Experimental Neurology 201 (2006) 519 – 524 www.elsevier.com/locate/yexnr
Brief Communication
Endogenous neurosteroids modulate epileptogenesis in a model of temporal lobe epilepsy Giuseppe Biagini a,⁎, Enrica Baldelli a , Daniela Longo a , Luca Pradelli a , Isabella Zini a , Michael A. Rogawski b , Massimo Avoli c,d b
a Dipartimento di Scienze Biomediche, Università di Modena e Reggio Emilia, 41100 Modena, Italy Epilepsy Research Section, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892-3702, USA c Dipartimento di Fisiologia Umana e Farmacologia ‘V. Erspamer’, Università di Roma ‘La Sapienza’, 00185 Roma, Italy d Montreal Neurological Institute and Department of Neurology and Neurosurgery, McGill University, Montreal, QC, Canada H3A 2B4
Received 17 January 2006; revised 28 March 2006; accepted 21 April 2006 Available online 14 June 2006
Abstract Neurosteroids modulate seizure susceptibility, but their role in the regulation of epileptogenesis is unknown. Status epilepticus (SE) induces temporal lobe epileptogenesis following a latent period in which glial cells are activated. Here, we found that P450scc, the rate-limiting enzyme in steroid synthesis, is upregulated in hippocampal glia during the latent period after pilocarpine-induced SE in rats. More prolonged SE was associated with greater P450scc expression and longer latencies to the development of seizures, suggesting that enhanced steroid synthesis retards epileptogenesis. The 5a-reductase inhibitor finasteride, which blocks neurosteroid synthesis, reduced the latent period, indicating that glia-derived neurosteroids may be antiepileptogenic. © 2006 Elsevier Inc. All rights reserved. Keywords: Epileptogenesis; Status epilepticus; Glia; Hippocampus; Neurosteroid; Pilocarpine
Introduction Cholesterol side-chain cleavage cytochrome P450 (P450scc) is the rate-limiting enzyme for steroidogenesis in all endocrine tissues. This enzyme, which converts cholesterol to pregnenolone, is also present in the central nervous system (CNS) where it plays a key role in the local synthesis of steroids, including GABAA receptor modulatory neurosteroids (Mellon and Griffin, 2002). These steroids potentiate GABA-activated chloride currents, thereby enhancing GABA-mediated inhibition (Lambert et al., 2003). Neurosteroids have a variety of CNS actions related to their effects on the GABAA receptor, including protective activity in diverse experimental seizure models (Rogawski and Reddy, 2004). Therefore, neurosteroids may represent endogenous modulators of seizure susceptibility. Such endogenous GABAA receptor modulatory neurosteroids include ⁎ Corresponding author. Dipartimento di Scienze Biomediche, Sezione di Fisiologia,Via Campi 287, Italy. Fax: +39 059 205 5363. E-mail address:
[email protected] (G. Biagini). 0014-4886/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.expneurol.2006.04.029
allopregnanolone, allotetrahydrodeoxycorticosterone, androstanol and androsterone, which are derived from their parent steroid hormones progesterone, deoxycorticosterone and testosterone by reduction at the 5- and 3-positions on the A-ring. The initial 5α-reduction is rate limiting for neurosteroid synthesis and can be blocked by the azasteroid finasteride, which prevents the anticonvulsant activity of the parent steroids progesterone and deoxycorticosterone (Rogawski and Reddy, 2004). In addition, it has recently been reported that finasteride aggravated seizures in a patient taking progesterone for the treatment of catamenial epilepsy (Herzog and Frye, 2003). Glial cell reaction is known to accompany brain damage caused by status epilepticus (SE) (Represa et al., 1995; Belluardo et al., 1996). This phenomenon has been mainly characterized in the CA3 subfield of the hippocampus, a region that is vulnerable to damage in several animal models of temporal lobe epilepsy (TLE) (Clifford et al., 1987; Represa et al., 1995). Interestingly, glial cell reaction develops in the first week after SE and slowly resolves in the following weeks (Represa et al., 1995; Belluardo et al., 1996; Borges et al., 2003). Thus, the time course of glial
520
Brief Communication
cell reaction overlaps the latent period of epileptogenesis, the interval following SE prior to the onset of spontaneous seizures (Leite et al., 1990). Glia are the most active steroidogenic cells in the brain inasmuch as they contain a greater diversity and express higher levels of steroidogenic enzymes than neurons. In particular, astrocytes and oligodendrocytes and to a lesser extent neurons express P450scc. Here, we sought to determine whether a link existed between glial activation, neurosteroids and seizure development in the latent period. To this end, we studied (i) the time course of P450scc expression in the pilocarpine model of TLE, (ii) the relationship between P450scc and the time of appearance of spontaneous seizures and, finally, (iii) the effects of finasteride on the duration of the latent period. Materials and methods Animals and treatments Male Sprague–Dawley rats (270–300 g, Harlan Italy, S. Pietro al Natisone, Italy) surviving (n = 157) pilocarpine-
