Hippocampal infection with HSV-1-derived vectors expressing an NMDAR1 antisense modifies behavior


  • Martín F. Adrover and Valerie Guyot-Revol contributed equally to this paper.

*Corresponding authors: A. L. Epstein, Centre de Génétique Moléculaire et Cellulaire, CGMC – UMR 5534 – CNRS, Univ. Claude Bernard Lyon 1, Bâtiment G. Mendel, 16 rue Raphaël Dubois, 69622 Villeurbanne Cedex, France. E-mail: epstein@ biomserv.univ-lyon1.fr. D. Jerusalinsky, Institute Cell Biol. & Neurosci, School of Medicine, Univ. of Buenos Aires, 2155 Paraguay street, 2nd floor, (1121) Buenos Aires, Argentina. E-mail: jerusali@mail.retina.ar


Herpes simplex virus-derived amplicon vectors simultaneously expressing the open reading frame encoding NR1 subunit of the NMDA receptor, either in sense or antisense orientation, as well as the open reading frame encoding the green fluorescent protein (GFP), as distinct transcription units, were constructed. Vector expression in cells was demonstrated by GFP-fluorescence, immunofluorescence, Western blots and RT-PCR. The vectors were inoculated into the dorsal hippocampus of adult male rats, which were then trained for habituation to an open field and for inhibitory avoidance to a foot-shock. Those animals injected with vectors expressing NR1 protein showed habituation to a new environment, and achieved the criteria for a step-down inhibitory avoidance to a foot-shock. In contrast, animals injected with vectors carrying the NR1 open reading frame in antisense position, showed neither habituation nor appropriate performance in the inhibitory avoidance task. There was no evidence for motor impairment or motivational disturbance, since all the animals exhibit similar behavior and performance in the training sessions. Hence, the impaired performance might be due to either amnesia or disability to record events. Transgene expression in brain, as revealed by GFP fluorescence, was mainly observed in pyramidal cells of CA1, but also in CA3. Therefore, our results strongly support the participation of hippocampal NR1 subunit in habituation to a new environment, but also in recording events for the inhibitory avoidance task. Hence, amplicon vectors appear to be useful tools to modify endogenous gene expression at a defined period, in restricted brain regions, and should allow investigating in vivo functions of genes.

The hippocampus has a fundamental role in learning and memory across multiple species. Hippocampal lesions in different rodent species produced impairment in spatial and odor discrimination tasks (Jarrard 1993; Morris et al. 1982; O'Keefe & Nadel 1978). O'Keefe and Nadel (1978) proposed that the hippocampus serves as a spatial map, and they found neurons (place cells) that gave support to this assumption.

Since the beginning of the 1990s it has been accepted that the NMDA receptor (NMDAR) has an important role in synaptic plasticity, with relevance to long-term-potentiation (LTP) (Bliss & Collingridge 1993), as well as in a wide variety of learning tasks (Collingridge & Singer 1990; Davis et al. 1992; Hlinák & Krjcí 1998; Morris 1990). Although it has been shown that robust learning can occur in animals with hippocampal lesions or NMDAR-dependent LTP blockade (Cain et al. 1997; Saucier & Cain 1995), a critical role for the NMDAR in learning and memory has been strongly suggested both by pharmacological blockade and by genetic manipulations. Gene knockouts (KO) have been useful for investigating some molecular and cellular basis for learning and memory (Silva et al. 1992a,b), but sometimes a non-restricted genetic deletion may lead to severe developmental defects or even to death. Despite an apparently normal development, interpretation of the results is frequently complicated because the phenotype, often expressing ubiquitous changes, could not be attributed to a particular tissue or cell. It is even more difficult to discard the influence on development, which could be responsible for the observed phenotype. A mutation might lead to compensatory responses, and these could also depend on the background genotype (Gerlai 1996a). In general, the phenotypical effect of a mutation might depend on the genetic background (for gene-targeting debate, see Crusio 1996; Gerlai 1996a,b; Lathe 1996). The KO mice for the NR1 subunit of the NMDAR died neonatally (Forrest et al. 1994; Li et al. 1994). The CA1-specific NR1-KO mice (McHugh et al. 1996; Tsien et al. 1996), obtained by using specific promoters, were particularly promising. However, we do not know how widespread and selective tissue-specific promoters are, or if the deletion in this particular case occurred only after certain periods of development. Hence, the challenge was to develop an alternative technology that led either to KO or to overexpression of a gene, with temporal and spatial specificity, to study in vivo functions in different mammalian species. Viral vectors appeared the ideal technology.

