Ž .Brain Research 756 1997 76–83

Research report

Ž .Adeno-associated virus AAV vector antisense gene transfer in vivodecreases GABA a containing receptors and increases inferior collicularA 1

seizure sensitivity

Xiao Xiao a,f, Thomas J. McCown a,b,c,e,) , Juan Li a,f, George R. Breese b,c,d,e, A. Leslie Morrow c,d,e,R. Jude Samulski a,d

a Gene Therapy Center, CB 7352, 7119 Thurston Building, UniÕersity of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USAb UNC Neuroscience Center, UniÕersity of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA

c Center for Alcohol Studies, UNC, Chapel Hill, NC 27599, USAd Department of Pharmacology,UNC, Chapel Hill, NC 27599, USA

e Department of Psychiatry, UNC, Chapel Hill, NC 27599, USAf Somatix Corporation, Chapel Hill, NC, USA

Accepted 7 January 1997

Abstract

Ž .In the inferior colliculus, adeno-associated virus AAV vectors are capable of gene transfer and stable, long-term expression, but itremained to be shown if this in vivo gene transfer could alter focal seizure sensitivity in the inferior colliculus. Because GABA receptors

Ž .directly modulate inferior collicular seizures, AAV vectors were constructed with a cytomegalovirus CMV promoter and a truncated,Žhuman GABA a cDNA in both the sense and antisense orientations. Seven days after collicular microinjection of the sense vectors 1A 1

9 .ml; 3=10 particlesrml , neurons exhibited GABA a-like immunoreactivity in amounts far exceeding endogenous concentrations.Aw3 xUnilateral or bilateral sense vector infusion had no effect on inferior collicular seizure parameters or on H zolpidem binding. In contrast,

Ž 8 .bilateral infusion of the antisense AAV-GABA a vector 1 ml; 3=10 particlesrml caused a 137% increase in the seizure duration.A 1w3 xMoreover, unilateral antisense vector infusion produced a localized, 48% decrease in H zolpidem binding. Thus, in the inferior

colliculus, antisense AAV-CMV vectors can reduce a specific receptor subunit protein and change receptor function that directlyinfluences in vivo seizure sensitivity.

Ž .Keywords: Adeno-associated virus; Gene therapy; Inferior colliculus; Seizure; g-Aminobutyric acid GABA ; Antisense

1. Introduction

Following the successful demonstration of gene transferw xand expression in the CNS 17 , subsequent investigators

have utilized a number of virus vectors to deliver a diverserange of genes targeted towards a wide variety of braindisorders. For example, using an adenovirus vector, Badie

w xet al. 1 found that adenovirus vector delivery of a p53gene significantly reduced the volume of a glial tumor.

w xUsing a herpes virus vector, Lawrence et al. 8 reportedthat vector delivery of a Bcl-2 gene protected hippocampal

) Corresponding author.

neurons from focal ischemic damage or adriamycin toxic-Ž .ity. Finally, using an adeno-associated virus AAV vector,

w xKaplitt et al. 7 showed that vector delivery of tyrosinehydroxylase, the rate-limiting enzyme for dopamine syn-thesis, partially reversed motor deficits induced by adopamine lesion. Although these studies used three differ-ent vectors to deliver three different genes for three differ-ent reasons, the studies have an important commonality. Ineach case, an effective method of gene transfer and expres-sion was coupled to a relevant animal model of the disor-der being targeted.

w xRecently, McCown et al. 13 found that AAV vectorscontaining an an E. coli b-galactosidase gene driven by a

Ž .cytomegalovirus CMV promoter produced stable, long-term gene transfer and expression in the inferior colliculus.

0006-8993r97r$17.00 Copyright q 1997 Elsevier Science B.V. All rights reserved.Ž .PII S0006-8993 97 00120-0

( )X. Xiao et al.rBrain Research 756 1997 76–83 77

Not only did the level of gene expression remain stableover a 3 month period, but the gene product was localizedalmost exclusively to neurons. These findings presented anexcellent opportunity to study the application of genetherapy to seizure disorders, because the inferior colliculuscontains a well characterized site of focal seizure genesisw x9,12 .