induced SE were used in four different experiments. Pilocarpine treatment was as described by Biagini et al. (2001). To prevent the effects caused by peripheral muscarinic receptor stimulation, we treated rats with subcutaneous (s.c.) scopolamine methylnitrate (1 mg/kg) 30 min before intraperitoneal (i.p.) pilocarpine (380 mg/kg). Seizure activity was scored according to the scale of Racine (1972) with modifications (Biagini et al., 2005). In particular, stage 0 corresponded to no response; stage 1, facial twitching; stage 2, myoclonic body jerks; stage 3, unilateral forelimb clonus; stage 4, clonic rearings; stage 5, generalized clonic–tonic convulsions; stage 6, continuous clonic–tonic convulsions for at least 1 h. All rats exhibited a stage 6 response to pilocarpine. In the experiment to determine the P450scc time course, SE was allowed to terminate spontaneously after 4–6 h. In the other experiments, SE was stopped with diazepam (20 mg/ kg, i.p.) at defined time intervals. The mortality was about 15%. After recovering from SE, rats were irritable and hypersensitive to sensory stimulation. The manifestation of spontaneous seizures (facial automatisms, head nodding, forelimb clonus, rearing and occasional generalized convulsions) was monitored by
Fig. 1. Time course of P450scc induction after pilocarpine-induced status epilepticus (SE). (A) P450scc immunoreactivity in the CA3 hippocampal subfield is enhanced at 1 day after SE (compare with control staining) and reaches a peak after 1 week, mostly in glial elements (inset). The increase persists in glial cells for at least 3 weeks, when P450scc-positive elements are visible only in the CA3 stratum lacunosum-moleculare (arrows). (B) Semiquantitative evaluation of the time course of P450scc immunoreactivity in glial cells and neurons in strata radiatum and lacunosum-moleculare and stratum pyramidale, respectively. Data points represent mean ± SEM of cell counts and field area (FA) values. *P b 0.05; **P b 0.01 vs. control-positive cell counts. #P b 0.05; ##P b 0.01 vs. control FAvalues. Games–Howell test was used for multiple comparisons. Scale bars, 50 μm.
Brief Communication
daily video recording for 6-h periods, starting 3 days after SE. Non-epileptic control (NEC) rats (n = 17) injected i.p. with saline did not develop any of these seizure manifestations. In the first experiment, pilocarpine-treated rats that had experienced a 4–6 h of SE and NECs were sacrificed 1, 3, 7, 14 and 21 days after the injection (n = 12, 9, 18, 10 and 11, respectively, for pilocarpine; n = 2 at each time point for NECs) to assess the time course of P450scc changes. In the second experiment, we investigated the induction of P450scc as a function of SE duration by interrupting SE after 1, 2 or 3 h (n = 19, 12 and 16, respectively); these animals were sacrificed 3 days later. In the third experiment, pilocarpine-treated rats experiencing 1 or 3 h SE (n = 13 and 14, respectively) were scored for the appearance of spontaneous seizures. In the last experiment, NECs (n = 4) and pilocarpine-treated rats experiencing 3 h SE (n = 10) were treated with daily s.c. injections of 100 mg/kg finasteride (Ivy Chiral Chemicals, NJ, USA) or with 30% hydroxypropyl-β-cyclodextrin in water (vehicle-treated, n = 3 for NECs and n = 13 for the pilocarpine group), starting 3 days after SE and continuing for 18 days. The animals were monitored for spontaneous seizures each day and were then anaesthetized with chloral hydrate (450 mg/kg, i.p.) and perfused via the ascending aorta with 100 ml
521
saline followed by Zamboni's fixative, as previously described (Biagini et al., 1993). Brains were kept overnight in the same fixative at 4°C and, after cryoprotection by immersion in 15% and 30% sucrose-phosphate buffer solutions, were frozen and cut horizontally with a freezing microtome. All experimental procedures were approved by Institutional Animal Care Committees and conformed to National Institutes of Health guidelines. Immunohistochemistry and image analysis The distribution of P450scc in pilocarpine-treated and NEC rats was studied with a polyclonal antibody (Chemicon, Tamecula, CA, USA) raised against amino acids 421–441 of rat P450scc. This antibody has been previously characterized and demonstrated to be specific for P450scc (Roby et al., 1991; King et al., 2002). Immunohistochemistry was carried out using the avidin–biotin complex technique and diaminobenzidine as chromogen (Biagini et al., 1993). Endogenous peroxidase was blocked by 0.1% phenylhydrazine in phosphate-buffered saline (PBS) for 20 min, followed by several washes in PBS before incubation with the primary antibody. Stained sections were analyzed by densitometry using KS300 image analysis software (Biagini et al., 2005).