Taking into account the neurotropism, the size of inserts they can accommodate and other appropriate characteristics of herpes simplex virus type-1 (HSV-1) (Simonato et al. 2000), we have built HSV-1 derived amplicon vectors to express NMDAR transgenes to the rat CNS. The NMDAR is a heteromeric complex mainly made of NR1 and NR2 subunits, where NR1 is an essential subunit, always present, which possesses most of the properties of the NMDAR-channel complex, while the NR2A-2D subunits are mainly modulatory, regulating channel gating and Mg2+ dependency (Monyer et al. 1992; Moriyoshi et al. 1991; Nakanishi 1992; Nakanishi et al. 1998).

Amplicon-type vectors carrying sequences encoding NR1, either in sense or antisense orientations, were developed with the objective of expressing the transcription units in vivo, in an attempt to further clarify the role of NMDAR. Amplicon is a helper-dependent HSV-1-derived vector made from a plasmid containing just one origin of replication (Ori-S) and one packaging sequence (a) from HSV-1, in addition to the transgenic transcription unit (Spaete & Frenkel 1982). Because this plasmid does not encode any further viral sequence, it can only be grown and packaged into HSV-1 particles in the presence of trans-acting functions provided by a HSV-1 helper. Thus, amplicon stock preparations used for gene transfer usually contain, in addition to the high-titer amplicon vector particles carrying the transgene, a variable degree of contaminating helper virus (Epstein 1995; Leib & Olivo 1994; Spaete & Frenkel 1982). We have used a recently developed system, the so-called HSV-1-LaL virus, as a helper (Logvinoff & Epstein 2001) that produces non-pathogenic vectors.

Double amplicon vectors carrying either NR1(+) sequence (expected to transcribe mRNA encoding the NR1 protein, and thus allowing expression of the protein), or NR1(–) sequence (expected to transcribe antisense NR1 RNA), and the green fluorescent protein (GFP) reporter gene, were constructed. The vectors were injected into the dorsal hippocampus of rats. Subsequently, rats were submitted to an open field paradigm and to an inhibitory avoidance task.

Materials and Methods


Baby hamster kidney (BHK-21) cells and African green monkey (VERO) cells were used. BHK-21 cells were grown in Glasgow medium supplemented with 10% fetal bovine serum (FBS), glutamine and antibiotics. VERO cells were grown in Dulbecco modified Eagle medium (DMEM) plus 10% FBS, glutamine and antibiotics. The neuroattenuated HSV-1 LaL strain (Logvinoff & Epstein 2000) was used to package the amplicon plasmids into virus particles. Stocks of HSV-1 LaL were grown and titrated by plaque assay on VERO cells as already described (Machuca et al. 1986).

The cerebral cortex was isolated from brains of E18 Wistar rats (according to Brewer et al. 1993; Brewer 1995). The cells were plated at a density of 4 × 106, either on poly D-lysine 35 mm plastic plates or on glass coverslips. The cultures were maintained for 4 days in Neurobasal (bicarbonate normal), supplemented with B27 (Gibco, Paisley, UK) at 37 °C, in a 5% CO2 incubator.

Amplicon plasmids

To construct the double amplicon plasmids carrying NR1 transcription units either in sense (+) or antisense (–) orientations, and the open reading frame (ORF) of GFP, the latest was first subcloned into the BamHI-SacI sites of a pA-SK amplicon plasmid (Tsitoura et al. 2002), to produce a novel amplicon plasmid called pA-EUA-1. Thus, this pA-EUA-1 plasmid carries the GFP ORF under the control of HSV-1 IE4 promoter, and a bovine growth hormone (BGH) polyadenylation sequence. The NR1 ORF (EcoRV-NruI from pN60) was cloned in either sense or antisense orientations, at the blunt-ended NotI site of a pCI plasmid (Promega, Mannheim, Germany), which is surrounded by the immediate early promoter of human cytomegalovirus (HCMV) and SV40 polyadenylation sequences. Lastly, the NR1(+) or NR1(–) transcription units (blunt-ended BglII-BamHI) were subcloned into the blunt-ended PstI site of pA-EUA-1, to obtain the amplicon plasmids pA-GFP-NR1(+) and pA-GFP-NR1(–), respectively. The control plasmid pA-GFP-LacZ was constructed in a similar way, except that a NotI fragment containing LacZ ORF was subcloned from plasmid pCMV-beta (BD Biosciences Clontech, Palo Alto, CA, USA), into the NotI site of pCI. In these double plasmids, the NR1 or LacZ transcription units are expressed under the control of HCMV promoter, whereas GFP is under the control of HSV-1 IE4 promoter. Similarly constructed single amplicon plasmids, expressing either GFP or NR1 protein alone (described in Cheli et al. 2002), were used in some experiments.

