Diazepam protects against rat hippocampal neuronal cell death induced by antisense oligodeoxynucleotide to GABAA receptor γ2 subunit

Brain Research 765 Ž1997. 21–29

Research report

Diazepam protects against rat hippocampal neuronal cell death induced by antisense oligodeoxynucleotide to GABA A receptor g 2 subunit Jesper Karle ) , Michael Robin Witt, Mogens Nielsen Research Institute of Biological Psychiatry, St. Hans Hospital, DK-4000 Roskilde, Denmark Accepted 1 April 1997

Abstract Antisense oligodeoxynucleotides ŽODNs. are used for the selective inhibition of gene expression. Antisense ODNs are promising tools for the investigation of physiological implications of proteins in the central nervous system of rodents in vivo. We have previously demonstrated that a phosphorothioate antisense ODN to the GABA A receptor g 2 subunit, but not sense or mismatch control ODNs, induces a decrease in ex vivo benzodiazepine receptor radioligand binding in rat hippocampus when infused into the hippocampus in vivo wKarle et al., Neurosci. Lett., 202 Ž1995. 97–100x. This effect is parallelled by a decrease in the number of GABA A receptors and an extensive loss of hippocampal neurones. There is increasing awareness of risks of toxic ‘non-antisense’ effects induced by ODNs, and in particular phosphorothioate ODNs. The present experiments were designed to investigate the specificity of effects induced by the g 2 subunit antisense ODN. The temporal development of changes in w 3 Hxflunitrazepam and w 3 Hxquinuclidinyl benzilate binding as well as in tissue protein levels supports the notion that the antisense ODN primarily acts by blocking the expression of the targeted receptor subunit protein. Furthermore, it is shown that a threshold for the elicitation of neurodegenerative changes exists. Finally, it is demonstrated that diazepam treatment of rats protects against the development of neuronal cell death induced by the antisense ODN. Collectively, the results support the hypothesis that the neurodegeneration induced by the antisense ODN is a consequence of diminished GABAergic inhibitory tonus following a selective down-regulation of g 2 subunit-containing GABA A receptor complexes. q 1997 Elsevier Science B.V. Keywords: GABA A receptor; g 2 subunit; Antisense oligodeoxynucleotide; Diazepam; Rat hippocampus; Radioligand binding

1. Introduction Antisense oligodeoxynucleotides ŽODNs. have been established as useful tools for the selective inhibition of gene expression in the brain of experimental animals in vivo Žfor reviews see w16,33,45x.. Several reports have demonstrated marked effects of antisense ODNs targeted to different neuronal proteins, e.g., neurotransmitter receptors w46,47,51x. The most important inhibitory neurotransmitter in the mammalian brain is g-aminobutyric acid ŽGABA.. GABA A receptors are multi-subunit complexes assembled from different combinations of polypeptide subunits of which several families and isoforms have been identified Ž a 1–6, b 1–4, g 1–3, d and r 1–2. Žfor reviews see w26,28,38,43x.. Each subunit isoform is encoded by a different gene.

) Corresponding author. Fax: Ž45. Ž46. 334-367; E-mail: [email protected]

0006-8993r97r$17.00 q 1997 Elsevier Science B.V. All rights reserved. PII S 0 0 0 6 - 8 9 9 3 Ž 9 7 . 0 0 4 6 3 - 0

Previous results obtained in our laboratory with an antisense ODN targeting the GABA A receptor g 2 subunit w19,20x are consistent with the notion that a g 2 subunit is an important constituent of GABA A receptors in the rat hippocampus w11,49x. The presence of a g 2 subunit is necessary for the detection of a classical benzodiazepine binding site within a GABA A receptor complex w29,34x. We have shown that an 18-mer phosphorothioate ODN targeted to the mRNA encoding the rat GABA A receptor g 2 subunit can markedly decrease benzodiazepine receptor radioligand binding in rat hippocampus w19x. Parallel decreases in the binding of radioligands to the GABA binding site and the ion channel domain of the GABA A receptor have indicated that blocked expression of the g 2 subunit in the rat hippocampus leads to a decrease in the number of GABA A receptors in that region. This may be a result of impaired or incomplete assembly of receptor complexes due to the lack of a g 2 subunit. A severe loss of hippocampal neurones following continuous antisense ODN treatment for 5 days may be consistent with a