Fig. 2. Dependence of P450scc induction on status epilepticus (SE) duration. P450scc immunoreactivity was determined 3 days after pilocarpine injection in rats experiencing 1, 2 and 3 h SE. (A) In rats experiencing 3 h SE, P450scc is induced in several regions (arrows) of the hippocampal formation, such as dentate gyrus (DG), several Cornu Ammonis (CA) subfields and entorhinal cortex (EC). (B) Field area (FA) and positive glial cell counts increase as the duration of SE increases from 1 to 3 h. Data points represent mean ± SEM of cell counts and FA values. *P b 0.05 vs. 3 h SE-positive cell counts; #P b 0.05 vs. 3 h SE FA values. Games–Howell test was used for multiple comparisons. Scale bar, 500 μm.
522
Brief Communication
P450scc immunostaining was quantified in the CA3 subfield as follows: glial cells numbers and areas were evaluated in fields measuring 0.28 mm2 in strata radiatum and lacunosum-moleculare, which are nearly devoid of neurons; neuron number and area were measured in 0.38 mm2 fields within the CA3 pyramidal layer, where there are few astrocytes. The sections (50 μm thick) were obtained 6.1 mm from bregma (Paxinos and Watson, 1998). Data from two different fields in each hippocampus of three sections were averaged. Background values were obtained from areas that did not contain any stained cells. Stained profiles were discriminated from background and the field area (FA) corresponding to the region covered by specifically stained cell profiles was measured (Biagini et al., 1993, 1995). Cell profile counts were determined in each field for profiles greater in diameter than a minimum cutoff value, taken to be 4 and 7 μm for glial cells and neurons, respectively. Statistical analysis
in the 1-h SE group compared with the 3-h SE group, respectively (Fig. 2B). Factors that affect the course of epileptogenesis were further investigated by determining the onset of seizures during the 30day period following SE. Surprisingly, a shorter duration of SE (1 h) was associated with a more rapid onset of seizures (Fig. 3A). This raised the possibility that the greater induction of P450scc or other neurosteroid synthesizing enzymes by the prolonged SE results in higher neurosteroid levels that interfere with epileptogenesis. To address this hypothesis, we treated rats that had experience 3 h of SE with finasteride to block neurosteroid synthesis during the latent period of epileptogenesis. As shown in Fig. 3B, finasteride treatment accelerated the development of seizures, indicating that neurosteroids retard epileptogenesis. Finally, we ruled out the possibility that the finasteride per se could exert a proconvulsant effect by treating control animals (n = 4) for 18 days with doses of finasteride similar to those used in pilocarpine-treated rats. None of these animals exhibited seizures.
