THE JOURNAL OF COMPARATIVE NEUROLOGY 321~19-32 (1992)
Effects of Prenatal Ethanol Exposure on the Development of Bergmann Glia and Astrocytes in the Rat Cerebellum: An Immunohistochemical Study ASHOK K. SHETTY AND DWIGHT E. PHILLIPS Department of Biology and WAMI Medical Education Program, Montana State University, Bozeman, Montana 59717-0346
ABSTRACT The consequences of prenatal ethanol exposure on the postnatal development of Bergmann glia and astrocytes in the rat cerebellum were investigated by using glial fibrillary acidic protein (GFAP) immunolabeling. Pregnant rats were either fed with an ethanol containing liquid diet (6.7% v/v) or pair-fed with an isocaloric diet throughout gestation. On postnatal day (PD) 15 and 22, parasagittal sections of the cerebellar vermis from female offspring were processed for GFAP immunohistochemistry to assess the development of Bergmann glia and astrocytes in lobules I, VII, and X and astrocytes in the central core of white matter. On PD 15, compared to control animals, ethanol exposed animals had fewer GFAP positive Bergmann glial fibers per unit length of molecular layer; a significantly greater percentage of morphologically immature Bergmann fibers; a significantly greater GFAP positive astrocytic area per unit area of internal granular layer and central white matter; and the astrocytic processes were wider and more closely packed. These glial changes were associated with significantly thicker external granular layer in all 3 lobules. However, no significant differences were seen between the ethanol exposed and control animals on PD 22, indicating "catch-up growth" in the ethanol exposed animals during the third postnatal week. These results suggest that prenatal ethanol exposure causes (1)delayed maturation of Bergmann glia, which in turn contributes to the delayed migration of granule cells; and (2) alterations in the normal postnatal development of aStrOCyteS. o 1992 Wiley-Liss, Inc. Key words: glial fibrillary acidic protein, radial glia, granule cell, alcohol and cerebellar cortex
Many studies have unequivocally demonstrated that gestational alcohol exposure is detrimental to brain development and function in the offspring (Streissguth et al., '78; Riley et al., '86; Streissguth, '86; West and Pierce, '86). The teratogenic effect of ethanol on the developing nervous system in experimental animals is mainly characterized by abnormalities in nerve cell proliferation, migration and maturation (Borges and Lewis, '83; Kennedy and Elliot, '86; Miller, '86, '88; Al-Rabiai and Miller, '89) as well as by delays in the acquisition of myelin (Samorajski et al., '86; Phillips, '89) and the maturation of glia (Kennedy and Mukerji, '86a,b; Phillips and Krueger, '90). Postmortem studies of children with fetal alcohol syndrome (FAS) show aberrant cytoarchitecture and neuroglial heterotopia indicative of disorganized migration of neuronal and glial elements (Clarren et al., '78; Peiffer et al., '79; Clarren, '86). Delayed migration and maturation of neurons could be due to defective development of glia particularly since radial glia, a type of astroglia, are proposed to play an important
o 1992 WILEY-LISS, INC.
role in the migration and differentiation of neurons (Rakic, '85; Lauder and McCarthy, '86; Manthorpe et al., '86). Though some studies have addressed the potential effects of alcohol on the development of glial cells in vitro (Davies and Vernadakis, '84; Kennedy and Mukerji, '86a,b; Chiappelli et al., '91; Davies and Cox, '91; Davies and Ross, '91) and in optic nerve in vivo (Phillips and Krueger, '901, there is generally a paucity of knowledge concerning the effects of in vivo alcohol exposure on the ontogenesis and maturation of astrocytes in different regions of the central nervous system. The cerebellum is a suitable region for studying neuronastroglia interactions during development because of the limited number of neuronal and astroglial lineages and because the regularly organized developing cell populations have been well characterized in terms of their morphology and the timing of their development (Altman, '69, '72; Accepted March 2 , 1992.
A.K. SHETTY AND D.E. PHILLIPS
20 Palay and Chan Palay, '74; Wilkin and Levi, '86; Berciano et al., '90). The two types of astroglia in the developing cerebellar cortex, Bergmann glia and velate astrocytes, interact differently with neurons (Ramon y Cajal, '11; Palay and Chan Palay, '74; Hatten et al., '84).The vertically oriented Bergmann glial fibers in the molecular layer are believed to play an important role in the guidance of migrating immature granule cells from the external granular layer (EGL) to the internal granular layer (IGL) (Rakic, '71; Rakic and Sidman, '73; Rakic, '85). On the other hand, velate astrocytes in the IGL are mainly responsible for the clustering of granule cell bodies and the organization of perisynaptic glia surrounding cerebellar glomeruli (Palay and Chan Palay, '74; Wilkin and Levi, '86). Many studies report evidence of alcohol induced abnormalities or delays in the migration of neurons from the EGL across the molecular layer to the IGL (Bauer-Moffett and Altman, '77; Chernoff, '77; Volk, '77; Kornguth et al., '79; Borges and Lewis, '83). Such effects could be due to defects or delays in the development and maturation of Bergmann glia. In addition, the reduced number of granular neurons in the IGL following alcohol exposure (Pierce et al., '89; Bonthius and West, '90; West et al., '90) may induce secondary changes in the velate astrocytes apart from the effects of alcohol on the genesis of these astrocytes. Despite these evidences, information regarding the effects of alcohol on the development and maturation of Bergmann glia and other astrocytes in the cerebellum is not available in the literature. Glial fibrillary acidic protein (GFAP) immunohistochemistry was used in the present investigation to examine the effects of prenatal ethanol exposure on the development of Bergmann glia, velate astrocytes in the IGL and astrocytes in the central white matter core of the rat cerebellum. GFAP is the major intermediate filament protein of Bergmann glia and astrocytes and has been widely used as a specific immunohistochemical marker for these cells in the cerebellum. Postnatal day (PD) 15 was chosen for the initial investigation, because by PD 15 (1)the GFAP positive axial filamentous cores of the Bergmann fibers are well developed and extend up to the external limiting membrane from the Purkinje cell layer; (2) GFAP is well expressed in the astrocytic processes in both the IGL and the central core of white matter (Ghandour et al., '81); and (3) the proposed Bergmann glial mediated migration of granule cells from the EGL to the IGL is active (Altman, '72). PD 22 was chosen for later analysis to determine if any changes that were present at PD 15 persisted in the maturation of Bergmann glia and granule cell migration or in the maturation of other astrocytes since by PD 22 the migration of granular neurons from the EGL to the IGL is completed (Wilkin and Levi, '86).
