Retrograde adenoviral vector targeting of nociresponsive pontospinal noradrenergic neurons in the rat in vivo.

The spinal dorsal horn receives a dense innervation of noradrenaline-containing fibers that originate from pontine neurons in the A5, locus coeruleus (LC), and A7 cell groups. These pontospinal neurons are believed to constitute a component of the endogenous analgesic system. We used an adenoviral vector with a catecholaminergic-selective promoter (AVV-PRS) to retrogradely label the noradrenergic neurons projecting to the lumbar (L4-L5) dorsal horn with enhanced green fluorescent protein (EGFP) or monomeric red fluorescent protein (mRFP). Retrogradely labeled neurons (145 +/- 12, n = 14) were found in A5-12%, LC-80% and A7-8% after injection of AVV-PRS-EGFP to the dorsal horn of L4-L5. These neurons were immunopositive for dopamine beta-hydroxylase, indicating that they were catecholaminergic. Retrograde labeling was optimal 7 days after injection, persisted for over 4 weeks, and was dependent on viral vector titer. The spinal topography of the noradrenergic projection was examined using EGFP- and mRFP-expressing adenoviral vectors. Pontospinal neurons provide bilateral innervation of the cord and there was little overlap in the distribution of neurons projecting to the cervical and lumbar regions. The axonal arbor of the pontospinal neurons was visualized with GFP immunocytochemistry to show projections to the inferior olive, cerebellum, thalamus, and cortex but not to the hippocampus or caudate putamen. Formalin testing evoked c-fos expression in these pontospinal neurons, suggesting that they were nociresponsive (A5-21%, LC-16%, and A7-26%, n = 8). Thus, we have developed a viral vector-based strategy to selectively, retrogradely target the pontospinal noradrenergic neurons that are likely to be involved in the descending control of nociception.


Surgical procedures
Wistar rats (130 -200 g) were anesthetized (i.m. or i.p.) with ketamine (5 mg / 100 g, Vetalar, Pharmacia, Northamptonshire, UK) and medetomidine (30 g / 100 g, Domitor, Pfizer, Kent, UK) until loss of hindpaw withdrawal reflex. The animal was placed in a stereotaxic frame and secured in atraumatic ear bars. Core temperature was monitored using a rectal thermistor and maintained at 37°C with a homeothermic blanket (Harvard Apparatus, Kent, UK). Aseptic surgical techniques were employed throughout.
Lumbar spinal exposure. The spinous process of T13 was identified by tracing the lowest rib to the vertebrae. Through a midline incision, the paraspinous muscles were detached from T12-L2 and a laminectomy was performed using fine rongeurs (Fine Science Tools, Heidelberg, Germany) to allow access to the spinal cord at L4 -5. The spinous process of L3 was fixed in a spinal clamp (Narishige, Tokyo, Japan) to minimize movement during injections.
Cervical spinal exposure. The head of the rat was angled down (incisor bar at ؊20 mm) and through a midline incision the paraspinous muscles were parted to define the prominent T2 spinous process that was clamped and axially distracted. This allowed a laminectomy to be performed to expose the dorsal aspects of C6 -C7 segments of the spinal cord. Adenoviral vector injection. Injections of AVV were made using a microcapillary pipette (calibrated in 1 L intervals; Sigma, St. Louis, MO), with a tip diameter of around 20 m. A 25 L syringe (Hamilton, Bonaduz, Switzerland) was connected to the pipette with tubing primed with mineral oil. Using a syringe driver (Genie; Kent Scientific, Litchfield, CT) the pipette was backfilled with AVV. The pipette was directed into the dorsal horn using a micromanipulator (Narishige) to coordinates 400 m lateral to the midline, 500 m deep to dorsal surface. The successful delivery of AVV was visually confirmed on each occasion by observing the movement of the oil-vector medium interface in the calibrated capillary. Unless otherwise stated, two pairs of bilateral injections (each 500 m apart in the rostrocaudal axis) of AVV (500 nL/injection over 2 minutes) were made into a single spinal segment (Fig. 1).
Recovery. The paraspinous muscles were sutured and the skin was closed with clips. Animals were given 1 mL sterile saline (i.p.) for fluid replacement, buprenorphine (2 g / 100 g s.c., Temgesic, Schering-Plough, Hertfordshire, UK) for pain relief and atipamezole (an alpha2-adrenoceptor antagonist, 0.1 mg / 100 g i.p., Antisedan, Pfizer) to reverse the medetomidine anesthesia. Animals were monitored in a recovery room for the first 24 hours following surgery.
Tissue fixation. Rats were sacrificed (at 7 days, unless otherwise stated) with an overdose of pentobarbital (20 mg / 100 g i.p., Euthatal, Merial Animal Health, Essex, UK). After an intracardiac dose of heparin (100 IU) the animal was perfused transcardially with 0.9% NaCl (1 mL/g) followed by 4% formaldehyde (Sigma) in 0.1 M phosphate buffer (PB, pH 7.4, 1 mL/g). After perfusion the L4 and L5 dorsal root ganglia were identified, marked, and their roots used as a reference for segmental identification of the spinal cord. The brain and spinal cord were removed and postfixed for 2 hours before overnight cryoprotection in 30% sucrose in 0.1 M PB. To distinguish orientation the tissue was marked with a needle puncture site.