Statistical analysis was performed using the Statistical Package for the Social Sciences, version 8.0 (Chicago, IL, U.S.A., 1998). Results were analyzed with one-way analysis of variance followed by the Games–Howell test for multiple comparisons. The Kaplan–Meier method was used to estimate the rate of onset of stage 5 spontaneous seizures after SE (Racine, 1972). These curves were compared by the log-rank test. Results P450scc immunoreactivity was studied in the CA3 strata radiatum and lacunosum-moleculare at intervals following a sustained (4–6 h) episode of SE induced by pilocarpine. As shown in Fig. 1, P450scc immunoreactivity in presumptive glial cells increased and then returned toward control values during the 3-week period following SE. Specifically, there was a doubling in cell counts by day 1 and a sustained tripling from days 3 to 7. Even at 21 days after SE, the number of P450scc-positive glial cells was still 2.5-fold the control value. At that time, immunoreactive glial cells were confined to the stratum lacunosum-moleculare (Fig. 1A, arrows). Glial FA values paralleled the changes in cell counts, but the magnitude of the elevation was greater, with the mean FA value at a sustained peak (i.e., from days 3–7) reaching approximately 9-fold the control level. In addition, there was an increase in P450scc immunoreactivity in the CA3 pyramidal cell layer following SE, but both magnitude and duration were less pronounced: significant increases in cell counts were found at days 3 and 7 post-SE whereas the FA values were only significantly increased at day 3. Next, we sought to determine the relationship between SE duration and the extent of the increase in P450scc immunoreactivity by arresting seizures after periods of SE from 1 to 3 h. These experiments showed a greater increase in P450scc immunoreactivity after 3 h SE than at 1 h in several regions of the hippocampal formation, including the dentate gyrus, CA1–2 subfields, subiculum and entorhinal cortex (Fig. 2A, arrows). Analysis of P450sccpositive cell numbers and FA values in the CA3 strata radiatum and lacunosum-moleculare demonstrated 1.5- and 3-fold lower values
Fig. 3. Latency to the onset of spontaneous seizures as a function of status epilepticus (SE) duration (A) and finasteride treatment (B). (A) Rats experiencing a short (1 h) episode of SE exhibited spontaneous seizures earlier than those exposed to SE for over 3 h (P b 0.01, log-rank test). (B) In rats experiencing 3 h SE, treatment with daily finasteride (100 mg/kg) during the 18-day period marked by arrows was associated with acceleration in the onset of spontaneous seizures compared with vehicle-treated animals (P b 0.01, log-rank test).
Brief Communication
Discussion Our study demonstrates that P450scc, a key enzyme in steroid synthesis, is upregulated in the CA3 hippocampal subfield following pilocarpine-induced SE. These changes occurred mainly in glial cells, although there was also a transient increase in neurons. The induction of P450scc was greater with more prolonged episodes of SE. Similar increases in P450scc staining were evident in other limbic areas, including the dentate gyrus, subiculum and entorhinal cortex, in the pilocarpine-treated rats. The changes occurred in a regional pattern that corresponds to areas of diffuse glial cell activation following SE-related neuronal damage (Represa et al., 1995; Belluardo et al., 1996). P450scc is well recognized to be expressed in type I astrocytes in the adult brain (Zwain and Yen, 1999; Shibuya et al., 2003). This enzyme is also found in certain neuronal populations, including hippocampal pyramidal and dentate granule cells (Zwain and Yen, 1999; Shibuya et al., 2003). The induction of P450scc in neurons was less pronounced than in glia. This difference can, in part, be attributed to the loss of hippocampal pyramidal neurons in the days and weeks following SE (Poirier et al., 2000). However, the extent of neuronal cell loss in CA3 is between 10% and 30% (Liu et al., 1994; Biagini et al., 2005). This amount of damage is unlikely to account for the dramatic difference between glia and neurons in P450scc immunoreactivity after SE (cf. Fig. 1). Thus, we are inclined to conclude that reactive astrocytes are likely to be a primary source of enhanced steroidogenesis after SE. By examining the relationship between SE duration and latent period length, we were surprised to find that the latency for the onset of seizures is greater after a more prolonged period of SE. In fact, approximately one-half of the animals experiencing 1 h SE exhibited spontaneous seizures one week after pilocarpine injection, whereas the group that had experienced 3 h SE did not achieve a similar incidence of seizures until 3 weeks after SE. The time course of the development of spontaneous seizures we observed with 3 h SE is similar to that reported in other studies (Leite et al., 1990). Because 3 h SE caused a greater induction of P450scc than 1 h SE, we hypothesized that increased neurosteroid synthesis was responsible for the slowing of epileptogenesis (delay in onset of spontaneous seizures) in the animals that experienced longer SE. To examine this possibility, rats that had experienced 3 h SE were treated with finasteride to block neurosteroid synthesis. We found that the latent period in the finasteride-treated animals was reduced compared with control animals and was comparable to that observed in rats experiencing 1 h SE. These results support the possibility that neurosteroids influence epileptogenesis during the latent period by delaying the development of spontaneous seizures. It is unlikely that the endogenously synthesized neurosteroids suppress seizures in this experiment solely as a result of their anticonvulsant activity (Kokate et al., 1994) because none of the animals stopped exhibiting seizures when finasteride was withdrawn. Thus, it seems that neurosteroids specifically delay epilepsy development, although this will need to be explored further. In the kindling model of epileptogenesis, it is recognized that agents that enhance GABA-mediated inhibition retard the development of the epileptic state (Burnham, 1989). The apparent antiepileptogenic effect of neurosteroids in the present situation could be comparable.
