Although a considerable number of reports have described the use of HSV and adenovirus as gene transfer agents in the peripheral nervous system, both are derived from disease-causing viruses that express structural proteins capable of eliciting an immune response, invariably leading to short-term gene expression (< 2 months) and cytotoxicity [6, 23, 24]. rAAV is a non-inflammatory vector that provides efficient and long-term expression in neurons in vivo and whose parent wild-type virus has never been associated with disease. In the present study, we have examined the ability of rAAV serotype 6 to deliver a transgene to the dorsal root ganglia of mice following five different routes of administration. We have carefully estimated the percentage of cells transduced and characterized the transduction profile in terms of cell size and colocalization with neuronal markers. When combined or compared with transgenic approaches, this vector may provide valuable information in the mechanistic study of pain.
The ability of rAAV2/6 to transduce sensory neurons from the periphery was examined first as this vector has been reported capable of retrograde axonal transport [18, 25]. Delivery into the triceps surae muscle of the hindlimb resulted in transduction of cells within the L4 DRG albeit at low levels. The transduction may have arisen through infection of intrafusal muscle fibers and Golgi tendon organs that are located in the muscle and send proprioceptive information to the spinal cord. This is in accordance with the observation that eGFP-positive fibers were located in the deeper lamina of the dorsal horn. Subcutaneous delivery of rAAV2/6 also resulted in low levels of eGFP-positive neurons in the L4 DRG. Curiously, transduction of the DRG following subcutaneous delivery has not been observed previously for rAAV, with rAAV serotype 2 (rAAV serotype 2 genome packaged in the serotype 2 capsid, rAAV2/2) failing to transduce DRG cells following subcutaneous injection into the hindpaw of rats . This is not surprising as rAAV2/2 is inefficient at retrograde transport in comparison to rAAV2/6 .
Cutaneous and subcutaneous delivery of HSV has frequently been used to transduce sensory neurons of rats and mice as proof-of-principle for gene therapy of chronic pain [16, 26]. We cannot compare the transduction observed here with previous reports as the vast majority has not quantified transduction efficiency. This is in part due to the use of secreted transgenic peptides that have had the capacity to exert paracrine functions on non-transduced cells. At least in one report, however, topical delivery of HSV was shown to achieve approximately 50% of total DRG neurons . Despite the relatively lower transduction observed in the present study, subcutaneous delivery of rAAV2/6, with further optimization, may be an alternative to HSV for the delivery of secreted transgenic peptides towards human gene therapy of pain. Indeed, rAAV is the only viral vector currently being examined in clinical trials for neurological disease .
Transgene delivery to nociceptive neurons can also be used to explore the nature of changes that lead to chronic pain states. For example, HSV has been used to investigate signaling pathways that result in altered ion channel functions that have become hyper-responsive to channel activation and pain signaling . Fink et al  have also used HSV to examine the role of TNFα signaling in chronic pain. With the goal of increasing the number of transduced nociceptive neurons that would be required for exploring the mechanisms of pain, we examined sciatic nerve and intrathecal routes administration of rAAV that have recently reported success in the transduction of rat DRG. Indeed, injection of rAAV2/6 into the sciatic nerve of C57BL/6 mice resulted in eGFP expression in more than 30% of the L4 DRG neurons. This is a substantial fraction of cells considering that the L4 DRG has afferent projections into the sciatic nerve and also other nerves of the lumbosacral plexus that were not injected (eg. pudendal nerve, obturator nerve, posterior cutaneous nerve of the thigh) that account for 40-50% of total L4/L5 DRG neurons . Interestingly, the efficiency and profile of transduction was similar for the three mouse strains examined, suggesting that rAAV2/6 transduction of DRG neurons can be applied to the mouse regardless of the model being used.
The level of transduction observed in this study was similar to sciatic nerve delivery of rAAV2/2 in the rat [13, 31, 32]. The serotype 6 rAAV used in the present report has previously been delivered to the sciatic nerve of rats to result in retrograde transport to motor neurons, however, transduction in the DRG was not examined . Other viral vectors have also been examined through this delivery route. HSV has resulted in efficient transduction of the L4 DRG, although declined dramatically after 3 weeks [8, 33, 34]. Adenovirus has also been administered to the sciatic nerve of rats and mice, although has led to transduction of predominantly Schwann cells on both accounts [35, 36].
