Genomic DNA was extracted from the tonsil tissue of a healthy adult male mongrel dog using the DNeasy Tissue Kit (Qiagen, Germany). Primers were generated using the sequences of human and mouse β-NGF genes to amplify partial dβ-NGF sequences. PCR using the primers NGF-1F (5'-TCAGCATTCCCTTGACACWG-3') and NGF-1R (5'-AGCCTTCCTGCTGAGCAC-3') was performed for 35 cycles at 94℃ for 1 min, 44℃ for 1min and 72℃ for 1 min (Fig. 1). PCR using the primers NGF-2F (5'-AGTTCTCGGTGTGCGACAG-3') and NGF-2R (5'-GCCCAGGAGAGTGTGGAG-3') was also performed for 35 cycles at 94℃ for 1 min, 55℃ for 1 min and 72℃ for 1 min (Fig. 1). A PCR product of approximately 600 bp was expected when using the primers NGF-1F and NGF-1R and 400 bp with the primers NGF-2F and primer NGF-2R. Overlapping PCR was performed to combine the PCR products of the partial dβ-NGF using NGF-1F and NGF-2R. This overlap PCR was performed for 35 cycles at 94℃ for 1 min, 58℃ for 1 min and 72℃ for 1 min. The size of this PCR product was confirmed to be approximately 660 bp (pcDNA1) using 1.5% agarose gel electrophoresis. The cloning of pcDNA1 was performed with pCR2.1 vector (Invitrogen, USA) and plasmid DNA was extracted from Escherichia coli TOP10 cells (Invitrogen, USA) with the plasmid purification kit (NucleoGen Biotechnology, Korea). The plasmids were sequenced by Takara-Korea Biomedical Inc. The remainder of dβ-NGF DNA was synthesized artificially based on the sequence of the predicted Canis familiaris nerve growth factor beta (5'-ATGTCCATGTTGTTCTACACTCTGATCACAGCTCTTCTGATCGGCATCCGGGCAGAACCGCATCCAGAGAGCCATGTCCCAGCAGGACACGCCATCCCCCACGCCCACTGGACTAAGCTTCAGCATTCCCTT-3'; GeneBank sequence entry XM_540250; NCBI, USA) (Fig. 1). Overlapping PCR was performed to combine the synthesized oligonucleotide and pcDNA1. The sense primer (Bgl2 + NGF-F) with a sequence of 5'-GGCAGATCTATGTCCATGTTG-3', and the antisense primer (EcoR1 + NGF-R) with a sequence of 5'-GGAGAATTCTCAGGCTCGTCT-3', were used for the overlap PCR. The sense primer had the BglII (TaKaRa Bio, Japan) restriction site and the antisense had the EcoR I (TaKaRa Bio, Japan) restriction site. PCR was performed for 35 cycles at 94℃ for 1 min, 53℃ for 1 min and 72℃ for 1 min. The obtained PCR product was expected to be 725 bp (pcDNA2). The pcDNA2 was cloned and the nucleotide sequence was analyzed. The pcDNA2 clone was named the presumed dog β-NGF (pdβ-NGF).

The CMV vector (phCMV1; Gene Therapy Systems, USA) was chosen to construct the recombinant pdβ-NGF (rpdβ-NGF) plasmid. The phCMV vector and pcDNA2 were digested with Bgl II and EcoR I restriction enzymes, and purified following separation on a 1.5% agarose gel. Ligation of the purified phCMV1 and pcDNA2 was performed using T4 DNA ligase (TaKaRa Bio, Japan). The rpdβ-NGF plasmid was transformed into Escherichia coli TOP10 cells and DNA extracted with a plasmid purification kit.

The rpdβ-NGF plasmid and a separate phCMV1 vector plasmid prepared for transfection were free of protein, RNA and chemical contamination (A₂₆₀/A₂₈₀ ratio of 1.9) and had a final concentration of 0.4 mg/ml.

A cationic liposome-mediated transfection technique (Gene therapy systems, USA) was carried out to deliver the two plasmids into Chinese hamster ovary (CHO) cells (Korean Cell Line Bank, Korea). Four days after the transfections, supernatants were collected and filtered through a 0.22-µm filter (Millipore, USA).

pdβ-NGF protein levels were measured using the Duoset Enzyme-Linked Immunosorbent Assay (ELISA) development system (R&D Systems, USA). This system used a sandwich ELISA method with anti-human β-NGF as the detection antibody and was performed according to the manufacturer's recommendations. The bioactivity of the pdβ-NGF protein was assessed using the rat pheochromocytoma cell line (PC12 cell; Seoul National University, Korea). PC12 cells were plated at a density of 5 × 10⁴ cells/ml in 24-well tissue culture plates (Falcon, USA). One ml of supernatant from the rpdβ-NGF-tranfected CHO cells was added to the PC12 cell cultures. Supernatant from CMV-transfected CHO cells, was used as a negative control. PC 12 cells were monitored daily by microscopic examination.

