NR1 is an essential subunit of the N-methyl-d-aspartate (NMDA) receptor, which at the spinal level is involved in injury-induced pain hypersensitivity and morphine tolerance. An in vitro luciferase assay was used to identify candidate and control (inactive) short interfering RNA (siRNA) sequences that are expressed by a recombinant adeno-associated virus (rAAV) plasmid. rAAV vectors targeting the NR1 subunit were prepared that express active or control (mismatch) siRNA sequences and injected into the mouse spinal cord dorsal horn (SCDH). Three weeks after vector administration, green fluorescent protein labeling of the ipsilateral SCDH confirmed the spatial localization of the viral transduction. Active siRNAs resulted in a 60 to 75% knockdown of NR1 mRNA and protein in the area of the virus injection. The spatial knockdown persisted for at least 6 months after a single administration of the vector. Neither the active nor the mismatch siRNAs resulted in cellular toxicity as measured by nuclear staining and cell integrity. The vector-derived knockdown of NR1 expression in SCDH did not alter acute thermal or mechanical stimulus paw-withdrawal thresholds. However, the vector-derived siRNA prevented the mechanical allodynia measured at 24 and 48 h after injection into the paw of the inflammatory agent, Complete Freund's adjuvant. These results demonstrate that vector-derived siRNAs can be used to produce an in vivo spatial knockdown of the expression and function of the NMDA receptor that is confined to the ipsilateral SCDH. Vector-derived siRNAs may have therapeutic potential for the management of injury-induced pain resulting from the activation of NMDA receptors in the SCDH.
RNA interference (RNAi) is an evolutionarily conserved mechanism for silencing gene expression by targeted degradation of mRNA. Because it allows a sequence-specific inhibition of gene expression, it is being widely applied in biology, including in vivo applications (Fire et al., 1998; Dorsett and Tuschl, 2004; Dykxhoorn et al., 2006).
RNAi is often more potent and longer lasting than other nucleic acid-based approaches and, compared to gene targeting by homologous recombination, is less expensive and timeconsuming and not limited to a particular mammalian species (i.e., the mouse) (Dorsett and Tuschl, 2004; Dykxhoorn et al., 2006). A key feature of this approach is that small double-stranded RNAs (small interfering RNAs, siRNAs) are generated by the action of an RNase-III-like enzyme called Dicer from a larger double-stranded RNA precursor or from short hairpin (sh) RNAs, which can be expressed from plasmids or viral vectors (Lieberman et al., 2003; Dorsett and Tuschl, 2004). mRNA degradation occurs when the antisense strand of the siRNA directs the RNA-induced silencing complex that contains the RNA endonuclease Ago2 to cleave the complementary sequence of the target mRNA (Dykxhoorn et al., 2006). Silencing by RNAi is typically incomplete—a knockdown rather than a knockout. Similar to other nucleic acid-based methods, any gene is a potential target for silencing by RNAi; therefore, the possibilities are virtually endless. However, major challenges to the in vivo application of RNAi include the efficient identification of precursor shRNAs that will be processed in vivo to siRNAs and target site delivery of the precursor shRNA. In this report, we describe an efficient in vitro method for identifying shRNAs that will be processed into active siRNAs and a method for precise and reproducible spatial delivery of these shRNAs to the target neurons. Our approach to this design and evaluation will be illustrated using siRNAs that target the NR1 subunit of the NMDA receptor in the SCDH. The NR1 subunit is required for this ionotropic glutamate receptor to function as ion channel throughout the CNS, including the SCDH (South et al., 2003). The contribution of the NMDA receptor to injury-induced pain hypersensitivity has been well established by pharmacological antagonism (Woolf and Thompson, 1991; Davis and Inturrisi, 2001), antisense oligonucleotides (Shimoyama et al., 2005), and a conditional knockout (South et al., 2003). Because the behavioral consequences of blocking, reducing, or deleting the NMDA receptor are well known, we focused on the histochemical effects of the delivery of siRNAs targeting the NR1 subunit in SCDH. We describe an approach to design, screen in vitro, and deliver in vivo vector-based siRNAs that are effective in significantly reducing NR1 gene expression.
