Acta Neuropathol (1997) 93 : 450– 460
© Springer-Verlag 1997
R E G U L A R PA P E R
John W. Conlee · Steven M. Shapiro
Development of cerebellar hypoplasia in jaundiced Gunn rats: a quantitative light microscopic analysis
Received: 1 March 1996 / Revised, accepted: 16 September 1996
Abstract The homozygous (jj) Gunn rat provides a model for hyperbilirubinemia which includes prominent cerebellar hypoplasia. Development of the Gunn rat cerebellum was examined with and without the additional effects of elevating brain bilirubin concentration to still higher levels via sulfadimethoxine (sulfa) administration. Homozygous (jj) Gunn rats and heterozygous (Nj) littermate controls (n = 32 each) were given 100 mg/kg sulfa or saline at postnatal days 3, 7, 17, and 30, and most were sacrificed 24 h later (n = 4 for each genotype at each age). Cerebellar volume, total volume and cell number for each deep cerebellar nucleus, densities for Purkinje and granule cells in the cerebellar cortex of lobules II, VI and IX, and the density of vacuolated Purkinje cells were all measured quantitatively. Cytoplasmic vacuolation provided an indication of bilirubin toxicity and was never observed in the Nj control rats. Vacuolated Purkinje cells were first observed in jj-saline rats at 18 days and were found only in the more anterior lobules of the cerebellum (II and VI). By contrast, vacuolated Purkinje cells were observed in jjsulfa rats at both 4 and 8 days, but only in the most posterior cerebellar lobule (IX). In all older jj rats, the decline in vacuolation was accompanied by significant necrosis and resorption of the Purkinje cells in the anterior lobules. Since the Purkinje cells in the posterior lobules are the first to differentiate in the cerebellum and are resistant to bilirubin toxicity in jj-saline rats, the results support the
J. W. Conlee Department of Veteran Affairs Medical Center, Salt Lake City, UT 84148, USA J. W. Conlee Department of Neurobiology and Anatomy, University of Utah School of Medicine, Salt Lake City, UT 84132, USA S. M. Shapiro (Y) Division of Child Neurology, Department of Neurology, Medical College of Virginia, Virginia Commonwealth University, Box 980211, Richmond, VA 23298–0211, USA Tel.: 1-804-828-7416; Fax: 1-804-828-5654; e-mail:
[email protected]
presence of a critical period when elevated brain bilirubin may be most toxic to neuronal development. The findings suggest that neurons undergoing differentiation at the time of bilirubin exposure are most susceptible to cell death, while cells that are slightly more or slightly less mature may show only transient changes. Key words Hyperbilirubinemia · Bilirubin encephalopathy · Kernicterus · Cerebellum · Purkinje cells
Introduction Despite the prevalence of neonatal jaundice and its risk of permanent brain damage, bilirubin neurotoxicity remains poorly understood [22]. An animal model often used to study the effects of hyperbilirubinemia is the Gunn rat [7, 10]. Homozygous (jj) Gunn rats have an inherited deficiency of hepatic UDP-glucuronyl transferase which produces an abnormal elevation in serum bilirubin and visible jaundice within 6 h after birth [10, 27]. Cerebellar hypoplasia is the most prominent abnormality seen in the Gunn rat, and the Purkinje cells are reported to be affected almost exclusively [20, 24]. Furthermore, the most severely damaged regions of the cerebellum have been shown to be the more anterior lobules, with the most posterior lobules showing the greatest resistance to bilirubin toxicity [12]. These differences in susceptibility correspond to the various degrees of Purkinje cell maturation during bilirubin exposure and have led to the notion that bilirubin-induced cerebellar hypoplasia is developmentally regulated [19]. Neural degeneration in the Gunn rat is first heralded by the formation of vacuoles and membranous cytoplasmic inclusions in the cerebellar Purkinje cells [21, 23]. Cerebellar abnormalities have also been reported to underlie some of the initial clinical symptoms of kernicterus in children [9, 14]. However, cerebellar hypoplasia has not been observed consistently in either clinical or experimental hyperbilirubinemia, nor have prominent cytoplasmic inclusions been reported elsewhere other than in the
451
Gunn rat [17]. This has raised questions regarding the generality of the neurotoxic effects in the Gunn rat cerebellum and their significance in the pathogenesis of bilirubin encephalopathy. The Gunn rat is the only model known to develop bilirubin encephalopathy spontaneously, and thus differs from all other models which rely upon artificial means of postnatal induction [27]. Experimental hyperbilirubinemia in other species must often be combined with anoxia to produce the acute necrosis and neuronal cytolysis characteristic of human kernicterus [4, 5, 15, 18]. Such lesions have been produced in the Gunn rat without anoxia when brain bilirubin levels were elevated acutely with sulfonamide [17]. Thus, both patterns of neuronal degeneration have been observed in the Gunn rat, and appear in response to differing intensities and durations of bilirubin exposure. The cerebellum of the jaundiced Gunn rat may thus provide a key to understanding bilirubin toxicity in the human brain. In the present study, the development of the cerebellum was examined in homozygous (jj) Gunn rats and heterozygous (Nj) Gunn rat littermate controls with and without sulfadimethoxine (sulfa) administration. Sulfa displaces bilirubin from its albumin binding sites, resulting in a net transfer of bilirubin from the serum into the brain and produces an acute encephalopathy [5, 17, 21]. Although several studies have examined the effects of sulfa on the Gunn rat cerebellum [19–21], none have compared the effects of bilirubin toxicity on the postnatal development of cerebellar lobules known to differ in their onset of Purkinje cell differentiation. Specifically, no other study has quantitatively compared the effects of sulfa administration on the developmental time course of Purkinje cell hypoplasia in both the anterior and posterior lobules of the cerebellum. Such an investigation could help to identify the role that bilirubin plays in the developmental expression of cerebellar abnormalities in the Gunn rat.