22

J. Karle et al.r Brain Research 765 (1997) 21–29

decrease in GABAergic inhibitory input to the neurones following GABA A receptor down-regulation. Control experiments carried out with either sense or mismatch ODN have demonstrated the nucleotide sequence specificity of the effect induced by the g 2 subunit antisense ODN w19x. Recently, Zhao et al. w52x found an increase of the seizure threshold for a benzodiazepine inverse agonist, but not for picrotoxin, in rats treated with intracerebroventricular Ži.c.v.. injections of a 17-mer phosphorothioate antisense ODN with a sequence closely resembling the one applied in our laboratory. Although successfully applied in numerous reported studies, the antisense technique is still in its infancy. The precise mechanism of action of antisense ODN is poorly understood. Presumably, antisense ODN inhibits protein synthesis upon hybridization to its targeted, complementary mRNA sequence w14,25x. There are several obstacles and pitfalls related to the use of the antisense technique w6,41,44x, that make results obtained with antisense ODNs, and in particular with phosphorothioate analogues, difficult to interpret w23,42x. There is growing awareness of both nucleotide ‘sequence independent’ and ‘sequence dependent’ non-antisense actions of phosphorothioate antisense ODNs, i.e., actions which cannot be attributed to a mechanism involving hybridization between ODN and a complementary mRNA w23,40,42x. Sequence independent toxic actions of phosphorothioate ODNs have been reported after prolonged administration into brain tissue w6,16x or after i.c.v. injections at high concentration w48x. Some phosphorothioate ODNs have been shown in vitro to directly interact with proteins, activate lymphocytes or exert antiproliferative actions w23,24x. The nature of the effects induced by the antisense ODN to the GABA A receptor g 2 subunit w19,20x, i.e., neurodegeneration, warrants thorough investigation in order to verify that the effects can be explained by a ‘true’ antisense mechanism. The aim of the present study was to investigate whether the changes induced by the GABA A receptor g 2 subunit antisense ODN in the hippocampus can be attributed to a blocked expression of the targeted gene. The temporal development of changes in hippocampal membrane binding of the benzodiazepine receptor radioligand w 3 Hxflunitrazepam Žw 3 HxFNM. was measured. Also, binding of a ligand to a receptor not targeted by the antisense ODN, the muscarinic acetylcholine receptor ligand w 3 Hxquinuclidinyl benzilate Žw 3 HxQNB. as well as changes in membrane protein content were estimated. Furthermore, experiments were designed to determine whether a threshold for the induction of deleterious neuronal effects by the antisense ODN exists. Finally, it is demonstrated that the development of antisense ODN induced hippocampal neurodegeneration can be prevented by systemic administration of diazepam, a benzodiazepine receptor agonist. The results support the notion that the initial event in the antisense ODN induced neuronal cell death is a reduc-

tion of the number of GABA A receptors in the hippocampus.

2. Materials and methods 2.1. Oligodeoxynucleotides All ODNs used in the present experiments were fully phosphorothioate modified. The antisense ODN is complementary to a region which spans the putative 5X translation start codon ŽAUG. of the mRNA encoding the rat GABA A receptor g 2 subunit Ž5X-TAT-TTG-GCG-AAC-TCA-TCG3X . w18–20,37x. To assess the specificity of the effects induced by the antisense ODN, two different control ODNs were used: a four-base mismatch ODN ŽMM4. in which the sequence of the four central bases of the antisense ODN was interchanged Ž5X-TAT-TTG-GAA-GCC-TCATCG-3X . or a two-base mismatch ODN ŽMM2., in which the two central bases of the antisense ODN sequence were interchanged Ž5X-TAT-TTG-GCA-GAC-TCA-TCG-3X .. These control ODNs were chosen in order to use ODN sequences resembling the antisense sequence as closely as possible. ODNs were manufactured by b-cyanoethyl phosphoramidite chemistry and phosphorothioate-modified by the Beaucage reagent ŽDNA Technology, Aarhus, Denmark.. ODNs were purified by reverse phase HPLC, ethanol precipitated and finally dissolved in sterile H 2 O for intrahippocampal infusion. The ODN sequences have been controlled for homologies to known rodent gene sequences in the EMBL database w13,19x. 2.2. Infusion of ODN into rat hippocampus Antisense or control ODN was infused continuously into the right hippocampus of male Wistar rats Žweight 225–250 g. ŽMoellegaard Breeding, Lille Skensved, Denmark. as previously described w19x. Briefly, rats were anaesthetized Žsodium pentobarbital, 50 mgrkg body weight intraperitoneally Ži.p... and a cannula was stereotactically implanted centrally in the hippocampus ŽA s 3.0 mm; L s q4.3 mm from bregma; depth: 5.0 mm inferior to skull.. The cannula was connected to an Alzet osmotic minipump ŽAlza, Palo Alto, CA, USA. via a polyethylene catheter. The untreated left hippocampi Žcontralateral to the ODN treated side. were used as internal controls. 2.3. Expt. 1: antisense effects as a function of ODN infusion time Groups of rats Ž n s 4–6 in each group. were infused continuously with antisense ODN Ž1.7 m grm l, 0.5 m l infusionrh. into the right hippocampus for periods of 1–7 days. Separate groups of animals Ž n s 9. received control mismatch ODN ŽMM4 or MM2. in the same concentration for a period of 5 days. Rats were sacrificed by decapitation

J. Karle et al.r Brain Research 765 (1997) 21–29

immediately after the scheduled infusion time. The brains were quickly removed and the right and left hippocampi dissected free and weighed. Correct placement of the infusion cannula in the right hippocampus was verified and the macroscopical appearance of individual hippocampi was noted. Individual hippocampi were kept cold Ž0–48C. until further processing on the same day. Binding of w 3 HxFNM Ž0.5 nM. and w 3 HxQNB Ž0.5 nM. was estimated in membrane preparations Ž2 mg original tissuerml. of both ODN treated and contralateral hippocampi as previously described w32x. Midazolam Ž10y5 M. and atropine Ž5 = 10y6 M. were used to define non-specific binding of w 3 HxFNM and w 3 HxQNB, respectively. Protein content in individual membrane preparations was estimated ŽBioRad w . using bovine serum albumin as standard.