Materials and Methods
All experimental procedures involving animals were approved by the Institutional Animal Care and Use Committee (protocol 0508-392A) at Weill Medical College of Cornell University.
Design of siRNAs. The mouse NMDA receptor 1 subunit (Grin1) sequence was obtained from GenBank (accession number NM_008169; 3031 base pairs). Candidate siRNA sequences were designed using the siRNA design (http://jura.wi.mit.edu/siRNAext/) from the Whitehead Institute for Biomedical Research (Massachusetts Institute of Technology, Cambridge, MA) (Yuan et al., 2004). This computational tool implements several algorithms to identify siRNAs with a high probability of silencing the target gene. Several filters can be used to refine the user's oligonucleotide sequence characteristics. The program presents information about properties of the 21-mers, including the thermodynamic stability of the double-stranded RNA duplex, guanine cytosine base content, and other features that may contribute to the effectiveness of an siRNA (Mittal, 2004). Similarity searching of candidate siRNA against the mRNA sequences of both the mouse and rat was conducted using the Blast program (National Center for Biotechnology Information). The corresponding DNA sequence of each siRNA was converted into a short hairpin by the addition of a loop sequence (TTCAAGAGA) (Brummelkamp et al., 2002) and end nucleotide overhangs that are compatible with the restriction enzymes BglII and XbaI. For future utility, we selected candidate nucleotide sequences that were also 100% homologous with rat NR1 subunit. The complementary oligodeoxynucleotide strands were purchased from Sigma-Genosys (The Woodlands, TX) and annealed to produce the desired DNA sequence. Each double-stranded DNA sequence was then inserted downstream of the H1 promoter in a plasmid containing the sequence information required to produce a recombinant adeno-associated virus (rAAV)-2 vector (provided by M. Kaplitt, Neurosurgery, Weill Medical College, Cornell University). This rAAV plasmid also contains the coding sequence for green fluorescent protein (GFP) under the direction of a chicken β-actin promoter (Musatov et al., 2006). The sequence of the inserts for each plasmid was confirmed.
Screening for Effective siRNAs. The psiCHECK Dual Luciferase Reporter Assay (Promega, Madison, WI) was used to screen each plasmid. The psiCHECK-2 vector contains the two reporter genes, Renilla luciferase and firefly luciferase. To create a target construct, the cDNA for NR1 was inserted into the multiple cloning region located 3′ to the synthetic Renilla gene translational stop codon between the AsisI and NotI restriction sites creating a fusion transcript of NR1 and Renilla sequences.
HEK 293 cells were cotransfected at 1:1 M ratio with the psiCHECK plasmid containing the target NR1 cDNA and the rAAV plasmid containing a sequence that allows expression of a candidate shRNA. Twenty-four hours after transfection, the cell lysate was collected for dual luciferase reporter assay that measures Renilla and firefly luciferase activity using a luminescence microplate reader (Clarity Luminescence Microplate Reader; Bio-Tek Instruments Inc.) This assay is used to screen the in vitro cleaving activity of five candidate siRNAs (labeled 1, 2, 4, 5, and 6), as well as any nonselective activity of a scrambled sequence MM-6 (a mismatch control) and the rAAV plasmid containing no added DNA sequence (control; C). MM-6 is a randomly scrambled sequence that contained the same nucleotides as found in siRNA 6. The ratio of the Renilla to firefly luciferase activity was used to determine whether or not the rAAV plasmid generated an shRNA that was processed to an active siRNA capable of cleaving the NR1-Renilla fusion mRNA.
Packaging of Active siRNAs into Viral Vectors. The rAAV viral vectors were prepared by cotransfecting HEK 293 cells with an rAAV plasmid containing a DNA sequence coding for an active or mismatch shRNA sequence and a helper plasmid encoding adenovirus genes E2A, E4, and VA RNA, as well as AAV genes for Rep and Cap (Musatov et al., 2002). Cells were collected 72 h after transfection and lysed, and virus was purified using a HiTrap heparin column (17-0407-01; GE Healthcare, Piscataway, NJ) as reported previously (Mastakov et al., 2002; Musatov et al., 2002). The genomic titer was determined by real-time polymerase chain reaction using an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA) and SYBR Green Master Mix (Applied Biosystems) and ranged from 1 to 3 × 1011 genomic copies/ml. Forward primer (5′-AAGCAGAAGAACGGCATCAAG-3′) and reverse primer (5′-CGGACTGGGTGCTCAGGTAG-3′) were designed to flank a 151-base pair region on the GFP gene.