Materials and methods Animals and treatment A total of 32 homozygous jaundiced (jj) and 32 heterozygous nonjaundiced (Nj) Gunn rats were obtained from jj male × Nj female matings in our Gunn rat breeding colony at the Medical College of Virginia, Virginia Commonwealth University, and studied at 4, 8, 18, and 31 days of age. Pregnant females were checked daily for new litters; 8 jj and 8 Nj control rats were examined at each age. Half of the animals in each age group were administered a single intraperitoneal injection of 100 mg/kg sulfadimethoxine (sulfa) diluted in normal saline (Albon, Hoffmann-La Roche, Nutley, N.J.); the other half received an equal volume of normal saline. The 16 animals in each age group were divided into four sets; each set of 4 littermates contained a single Nj-sulfa-, Nj-saline-, jj-sulfa-, and jj-saline-injected rat. To reduce effects due to interlitter variation, each set of 4 littermates was derived from a different litter for the 18-day and 31-day age groups; the 4-day-old animals came from three litters, while the 8-day animals came from two litters. The sulfa or saline injection was given 24 h before sacrifice to rats at postnatal days 3, 7, and 17, and to one set of the 30-day-old rats. The remaining three sets of 31-day-old rats were injected at 17 days, but killed at 31 days because of our previous observation that
sulfa injection into 30-day-old jj rats produces no behavioral or electrophysiological abnormalities (S. Shapiro, personal observation). Animal protocols used in this study were approved by the Medical College of Virginia Animals Care and Use Committee and followed guidelines established by the National Institutes of Health.
Histology and morphometry At the specified ages, animals were anesthetized with methoxyflurane gas and perfused transcardially with phosphate-buffered saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Brain stems were embedded in paraffin and serial sections were cut at 10 µm in the transverse plane; sets of every 10th (for the 4 and 8 day rats) or 20th (for the 18 and 31 day rats) section were mounted on subbed slides and stained with 0.05% thionin. Prior to the quantitative analysis, all slides were coded to avoid observer bias. The effects of hyperbilirubinemia and sulfa on the postnatal development of the cerebellum were analyzed using a computerized image analysis system (BioQuant Advanced Image Analysis System, R & M Biometrics, Nashville, Tenn.). The following anatomical features were examined morphometrically: (1) total cerebellar volume, (2) total volume for each main subdivision of the deep cerebellar nuclei, (3) total neuron number in each subdivision of the deep cerebellar nuclei, (4) Purkinje cell density in each of three regions of the cerebellar cortex, (5) density of Purkinje cells containing vacuoles and other abnormal cytoplasmic inclusions, and (6) granule cell density for corresponding regions of the cerebellar cortex in which a Purkinje cell density was determined. The total volume of the cerebellum and the total volume of each of the deep cerebellar nuclei were found using similar quantitative procedures. Estimates of total cerebellar volume were determined by tracing the perimeter of the cerebellum contained within each section at a final magnification of × 40. The area (in µm2) of the cerebellum was summed across all sections containing the cerebellum and multiplied first by the section thickness (i.e., 10 µm), and then by the sampling interval (i.e., 10 or 20 depending upon the age). This yielded the total cerebellar volume (in µm3) which was converted to mm3. Estimates of total volume for each of the deep cerebellar nuclei were found in the same manner, but their perimeters were traced at a final magnification of × 100. The deep cerebellar nuclei were subdivided cytoarchitectonically into medial, posterior and anterior interposed, and lateral subdivisions [23], with the posterior and anterior components of the interposed nucleus combined and treated as a single deep cerebellar nucleus. The total number of cells in each of the deep cerebellar nuclei was estimated from the density of neurons in each subdivision. Neuron density was estimated in each section by counting the number of cells containing a visible nucleus within a 5000-µm2 area of each subdivision. The neuron density (cells/µm2) was then multiplied by the area (µm2) of the subdivision to yield the total number of cells in each cerebellar nucleus for that section. The number of cells per section was summed across all sections containing