The amplicon plasmids used to prepare the corresponding amplicon vectors are shown in Fig. 1(a). Each of these plasmids carries two independent transgenes, either NR1(+) and GFP (pA-GFP-NR1(+)), or NR1(–) and GFP (pA-GFP-NR1(–)). A third plasmid, pA-GFP-LacZ, carries LacZ ORF instead of NR1.

Figure 1.

Amplicon plasmids and their expression. (a) Diagrams of pA-GFP-NR1(+) and pA-GFP-NR1(–) amplicon plasmids, which became the genome of the corresponding vectors. Each amplicon plasmid has an origin of replication (ORI-S) and a cleavage-packaging sequence (a) from HSV-1. The arrow over the HCMV sequence indicates the direction of transcription. The arrowhead in the NR1 fragments indicate the orientation of the open reading frame. HCMV: immediate-early enhancer promoter of human cytomegalovirus. prIE4: immediate-early 4 promoter of HSV-1. AmpR: ampicillin resistance gene. ColE1: bacterial origin of replication of all plasmids. pA (BGH) and pA (SV40): polyadenylation signals from bovine growth hormone gene and SV40, respectively. (b) Expression of GFP and NR1 protein in BHK cells transfected with pA-GFP-NR1(+) plasmid (as described in Materials and Methods). Photomicrographs taken 24 h post-transfection, under a Zeiss LSM510 confocal microscope. (i) fluorescence from GFP (green filter, 488 nm), (ii) immunofluorescence after treatment with anti-NR1 antibody, and TRITC-labeled IgG as secondary Ab (red filter, 543 nm), (iii) Merged. Notice that both signals do not completely overlap, although both proteins were coexpressed in the same cell.

Amplicon vectors

Amplicon vector stocks were prepared as already described (Lowenstein et al. 1994) using the amplicon plasmids (Fig. 1a), and the highly neuroattenuated HSV-1 LaL virus as helper (Logvinoff & Epstein 2000). Briefly, BHK-21 cells were transfected with 5 µg of either amplicon plasmid. One day later, transfected cells were superinfected at a multiplicity of infection of 1 plaque forming unit (PFU) per cell, with HSV-1 LaL as helper virus. When cytopathic effect was maximum, cells were collected by centrifugation and disrupted by freezing and thawing three times to release the vector stocks, and were centrifuged at 1000 g for 10 min to pellet the cell debris. Helper virus particles in the supernatants were titrated by plaque assay on VERO cells (Machuca et al. 1986). To titrate the vector particles expressing GFP (A-GFP-NR1(+), A-GFP-NR1(–), A-GFP-LacZ and A-GFP), VERO cells were infected with serial dilutions of the corresponding vector stock. The following day, infection was stopped, and cells were fixed with 1% paraformaldehide. Cells expressing fluorescent GFP were scored directly under a Zeiss Axioscope fluorescence microscope (Carl Zeiss, Oberkochen, Germany). The titers of the different amplicon vector stocks ranged from 5.106 to 5.107 particles/ml.

VERO cells were transfected using Lipofectamine (Gibco), whereas BHK-21 cells were transfected using Effectene transfection reagent (Qiagen Gmbh, Hilden, Germany), according to the manufacturer's instructions.


BHK-21 cells were plated into 16 mm diameter well containing a coverslip, at a density of 4.104 cells per well, and were either transfected the following day with 0.4 µg of plasmid DNA, or infected with the corresponding vectors. The day after, the cells were washed twice with PBS, and fixed for 5 min with 1.5% formaldehyde in PBS containing 2% sucrose. After three washes with PBS, cells were permeabilized by immersion in a PBS solution containing 10% sucrose and 0.5% Nonidet P-40, for 5 min. Cells were stained with rabbit anti-NR1 antibody (1 : 50 in PBS, plus 1% newborn calf serum; Chemicon International, Temecula, CA, USA), using tetramethylrhodamine isothiocyanate (TRITC)-labeled antirabbit IgG as secondary antibody. The stained cells were washed three times with PBS plus 1% newborn calf serum, then mounted in fluoromount and examined with a Zeiss LSM510 confocal microscope (Carl Zeiss).

Primary cultures from cerebral cortex of rats at E18 were exposed to A-GFP-NR1(+), A-GFP-NR1(–) or A-GFP, for 2 h at 37 °C (5% CO2). After infection, the medium containing the vectors was removed, and the cells were incubated for the next 24 h in fresh medium (Neurobasal/B27).