In Vivo Delivery. The viral vectors (rAAV.shRNA NR1) were injected into the SCDH as described by South et al. (2003). The mice were anesthetized with a ketamine-xyaline mixture, and the dorsal lumbar L2 and part of L3 spinous processes were removed by laminectomy. The virus was injected intraparenchymally (IPI) into the SCDH, targeting the lumbar enlargement. Three unilateral (right side) injections of 1 μl each were administered 0.6 mm apart at a depth of 0.3 mm from the dorsal border and 0.5 mm from the midline using a glass pipette with a 40-μm-diameter tip attached to a 5-μl Hamilton syringe. The syringe was mounted on a microinjector (David Kopf Instruments, Tujunga, CA) attached to a stereotaxic unit. After injection, the dorsal musculature was sutured, and the skin was closed.
Post-Injection Evaluation of GFP and NR1 Expression. At least 3 weeks and up to 6 months after administration, the mice were anesthetized with pentobarbital and then perfused transcardially with saline followed by 4% paraformaldehyde in phosphate-buffered saline. The spinal cord was dissected and placed in 4% paraformaldehyde for 1 h before being transferred to 30% sucrose for cryoprotection. After 72 h in sucrose, the lumbar spinal cord was cut into four 3-mm sections and frozen into a mold with optimal cutting temperature compound. Twenty-micrometer cryostat (Leica, Wetzlar, Germany) sections were prepared for immunohistochemistry and in situ hybridization.
Immunohistochemistry. The spinal cord sections were incubated with rabbit polyclonal antibodies to NR1 (1:1000, 07-362; Upstate Biotechnology, Charlottesville, VA) or GFP (1:1000, A-6455; Invitrogen, Carlsbad, CA). The sections were then incubated in biotinylated goat anti-rabbit IgG (1:250, BA-1000; Vector Laboratories, Burlingame, CA), and immunoreactivity was detected with the avidin-biotin-peroxidase complex 3,3-diaminobenzidinetetra-hydrochloride technique (Hsu et al., 1981).
For double fluorescent immunohistochemistry, spinal cord sections were first incubated overnight with antibodies to NR1 (1:1000, 07-362; Upstate Biotechnology). Sections were then incubated with biotinylated goat anti-rabbit IgG (1:250, BA-1000; Vector Laboratories). The immunolabeling was detected by the ABC technique (Hsu et al., 1981) and Cy3-tyramide as chromogen (NEL744001KT; PerkinElmer Life and Analytical Sciences, Wellesley, MA). The sections were then incubated overnight with antibodies to GFP (1:1000, A-6455; Invitrogen). Alexa Fluor 488 goat anti-rabbit IgG (1:250, A-11034; Invitrogen) was then applied as the secondary antibody. Finally, sections were mounted in the anti-fading mounting medium GelMount (T3605; Invitrogen) containing 0.1% TO-PRO-3 iodide (642/661) (Invitrogen). We were unsuccessful in attempts to colocalize NR1 and GFP using a nonrabbit-derived GFP antisera due to the weak labeling of GFP. Therefore, we evaluated a series of dilutions of the NR1 antisera and found that the conditions described above allowed us to colocalize the rabbit-derived NR1 and GFP without interference from the GFP fluorescent labeling (see Supplemental Fig. 1). The approach has been used successfully with other antisera (Dai et al., 2002). To test the specificity of the immunolabeling, control slides were exposed to diluted normal goat serum instead of the primary antibody.