MATERIALS AND METHODS Animals and diet regimen Nulliparous female Sprague-Dawley rats (Holtzmann/ Sasco Co.) weighing from 220 to 300 g were caged with individual male rats for breeding in the evening. The next morning those female rats with an ejected vaginal plug (gestational day 0 ) were divided into two weight-matched groups: (1)ethanol consuming dams-these animals were given free access to liquid diet (Bioserv Inc., Frenchtown, N.J.) containing 37.5%ethanol derived calories (6.7%viv); and (2) pair-fed dams-these animals received the same
volume of liquid diet containing isocaloric amounts of maltose-dextrin in place of ethanol. A chow-fed control group was not included in this study since a number of previous studies and a pilot study in our laboratory (with the same diet from Bioserv Inc.) clearly indicated that there were no significant differences between the chow-fed and pair-fed control dams in either weight gain during pregnancy or in litter size. In addition, significant differences were not shown between the chow-fed and pair-fed control offspring in terms of body or brain weights and in terms of the appearance of developmental markers such as eye opening, righting reflex and dental eruption (Miller, '88; Gottesfeld et al., '89; Miller and Potempa, '90). All dams received fresh diet at 9.30 A.M. daily, beginning on the 0 gestational day (GD) and continuing to parturition. The dams were housed individually under constant conditions of temperature, humidity and lighting (12 hour light-dark cycle). The ethanol containing liquid diet was available ad libitum throughout the day. Each ethanol fed rat was paired by weight with a rat fed control diet and the two were pair-fed; that is, each control rat was given the same volume of liquid diet as that consumed by the weight matched ethanol fed rat on the respective gestation day. Thus, during the entire gestation period, both the ethanol fed and control dams consumed diets of identical caloric content. The blood alcohol concentration (BAC) in the ethanol exposed dams was determined from blood serum by using a Sigma diagnostic kit (#332 UV).Blood samples were taken from the tail veins of ethanol fed rats at 10 P.M. on GD16,6 hours after the start of the dark cycle when most animals have consumed a large portion of their diet. On the day of birth, neonates from each diet group were pooled, randomly culled to 8 pups per litter (4 males and 4 females), and fostered to untreated dams which gave birth within the same period, so that no pups were reared by their natural mothers.
Tissue processing On PD 15 and 22 (day of birth = PD 11, six females from each group (pair-fed control and alcohol exposed offspring) were sacrificed. Each animal was weighed, then anesthetized with ether prior to sacrifice by transcardiac perfusion with 4% paraformaldehyde in 0.1M cacodylate buffer (pH = 7.4). The brain was dissected from the cranial cavity and immersed in the same fixative for 2 hours. The fixed brains were weighed after trimming off olfactory bulbs, cerebellar flocculi and spinal cord. Cerebella were then dissected out and postfixed in the same primary fixative for 18 hours at 4°C before immersion in 0.1M phosphate buffered saline (PBS, pH = 7.4).Fifty micron thick parasagittal sections near the middle of the cerebellar vermis were cut on a vibratome, collected in 0.1M PBS and washed twice.