23

2.6. Statistical analysis In Expt. 1 a two-way analysis of variance ŽANOVA. was carried out for each measured parameter with infusion day and treatment Žantisense vs. control. as factors. Subsequently, a one-way ANOVA was performed for each day in order to determine differences between values of antisense ODN treated hippocampi and control values. Data passed tests of normality and equal variance. In the experiments involving mismatched control ODNs and in Expt. 2 groups were compared by the Mann–Whitney U-test.

3. Results 3.1. Expt. 1: antisense effects as a function of time

2.4. Expt. 2: reÕersibility of effects induced by antisense ODN; threshold for cell injury Antisense ODN was infused into the right hippocampus Ž1.7 m grm l, 0.5 m l infusionrh. of two groups of rats Ž n s 6 in each group. for either 2 or 3 consecutive days. At the end of the scheduled infusion time, ODN infusion was terminated by explantation of osmotic minipumps and polyethylene catheters from rats under light ether anaesthesia. Infusion cannulae were left in situ. On day 6 post surgery, i.e., 4 or 3 days, respectively, after discontinuation of ODN infusion, rats were sacrificed and binding experiments with w 3 HxFNM and w 3 HxQNB as well as estimation of tissue protein content were carried out on hippocampal membrane preparations as described above.

To characterize the development over time of effects induced by the GABA A receptor g 2 subunit antisense ODN following continuous intrahippocampal infusion, changes in hippocampal membrane binding of w 3 HxFNM and w 3 HxQNB as well as changes in protein content in membrane preparations were determined in groups of rats treated with antisense ODN for 1–7 days. The changes in radioligand binding and in protein content in antisense treated hippocampi compared to the contralateral Žuntreated. hippocampi are shown as a function of infusion time in Fig. 1. A two-way ANOVA revealed that the time course of each measured parameter is significantly different from that of the control contralateral hippocampi. A gradual decrease was found in the specific binding of w 3 HxFNM throughout the duration of the experiment. The

2.5. Expt. 3: antisense ODN infusion combined with diazepam treatment Rats were infused continuously with antisense ODN Žsame dosage as above. into the right hippocampus for 5 days. One group of rats received one daily i.p. injection of diazepam ŽApozepam w , AL Pharma, Denmark. Ž10 mgrkg body weight. from the day before the start of ODN infusion Žday y1. until the day before sacrifice Žday 4 post surgery.. Control rats received the similar amount of antisense ODN without other treatment. Following the scheduled infusion time animals were perfused transcardially with NaCl Ž0.9%. followed by paraformaldehyde Ž4% in 0.1 M phosphate buffer; pH 7.4. under sodium pentobarbital anaesthesia, prior to decapitation. Whole brains were dissected free and postfixed in paraformaldehyde Ž4%. for approx. 1 week. Whole brains were quickly frozen and cut in 30 m m thick horizontal sections by means of a microtome. The brain sections were fixed and haematoxylin-eosin stained according to standard procedures for histological examination. The experiment was repeated twice involving a total of 12–16 rats in each group.

Fig. 1. Effect of intrahippocampal antisense ODN infusion on w 3 HxFNM binding ŽB., w 3 HxQNB binding Žv . and membrane protein concentration Žw. as a function of time. For experimental details refer to Section 2. Data are presented as mean"S.D. Ž ns 4–6 in each group of rats. and shown as percentage of results from contralateral control hippocampi. A two-way ANOVA revealed that curves are significantly different from those of control hippocampi, i.e., there is a significant interaction effect between days and treatment Žantisense vs. control. Ž P - 0.05.. ) P - 0.05 compared to results obtained with control hippocampi Žone-way ANOVA..

24

J. Karle et al.r Brain Research 765 (1997) 21–29

tissue changes. Decreases in protein content therefore account for a major part of the decreases in radioligand binding found in the later stages of Expt. 1. Infusion of mismatch control ODN Žeither MM4 or MM2. for 5 consecutive days had no significant effect on w 3 HxFNM binding ŽFig. 2.. Similarly, the control ODNs had no effect on w 3 HxQNB binding or tissue protein content Ždata not shown.. We have previously found a 12% reduction of w 3 HxFNM binding induced by MM4 w19x; this may be attributed to sequence independent toxic effects of the infused control ODN. A representative brain section from a rat treated with MM2 for 5 days is shown in Fig. 3. The MM2 treated hippocampus appears intact except for a small cell deficit reflecting local injury at the site of ODN infusion. 3.2. Expt. 2: reÕersibility of effects induced by antisense ODN; threshold for cell injury

Fig. 2. w 3 HxFNM Ž0.5 nM. binding to hippocampal membrane preparations following intrahippocampal infusion of antisense or control mismatch ŽMM4, MM2. ODN for 5 days. Data are expressed as percentage of untreated controls Žmean"S.D.; ns9.. Absolute value for untreated control: 5990"530 dpmr2 mg tissue. ) P - 0.01 vs. untreated control, MM4 or MM2 ŽMann–Whitney U-test.; n.s., not significantly different from untreated control.