The cell-size profile of transduced cells was next characterized to determine whether nociceptors were efficiently targeted. Approximately 90% of eGFP-positive cells were small-sized (<300 μm2) corresponding to putative nociceptive neurons. As only one half the total DRG cell population of mice are less than 300 μm2 , we can infer that there was a transduction preference for this small nociceptive cell population. Indeed, more frequent colocalization of eGFP was observed with markers of small DRG neurons (TRPV1, Substance P, CGRP, TrkA and IB4) than with large neurons (NF200) in the DRG. eGFP-positive fibers were present mainly in the superficial lamina of the spinal cord and colocalized extensively with CGRP1 (lamina I and II outer) and less with IB4 (lamina II inner) suggesting an enriched transduction of peptidergic neurons following this route of administration. Perhaps the apparent transduction preference of nociceptors is attributed to the degree of myelination that may impede entry of the virion into larger fibers following injection into the nerve. Further work would be required to confirm this.
Intrathecal administration into mice was next examined as this has previously been reported to result in efficient rAAV-mediated gene transfer to the DRG of rats [9, 10]. To negate the need of performing a laminectomy in the relatively smaller rodent, a four centimeter catheter was inserted into the cistern magna and moved to the level of the lumbar spinal cord [37, 38]. rAAV2/6 delivery resulted in the highest transduction observed in the present study, reaching approximately 60% of L4 DRG neurons. eGFP-positive cells colocalized with markers of nociceptive and non-nociceptive cells, however, upon analysis of cell area, there was a tendency for less small-sized neurons (<300 μm2) (P = 0.15) and more medium-sized neurons (300 - 700 μm2) (P = 0.18) than for the sciatic nerve delivery. Importantly, we found that transduction efficiency was unaltered following nerve ligation, demonstrating that rAAV2/6 can be used in models of neuropathic pain.
The results observed here in mice are similar to that obtained with rAAV serotype 8 (rAAV serotype 2 genome packaged in the serotype 8 capsid) in rats, that led to high levels of transduction of TRPV1, Substance P, CGRP and IB4 positive cells within the DRG . Curiously, rAAV2/2 failed to transduce DRG cells following this mode of delivery  demonstrating different cellular tropisms among rAAV serotypes. Adenovirus has also been delivered intrathecally for chronic pain studies to result in efficient (although transient) expression of proteins within the cerebrospinal fluid, such as beta-endorphin  and interleukin-10 , although transduction was not examined in the DRG.
Although intrathecal delivery results in transduction of more DRG neurons, the sciatic nerve injection still has advantages for studies of neuropathic pain. Unlike intrathecal administration, injection of the sciatic nerve can be performed unilaterally, which allows the non/vehicle-treated side to serve as a useful internal control for behavior or histology. This specificity of the nerve injection is further exemplified by the finding that only this injection delivery route failed to result in spread of the vector into the blood stream, whereas intrathecal administration gave considerable transduction in the liver. Direct nerve injection can also be used to target injured versus non-injured nerves, for example, in the spared nerve injury model , where the spread of rAAV2/6 following intrathecal delivery would preclude this. Intrathecal administration may therefore be more suited to behavioral studies where the greater number of cells transduced may allow better discrimination of changes to pain perception in the whole animal.
Intravenous administration via the tail vein was also examined as this had previously resulted in transgene expression in the fibers of the spinal cord dorsal horn with rAAV2/8  and rAAV2/6 . This mode of delivery resulted in eGFP-positive cells within the DRG at lumbar, thoracic and cervical levels of spinal cord, and colocalized with the marker of non-nociceptive neurons, NF200. This novel mode of transduction may have resulted through infection of nerve endings within the muscle. This follows the reasoning that 1) skeletal muscle is highly vascularized and 2) the transduction pattern in the deeper lamina of the spinal cord is similar to that observed after intramuscular rAAV2/6 delivery.
One caveat of this study has been that eGFP expression was examined only at the designated three week time point. Previously we have reported expression in neurons at least 8 weeks following injection with the rAAV2/6 vector  and also observed stable expression for at least 5 months in spinal cord motor neurons (unpublished observations). Taken together with the multitude of studies demonstrating long-term transgene expression for rAAV in neurons, we believe that the rAAV2/6-mediated transgene expression in nociceptive neurons would also be stable through time.