GFAP immunohistochemistry The sections were treated with a solution of 0.1M PBS containing 3% hydrogen peroxide and 20% methanol for 15 minutes to remove endogenous peroxidase, then washed thoroughly for at least 2 hours with 8 changes of PBS containing 0.3% Triton X-100. GFAP was labeled in freefloating sections by using a polyclonal antiserum obtained from Sigma Chemicals (St. Louis, MO). The sections were first incubated in 10% normal goat serum (GIBCO BRL, NY) for 30 minutes followed by incubation in GFAP
21
EFFECTS OF ALCOHOL ON CEREBELLAR ASTROGLIA antibody (developed in rabbit) at a dilution of 1 : l O O in 0.01M PBS containing 0.1% normal goat serum and 1% Triton X-100. Sections were incubated for 20 hours at room temperature with gentle agitation in a laboratory shaker. Following a thorough wash in PBS, the sections were incubated in peroxidase conjugated goat anti-rabbit IgG (Sigma Chemicals) at a dilution of 1500 in 0.01M PBS for 2 hours at room temperature. The sections were then washed three times in PBS and rinsed once in 0.05M Tris buffer followed by incubation for 10 minutes in a solution containing 0.05% DAB (3,3 diaminobenzidine tetrachloride) in 0.05M Tris buffer. The peroxidase reaction was completed by treating the sections for 5 minutes with 0.05M Tris buffer solution containing 0.05% DAB and 0.01% hydrogen peroxide. After another thorough wash in Tris buffer, sections were mounted on gelatin treated slides, air dried, dehydrated, cleared, and coverslipped with Permount. The immunohistochemical staining of sections from ethanol exposed and pair-fed control animals was done at the same time in parallel by using the same dilutions of primary and secondary antibody solutions, the same number of washes and the same concentrations of DAB and hydrogen peroxide. The same procedure was followed for negative control sections except that the primary antibody was omitted and the sections were incubated in normal goat serum only. Neither immunostaining nor any recognizable background staining was observed under these conditions in the negative control sections (Figs. lg,h). A few sections from each animal were counter-stained with Mayer's hematoxylin (Sigma Chemicals) to visualize the layers in the cerebellar cortex. Observations and photography were done with an Olympus photomicroscope.
Morphometric analysis Morphometric studies were performed by using a bit pad (Sigma Scan, Jandel Scientific, Corte Madera, CA) or an image analysis system (MCID, Microcomputer Imaging Device, Imaging Research Inc., St. Catharines, Ontario, Canada) on lobules I, VII, and X of cerebellar vermis as defined by Larsell ('52) and on the central core of white matter. Three sections of cerebellar vermis were evaluated for each animal. All data were collected blind to experimental codes and means were calculated for each animal individually before the means were determined for the six animals per group. Statistical comparisons were done on measurements and counts from 6 experimental and 6 control animals at each age with a Student's t-test (two tailed). The density of GFAP positive Bergmann glial fibers per unit length of folia surface in the molecular layer was determined at a consistent area in each of the 3 lobules. ) Fibers were counted in a camera-lucida image ( 1 9 3 ~by using a bit pad with a lighted cursor to measure the length of the area and to mark the counts. The fibers were characterized as either mature or immature types based on their morphology. GFAP positive astrocytic area per unit area of tissue was measured by MCID densitometric analysis in the IGL of three vermal lobules and in the central core of the white matter. The microscopic image was transferred to the videoscreen by focussing the appropriate area in the immunostained section with an Olympus microscope equipped with a video camera. The same intensity of light was used to focus all the samples measured. The densitometric program has a continuous gray scale composed of 255 levels with the
densest gray level assigned the lowest number and the lightest level assigned the highest number. The background and target values were set to 255 and 172, respectively, following digitization of the original gray value image in the videoscreen. Those values were determined by selecting the background and target areas in several sections from both ethanol exposed and pair-fed control animals before commencing the measurements on coded slides for statistical analysis. This scale eliminated the background staining completely and retained all the target (GFAP positive) structures in the range 0-172. The binary image of GFAP positive elements was then generated by selecting a suitable threshold value (which varied from 215 to 225) to include all the GFAF' positive structures without any background. The final binary image was cross-checked with the original gray value image by alternating the two images on the videoscreen. Finally, the image was frozen and the area occupied by the GFAP positive structures in the field (0.04288mm2in area) was measured by scanning the whole field on the videoscreen. The data was then stored for further calculations and statistical analysis. The cross-sectional area of the lobules was measured on a bit pad by tracing the camera-lucida image of the lobules (37x1 in immunocytochemically stained parasagittal sections of the vermis. Each lobule was marked uniformly (Pierce and West, '87) and three sections were measured for each animal. The thickness of the EGL was measured on the bit pad (940 x 1 by taking three measurements at a consistent region in each of the three lobules from hematoxylin counter-stained midvermal sections in every PD 15 animal. The thickness of the EGL was not measured in PD 22 animals because the EGL in the same regions measured on PD 15 was either not present or was only one cell thick in both the ethanol exposed and pair-fed control animals.
RESULTS The weight gain by ethanol consuming rats during pregnancy was not significantly different than that of pair-fed control animals. From GD1 to GD20, the ethanol fed rats averaged 39.9%in weight gain compared to 35.9% in the pair-fed control animals ( P < 0.3). The mean daily ethanol consumption in the ethanol exposed dams was 10.9 0.6 gikg body weight during GD 6-20, while the average BAC on GD16 was 139.8 43.7 mgidl (n = 6). The brain weights of the pups exposed to ethanol prenatally were slightly less than the pair-fed control pups at both ages studied but the differences were not significant statistically (6% less on PD 15, P < 0.06 and 4%less on PD 2 2 , P < 0.1).