decrease is statistically significant Ž P - 0.05, one-way ANOVA. at 2 days of antisense ODN infusion; a tendency towards a decrease was observed after 1 day. Previously, Scatchard analyses have shown that the reduction in w 3 HxFNM binding is mainly due to a reduced Bmax value w19x. The binding of w 3 HxQNB and tissue protein content were unchanged during the first 2 days of antisense treatment, but decreased subsequently. The occurrence of these changes coincides with the development of histological signs of neuronal cell death, which occur after approx. 4 days of antisense infusion Ždata not shown.. The radioligand binding data are presented as dpm per 2 mg original tissue. Protein concentration is shown as an individual parameter which probably reflects degenerative

Table 1 shows the results of hippocampal membrane binding of w 3 HxFNM and w 3 HxQNB as well as tissue protein content estimated when intrahippocampal antisense ODN infusion was discontinued after 2 or 3 days, and rats were left untreated for 4 or 3 days, respectively, until sacrifice Žday 6 post surgery.. For comparison, data from rats infused continuously for a total of 6 days are shown in Table 1. Two days of antisense treatment are sufficient to cause a decrease in w 3 HxFNM binding, when measured immediately after the treatment ŽFig. 1.. After 2 days of antisense ODN infusion followed by a 4 day infusion-free period, there were no changes in radioligand binding or in tissue protein content. This indicates that following 2 days of treatment the effects induced by the antisense ODN are reversible. In contrast, 3 days of antisense treatment elicited pronounced decreases in the binding of both radioligands as well as a marked decrease in tissue protein concentration. These changes remained apparent after the following 3 day infusion-free period and were similar to those obtained after 6 days of antisense ODN treatment ŽTable 1.. These results indicate that 3 days of antisense treatment had induced irreversible effects in the hippocampus and that the ‘point of no return’ for the induction of irreversible effects is between 72 and 96 h of antisense ODN treatment. 3.3. Expt. 3: antisense ODN infusion combined with diazepam treatment Fig. 3 shows a section of a rat brain treated with intrahippocampal antisense ODN infusion for 5 days. A

Fig. 3. Representative photomicrographs showing horizontal sections from rat brains infused into the right hippocampus with antisense ODN ŽA,B. or mismatch ODN MM2 ŽC. for 5 days. The brain section shown in A is from a rat which did not receive treatment apart from the antisense ODN infusion. A widespread loss of neurones and intense infiltration with monocytermacrophage-like cells can be seen throughout the right hippocampal formation. The ODN infusion site is indicated by the arrow. B: a brain section from a rat which received daily injections of diazepam Ž10 mgrkg i.p.. during the antisense ODN treatment period. Only slight tissue changes are visible in the vicinity of the infusion site Žarrow.. C: a minor deficit immediately adjacent to the control mismatch ODN ŽMM2. infusion site Žarrow.. Scale bar: 0.33 mm.

J. Karle et al.r Brain Research 765 (1997) 21–29

25

J. Karle et al.r Brain Research 765 (1997) 21–29

26

Table 1 Threshold for reversibility of effects induced by antisense ODN to GABA A receptor g 2 subunit in rat hippocampus Duration of antisense ODN infusion 2 days w 3 HxFNM binding w 3 HxQNB binding Protein conc. Ž m grmg.

3 days

6 days

Antisense

Contralat.

Antisense

Contralat.

Antisense

Contralat.

6590 " 700 10 830 " 2180 63 " 6

7200 " 570 12 860 " 1160 69 " 4

4190 " 670 a Ž57%. 5140 " 850 a Ž47%. 44 " 3 a

7370 " 450 10 980 " 1320 60 " 3

4050 " 360 a Ž57%. 6640 " 520 a 50 " 3 a

7060 " 540 11 940 " 1190 65 " 4

Results of binding experiments with w 3 HxFNM and w 3 HxQNB as well as protein concentration in membrane preparations from antisense ODN treated and contralateral Žcontrol. hippocampi. Antisense ODN was infused continuously into the right hippocampus for 2, 3 or 6 days, respectively; after the 2 or 3 day infusion period the infusion was discontinued by removal of osmotic minipumps. All rats were sacrificed 6 days after surgery Ži.e., 4 or 3 days after discontinuation of antisense ODN infusion, respectively. Data are expressed as mean " S.D. Ž n s 6.. Binding data are shown as dpmr2 mg original tissue. a P s 0.001 vs. contralateral, Mann–Whitney U-test.

marked loss of neurones as well as intense infiltration with monocytermacrophage-like cells are visible throughout the ipsilateral hippocampal formation. The changes are similar to those previously described w19,20x. The reproducibility and the reliability of the neuronal cell loss have been confirmed in a number of subsequent experiments.