NR1 in Situ Hybridization. Digoxigenin-labeled antisense or sense riboprobes were synthesized by in vitro transcription from a 2.2-kb region between exons 11 and 22 on the NR1 cDNA (11 175 025 910; Roche Diagnostics, Indianapolis, IN). The hybridization procedure was performed as described earlier (Simmons et al., 1989). The digoxigenin nucleic acid detection kit (11 175 041 910; Roche Diagnostics) was adapted for nonradioactive detection. In brief, after posthybridization washes, the slides were rinsed in washing buffer [0.1 M maleic acid, 0.15 M NaCl, and 0.3% (v/v) Tween 20, pH 7.5], incubated for 30 min in blocking solution (11 175 041 910; Roche Diagnostics) and 2 h in antibody solution (11 175 041 910; Roche Diagnostics). The slides were then washed twice in buffer for 15 min, equilibrated in detection buffer (0.1 M Tris-HCl, and 0.1 M NaCl, pH 9.5) for 2 to 5 min before incubating with the color substrate (11 175 041 910; Roche Diagnostics) overnight in a humidified chamber. After the reaction was complete, the slides were rinsed in water and dehydrated in a graded ethanol series, xylenes, and mounted in Permount mounting medium (Thermo Fisher Scientific, Inc., Waltham, MA).
Behavioral Evaluation. These studies were conducted in groups of 10 mice per treatment by an individual who was blinded as to the treatments. The behavioral studies were repeated once.
Motor Function Tests. To assess the effects of viral vector administration on motor function, reflexes for surface righting, placing/stepping, and grasping/climbing were evaluated as described by Kest et al. (1997). In the surface righting test, the mouse was placed horizontally on its back. An untreated mouse regains its normal upright position immediately. A latency of greater than 1.5 s indicates impairment. The placing/stepping test was performed by drawing the dorsum of either hind paw over the edge of a table top. An untreated mouse lifts the paw and “steps” onto the table top. The absence of such a behavior indicates impairment. Grasping/climbing ability on a wire grid inclined at 90° was determined. An untreated mouse can remain on the grid for at least 30 s.
Mechanical Stimulus Threshold. The threshold to a non-noxious mechanical stimulus was assessed using a set of von Frey filaments. The animal was placed in a Plexiglas cage with mesh flooring suspended above the researcher and left to acclimate for 15 min. The 50% g withdrawal threshold was determined using the up-down method of Dixon (Chaplan et al., 1994). Starting with a von Frey filament of 1 g, filaments were applied perpendicularly against the midplantar surface of the hind paw in a sequential ascending or descending order until the threshold for each mouse was determined. The threshold value is calculated from six determinations (Chaplan et al., 1994).
Thermal (Heat) Paw-Withdrawal Threshold. Thermal paw withdrawal was assessed using a thermal nociceptive stimulus (Hargreaves et al., 1988). Animals were placed in a Plexiglas cage and placed on a preheated glass plate maintained at 30°C and left to acclimate for 30 min. A radiant thermal stimulus was focused on the midplantar surface of the hind paw, and the latency in seconds for the withdrawal of the paw from the heat source was determined automatically. Each paw was evaluated separately, and a maximal cutoff of 20 s was employed. The threshold value is the average of three determinations.
CFA-Induced Mechanical Allodynia. Fifteen microliters of Complete Freund's adjuvant (CFA, 1 mg/ml heat-killed Mycobacterium tuberculosis in 85% paraffin oil and 15% mannide monoleate; Sigma-Aldrich, St. Louis, MO) was injected into the right hind paw of a lightly restrained mouse. The mice had received either the mismatch vector (MM-6) or siRNA vector (siRNA 6) in the ipsilateral SCDH 3 weeks before the intraplantar injection of CFA. Baseline mechanical stimulus paw threshold was measured as described above, and paw size was determined using a caliper just before the intraplantar injection of CFA (baseline) and at 24 and 48 h after CFA.
Quantification and Statistics. The reduction in NR1 immunolabeling and mRNA was estimated using the Metamorph software program (Universal Imaging, Downingtown, PA) as described previously (South et al., 2003). In brief, 24-bit color images of spinal sections were collected by a Nikon Eclipse 80i microscope (Nikon, Melville, NY) at a magnification of 32×. The integrated gray level intensity per unit area was measured and compared for the ipsilateral and contralateral SCDH. The histochemical and behavioral data were analyzed by an analysis of variance followed by the Student-Newman-Keuls test (multiple groups) or the t test (two groups) using the InStat program (version 3.00; GraphPad Software Inc., San Diego, CA).
