293T cells (CRL-11268), human osteosarcoma (HOS) cells (CRL-1543) and BHK21 cells (CCL-10) were obtained from the American Type Culture Collection (ATCC). DMEM/10% FBS (Invitrogen) was used to propagate 293T and HOS cells and Eagle's Minimum Essential Medium (EMEM)/10% FBS for BHK21 cells. Human neuroblastoma SH-SY5Y cells (ATCC CRL-2266) were grown to 70% confluence and maintained in a growth medium containing a 1:1 mixture of EMEM and Ham's F12 medium (90%) (Invitrogen), supplemented with 10% FBS (HyClone). Medium was renewed every 2 days. The NSC-34 neuroblastoma-spinal cord hybrid cell line 24 was provided by Dr. Neil Cashman, Toronto. NSC-34 cells were propagated in DMEM/10% FBS.

Astrocytes were isolated from embryonic day 15 mouse cerebral cortices and kept in 75-cm² flasks coated with poly-D-lysine containing a 1:1 mixture of DMEM and Ham's F12 medium supplemented with 5% FBS, 5% horse serum, 2 mM GlutaMAX, 100 U/ml penicillin and 100 mg/ml streptomycin. For transduction, cells were harvested and plated into 8-well chamber slides (Nalge Nunc) coated with poly-D-lysine.

Spinal cords were obtained under sterile conditions from 15-day Sprague Dawley rat embryos. Dorsal root ganglia (DRGs) and perineural membranes were removed and cords cut into approximately 2-mm sections, which were then triturated and dissociated in a 0.05% trypsin/0.53 mM EDTA solution. Cells were collected, centrifuged for 5 min at 1800 rpm, and resuspended in complete growth medium made in supplemented Neurobasal Medium (Invitrogen-GIBCO). Cells were plated on glass coverslips in multi-well cell culture plates pre-coated with poly-L-lysine (24 h, 0.005% in H₂O).

DRGs were carefully removed from 15-day-old Sprague-Dawley rat embryo cords and cultured following an established protocol 25. Briefly, DRG explants were plated into the inner compartments of Campenot compartmentalized chambers 26 in a small volume of medium and allowed to adhere for two hours. The chambers were prepared as follows: 35-mm cell culture dishes were pre-coated with a diluted collagen solution (three parts of sterile distilled H₂O to one part of rat tail collagen type I – 2 mg/ml – Roche Diagnostics Corp.). Scratches were then made in the collagen substratum using a pin rake, to guide axon growth. A drop of medium was placed onto the plate before setting the divider onto the culture dish, in order to facilitate axon growth underneath the silicon grease barriers. Teflon chamber dividers (Tyler Research) were then carefully attached to the collagen-coated dishes using silicone vacuum grease (Dow Corning). After 1 h in the incubator, the inner compartments of the chambers were filled with culture medium to check for leakage. Chambers that showed any leakage were discarded and those that appeared well-sealed were filled with culture medium. Fluorescence was also used to demonstrate the absence of leakage between chambers, by adding a fluorescent tracer (DiI – Molecular Probes, Invitrogen Corporation) in the inner compartment. The compartments were then filled with growth medium, consisting of Neurobasal medium supplemented with B-27 additive (Invitrogen), 50 ng/ml of Nerve Growth Factor 2.5S (NGF) (Invitrogen), and 40 μM of 5'-Fluoro-2'-deoxyuridine – FUDR (Sigma-Aldrich). The outer compartments were filled with growth medium supplemented with 100–200 ng/ml of NGF to coax neurite growth into these compartments. Cells were re-fed every other day. The outer compartments were also re-filled every other day with growth medium supplemented with NGF.

SH-SY5Y cells and mixed spinal cord cell cultures were processed for β-Gal staining 3 days after transduction using an X-Gal staining kit (Invitrogen). In order to identify neurons and glia in the cultures transduced with the EGFP-encoding vectors, ICC was performed with antibody directed against microtubule associated protein 2 (Map-2) and glial fibrillary acidic protein (GFAP). The Map-2 antibody (Covance Research Products) was used at a concentration of 1:10,000 and anti-GFAP antibody (Promega) at a concentration of 1:1000. Fluorescent cells were visualized using a Nikon E400 upright microscope.