In rats, the sensorimotor area of the inferior collicularw xcortex is capable of focal seizure genesis 12 . In this area,

a brief electrical stimulus produces post-stimulus wildrunning behaviors that coincide with localized afterdis-charge, yet when these seizures are elicited twice a day,there is a gradual increase in seizure duration until finallyon a given trial, the wild running seizure is followed by theappearance of generalized tonic-clonic, or myoclonic,seizure activity. This generalized seizure activity coincideswith afterdischarge activity in the forebrain and, analogousto limbic seizure kindling, proves to be a permanent

w xchange in seizure severity 9,10 . From these well charac-terized properties, it is possible to evaluate each compo-nent of the seizure complex: the seizure initiation thresh-old, the seizure duration, and the number of stimulationsnecessary to produce permanent seizure generalization intothe forebrain.

In brain, the major inhibitory neurotransmitter is g-Ž .aminobutyric acid GABA , and in the inferior collicular

cortex, as well as other brain areas, GABA receptorsdirectly modulate seizure sensitivity. Increasing GABAA

receptor activity attenuates seizure genesis, while blockadeof GABA receptor function increases seizure sensitivityAw x5,11 . GABA receptors are ligand-gated ion channelsA

whose structure is a proposed pentameric assembly ofŽdifferent a , b , g , and d receptor subnits see Sieghart

w x .20 for review . In the inferior colliculus, the predominantw xreceptor subunit proteins are a , b , g 4 , all potential1 2 2

targets of manipulation by gene transfer. Therefore, it wasnow possible to answer the question, can AAV vector-mediated gene transfer and expression alter GABA recep-A

tor function in neurons that directly influence inferiorcollicular seizures?

2. Materials and methods

2.1. Animals

All of the animals were pathogen-free, male or femaleSprague–Dawley rats obtained from Charles Rivers Asso-

Ž .ciates Raleigh, NC , or bred within the animal colony.The animals were maintained in a 12 h light–dark cycleand had free access to water and food. All care andprocedures were in accordance with the ‘‘Guide for the

ŽCare and Use of Laboratory Animals’’ DHHS PublicationŽ . .No. NIH 85-23 , and all procedures received prior ap-

proval by the IACUC at the university.

2.2. AAV Õector preparation

The AAV-LacZ vector used in these studies has beenw xpreviously described 13 . The AAV vectors containing the

truncated human GABA a cDNA were constructed inA 1

either the sense or antisense orientation with respect to theCMV promoter. A truncated human GABA a DNAA 1w x Ž .15,16 y165 to q707 with respect to the start codonwas excised from plasmid p33 with restriction enzymesBssHII and HindIII and blunt-ended with Klenow enzyme.This fragment was ligated with a 4.2 kb blunt-ended NotIfragment from plasmid pdx31-LacZ5 and transformed into

Ž .E. coli DH10B strain Gibco . Numerous clones werescreened and analyzed with restriction enzymes. The sense

Ž .plasmid was designated pAAV-DNA GABA a q , while1

the antisense plasmid was designated pAAV-DNA GABAŽ .a y .1

Both sense and antisense AAV-GABA a viral parti-A 1

cles were produced by cotransfecting vector plasmidsŽ Ž . Ž .pAAV-DNA GABA a q , pAAV-DNA GABA a y ,1 1

w x.pdx31-LacZ 14 with the helper plasmid AAVrAd intow xadenovirus infected 293 cells 19 . Briefly, 25 mg of

Ž .plasmid DNA 12.5 mg vector plus 12.5 mg helper wastransfected by calcium phosphate precipitation into 293cells at 80% confluency in DMEM plus 10% FBS. Themedium was replaced after 8–12 h transfection with freshDMEM plus 10% FBS. Adenovirus 5 was added to the

Ž .cells at 1 m.o.i. multiplicity of infection . After 2.5 days,the cells were harvested and subjected to three freezerthawcycles. Cell debris was removed by low speed centrifuga-tion. The supernatant containing AAV-GABA a particles1

was added to an equal volume of cold, saturated ammo-Ž .nium sulfate solution pH 7.0 . The sample was placed on

ice for 10 min and centrifuged at 15 000=g for 10 min.The pellet was redissolved in cesium chloride-PBS solu-

Ž .tion density 1.38 grml and centrifuged at 40 000 r.p.m.for 48 h, separating the AAV from the adenovirus. TheAAV band was collected, dialyzed against PBS buffer andheated at 568C for 30 min which inactivates any aden-ovirus. The virus preparations were aliquoted and stored aty208C until use. All of the studies utilized a single stockof the respective virus preparations.