PsiCHECK Luciferase Assay. The psiCHECK luciferase assay was used as an in vitro method for screening each candidate siRNA sequence. As shown in Fig. 1, each of the five rAAV plasmids expressing candidate shRNAs resulted in a significantly (p < 0.001) reduced ratio of Renilla/firefly luciferase compared with the control plasmid normalized to a ratio of 100%. The reduction in the ratio varied from 18 to 65% (Fig. 1). The three most effective siRNAs (4, 5, and 6) decreased the ratio by 60, 58, and 65%, respectively. In contrast, MM-6 did not decrease the ratio.
The four sequences selected for in vivo studies had the following sequences where the antisense strand is underlined and the stem loop sequence is in italics: siRNA 4, 5′-GTACCCATGTCATCCCAAA TTCAAGAGA TTTGGGATGACATGGGTAC-3′; siRNA 5, 5′-GGAACGGAATGATGGGAGA TTCAAGAGA TCTCCCATCATTCCGTTCC-3′; siRNA 6, 5′-GAATGTCCATCTACTCTGA TTCAAGAGA TCAGAGTAGATGGACATTC-3′; and MM-6, 5′-GTATGACCATGTACTCTCA TTCAAGAGA TGAGAGTACATGGTCATAC-3′.
Viral Transduction in the SCDH Measured by GFP Labeling. The ability of the rAAV.shRNA NR1 vector to deliver each of the four siRNAs in vivo was evaluated following unilateral IPI into the SCDH of mice. Four mice received vector 4, two vector 5, 16 vector 6, and eight vector MM-6. Approximately 3 weeks after IPI of vector 4, 6, or MM-6, a highly localized pattern of GFP expression was observed that was spatially restricted to the ipsilateral dorsal horn (Fig. 2, A, D, and G), with rare extension into the ipsilateral ventral horn (Fig. 2D). Both punctate and diffuse patterns of labeling were seen with identifiable cell body profiles. However, definitive cell types could not be determined. The three concurrent IPI of virus that were separated by 0.6 mm resulted in a GFP-labeling pattern that extended rostrocaudally for approximately 3 mm (2.9 ± 0.6 mm). This distance encompasses the L4, L5, and L6 spinal segments.
NR1 mRNA and Protein Expression. In adjacent sections, NR1 mRNA labeling by in situ hybridization and NR1 protein labeling by immunohistochemistry revealed an almost complete depletion of NR1 mRNA and protein in the ipsilateral dorsal horn as a consequence of the administration of vectors 4 and 6 identified as active siRNA sequences by the luciferase assay (Fig. 2, B and C, E and F). Vector 5 also resulted in an siRNA with in vivo activity using these labeling criteria (data not shown). In contrast, vector MM-6 did not affect NR1 mRNA or protein labeling (Fig. 2, H and I). With all active siRNAs, the contralateral dorsal horn, which was devoid of GFP labeling, did not show any alteration in NR1 mRNA or protein (Fig. 2, B–I).
To examine the colocalization of GFP and NR1, confocal double immunofluorescence was performed. On the contralateral side of the SCDH, labeling of NR1 was evident without the presence of GFP labeling (see Supplemental Fig. 1, A, B, D, and E). On the ipsilateral side, GFP labeling is seen in both the control vector MM-6 and vector 6-transduced cells (Fig. 3, A and F). In contrast, a substantial reduction in NR1 immunolabeling is seen on the ipsilateral side of vector 6-transduced cells (Fig. 3G), whereas cells transduced with vector MM-6 show normal NR1 expression (Fig. 3B). As a result, the merged images show the colocalization of GFP and NR1 in vector MM-6-transduced cells (Fig. 3, C, arrowheads, and D, Enlarged). Very little colocalization is seen in vector 6-transduced cells (Fig. 3, H and I, Enlarged). The rAAV vector is neurotropic (Kaspar et al., 2002; South et al., 2003), and these results are consistent with prior observations that the vector-derived expression of GFP, like NR1, is confined to neurons. No cell loss was seen on the viral injected side in either vector 6 or MM-6 groups compared with the contralateral side as indicated by staining with a nuclear dye, TO-PRO-3 (Fig. 3, E and J; Supplemental Fig. 1, C and F). The patterns of GFP expression and NR1 mRNA and protein knockdown seen at 3 weeks after IPI were still evident in tissue sections from two mice collected at 2 months (not shown) and 6 months after administration of vector 6 (compare Fig. 2, D–F, with 4, A–C).