*
*
GFAP immunoreactivity Immunostaining with the antibody directed against GFAP resulted in labeling of Bergmann glial fibers in the molecular layer and astrocytes in the internal granular layer and white matter of the cerebellums in both ethanol exposed and pair-fed control animals. The intensity of staining was greater in the Bergmann glial fibers and astrocytic processes compared to the somata of the cells. Morphology o f Bergmann glial fibers. In both ethanol exposed and control offspring, GFAP positive Bergmann glial fibers appeared as regularly spaced parallel palisades oriented radially in the molecular layer, extending from the upper limit of the Purkinje cell layer towards the pial
22
Fig. 1. Immunoreactivity for glial fibrillary acidic protein (GFAP) in the cortex of cerebellar vermis on PD 15. a, c, and e are from lobules I, VII, and X, respectively, of ethanol exposed animals. b, d, and f are from lobules I, VII, and X, respectively, of pair-fed control animals. Note the vertically oriented Bergmann glial fibers in the molecular
A.K. SHETTY AND D.E. PHILLIPS
layer (ML) and astrocytic processes in the internal granular layer (IGL). The thicker external granular layer (arrows) and fewer Bergmann fibers are clearly evident in lobules VII (c) and X (el of ethanol exposed animals compared to control animals (d,f). x 170.
EFFECTS OF ALCOHOL ON CEREBELLAR ASTROGLIA
23
Fig. 1 (continued). g,h: Photomicrographs of negative control seetions of cerebellum used in the GFAP immunohistochemicalprocedure. Neither specific immunostaining nor non-specific background staining are seen. The different layers of cerebellar cortex and central white
matter core are recognizable. EGL = external granular layer; ML = molecular layer; IGL = internal granular layer; WM = white matter; CWM = central white matter core of cerebellum. x 185.
surface where the glial endfeet formed a continuous glial limiting membrane (Fig. la-f). Two types of GFAP positive Bergmann fibers could be characterized in both groups on PD 15: (1)thin and smooth fibers with club shaped glial end feet (growth cones) terminating either in the upper part of molecular layer or lower half of the EGL; (2) thick, smooth, and slightly twisted fibers with end feet either touching or ending close t o the external limiting membrane (ELM).The thin fibers appeared to be new filiform processes growing
towards the surface of the cerebellum (Das, '76; Shiga et al., '831, whereas the thick fibers with variable end feet (conical, club shaped or flattened) appeared more like twisted ribbons than the fine caliber filiform processes. Since it is known that mature GFAP positive Bergmann fibers extend to the ELM and that the expression of GFAP gradually increases during postnatal development (hence thickness of the axial core of these fibers, Ghandour et al., '811, we classified the thin fibers that did not extend to the ELM as
24
Fig. 2. Magnified view of GFAP immunoreactivity in the cerebellar cortex on postnatal day (PD) 15. a and b show GFAF' positive Bergmann glial fibers in an ethanol exposed animal and a pair-fed control animal, respectively. Note that there are fewer Bergmann fibers in the ethanol exposed animal than in the pair-fed control animal. Arrows indicate fine caliber (immature) Bergmann fibers in the external granular layer (EGL) falling short of the external limiting membrane while arrowheads indicate the thicker and slightly twisted (more mature) Bergmann fibers terminating in or close to the external limiting membrane. A comparatively greater number of immature
A.K. SHETTY AND D.E. PHILLIPS
GFAP positive Bergmann fibers are evident in the ethanol exposed animal (a).ML = molecular layer. c and d show GFAP immunopositive structures in the internal granular layer of an ethanol exposed animal and a control animal, respectively. In the ethanol exposed animal, the cell bodies of astrocytes (arrows) are stained and astrocytic processes are thicker, and greater in number whereas in the control animal only faintly stained thinner astrocytic processes without somata are obvious. ~ 3 4 0 .
25
EFFECTS OF ALCOHOL ON CEREBELLAR ASTROGLIA immature types and the thick fibers that end either in or close to the ELM as mature types. Based on these criteria, the qualitative differencesapparent between the two groups of animals on PD 15 were subtle although detailed evaluation of the two groups indicated that there was a relatively greater number of immature fibers than mature fibers in the ethanol exposed animals (Fig. 2a,b). Counts of the immature and mature fiber types revealed that the ethanol exposed animals had a significantly greater percentage of immature fibers than control animals on PD 15 in all three lobules studied (Table 1). No differences were seen in the morphology of individual GFAP positive Bergmann glial fibers on PD 22. The maturation of Bergmann fibers in both ethanol exposed and control animals appeared similar (Fig. 3). Virtually all of the fibers were of the mature type in both groups so relative counts of mature and immature types were not done.
Number of Bergmann glial fibers. There were fewer fibers per unit length in ethanol exposed animals on PD 15 in all three lobules studied (Fig. 4).The differences were
Fig. 3. GFAF' immunoreactivity in the PD 22 cerebellar cortex (Lobule X) of an ethanol exposed animal (a)and a pair-fed control animal (b).Thick and mature Bergmann glial fibers with end feet (arrows) touching the external limiting membrane are structurally
similar in both groups. ML = molecular layer; IGL = internal granular layer. The external granular layer is missing in these particular examples in both groups. x 185.
