Proc. Natl. Acad. Sci. USA Vol. 90, pp. 7386-7390, August 1993 Neurobiology Long-term depression of glutamate-induced y-aminobutyric acid release in by insulin-like growth factor I (olivo-cerebeilar pathway//electrical stimulation/microdialysis/pontine nudeus) MANUEL A. CASTRO-ALAMANCOS AND IGNACIO TORRES-ALEMAN Laboratories of Psychobiology and Cellular and Molecular Neuroendocrinology, Cajal Institute, Consejo Superior de Investigaciones Cientificas, Madrid, Spain Communicated by William T. Greenough, April 19, 1993 (received for review November 18, 1992)

ABSTRACT We tested the possibility that insulin-like We decided to investigate whether IGF-I might be involved growth factor I (IGF-I) acts as a neuromodulator in the adult in the modulation of neurotransmitter function in the adult cerebellar cortex since previous observations indicated that cerebellar cortex since this area encompasses the climbing IGF-I is located in the olivo-cerebellar system encompassing the fiber terminal fields that synapse onto the dendritic arbors of inferior olive and Purkinje cells. We found that conjoint Purkinje cells. While Purkinje cells use y-aminobutyric acid administration ofIGF-I and glutamate through a microdialysis (GABA) as their main neurotransmitter, they receive a glu- probe stereotaxically implanted into the cerebellar cortex and tamatergic input from projecting granule cells through the greatly depressed the release of Y-ami- parallel fiber system. In addition, Purkinje cells receive nobutyric acid (GABA), which normally follows a glutamate another excitatory input from inferior olivary neurons pulse. This inhibition was dose-dependent and long-lasting. through afferents (18). Moreover, the effect was specific for glutamate since KCI- Our working hypothesis was that IGF-I released by inferior induced GABA release was not modified by IGF-I. Basic olive axons would bind to IGF-I receptors in Purkinje cells (3, fibroblast growth factor, another growth-related peptide pres- 9) and modulate their activity. Thus, we first studied whether electrical stimulation of the inferior olive would elicit IGF-I ent in the cerebellum, did not alter the response of GABA to release in the cerebellar cortex. We then explored the effects glutamate stimulation. In addition, electrical stimulation ofthe of exogenous application of IGF-I on the release of GABA in inferior olivary complex significantly raised IGF-I levels in the response to glutamate stimulation in the cerebellum. Finally, cerebellar cortex. Interestingly, when the inferior olive was we determined whether electrical stimulation of the inferior stimulated in conjunction with glutamate administration, olive might modulate the release of GABA induced by GABA release by cerebellar cells in response to subsequent glutamate in a way similar to that seen after exogenous glutamate pulses was depressed in a manner reminiscent ofthat application of IGF-I. seen after IGF-I. These rmdings indicate that IGF-I produces a long-lasting depression of GABA release by Purkinje cells in response to glutamate. IGF-I might be present in climbing fiber MATERIALS AND METHODS terminals and/or cells within the cerebellar cortex and thereby Materials. Recombinant IGF-I and basic fibroblast growth might affect function. Whether this IGF-I- factor (b-FGF) were purchased from Boehringer Mannheim; induced impairment of glutamate stimulation of Purkinje cells [1251]iodine (2000 Ci/mmol; 1 Ci = 37 GBq) was from New underlies functionally plastic processes such as long-term de- England Nuclear. IGF-I was iodinated to a specific activity pression is open to question. of =2000 Ci/mmol by using the lactoperoxidase method. All other reagents were obtained from Sigma unless otherwise Insulin-like growth factor I (IGF-I) is a trophic and neuro- indicated. modulatory peptide expressed by specific populations of Animals. Wistar rats from our in-house colony were used. neurons in the adult rat brain. Relatively high levels of IGF-I Animals were kept in a 12-h dark/light cycle and received are found in both the developing and the adult cerebellum food and water ad libitum. Animals were handled according At least of the IGF-I content in the adult cerebel- to National Institutes of Health guidelines on animal care. (1-7). part Microdialysis Procedure. Dialysis probes were of concen- lum originates from climbing fiber afferents arising from the tric design using a cuprophan hollow dialysis membrane inferior olivary complex since chemical and surgical lesions sealed at one end with epoxy resin. The other end of the ofthe olivo-cerebellar pathway produce a drastic decrease in membrane was inserted into a 25-gauge stainless steel tube cerebellar IGF-I levels (8). Moreover, the inferior olivary and sealed. A hollow fused silica fiber (140 ,um, o.d.; Poly- complex expresses high levels of IGF-I (9). micro Technologies, Phoenix), which served as the inlet, was In the developing brain, most of the effects of IGF-I on inserted through the length of the 25-gauge stainless steel nerve cells are of a trophic type (10-13) including those on tube to the end of the dialysis membrane. The liquid flows in Purkinje cells in the cerebellar cortex (14). On the other hand, the probe through the fused silica fiber and leaves the fiber at neuromodulatory actions of this peptide in areas such as the the end of the dialysis membrane. hippocampus and the hypothalamus in the adult central Rats were anesthetized with sodium pentobarbital (50 nervous system have been reported by several authors (15, mg/kg, i.p.) and dialysis probes were placed into the cere- 16). Others have shown release of bioactive IGF-I in re- bellar cortex (coordinates: AP = -11.5, L = 2.5; ref. 19) or sponse to neural stimulation (17). These data suggest that into the cerebellar cortex and deep cerebellar nuclei (coor- IGF-I may be a trophic factor during neurogenesis and a dinates: AP = -11.5, L = 2.0, V = 6.5). To dialyze the neuromodulator and/or plasticity mediator in the adult brain cerebeilar cortex, the dialysis membrane extended ventrally (9-14). from the pia 3 mm into the cerebellum. When dialyzing both

