The first two authors contributed equally to this article.
Address corresponding and reprint requests to Dr. S.C. Baraban at Box 0520, Department of Neurological Surgery, University of California, San Francisco, 503 Parnassus Ave., San Francisco, CA, U.S.A. E-mail: email@example.com
Summary: Purpose: In patients with tuberous sclerosis complex (TSC), a wide range of neurologic abnormalities develop, including mental retardation and seizures. Brains from TSC patients are characterized by the presence of cortical tubers, large dysmorphic neurons, and abnormal cytomegalic cells. Although analysis of human TSC brain samples led to the identification of these abnormal cell types, very little is known about how these cells function. In an effort to model TSC-associated CNS abnormalities (and ultimately to analyze the electrophysiologic properties of abnormal cells), we examined Eker rats carrying a Tsc2 mutation. Anatomic studies, including standard histologic stains and immunocytochemistry, were performed on young Eker rats exposed to a carcinogen in utero or aged untreated Eker rats (18–24 months old).
Methods: Pregnant TSC2+/− females were injected once a day with hydroquinone (HQ), and offspring were killed at postnatal day P14 or P28. Coronal tissue sections throughout the CNS were prepared and stained for cresyl violet. In separate studies, brains of old untreated Eker rats were sectioned for anatomic analysis by using standard immunohistochemical techniques.
Results: Tissue sections stained with cresyl violet did not reveal any gross differences between HQ-treated Eker (Tsc2Ek/+) rats and siblings (Tsc2+/+). However, two classes of abnormal giant cells were observed in brain sections from untreated aged Eker rats: (a) large dysmorphic pyramid-like cells immunoreactive for NeuN, tuberin, and EAAC-1 in layers IV–VI; and (b) abnormal cytomegalic cells immunoreactive for glial fibrillary acidic protein (GFAP), vimentin, and nestin in deep cortical layers or along the white matter. In addition, large subependymal astrocytomas were observed in four animals.
Conclusions: Our data suggest that cortical tuber formation in Eker rats is a rare event and that prenatal exposure to a nongenotoxic carcinogen such as HQ is not sufficient to induce tuber formation. However, with advanced age, an increased likelihood of astrocytoma formation and the emergence of dysmorphic neurons and cytomegalic cells in the Eker rat brain might exist; each of these abnormalities mimics those seen clinically and could contribute to neurologic problems associated with TSC. Further analysis of this rodent model may be warranted.
Children with tuberous sclerosis complex (TSC) have mental retardation, autism, cognitive impairment, and intractable seizures. TSC patients also are characterized by the development of benign tumors (hamartomas) and focal dysplasias in multiple organ systems (1–3). In the brain, TSC lesions typically include cortical tubers, subependymal nodules, and giant cell astrocytomas (4–6). Cortical tubers have been identified as an epileptogenic focus in TSC patients and are often the target of resective surgical procedures (7–9). Tubers contain abnormal cells with phenotypes classified as neuronal (large dysmorphic neurons; DNs) or astrocytic (abnormal giant cells; GCs). Despite considerable interest in how abnormal cells contribute to an epileptic condition in TSC patients, little is known about how these cells function. To examine abnormal cells, and in an initial effort to establish a potential animal model for future electrophysiologic analysis of their role in seizure genesis, we studied the Eker rat model of TSC (10). In Long–Evans Eker rats, one of the Tsc2 alleles is inactivated, and evidence for a cortical tuber, containing abnormal cytomegalic cells, was reported in one 19-month-old carrier (1). However, identification of abnormal cell types reminiscent of those seen in patients with TSC has not been reported.
It is well documented that exposure to chemical carcinogens or ionizing radiation results in a high incidence of renal carcinomas in the Eker rat (11–13). These findings suggest that two “hits” (one inherited, one somatic) are necessary to produce tumors. Whether a “second hit” leads to development of cortical tubers or collections of abnormal cells in the brains of Eker rats has not been investigated. Here we examined whether (a) prenatal exposure to hydroquinone (HQ), a benzene metabolite and carcinogen shown to induce renal tumors in Eker rats (11), or (b) the aging process can serve as second hits leading to brain abnormalities in the Long–Evans Eker rat.