Identification of spinal injection sites
The spinal cord from rats at 1-14 day post-AVV injection was cut in either coronal, longitudinal, or sagittal planes. The injection tracks were typically visible allowing confirmation of the location of the injection site within the target area of the dorsal horn. Immunocytochemistry for hexon protein, a com-ponent of the adenoviral coat, determined the extent of spread of the injected AVV. Examination of spinal cord tissue at the earliest timepoint (24 hours after the injection) indicated that the spread of AVV was confined to the ipsilateral dorsal horn (see Fig. 1). The expression of hexon viral protein was seen in a reduced distribution at day 3, and by day 7 only a "halo" region around the injection site (within 100 m) was observed.

Immunohistochemistry
Tissue sections were cut at 40-m intervals using a freezing microtome and either serially mounted or left free-floating for fluorescence immunohistochemistry (IHC). Mounted sections were ringed with a DAKO pen (DAKO, Ely, UK) for on-slide incubations. Tissue sections were washed (؋3) in 0.1 M phosphate-buffered 0.9% NaCl (pH 7.4, PBS) and permeabilized in 50% ethanol for 30 minutes before further washing. The tissue was incubated with primary antibodies (see Table   Figure 1. Retrograde targeting of brainstem noradrenergic neurons using spinal injection of AVV. a: Schematic of experimental approach. AVV-PRS-EGFP is injected into the dorsal horn of the lumbar spinal cord. The AVV is taken up by axon terminals of the pontospinal noradrenergic neurons and retrogradely transported. EGFP is expressed under the control of the catecholaminergic selective promoter PRS to enable visualization of the neuronal anatomy. b: The location and depth of the injection site is indicated by hexon protein-immunoreactivity (Cy3 fluorescence) in the spinal cord (1 day after unilateral AVV-injection). c: Overlaid brightfield and fluorescence images shows the spread of the AAV to be limited to the dorsal horn. Inset shows Rexed's laminae, red line indicates the injection trajectory. Scale bar ‫؍‬ 100 m. 1) in PBS with 5% horse serum (HS), 0.3% Triton X-100, and 0.01% sodium azide for 24 -72 hours at room temperature. Mounted sections were kept in a humidified chamber (RA Lamb, Eastbourne, UK) and free-floating sections were continuously agitated. After further washing, sections were incubated overnight with biotinylated secondary antibody (all Jackson ImmunoResearch, West Grove, PA; see Table 1) diluted 1:1,000 in PBS with 2% HS and 0.3% Triton. Sections were washed, incubated with Streptavidin Cy3 (1:1,000 in PBS, Sigma) for 4 hours before a final wash. Additionally some sections were incubated with DAPI (4,6-diamidino-2phenylindole dilactate, 1 g/mL, Invitrogen, La Jolla, CA) for 15 minutes to delineate the nucleus. Controls were routinely run, by omitting primary antibodies, to ensure the specificity of immunostaining.
The mouse monoclonal anti-DBH antibody (Chemicon, Temecula, CA; MAB308) was raised against purified bovine DBH. The distribution of labeled neurons within the brainstem closely resembled that described for catecholaminergic neurons reported by previous investigators in the rat (Dahlstrom and Fuxe, 1964;Swanson, 1976;Grzanna and Molliver, 1980) and specifically in studies using this particular antibody (Rinaman, 2001; Espana and Berridge, 2006). In the absence of a commercial source of purified bovine DBH, antibody specificity was confirmed using a previously described preabsorption protocol (Dvoryanchikov et al., 2007). In brief, 1:10,000 dilutions of the anti-DBH antibody were preincubated in wells at room temperature either with or without a 40-m section of fixed rat adrenal medulla. The aliquots were then aspirated and used in parallel according to the previously described IHC protocol against serial sections of pontine tissue containing the LC. The fluorescence labeling of LC neurons was abolished by the preincubation of antibody with the adrenal medulla.
The anti-c-fos polyclonal rabbit IgG (Santa Cruz Biologicals, Santa Cruz, CA; sc-52) was raised against a N-terminus pep-tide from human c-fos (amino acids 3-16, sc-52P). This antibody recognized a 62-kDa protein, inducible by phorbol ester application, on Western blot corresponding to the expected molecular weight of c-fos (manufacturer's technical datasheet). The immunoreactivity for c-fos induced in response to formalin testing was eliminated in both the spinal cord and the locus coeruleus by overnight preabsorption of the anti-c-fos antiserum with 0.1 g/mL of the peptide antigen sc-52P.
The anti-GFP rabbit polyclonal IgG (Invitrogen, A11122) was raised against GFP protein extracted from Aequorea victoria and affinity column purified; its specificity has been confirmed in rat neurons transfected with GFP-expressing vectors (Card et al., 2006). In our control experiments no labeling was seen in brain tissue from animals that had not been transfected with AVV-PRS-EGFP.
The antihexon goat polyclonal IgG (Biodesign International, B65101G) was raised against hexon protein from adenovirus type 2 and also recognizes hexon from adenovirus types 5 and 6. It identifies a protein of 105 kDa on Western blot corresponding to the expected molecular mass of hexon protein (manufacturer's technical datasheet). Specificity of this antibody for our adenoviral vector was tested by preincubation of antihexon antibody (1:1,000) with Ad-PRS-EGFP (4 ؋ 10 9 TU/mL) overnight. Subsequent IHC using control and preabsorbed antihexon antibody (1:2,000) on serial sections of spinal cord injection sites (24 hours after injection of Ad-PRS-EGFP) showed that preabsorption eliminated the fluorescence labeling surrounding the injection site.