Western blots

Transfected or infected cell cultures were homogeneised in cold Tris-HCl 25 mM buffer, pH 7.4, containing, 5 mM EDTA, 1 mM EGTA, 1 mM PMSF and 10 µg/ml bacitracine. Aliquots of 25 µg protein (Bradford 1976) were submitted to unidirectional electrophoresis in 7.5% SDS-PAGE. NR1 subunit was detected with rabbit polyclonal affinity-purified anti-NR1 (Chemicon), and a secondary antibody, goat antirabbit IgG-horseradish peroxidase conjugate, revealed with a chemiluminiscent protein detection system (Amersham Pharmacia Biotech, Buckinghamshire, UK). The bands in the film were quantified by computerized densitometry (MCID program, Imaging Research Inc., St. Catharines, Canada).


Total RNA, from either transfected or infected cells, was obtained in Trizol Reagent (Gibco), and treated with DNase I (Roche Diagnostics GmbH, Mannheim, Germany) to eliminate DNA contamination. Reverse transcription was performed on 1 µg of RNA, using Omniscript reverse transcriptase kit (Qiagen) according to the manufacturer's instructions, and using a gene-specific primer (NR1 Antisense-nt1062: 5′AACCTGCAG- AACCGCAAG 3′). PCR reactions were performed with 2 µl of the RT-product in 50 µl Taq polymerase buffer (Sigma-Aldrich Chemie Gmbh, Steinheim, Germany), 2 mM MgCl2, 0.2 mM of each dNTP (Roche), 1 Unit Taq DNA polymerase (Sigma), 0.2 µM of each primer (NR1 Antisense and NR1 Sense-nt1396: 5′ GCTTGATGAGCAGGTCTATGC 3′) and 0.5 µCi of [32αP]-dCTP (Amersham). Reactions were run for 27 cycles (1 min at 95 °C, 1 min at 55 °C and 1.5 min at 72 °C, per cycle). Samples were purified using Micro Bio-Spin Chromatography Columns (Bio-Rad, Hercules, CA, USA), previously washed with TBE buffer and run on 5% agarose gel, then fixed in 7% acetic acid for 10 min, dried and exposed to the film.

Surgery and in vivo vector infusion

Adult male Wistar rats (weighing 200–250 g) were maintained in a 12:12 h light:dark cycle, 8 animals per cage, with food and water ad libitum.

The vectors were infused into the dorsal hippocampus through a 0.4-mm diameter hole in the skull, under 0.2 ml/100 g of body weight, of ketamine (50% solution; Kensol®):xilacine (2% solution; Rompun®) (3:1, intra-muscular (i.m) anesthesia. The appropriate coordinates (AP − 4.0; LL – 3.0; DV − 2.3) were determined according to the Atlas of Paxinos and Watson (1996), using a stereotaxic apparatus for small animals (David Kopf Instruments, Tujunga, CA, USA). The maximum volume injected was 0.8 µl per side, containing 5.106−5.107 vector particles/ml approximately, estimated from the GFP fluorescence for either vector stock (the injection lasted 2 min, and 2 more min to remove the needle).

Behavioral assays and histology

Five days after surgery, each rat was put into an open field (1.2 × 1.2 × 0.40 m height), with 64 quadrants in the floor. Rats freely explored the open field for 2 min, and the number of crossings from one quadrant to another, the number of rearings and grooming were observed and quantified. To evaluate habituation, a test session was performed 24 h after the training session. Subsequently, the same animals were trained for an inhibitory avoidance task. Six days after injection (one day after the first open field session), each rat was placed on an isolated platform (2.5 cm high × 8.0 cm wide × 30.0 cm long), on one side of an acrylic box (40 cm high × 30 cm wide × 60 cm long), whose floor was an electrifiable grid made of 3.0 mm diameter bronze parallel bars, spaced 5.0 mm apart. Latency to step down from the platform placing the four paws onto the grid was measured. In the training session, immediately upon stepping down, the animals received a 2 second 0.5 mA shock. In the test session performed 24 h later, no shock was given and the step down latency was recorded. Memory performance was evaluated by comparing test to training latencies. A vehicle-injected group, corresponding to the sham-operated control, and a group of naive rats were also trained and tested. The assays were carried out in a double-blind manner and were performed under a low intensity red light (less than 1 lux).

