of the brain. As a consequence, not only is the function of neurons and other cells impaired, but also the structure and function of at least a small fraction of brain microvessels may be defective. The manifestations of BBB dysfunction induced by both agents that were observed with morphological methods do not explain the underlying molecular mechanisms of this process. The elucidation of this complex problem requires further experimental studies, including quantitative genetic and molecular biology methods. Acknowledgments We wish to express our appreciation to Ms. M. Stoddard Marlow for meticulous editorial revisions and to Ms. J. Kay for secretarial assistance. This work was supported by the NewYork State Office of Mental Retardation and Developmental Disabilities.

References ALLEN, G.

& C O U R C H E S N E , E . (2003) Differential effects of developmental cerebellar abnormality on cognitive and motor functions in the cerebellum: an fMRI study of autism. American Journal of Psychiatry 160, 262–273.

BANKS, W. A., KASTIN, A. J., BRENNAN, J. M. & V A L L A N C E , K . L . (2003) Absorptive endocytosis of HIV-1

gp120 by blood-brain barrier is enhanced by lipopolysaccharide. Experimental Neurology 156, 165–171. BOLZ, S., FARRELL, C. L., DIETZ, K

& WOLBURG, H.

(1996) Subcellular distribution of glucose transporter (GLUT–1) during development of the blood-brain barrier in rats. Cell and Tissue Research 284, 355–365. CORNFORD, E. M., HYMAN, S.

& PARDRIDGE, W. M.

(1993) An electron microscopic immunogold analysis of developmental up-regulation of the blood-brain barrier Glut1 glucose transporter. Journal of Cerebral Blood Flow and Metabolism 13, 841–854.

J Neurocytol (2005) 34:435–446 CHRISTIANSON, A. L., CHESLER, N. & KROMBERG, J. G . (1994) Fetal valproate syndrome: clinical and neurodevelop-

mental features in two sibling pairs. Developmental Medicine and Child Neurology 36, 361–369. DERMIETZEL, R., KRAUSE, D., KREMER, M., WANG, C . & S T E V E N S O N , B . (1992) Pattern of glucose transporter

(Glut1) expression in embryonic brains is related to maturation of blood-brain barrier tightness. Developmental Biology 193, 152–163. DE VIVO, D. C., TRIFILETTI, R., JACOBSON, R. I. & H A R I K , S . I . (1990) Glucose transporter deficiency causing

persistent hypoglycorrhachia: A unique cause of infantile seizures and acquired microcephaly. Annals of Neurology 28, 414–415. D O B R O G O W S K A , D . H . & V O R B R O D T , A . W . (1999) Quantitative immunocytochemical study of blood-brain barrier glucose transporter (Glut-1) in four regions of mouse brain. Journal of Histochemistry and Cytochemistry 47, 1021–1029. D R E W E S , L . R . (1998) Biology of the blood-brain glucose transporter. In Introduction to the Blood-Brain Barrier (edited by Pardridge, W. M.) pp. 165–174. Cambridge: Cambridge University Press.

445 LIPPOLDT, A., KNIESEL, U., LIEBNER, S., KALBACHER, H., KIRSCH, T., WOLBURG, H. & H A L L E R , H . (2000) Structural alterations of tight junctions

are associated with loss of polarity in stroke-prone spontaneously hypertensive rat blood-brain barrier endothelial cells. Brain Research 885, 251–261. M C I L V A N E , W . J . & C A T A L D O , M . F . (1996) On the clinical relevance of animal models for the study of human mental retardation. Mental Retardation and Developmental Disability Research Review 2, 188–196. M U R A T O R I , F . , C E S A R I , A . & C A S E L L A , C . (2001) Autism and cerebellum. An unusual finding with MRI. Panminerva Medicine 43, 311–315. OMTZIGT, J. G., NAU, H., LOS, F. J., PIJPERS, L. & L I N D H O U T , D . (1992) The disposition of valproate and its

metabolites in the late first trimester and early second trimester of pregnancy in maternal serum, urine, and amniotic fluid: Effect of dose, co-medication, and the presence of spina bifida. European Journal of Clinical Pharmacology 43, 381–388. PASCUAL, J.M., VAN HEERTUM, R.L., WANG, D., E N G E L S T A D , K . & D E V I V O , D . C . (2002) Imaging the