The infectious titer of the AAV-CMV LacZ vectorswere determined by x-gal staining as previously describedw x Ž .13 . The titers of AAV-GABA a q or AAV-GABA1Ž .a y vectors were determined as virus particle numbers1

by Southern hybridization. Briefly, 2 ml of virus prepara-tion was digested with 10 units of DNase I at 378C for 30min. The reaction was stopped by EDTA and followed by2 phenol-chloroform extractions. After precipitation with0.3 M sodium acetate and 2.5 vol. of ethanol, the viral

ŽDNA was dissolved in 15 ml 150 mM NaOHr1 mM.EDTA and separated on a 1% denaturing agarose gel.

Southern hybridization was carried out with a 32 P-labeledhuman GABA a fragment. The viral particle numbersA 1

( )X. Xiao et al.rBrain Research 756 1997 76–8378

were determined by quantitating the radioactivity gener-ated from the 2.4 kb viral DNA band using plasmid

Ž .pAAV-DNA GABA a q as the copy number standard.1

This virus particle determination generally results in titers1000 times higher than the infectious titer determinations.

2.3. Electrode implantation and AAV microinjection

Prior to surgery, the animals were anesthetized with 50mgrkg pentobarbital and placed into a stereotaxic frame.The first manipulation tested if the AAV vectors wouldalter the seizure threshold or the seizure duration aftermicroinjection in between the stimulating electrodes. For

Žthis group, two insulated stainless steel wires 0.015 inch.diameter were wrapped around a 26-gauge stainless steel

guide cannula and secured with epoxy. The wires termi-nated 2.0 mm below the tip of the stainless steel cannulaguide and had a tip separation of 400–500 mm. Thiselectrodercannula guide was then implanted into the

Žseizure sensitive site of the inferior collicular cortex IAL0.2 mm, lateral 1.6 mm, vertical 3.5 mm, according to the

w x.atlas of Paxinos and Watson 18 . The electrodercannulaguide was secured to three screws in the skull with cranio-plastic cement. All implants were made unilaterally, andthe animals were allowed at least 7 days to recover fromthe surgery. After the baseline seizure stimulation thresh-

Žold current and seizure duration was determined 3–4. Ž 9tests , 1 ml of sense AAV-GABA a vectors 3=101

. Ž 8particlesrml , antisense AAV-GABA a vectors 3=101

. Ž 5particlesrml or AAV-LacZ vectors 1=10 transducing.unitsrml; 1 transducing units1000 particles were in-

fused at a rate of 1 ml over 9 min. The injector was left inplace for 1 min post infusion to allow diffusion from theinjection site, and the animals were tested 7 days posttreatment. Thus, this procedure allows within-subject com-parisons.

The second manipulation tested if bilateral treatmentwith AAV vectors would alter the wild running seizureduration or the number of stimulations necessary to pro-duce kindled generalized seizure activity. For this group,

Žthe animals were anesthetized with pentobarbital 50.mgrkg and placed into a stereotaxic frame. The skull was

exposed and holes were drilled bilaterally over the samecollicular sites as for the unilateral group. A 32-gaugestainless steel tube was lowered into each site and 1 ml of

Ž 9 .sense AAV-GABA a vectors 3=10 particlesrml or1Ž 8 .antisense AAV-GABA a vectors 3=10 particlesrml1

were infused at a rate of 1 ml over 9 min. The injector wasleft in place for 1 min post infusion to allow diffusionfrom the injection site. One hole was filled with bone wax,

Ž Ywhile on the other side, a tripolar electrode 0.015 stain-less steel, insulated except for the tip cross-section, 400