The publication costs of this article were defrayed in part by page charge Abbreviations: IGF-I, insulin-like growth factor I; GABA, y.ami- payment. This article must therefore be hereby marked "advertisement" nobutyric acid; bFGF, basic fibroblast growth factor; LTD, long- in accordance with 18 U.S.C. §1734 solely to indicate this fact. term depression. 7386 Downloaded by guest on September 25, 2021 Neurobiology: Castro-Alamancos and Torres-Aleman Proc. Natl. Acad. Sci. USA 90 (1993) 7387 the cerebellar cortex and the deep cerebellar nuclei, the dialysis membrane extended from the pia 6.5 mm into the cerebellum. The latter type of probe was used to measure levels of GABA released by Purkinje cells into the deep cerebellar nuclei and at the same time to stimulate Purkinje cell soma and dendrites located in the cerebellar cortex (18). Krebs-Ringer bicarbonate buffer was pumped (syringe pump; Harvard Apparatus) through the dialysis probe at a rate of 2 ,ul/min. After 60-80 min for equilibration, samples were collected every 10 min for the duration of the experi- ment. Four baseline samples followed by infusion of the different test substances for a 40-min period were collected in each experiment. A new baseline period for baseline recov- ery followed and this process was repeated up to three more times. Baseline levels were arbitrarily taken as 100% (see figures). Glutamate pulses were applied at 50 and 5 mM, IGF-I was applied at 100 and 1 nM, bFGF was applied at 100 nM, and KCI was applied at 100 mM. Electrical Stimulation of the Inferior Olive and Pontine . After implantation of the dialysis probe in the K cerebellar cortex, a monopolar electrode (0.2 mm diameter; a Rhodes Medical Instruments, Tujunga, CA) was placed into the contralateral inferior olive (AP = -12.5, L = 0.7, V = 9.0) or into the contralateral pontine nucleus (AP = -7.0, L = 1.0; V = 9.5). Placement of the electrode was aided by recording field potentials (P15 Preamplifier; Grass) from the molecular layer with an electrode placed adjacent to the dialysis probe. Monophasic cathodic pulses (3-msec pulse) of 100, 200, and 400 ,uA at a frequency of 5 or 20 Hz were applied for 10 min. Samples for IGF-I determination were collected every 10 min all throughout the procedure. b. ~~~~~4 When inferior olive stimulation was applied in conjunction with glutamate (5 mM), the electrical stimulation lasted 20 FIG. 1. Representative photographs showing the position at min (corresponding to two dialysates of 20 ,ul) and consisted which the microdialysis probe and the microelectrode were placed in of 3-msec pulses at 5 Hz and 300 ILA. the cerebellum (the tract of the probe is indicated by arrowheads) (a) HPLC of GABA. Cerebellar dialysates were analyzed for and in the inferior olive (arrowhead) (b), respectively. GABA content by HPLC (Perkin-Elmer) in a 5-p.m C18 column and a nonlinear gradient with 0.5 M phosphate buffer nucleus, elicited a significant increase in IGF-I levels in the (pH 5.5) and 100o methanol as described by Joseph and dialysate obtained from the cerebellar cortex. The increase Marsden (20). was slightly more effective at 5 Hz as compared to 20 Hz (P IGF-I Radioimmunoassay. The assay procedure has been described in detail (21). Polyclonal rabbit anti-IGF-I (UB3- 189) was a generous gift of The National Pituitary Agency of National Institutes of Health, Chapel Hill (L. Underwood co) 150 and J. Van Wyk). The sensitivity of the assay was 125 pg of 70) IGF-I per ml. Microdialysis samples were directly assayed. Statistics. A one-way analysis of variance followed by a 130 * post-hoc Tukey's test was used when comparing several so Iw groups. A Student's t test was used to determine significant differences between two groups. 0)0 RESULTS Fig. 1 b and a illustrates the localization of the stimulating 90 electrode in the inferior olive and of the microdialysis probe 1 2 3 4 5 6 in the cerebellum, respectively, from two representative Sample A animals. The recovery of GABA through the microdialysis FIG. 2. Electrical stimulation of the inferior olivary nucleus probe was 15.1 ± 0.63% (mean ± SEM) and the recovery of resulted in an increase in the levels of IGF-I in the molecular layer IGF-I was 6.2 + 3.2%. These recovery rates are similar to of the cerebellum. Electrical pulses of 5 (open circles) and 20 (solid those obtained by other authors (4). Values shown were not circles) Hz with current intensities of 100-400 uA elicited significant corrected for percent of recovery. The mean basal level of increases in immunoreactive IGF-I levels in the microdialysate. No GABA in the dialysates was 647 ± 77 pg/20 ,ul and of IGF-I