Data analysis and photomicrography
All sections were mounted and coverslipped with fluorescent mounting medium (DAKO). Tissue was examined and representative images obtained using a conventional fluorescence microscope (Zeiss Axioskop 2) or with a confocal microscope (Leica DMIRBE and TCSNT). Images were acquired and processed using the respective company software (to adjust contrast and brightness) and, if required, images were further processed using Adobe Photoshop CS2 (Adobe Systems, San Jose, CA) to annotate structures and add scale bars.
Retrogradely labeled pontospinal noradrenergic neurons were counted as being present within each section when the nucleus could be identified within the cell body and a primary dendrite was visible. Counts of DBH-positive cells were made from noncontiguous sections (1-in-4). Using DAPI staining (n ‫؍‬ 2 rats) we obtained an accurate estimate for the size of the nucleus in the retrogradely labeled NA neurons (9.6 ؎ 0.6 m, n ‫؍‬ 32) and used this average dimension to Abercrombie (1946) correct all NA neuron cell counts. Data reported as mean ؎ SEM. Statistical significance was assessed using unpaired t-test or one-way ANOVA with Bonferroni's post-hoc test (Prism4, GraphPad, San Diego, CA), differences were considered significant at P < 0.05.

Experimental protocols
Targeting noradrenergic neurons projecting to the lumbar dorsal horn. To test whether the AVV could retrogradely target the pontospinal noradrenergic neurons we injected AVV-PRS-EGFP (3 ؋ 10 10 TU/mL) bilaterally into the lumbar (L4 -5) dorsal horn of spinal cord (n ‫؍‬ 14). The spinal cord, brainstem, and forebrain were examined for the presence of EGFP-positive neurons (the morphology of the cell somata was examined in detail in nine of these animals). DBH IHC was used to examine whether the EGFP expression was restricted to catecholaminergic neurons (n ‫؍‬ 6). An indication of the relative contributions of each of the pontine NA nuclei to the pontospinal projection was obtained from counts of DBHpositive neurons in each region. These counts were determined directly for the A5 and A7 regions (n ‫؍‬ 3 rats) and we used previously obtained values for the LC (Loughlin et al., 1986). This allowed the calculation of the proportion of neurons in each region that projected to the lumbar spinal cord.
We compared AVV-PRS-EGFP with the conventional retrograde tracers FluoroGold (FG, Fluorochrome, Denver, CO), cholera toxin b subunit (CTb, List Biological Laboratories, Campbell, CA), and red fluorescent latex microspheres (Retrobeads, Lumafluor, Naples, FL). For these experiments AVV (3 ؋ 10 10 TU/mL) was injected (at two sites bilaterally) into the L4/L5 spinal cord (n ‫؍‬ 9 rats). To obtain an estimate of the size of the total population of NA neurons projecting to the lumbar spinal cord a pair of FG (5%, 50 nL each side, n ‫؍‬ 2 rats) or CTb (1%, 500 nL, n ‫؍‬ 4 rats) injections was made into the spinal dorsal horn, interleaved between the AVV injections. We also made use of fluorescent microspheres, which are known to have a more restricted distribution from the injections site to identify the NA neurons projecting specifically to the L4 -5 dorsal horn (100 nL, at two sites bilaterally) interleaved with the AVV injections in the rostrocaudal axis (n ‫؍‬ 3 rats). One further animal received co-injections (500 nL, two sites bilaterally) of a mixture of AAV-PRS-EGFP (3 ؋ 10 10 TU/mL) and red microspheres (1:2 dilution). For the CTb injections retrogradely labeled brainstem neurons were detected by IHC for CTb (1:2,000, 3 days at 4°C, goat anti-CTb, List Biological Laboratories) and revealed by incubation with fluorescent secondary antibody (antigoat Alexa 594, 1:200, overnight, Invitrogen). All brainstem tissue was processed for DBH IHC to determine the numbers of noradrenergic neurons labeled by FG/CTb/beads.