Diazepam treatment was able to prevent the development of hippocampal neuronal cell death when administered once daily to rats which concurrently received intrahippocampal antisense ODN infusion ŽFigs. 3 and 4.. In the right hippocampus of the majority of diazepam treated rats Ž9r12. there were virtually no histological changes except for a lesion corresponding to the ODN infusion site and cellular changes in its immediate surroundings ŽFig. 3.. In two brains from diazepam treated rats evidence was found of some protection against the deleterious effects of the antisense ODN ŽFig. 4., while in one brain there was no sign of neuroprotection by diazepam. In a minority Ž3r16. of brains treated with antisense ODN alone there was no or partial cell loss ŽFig. 4., probably due to technical flaws.

4. Discussion

Fig. 4. Comparison between the histological examinations of hippocampal tissue damage Žreflecting neuronal cell loss. in brain sections from rats treated with intrahippocampal antisense ODN infusion with or without concurrent i.p. diazepam treatment. Results are expressed as percentage of the total number of rats in each group showing no, partial or total damage Žantisense: ns16; antisenseqdiazepam: ns12.. Refer to text for explanation.

The reported experiments characterize changes following intrahippocampal infusion of a phosphorothioate ODN antisense to the GABA A receptor g 2 subunit mRNA. The results of Expt. 1 demonstrate that the antisense ODN induces a reduction in benzodiazepine receptor radioligand Žw 3 HxFNM. binding which is positively correlated to the duration of antisense ODN infusion. This supports the notion that a blockade of the expression of the g 2 subunit leads to a down-regulation of benzodiazepine binding sites. The muscarinic acetylcholine receptor radioligand w 3 HxQNB was chosen in order to estimate the binding to a receptor not targeted by the antisense ODN treatment. Changes in w 3 HxQNB binding may therefore serve as an indicator of secondary changes in the brain tissue. In a previous study w19x, the binding of w 3 HxQNB was decreased in hippocampal membranes following 5 days of antisense ODN treatment, although to a lesser extent than the binding of ligands specific for the GABA A receptor complex. To elucidate the mechanism of action of the antisense ODN we wanted to determine whether the antisense ODN primarily had an effect on GABA A receptors,

J. Karle et al.r Brain Research 765 (1997) 21–29

as indicated by a change in w 3 HxFNM receptor binding, and whether reduced w 3 HxQNB binding as well as tissue injury were secondary to a change of GABA A receptors. The decrease in w 3 HxFNM binding observed from day 2 as opposed to the delayed occurrence of decreased w 3 HxQNB binding and tissue protein concentration suggests that the latter changes are secondary to a GABA A receptor-specific effect. The mechanism of action might be a decreased GABAergic tonus caused by a selective down-regulation of functional GABA A receptors, as previously hypothesized w19x. To validate the effects obtained with the antisense ODN, we have applied control ODNs with nucleotide sequences which resemble that of the antisense ODN as closely as possible. The finding that the mismatched control ODNs did not induce effects comparable to the effects induced by the antisense ODN has confirmed the nucleotide sequence specificity of the effect. The resemblance between the antisense ODN and the applied control ODNs makes it very unlikely that ‘sequence dependent’ non-antisense actions are responsible for the effects induced by the antisense ODN. An ODN with a change of only two bases ŽMM2. had no significant effect. This result supports the notion that the antisense ODN presumably exerts its action at the level of mRNA, and suggests that unknown mRNAs with homologies to the target mRNA have not been affected by the treatment. It is plausible that a two-base mismatch can result in the lack of activity of an ODN; on average one mismatched base pair results in a 500-fold reduction of hybridization affinity w7,45x. Reversibility of antisense effects has been reported Že.g. w27x., illustrating the temporary profile of inhibition of gene expression by means of antisense ODN. In accordance with the presumed mechanism of action, an antisense effect should be reversible, unless it leads to irreversible secondary changes. This is most likely the case in the present experiments, in which the blocked expression of the targeted protein leads to neurodegeneration. The degree of reversibility of the effects induced by the antisense ODN to the GABA A receptor g 2 subunit was estimated in Expt. 2. Based on the patterns of time curves presented in Fig. 1 we hypothesized that a threshold for the induction of processes leading to neuronal cell death exists. Since the decrease in binding to the muscarinic acetylcholine receptor and the loss of tissue protein appear after 3–4 days of antisense ODN treatment ŽFig. 1., the threshold for the induction of irreversible effects was expected to be found within that timeframe. Phosphorothioate antisense ODNs have been shown to be intact for at least 24 h in the presence of rat brain tissue w48x and found to have a half-life of approx. 19 h in rat cerebrospinal fluid w5x. Therefore, a minimum of 24 h should be added to the ODN infusion time in order to estimate the total length of the antisense ODN exposure in Expt. 2. Based on the results obtained after discontinuation of antisense treatment ŽTable 1. it is concluded that a threshold for tissue