each subdivision, and this value was multiplied by the sampling interval (i.e., 10 or 20) to estimate the total number of cells present in each of the deep cerebellar nuclei. These values were not corrected for split cell errors, which occur when structures have been cut near the surface of the section and are present for counting in two adjacent sections [13]. However, relative differences in total cell number were interpreted to reflect true differences between the age groups and treatment conditions. Abnormalities in the cerebellar cortex were examined by measuring the density of Purkinje and granule cells, as well as the density of Purkinje cells containing vacuoles [24]. Vacuoles are considered to be a strong indicator of Purkinje cell degeneration due to bilirubin toxicity [20], and they appeared in the cytoplasm as large balloon-like swellings, as well as small vesicular inclusions. Thus, the Purkinje cell density was used to estimate the observable number of Purkinje cells present, both normal and vacuolated, while the density of vacuolated Purkinje cells provided an estimate of the
452 number of cells undergoing degeneration at a given age. Densities for both the Purkinje and granule cells in the cerebellar cortex were determined in sections through the vermis of lobules II, VI, and IX [25]. In the two younger age groups, Purkinje cells containing a nucleus were counted at a final magnification of × 1000 over four consecutive sample lengths of 150 µm each. In the two older age groups, such cells were counted at a final magnification of × 400 over four sample lengths of 400 µm each. The number of vacuolated Purkinje cells was also determined in each sample. Across all age groups, cell density in the granule cell layer was found for the region immediately below the Purkinje cell layer by counting the number of granule cells at a final magnification of × 1000 in four separate sample areas of 1000 µm2 each. For each lobule, the average density of both the Purkinje (cells/µm) and the granule (cells/µm2) cells was determined by averaging across the four separate samples of each cell type.
cerebellar volume in both groups of jj animals was markedly smaller than in either group of Nj animals (Fig. 1). A one-way multivariate analysis of variance demonstrated a significant main effect for group (Rao’s R = 3.121; df =
For comparison of total cerebellar volumes, a multivariate analysis of variance was performed using four between group variables (Njsulfa, Nj-saline, jj-sulfa, jj-saline) and four age-dependent variables (4, 8, 18, and 31 days). Nuclear volumes, absolute cell numbers, and cell densities were compared using group (Nj-sulfa, Nj-saline, jjsulfa, jj-saline) × age (4, 8, 18, 31 day) × structure three-way analyses of variance (ANOVA) with repeated measures on the latter factor. The multivariate analysis was used to compare multiple dependent variables without repeated measures. The three-way ANOVAs compared multiple dependent variables in which the anatomical variables were specified as a repeated measures factor. The F statistic used in both analyses is extremely robust and will permit even major violations to the assumption of homogeneity of variance without compromising its validity [28]. Thus, possible violations to this assumption due to the relatively small sample size (four animals/condition) were not expected to invalidate the analysis. All post hoc comparisons were conducted using the Scheffé test with alpha set at P < 0.05.
Results
Nuclear volume (mm3 ± SEM)
Statistical analysis
Total cerebellar volume (mm3 ± SEM)
Total cerebellar volume was similar for all groups of Nj and jj rats at both 4 and 8 days of age (Fig. 1). At 18 days, mean cerebellar volume in both groups of jj animals was smaller than that in the Nj rats, although this difference was not statistically significant. By 31 days of age, mean
Age (days) Age (days)
Fig. 1 Mean total cerebellar volume for each treatment group of Nj and jj rats at each age studied
Fig. 2 Mean volume for lateral (A), interposed (B), and medial (C) subdivisions of the deep cerebellar nuclei for all Nj and jj rats at each age studied
Total cell number (× 103 ± SEM)
453
Age (days)
Fig. 3 Mean total cell number in lateral (A), interposed (B), and medial (C) subdivisions of the deep cerebellar nuclei for all Nj and jj rats at each age studied
12,24; P < 0.0085). Post hoc analyses showed that the mean cerebellar volume in the jj-sulfa rats was significantly smaller than in either the Nj-saline (P < 0.006) or the Njsulfa (P < 0.022) rats.