TABLE 1. Percentage (%) of Mature and Immature Bergniann Fibers in Different Lobules on Postnatal day 15'
Region of cerebellar vermis Lobule I Mature fibers Immature fibers Lobule VII Mature fibers Immature fibers Lobule X Mature fibers Immature fibers
ETOH exposed (n = 6)
Control (n = 6)
45.3 (11) 54.7 (11)
81.4 (9.3) 18.6 19.3)
Effects of Prenatal Ethanol Exposure on the Development of Bergmann Glia and Astrocytes in the Rat Cerebellum: An Immunohistochemical Study ASHOK K. SHETTY AND DWIGHT E. PHILLIPS Department of Biology and WAMI Medical Education Program, Montana State University, Bozeman, Montana 59717-0346
ABSTRACT The consequences of prenatal ethanol exposure on the postnatal development of Bergmann glia and astrocytes in the rat cerebellum were investigated by using glial fibrillary acidic protein (GFAP) immunolabeling. Pregnant rats were either fed with an ethanol containing liquid diet (6.7% v/v) or pair-fed with an isocaloric diet throughout gestation. On postnatal day (PD) 15 and 22, parasagittal sections of the cerebellar vermis from female offspring were processed for GFAP immunohistochemistry to assess the development of Bergmann glia and astrocytes in lobules I, VII, and X and astrocytes in the central core of white matter. On PD 15, compared to control animals, ethanol exposed animals had fewer GFAP positive Bergmann glial fibers per unit length of molecular layer; a significantly greater percentage of morphologically immature Bergmann fibers; a significantly greater GFAP positive astrocytic area per unit area of internal granular layer and central white matter; and the astrocytic processes were wider and more closely packed. These glial changes were associated with significantly thicker external granular layer in all 3 lobules. However, no significant differences were seen between the ethanol exposed and control animals on PD 22, indicating "catch-up growth" in the ethanol exposed animals during the third postnatal week. These results suggest that prenatal ethanol exposure causes (1)delayed maturation of Bergmann glia, which in turn contributes to the delayed migration of granule cells; and (2) alterations in the normal postnatal development of aStrOCyteS. o 1992 Wiley-Liss, Inc. Key words: glial fibrillary acidic protein, radial glia, granule cell, alcohol and cerebellar cortex
Many studies have unequivocally demonstrated that gestational alcohol exposure is detrimental to brain development and function in the offspring (Streissguth et al., '78; Riley et al., '86; Streissguth, '86; West and Pierce, '86). The teratogenic effect of ethanol on the developing nervous system in experimental animals is mainly characterized by abnormalities in nerve cell proliferation, migration and maturation (Borges and Lewis, '83; Kennedy and Elliot, '86; Miller, '86, '88; Al-Rabiai and Miller, '89) as well as by delays in the acquisition of myelin (Samorajski et al., '86; Phillips, '89) and the maturation of glia (Kennedy and Mukerji, '86a,b; Phillips and Krueger, '90). Postmortem studies of children with fetal alcohol syndrome (FAS) show aberrant cytoarchitecture and neuroglial heterotopia indicative of disorganized migration of neuronal and glial elements (Clarren et al., '78; Peiffer et al., '79; Clarren, '86). Delayed migration and maturation of neurons could be due to defective development of glia particularly since radial glia, a type of astroglia, are proposed to play an important
o 1992 WILEY-LISS, INC.
role in the migration and differentiation of neurons (Rakic, '85; Lauder and McCarthy, '86; Manthorpe et al., '86). Though some studies have addressed the potential effects of alcohol on the development of glial cells in vitro (Davies and Vernadakis, '84; Kennedy and Mukerji, '86a,b; Chiappelli et al., '91; Davies and Cox, '91; Davies and Ross, '91) and in optic nerve in vivo (Phillips and Krueger, '901, there is generally a paucity of knowledge concerning the effects of in vivo alcohol exposure on the ontogenesis and maturation of astrocytes in different regions of the central nervous system. The cerebellum is a suitable region for studying neuronastroglia interactions during development because of the limited number of neuronal and astroglial lineages and because the regularly organized developing cell populations have been well characterized in terms of their morphology and the timing of their development (Altman, '69, '72; Accepted March 2 , 1992.
A.K. SHETTY AND D.E. PHILLIPS
20 Palay and Chan Palay, '74; Wilkin and Levi, '86; Berciano et al., '90). The two types of astroglia in the developing cerebellar cortex, Bergmann glia and velate astrocytes, interact differently with neurons (Ramon y Cajal, '11; Palay and Chan Palay, '74; Hatten et al., '84).The vertically oriented Bergmann glial fibers in the molecular layer are believed to play an important role in the guidance of migrating immature granule cells from the external granular layer (EGL) to the internal granular layer (IGL) (Rakic, '71; Rakic and Sidman, '73; Rakic, '85). On the other hand, velate astrocytes in the IGL are mainly responsible for the clustering of granule cell bodies and the organization of perisynaptic glia surrounding cerebellar glomeruli (Palay and Chan Palay, '74; Wilkin and Levi, '86). Many studies report evidence of alcohol induced abnormalities or delays in the migration of neurons from the EGL across the molecular layer to the IGL (Bauer-Moffett and Altman, '77; Chernoff, '77; Volk, '77; Kornguth et al., '79; Borges and Lewis, '83). Such effects could be due to defects or delays in the development and maturation of Bergmann glia. In addition, the reduced number of granular neurons in the IGL following alcohol exposure (Pierce et al., '89; Bonthius and West, '90; West et al., '90) may induce secondary changes in the velate astrocytes apart from the effects of alcohol on the genesis of these astrocytes. Despite these evidences, information regarding the effects of alcohol on the development and maturation of Bergmann glia and other astrocytes in the cerebellum is not available in the literature. Glial fibrillary acidic protein (GFAP) immunohistochemistry was used in the present investigation to examine the effects of prenatal ethanol exposure on the development of Bergmann glia, velate astrocytes in the IGL and astrocytes in the central white matter core of the rat cerebellum. GFAP is the major intermediate filament protein of Bergmann glia and astrocytes and has been widely used as a specific immunohistochemical marker for these cells in the cerebellum. Postnatal day (PD) 15 was chosen for the initial investigation, because by PD 15 (1)the GFAP positive axial filamentous cores of the Bergmann fibers are well developed and extend up to the external limiting membrane from the Purkinje cell layer; (2) GFAP is well expressed in the astrocytic processes in both the IGL and the central core of white matter (Ghandour et al., '81); and (3) the proposed Bergmann glial mediated migration of granule cells from the EGL to the IGL is active (Altman, '72). PD 22 was chosen for later analysis to determine if any changes that were present at PD 15 persisted in the maturation of Bergmann glia and granule cell migration or in the maturation of other astrocytes since by PD 22 the migration of granular neurons from the EGL to the IGL is completed (Wilkin and Levi, '86).