.mm vertical tip separation was lowered into the collicularŽ .cortex same coordinates as the microinjection . This stim-

ulating electrode was secured to three screws in the skullwith cranioplastic cement, and the animals were allowed 7days to recover. As before, seizure testing was initiated 7days post treatment. Finally, for the immunohistochemistry

Ž 9Fig. 1. GABA a-like immunoreactivity in the inferior colliculus 7 days after a unilateral infusion of sense AAV-GABA a vectors 1 ml; 3=10A 1.particlesrml . Notice that light to negligible staining is present in the right, uninjected side, because large amounts of vector-derived product required early

termination of the DAB reaction process. Scale bars500 mm.

( )X. Xiao et al.rBrain Research 756 1997 76–83 79

Ž 9Fig. 2. GABA a-like immunoreactivity in the inferior colliculus 7 days after a unilateral infusion of sense AAV-GABA a vectors 1 ml; 3=10A 1. Ž . Ž .particlesrml . A and an enlargement B are 500 mm rostral to the injection. C and an enlargement D are through the injection site, while E and anŽ .enlargement F are 800 mm caudal to the injection site. The enlargements illustrate the localization of gene product to neurons with no obvious labeling of

Ž . Ž .astrocytes or microglia. Scale bars500 mm in A, C and E ; 500 mm in B, D and F .

( )X. Xiao et al.rBrain Research 756 1997 76–8380

Table 1Effects of unilateral treatment with sense or antisense AAV-GABA a or1

AAV-LacZ vectors on inferior collicular seizure sensitivity

Change in Change inseizure threshold wild runningcurrent from duration frompretreatment pretreatment

a aŽ . Ž .control mm control s

Ž .Sense AAV-GABA a1 ns4 y5q5 0.4q0.5bŽ .Antisense AAV-GABA a ns6 0q7.3 2.5q1.91

AAV-LacZ 8q6.3 0.4q0.6

a After the baseline seizure threshold and duration were determinedanimals received an infusion of one of the AAV vectors in the area of the

Ž .stimulating electrode see Section 2 . Seven days later, the seizurethreshold and duration were determined. Since each subject served as itsown control, these values reflect the absolute change in threshold currentŽ . Ž .mA or wild running duration s from pretreatment control values, notthe individual values.b P -0.05.

w3 xor H zolpidem autoradiography, animals received unilat-eral AAV vector infusions as described above.

2.4. Stimulation procedure

For the electrodercannula animals, the baseline seizurestimulation threshold current and wild running durationwere determined after recovery from surgery, as previously

w xdescribed 9 . Briefly, animals were connected to a GrassŽModel SD9 stimulator and stimulation 30 Hz, 1.5 ms

.duration, monophasic square wave was initiated at 80 mAŽall stimulation currents were continuously monitored onan oscilloscope by measuring the voltage drop across a 10

.k V resistor . The stimulation current was increased 20mA every 5 s until the first appearance of wild runningbehavior. At this instance, the stimulation was terminated,and the post-stimulus wild running duration was timed.The animals were tested once per day, and 3–5 days wererequired to establish a stable, control baseline seizurethreshold and wild running duration. The animals thenreceived vector infusions as described above. Seven dayslater, the seizure stimulation threshold and wild runningduration were determined. For the bilateral infusions,seizure testing was initiated 7 days after vector infusionand electrode implantation. As above, the seizure thresholdcurrent and wild running duration were determined, afterwhich the seizure kindling rate was determined. For thekindling process, the animals received two stimulations perday, one in the morning and one in the afternoon. Whenthe animals exhibited two consecutive episodes of wildrunning followed by forelimb tonus and hindlimb clonusor myoclonic jerks of the neck and forelimbs, the animalswere considered kindled. At this point, the appearance of

w xgeneralized seizure activity is permanent 9 .