The absolute volume and total neuron number were significantly reduced in each of the deep cerebellar nuclei of the jj rats (Figs. 2, 3). Data from both the saline- and sulfa-treated animals were similar within each group of Nj and jj rats, and thus were combined. The volumes for each of the deep cerebellar nuclei were comparable in the Nj and jj rats at 4, 8, and 18 days of age, but were smaller in the 31-day jj animals (Fig. 2). A three-way ANOVA demonstrated a highly significant main effect for group (F = 32.533; df = 1,56; P < 0.001), and a significant group by age interaction (F = 13.050; df = 3,56; P < 0.001). Post hoc tests revealed that the total volume of the deep cerebellar nuclei, averaged across all three subdivisions, was only significantly smaller (P < 0.001) in the 31-day jj rats. The mean number of neurons in each of the deep cerebellar nuclei was comparable between groups at 4 days of age, but was reduced in all of the older jj rats (Fig. 3). A three-way ANOVA showed a significant main effect for group (F = 9.525; df = 1,56; P < 0.003). Post hoc analysis showed that this was a significant reduction when averaged across all the ages studied (P < 0.003). In the cerebellar cortex, Nj and jj rats both showed a similar sequence of developmental changes, but only the jj rats displayed significant vacuolation and degeneration of the Purkinje cells. At postnatal day 4, all regions of the cerebellar cortex contained a greatly thickened external granule cell layer, a thin molecular layer, a Purkinje cell layer two to four cells deep, and a sparsely populated internal granule cell layer (Figs. 4, 5). Vacuolated Purkinje cells were observed in two of the jj-sulfa rats at this age, but only in lobule IX. At postnatal day 8, Purkinje cells mostly formed a monolayer in lobule IX; by contrast, those in the more anterior lobules had not matured much beyond the Purkinje cells observed in the 4-day-old animals (Figs. 4, 5). Purkinje cell vacuoles were observed in all lobules of all the 8-day-old jj-sulfa rats (Fig. 4D); only one jj-saline rat showed vacuolization at this age, and only in lobule IX. At postnatal day 18, a monolayer of Purkinje cells had formed in all lobules; however, a thin layer of external granule cells was still present in lobules II and VI (Figs. 4E, F; 5E, F). At this age, vacuolated and degenerated Purkinje cells were abundant across all lobules in the jj-sulfa rats, and in lobules II and VI of the jjsaline rats (Figs. 4F, 5F). By day 31, neither group of Nj or jj rats showed an external granule cell layer and the cerebellum appeared fully mature (Figs. 4G, H; 5G, H). Vacuolated and degenerated Purkinje cells were observed in all lobules of the jj-sulfa rats, and in lobules II and VI of the jj-saline animals (Figs. 4H; 5H). The Nj and jj rats both showed a highly significant decrease in Purkinje cell density between 4 and 31 days of age. However, the Purkinje cell density in the jj rats was abnormally reduced as a function of both age and position in the cerebellum. Since Purkinje cell density did not differ in the animals given sulfa or saline, this parameter was combined within groups for further analysis. Purkinje cell density (cells/mm) in lobules II, VI, and IX of both the Nj and jj rats averaged 141 cells in the 4-day animals, and 80 cells in the 8-day rats (Fig. 6A). Statistical comparisons in
454
Fig. 4 A–H Bright-field photomicrographs of cerebellar cortex in the vermis of lobule IX. A, B Four-day Nj- sulfadimethoxine (sulfa) (A) and jj-sulfa (B) rats. Cerebellar cortex was largely unaffected in jj rats. C, D Eight-day Nj-sulfa (C) jj-sulfa (D) rats. Inset shows Purkinje cell layer at higher magnification. Purkinje cells are highly vacuolated in jj-sulfa, but not Nj-sulfa rats. E, F Eighteenday Nj-sulfa (E) and jj-sulfa (F) rats. Note vacuolization and necrosis in PCL and thinner ML in jj rats. G, H Thirty-one-day Njsulfa (G) and jj-sulfa (H) rats. PCL and ML are comparable between Nj and jj rats in lobule IX. (EGCL external granule cell layer, ML molecular layer, PCL Purkinje cell layer, GCL granule cell layer). A–H × 240; insets in C, D × 520
the 4- and 8-day-old animals did not reveal any differences in Purkinje cell density between the Nj and jj rats. In the 18- and 31-day animals, the density of Purkinje cells averaged 39 cells across the lobules in the Nj rats and 30 cells in the jj rats (Fig. 6B, C). The density of Purkinje cells in the jj rats was significantly reduced only for lobules II and VI of the cerebellum; Purkinje cell density in lobule IX was unaffected (Fig. 6B, C). The results of the three-way ANOVA comparing the 18- and 31-day Nj and jj rats yielded a significant main effect for group (F =
455