27

injury does exist, and that the point of no return is after 3–4 days of antisense ODN treatment. It is not possible to precisely estimate the decrease in benzodiazepine receptors or in the g 2 subunit per se at the critical point, since after 3–4 days of antisense treatment it is no longer possible to distinguish between primary and secondary effects in the affected tissue ŽFig. 1.. Based on the results of Expts. 1 and 2 and assuming an even distribution of the antisense ODN in the majority of the hippocampus w50x, we speculate that a decrease in the number of benzodiazepine receptors in the order of 25% is necessary for the induction of processes leading to neuronal cell death in the hippocampus. We believe that the antisense induced neuronal cell death is not a result of a direct neurotoxic action of the antisense ODN. The development of neuronal deleterious effects is more likely dependent on a down-regulation of receptors below a threshold. This notion is supported by the need for continuous delivery of ODN in order to induce neuronal cell loss. This is in contrast to the hippocampal neurodegeneration induced by a single injection of an excitotoxin, e.g., kainic acid w2,39x, quinolinic acid w21x or ibotenic acid w22x, or tetanus toxin w1x. Also, behavioural changes of antisense ODN treated rats develop subtly ŽJ. Karle, submitted. as opposed to the seizural activity, which often occurs shortly after intrahippocampal toxin treatment w1,3,10x. The differences are most likely not explained merely by limited distribution of the infused antisense ODN, as it has been shown that phosphorothioate ODN can distribute to the entire dorsal hippocampus after one intrahippocampal bolus injection w50x. In Expt. 3 it was investigated whether the deleterious effects of the antisense ODN could be prevented by simultaneous treatment of rats with a benzodiazepine receptor agonist. It was shown that diazepam could effectively protect against the development of neuronal cell damage ŽFigs. 3 and 4.. Diazepam may counteract the effects induced by the g 2 subunit antisense ODN by several mechanisms. A straightforward explanation would be that diazepam exerted its effect via a potentiation of GABAgated chloride influx through remaining GABA A receptor regulated ion channels, thereby restoring an inhibitory input sufficient for neuronal survival. Diazepam has been reported to protect CA1 hippocampal neurones against degeneration induced by ischaemia w17,36x. There is no general consensus on the effects of chronic benzodiazepine treatment on the number and properties of GABA A rbenzodiazepine receptors Žfor review see w9x.. Both a decrease w30,35x and an increase w8x as well as no alteration w4,12,31x of benzodiazepine receptor radioligand binding have been found. Recent results with different experimental designs in vivo w52x as well as in vitro w53x using g 2 antisense ODNs very similar to the one applied in our laboratory support the usefulness of antisense strategies for the characterization of GABA A receptors. The validation of the antisense

28

J. Karle et al.r Brain Research 765 (1997) 21–29

technique as a tool to specifically block the expression of different GABA A receptor subunits may lead to the dissection of the GABA A receptor complex at the molecular level and provide insight into the structural and functional roles played by specific receptor subunits w26x. This would help us understand the implications of different GABA A receptor subtypes in the brain and lead to improved, receptor subunitrsubtype selective approaches for pharmacological interference with the GABA A receptors. Improvement of in vivo antisense ODN technology may also have implications for antisense approaches to gene therapy of central nervous system disorders, e.g., brain tumours w15x.

Acknowledgements The authors wish to thank Ms. H. Soendergaard, Ms. G. Jensen, Ms. G. Ward and Mr. J. Raun for skilful technical assistance. The study was supported by the Danish Hospital Foundation for Medical Research, Region of Copenhagen, the Faroe Islands and Greenland, the Lundbeck Foundation, PharmaBiotec Research Center, the Eli Lilly Fund for Psychiatric Research and the Carlsberg Foundation.

References w1x G. Bagetta, M.T. Corasaniti, G. Nistico, ` N.G. Bowery, Behavioural and neuropathological effects produced by tetanus toxin injected into the hippocampus of rats, Neuropharmacology 29 Ž1990. 765–770. w2x Y. Ben-Ari, Limbic seizure and brain damage produced by kainic acid: mechanisms and relevance to human temporal lobe epilepsy, Neuroscience 14 Ž1985. 375–403. w3x P. Britton, P.S. Whitton, L.J. Fowler, N.G. Bowery, Tetanus toxininduced effects on extracellular amino acid levels in rat hippocampus: an in vivo microdialysis study, J. Neurochem. 67 Ž1996. 324–329. w4x C. Bræstrup, M. Nielsen, R.F. Squires, No changes in rat benzodiazepine receptors after withdrawal from continuous treatment with lorazepam and diazepam, Life Sci. 24 Ž1979. 347–350. w5x J.M. Campbell, T.A. Bacon, E. Wickstrom, Oligodeoxynucleoside phosphorothioate stability in subcellular extracts, culture media, sera and cerebrospinal fluid, J. Biochem. Biophys. Methods 20 Ž1990. 259–267. w6x B.J. Chiasson, J.N. Armstrong, M.L. Hooper, P.R. Murphy, H.A. Robertson, The application of antisense oligonucleotide technology to the brain: some pitfalls, Cell. Mol. Neurobiol. 14 Ž1994. 507–521. w7x S.T. Crooke, Progress toward oligonucleotide therapeutics: pharmacodynamic properties, FASEB J. 7 Ž1993. 533–539. w8x P. DiStefano, K.R. Casse, D. Colello, H.B. Boxman, Increased specific binding of 3 H-diazepam in rat brain following chronic diazepam administration, Cell Biol. Int. Rep. 3 Ž1979. 163–167. w9x M. Farrant, T.T. Gibbs, D.H. Farb, Molecular and cellular mechanisms of GABArbenzodiazepine-receptor regulation: electrophysiological and biochemical studies, Neurochem. Res. 15 Ž1990. 175– 191. w10x E.D. French, C. Aldinio, R. Schwarcz, Intrahippocampal kainic acid, seizures and local neuronal degeneration: relationships assessed in unanesthetized rats, Neuroscience 7 Ž1982. 2525–2536.