2.5. Immunohistochemistry

Seven days after AAV-GABA a vector infusion,1

animals were anesthetized with 100 mgrkg pentobarbital

Ž .i.p. and perfused transcardially with ice-cold 100 mMŽ . Ž .sodium phosphate-buffered saline PBS pHs7.4 , fol-

lowed by 4% paraformaldehyde in 100 mM phosphateŽ .buffer pHs7.4 . After overnight fixation in the para-

Žformaldehyde-phosphate buffer, vibratome sections 40 mm.thick were taken and rinsed in PBS. Tissue sections were

incubated in 10% normal horse serum and 0.2% TritonX-100 in PBS for 30 min. Next, sections were incubatedwith a monoclonal antibody to GABA receptor a sub-A

Ž .units 1:250 dilution; Chemicon in 3% normal horseserum, 0.2% Triton X-100 and PBS for 48–72 h at 408C.Tissue sections were then rinsed in PBS and processedthrough secondary biotinylated horse anti-mouse antibodyand avidin-biotin complex using a Vectastain Mouse Elite

Ž .ABC Kit Vector Laboratories, Burlingame, CA . Visual-ization of FLI was achieved by nickelrcobalt enhancement

X w xof 3,3 -diaminobenzidine tetrahydrochloride 6 .

[ 3 ]2.6. H Zolpidem autoradiography

The zolpidem autoradiography was performed as previ-w xously described 3 . First, rats received unilateral microin-

fusions of either the sense or antisense AAV-GABA a1Ž 9vectors 1 ml over 9 min; sense 3=10 particlesrml,

8 .antisense 3=10 particlesrml and 7 days later weresacrificed. Ten micrometer thick tissue sections were thawmounted onto slides and stored at y808C. The slides were

Ž .preincubated for 15 min 50 mM Tris, pH 7.5 and thenw3 x Žincubated for 30 min in 5 nM H zolpidem 50 mM Tris,

.120 mM NaCl, pH 7.5 . The slides were washed for 2 minin 50 mM Tris, pH 7.5, followed by a 15 s wash indistilled water. Non-specific binding was determined by

w3 xaddition of 1 mM zolpidem. Once dry, the slides and HŽ .standards were opposed to Hyperfilm Amersham . Six

Fig. 3. The effects of sense or antisense truncated human AAV-GABAŽ .a AAV-GABA on inferior collicular wild running seizure duration 71

Ž .days after bilateral infusions see Section 2 . Note that the sense constructŽdid not alter the wild running seizure duration control y6.4"0.4 s,

.ns7; sense AAV y6.6"0.6 s, ns5; mean"S.E.M. . In contrast, theantisense AAV produced a significant 137% increase in the wild running

Ž .seizure duration 21.8"5.3 s, ns6 .

( )X. Xiao et al.rBrain Research 756 1997 76–83 81

weeks later, the films were developed and the opticalŽdensities were measured by digital analysis Imaging Tech-

.nology FG-100 image analysis system . Amounts of ra-

dioactivity in brain regions were determined by relatingoptical density measurements in the sections to those in thew3 xH standards.

w3 xFig. 4. Autoradiographs of H zolpidem binding to brain sections through the inferior colliculus 7 days after the infusion of antisense AAV-GABA a1Ž . Ž . w3 xA or sense vectors B . Note that in A, the light area indicated by the arrow illustrates a localized decrease of H zolpidem binding. This section iscaudal to the injector site, where there was no tissue damage from the injector. In contrast, the arrow in B identifies the injector site for the sense

w3 xAAV-GABA a , and as seen there is no change in the H zolpidem binding density outside the injector tract, compared to the uninjected, contralateral1Žw3 x .side. C represents non-specific binding of H zolpidem in the presence of 1 mM zolpidem .

( )X. Xiao et al.rBrain Research 756 1997 76–8382

2.7. Statistics

For the comparisons in studies with microinfusionsbetween the electrodes, pretreatment values were com-pared to the post-treatment values using a paired t-test,

w xbecause each animal served as its own control 21 . Com-w3 xparisons for the bilateral infusions and the H zolpidem

binding densities were made with a two-tailed t-test.