AVV injection into lumbar cerebrospinal fluid.
To test whether intraparenchymal delivery was required for retrograde labeling AVV-PRS-EGFP was injected intrathecally into the cerebrospinal fluid, over the lumbar dorsal horn (3 ؋ 10 10 TU/mL, 2 L, n ‫؍‬ 2).
Topographical organization of the pontospinal noradrenergic projection. Injections of AVV-PRS-EGFP or -mRFP were made into discrete locations within the spinal cord to examine the topography of the pontospinal NA projection.

Animals were culled 14 days after AVV injection. Retrogradely labeled brainstem neurons were counted and the proportion of double-labeled cells was expressed as a percentage of the total number of mRFP-positive neurons.
Lateralization of pontospinal noradrenergic projection. Rats (n ‫؍‬ 3) received four unilateral injections of AVV-PRS-EGFP (3 ؋ 10 10 TU/mL) into L4/L5 dorsal horn. The distribution of retrogradely labeled neurons (ipsi-vs. contralateral) was expressed as percentage of the total number of neurons in the A5, LC, and A7 groups.
Projections of pontospinal NA neurons to other brain regions. The extent of the axonal arbor of the pontospinal NA neurons that innervate dorsal horn of the L4/L5 spinal segments were revealed using anti-EGFP IHC. Rats (n ‫؍‬ 5) had bilateral spinal injections of AVV-PRS-EGFP (3 ؋ 10 10 TU/mL) and were culled 14 days later. Brain tissue was cut in the coronal plane, while spinal cord tissue was cut in coronal, longitudinal, and sagittal planes. To contrast the pattern of distribution of noradrenergic fibers with the EGFP containing projections from pontospinal neurons, alternate sections were processed for anti-GFP and anti-DBH IHC (in two animals). The density of the EGFP containing fiber projections revealed by anti-GFP IHC was scored for each animal on a rating scale from 0 to 5, with 0 representing a total absence of fibers to 5 being densest (e.g., principal nucleus of the olive).
To test whether the pontospinal NA neurons had multiple projection targets a further two animals received both lumbar spinal injections of AVV-PRS-EGFP and injections of red fluorescent beads (four unilateral injections of 100 nL) into sites within the ventral posterolateral thalamic nucleus (micropipette lowered vertically through a craniotomy burr hole to injection sites located between bregma ؊2.3 to ؊3.5 mm, ML 3-3.6 mm, and depth 6 -6.4 mm; coordinates after Paxinos and Watson, 2005). Animals were culled 14 days after AVV and bead injections. The bead injection sites were confirmed as being in the thalamic VPL and the numbers of retrogradely labeled neurons were counted in A5, LC, and A7 regions. The proportion of bead and EGFP double-labeled cells was expressed as percentage of the total number of EGFP-positive neurons.
Activation of pontospinal NA neurons during the formalin test. All animals were handled daily and habituated to the testing room, experimental equipment, and handlers. Rats (n ‫؍‬ 10) received bilateral injections of AVV-PRS-EGFP (3 ؋ 10 10 TU/mL) into the L4/L5 dorsal horn. One week later the animals were divided into two groups for the formalin test, a test group (n ‫؍‬ 8) and a control group (n ‫؍‬ 2). Rats were acclimatized for 30 minutes to a clear Plexiglas testing chamber before either formalin (5% neutral buffered) or NaCl (0.9%) was injected subcutaneously (50 L, 30G needle) to raise a bleb on the dorsal surface of the right hindpaw.
Rats were replaced in the testing chamber and the numbers of flinches and foot lifts were tallied over 1-minute periods, initially every 2 minutes for the first 10 minutes and then every 5 minutes for the remainder of the 60 minutes. Animals were culled 2 hours after the end of the observation period to allow optimal c-fos expression. Brainstem tissue was processed for c-fos and DBH IHC in all rats and counts were made of retrogradely labeled pontospinal neurons showing c-fos expression across each cell group. In addition, spinal c-fos expression was quantified (control [n ‫؍‬ 2] and formalin test rats [n ‫؍‬ 3]) by tallying the positive neurons from 10 coronal, nonsequential, spinal cord sections (40 m) from L4 -5 with the greatest numbers of c-fos positive nuclei. Comparisons were made between c-fos counts in control and test group animals and between ipsi-and contralateral dorsal horns.
The retrogradely labeled neurons showed DBH-ir (A5, 97 ؎ 3%; LC, 97 ؎ 4%; and A7, 100 ؎ 0%, n ‫؍‬ 6, see Fig. 3), indicating that the PRS promoter restricted the expression of EGFP to catecholaminergic neurons. The retrogradely labeled LC neurons that we were unable to unequivocally determine as being DBH-ir (1.6%, 10/622, n ‫؍‬ 6 rats) were found in the core of the nucleus and had morphologies consistent with being noradrenergic. However, because of the density of DBH staining seen in the closely packed cells in the core of the LC it was occasionally difficult to distinguish unequivocally be-tween low fluorescence signal in the labeled cell and that from adjacent somata. Given the location of these cells within the core of the LC, it would seem likely that they represent NA neurons with faint DBH-ir.
The A5 and A7 regions were identified using the Paxinos and Watson atlas (2005) 300 m radius, Fig. 1). It should also be noted that there was no pontine EGFP expression seen in the FG/ CTb experiments, indicating an apparent incompatibility in the use of these conventional tracers with our adenoviral vector (as suggested previously for FG: Schramm, 2006).
The spinally injected red latex microspheres showed a similar restricted distribution in the cord to the AVV-PRS-EGFP (n ‫؍‬ 3). Within the pontine NA nuclei the beads and AVV-PRS-EGFP labeled similar numbers of neurons (78 ؎ 25 vs. 110 ؎ 49 respectively, n ‫؍‬ 3, NS). Most of the bead-positive neurons in the A5, LC, and A7 areas were DBH-positive (82%). The distribution of bead-positive neurons across the NA cell groups was essentially identical to that seen with the viral vector with most located in the LC (79%) and the remainder in A5 (15%) and A7 (5%). The interleaved injections (n ‫؍‬ 3 rats) showed 18 ؎ 4% double labeling of EGFP neurons with beads. By contrast the co-injection of AVV with beads showed 60% colocalization (n ‫؍‬ 1 rat, note there was an apparent loss of efficacy of AVV labeling with this co-injection approach with only 10 neurons being EGFP-positive).