MATERIALS AND METHODS
Founder rats carrying the Eker mutation were kindly provided by Dr. Cheryl L. Walker (University of Texas M.D. Anderson Cancer Center, Smithville, TX, U.S.A.). Eker rats used for these studies were originally bred on a Long–Evans background. All animals were housed and treated in accordance with University of California, San Francisco, animal care guidelines. Eker rats were identified as carriers (Tsc2Ek/+) by using reverse transcriptase-polymerase chain reaction (RT-PCR) detection of the Tsc2 gene mutation (14). No evidence for abnormal behaviors or seizures was evident in routine animal observation. Renal tumors were found in all Tsc2Ek/+ carriers, but in none of their Tsc2+/+siblings.
In some studies, pregnant female Eker rats were administered intraperitoneal (i.p.) injections of HQ sonicated in corn oil. Subconvulsive and sublethal doses were chosen according to published carcinogenicity and toxicity data (11,15,16). Rats were administered HQ once daily at intervals indicated in the Results section. All offspring were genotyped between postnatal days 4 (P4) and P7.
Tissue preparation and immunohistochemistry
For immunohistochemistry or histology, rats were anesthetized with ketamine/xylazine and then perfused transcardially with chilled 1× phosphate-buffered saline (PBS) followed by 4% freshly prepared paraformaldehyde. The brains were immediately removed, postfixed at 4°C, rinsed in PBS, and then cryoprotected in 30% sucrose (in 1× PBS). Floating sections (40 μm) were cut by using a cryostat (Leica, Microsystems AG, Wetzlar, Germany). Sections were permeabilized in PBS with 0.1% Triton X-100. Free-floating sections were first pretreated with 1% H2O2 in PBS for 15 min. All subsequent steps were performed with PBS and 0.2% Triton X-100 (PBT). Sections were blocked with 10% normal goat serum (NGS) in PBT (1–2 h), and then primary antibodies were added directly to the sections in blocking solution and incubated for 15 to 18 h. After washing with PBT, sections were incubated with biotinylated goat anti-rabbit immunoglobulin G (IgG) or biotinylated horse anti-mouse IgG (Vector Labs, Burlingame, CA, U.S.A.; 1:500) for 2 h in PBT/10% macromolecular complex (ABC complex; Vector Labs; 1:500) in PBS/10% NGS for 2 h. This was followed by incubation with a biotinylated secondary antibody (1:500). Immunoreactivity was visualized by using the Vectastain ABC kit and DAB (diaminobenzidine) exposure. Sections were washed and mounted in ProLong (Molecular Bioprobes, Eugene, OR, U.S.A.). In some cases, cell bodies were visualized in slide-mounted sections through standard cresyl violet staining. Examination of tissue sections was performed by an investigator blinded to the status of the animal. All sections were examined by using a microscope equipped with a SPOT cooled charge-coupled device camera.
Free-floating sections were immunostained for monoclonal antibodies: mouse anti-EAAC-1 (Chemicon Temecula, CA, U.S.A.; 1:500), mouse antinestin (Chemicon; 1:250), mouse anti-NeuN (Chemicon; 1:500), and mouse antivimentin (Chemicon, 1:500), and with polyclonal antibodies rabbit antiglial fibrillary acidic protein (GFAP; Chemicon, 1:1,000) and rabbit antituberin C-20 (Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A., 1:750).
Prenatal treatment with a carcinogen
Pregnant Eker rats heterozygous for the Tsc2 mutation were treated with HQ, a carcinogen previously shown to induce renal adenomas and carcinomas in these animals (11). Rats were separated into two treatment groups that spanned “early” (embryonic days 8–19; n = 9) and “late” (E15–E20; n = 5) neurodevelopmental epochs in the rat. Daily intraperitoneal (i.p.) injections of HQ were made at drug concentrations of 50, 100, or 150 mg/kg. At 150 mg/kg, prenatal HQ administration caused resorption or abortion of fetal development in some litters (n = 2). Surviving offspring were genotyped and killed at postnatal days 14 or 28 (e.g., young adult). Serial tissue sections throughout the CNS were prepared and stained with cresyl violet (Fig. 1). Cortical tubers were not observed in any HQ-treated offspring examined (n = 9). Histologic examination of cortical and hippocampal architecture did not reveal any major differences between HQ-treated Eker (Tsc2Ek/+) rats and age-matched siblings (Tsc2+/+).