w11x J.-M. Fritschy, H. Mohler, GABA A receptor heterogeneity in the adult rat brain: differential regional and cellular distribution of seven major subunits, J. Comp. Neurol. 359 Ž1995. 154–194. w12x D.W. Gallager, J.M. Lakoski, S.F. Gonsales, S.L. Rauch, Chronic benzodiazepine treatment decreases postsynaptic GABA sensitivity, Nature 308 Ž1984. 74–77. w13x GCG, Program Manual for the Wisconsin Package, Genetics Computer Group, Madison, WI, 1994. w14x M.K. Ghosh, J.S. Cohen, Oligodeoxynucleotides as antisense inhibitors of gene expression, Prog. Nucleic Acid Res. Mol. Biol. 42 Ž1992. 79–106. w15x W.A. Hall, E.P. Flores, W.C. Low, Antisense oligonucleotides for central nervous system tumours, Neurosurgery 38 Ž1996. 376–383. w16x M. Heilig, K.-H. Schlingensiepen, Antisense oligodeoxynucleotides as novel neuropharmacological tools for selective expression blockade in the brain, in: Genetic Manipulation of the Nervous System, Academic Press, 1996, pp. 249–268. w17x F.F. Johansen, N.H. Diemer, Enhancement of GABA neurotransmission after cerebral ischemia in the rat reduces loss of hippocampal CA1 pyramidal cells, Acta Neurol. Scand. 84 Ž1991. 1–6. w18x J. Karle, M. Nielsen, Modest reduction of benzodiazepine binding in rat brain in vivo induced by antisense oligonucleotide to GABA A receptor g 2 subunit subtype, Eur. J. Pharmacol. Mol. Pharmacol. 291 Ž1995. 439–441. w19x J. Karle, M.-R. Witt, M. Nielsen, Antisense oligonucleotide to GABA A receptor g 2 subunit induces loss of neurones in rat hippocampus, Neurosci. Lett. 202 Ž1995. 97–100. w20x J. Karle, M.-R. Witt, M. Nielsen, The use of in vivo antisense oligonucleotide technology for the investigation of brain GABA A receptors, Neurochem. Int., Ž1996. in press. w21x G. Keilhoff, G. Wolf, F. Stastny, W. Schmidt, Quinolinate neurotoxicity and glutamatergic structures, Neuroscience 34 Ž1990. 235–242. w22x C. Kohler, R. Schwarcz, Comparison of ibotenate and kainate ¨ neurotoxicity in rat brain: a histological study, Neuroscience 8 Ž1983. 819–835. w23x A.M. Krieg, C.A. Stein, Phosphorothioate oligodeoxynucleotides: antisense or anti-protein?, Antisense Res. Dev. 5 Ž1995. 241. w24x A.M. Krieg, A.-K. Yi, S. Matson, T.J. Waldschmidt, G.A. Bishop, R. Teasdale, G.A. Koretzky, D.M. Klinman, CpG motifs in bacterial DNA trigger direct B-cell activation, Nature 374 Ž1995. 546–549. w25x J.P. Leonetti, G. Degols, J.P. Clarenc, N. Mechti, B. Lebleu, Cell delivery and mechanisms of action of antisense oligonucleotides, Prog. Nucleic Acid Res. Mol. Biol. 44 Ž1993. 143–166. w26x R.L. Macdonald, R.W. Olsen, GABA A receptor channels, Annu. Rev. Neurosci. 17 Ž1994. 569–602. w27x M.M. McCarthy, D.B. Masters, K. Rimvall, S. Schwartz-Giblin, D.W. Pfaff, Intracerebral administration of antisense oligodeoxynucleotides to GAD65 and GAD67 mRNAs modulate reproductive behavior in the female rat, Brain Res. 636 Ž1994. 209–220. w28x R. McKernan, P.J. Whiting, Which GABA A receptor subtypes really occur in the brain?, Trends Neurosci. 19 Ž1996. 139–143. w29x S.J. Mihic, P.J. Whiting, R.L. Klein, K.A. Wafford, R.A. Harris, A single amino acid of the human g-aminobutyric acid type A receptor g 2 subunit determines benzodiazepine efficacy, J. Biol. Chem. 209 Ž1994. 32768–32773. w30x L.G. Miller, D.J. Greenblatt, J.G. Barnhill, R.I. Shader, Chronic benzodiazepine administration I. Tolerance is associated with benzodiazepine receptor downregulation and decreased g-aminobutyric acid A receptor function, J. Pharmacol. Exp. Ther. 246 Ž1988. 170–176. w31x H. Mohler, T. Okada, S.J. Enna, Benzodiazepine and neurotrans¨ mitter binding in rat brain after chronic administration of diazepam and phenobarbital, Brain Res. 156 Ž1978. 391–395. w32x M. Nielsen, C. Braestrup, The apparent target size of rat benzodiazepine receptor, acetylcholinesterase, and pyruvate kinase is highly influenced by experimental conditions, J. Biol. Chem. 263 Ž1988. 11900–11906.