3. Results

3.1. AAV-mediated gene transfer and expression

AAV vectors were constructed with a CMV promoterŽand a truncated, human GABA a cDNA AAV-GABAA 1

.a in both the sense and antisense configurations. When1

the sense AAV-GABA a vectors were microinjected into1Ž 9 .the inferior colliculus 1 ml; 3=10 particlesrml , 7 days

later substantial protein expression could be seen almostŽ .exclusively in neurons see Fig. 1 and Fig. 2 . The amount

of vector derived protein exceeded endogenous GABA a

protein to such a degree that immunohistochemical stain-ing for endogenous protein is barely visible in the con-

Ž .tralateral, uninjected colliculus see Fig. 1 . Clearly,AAV-GABA a vectors transfer and express this receptor1

subunit protein in vivo at levels similar to previous find-w xings for the b-galactosidase reporter gene, LacZ 13 .

3.2. Effects of sense and antisense AAV-GABAa Õectors1

on inferior collicular seizure sensitiÕity

Unilateral microinjection of the sense AAV-GABA a1Ž 9 .vectors 1 ml; 3=10 particlesrml or AAV-CMV-LacZŽ 8 .vectors 1 ml; 6.8=10 particlesrml in the vicinity of

the stimulation electrode did not alter the seizure thresholdcurrent or the wild running seizure duration 7 days laterŽ .see Table 1 . Moreover, in both cases, there was cleargene product expression in neurons near the electrodeŽ .data not shown . When the antisense vectors were infusedinto the vicinity of the stimulating electrode, the thresholdseizure current did not change, but there was a small,

Žsignificant increase in the wild running duration see Table.1 .

Next, rats received bilateral infusions of the sense orŽ .antisense AAV-GABA a vectors 1 mlrside just prior1

to the electrode implantation and were tested 1 week later.As seen in Fig. 3, the sense infusions did not alter theinitial wild running duration in comparison to uninjectedcontrols, but the antisense group exhibited a significant,137% increase in the wild running duration in comparisonto the control group. Moreover, this antisense effect provedto be quite selective. Subsequent testing showed that therewere no significant differences between the three groups inthe number of stimulations necessary to kindle generalized

Žseizure activity uninjected controls17"2; sense AAV-GABA a s17"3; antisense AAV-GABA a s14"11 1Ž ..stimulations to kindling"S.E.M. .

3.3. Effects of sense and antisense AAV-GABA a Õectors1[ 3 ]on H zolpidem binding

Zolpidem binding depends upon the presence of a ,1w xb , and g subunit GABA receptor proteins 2,20 , soX 2 A

perturbations in the assembly of this receptor might bew3 x Ž 9reflected in H zolpidem binding. The sense 1 ml; 3=10. Ž 8 .particlesrml or antisense 1 ml; 3=10 particlesrml

AAV-GABA a vectors were infused unilaterally and 71w3 xdays later, brain sections were taken for H zolpidem

binding. As seen in Fig. 4, the sense infusion did not alterw3 x ŽH zolpidem binding density 104"4% contralateral side

."S.E.M.; ns4 . In contrast, the antisense vectors causedw3 xa localized, significant decrease in H zolpidem binding

density, when compared to the untreated, contralateral sideŽ .52"6% of contralateral side"S.E.M., ns4, P-0.05 .

4. Discussion

The results from these studies support three primaryconclusions. First, AAV vectors can transfer receptor sub-unit genes into neurons of the inferior colliculus and, underCMV promoter control, express substantial amounts of thegene product. Secondly, in the inferior colliculus the levelof gene expression is sufficiently high, such that antisenseconstructs can influence the function of native receptors,and finally, this AAV-mediated gene transfer occurs inneurons that directly influence the seizure process in theinferior collicular cortex.

Similar to previous studies using an E. coli LacZreporter gene, AAV-GABA a vectors transferred and1

expressed substantial gene product in the inferior collicu-lus. Like AAV-CMV-LacZ studies, neurons were the over-whelming cell type exhibiting the gene product. No ex-pression was found in astrocytes or microglia, cells thatcan be distinguished from neurons on a morphologicalbasis. Clearly, the AAV-CMV vector is tropic for neurons,but it remains to be determined if the vector, the promoteror both are responsible for this selectivity. As importantly,these findings demonstrate that using AAV-CMV vectors,receptor subunit proteins can be expressed in neurons invivo. In the case of the truncated, human GABA a1

protein, substantial protein expression did not cause anychanges in measures of focal seizure activity. The simplestexplanation for these results is that the truncated, humanGABA a protein is not functional. Alternatively, the1

protein might be functional, but just does not assemblewith native receptors or increase receptor number. Theresolution of these questions will require further investiga-tions with a full length protein.