The pNL-EGFP/CMV/WPREΔ U3 lentivirus vector plasmid was described before 27. In the pNL-EGFP/lacZco/WPREΔ U3 lentiviral vector plasmid 28, a codon-optimized lacZ gene sequence encoding β-galactosidase (β-Gal) 29 was used to replace the EGFP transgene sequence. The VSV-G-encoding pLTR-G plasmid was described before 30. Plasmids encoding glycoproteins of the Rabies virus PV strain (GenBank Accession number: A14671), Duvenhage virus (DuvSAF2 strain) (GenBank Accession number: AF298147), European bat Lyssavirus (EBL1FRA strain) (GenBank Accession number: AF298143) and Lagos bat Lyssavirus (LagNGA strain) (GenBank Accession number: AF298148) 23 were constructed as follows: Total RNA was extracted from BHK-21 cells infected with the various Lyssavirus isolates and reverse-transcribed using an ImProm-II Reverse Transcription System (Promega) and primers corresponding to the 3'-untranslated regions of the various viral RNAs. The primers used for reverse transcription were: Rabies PV: 5' CGG GAT CCG GCC AGC TCT CAC AGT CCG GT; DuvSAF: 5' CGG GAT CCC TCT CAC TCC CTT GTT GAT GG; EBL1: 5' CGG GAT CCT GCT TAT GAC TCA CAA GTA GT; LagNGA: 5' GGA ATT CTT GTT ACC ATG AGT CAA CTA AAA. The glycoprotein coding regions were subsequently PCR amplified using AccuPrime™ Pfx DNA Polymerase (Invitrogen) and subcloned into pLTR 30 to yield pLTR-PV, pLTR-DuvSAF, pLTR-EBL1 and pLTR-LagNGA, respectively. The primers used for PCR were as follows: Rabies PV: 5' GGA ATT CCA AGG AAA GAT GGT TCC TCA G (sense), 5' CGG GAT CCG GCC AGC TCT CAC AGT CCG GT (antisense); DuvSAF: 5' GGA ATT CAC CAT GCC ACT CAA TGC AGT CA (sense), 5' CGG GAT CCC TCT CAC TCC CTT GTT GAT GG (antisense); EBL1: 5' GGA ATT CAC CAT GTT ACT CTC TAC CGC CA (sense), 5' CGG GAT CCT GCT TAT GAC TCA CAA GTA GT (antisense); LagNGA: 5' CCC CCG GGA TCA GAC ATT AGA GCT ACC CT (sense), 5' GGA ATT CTT GTT ACC ATG AGT CAA CTA AAA (antisense). All glycoprotein-encoding sequences were analyzed by DNA sequencing.

For histological analyses, rats were deeply anesthetized with sodium pentobarbital (100 mg/kg, I.P.) and transcardially perfused with a 0.9% saline solution, followed by 4% paraformaldehyde (Sigma-Aldrich) in phosphate buffer saline (PBS), pH 7.4. Sciatic nerves, DRGs, and spinal cords were dissected, post-fixed for 24 h, and transferred to a 30% solution of sucrose. Tissues were first processed for β-Gal staining and then frozen in optimal cutting temperature gel (OCT; Sakura Finetek USA) and stored until histological processing. Nerves, DRGs, and spinal cords were cut at 25 μm in transverse sections on a cryostat (Leica Microsystems) and mounted onto slides. Counter-staining was performed on transverse sections using Eosin. Finally, the slides were mounted with coverslips using Permount mounting medium (Fisher Scientific).

To first determine the efficiency with which these vectors could transduce neuronal cells, we used SH-SY5Y cells. In this experiment, vector MOIs were adjusted based on functional titers (TU). Overall, Rabies PV-pseudotyped vectors performed better than the other pseudotypes in this assay, transducing 20.76 (± 1.1) and 36.78 (± 1.05) % of the cells at MOIs 0.1 and 1, respectively (Figure 1). Moreover, it was interesting to note that only the transduction efficiency of Rabies PV-treated cells significantly increased as a result of a higher MOI (two-way ANOVA, p < 0.05).