MATERIALS AND METHODS Animals and diet regimen Nulliparous female Sprague-Dawley rats (Holtzmann/ Sasco Co.) weighing from 220 to 300 g were caged with individual male rats for breeding in the evening. The next morning those female rats with an ejected vaginal plug (gestational day 0 ) were divided into two weight-matched groups: (1)ethanol consuming dams-these animals were given free access to liquid diet (Bioserv Inc., Frenchtown, N.J.) containing 37.5%ethanol derived calories (6.7%viv); and (2) pair-fed dams-these animals received the same
volume of liquid diet containing isocaloric amounts of maltose-dextrin in place of ethanol. A chow-fed control group was not included in this study since a number of previous studies and a pilot study in our laboratory (with the same diet from Bioserv Inc.) clearly indicated that there were no significant differences between the chow-fed and pair-fed control dams in either weight gain during pregnancy or in litter size. In addition, significant differences were not shown between the chow-fed and pair-fed control offspring in terms of body or brain weights and in terms of the appearance of developmental markers such as eye opening, righting reflex and dental eruption (Miller, '88; Gottesfeld et al., '89; Miller and Potempa, '90). All dams received fresh diet at 9.30 A.M. daily, beginning on the 0 gestational day (GD) and continuing to parturition. The dams were housed individually under constant conditions of temperature, humidity and lighting (12 hour light-dark cycle). The ethanol containing liquid diet was available ad libitum throughout the day. Each ethanol fed rat was paired by weight with a rat fed control diet and the two were pair-fed; that is, each control rat was given the same volume of liquid diet as that consumed by the weight matched ethanol fed rat on the respective gestation day. Thus, during the entire gestation period, both the ethanol fed and control dams consumed diets of identical caloric content. The blood alcohol concentration (BAC) in the ethanol exposed dams was determined from blood serum by using a Sigma diagnostic kit (#332 UV).Blood samples were taken from the tail veins of ethanol fed rats at 10 P.M. on GD16,6 hours after the start of the dark cycle when most animals have consumed a large portion of their diet. On the day of birth, neonates from each diet group were pooled, randomly culled to 8 pups per litter (4 males and 4 females), and fostered to untreated dams which gave birth within the same period, so that no pups were reared by their natural mothers.
Tissue processing On PD 15 and 22 (day of birth = PD 11, six females from each group (pair-fed control and alcohol exposed offspring) were sacrificed. Each animal was weighed, then anesthetized with ether prior to sacrifice by transcardiac perfusion with 4% paraformaldehyde in 0.1M cacodylate buffer (pH = 7.4). The brain was dissected from the cranial cavity and immersed in the same fixative for 2 hours. The fixed brains were weighed after trimming off olfactory bulbs, cerebellar flocculi and spinal cord. Cerebella were then dissected out and postfixed in the same primary fixative for 18 hours at 4°C before immersion in 0.1M phosphate buffered saline (PBS, pH = 7.4).Fifty micron thick parasagittal sections near the middle of the cerebellar vermis were cut on a vibratome, collected in 0.1M PBS and washed twice.