Brain abnormalities in aged Eker rats
Next we examined coronal brain sections beginning at the level of caudate putamen and ending at the level of auditory cortex/rostral dentate gyrus. Cresyl violet–stained sections from untreated aged Eker rats (18–24 months old; n = 9) bred at UCSF revealed no gross changes in structure, lamination, or development of tubers. Interestingly, closer examination of histologic sections indicated the presence of two types of abnormal cortical cells in all aged Tsc2Ek/+ rats. First, abnormally large pyramid-shaped cells >40 μm in diameter were found in cortical layers IV–VI (Fig. 2A and B). These cells were rare (∼15–30 per animal) and did not exhibit any preferential regional localization but were consistently observed in all Eker carriers but in no age-matched siblings. The large pyramid-shaped cells stained with a neuron-specific antibody (NeuN; Fig. 2C), suggesting a neuronal phenotype and DN classification. Large DNs were strongly tuberin immunoreactive similar to recent clinical findings (6) (Fig. 2D). DNs also expressed a neuronal glutamate transporter protein (EAAC-1; Fig. 2E), indicative of a glutamatergic phenotype. Second, cresyl violet–stained sections double labeled with an astrocyte-specific antibody (i.e., GFAP) indicated the presence of abnormal giant cytomegalic cells (GCs) in deep cortical layers (Fig. 3A). These cells were irregularly shaped and extended intensely GFAP-immunoreactive processes. GCs also exhibited strong vimentin immunoreactivity (Fig. 3C), further consistent with an astrocytic phenotype. Nestin, a marker of immature cells, was strongly expressed in single GCs in deep cortical layers (Fig. 3B). Nestin-positive GCs also could be found in clusters along cortical layer VI or in hamartoma-like formations in the subcortical white matter (Fig. 3D). These cells were rare (∼10–30 cells/cluster per animal) but were consistently observed in all Eker carriers but in no age-matched siblings.
In addition, large tumors connected to the brainstem/cerebellum were observed in four aged Eker heterozygous rats (Fig. 4). Generally, tumors were >0.5 mm in diameter and appeared to grow in an expansive, noninfiltrative fashion, largely sparing nearby brain histoarchitecture. These lesions contained varying proportions of large cells with abundant eosinophilic cytoplasm, large lobulated or multiple nuclei, and intermingled with dense fibrillary processes. Tumors appeared to resemble subependymal giant cell astrocytomas characteristic of patients with TSC.
Analysis of human TSC brain samples has yielded much information on the phenotype and morphology of abnormal cell types. Unfortunately, human TSC tissue is genetically heterogeneous, difficult to obtain, and not easily amenable to electrophysiologic studies. One approach to circumvent these problems is to use animal models of TSC. Several mouse models have been generated by targeted deletion of Tsc1 or Tsc2 genes (17,18), and a rat featuring a spontaneous germline mutation was identified in 1961 (19). Homozygous deletion of TSC genes in rodents invariably leads to significant disruption of early embryonic development and lethality. Thus analysis of the roles played by TSC gene mutations in brain development and function has focused largely on heterozygous offspring. It is worth noting, however, that recent conditional knockout of TSC genes resulted in novel mouse mutants that may provide new insights into seizure development (20,21). Here we describe the presence of abnormal cell types in aged heterozygous Long–Evans Eker rats, a Tsc2 mutant, that closely resemble cells seen in human TSC brain samples. We also provided evidence that a “second hit” with prenatal HQ does not induce cortical tuber formation in this model. Although tubers did not develop in these animals, the presence of abnormal cell types with phenotypes similar to those reported clinically suggests that the aged Long–Evans Eker rat could be useful model to study molecular/anatomic properties of these cells in more detail.