Factors influencing retrograde AVV transduction
Effect of AVV titer. The efficiency of retrograde labeling was dependent on the titer of AVV injected into the spinal cord. Serial dilution of the AVV injectate produced reductions in the numbers of retrogradely labeled neurons (see Fig. 4a) from 145 ؎ 12 (300 ؋ 10 8 TU/mL, control) to 49 ؎ 19 (at 100 ؋ 10 8 , P < 0.0001) and down to as few as 5 ؎ 3 (at 3 ؋ 10 8 TU/mL). Interestingly, at the highest AVV titers (1,000 ؋ 10 8 TU/mL) there was a dramatic fall in the numbers of retrogradely labeled neurons.
Influence of time postinjection. The first pontine retrograde labeling could be seen 24 -48 hours after spinal injection of AVV-PRS-EGFP (Fig. 4b). The numbers of labeled pontine neurons increased with longer expression periods up to day 7, when it peaked and no further increase in the numbers of EGFP-positive neurons was seen at day 14. The visible extent of the dendritic tree of labeled neurons appeared greater at day 14 compared to that seen at earlier timepoints, suggesting the presence of higher concentrations of EGFP in the processes.
AVV injection into lumbar cerebrospinal fluid. One week after intrathecal injection of AVV-PRS-EGFP (2 L, 3 ؋ 10 10 TU/mL, n ‫؍‬ 2 rats) no EGFP-positive neurons were seen in the pons or elsewhere in the brain or spinal cord, indicating that intraparenchymal spinal injection of this AVV is needed for retrograde labeling and importantly that the pontine labeling is not a consequence of rostral spread of vector carried in the CSF.

Topographical organization of the pontospinal noradrenergic projection
Unilateral spinal injection of AVV demonstrated a bilateral projection from the LC with an ipsilateral predominance (62 ؎ 4% of the labeled pontospinal NA neurons being ipsilateral, n ‫؍‬ 3 rats; P < 0.05, unpaired t-test). Some of the pontospinal neurons (4%) were seen to project bilaterally to the lumbar dorsal horn using ipsi-and contralateral injections of AVV expressing EGFP or mRFP (see Fig. 5

Axonal projections of pontospinal NA neurons
Using GFP IHC it was possible to amplify the EGFP fluorescence signal and reveal fine, distal processes (including axon terminal fields) of the retrogradely labeled pontospinal NA neurons throughout the neuroaxis (Figs. 6, 7; Table 2 for more details). DBH IHC allowed comparisons to be made with the overall distribution of NA projections in the brain and spinal cord and confirmed the GFP containing fibers to be DBHpositive.