523
The dose of finasteride we used was chosen because it previously has been shown to nearly completely inhibit the anticonvulsant activity of neurosteroid precursors by substantially blocking their conversion to neurosteroids (Kokate et al., 1999; Reddy et al., 2004). Moreover, this dose of finasteride when administered according to the same protocol did not induce seizures in control animals. Thus, finasteride does not appear to have unspecific proconvulsant actions that could artefactually account for the acceleration in seizure onset. Overall, our results suggest that induction of neurosteroid synthesis by severe SE exerts a suppressive effect on epileptogenesis during the latent period. Although some steroid synthetic enzymes including P450scc are expressed in neurons, the increased expression of P450scc following SE is more persistent in glia. Therefore, we propose that neurosteroids involved in regulating epileptogenesis may largely originate from glia thus reflecting glial cell activation in damaged tissue. In line with this view, glial activation (maximal during the first two to three weeks following SE) corresponds to the latent period. If neurosteroids do serve as regulators of epileptogenesis, this raises the possibility that exogenously administered neurosteroids (or synthetic analogs with similar activity) could have clinical utility in the prevention of epileptogenesis, in the setting of SE or in other situations where there is a risk for the development of epilepsy, such as in traumatic brain injury. Acknowledgments This work was supported by the Italian Ministry of Education, University and Research (Fondo per gli Investimenti della Ricerca di Base RBNE01NR34_011, Research Project of Relevant National Interest 2003060538_003), Pierfranco and Luisa Mariani Foundation (R-06-50) and the Canadian Institutes of Health Research (Grant MT-8109).
References Belluardo, N., Mudo, G., Jiang, X.H., Condorelli, D.F., 1996. Induction of astroglial gene expression by experimental seizures in the rat: spatiotemporal patterns of the early stages. Glia 16, 174–186. Biagini, G., Pich, E.M., Carani, C., Marrana, P., Gustafsson, J.A., Fuxe, K., Agnati, L.F., 1993. Indole-pyruvic acid, a tryptophan ketoanalogue, antagonizes the endocrine but not the behavioral effects of repeated stress in a model of depression. Biol. Psychiatry 33, 712–719. Biagini, G., Sala, D., Zini, I., 1995. Diethyldithiocarbamate, a superoxide dismutase inhibitor, counteracts the maturation of ischemic-like lesions caused by endothelin-1 intranstriatal injection. Neurosci. Lett. 190, 212–216. Biagini, G., Avoli, M., Marcinkiewicz, J., Marcinkievicz, M., 2001. Brainderived neurotrophic factor superinduction parallels antiepileptic-neuroprotective treatment in the pilocarpine epilepsy model. J. Neurochem. 76, 1814–1822. Biagini, G., D'Arcangelo, G., Baldelli, E., D'Antuono, M., Tancredi, V., Avoli, M., 2005. Impaired activation of CA3 pyramidal neurons in the epileptic hippocampus. Neuromol. Med. 7, 325–342. Borges, K., Gearing, M., McDermott, D.L., Smith, A.B., Almonte, A.G., Wainer, B.H., Dingledine, R., 2003. Neuronal and glial pathological changes during epileptogenesis in the mouse pilocarpine model. Exp. Neurol. 182, 21–34. Burnham, W.M., 1989. The GABA hypothesis of kindling: recent assay studies. Neurosci. Biobehav. Rev. 13, 281–288.