Quantification of the Knockdown of NR1 mRNA and Protein. To determine the extent of knockdown of NR1 mRNA and protein, we analyzed the immunohistochemistry and in situ images using Metamorph software. The changes in the NR1 mRNA or protein levels were estimated by measuring the integrated gray level per unit area in the whole SCDH region of the ipsilateral side and comparing to the corresponding contralateral side. As shown in Fig. 5, the active vectors significantly reduced the NR1 mRNA and protein labeling on the ipsilateral side by 60 to 75% (p < 0.01), whereas there was no significant change after vector MM-6. No significant difference was observed among the three active siRNAs in terms of the magnitude of the knockdown for each against MM-6 and against each other.
Effects of the Viral Vector and the Knockdown of NR1 Subunit on Motor Function, Acute Thermal, and Mechanical Stimuli. None of the mice that received either the mismatch (vector MM-6) or siRNA (vector 6) showed any deficits in motor function as assessed by the three reflex tests described under Materials and Methods.
Figure 6, A and B, shows that injection of the mismatch (vector MM-6) or the siRNA (vector 6) into the SCDH had no effect on acute noninjury-inducing thermal (Fig. 6A) or mechanical (Fig. 6B) stimuli applied to either the ipsilateral (right) or contralateral (left) paws. Compared with the responses before administration of the viral vector, neither vector altered the latency for thermal paw-withdrawal (Fig. 6A) or the mechanical stimulus threshold (Fig. 6B). When the paw ipsilateral to the vector injection into the SCDH was compared with the contralateral paw, no difference was observed. Neither the procedure for delivering the virus (Fig. 6, A and B) nor the spatial knockdown of NR1 (Fig. 6B) affected noninjury-inducing stimuli.
Effects of the Knockdown of NR1 Subunit on Injury-Induced Mechanical Allodynia Resulting from CFA-Induced Inflammation. The intraplantar injection of CFA resulted in an equivalent edema in the ipsilateral paws of both the mice treated with the mismatch vector MM-6 and the siRNA vector 6 (Fig. 7). In contrast, only the MM-6-treated mice showed mechanical allodynia as measured as a reduction in the mechanical threshold (50% g threshold) using von Frey hairs applied to the paw (Fig. 7). The mice that received siRNA vector 6 were protected from CFA-induced mechanical allodynia (p < 0.05).
Previous studies from this laboratory have shown that an antisense oligonucleotide targeting the NR1 subunit (Shimoyama et al., 2005) or a conditional deletion of this subunit at the genomic level in SCDH (South et al., 2003) results in a decrease or abolition of NMDA receptor mediated injury-induced pain, central sensitization and morphine tolerance. Thus, reduction of NMDA receptor expression at the gene or mRNA level has well established consequences for SCDH neuronal function.
This study clearly demonstrates some of the advantages of using RNAi to target gene expression. First, we show that in vivo, RNAi can be controlled both temporally and spatially avoiding the complications resulting from constitutive gene deletion, including deletion of the gene in systems other than those of interest, gene redundancy during development or embryonic or early postnatal lethality (South et al., 2003). We were able to achieve a reduction of NR1 mRNA and protein that averaged 60 to 75%. This reduction is greater than the 30% reduction in NR1mRNA we observed with antisense (Shimoyama et al., 2005) and nearly comparable to the 80% reduction in NR1 mRNA and protein that results from a conditional deletion of the NR1 gene using the Cre/loxP technology (South et al., 2003). This siRNA-induced reduction in gene expression occurred after a single treatment with the viral vector and avoided the repeated injections over several days as required with antisense (Shimoyama et al., 2005) or the necessity of generating a mutant mouse with the gene of interest floxed (South et al., 2003). Thus, our approach is much more facile than the conditional gene deletion approach and can be easily extended to other species including the rat (Tan et al., 2005).