The significant, selective changes induced by the anti-sense vectors demonstrate that given high levels of geneexpression, antisense gene delivery can alter neural func-tion without producing neural damage or toxicity. Theantisense AAV-GABA a vectors more than doubled the1

seizure duration, yet did not alter the seizure threshold or

( )X. Xiao et al.rBrain Research 756 1997 76–83 83

the number of stimulations for kindling. This change couldnot be attributed to the AAV vector, because the sensevector had no effects. Also, it is unlikely that neuraldamage or toxicity played a role in these changes. Onlyone component of the seizure was altered, and the natureof this change required neural activity to increase, notdecrease, as would be expected with toxicity. Moreover,the antisense-induced changes are compatible with a reduc-tion in a containing GABA receptors in the inferior1 A

colliculus, a conclusion reinforced by the localized reduc-w3 xtion in H zolpidem binding density. Thus, from antisense

gene delivery, it appears that benzodiazepine type 1 recep-tors influence a specific component of inferior collicularseizures, the seizure duration.

As importantly, the antisense studies show that AAV-mediated gene transfer and expression can influence infe-rior collicular seizures and, in the process, reveal basicproperties of the collicular seizure network. Assuming thatthe antisense vectors reduced the number of GABAA

receptors with a subunits, the net result would be a loss1

of inhibition. For this loss of inhibition to increase theseizure duration, it is likely that the inhibitory receptors areon excitatory neurons that directly modulate seizure out-put. In such a case, the loss of inhibitory receptors on theexcitatory output neurons would lead to an increasedseizure duration. Therefore, AAV-mediated gene transferand expression appear to influence neurons that directlymodulate inferior collicular seizure sensitivity. Because thebilateral antisense effects far exceeded the effects of aunilateral infusion, the contralateral colliculus also mustsubserve a role in the seizure process.

In conclusion, AAV gene transfer and expression of anantisense construct showed that GABA receptors contain-A

ing a subunits subserve a specific role in the modulation1

of inferior collicular seizures. Also, these studies demon-strate that in a brain area with high, stable levels ofexpression, antisense vectors can alter selectively endoge-nous receptor function. Finally, these studies illustrate thatAAV-mediated gene transfer into the inferior colliculusprovides an excellent model to test gene therapy targets forthe treatment of focal seizure genesis.

Acknowledgements

These studies were supported in part by GrantsAA09122, HL51818, HD03110, and NS 35633.

References

w x1 Badie, B., Drazan, K.E., Kramar, M.H., Shaked, A. and Black, K.L.,Adenovirus-mediated p53 gene delivery inhibits 9L glioma growth

Ž .in rats, Neurol. Res., 17 1995 209–215.w x2 Carter, D.B., Thomsen, D.R., Im, W.B., Lennon, D.J., Ngo, D.M.,

Gale, W., Im, H.K., Seeburg, P.H. and Smith, M.W., Functionalexpression of GABA chloride channels and benzodiazepine bind-A

ing sites in baculovirus infected insect cells, BiorTechnology, 10Ž .1992 679–681.

w x3 Duncan, G.E., Breese, G.R., Criswell, H.E., McCown, T.J., Herbert,w3 xJ.S., Devaud, L.L. and Morrow, A.L., Distribution of H zolpidem

binding sites in relation to mRNA encoding the a1, b 2 and g 2subunits for the GABA receptor in rat brain, Neuroscience, 64AŽ .1995 1113–1128.

w x4 Fritschy, J. and Mohler, H., GABA -receptor heterogeneity in theA

adult rat brain: differential regional and cellular distribution of sevenŽ .major subunits, J. Comp. Neurol., 359 1995 154–194.

w x5 Frye, G.D., McCown, T.J. and Breese, G.R., GABA-mimetic block-ade of ethanol withdrawal-audiogenic seizures in the inferior collicu-lus: comparison with the substantia nigra and medial septum, J.