Fig. 5 A–H Bright-field photomicrographs of cerebellar cortex in the vermis of lobule VI. A, B Four-day Nj-sulfa (A) and jj-sulfa (B) rats. Cerebellar cortex was unaffected in jj rats. C, D Eight-day Njsulfa (C) and jj-saline (D) rats. Cerebellar cortex in jj-saline rats appeared normal, without vacuolation. E, F Eighteen-day Nj-sulfa (E) and jj-sulfa (F) rats. Note vacuolation and necrosis in PCL of jj rat. G, H Thirty-one-day Nj-sulfa (G) and jj-saline (H) rats. Note near complete degeneration of the PCL and greatly thinned ML in jj rat. (EGCL external granule cell layer, ML molecular layer, PCL Purkinje cell layer, GCL granule cell layer). A–H × 240
14.489; df = 1,28; P < 0.001), and a significant group by lobule interaction (F = 8.070; df = 2,56; P < 0.001). The density of vacuolated Purkinje cells was found to depend upon the treatment group, age, and lobule of the cerebellum (Fig. 7). A three-way ANOVA demonstrated a highly significant main effect for group (F = 19.287; df = 3,48; P < 0.001). Post hoc analyses indicated that the density of vacuolated Purkinje cells across lobules was significantly greater in the jj-sulfa rats than in the jj-saline animals, or in either group of Nj rats (P < 0.001 for each
456
Purkinje cell number/mm (± SEM)
Vacuolated Purkinje cells/mm (± SEM)
Lobule IX
Lobule VI
Lobule II
Age (days)
Fig. 7 Mean density of vacuolated Purkinje cells plotted as a function of age and treatment group in cerebellar lobules IX (A), VI (B), and II (C)
Cerebellar lobule
Fig. 6 Mean Purkinje cell density in lobules II, VI, and IX of Nj and jj rats at postnatal days 4 and 8 (A), 18 (B), and 31 (C). Note that the abscissa changes between A and B
group). Highly significant two-way interactions of group and age (F = 5.905; df = 9,48; P < 0.001), and three-way interactions between group, age, and lobule (F = 3.244; df = 18,96; P < 0.001) were also shown. In the jj-saline rats, vacuolated Purkinje cells were seldom observed at 4 and
8 days, increased by 18 days, and declined to near zero by 31 days of age. In the 18-day jj-saline rats, vacuoles were largely confined to lobules II and VI. By contrast, vacuolated Purkinje cells were most prevalent in the jj-sulfa rats near 8 days (Fig. 7). When sulfa was administered on day 7, there was a highly significant increase in the density of vacuolated Purkinje cells in lobules II (Fig. 7C) and VI (Fig. 7B), as well as the first appearance of vacuolated cells in lobule IX (Fig. 7A). Thereafter, sulfa administration did not significantly increase the degree of vacuolation over the saline-injected jj rats, including the 31-day
457 Lobule 9
Granule cell density/1000 µm2 (± SEM)
% Vacuolated Purkinje cells/mm (± SEM)
Lobule 9
Lobule 6
Lobule 6
Lobule 2
Lobule 2
Age (days)
Fig. 9 Mean density of granule cells plotted as a function of age and genotype in the internal granule cell layer of lobules IX (A), VI (B), and II (C) Age (days)
Fig. 8 Mean percent of vacuolated Purkinje cells per millimeter of cerebellum plotted as a function of age and treatment group in lobules IX (A), VI (B), and II (C)
rats injected either 1 or 14 days before sacrifice. Purkinje cell vacuoles were never observed in the Nj rats. Since Purkinje cell density declined with age in all animals (see Fig. 6), the lower density of vacuolated cells over time may have simply reflected the smaller number of surviving cells. This was examined by plotting the percentage of vacuolated Purkinje cells relative to the total number present per millimeter of tissue (Fig. 8). These results showed that about 70% of the Purkinje cells were vacuolated in all lobules of the jj-sulfa rats at 4 and 8 days, while very few vacuoles were observed in the jj rats
given saline. In the 18-day jj-sulfa rats, vacuoles were observed in 35% of the Purkinje cells in lobule IX (Fig. 8A) and 80% of the Purkinje cells in lobules II and VI (Fig. 8B, C). However, 65% of the Purkinje cells in lobules II and VI of the 18-day jj-saline rats were also vacuolated, indicating that sulfa administration augmented vacuolation by only about 15%. At 31 days, about 80% of the Purkinje cells were vacuolated in lobule II of the jj-sulfa rats, compared to 30% in the jj-saline rats (Fig. 8C). In lobule VI, the percentage of vacuolated Purkinje cells had decreased to 25% in both the sulfa- and saline-treated jj animals (Fig. 8B). In lobule IX, 25% of the Purkinje cells were vacuolated in the 31-day jj-sulfa rats, while the jjsaline rats had none (Fig. 8A). Thus, the significant decrease in Purkinje cell density seen in lobules II and VI of the 18- and 31-day-old jj-sulfa rats was accompanied by a
458
corresponding decrease in the percentage of vacuolated Purkinje cells only in lobule VI. The percentage of vacuolated Purkinje cells in lobule II actually increased from day 8, until nearly all were vacuolated at 31 days. The density of granule cells in the internal granule cell layer did not differ significantly between the treatment groups at any age studied (Fig. 9). The density of granule cells increased uniformly between 4 and 31 days, and was similar in both the jj and Nj rats.