GFAP immunohistochemistry The sections were treated with a solution of 0.1M PBS containing 3% hydrogen peroxide and 20% methanol for 15 minutes to remove endogenous peroxidase, then washed thoroughly for at least 2 hours with 8 changes of PBS containing 0.3% Triton X-100. GFAP was labeled in freefloating sections by using a polyclonal antiserum obtained from Sigma Chemicals (St. Louis, MO). The sections were first incubated in 10% normal goat serum (GIBCO BRL, NY) for 30 minutes followed by incubation in GFAP
21
EFFECTS OF ALCOHOL ON CEREBELLAR ASTROGLIA antibody (developed in rabbit) at a dilution of 1 : l O O in 0.01M PBS containing 0.1% normal goat serum and 1% Triton X-100. Sections were incubated for 20 hours at room temperature with gentle agitation in a laboratory shaker. Following a thorough wash in PBS, the sections were incubated in peroxidase conjugated goat anti-rabbit IgG (Sigma Chemicals) at a dilution of 1500 in 0.01M PBS for 2 hours at room temperature. The sections were then washed three times in PBS and rinsed once in 0.05M Tris buffer followed by incubation for 10 minutes in a solution containing 0.05% DAB (3,3 diaminobenzidine tetrachloride) in 0.05M Tris buffer. The peroxidase reaction was completed by treating the sections for 5 minutes with 0.05M Tris buffer solution containing 0.05% DAB and 0.01% hydrogen peroxide. After another thorough wash in Tris buffer, sections were mounted on gelatin treated slides, air dried, dehydrated, cleared, and coverslipped with Permount. The immunohistochemical staining of sections from ethanol exposed and pair-fed control animals was done at the same time in parallel by using the same dilutions of primary and secondary antibody solutions, the same number of washes and the same concentrations of DAB and hydrogen peroxide. The same procedure was followed for negative control sections except that the primary antibody was omitted and the sections were incubated in normal goat serum only. Neither immunostaining nor any recognizable background staining was observed under these conditions in the negative control sections (Figs. lg,h). A few sections from each animal were counter-stained with Mayer's hematoxylin (Sigma Chemicals) to visualize the layers in the cerebellar cortex. Observations and photography were done with an Olympus photomicroscope.
Morphometric analysis Morphometric studies were performed by using a bit pad (Sigma Scan, Jandel Scientific, Corte Madera, CA) or an image analysis system (MCID, Microcomputer Imaging Device, Imaging Research Inc., St. Catharines, Ontario, Canada) on lobules I, VII, and X of cerebellar vermis as defined by Larsell ('52) and on the central core of white matter. Three sections of cerebellar vermis were evaluated for each animal. All data were collected blind to experimental codes and means were calculated for each animal individually before the means were determined for the six animals per group. Statistical comparisons were done on measurements and counts from 6 experimental and 6 control animals at each age with a Student's t-test (two tailed). The density of GFAP positive Bergmann glial fibers per unit length of folia surface in the molecular layer was determined at a consistent area in each of the 3 lobules. ) Fibers were counted in a camera-lucida image ( 1 9 3 ~by using a bit pad with a lighted cursor to measure the length of the area and to mark the counts. The fibers were characterized as either mature or immature types based on their morphology. GFAP positive astrocytic area per unit area of tissue was measured by MCID densitometric analysis in the IGL of three vermal lobules and in the central core of the white matter. The microscopic image was transferred to the videoscreen by focussing the appropriate area in the immunostained section with an Olympus microscope equipped with a video camera. The same intensity of light was used to focus all the samples measured. The densitometric program has a continuous gray scale composed of 255 levels with the
densest gray level assigned the lowest number and the lightest level assigned the highest number. The background and target values were set to 255 and 172, respectively, following digitization of the original gray value image in the videoscreen. Those values were determined by selecting the background and target areas in several sections from both ethanol exposed and pair-fed control animals before commencing the measurements on coded slides for statistical analysis. This scale eliminated the background staining completely and retained all the target (GFAP positive) structures in the range 0-172. The binary image of GFAP positive elements was then generated by selecting a suitable threshold value (which varied from 215 to 225) to include all the GFAF' positive structures without any background. The final binary image was cross-checked with the original gray value image by alternating the two images on the videoscreen. Finally, the image was frozen and the area occupied by the GFAP positive structures in the field (0.04288mm2in area) was measured by scanning the whole field on the videoscreen. The data was then stored for further calculations and statistical analysis. The cross-sectional area of the lobules was measured on a bit pad by tracing the camera-lucida image of the lobules (37x1 in immunocytochemically stained parasagittal sections of the vermis. Each lobule was marked uniformly (Pierce and West, '87) and three sections were measured for each animal. The thickness of the EGL was measured on the bit pad (940 x 1 by taking three measurements at a consistent region in each of the three lobules from hematoxylin counter-stained midvermal sections in every PD 15 animal. The thickness of the EGL was not measured in PD 22 animals because the EGL in the same regions measured on PD 15 was either not present or was only one cell thick in both the ethanol exposed and pair-fed control animals.
RESULTS The weight gain by ethanol consuming rats during pregnancy was not significantly different than that of pair-fed control animals. From GD1 to GD20, the ethanol fed rats averaged 39.9%in weight gain compared to 35.9% in the pair-fed control animals ( P < 0.3). The mean daily ethanol consumption in the ethanol exposed dams was 10.9 0.6 gikg body weight during GD 6-20, while the average BAC on GD16 was 139.8 43.7 mgidl (n = 6). The brain weights of the pups exposed to ethanol prenatally were slightly less than the pair-fed control pups at both ages studied but the differences were not significant statistically (6% less on PD 15, P < 0.06 and 4%less on PD 2 2 , P < 0.1).