Some of the animals were killed five days after injection, without any other treatment. However, the trained rats were killed after the last test, at day seven. The brains were gently removed, washed in cold 0.32 M sucrose, and kept at − 70 °C until used. The animals included in the analyses were those in which the hippocampal injection was clearly indicated by the trace and lesion caused by the needle. 20 µm thick brain sections were obtained with a cryostat, they were mounted in gelatinized glass slides, fixed and stained for optic microscopy (hematoxilin-eosin). The fluorescence was observed under a photomicroscope Zeiss Axiophot (Carl Zeiss). The photomicrographs were obtained in 600 ASA color film.

All the procedures (housing, handling, surgical and behavioral methods) involving animals were carried out in accordance with the European Communities Council Directive of 24 November 1986 (86/609/EEC) and the School of Veterinary of Buenos Aires University Review Committee for use of Animal Subjects.


Transgenic expression from amplicon plasmids and vectors in cultured cells

The amplicon plasmids and vectors used in these experiments were expected to express both mRNA for GFP, and either NR1 mRNA, or antisense RNA molecules to NR1 mRNA.

The expression of NR1 and GFP was assessed in culture cells by Western blot using specific antibodies against either protein, by GFP fluorescence and by immunocytochemistry.

Immunofluorescence studies in BHK cells revealed that those cells transfected with pA-GFP-NR1(+) simultaneously expressed both proteins in the same cell (Fig. 1b, red and green fluorescence). Furthermore, NR1 and GFP did not appear to colocalize to the same subcellular compartment. Similar experiments performed infecting VERO cells (not shown) and cortical neurons in primary cultures with A-GFP-NR1(+) vectors, also showed expression of both proteins in the same cell (Fig. 2). On the other hand, infection of neurons with A-GFP-NR1(–) vector shows expression of GFP alone (Fig. 2).

Figure 2.

Expression of GFP and NR1 proteins in cultured neurons infected with the double vectors. Primary cultures of neurons from rat cerebral cortex (E19) were infected with either A-GFP-NR1(+) or A-GFP-NR1(–) vectors. After 24 h the cells were fixed, treated with anti-NR1 Ab, and stained by using TRITC-labeled IgG as secondary Ab. The cells were examined with an Olympus microscope. Red and green fluorescence corresponds to NR1 and GFP proteins, respectively. Cultured neurons infected with (a) A-GFP-NR1-(–), amplification: 200x; (b) A-GFP-NR1(+), 200x; (c) A-GFP-NR1(+), 400x. Notice that GFP was expressed from both vectors, whereas NR1 was only expressed in the case of the A-GFP-NR1(+) vector.

Western blots of VERO cells infected with A-GFP-NR1(+) vector showed strong expression of both NR1 and GFP proteins (Fig. 3, lane 2). On the other hand, VERO cells either not infected (lane 1), or infected with A-GFP-NR1(–) or A-GFP vectors (lanes 3 and 4, respectively), did not show any reactive band to NR1-specific antibodies.

Figure 3.

Expression of NR1 and GFP in VERO cells infected with A-GFP, A-GFP-NR1(+) or A-GFP-NR1(–) vectors, and revealed by Western blot. Cells either not infected (lane 1) or infected with A-GFP-NR1(+) (lane 2), A-GFP-NR1(–) (lane 3),or A-GFP (lane 4), blotted using anti-NR1, antiactin and anti-GFP specific antibodies. Notice that NR1 protein is observed only after infection with A-GFP-NR1(+) vector.

Taken together, these experiments demonstrate that our vectors express NR1 and GFP proteins in vitro, and that this expression can be readily detected by immunofluorescence and by Western blots.

Expression of NR1 antisense RNA molecules was assessed by RT-PCR in BHK-21 cells infected with A-GFP-NR1(–) vector, using specific primers for antisense RNA. Figure 4 shows an amplified DNA band in cells infected with A-GFP-NR1(–) vector (lane 3). In contrast, no amplification was observed in cells infected with A-GFP-NR1(+) (lane 2), or in non-infected cells (lane 4), or in cells infected with A-GFP-NR1(–) but without reverse transcription (lane 5) either. Lane 6 shows a positive PCR control made on pA-GFP-NR1(–) plasmid.

Figure 4.

Synthesis of NR1 antisense RNA from A-GFP-NR1(–) vector. Total RNA was extracted from BHK cells, and a fragment of 334 bp was amplified by RT-PCR. Lane 1: molecular weight markers. Lane 2: cells infected with A-GFP-NR1(+). Lane 4: control non-infected cells. Lanes 3 and 5: cells infected with A-GFP-NR1(–), and amplification performed with (lane 3) or without (lane 5) reverse transcription. Lane 6: control PCR performed on pA-GFP-NR1(–) plasmid. Lane 7: H2O RT-PCR.

