We conducted all experiments using adult male C57BL/6J mice (20–30 g; Jackson Laboratories, Bar Harbor, ME). All mice were on a 12:12 hr light:dark cycle with food and water available ad libitum. We followed the International Association for the Study of Pain (IASP) guidelines for investigations of pain in animals (Zimmerman, 1983
). The Institutional Animal Care and Use Committee of the University of Maryland School of Medicine approved these experiments. In accordance with these guidelines, we used the minimum numbers of animals to meet the rigor necessary for this series of experiments (see ).
Study Variables and Number of Animals Used for Each Experiment
Animal Model Generation
Age-matched mice received a single 50 mg/kg intravenous (IV) dose of 2′, 3′-didehydro-3′-deoxythymidine (Sigma Aldrich, St. Louis, MO; brand name: Zerit; generic name: stavudine) into the tail vein. Control mice received a weight-based dose-equivalent volume of physiological saline vehicle via tail vein. We selected stavudine as the agent as this is a widely prescribed NRTI in clinical use and is associated with significant neuropathic pain. Although patients are generally administered stavudine orally, previous studies have shown that both oral and intravenous administration routes produce similar nocifensive behavioral profiles in rodents (Joseph et al., 2004
). Thus, we chose to use the intravenous route to minimize the handling stress to the animals associated with daily oral gavage.
For the injection, the mice were placed in a Broome Style Rodent Restrainer (Plas Labs, Lansing, MI) with the tail extending from the end. The tail was vasodilated by immersion in a warm water bath (40–42°C) for 15–30 s prior to injection. A 100-μl Hamilton syringe with a ½ inch 30g needle was used for the injection. The lateral tail vein was located and the tail was immobilized between the thumb and forefinger. The needle was inserted, bevel up, at a 10° angle in the rostral direction. We injected the solution slowly while watching closely for the vein to blanch and to ensure that there was no detectable swelling of the tail near the injection site. Following needle removal, we applied pressure to the injection site for 15–30 s to stop bleeding and avoid hematoma formation. Total weight-based injectate volumes for drug- and vehicle-treated animals ranged from 40 to 60 μl.
Nocifensive Behavioral Testing
The nocifensive behavior of paw withdrawal from a mechanical stimulus was used to assess the development of tactile allodynia. A series of von Frey filaments (Touch Test Sensory Evaluator Kit, myNeurolab.com, St. Louis, MO), with bending forces that ranged from 0.04g to 1.40g, was used to deliver the tactile stimuli. Naïve mice were tested before drug administration to determine their tactile threshold, defined as the fiber with the smallest bending force that elicited three aversive responses (paw withdrawal) out of five trials. Tactile allodynia was determined to be present if the response threshold shifted to the left, such that a previously nonnoxious fiber with a bending force less than the naïve threshold fiber elicited three aversive responses out of five trials. Two groups of mice (drug group n = 6 and vehicle group n = 6) were tested pre-drug (naïve) and then 1, 7, 14, 21, and 28 days after drug administration to observe changes in their behavioral responses over time.
For behavioral testing, the mice were placed in individual Plexiglas cubicles (8.5 cm in length × 4 cm in height × 4 cm in width) on an elevated wire mesh platform and allowed to acclimate for approximately 1 hr, during the course of which exploratory and grooming activity ended. Each von Frey filament was applied to the hind paw until the filament just bent and was held in place for 5 s or until the mouse withdrew its paw. Each filament was tested five times on each hind paw. A positive (aversive) response to the stimulus was defined as a brisk withdrawal, with or without shaking or licking, of the hind paw either during or immediately upon removal of the filament application. The tactile stimuli were applied to the plantar surface of the hind paw, starting with the 0.4g filament. If the 0.4g filament elicited three positive responses out of five trials, then the mouse was tested moving downward through the filament series toward 0.04g until the filament with the smallest bending force to elicit three positive responses was identified and recorded as the threshold fiber. If the 0.4g filament did not elicit three positive responses, then the mouse was tested moving upward through the series toward 1.4g until the filament with the smallest bending force to elicit three positive responses was identified and recorded as the threshold fiber. Mice that are naïve to experimental treatment typically show a positive response to the 0.6g–1.4g fibers, while mice that are treated with stavudine typically respond positively to the filaments in the 0.04g–0.4g range. Because stavudine is a systemic drug, the thresholds of both hind paws were averaged for each mouse. In all nocifensive tests, the observer was blind to condition. Data are presented as the mean gram bending force ± standard error of measurement (SEM).