Discussion The results demonstrate that cerebellar hypoplasia in the jj Gunn rat begins largely after 8 postnatal days. This observation is consistent with earlier studies, which identified 6–10 days after birth as the critical period for bilirubin-induced cerebellar hypoplasia [11, 12, 19, 24]. In the present study, a developmental arrest in the total volume of the cerebellum was accompanied by significant reductions in the volume and number of neurons in the deep cerebellar nuclei, and by a marked reduction in the density of Purkinje cells in the cerebellar cortex. The volume of the deep cerebellar nuclei was significantly reduced only by 31 days, and the decrease in total neuron number was significant only when averaged across all ages studied. These findings suggest that the effects of bilirubin on the deep cerebellar nuclei may be secondary to the decrease in Purkinje cell density, which was shown to be both age and position dependent; at 4 and 8 days after birth, both Nj and jj rats had similar numbers of Purkinje cells. However, at 18 and 31 days the jj rats demonstrated significantly fewer Purkinje cells in lobules II and VI, while lobule IX remained unaffected. Others have also observed the absence of abnormalities in lobule IX, and have suggested that the earlier maturation of the Purkinje cells in this region may underlie its resistance to bilirubin toxicity [12]. On the other hand, the Purkinje cells in the more anterior lobules (II and VI) differentiate several days later [1], and this may play a key role in their susceptibility to hyperbilirubinemia. Thus, the pronounced cerebellar hypoplasia found in the Gunn rat could be related to bilirubin exposure during a critical phase of Purkinje cell development [19]. In addition to Purkinje cell degeneration, the most striking histological feature of the Gunn rat cerebellum is the vacuolation and extreme ballooning of the Purkinje cell cytoplasm [20]. The translucent vacuoles seen in the present study were observed in paraffin sections, giving the Purkinje cells a “moth-eaten” appearance. However, in studies in which lipid extraction was minimized, such vacuoles were shown to contain membranous bodies which were possibly produced by the interaction of bilirubin with the newly formed cell membranes of developing Purkinje cells [20, 22]. The formation of vacuoles during bilirubin exposure is believed to damage Purkinje cells and cause the death of those most severely affected. The quantitative effects of hyperbilirubinemia on the proportions of abnormal Purkinje cells in jj rats given
saline or sulfa strongly agree with the results of an earlier study by Schutta and Johnson [21]. That study and the present one demonstrate that the formation of Purkinje cell vacuoles and membranous cytoplasmic inclusions is enhanced in jj rats given sulfa. This suggests that while spontaneous hyperbilirubinemia induces the formation of Purkinje cell vacuoles, this process is acutely increased by the elevation of brain bilirubin with agents, e.g., sulfa, which displace bilirubin from blood into brain tissue. Sulfa has been shown to increase the number of degenerating Purkinje cells as well as the number of abnormal membranous cytoplasmic inclusions [23], indicating that the formation of cytoplasmic inclusions may underlie Purkinje cell degeneration [17]. However, since Purkinje cell abnormalities are not observed in untreated jj rats before postnatal day 3 and actual necrosis of the Purkinje cells is not observed before day 10 [20], vacuolation and Purkinje cell degeneration appear to be protracted processes which are accelerated by sulfa administration. Animals in the present study were given sulfa, which displaces bilirubin into the brain from its serum albumin binding sites [3, 17]. Sulfa did not increase the number of degenerated Purkinje cells 24 h after administration, but did influence the degree of Purkinje cell vacuolation in the jj animals. This vacuolation is believed to be due to the release of bound bilirubin by sulfa because increased serum bilirubin levels resulting from hemolysis or from novobiocin injection also produce vacuoles and cytoplasmic inclusions in the Purkinje cells [24, 26]. Conversely, these changes are prevented by phototherapy, which reduces the levels of serum and brain bilirubin [11, 12]. In the present study, sulfa injections on day 7 induced significant vacuolation in all cerebellar lobules including lobule IX, which is usually unaffected in the Gunn rat. Thus, a period of increased susceptibility for vacuolation was revealed at day 7, with sulfa administration prior to this time having much less effect. Sulfa-induced vacuolation either remained high or increased still further in the more anterior lobules after day 8, but decreased