changes in IGF-I levels were found after stimulation of the pontine was 5.9 ± 0.17 pg/20 p.l. Basal levels ofIGF-I were very close nuclei when using 5 (open triangles) and 20 (solid triangles) Hz. Since to the detection limits of the radioimmunoassay and in some no differences in the response of IGF-I to different current intensities samples below them. These were with similar frequencies were appreciated, they were pooled and samples arbitrarily assigned shown together. The electrical stimulation was given for 10 min a value corresponding to the detection limit of the assay (125 during collection of the fourth sample (arrowhead). The time corre- pg/ml). sponding to each sample is 10 min. Results shown are the mean ± Release of IGF-I in the Cerebeilar Cortex After Electrical SEM of 12 stimulations done in six animals with 5 Hz and of eight Stimulation of the Inferior Olive. As shown in Fig. 2, electrical challenges done in four animals with 20 Hz. *, P < 0.05 vs. stimulation of the inferior olive, but not of the pontine prestimulation levels. Downloaded by guest on September 25, 2021 7388 Neurobiology: Castro-Alamancos and Torres-Aleman Proc. Natl. Acad. Sci. USA 90 (1993) < 0.05 and P = 0.05, respectively), even though both verely impaired in stimulating GABA release when the neu- frequencies were effective in eliciting IGF-I release. Since rons were previously exposed to 100 nM but not 1 nM IGF-I current intensities ranging from 100 to 400 uA produced (Figs. 3 b and c and 4). This indicates a long-term effect of identical responses, results were pooled and are shown IGF-I, which is also dose-dependent. bFGF, another growth together. We also tried higher-frequency stimulation of the factor active in cerebellum (14), did not interfere in the pontine nucleus (300 Hz with a 0.2-msec pulse) and found no GABA response to glutamate stimulation (Figs. 3d and 4), change in IGF-I baseline levels in the contralateral cerebellar which reinforces the specificity of the effect of IGF-I on cortex. GABA release. Glutamate Stimulates GABA Release in Cerebellum. Ad- IGF-I Does Not Inhibit KCl-Induced GABA Release in ministration of 50 mM glutamate through the dialysis probe Cerebellum. To determine whether another type of stimula- inserted into the cerebellar cortex and deep cerebellar nuclei tion ofGABA release might be affected by IGF-I, we exposed elicited a 9-fold increase in GABA release (Fig. 3a). Subse- the cerebellar neurons to depolarizing concentrations of KCl quent pulses of50 mM glutamate produced identical peaks of (100 mM) in the presence and absence of 100 nM IGF-I. As GABA, suggesting that cerebellar neurons were not damaged shown in Fig. 5, IGF-I did not impair the response of GABA by prolonged exposure to glutamate and that cerebellar to repeated KCl stimulation. Furthermore, the GABA re- GABAergic neurons are able to respond to repeated gluta- sponse to KCl was completely normal even when the KCl mate challenges for long periods oftime without showing any pulse was applied to animals showing a depressed GABA decline. response to glutamate due to prior exposure to IGF-I (data Conjoint Administration of IGF-I and Glutamate Inhibits not shown). These data suggest a specificity of action for Glutamate-Induced GABA Release in the Cerebellum. Admin- IGF-I upon glutamate-dependent GABA responses. istration of IGF-I (1 and 100 nM) to the cerebellum through Stimulation ofthe Inferior Olive Inhibits Glutamate-Induced the microdialysis probe did not alter baseline GABA release. GABA Release in a Manner Similar to that Produced by IGF-I. Similarly, application of 100 nM IGF-I alone did not modify Fig. 6 shows the effect of conjoint application of glutamate the GABA response to glutamate challenges delivered sub- and electrical stimulation of the inferior olive (5 Hz, 300 ,uA, sequently (data not shown, but see Fig. 4a). On the contrary, 3-msec pulse, for 20 min) on glutamate-induced GABA when IGF-I was given simultaneously with glutamate a responses. Four consecutive pulses of 5 mM glutamate significant depression of the GABA response to glutamate produced consistent GABA peaks very similar to those found was found (P < 0.0002, Figs. 3b and 4). The inhibition was after the previously tested dose of 50 mM (data not shown, dose-dependent since 100 nM IGF-I was significantly more but see Fig. 3a). On the contrary, electrical stimulation ofthe effective than 1 nM (P < 0.002; Figs. 3 b and c and 4). inferior olive at the same time that the first glutamate pulse Furthermore, subsequent glutamate pulses were also se- was delivered elicited a depression of GABA responses to -a 1000 100011