In this set of experiments, MOIs were adjusted based on particle titers determined by quantitative RT-PCR of virion-derived RNA to rule out effects caused by differences in receptor levels in target cells. Particle titer-adjusted pseudotypes were tested in the NSC-34 neuroblastoma-spinal cord hybrid cell line 24. Vectors encoding EGFP were used to facilitate the analysis of transduced cells by flow cytometry (FACS). HOS cells that are easily transduced by lentiviral vectors 3036 were tested in parallel to compare the performance of the various pseudotypes (data not shown). Cells were processed for FACS 3 days after treatment as described 33. The results presented in Figure 2 indicate that at an MOI of 5 × 10³ vector particles per cell, Rabies PV pseudotypes transduced 49.9% of NSC-34 cells, while the performance of all other pseudotypes was considerably lower, including a VSV-G pseudotype that was used for comparison.

The vectors were then tested in mixed spinal cord cultures to assess their potential specificity for neurons as opposed to glial cells. Cultures from 15 day-old rat embryos were transduced and processed 3 days later for cell identification. Two different kinds of vectors were used: vectors encoding β-Gal (10³ vector particles per cell) and vectors encoding EGFP (10⁴ vector particles per cell). Overall, the transduction efficiencies in mixed spinal cord cultures were similar to those observed in SH-SY5Y and NSC-34 cells, i.e. vectors pseudotyped with the Rabies PV glycoprotein revealed the highest transduction efficiency to neurons (70.35 ± 9.66% of transduced cells) (Figures 3A and 3B) followed by the others. The difference between vectors reached significance in all groups (p < 0.05).

Because all the vectors were able to transduce glial cells at some level in mixed cultures (Figure 3C – dark gray bars), we decided to investigate this effect in more detail using primary pure astrocytes. Indeed the results presented in Figure 3C show that all pseudotypes have the ability to transduce astrocytes in either mixed or pure astrocyte populations. Interestingly, the effect of the culture type (mixed or pure) did not affect the pattern of transduction in the Rabies PV, LAgNGA, and EBL1 conditions. The percentage of transduced astrocytes by VSV-G and DuvSAF pseudotyped vectors, however, significantly decreased in pure astrocytes compared to mixed cultures (Figure 3C). A two-way ANOVA analyzing vector and culture type revealed a significant effect in these particular groups (*, ** p < 0.05).

Aiming at using these pseudotypes as vehicles for CNS targeting by peripheral injection, we then evaluated their uptake when peripherally applied to axons terminals. For this purpose, DRG explants grown in compartmented chambers were used. The MOIs of vectors were normalized based on particle titers and 10 μl (2 × 10⁸ vector particles total) of each pseudotyped vector were added to the chambers. Only one outer compartment of the chambers containing axon terminals was treated and, 3 days later, fluorescence was assessed in the central compartment containing the cell bodies. When vectors were directly added to the inner compartment containing the DRG cell bodies, fluorescence could be detected with all the pseudotypes (data not shown). On the other hand, when vectors were added to the compartments containing the axon terminals, fluorescence could only be detected in DRG explants of chambers treated with the Rabies PV and the DuvSAF vectors (Figure 4), indicating their ability to be taken up and be retrogradely transported to the cell bodies of the DRG explants.

In order to assess the uptake and retrograde axonal transport of our pseudotyped lentiviral vectors in vivo, we next used the remote viral gene delivery model 3741. 2.5 × 10⁴ TU of the different pseudotypes in a volume of 2 μl were injected into the crushed sciatic nerve of rats. To allow optimal transgene expression, animals were euthanized after 4 weeks and sciatic nerves, ipsilateral lumbar DRGs, and lumbar spinal cords were removed for analysis. Sections were stained using X-Gal to visualize β-Gal expressing cells. While all the vectors mediated some kind of retrograde transport to DRGs in vivo, β-Gal transgene expression was only observed in ventral horn motor neurons of the spinal cords injected with the Rabies PV-pseudotyped vector (Figure 5).