In order to test whether expression of antisense RNA from NR1(–) transgene was able to inhibit synthesis of NR1 protein, BHK-21 cells were cotransfected with a constant dose of pA-NR1(+), a plasmid expressing only NR1 protein (Cheli et al. 2002), and increasing amounts of pA-GFP-NR1(–). These experiments showed a significant and correlated decrease in NR1 protein, without any significant change on actin expression, even at the lowest concentration of the antisense plasmid (Fig. 5, lanes 3 and 4). The expression of pA-GFP-NR1(–) plasmid in these cells was assessed with GFP specific antibodies. As control, cells cotransfected with a constant amount of pA-NR1(+) and increasing doses of pA-GFP-LacZ plasmid, showed no significant changes in NR1 protein at the lowest concentration of pA-GFP-LacZ (lane 6), whereas there was only a slight reduction at the highest concentration of the plasmid (lane 7). The specific inhibition of NR1 expression by NR1(–) antisense RNA was also demonstrated in a similar experiment, where the total amount of transfected DNA was rendered constant by adding supplementary amounts of pA-GFP-LacZ to a mixture of pA-GFP-NR1(+) and pA-GFP-NR1(–) plasmids (data not shown). Taken together, these results strongly support the fact that NR1 expression was specifically inhibited by the NR1(–) transgene carried by pA-GFP-NR1(–).

Figure 5.

Inhibition of NR1 protein expression from pA-NR1(+) plasmid by cotransfection with pA-GFP-NR1(–). Lane 1 corresponds to non-transfected BHK cells. Lane 2 and 5: BHK cells were transfected with pA-NR1 (+) plasmid alone, or were cotransfected with pA-NR1(+) and increasing doses of either pA-GFP-NR1(–): lanes 3 and 4, or pA-GFP-LacZ: lanes 6 and 7. Proteins were blotted and revealed by anti-NR1, anti-GFP and antiactin specific antibodies, using an IgG-horseradish peroxidase conjugate secondary antibody. The cells transfected with pA-NR1(+) and cotransfected with pA-GFP-NR1(–) showed a significant decrease in the band corresponding to NR1 protein.

In vivo studies

Vectors from the same stocks were infused into the brain of adult male Wistar rats to study expression and behavior. The animals received each of the double vectors into the dorsal hippocampus and were killed 5 days after infection. The brains were used for histology (Fig. 6 a), and to observe the fluorescence (Fig. 6b-e). Figure 6(b,c) corresponds to the fluorescence in brain slices from rats infected with A-GFP-NR1(+), while Fig. 6 (d,e) corresponds to a rat injected with A-GFP-NR1(–) 5 days after injection. There was a remarkable GFP fluorescence in pyramidal cells, most of them in the CA1 region (Fig. 6c,d), and a few in CA3. A particularly intense fluorescence can be observed close to, and likely inside the cell nucleus (Fig. 6e). The expression was evident 48 h after injection, but was even higher at 72 h, appearing stable through the following 4 days (not shown).

Figure 6.

In vivo expression after injection of the vectors into the dorsal hippocampus of rats. Photomicrographs from rat brain slices inoculated with either A-GFP-NR1(–) or A-GFP-NR1(+), observed by transmitted visible light and by fluorescence. (a) Hematoxilin-eosin stained, magnification 25 × ; (b) A-GFP-NR1(+) in CA1, 200 × ; (c) A-GFP-NR1(+), 400 × ; (d) A-GFP-NR1(–), 400 × ; (e) A-GFP-NR1(–), 600x. Most cells exhibiting fluorescence are pyramidal neurons (the photographs have been amplified from 35 mm negatives, with same magnification, though the only difference is the microscope magnification).

The rats were injected with either the vehicle or each of the double vectors five days before performing the behavioral assays. All the injected animals appeared to behave as naive rats in the first session in the open field. The number of rearings and crossings from one quadrant to the other significantly diminished in the test compared to the training session, for naive animals as well as for sham operated (vehicle-injected, for crossings, P < 0.05; for rearings, P < 0.02; n = 13) (control groups) (Fig. 7a, Table 1). The animals inoculated with A-GFP-NR1(+) vector also showed a significant decrease in the number of crossings and rearings in the test session (Fig. 7a; for crossings, P < 0.05; Table 1; for rearings, P < 0.02; n = 12) (spending significantly longer time in central quadrants; not shown). On the other hand, the rats infused with A-GFP-NR1(–) performed as control animals in the training session, while in the test session showed no decrease in crossings (Fig. 7a), neither in rearings (Table 1). The grooming behavior was almost absent during the observation period.