800 800+ 6004 600 .

400 4004

(a 200 I ***... ** .....* - 200 to _ I

0- ...... ; U-nE.E T w w Glu Glu Glu Glu Glu Glu Glu Glu + IGF-I C)000 O' ' 0) b~~~~~~ 1000 I d Boo t 800

600+ 600 4001- 400t 200- 200

o I ...... U Ww w| . . . . .*...... WW ss *w ww ** ** ** ** ** ** * ** * **** **** **** **** Glu Glu Glu Glu Glu Glu Glu Glu + + IGF-I FGF

FIG. 3. Release of GABA from cerebellar neurons in response to glutamate and its modulation by IGF-I. (a) Repeated 50 mM pulses of glutamate (Glu) produced identical GABA peaks for at least four consecutive challenges. (b) A dose of 1 nM IGF-I impaired the GABA response to conjunctive administration of 50 mM glutamate but did not alter subsequent responses to glutamate pulses. (c) However, a dose of 100 nM IGF-I coadministered with 50 mM glutamate produced an almost complete blockage of the GABA peak and a drastic attenuation of subsequent glutamate challenges for at least three more pulses. (d) A dose of 100 nM bFGF did not alter the response of GABA to 50 mM glutamate coadministered with bFGF or to subsequent glutamate pulses. The time corresponding to each sample is 10 min. Pulses of each test substance were applied for a time corresponding to four samples (40 min). Results shown are from representative animals. Asterisks indicate pulses. Downloaded by guest on September 25, 2021 Neurobiology: Castro-Alamancos and Torres-Aleman Proc. Natl. Acad Sci. USA 90 (1993) 7389

a 500 T a) co 3000 a)

: < 300- 2000 - C) a) co ._-0) 75 iooo -l- 0 CD .- 100 .coo I0- 0 a . KCL KCL KCL KCL C 0- b + C6 IGF-I a) TI- co FIG. 5. KCl stimulation of cerebellar GABA release is not modified by IGF-I. No difference in the response of GABA to 100 L0) mM KCl was seen when given in combination with 100 nM IGF-I. < 2000 Similarly, the response of GABA to subsequent challenges of 100 mM KCl was not modified by previous exposure to IGF-I. Results shown are of a representative experiment. The pattern of GABA ** release was identical when 100 mM KCI was given alone (data not 1000 shown). This experiment was repeated with another animal, with identical results. The time corresponding to each sample is 10 min (tick marks on X axis). Pulses ofeach test substance were applied for a time corresponding to four samples (40 min, asterisks). 0 1 1 1 Glu Glu Glu Glu + + + contain IGF-I receptors (3, 9, 22). (ii) Developing Purkinje IGF-I IGF-I FGF neurons are specifically dependent on IGF-I to survive and (1 nM) (100 nM) (100 nM) differentiate in vitro (14). (iii) Purkinje cells constitute the main targets of climbing fibers, and we found depressed FIG. 4. Inhibition of glutamate-stimulated GABA release by GABA responses after conjoint stimulation ofclimbing fibers IGF-I is dose-dependent and long-lasting. (a) Conjoint application of and glutamate administration. (iv) From a mere quantitative IGF-I, but not bFGF, and glutamate (50 mM) significantly inhibited perspective, Purkinje cells constitute the majority of the glutamate-stimulated GABA release in a dose-dependent fashion, as compared to glutamate alone (Glu). (b) The release of GABA in GABAergic cells present in the cerebellum (18). response to subsequent challenges with 50 mM glutamate was Our results further indicate that both IGF-I and inferior significantly depressed after previous conjoint exposure to 100 nM, olive stimulation produced a long-lasting depression of but not 1 nM IGF-I, and glutamate, as compared to glutamate alone. GABA release by cerebellar cells in response