EL-KHODOR, B., KOZLOWSKI, P. B., RABE, A. & LEE, M . H . (2003) Prenatal valproic acid and lipopolysaccharide

metabolic footprint of Glut 1 deficiency on the brain. Annals of Neurology 52, 458–464.

synergistically produce behavioral and neuroanatomical changes. Developmental Psychobiology 43, 253. E N G E L H A R D T , B . & R I S A U , W . (1995) Development of the blood-brain barrier. In new concepts of a blood-brain barrier (edited by Greenwood, J., Begley, D.J. & Segal, M.B.) pp.11–31. New York and London: Plenum Press. F A R R E L L , C . L . & P A R D R I D G E , W . M . (1991) Blood-brain barrier glucose transporter is asymmetrically distributed on brain capillary endothelial luminal and abluminal membranes: An electron microscopic immunogold study. Proceedings of the National Academy of Sciences USA 88, 5779–5783.

PLETNIKOV, M. V., MORAN, T. H. & CARBONE, K. M . (2002) Borna disease virus infection of the neonatal rat:

FATEMI, S. H., EARLE, J., KANODIA, R., KIST, D., EMAMIAN, E. S., PATTERSON, P.H., SHI, L.M. & S I D W E L L , R . (2002) Prenatal viral infection leads to pyramidal

cell atrophy and macrocephaly in adulthood: Implications for genesis of autism and schizophrenia. Cellular and Molecular Neurobiology 22, 25–33. GIBBS, J. P., ADEYEYE, M. C., YANG, Z. & SHEN, D . D . (2004) Valproic acid uptake by bovine brain microvessel

endothelial cells: role of active transport. Epilepsy Research 58, 53–66. G O R D O N , N . & N E W T O N , R . W . (2003) Glucose transporter type 1 (GLUT-1) deficiency. Brain Development 25, 477–480. HORNIG, M., WEISSENBOCK, H., HORSCROFT, N. & L I P K I N W I (1999) An infection-based model of neurodevelop-

mental damage. Proceedings of the National Academy of Sciences USA 96, 12102–12107. H O R N I G , M . & L I P K I N , W . I . (2001) Infectious and immune factors in the pathogenesis of neurodevelopmental disorders: Epidemiology, hypotheses, and animal models. Mental Retardation and Developmental Disability Research Review 7, 200–210. INGRAM, J. L., PECKHAM, S. M., TISDALE, B. & R O D I E R , P . M . (2000) Prenatal exposure of rats to valproic

acid reproduces the cerebellar anomalies associated with autism. Neurotoxicology and Teratology 22, 319–324. KAUFMANN, W. E., COOPER, K. L., MOSTOFSKY, S. H., CAPONE, G. T., KATES, W. R., NEWSCHAFFER, C. J., BUKELIS, I., STUMP, M . H . , J A N N , A . E . & L A N H A M , D . C . (2003)

Specificity of cerebellar vermian abnormalities in autism: A quantitative magnetic resonance imaging study. Journal of Children Neurology 18, 463–470.

Developmental brain injury model of autism spectrum disorders. Frontiers of Biosciences 7, D593–D607. SCHNEIDER, M. L., MOORE, C. F. & KRAEMER, G. W . (2001a) Moderate alcohol during pregnancy: Learning and

behavior in adolescent rhesus monkeys. Alcohol Clinical and Experimental Research 25, 1383–1392. S C H N E I D E R , T . , L A B U Z , D . & P R Z E W L O C K I , R . (2001b) Nociceptive changes in rats after prenatal exposure to valproic acid. Polish Journal of Pharmacology 53, 531–534. S O B A N I E C - L O T O W S K A , M . E . & S O B A N I E C , W . (1996) Morphological features of encephalopathy after chronic administration of the antiepileptic drug valproate to rats. A transmission electron microscopic study of capillaries in the cerebellar cortex. Experimental Toxicology and Pathology 48, 65–75. TARNAWSKI, J., WEN, G.Y., HEANEY, A., MA, S., EL-KHODOR, B.F., RABE, A., KOZLOWSKI, P.B. & L E E , M . H . (2004) Combined exposure to lipopolysaccha-