Ten mongrel dogs (5 males and 5 females) roughly 2 years of age were used in this experiment. The body weights ranged from 4 to 6 kg. Among them, two dogs were in the negative control group, four dogs were in positive control group and four dogs were in the experimental group with gene therapy. All of the dogs were clinically judged to be in good health and neurologically normal, and had their own admission number from the Institute of Laboratory Animal Resources, Seoul National University (SNU-060623-1). During the experiment, all of the dogs were cared for according to the Animal Care and Use Guidelines (Institute of Laboratory Animal Resources, Seoul National University, Korea). Body weights of test dogs were measured every morning during the test period.

The CMV vector containing rpdβ-NGF was prepared in advance. The cationic polymer transfection reagent (Polyplus transfection, France) was used to transport these plasmids into the intrathecal region. Each plasmid was condensed with in vivo-jetPET-Gal reagent at a 10-N/P ratio (measure of the ionic balance of the complexes). First, the prepared plasmids were diluted with 200 µl of 5% glucose (w/v) and an appropriate amount of in vivo-jetPET-Gal reagent in 200 µl of 5% glucose (w/v). Second, 200 µl of in vivo-jetPET-Gal solution was added to the plasmid solution followed by incubation for 15 min at room temperature. Third, the mixture was injected to the dogs of the gene therapy group (n = 4) through intrathecal injection using a 27-gauge needle. Before this administration, the dogs of the gene therapy group were anesthetized with zoletil.

Twenty-four hours after vector inoculation, the dogs from the gene therapy group (n = 4) and the positive control group (n = 4) were intoxicated with pyridoxine (Sigma, France). The pyridoxine was prepared in distilled water (100 mg/ml) immediately before injection, and administered at 150 mg/kg subcutaneously once a day in the morning, for 7 days. Dogs in the negative control group (n = 4) received vehicle (isoosmotic sterile aqueous solution of sodium chloride).

Postural reaction (wheelbarrowing, hopping, extensor postural thrust, placing, tonic neck reaction and proprioceptive positioning) assessments were done on all dogs every morning during the test period.

All of the dogs were preanesthetized with atropine (0.1 mg/kg of body weight, IM). Anesthesia was induced with diazepam and was maintained with isoflurane through a semiclosed system. Subcutaneous temperature was maintained at 37~38℃. Neuropack2 (Nihon Kohden, Japan) was used for all recordings. All measurements were performed in the left hindlimb. M wave was recorded for the tibial nerve, using 1 Hz, 0.5 ms, supramaximal stimulus. Stimulating electrodes were positioned in the distal tibial nerve. The recording electrode was positioned in the plantar interosseous muscle. The ground electrode was positioned between the stimulating electrode and the recording electrode. The recording electrode was a bipolar needle electrode. The Hoffman (H)-reflex was recorded using 1 Hz, 0.5 ms, submaximal stimulus. The stimulating electrode was positioned in the tibial nerve adjacent to the hook and the recording and ground electrodes were positioned in the same site of the tibial nerve where the M wave was measured. All measurements were performed at least eight times. Electrophysiological recordings were performed twice, once before the experiment and once after the test period.

After the experimental period (10 days from the start of the experiment), the dogs were anesthetized with a high dose of Tiletamine/zolazepam and propofol, and perfused transcardially with 0.1 M phosphate-buffered saline (PBS), followed by 4% paraformaldehyde in 0.1 M PBS to induce euthanasia. After perfusion, tissues (lumbar spinal cord (L₄), left and right dorsal root ganglia of L₄ and sciatic nerve) were quickly removed and post-fixed for 4~6 h in the same fixative at 4℃ and embedded in paraffin. The tissues were sectioned serially with a thickness of 5 µm using a microtome (Reichert-Jung, Germany) and floated onto gelatine-coated slides. Next, they were deparaffinized in xylene, rehydrated in a descending ethanol series, and stained with hematoxylin and eosin. The sections were observed using an Olympus BX51 microscope (Olympus, Japan) attached to a IMT2000 digital camera (iMTechnology, Korea) and images were captured using Adobe Photoshop version 6.0 software via IMT2000.

A Paired t-test was done for the analysis of body weights, M wave and H-reflex amplitudes before and after the pharmacologic treatment. The level of significance was set at p < 0.05.