sharply in lobule IX. By contrast, jj rats given saline showed a peak in vacuolation near day 18, which would indicate that their period of greatest susceptibility most likely occurs between 8 and 18 days of age. All studies agree that the anterior lobules of the Gunn rat cerebellum are the most severely affected. The present results further show that vacuolation in lobules II and VI becomes most severe between 8 and 18 days of age and is accompanied by degeneration of up to half of the Purkinje cells. Sulfa produced the greatest degree of vacuolation when administered on day 7. This corresponds to the critical period that others have described both for the prevention of bilirubin-induced degeneration of Purkinje cells using phototherapy [12], and for the production of Purkinje cell degeneration during transient hyperbilirubinemia induced by novobiocin [11]. Dendritic outgrowth in the Purkinje cells normally begins around days 6–7 in lobule II and days 8–9 in lobule VI [1, 6]. Thus, there is considerable overlap between the time when bilirubin-induced vacuolation first begins, and the time when the Purkinje
459
cells begin to elaborate dendrites and receive presynaptic innervation. It has been suggested that the binding of bilirubin to lipid membranes could disrupt the formation of the Purkinje cell membranes and result in the production of membranous cytoplasmic inclusions and vacuoles as part of the degenerative process [24, 26]. Several studies have reported that the most posterior lobules of the cerebellum (lobules IX and X) are immune to hypoplasia in the Gunn rat [12, 19, 20]. Since these lobules are known to develop before the more anterior ones [1, 6], it was concluded that the more mature regions of the cerebellum are resistant to bilirubin toxicity. The present results further point toward a developmental dependence of cerebellar hypoplasia in hyperbilirubinemia. Potentiation of Purkinje cell vacuolation by sulfonamides shows that elevated levels of brain bilirubin are capable of producing abnormalities near day 8 in all regions of the cerebellum, including lobule IX. However, Purkinje cell vacuolation in the jj rats given saline appears to begin in lobules II and VI only by around day 8, and to peak near day 18 when about half of the Purkinje cells have degenerated. Since the Purkinje cells in lobule IX normally initiate dendritic outgrowth by days 4–5, these cells are more functional before the onset of bilirubin toxicity. This level of maturity may protect them from the toxic effects of bilirubin, thus preventing vacuolation and degeneration in lobule IX of the untreated Gunn rat. The susceptibility of Purkinje cells to vacuolation in lobule IX of the 8 day jj rats given sulfa suggests that the posterior cerebellum would show more hypoplasia if brain bilirubin levels were higher during the first postnatal week. Total serum bilirubin has been shown to increase rapidly in the Gunn rat within the first 24 h after birth to achieve levels which are eight to ten times greater than normal [19, 21]. This elevation is maintained over a 3week period, during which time the brain bilirubin level decreases from an approximately eightfold elevation to about fourfold. Yet, the eightfold increase in brain bilirubin during the first postnatal week does not appreciably affect lobule IX. Thus, the absence of significant vacuolation in the 4 day sulfa rats may be related more to the timing of the exposure to elevated bilirubin. Since Purkinje cells in lobule IX undergo dendritogenesis during days 4–5 [6], an earlier sulfonamide injection on day 2 may be more likely to produce a peak in brain bilirubin before most dendritic outgrowth has begun. Clinical evidence supports at least two effects of bilirubin neurotoxicity: (1) kernicterus which leads to permanent brain damage and death, and (2) bilirubin encephalopathy which produces transient changes in brain function [8]. It is not known whether these expressions of bilirubin neurotoxicity lie on a continuum or are the effects of separate neuropathological mechanisms. Of the numerous factors which interact to produce bilirubin neurotoxicity, foremost are the level of unbound bilirubin, brain pH, blood-brain barrier permeability to bound and unbound bilirubin, and the susceptibility of the brain to a given bilirubin exposure [27]. The present findings support the growing evidence that neurons undergoing early differen-
tiation at the time of the exposure are highly susceptible to bilirubin neurotoxicity and cell death, while slightly more or less mature cells may show only transient changes. Acknowledgements This work was supported by NIH-NIDCD grant R01-DC00369 to S.M.S. We thank Lillian Gerity and Margaret Bennett for their skilled technical assistance.