Although synthetic siRNAs for several mammalian genes are available commercially e.g., (Czauderna et al., 2003; Sørensen et al., 2003), we used the Whitehead propriety program to design the siRNAs. This approach allowed considerable control over the design algorithm. Using the published criteria (Mittal, 2004) and the National Center for Biotechnology Information Blast program, we were able to obtain sequences homologous to both rat and mouse NR1 mRNA sequences. Therefore, unlike a constitutive knockout approach, which is usually limited to a particular mammalian species, we can also explore the effects of these siRNA sequences and the consequences of NR1 gene knockdown in both mouse and rat. Furthermore, this approach can be used to efficiently design, test and validate siRNAs that target other candidate genes that may be involved in spinal pain processing.
A facile method for evaluating the efficacy of potential siRNA and control sequences utilized the in vitro psiCHECK luciferase test. The three shRNAs sequences that produced the greatest decrease in the Renilla luciferase signal in vitro were found to be highly effective at reducing NR1 gene expression in vivo. Rather than packaging and purifying virus for each potential siRNA sequence, which is costly and time consuming, we were able to select the sequences of interest using the in vitro luciferase assay. Furthermore, by screening the siRNA sequences directly in the AAV plasmid, we avoided any confounds that might occur when sequences active in vitro are then subcloned into a vector for virus production.
Two weeks after the direct intraparenchymal injection of rAAV vectors into the SCDH, significant transduction is observed that is confined to neurons without evidence of cellular toxicity at both the light microscope and ultrastructural levels (South et al., 2003). Although, ultrastructural studies were not conducted in the present report, a previous investigation showed that the loss of NR1 resulting from gene deletion is confined to somata and dendrites in SCDH. Here we show that rAAV viral vectors expressing an siRNA induced a spatially localized knockdown of NR1 gene expression in SCDH neurons of adult mice, whereas a control virus expressing a mismatch sequence failed to do so. Our injection protocol resulted in the rostrocaudal spread of viral transduction that included the three lumbar segments (L4–L6) that comprise the sciatic sensory distribution from the paw to the SCDH. The expression of GFP and decreases of NR1 mRNA and protein were confined to the ipsilateral dorsal horn, with only minimal expression detected in the ipsilateral ventral horn or on the contralateral side. Thus, siRNA delivered to the SCDH by a vector results in a reproducible spatial localization that is difficult to achieve or verify with pharmacological agents and not possible with conventional gene targeting.
Several recent reports have demonstrated the utility of rAAV vectors for gene silencing in the CNS using RNAi (Hommel et al., 2003; Babcock et al., 2005; Machida et al., 2006; Musatov et al., 2006). rAAV vectors can be produced in high titers and can transduce postmitotic neurons in the brain and SCDH in a highly spatially localized manner (Kaplitt et al., 1994; Kaspar et al., 2002; South et al., 2003).
The shRNAs delivered in vivo by our rAAV vectors were converted to siRNAs that resulted in the prolonged knockdown of NR1 expression. Although we did not test expression beyond 6 months, stable GFP expression was seen at 10 months after in vivo administration of related rAAV vectors (South et al., 2003). Long-lasting and stable expression can be especially important when exploring the therapeutic opportunities for the use of RNAi to knock down the expression of genes contributing to persistent conditions such as chronic pain.
Although RNAi offers several important advantages for gene targeting or knockdown, there is evidence of nonspecific effects with siRNA, including activation of the interferon pathway (Bridge et al., 2003; Sledz et al., 2003) resulting in general decrease in protein synthesis and eventual cell death. In addition, shRNAs derived from adeno-associated virus type 8 can down-regulate liver-derived microRNAs, resulting in fatal liver toxicity within 1 month after intravenous infusion of the vector (Grimm et al., 2006). We did not observe any significant local cell loss or systemic toxicity up to 6 months after the injection of our vectors. Nuclear staining with TO-PRO-3 demonstrated normal number and morphology of nuclei when vector-injected and noninjected SCDH were compared. In addition, no differences were observed in the levels of NR1 protein when the ipsilateral and contralateral SCDH of mice injected with the control vector were compared. These observations, together with the direct demonstration of a substantial loss of NR1 mRNA, indicate that the decrease in NR1 expression resulted from an RNAi-induced cleavage of NR1 mRNA and not from nonspecific secondary changes in cellular processes or oversaturation of small RNA pathways. We do not know whether we avoided the latter toxicity because the shRNAs expressed by our vector did not reach a concentration that was sufficient to compete with endogenous small RNA pathways or because in vivo these pathways were less critical for the survival of neurons than hepatic cells.