ride and valproic acid during early embryonic development in rats. Program No. 359.12. Society for Neuroscience Abstracts Viever/Itinerary Planner, Washington DC. ´ RI, P., A ´ B R A HA ´ M, C. S., SPEER, C. P., T E M E S VA ´ C S , J . & M E G Y E R I , P . (1993) Escherichia coli 0111 K O VA B4 lipopolysaccharide given intracisternally induces blood-brain barrier opening during experimental neonatal meningitis in piglets. Pediatric Research 34, 182–186. ´ N Y I , Z . , PA ´ Z MA ´ N Y , T . , NE´ M E T H , V E S Z E L K A , S . , U R BA ´ L, I., DUNG, N.T.K., A ´ B R A HA ´ M, C.S., L . , O BA ´ , G . & D E L I , M . A . (2003) Human serum amyloid S Z A BO component attenuates the bacterial lipopolysaccharide-induced increase in blood-brain barrier permeability in mice. Neuroscience Letters 352, 57–60. V O R B R O D T , A . W . (1993). Morphological evidence of the functional polarization of brain microvascular endothelium. In The Blood-Brain Barrier (edited by Pardridge, W.M.) pp. 137–164. New York: Raven Press. V O R B R O D T , A . W . (1995) The application of quantitative immunocytochemistry for the evaluation of blood-brain barrier (BBB) to endogenous albumin. In New Concepts of a Blood-Brain Barrier (edited by Greenwood, J., Begley, , D.J. & Segal, M. B.) pp. 39–46. New York and London: Plenum Press.

Springer

446

J Neurocytol (2005) 34:435–446

VORBRODT, A. W.

& D O B R O G O W S K A , D . H . (1994) Immunocytochemical evaluation of blood-brain barrier to endogenous albumin in adult, newborn and aged mice. Folia Hictochemica et Cytobiologica 32, 63–70.

VORBRODT, A. W., TARNAWSKI, M. &

DOBROGOWSKA, D. H., L O S S I N S K Y , A . S . (1994) A

quantitative immunocytochemical study of the osmotic opening of the blood-brain barrier to endogenous albumin. Journal of Neurocytology 23, 792–800. VORBRODT, A. W., DOBROGOWSKA, D. H., UENO, M. & L O S S I N S K Y , A . S . (1995) Immunocytochemical studies

of protamine-induced blood-brain barrier opening to endogenous albumin. Acta Neuropathologica 89, 491–499. VORBRODT, A. W., DOBROGOWSKA, D. H., TARNAWSKI, M., MEEKER, H. C. & CARP, R. I . (1997) Immunocytochemical evaluation of blood-brain

barrier to endogenous albumin in scrapie-infected mice. Acta Neuropathologica 93, 341– 348. VORBRODT, A. W., DOBROGOWSKA, D. H., MEEKER, H . C . & C A R P , R . I . (1999) Immunogold study of regional

differences in the distribution of glucose transporter (GLUT-1) in mouse brain associated with physiological and accelerated aging and scrapie infection. Journal of Neurocytology 28, 711–719. VORBRODT, A. W., DOBROGOWSKA, D. H., KOZLOWSKI, P., TARNAWSKI, M., DUMAS, R . & R A B E , A . (2001a) Effect of a single embryonic expo-

Springer

sure to alcohol on glucose transporter (Glut-1) distribution in brain vessels of aged mouse. Journal of Neurocytology 30, 167–174. VORBRODT, A. W., DOBROGOWSKA, D. H., TARNAWSKI, M., MEEKER, H. C. & CARP, R . I . (2001b) Quantitative immunogold study of glucose

transporter (GLUT-1) in five brain regions of scrapie-infected mice showing obesity and reduced glucose tolerance. Acta Neuropathologica 102, 278–284. VORHEES, C. V., WEISENBURGER, W. P. & MINCK, D. R . (2001) Neurobehavioral teratogenic effects of thalido-

mide in rats. Neurotoxicology and Teratology 23, 255– 264. WILLIAMS, G., KING, J., CUNNINGHAM, M., STEPHAN, M . , K E R R , B . & H E R S H , J . H . (2001) Fetal valproate

syndrome and autism: Additional evidence of an association. Developmental Medicine and Children Neurology 43, 202– 206. XAIO, H., BANKS, W.A., NIEHOFF, M.L. & MORLEY, J . E . (2001) Effect of LPS on the permeability of the blood-brain barrier to insulin. Brain Research 896, 36–42. XIA, M., WANG, D., PASCUAL, J. M., YANG, H., S T I N E R , S . & D E V I V O , D . C . (2004) Effects of

valproic acid on Glut-1 gene expression. Program No. 197.9 Society of Neuroscience Abstract Viewer/Itinerary Planner. Washington, DC.