to glutamate This indicates a long-lasting effect of IGF-I upon glutamate-induced when coadministered with this excitatory neurotransmitter. GABA release. bFGF had no effect. All the GABA peaks were Thus, a single exposure to 100 nM IGF-I not only blocked the pooled and shown together. Results shown are the mean ± SEM of peak of GABA in response to coadministered glutamate but at least three animals except in the case of conjoint application of also severely impaired at least three subsequent GABA bFGF where a single animal was tested. *, P = 0.0002; **, P < 0.0001 responses. The was for (vs. glutamate alone, by Tukey's test). inhibitory action of IGF-I specific glutamate since KCl stimulation of GABA release was un- subsequent glutamate challenges, without affecting the first altered. Moreover, the specificity of IGF-I was further sup- the that another factor peak of GABA (Fig. 6). This effect was similar, although not ported by fact bFGF, growth present in cerebellum (23), did not alter the GABA response to identical, to that seen after conjoint application of IGF-I and glutamate (see Figs. 3 and 4). glutamate. Similarly, electrical stimulation of the inferior olive elicited long-lasting effects on the GABA response to subsequent glutamate pulses. In this case, the depression of DISCUSSION GABA responses was smaller than when 100 nM IGF-I was used. These results show that IGF-I depresses glutamate-induced These findings open the intriguing possibility that IGF-I GABA release of cerebellar cells in a long-lasting and dose- may be involved in the learning-related phenomenon of dependent manner. Moreover, electrical stimulation of infe- cerebellar long-term depression (LTD, see ref. 24). LTD is rior olivary neurons that induces release of IGF-I in the characterized as a protracted inhibition of the glutamatergic cerebellar cortex also depresses glutamate-induced GABA synapse between parallel fiber afferents and Purkinje cell release of cerebellar cells in a way similar to that seen after dendrites that follows concurrent stimulation of climbing administration of IGF-I. These data indicate that IGF-I fiber and parallel fiber terminals (24). Thus, the final outcome released in the cerebellar cortex after climbing fiber stimu- of LTD is a long-lasting decrease in the sensitivity of the lation interacts with target GABAergic cells, probably the Purkinje cell to glutamate. What we have found now is that Purkinje neuron, to modulate their response to glutamate. IGF-I elicits a long-term depression in the response of Thus, IGF-I appears to fulfill a role as a neuromodulator in cerebellar GABAergic neurons to glutamate only when given the cerebellar cortex. simultaneously with glutamate. In addition, electrical stim- Although we still do not have direct evidence on which ulation ofthe inferior olive triggers the release of IGF-I in the GABAergic cells in the cerebellum are the ones that are cerebellar cortex and produces a long-term depression of the responding to the modulatory effects of IGF-I, the available GABA response to glutamate in a manner reminiscent to that evidence strongly supports that the Purkinje neuron is the produced by IGF-I when coadministered with glutamate. It is target cell for IGF-I. (i) Purkinje cells in the adult cerebellum important to point out that a hallmark of LTD is the necessity Downloaded by guest on September 25, 2021 7390 Neurobiology: Castro-Alamancos and Torres-Aleman Proc. Natl. Acad Sci. USA 90 (1993)