Ž .Pharmacol. Exp. Ther., 227 1983 663–669.w x6 Hsu, S.M. and Soban, E., Color modification of diaminobenzidine

Ž .DAB precipitation by metallic ions and its application for doubleŽ .immunohistochemistry, J. Histochem. Cytochem., 30 1982 1079–

1082.w x7 Kaplitt, M.G., Leone, P., Samulski, R.J., Xiao, X., Pfaff, D.W.,

O’Malley, K.L. and During, M.J., Long-term gene expression andphenotypic correction using adeno-associated virus vectors in the

Ž .mammalian brain, Nature Genet., 8 1994 148–154.w x8 Lawrence, M.S., Ho, D.Y., Sun, G.H., Steinberg, G.K. and Sapol-

sky, R.M., Overexpression of Bcl-2 with herpes simplex virusvectors protects CNS neurons against neurological insults in vitro

Ž .and in vivo, J. Neurosci., 16 1996 486–496.w x9 McCown, T.J., Greenwood, R.S., Frye, G.D. and Breese, G.R.,

Electrically elicited seizures from the inferior colliculus: a potentialŽ .site for the genesis of epilepsy? Exp. Neurol., 86 1984 527–542.

w x10 McCown, T.J., Greenwood, R.S. and Breese, G.R., Inferior collicu-Ž .lar interactions with limbic seizure activity, Epilepsia, 28 1987

234–241.w x11 McCown, T.J., Givens, B.S. and Breese, G.R., Amino acid influ-

ences on seizures elicited within the inferior colliculus, J. Pharma-Ž .col. Exp. Ther., 243 1987 603–608.

w x12 McCown, T.J., Duncan, G.E. and Breese, G.R., Neuroanatomicalcharacterization of inferior collicular seizure genesis: 2-deoxyglu-

Ž .cose and stimulation mapping, Brain Res., 567 1991 25–32.w x13 McCown, T.J., Xiao, X., Li, J., Breese, G.R. and Samulski, R.J.,

Differential and persistent expression patterns of CNS gene transferŽ . Ž .by an adeno-associated virus AAV vector, Brain Res., 713 1996

99–107.w x14 McLaughlin, S.K., Collis, P., Hermonat, P.L. and Muzyczka, N.,

Adeno-associated virus general transduction vectors: analysis ofŽ .proviral structure, J. Virol., 62 1988 1963–1973.

w x15 Montpied, P., Martin, B.M., Cottingham, S.L., Stubblefield, B.K.,Ginns, E.I. and Paul, S.M., Regional distribution of the

Ž .GABA rbenzodiazepine receptor a subunit mRNA in rat brain,AŽ .J. Neurochem., 51 1988 1651–1654.

w x16 Montpied, P., Ginns, E.I., Martin, B.M., Stetler, D., O’Carroll,A.M., Lolait, S.J., Mahan, L.C. and Paul, S.M., Multiple GABAA

receptor a subunit mRNAs revealed by developmental and regionalŽ .expression in rat, chicken and human brain, FEBS Lett., 258 1989

94–98.w x17 Palella, T.D., Hidaka, Y., Silverman, L.J., Levine, M., Glorioso, J.

and Kelley, W.N., Expression of human HPRT mRNA in brain ofmice infected with recombinant herpes simplex virus-1 vector, Gene,

Ž .80 1988 137–144.w x18 Paxinos, G. and Watson, C., The Rat Brain in Stereotaxic Coordi-

nates, Academic Press, New York, 1986.w x19 Samulski, R.J., Chang, L.-S. and Shenk, T., Helper-free stocks of

recombinant adeno-associated viruses: normal integration does notŽ .require viral gene expression, J. Virol., 63 1989 3822–3828.

w x20 Sieghart, W., Structure and pharmacology of g-aminobutyric acidAŽ .receptor subtypes, Pharmacol. ReÕ., 47 1995 181–234.

w x21 Winer, N., Statistical Principles in Experimental Design, McGraw-Hill, New York, 1971.