The Eker rat was initially described as a model for hereditary renal cell carcinoma (19) and features an autosomal dominant mutation in the Tsc2 gene encoding tuberin. Tuberin, a guanosine triphosphatase (GTPase)-activating protein, is thought to complex with hamartin (Tsc1) to regulate cell growth (22). Although no neurologic symptoms have been reported in the four decades since Eker animals were first identified, specific TSC-like abnormalities are found in the brains of these animals. Most common are subcortical hamartomas, subependymal hamartomas, and anaplastic gangliogliomas. Each of these lesions has been reported in Eker carriers bred on either a Fischer 344 or Long–Evans background. Cells within these lesions exhibit neuronal and glial features. For example, large cells with abundant eosinophilic cytoplasm found within subependymal lesions of the Fischer 344 Eker rat express GFAP but not neurofilament (10). In contrast, markers of neural proliferation and migration (CRMP4 and DCX) were observed in the cortex of Fischer 344 Eker rats (23). Interestingly, evidence for a cortical tuber has been reported only for the Long–Evans Eker rat (one of 19 animals). In this case, a tuber was found in the primary motor cortex and contained many cytomegalic cells immunoreactive for neurofilament or other neuronal markers (1). Large cytomegalic cells within the tuber were only weakly immunoreactive for tuberin (in contrast to clinical reports) and GFAP (indicative of a nonastrocytic phenotype). By using prenatal exposure to HQ, a carcinogen known to induce “second hit” renal carcinomas in Eker carriers, we were unable to generate cortical tubers in any HQ–Long–Evans Eker rat offspring. These findings do not necessarily exclude the possibility that tubers can be induced in Eker rats, but suggest that (a) HQ may have a renal-specific site of action, (b) a prenatal “second hit” that targets neuron development/migration may be more appropriate, or (c) tuber formation in rodents is so rare as to preclude the use of these animals for further laboratory study of tuber cell function. Furthermore, the lack of an epileptic phenotype in Eker rats may indicate that generation of seizures requires a “critical mass” of abnormal cells (or tubers) as seen in humans with TSC. Whether this critical mass can be attained in the Eker rat via a second genetically based mutation is an additional possibility.
Interestingly, certain pathologic features of abnormal brain cells in human TSC are closely recapitulated in the untreated Long–Evans Eker rat (1,6,19). Although cells within cortical tubers or hamartomas have been described in Eker rats (1), identification and immunohistochemical analysis of abnormal cortical cell types (e.g., dysmorphic neurons and GCs) in nontuber cortex was not reported. Here we demonstrate that these abnormal cells, like the cortical tuber, are prominent in older animals, suggesting, for the first time, that the aging process can serve as a “second hit” in rats with an inherited TSC2 mutation. First, abnormally large pyramid-like neurons located in deep cortical layers resemble cells previously designated large DNs. DNs identified in the aged Eker rat cortex exhibited an unusually large soma, hints of multinucleation (see Fig. 2D), and a clear expression of the neuron-specific antigen NeuN. That these cells also express tuberin was not reported in previous descriptions (6) but is consistent with recent findings of strongly tuberin-immunoreactive cells in human TSC cortex. Staining with an antibody to a glutamate transporter (EAAC-1) provides the first evidence that DNs could provide some type of abnormal glutamate-mediated excitability to the TSC brain. The presence of an abnormal collection of glutamatergic DNs is consistent with a hypothesis that these cells play a role in the generation of abnormal seizure discharge associated with TSC lesions. Second, our identification of an abnormally large astrocyte-like cell type in the aged Eker rat brain again resembles large cytomegalic astrocytic GCs reported in human TSC cortex (24,25). Consistent with clinical classification of GCs as astrocytic, our immunohistochemical staining confirmed that GCs in the Eker rat brain express astrocyte-specific markers (e.g., GFAP, vimentin, and nestin). The presence of nestin-positive GC cell clusters in deep cortical layers and subcortical white matter tracts also is consistent with a hypothesis first raised by Crino et al. (25) that these clusters may highlight the pathway of abnormally migrating cells destined to form tubers (or other abnormal collections of TSC cells). The specific physiologic role(s) played by DNs and GCs can be elucidated through direct electrophysiologic analysis of their intrinsic and synaptic functions. Although it is tempting to speculate that recordings from these cells can be achieved in the Eker rat, their scarcity (probably no more than a few dozen cells per brain) precludes the successful identification and patch-recording required to study their function in vitro. Nonetheless, the striking histologic and immunohistochemical similarities between these abnormal cortical cells and those found in human TSC brain samples, as well as the presence of subependymal astrocytomas, highlight the potential usefulness of Long–Evans Eker rats as an experimental model to study the pathogenesis of TSC lesions.
Acknowledgment: This work was supported with funds from the Tuberous Sclerosis Alliance (S.C.B.). We thank Dr. Tarik Tahan for analysis and classification of tumors.