q1) a adult olivo-cerebellar system. A functional relationship be- co 1000 - tween the modulatory action of IGF-I on glutamatergic a) co transmission in the cerebellum with a plasticity-related mech- Q) 800 - anism such as LTD is also suggested.

600 - We thank Dr. A. Arevalo for her advice in the use of the HPLC co system. We acknowledge the gift of the polyclonal anti-IGF-I anti- a) 0) 400 body through the National Hormone and Pituitary Program of the a National Institutes of Health. We also thank Dr. L. M. Garcia- Segura for his criticism. This work has been funded by grants from coC. 200 0 Fondo de Investigaciones Sanitarias de la Seguridad Social (92/0283) 10- and Comunidad de Madrid (146/92). . v(I Xr . AA A AA A Glu Glu Glu Glu 1. Andersson, I. K., Edwall, D., Norstedt, G., Rozell, B., Skott- Acta + ner, A. & Hansson, H. A. (1988) Physiol. Scand. 132, 1.0. Stim. 167-173. 0) 2. Rotwein, P., Burgess, S. K., Milbrandt, J. D. & Krause, J. E. Ca 2000- b (1988) Proc. Natl. Acad. Sci. USA 85, 265-269. 3. Werther, G. A., Abate, M., Hogg, A., Cheesman, H., Oldfield, B., Hards, D., Hudson, P., Power, B., Freed, K. & Herington, 1500-. A. C. (1990) Mol. Endocrinol. 4, 773-778. 4. Yamaguchi, F., Itano, T., Mizobuchi, M., Miyamoto, O., Janjua, N. A., Matsui, H., Tokuda, M., Ohmoto, T., CD 1000 Hosokawa, K. & Hatase, 0. (1990) Brain Res. 533, 344-347. 5. Bach, M. A., Shen-Orr, Z., Lowe, W. L., Roberts, C. T. & LeRoith, D. (1991) Mol. Brain Res. 10, 43-48. 6. Bondy, C. A. (1991) J. Neurosci. 11, 3442-3455. 0, 500 7. Garcia-Segura, L. M., Perez, J., Pons, S., Rejas, M. T. & Torres-Aleman, I. (1991) Brain Res. 560, 167-174. Ca 8. Torres-Aleman, I., Pons, S. & Garcia-Segura, L. M. (1991) < 0 Glu Glu Glu Brain Res. 564, 348-351. + after 9. Bondy, C., Werner, H., Roberts, C. T. & LeRoith, D. (1992) 1.0. Stim. 1.0. Stim. Neuroscience 46, 909-923. 10. Lenoir, D. & Honegger, P. (1983) Dev. Brain Res. 7, 205-213. FIG. 6. Stimulation of the inferior olive together with glutamate 11. DiCicco-Bloom, E. & Black, I. B. (1988) Proc. Natl. Acad. Sci. depressed subsequent GABA responses in a manner reminiscent of USA 85, 4066-4070. that produced by IGF-I. (a) Costimulation of the inferior olive (I.O., 12. McMorris, F. A. & Dubois-Dalcq, M. (1988) J. Neurosci. Res. 5 Hz, 300 ,IA, 3 msec) and administration of glutamate (5 mM) 21, 199-209. depressed subsequent GABA responses to glutamate