J. Karle et al.r Brain Research 765 (1997) 21–29 w33x P.M. Pilowsky, S. Suzuki, J.B. Minson, Antisense oligonucleotides: a new tool in neuroscience, Clin. Exp. Pharmacol. Physiol. 21 Ž1994. 935–944. w34x D.B. Pritchett, H. Sontheimer, B.D. Shivers, S. Ymer, H. Kettenmann, P.R. Schofield, P.H. Seeburg, Importance of a novel GABA A receptor subunit for benzodiazepine pharmacology, Nature 338 Ž1989. 582–585. w35x H.C. Rosenberg, T.H. Chiu, Tolerance during chronic benzodiazepine treatment associated with decreased receptor density, Eur. J. Pharmacol. 70 Ž1981. 453–460. w36x R.D. Schwartz, X. Yu, M.R. Katzman, D.M. Hayden-Hixson, J.M. Perry, Diazepam, given postischemia, protects selectively vulnerable neurons in the rat hippocampus and striatum, J. Neurosci. 15 Ž1995. 529–539. w37x B.D. Shivers, I. Killisch, R. Sprengel, H. Sontheimer, M. Kohler, ¨ P.R. Schofield, P.H. Seeburg, Two novel GABA A receptor subunits exist in distinct neuronal subpopulations, Neuron 3 Ž1989. 327–337. w38x W. Sieghart, Structure and pharmacology of g-aminobutyric acid A receptor subtypes, Pharmacol. Rev. 47 Ž1995. 181–234. w39x G. Sperk, Kainic acid seizures in the rat, Prog. Neurobiol. 42 Ž1994. 1–32. w40x C.A. Stein, Does antisense exist?, Nature Med. 1 Ž1995. 1119–1121. w41x C.A. Stein, Y.-C. Cheng, Antisense oligonucleotides as therapeutic agents – is the bullet really magical?, Science 261 Ž1993. 1004–1012. w42x C.A. Stein, A.M. Krieg, Problems in interpretation of data derived from in vitro and in vivo use of antisense oligodeoxynucleotides ŽEditorial., Antisense Res. Dev. 4 Ž1994. 67–69. w43x F.A. Stephenson, The GABA A receptors, Biochem. J. 310 Ž1995. 1–9. w44x R.W. Wagner, Gene inhibition using antisense oligodeoxynucleotides, Nature 372 Ž1994. 333–335. w45x C. Wahlestedt, Antisense oligonucleotide strategies in neuropharmacology, Trends Pharmacol. Sci. 15 Ž1994. 42–46.

29

w46x C. Wahlestedt, E. Golanov, S. Yamamoto, F. Yee, H. Ericson, H. Yoo, C.E. Inturrisi, D.J. Reis, Antisense oligodeoxynucleotides to NMDA-R1 receptor channel protect cortical neurons from excitotoxicity and reduce focal ischaemic infarctions, Nature 363 Ž1993. 260–263. w47x B. Weiss, L.-W. Zhou, S.-P. Zhang, Z.-H. Qin, Antisense oligodeoxynucleotide inhibits D2 dopamine receptor-mediated behavior and D2 messenger RNA, Neuroscience 55 Ž1993. 607–612. w48x L. Whitesell, D. Geselowitz, C. Chavany, B. Fahmy, S. Walbridge, J.R. Alger, L.M. Neckers, Stability, clearance, and disposition of intraventricularly administered oligodeoxynucleotides: implications for therapeutic application within the central nervous system, Proc. Natl. Acad. Sci. USA 90 Ž1993. 4665–4669. w49x W. Wisden, D.J. Laurie, H. Monyer, P.H. Seeburg, The distribution of 13 GABA A receptor subunit mRNAs in the rat brain I. Telencephalon, diencephalon, mesencephalon, J. Neurosci. 12 Ž1992. 1040–1062. w50x Y. Yaida, T.S. Nowak Jr., Distribution of phosphodiester and phosphorothioate oligonucleotides in rat brain after intraventricular and intrahippocampal administration determined by in situ hybridization, Regul. Pept. 59 Ž1995. 193–199. w51x M. Zhang, I. Creese, Antisense oligodeoxynucleotide reduces brain dopamine D 2 receptors: behavioral correlates, Neurosci. Lett. 161 Ž1993. 223–226. w52x T. Zhao, H.C. Rosenberg, T.H. Chiu, Treatment with an antisense oligodeoxynucleotide to the GABA A receptor g 2 subunit increases convulsive threshold for b-CCM, a benzodiazepine ‘inverse agonist’, in rats, Eur. J. Pharmacol. 306 Ž1996. 61–66. w53x W.J. Zhu, J.F. Wang, S. Vicini, D.R. Grayson, a 6 and g 2 subunit antisense oligodeoxynucleotides alter g-aminobutyric acid receptor pharmacology in cerebellar granule neurons, Mol. Pharmacol. 50 Ž1996. 23–33.