could be observed. In transverse sections of L4/L5, there was dense GFP-ir in the superficial dorsal horn (SDH, laminae I and II), moderate density of axonal projections in the deep dorsal horn (DDH, laminae III-V) and the ventral horn (laminae VII-IX). Moderate to dense labeling was observed around the central canal (area X). The densest GFP-ir labeling in the white matter of the spinal cord was found in the ventral funiculus, with light to moderate labeling in the lateral funiculus and gracile fasciculus. This pattern was confirmed in the longitudinal and sagittal sections, where long segments of GFP-ir axons were
observed running close to ventral, lateral, and dorsal surfaces, with the majority observed in the ventral funiculus. Occasionally, these axons could be seen to send collaterals at right angles to project into the gray matter (see Fig. 7b).
Brain. In the telencephalon, light to moderate GFP-ir was observed in the neocortex (particularly in the cingulate, frontal, insula, and piriform areas) and globus pallidus. By contrast, relatively few axonal projections were observed in the hippocampus, caudate putamen, nucleus accumbens, and entorhinal cortex despite strong DBH-ir labeling. In the diencephalon the highest density of GFP-ir was observed in thalamus. However, in the hypothalamus only light to moderate GFP-ir was observed, in contrast with the strong DBH-ir.
In the cerebellum there was a moderate density of GFP-ir labeling compared to the denser DBH labeling. The brainstem showed moderate to dense GFP-ir staining, especially in the inferior olive, pontine nuclei, periaqueductal gray, paramedian raphe, ventral tegmental area, deep mesencephalic, and anterior pretectal nucleus. The GFP-ir in the inferior olive was strongest in the principal nucleus.
To test whether these divergent projections originated from the pontospinal NA neurons we combined spinal AVV-PRS-EGFP with injections of red fluorescent beads into the VPL nucleus of the thalamus, a territory that we had noted to have a dense projection of EGFP-containing fibers. These thalamic injections retrogradely labeled neurons in the core of the LC and double labeling with beads was seen in 26 ؎ 3% of the EGFP-positive pontospinal NA neurons (n ‫؍‬ 2, Fig. 8). No retrograde bead labeling at all was seen in the A7 region and few neurons were seen in the A5 region with no double labeling of pontospinal neurons.

DISCUSSION
In this study we demonstrated the ability of an AVV containing a catecholaminergic cell-specific promoter to selectively retrogradely target pontospinal noradrenergic neurons projecting to the lumbar dorsal horn. This retrograde targeting from the lumbar L4 -5 segment identifies a relatively small pool (around 4%) of the total number of pontine noradrenergic neurons in the LC and A7 areas (and <2% of A5). Using this vector to express fluorescent markers such as EGFP and immunohistochemistry we have obtained a Golgi-like visualization of the anatomy and projections of the pontospinal noradrenergic neurons. Furthermore, we have been able to show that these pontospinal noradrenergic neurons are activated (express c-fos) during a noxious stimulus applied to the hindpaw, suggesting that these are nociresponsive and are therefore likely to constitute a component of the descending antinociceptive control system (Jones, 1991;Millan, 1997Millan, , 2002Pertovaara, 2006).