We found that neither the siRNA-induced knockdown of NR1 expression in SCDH by vector 6 nor the injection of vector MM-6 expressing a mismatch sequence altered the mechanical paw-withdrawal or the thermal paw-withdrawal thresholds. These results are consistent with previous studies that have demonstrated that NMDA receptors in SCDH mediate the responses to injury-inducing stimuli but not to nondamaging noxious stimuli (Woolf and Costigan, 1999; South et al., 2003). The application of acute high-intensity nondamaging thermal stimuli to the paw activates high-threshold primary afferents, C and Aδ fibers, that rapidly respond with well localized pain. These responses along with those to normal mechanical (tactile) stimuli do not require the presence of functional NMDA receptors in the SCDH (South et al., 2003). Rather, synaptic transmission after these stimuli is mediated by AMPA receptors, which generate fast excitatory postsynaptic currents (Woolf and Costigan, 1999).
In contrast, the siRNA-induced knockdown of NR1 expression in SCDH by vector 6 completely prevented the mechanical allodynia that was seen at 24 and 48 h after CFA in the MM-6 controls. Noxious tissue-damaging stimuli, such as the inflammatory agent CFA, generate a complex cascade of transmitters and modulators in SCDH that initiate activity-dependent neuroplastic changes that result in pain hypersensitivity (Woolf and Costigan, 1999). NMDA receptors play a prominent role in this cascade, which is initiated by a Ca2+ influx through open NMDA receptors.
NMDA receptor antagonists or deletion of the NR1 subunit from the SCDH can block central sensitization and the consequent pain hypersensitivity (Woolf and Thompson, 1991; Woolf and Costigan, 1999; South et al., 2003). The observation that both groups of mice had an equivalent increase in paw size indicates that the CFA-induced peripheral inflammation was of the same intensity in both vector 6- and MM-6-treated mice. However, the central processing of pain signals from the periphery was altered in the SCDH of mice where the expression of the NR1 subunit of the NMDA receptor was decreased by the vector 6 siRNA.
Our results demonstrate that siRNA targeting the NMDA receptor can prevent injury-induced pain hypersensitivity and suggest that this approach may provide a useful approach for both target validation and therapeutic intervention for injury-induced pain at CNS sites. Vector-delivered siRNA provides an important new tool for pain research that extends beyond the gene expression targeted in this report.
We thank Drs. T. Tuschl, J. Pena and M. Landthaler for the assistance with the design and screening of the siRNAs and for helpful comments on the manuscript, Drs. M. Kaplitt and S. Musatov for the rAAV plasmid and for assistance with the production of the vectors, and the late Dr. H. Robertson for assistance with the initiation of this project.
This work was supported in part by National Institute on Drug Abuse (NIDA) Grants DA001457 and DA000198 (to C.E.I.), NIDA Training Grant DA007274 (to S.M.G.), and NIDA Center Grant DA005130.
S.M.G. and Q.X. contributed equally to this work.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: RNAi, RNA interference; NMDA, N-methyl-d-aspartate; NR1, NR1 subunit of the NMDA receptor; siRNA, short interfering RNA; rAAV, recombinant adeno-associated virus; SCDH, spinal cord dorsal horn; GFP, green fluorescent protein; shRNA, short hairpin RNA; CNS, central nervous system; HEK, human embryonic kidney; CFA, Complete Freund's adjuvant; IPI, intraparenchymal injection.
- Received March 20, 2007.
- Accepted June 4, 2007.
- The American Society for Pharmacology and Experimental Therapeutics