The results of nucleotide sequence analysis showed that the gene cloned in pcDNA1 had a high degree of sequence homology with other mammalian β-NGF genes. The pcDNA1 sequence shared 86% and 83% sequence homology with that of human (GeneBank sequence entry NM_002506; NCBI, USA) and mouse (GeneBank sequence entry NM_013609; NCBI, USA) β-NGF sequences, respectively. The remaining part of dβ-NGF DNA, which was synthesized artificially, was included in the 5' region of the dβ-NGF open reading frame (ORF) and contained 132 base pairs. Overlapping PCR was performed to combine the synthesized oligonucleotide and pcDNA1, and the obtained PCR product was 725 bp (pcDNA2). Again, the nucleotide sequence analysis showed that the cloned gene had a high degree of sequence homology with other mammalian β-NGF genes. The pcDNA2 clone was named the presumed dog β-NGF (pdβ-NGF) (Fig. 1). With the additional sequence contributed by the syntesized oligoneucleotide, the shared homology changed to 85% and 81% compared to human (GeneBank sequence entry NM_002506; NCBI, USA) and mouse (GeneBank sequence entry NM_013609; NCBI, USA) β-NGF sequences, respectively. The deduced pdβ-NGF amino acid sequence shared 90% and 82% homology with that of human (GeneBank sequence entry NM_002506; NCBI, USA) and mouse (GeneBank sequence entry NM_013609; NCBI, USA), respectively.

Four days after transfection to CHO cells, pdβ-NGF protein was obtained from the supernatant. The filtered supernatant was measured with sandwich ELISA of human β-NGF. The results indicated that 53 pg/ml of pdβ-NGF protein existed in the supernatant. The bioactivity of pdβ-NGF protein was assessed using the rat pheochromocytoma cell line (PC12 cell; Seoul National University, Korea). Seven days after treatment with filtered supernatant, a small number of PC12 cells had neurite outgrowth, while the PC12 cells in the negative control group maintained their original morphology (Fig. 2).

The weight measurements showed that there was weight loss only in the positive control group. There were no weight changes in the negative control group or the gene therapy group. The difference in body weight of the positive control group was statistically significant (p < 0.05). The differences in body weight of the negative control group and the gene therapy group were not statistically significant (p < 0.05).

All the dogs in the positive control group developed a neurological disorder, characterized by ataxia involving first, and most prominently, the hindquarters. All of the dogs in the positive control group started to show proprioceptive abnormalities involving the hindquarters as detected by the postural reaction test (wheelbarrowing, hopping, extensor postural thrust, placing, tonic neck reaction and proprioceptive positioning) on the third day of pyridoxine injection. On the fourth day of pyridoxine injection, all dogs held their hindlimb stiffly when standing. These conditions were maintained until the end of the pyridoxine injection. On the other hand, all of the dogs in the negative control group and the gene therapy group were normal during the postural reaction test.

Electrophysiological readings were recorded to measure M wave and H reflex in all treatment groups. The M wave amplitude of all the dogs in the negative control group, the positive control group, and the gene therapy group showed no remarkable change before and after the pyridoxine administration as confirmed by statistical analysis (p < 0.05). However, there was a remarkable change in H reflex before and after the pyridoxine intoxication in the positive control group. Before the pyridoxine intoxication, the amplitude of H reflex was 0.52 ± 0.06 mV. After the pyridoxine intoxication, however, there was no consistently detectable H reflex in the positive control group. The H reflexes in the negative control group and the gene therapy group did not change before and after the pyridoxine intoxication as confirmed by statistical analysis (p < 0.05).

Histopathologically, there were no lesions in the lateral, dorsal or ventral funiculus, or in the gray matter of L₄ in the negative control group (Fig. 3A). The axons and myelin was disrupted with vacuolation in the positive control group (Fig. 3B). In the gene therapy group, swollen axons were occasionally seen in the dorsal funiculi of L₄ (Fig. 3C).

There were no lesions in the dorsal root ganglia (DRG) of L₄ in the negative control group (Fig. 4A). However, severe chromatolysis was observed in the neurons of DRG of L₄ in the positive control group. Vacuolation was also observed in the neurons. Occasionally, some neurons were necrotic, and were characterized by pyknotic nuclei and eosinophilic cytoplasm (Fig. 4B). Some neurons had pyknotic nuclei and eosinophilic cytoplasm in the gene therapy group (Fig. 4C).

There were no lesions in the axons or myelin in peripheral nerves (sciatic nerve) of the negative control group (Fig. 5A). Severe vacuolation was seen in the myelin in peripheral nerves (sciatic nerve) of the positive control group (Fig. 5B). There was mild vacuolation in the myelin in peripheral nerves (sciatic nerve) of the gene therapy group (Fig. 5C).