Organization of the pontospinal noradrenergic system
In agreement with a number of previous studies, we have shown that the dense noradrenergic innervation of the dorsal  horn originates from neurons whose cell bodies are located in the pontine A5, LC, and A7 regions (Westlund et al., 1983;Fritschy and Grzanna, 1990;Proudfit, 1991a,b, 1993). In total, our viral vector identifies around 150 pontine neurons with terminal projections to the L4 -5 dorsal horn. Previous ablation studies using intrathecal administration of anti-DBH-saporin showed loss of the majority of neurons in LC (Jasmin et al., 2003a). However, those authors concluded that their toxin had spread in the CSF to reach higher centers such as the cortex, thus lesioning a large proportion of the NA neurons. The AVV-PRS-EGFP by contrast requires intraparenchymal administration to target the pontospinal neurons so there is no spread in the CSF to higher centers.
The majority of our retrogradely labeled neurons were located in LC (80%), with the remainder in A5 (12%) and A7 (8%). This suggests that the predominant source of the neurons projecting to the dorsal horn is the LC. However, when considered as a proportion of the total number of noradrenergic neurons in each nucleus, the distribution becomes more even across the pontine nuclei LC (4.4%) and A7 (4%) and with relatively fewer in A5 (1.4%). These proportions suggest that the LC and A7 nuclei may have topographical organizations with similar weightings for the hindlimb spinal segments.
Our adenoviral vector strategy identified comparatively fewer pontospinal noradrenergic neurons than chemical retrograde tracers (FG/CTb injected into the same region of the spinal cord) with Ϸ50 -70% the number of LC neurons (115 ؎ 10 [AAV] vs. 244 ؎ 8 [FG] or 161 ؎ 51 [CTb]) and relatively smaller proportions of neurons in the A5 and A7 areas. It should be noted, however, that both FG and CTb spread considerably further within the cord parenchyma than the AVV to extend several segments rostrocaudally and throughout the full dorsoventral extent of the cord, thus they are likely to have labeled the pool of NA neurons projecting to the lumbar enlargement. In particular, the intermediolateral cell column, which terminates at L2, would have been included in the field of the conventional tracer injection but not in our AVV injections. This may account for the differences in the relative distribution of retrogradely labeled neurons across the NA nuclei, particularly in the case of the A5 region (and in the rostral ventrolateral medulla,) which have prominent projections to the sympathetic preganglionic neurons in the intermediolateral cell column (Loewy et al., 1979). In addition, it seems likely that our viral vector is only taken up by axon terminal fields (see below and Ridoux et al., 1994;Liu et al., 1997b), unlike FG or CTb, which are known to be taken up by axons of passage and therefore are likely to label noradrenergic neurons with projections traveling to more caudal spinal targets. By contrast, retrograde labeling with fluorescent latex microspheres (whose distribution was more limited in the spinal cord) identified comparable numbers of neurons as the AVV in the pontine A5, LC, and A7 areas with a similar proportional distribution across these nuclei. Hence, it would appear that dorsal horn injections of our AVV produces focal transduction of the NA projection neurons that is of comparable efficacy to conventional tracers and, based on these findings, we have no reason to believe that our AVV is selectively targeting a specific subgroup of the pontospinal neu- rons. Thus, in agreement with previous studies (Guyenet, 1980;Loughlin et al., 1986), we find that these pontospinal projections originate from a limited subset of the total population of NA neurons.
The topography of the pontospinal projection was characterized using AVVs expressing either EGFP or mRFP. This showed that the noradrenergic projection to the cord from the pons is bilateral, with an ipsilateral predominance (62%) in agreement with previous functional studies showing bilateral analgesic effects from unilateral stimulation of LC (Jones and Gebhart, 1986b). A proportion of LC neurons were found to have bilateral projection fields (at least 4% of the pontospinal neurons). We found relatively little overlap in the neurons projecting to the cervical and lumbar spinal dorsal horn (L4 -5 and C5-6, 1% colocalization). By contrast, co-injection of the vectors (or fluorescent microspheres) in the same sites in the lumbar spine produced Ϸ60% colocalization. This indicates that there is a rostrocaudal topographical organization of the pontospinal noradrenergic projection, with little overlap in the populations of neurons innervating spinal territories receiving sensory inputs from the fore-and hindlimbs. In addition, the fact that we observed little double labeling when making combined AVV injections to lumbar and cervical cord suggests that the AVV is only retrogradely transported from the terminal axonal field rather than by axons of passage (unlike conventional retrograde tracers). This makes the AVV a particularly selective tool for retrograde targeting.
We noted that the EGFP expression allowed the axons of the NA neurons to be identified and this could be enhanced by IHC to give a Golgi-like visualization of distal projections (note there was no increase in the numbers of GFP-positive labeled noradrenergic cell bodies after this immuno-enhancement). A similar ability to define the axonal processes of medullary NA neurons of the (C1 group) has been reported using direct injection of a lentiviral vector employing the PRS promoter to drive EGFP expression (Card et al., 2006). We found that the pontospinal neurons showed the expected pattern of axon terminal fields in the spinal cord. When comparing the density of the EGFP-containing fibers with those revealed by DBH IHC it was apparent that even in territories with relatively dense EGFP-containing projections (such as the superficial dorsal horn) there was a greater density of DBH-containing fibers. Given the distance of these axonal terminals from the NA somata, where the EGFP synthesis occurs, this disparity is likely to be a consequence of a concentration threshold effect for the visualization of EGFP-containing axons by immunocytochemistry.
Interestingly, axonal projections from the pontospinal NA neurons were also seen in other regions of the neuroaxis, such as the inferior olive, periaqueductal gray, cerebellum, thalamus, and some regions of the cortex (insular, cingulate, and piriform). Using the injection of fluorescent microspheres to the thalamic VPL we were able to retrogradely label neurons in the LC (as has been previously shown; Simpson et al., 1997;Voisin et al., 2005) and the presence of double labeling confirmed that some pontospinal LC neurons also project to the thalamus. We found no evidence of double labeling of pontospinal A5 or A7 neurons from the VPL. The distribution pattern of EGFP-containing fibers from the pontospinal neurons was specific, as there was no projection seen to some regions that have a dense noradrenergic innervation (DBH IHC), such as the hippocampus, caudate putamen, nucleus accumbens, and entorhinal cortex. This restricted distribution is in agreement with previous findings showing a topographical organization of the neurons in LC (Loughlin et al., 1986). The distributed nature of the NA system and the lack of conventional synaptic terminals (Descarries et al., 1977) have led to the inference that these neurons exert a diffuse global effect whose synchronized activation is important for generalized phenomena such as arousal (see Dismukes, 1977). However, the anatomical organization of the NA neurons projecting to the spinal cord suggests that they exert influence over a limited range of neural territories that share common functional specificity, e.g., pain matrix (insular and cingulate cortices, thalamus, periaqueductal gray, parabrachial nucleus, and spinal dorsal horn) or motor control (cerebellum, thalamus, and inferior olive).