Figure 7.

(a) Crossings in the open field after injection of the vectors into the dorsal hippocampus. Rats were injected with either A-GFP-NR1(+), A-GFP-NR1(–) or the vehicle. After 5 days, each animal was let to freely explore an open field for 2 min. The test session was performed 24 h later. Bars represent median of crossings (of 64 quadrants on the floor) per session, with interquartile ranges. ⋆Statistically significant differences by paired-Wilcoxon test (P < 0.05). ♠Statistically significant differences between treatments by Kruskal–Wallis one-way anova (P < 0.05). Numbers written in the bars represent the number of animals in each group. (b) Performance in the inhibitory avoidance task 6 days after injection of the vectors into the dorsal hippocampus. Rats were injected with either A-GFP-NR1(+), A-GFP-NR1(–) or the vehicle. Bars represent median of latencies (with interquartile ranges) to get down from an isolated platform onto a grid, where the animal got a mild shock in the training session; the test was performed 24 h later. ⋆Statistically significant differences by paired-Wilcoxon test (P < 0.05). ♠Statistically significant differences between treatments by Kruskal–Wallis one-way anova (P < 0.05). Numbers written in the bars represent the number of animals in each group.

After the open field task, the animals were trained for an inhibitory avoidance task in one step. All of them behaved in a similar way in the training session: there were no significant differences among training latencies to step down from the platform onto the grid, where the rat would receive a foot-shock (Fig. 7b). The rats injected with vectors carrying the NR1(+) transgene, as well as vehicle infused animals, achieved the criteria for the inhibitory avoidance learning, remaining significantly longer on the isolated platform in the test session. On the other hand, rats injected with A-GFP-NR1(–) vectors showed no significant differences between test and training latencies; therefore, they appeared unable to remember the task (Fig. 7b).


Amplicon vectors with the adequate transcription units were constructed and a suitable production was achieved and titrated by counting fluorescent cells in culture. The amplicon vector stocks were prepared using the novel HSV-1 LaL as helper virus (Logvinoff & Epstein 2000). Previous studies (Logvinoff & Epstein 2001; C. Zaupa & A. Epstein, personal communication) demonstrated that the HSV-1 LaL virus is highly neuroattenuated: mice infected with very high doses of it, directly into the brain, showed no symptoms or evidence of any clinical signs of disease, as opposed to mice infected with the wild type virus who died four to six days postinfection.

The amplicon plasmids and vectors were designed to express either GFP and NR1 proteins, or GFP and antisense RNA to NR1 mRNA (Fig. 1a). The GFP gene was conspicuously expressed from the vectors, both in cultured cells (Fig. 1b) and in rat brain in vivo, where numerous cells, mainly pyramidal neurons of CA1 in the dorsal hippocampus, showed GFP fluorescence (Fig. 6). Even though most of the fluorescent cells in the brain appeared to be intact neurons, further studies are required to evaluate the citotoxicity after longer periods, and also the possible expression in neuroglial cells.

In both cell lines and cultured neurons, either transfected with pA-GFP-NR1(+) plasmid or infected with the corresponding vector, NR1 and GFP were expressed in the same cell, though both proteins appeared not to be identically distributed. In a previous paper we have shown that NR1 protein synthesized from our vectors had the same electrophoretic mobility of the endogenous NR1 from PC12 cells, and was also recognized by the same anti-NR1 specific antibodies, corroborating its identity with the endogenous protein (Cheli et al. 2002). Although NR1 was expressed from those vectors carrying NR1(+) ORF, we do not know whether it would be assembled as a membrane receptor. Barría and Malinow (2002) have shown that exogenously expressed NR1 in culture neurons transfected with the corresponding plasmid was not assembled as NMDAR unless coexpressed with NR2 subunits. Therefore, it is likely that exogenously expressed NR1 would become part of the receptor complex in the cell surface, only in cells expressing NR2.

The antisense RNA expression from the A-GFP-NR1(–) vector was demonstrated by RT-PCR (Fig. 4). The ability of this RNA to specifically and efficiently inhibit the synthesis of the NR1 protein was shown in BHK-21 cells cotransfected with pA-NR1(+) and increasing amounts of pA-GFP-NR1(–). The expression of GFP from the transgene, as well as the endogenous actin, was not significantly modified, while there was a dose-dependent decrease in NR1 (Fig. 5). The results strongly support the fact that NR1 expression was specifically reduced by the NR1(–) transgene. The A-GFP-NR1(–) vectors did not induce exogenous or endogenous synthesis of NR1 protein. Hence, taken together, these results demonstrate that our vectors behaved as expected, specifically expressing either NR1 and GFP protein, or NR1 antisense and GFP.