524
Brief Communication
Clifford, D.B., Olney, J.W., Maniotis, A., Collins, R.C., Zorumski, C.F., 1987. The functional anatomy and pathology of lithium-pilocarpine and high-dose pilocarpine seizures. Neuroscience 23, 953–968. Herzog, A.G., Frye, C.A., 2003. Seizure exacerbation associated with inhibition of progesterone metabolism. Ann. Neurol. 53, 390–391. King, S.R., Manna, P.R., Ishii, T., Syapin, P.J., Ginsberg, S.D., Wilson, K., Walsh, L.P., Parker, K.L., Stocco, D.M., Smith, R.G., Lamb, D.J., 2002. An essential component in steroid synthesis, the steroidogenic acute regulatory protein, is expressed in discrete regions of the brain. J. Neurosci. 22, 10613–10620. Kokate, T.G., Svensson, B.E., Rogawski, M.A., 1994. Anticonvulsant activity of neurosteroids: correlation with γ-aminobutyric acid-evoked chloride current potentiation. J. Pharmacol. Exp. Ther. 270, 1223–1229. Kokate, T.G., Banks, M.K., Magee, T., Yamaguchi, S., Rogawski, M.A., 1999. Finasteride, a 5α-reductase inhibitor, blocks the anticonvulsant activity of progesterone in mice. J. Pharmacol. Exp. Ther. 288, 679–684. Lambert, J.J., Belelli, D., Peden, D.R., Vardy, A.W., Peters, J.A., 2003. Neurosteroid modulation of GABAA receptors. Prog. Neurobiol. 71, 67–80. Leite, J.P., Bortolotto, Z.A., Cavalheiro, E.A., 1990. Spontaneous recurrent seizures in rats: an experimental model of partial epilepsy. Neurosci. Biobehav. Rev. 14, 511–517. Liu, Z., Nagao, T., Desjardins, G.C., Gloor, P., Avoli, M., 1994. Quantitative evaluation of neuronal loss in the dorsal hippocampus in rats with long-term pilocarpine seizures. Epilepsy Res. 17, 237–247. Mellon, S.H., Griffin, L.D., 2002. Neurosteroids: biochemistry and clinical significance. Trends Endocrinol. Metab. 13, 35–43.
Paxinos, G., Watson, C., 1998. The Rat Brain in Stereotaxic Coordinates. 4th ed. Academic Press, San Diego. Poirier, J.L., Čapek, R., De Koninck, Y., 2000. Differential progression of Dark Neuron and Fluoro-Jade labelling in the rat hippocampus following pilocarpine-induced status epilepticus. Neuroscience 97, 59–68. Racine, R.J., 1972. Modification of seizure activity by electrical stimulation: II. Motor seizure. Electroencephalogr. Clin. Neurophysiol. 32, 281–294. Reddy, D.S., Castaneda, D.C., O'Malley, B.W., Rogawski, M.A., 2004. Anticonvulsant activity of progesterone and neurosteroids in progesterone receptor knockout mice. J. Pharmacol. Exp. Ther. 310, 230–239. Represa, A., Niquet, J., Pollard, H., Ben-Ari, Y., 1995. Cell death, gliosis, and synaptic remodeling in the hippocampus of epileptic rats. J. Neurobiol. 26, 413–425. Roby, K.F., Larsen, D., Deb, S., Soares, M.J., 1991. Generation and characterization of antipeptide antibodies to rat cytochrome P-450 side-chain cleavage enzyme. Mol. Cell. Endocrinol. 79, 13–20. Rogawski, M.A., Reddy, D.S., 2004. Neurosteroids: endogenous modulators of seizure susceptibility. In: Rho, J.M., Sankar, R., Cavazos, J. (Eds.), Epilepsy: Scientific Foundations of Clinical Treatment. Marcel Dekker, New York, pp. 319–355. Shibuya, K., Takata, N., Hojo, Y., Furukawa, A., Yasumatsu, N., Kimoto, T., Enami, T., Suzuki, K., Tanabe, N., Ishii, H., Mukai, H., Takahashi, T., Hattori, T.A., Kawato, S., 2003. Hippocampal cytochrome P450s synthesize brain neurosteroids which are paracrine neuromodulators of synaptic signal transduction. Biochim. Biophys. Acta 1619, 301–316. Zwain, I.H., Yen, S.S., 1999. Neurosteroidogenesis in astrocytes, oligodendrocytes, and neurons of cerebral cortex of rat brain. Endocrinology 140, 3843–3852.