pulses. Results 13. Torres-Aleman, I., Naftolin, F. & Robbins, R. (1990) Neuro- shown are of a representative animal. The time corresponding to science 35, 601-608. each sample is 10 min. Pulses ofeach test substance were applied for 14. Torres-Aleman, I., Pons, S. & Santos-Benito, F. F. (1992) Eur. a time corresponding to two samples (20 min). (b) Stimulation of the J. Neurosci. 4, 864-869. inferior olive plus glutamate administration produced a significant 15. Berelowitz, M., Szabo, M., Frohman, L. A., Firestone, S., inhibition of the release of GABA in response to subsequent pulses Chu, L. & Hintz, R. L. (1981) Science 212, 1279-1281. of glutamate. The effect of inferior olive stimulation was similar to 16. Araujo, D. M., Lapchak, P. A., Collier, B., Chabot, J. G. & that seen with IGF-I, as shown in Figs. 3 and 4. The peaks ofGABA Quirion, R. (1989) Brain Res. 484, 130-138. in response to subsequent pulses of glutamate were pooled and 17. Sara, V. R., Uvnas-Moberg, K., Uvnas, B., Hall, K., Wetter- shown together. Results are the mean ± SEM of three control rats berg, L., Posloncec, B. & Goiny, M. (1982) Acta Physiol. and three inferior olive costimulated rats. * and **, P < 0.05 (vs. Scand. 115, 467-470. glutamate and glutamate plus inferior olive stimulation, respective- 18. Ito, M. (1984) The Cerebellum and Neural Control (Raven, ly). New York). 19. Paxinos, G. & Watson, C. (1982) The Rat Brain in Stereotaxic of a temporal association ofthe stimulation of climbing fibers Coordinates (Academic, New York). and parallel fibers (or glutamate administration). Thus, our 20. Joseph, M. H. & Marsden, C. A. (1986) in HPLC of Small results support the possibility that IGF-I is participating as a Molecules, ed. Lim, C. K. (IRL, Oxford), pp. 13-27. 21. Pons, S., Rejas, M. T. & Torres-Aleman, I. (1991) Dev. Brain climbing fiber messenger in LTD of the cerebellar cortex Res. 62, 169-175. since IGF-I modulates the GABA response to glutamate only 22. Komoly, S., Hudson, L. D., Webster, H. DeF. & Bondy, C. A. when given simultaneously with glutamate. (1992) Proc. Natl. Acad. Sci. USA 89, 1894-1898. In conclusion, IGF-I, a classical growth factor for devel- 23. Riva, M. A. & Mocchetti, I. (1991) Dev. Brain Res. 62, 45-50. oping nerve cells, appears to act as a neuromodulator in the 24. Ito, M. (1989) Annu. Rev. Neurosci. 12, 85-102. Downloaded by guest on September 25, 2021