*
*
GFAP immunoreactivity Immunostaining with the antibody directed against GFAP resulted in labeling of Bergmann glial fibers in the molecular layer and astrocytes in the internal granular layer and white matter of the cerebellums in both ethanol exposed and pair-fed control animals. The intensity of staining was greater in the Bergmann glial fibers and astrocytic processes compared to the somata of the cells. Morphology o f Bergmann glial fibers. In both ethanol exposed and control offspring, GFAP positive Bergmann glial fibers appeared as regularly spaced parallel palisades oriented radially in the molecular layer, extending from the upper limit of the Purkinje cell layer towards the pial
22
Fig. 1. Immunoreactivity for glial fibrillary acidic protein (GFAP) in the cortex of cerebellar vermis on PD 15. a, c, and e are from lobules I, VII, and X, respectively, of ethanol exposed animals. b, d, and f are from lobules I, VII, and X, respectively, of pair-fed control animals. Note the vertically oriented Bergmann glial fibers in the molecular
A.K. SHETTY AND D.E. PHILLIPS
layer (ML) and astrocytic processes in the internal granular layer (IGL). The thicker external granular layer (arrows) and fewer Bergmann fibers are clearly evident in lobules VII (c) and X (el of ethanol exposed animals compared to control animals (d,f). x 170.
EFFECTS OF ALCOHOL ON CEREBELLAR ASTROGLIA
23
Fig. 1 (continued). g,h: Photomicrographs of negative control seetions of cerebellum used in the GFAP immunohistochemicalprocedure. Neither specific immunostaining nor non-specific background staining are seen. The different layers of cerebellar cortex and central white
matter core are recognizable. EGL = external granular layer; ML = molecular layer; IGL = internal granular layer; WM = white matter; CWM = central white matter core of cerebellum. x 185.
surface where the glial endfeet formed a continuous glial limiting membrane (Fig. la-f). Two types of GFAP positive Bergmann fibers could be characterized in both groups on PD 15: (1)thin and smooth fibers with club shaped glial end feet (growth cones) terminating either in the upper part of molecular layer or lower half of the EGL; (2) thick, smooth, and slightly twisted fibers with end feet either touching or ending close t o the external limiting membrane (ELM).The thin fibers appeared to be new filiform processes growing
towards the surface of the cerebellum (Das, '76; Shiga et al., '831, whereas the thick fibers with variable end feet (conical, club shaped or flattened) appeared more like twisted ribbons than the fine caliber filiform processes. Since it is known that mature GFAP positive Bergmann fibers extend to the ELM and that the expression of GFAP gradually increases during postnatal development (hence thickness of the axial core of these fibers, Ghandour et al., '811, we classified the thin fibers that did not extend to the ELM as
24
Fig. 2. Magnified view of GFAP immunoreactivity in the cerebellar cortex on postnatal day (PD) 15. a and b show GFAF' positive Bergmann glial fibers in an ethanol exposed animal and a pair-fed control animal, respectively. Note that there are fewer Bergmann fibers in the ethanol exposed animal than in the pair-fed control animal. Arrows indicate fine caliber (immature) Bergmann fibers in the external granular layer (EGL) falling short of the external limiting membrane while arrowheads indicate the thicker and slightly twisted (more mature) Bergmann fibers terminating in or close to the external limiting membrane. A comparatively greater number of immature
A.K. SHETTY AND D.E. PHILLIPS
GFAP positive Bergmann fibers are evident in the ethanol exposed animal (a).ML = molecular layer. c and d show GFAP immunopositive structures in the internal granular layer of an ethanol exposed animal and a control animal, respectively. In the ethanol exposed animal, the cell bodies of astrocytes (arrows) are stained and astrocytic processes are thicker, and greater in number whereas in the control animal only faintly stained thinner astrocytic processes without somata are obvious. ~ 3 4 0 .
25
EFFECTS OF ALCOHOL ON CEREBELLAR ASTROGLIA immature types and the thick fibers that end either in or close to the ELM as mature types. Based on these criteria, the qualitative differencesapparent between the two groups of animals on PD 15 were subtle although detailed evaluation of the two groups indicated that there was a relatively greater number of immature fibers than mature fibers in the ethanol exposed animals (Fig. 2a,b). Counts of the immature and mature fiber types revealed that the ethanol exposed animals had a significantly greater percentage of immature fibers than control animals on PD 15 in all three lobules studied (Table 1). No differences were seen in the morphology of individual GFAP positive Bergmann glial fibers on PD 22. The maturation of Bergmann fibers in both ethanol exposed and control animals appeared similar (Fig. 3). Virtually all of the fibers were of the mature type in both groups so relative counts of mature and immature types were not done.
Number of Bergmann glial fibers. There were fewer fibers per unit length in ethanol exposed animals on PD 15 in all three lobules studied (Fig. 4).The differences were
Fig. 3. GFAF' immunoreactivity in the PD 22 cerebellar cortex (Lobule X) of an ethanol exposed animal (a)and a pair-fed control animal (b).Thick and mature Bergmann glial fibers with end feet (arrows) touching the external limiting membrane are structurally
similar in both groups. ML = molecular layer; IGL = internal granular layer. The external granular layer is missing in these particular examples in both groups. x 185.
TABLE 1. Percentage (%) of Mature and Immature Bergniann Fibers in Different Lobules on Postnatal day 15'
Region of cerebellar vermis Lobule I Mature fibers Immature fibers Lobule VII Mature fibers Immature fibers Lobule X Mature fibers Immature fibers
ETOH exposed (n = 6)
Control (n = 6)
45.3 (11) 54.7 (11)
81.4 (9.3) 18.6 19.3)