It has been reported that a single dose of antisense oligonucleotide specific for NR1, unilaterally applied into the rat neostriatum, reduced NR1 mRNA and NR1 protein, leading to apomorphine-induced ipsilateral rotation (Lai et al. 2000). Kammesheidt et al. (1997) using an adenovirus-vector was able to express NR1 antisense oligonucleotides. However, immunological methods failed to demonstrate a reduction of NR1 protein, but whole-cell recording showed that scarce neurons in the transduced area were deficient in NMDA receptor-mediated synaptic current.

Although the NR1 subunit already assembled with NR2 in the cell surface appeared to have a relatively long half-life of about 20 h, there is another pool of NR1 with a short half-life of about 2 h which represents half the total amount of NR1 (Huh & Wenthold 1999). This could serve as a reserve and would support a mechanism for receptor insertion into the synaptic membrane when rapid changes in the receptor number are required, as it has been proposed for LTP (Liao et al. 1995). In our in vitro experiments with the A-GFP-NR1(–) vector, the speed of disappearence of NR1 was in agreement with a relatively short half-life pool for NR1. In a previous paper, we have shown a slight but reproducible reduction in hippocampal NR1 subunit in animals infected with single vectors carrying the NR1(–) sequence compared to those injected with vehicle, A-GFP or A-NR1(+) vectors (Cheli et al. 2002). Thus, the significant effects observed in behavior could be attributed to perturbations in NR1 turnover in certain cells of the neural network. For instance, almost all of the Schaffer collateral-commisural synapses on CA1 pyramidal cell spines possess NMDAR in the adult rat. It might be speculated that, in animals treated with A-GFP-NR1(–) vectors, the infected neurons would have been depleted of a short half-life pool of NR1, thus reducing its availability for assembling to NR2 subunits and being inserted into the synapse. But we do not know how many cells were actually infected, and thus, affected by vectors expressing the antisense. Taking into account the vector titre, an estimation of infection efficiency and consequent expression, it is likely that a reduction in NR1 expression would have taken place just in a few cells. However, this putative decrease was clearly enough to bring about behavioral modifications.

Modifications in NMDAR were evident after only one trial in the water maze (Jerusalinsky et al. 2000). This allowed us to suggest that the NMDAR could be particularly sensitive to novelty. To further explore this we decided to begin our behavioral studies with a simple hippocampal-dependent, novelty related task: habituation to an open field five days after vector infusion into the dorsal hippocampus. The rats injected with A-GFP-NR1(+) vector showed habituation as vehicle injected animals since the number of crossings and rearings significantly decreased in the test session. Thus, our amplicon vector did not have any effect on this task performance itself. On the other hand, animals injected with A-GFP-NR1(–) vectors did not show any decrease in crossings and rearings in the test compared to the training session, i.e. exhibited impaired habituation. These results are in agreement with those obtained with A-NR1(–) vectors (Cheli et al. 2002), strengthening the hypothesis that NR1 subunit in the dorsal hippocampus is involved in mechanisms related to habituation to a new environment.

To further study the participation of NMDAR in the dorsal hippocampus in learning and memory, we have trained infected animals in an inhibitory avoidance task. Again, the control groups as well as the A-GFP-NR1(+) vector injected rats learned, and remembered 24 h later to stay on the platform longer to avoid a mild foot-shock, while the A-GFP-NR1(–) injected rats failed to achieve this and could not learn/remember the task. This suggests that the NMDAR in the dorsal hippocampus is involved in recording the events to perform this task. An alternative hypothesis, suggested by Cain (1997, 1998) may be that naive rats trained under NMDAR antagonists display sensorimotor disturbances which interfere with their ability to acquire the task. We expect that further studies using the HSV-1 derived vectors, would allow clarification of this. We conclude that the constructed A-vectors are useful tools to modify gene expression in restricted regions of the CNS, at a defined time, and will allow investigation of in vivo functions, including behavior, without the problems associated with gene targeting or transgenic techniques.


This work has been supported by ECOS/SECYT Cooperation Program between France and Argentina, by the ‘Association Française contre les Myopathies’ (AFM), and the CNRS (France), the CONICET and FONCYT (Argentina) and the John S. Guggenheim Foundation. We are grateful to S. Nakanishi and R. Malinow for facilitating the plasmids with the NR1 ORF and to M. Besio Moreno and L. Pasquini for generously teaching and helping us with primary cultures.