Methodological considerations
In agreement with previous studies, we have shown AVV to be capable of retrograde transduction of neurons (Akli et al., 1993;Ridoux et al., 1994;Liu et al., 1997b;Lonergan et al., 2005;Tomioka and Rockland, 2006). Following spinal injection of the AVV, EGFP-positive pontine neurons were first seen after 2-3 days and labeling was maximal at 7 days. Efficient retrograde labeling required higher titers of spinal AVV (>10 10 TU/mL). However, above this level a significant fall-off was seen in the numbers of transfected pontine neurons, perhaps because local spinal reactions around the injection site damaged axon terminals before transduction could be initiated. Such an inflammatory effect has been previously reported following spinal injections of high concentrations of AVV (Liu et al., 1997b). The cell-specific promoter PRS successfully restricted expression of EGFP to pontine noradrenergic neurons as shown by DBH IHC. Previous spinal administration of AVV with the cytomegalovirus promoter has shown labeling of brainstem neurons in the red nucleus, vestibular nuclei, within the pontine reticular formation, and the locus coeruleus (Liu et al., 1997b). By contrast, we only saw expression in the A5, LC, and A7 cell groups with the PRS promoter. There is evidence that the PRS promoter is active in noncatecholaminergic neurons that express the Phox2 transcription factor, such as cholinergic autonomic neurons in the brainstem in areas like the dorsal vagal motor nucleus (Tiveron et al., 1996;Brunet and Pattyn, 2002;Lonergan et al., 2005). Following spinal administration of AVV-PRS-EGFP, we saw no expression in these brainstem cholinergic groups, as they do not project to the spinal cord. Furthermore, the spinal motoneurons and sympathetic preganglionic neurons, which are cholinergic, do not express the Phox2 transcription factor (Tiveron et al., 1996) and we saw no expression of EGFP in these cell groups following viral injection.
It is worth noting that following spinal injection of AVV the rats showed a prompt recovery of motor and sensory func-tion, such that their rotarod test was normal, as were their responses to thermal and mechanical sensory stimuli applied to the hindpaw (unpublished observation). Furthermore, rats transfected with AVV-PRS-EGFP showed typical biphasic nociceptive behavioral responses to formalin testing. Thus, our spinal dorsal horn injections of AVV were not associated with discernable gross motor or sensory deficits. Also, within the brainstem the retrograde transduction with AVV of the NA neurons appeared to be well tolerated. These neurons are capable of producing increased levels of c-fos following noxious stimulation. We saw no change in the numbers of labeled pontine NA neurons at timepoints of up to 1 month, indicating that there was no cell death as a consequence of AVV transduction and that the gene product continues to be transcribed for this time period. This is in contrast to the experience with high titers of AVVs when used by direct injection into the target cell groups where gene expression appears short-lived, in part because of local immune activation (Thomas et al., 2001;Kugler et al., 2003). We have also made electrophysiological recordings from transfected LC cells in both acute slices and also in slice cultures of the pons (unpublished observation) and the neurons appear healthy with firing patterns and intrinsic conductances similar to those previously reported (Williams and Marshall, 1987). The use of replication deficient viral vectors offers a number of useful features compared to conventional neuroanatomical tracing techniques. Through the use of cell-specific promoters it is possible to express the gene of interest selectively in a particular phenotype of cell, e.g., neuronal or glial (Lonergan et al., 2005;Card et al., 2006). Furthermore, there are a number of different fluorophores (e.g., EGFP or mRFP) or histochemical markers (i.e., beta-galactosidase) that can be expressed and used to label different cell populations using the same type of viral vector and promoter system. They also have the potential to allow anatomically targeted functional manipulations through the expression of genes that alter neuronal function (Johns et al., 1999). In the exploration of the noradrenergic system, this could simply be an extension of the lesion experiments, which have used either metabolic toxins such as DSP4 (Zhong et al., 1985)

Functional activation
To examine the role of the pontospinal NA neurons in nociception, we employed the formalin test (Dubuisson and Dennis, 1977) as a model of a persistent acute pain, as it has previously been shown that the endogenous NA analgesic system is actively involved in attenuating the response to this stimulus (Sugimoto et al., 1986;Liu et al., 1997a;Omote et al., 1998). It has also been shown that LC NA neurons show increased expression of c-fos in response to noxious stimulation such as hindpaw injections of formalin (Baulmann et al.,