References 1. Altman J (1972) Postnatal development of the cerebellar cortex in the rat. II. Phases in the maturation of the Purkinje cells and of the molecular layer. J Comp Neurol 145: 399–464 2. Reference deleted 3. Blanc WA, Johnson L (1959) Studies on kernicterus. Relationship with sulfonamide intoxication, report on kernicterus in rats with glucuronyl transferase deficiency and review of pathogenesis. J Neuropathol Exp Neurol 18: 165–187 4. Chen HC, Lien IN, Lu TC (1965) Kernicterus in newborn rabbits. Am J Pathol 46: 331–343 5. Diamond I, Schmid R (1966) Experimental bilirubin encephalopathy: the mode of entry of bilirubin-14C into the nervous system. J Clin Invest 45: 678–689 6. Goodlett CR, Hamre KM, West JR (1990) Regional differences in the timing of dendritic outgrowth of Purkinje cells in the vermal cerebellum demonstrated by MAP2 immunocytochemistry. Dev Brain Res 53: 131–134 7. Gunn CK (1938) Hereditary acholuric jaundice in new mutant strain of rats. J Hered 29: 137–139 8. Hansen TWR (1994) Bilirubin in the brain. Clin Pediatr 33: 452–459 9. Haymaker W, Margoles C, Pentschew A, Jacob H, Lindenberg R, Arroyo LS (1961) Pathology of kernicterus and posticteric encephalopathy. In: Swinyard CA (ed) Kernicterus and its importance in cerebral palsy. Thomas, Springfield, pp 221–228 10. Johnson L, Sarmiento F, Blanc WA, Day R (1959) Kernicterus in rats with inherited deficiency of glucuronyl transferase. Am J Dis Child 101: 322–349 11. Keino H, Kashiwamata S (1989) Critical period of bilirubin-induced cerebellar hypoplasia in a new Sprague-Dawley strain of jaundiced Gunn rats. Neurosci Res 6: 209–215 12. Keino H, Sato H, Semba R, Aoki E, Kashiwamata S (1985/86) Mode of prevention of phototherapy of cerebellar hypoplasia in a new Sprague-Dawley strain of jaundiced Gunn rats. Pediatr Neurosci 12: 145–150 13. Konigsmark B (1970) Methods for the counting of neurons. In: Nauta WJH, Ebbesson SOE (eds) Contemporary research methods in neuroanatomy. Springer, Berlin Heidelberg New York, pp 315–340 14. Labrune PH, Myara A, Francoula J, Trivin F, Odievre M (1992) Cerebellar symptoms as the presenting manifestations of bilirubin encephalopathy in children with Crigler-Najjar type I disease. Pediatrics 89: 768–770 15. Lucey JF, Hibbard E, Behrman RD, Esquivel de Gallardo FO, Windle WF (1964) Kernicterus in asphyxiated newborn Rhesus monkeys. Exp Neurol 9: 43–58 16. Reference deleted 17. Rose AL, Wisniewski H (1979) Acute bilirubin encephalopathy induced with sulfadimethoxine in Gunn rats. J Neuropathol Exp Neurol 38: 152–164 18. Rozdilsky B, Olszewski J (1961) Experimental study of the toxicity of bilirubin in newborn animals. J Neuropathol Exp Neurol 20: 193–205 19. Sawasaki Y, Yamada N, Nakajima H (1976) Developmental features of cerebellar hypoplasia and brain bilirubin levels in a mutant (Gunn) rat with hereditary hyperbilirubinemia. J Neurochem 27: 577–583 20. Schutta HS, Johnson L (1967) Bilirubin encephalopathy in the Gunn rat: a fine structure study of the cerebellar cortex. J Neuropathol Exp Neurol 26: 377–396
460 21. Schutta HS, Johnson L (1969) Clinical signs and morphologic abnormalities in Gunn rats treated with sulfadimethoxine. J Pediatr 75: 1070–1079 22. Schutta HS, Johnson L (1970) Mitochondrial abnormalities in bilirubin encephalopathy. J Neuropathol Exp Neurol 29: 296– 305 23. Schutta HS, Johnson L (1971) Electron microscopic observations on acute bilirubin encephalopathy in Gunn rats induced by sulfadimethoxine. Lab Invest 24: 82–89 24. Takagishi Y, Yamamura H (1987) The critical period of Purkinje cell degeneration and cerebellar hypoplasia due to bilirubin. Acta Neuropathol (Berl) 75: 41–45
25. Voogd J, Geritts NM, Marani E (1985) Cerebellum. In: Paxinos G (ed) The rat nervous system, vol 2: Hindbrain and spinal cord. Academic Press, New York, 251–291 26. Weil MC, Menkes JH (1975) Bilirubin interaction with gangliosides: possible mechanism in kernicterus. Pediatr Res 9: 791–793 27. Wennberg RP (1993) Animal models of bilirubin encephalopathy. Adv Vet Sci Comp Med 37: 87–113 28. Winer BJ (1971) Statistical principals in experimental design. McGraw-Hill, New York