Once nocifensive responses were established in drug-treated animals, two early time points were identified as having potential to illuminate molecular targets involved in the development of allodynia. Day 1 was selected because of the robust nocifensive behavioral response present 24 hr following drug administration. Day 3 was selected as a likely early point to capture molecules that are involved in the development of neuronal plasticity in the spinal dorsal horn to sustain the robust tactile allodynia. Separate cohorts of animals were used for each time point for microarray and quantitative polymerase chain reaction (qPCR; see ). The mice were administered weight-based doses of stavudine or physiological saline (vehicle) on Day 0 for RNA extraction on Days 1 and 3 after drug treatment for both the microarray analysis and qPCR assay.
For the tissue harvest, the mice were euthanized by cervical dislocation followed by rapid removal of the spinal cords, which were immediately flash frozen on dry ice. The lumbar region of each spinal cord was isolated and dorsal horn dissected from ventral horn. A standard TRIzol extraction protocol (Invitrogen, Carlsbad, CA) was used to extract RNA from the dorsal horn tissue samples, with each sample representing a single experimental animal. The RNA was further purified using the RNeasy Mini Kit (Qiagen, Valencia, CA). The quality and purity of the samples was analyzed by spectrophotometry and using the Experion RNA StdSens analysis kit (Bio-Rad, Hercules, CA). Samples with a 260/280 ratio of approximately 2.0 and two sharp peaks that corresponded to the 18S and 28S RNA on the RNA gel were considered of sufficient purity to be used in the microarray analysis and qPCR assays.
RNA Labeling and Microarray Processing
All microarrays were processed by one person in the same laboratory, following a standardized lab protocol to minimize nonbiological technical bias. Purified total RNA (3 μg) was reverse transcribed into cDNA using a 3′–Amplification One-Cycle cDNA Synthesis Kit (Affymetrix, Affymetrix, Inc., Santa Clara, CA, P/N 900431). The synthesized double-stranded cDNA was then purified using a GeneChip sample cleanup module (Affymetrix, P/N 900371) and used as a template in the subsequent synthesis of the biotin-labeled cRNA. A GeneChip IVT Labeling Kit (Affymetrix, P/N 900449) was used to synthesize the biotin-labeled cRNA and quantified per manufacturer guidelines. To monitor the labeling process independently from the quality of the starting RNA samples, a set of poly-A controls (Affymetrix, P/N 900433) were amplified and labeled together with all samples.
Gene-expression analysis was performed with the Affymetrix GeneChip Mouse Genome 430 2.0 array (Affymetrix, P/N 900496), which contain 45,000 probe sets and reports the gene-expression level of transcripts and variants from over 34,000 mouse genes. A total of 15 μg of synthesized cRNA from each sample was hybridized on the gene chip in a hybridization oven at 45°C for 16 hr at 60 rpm. All probe arrays were scanned with an Affymetrix GeneChip Scanner 3000 with digital image data processed using the Affymetrix GeneChip Operating Software 1.4 (GCOS, Affymetrix).
Microarray Data Analysis
The microarray data were provided as a flat file (.dtt). The .dtt file was extracted using the GeneChip Operating Software, and raw data files (.cel) for each hybridization were generated. Raw .cel files were then subjected to background correction and intrachip normalized using JMP Genomics v3.2 software (SAS, Cary, NC). Briefly, background correction is the process of correcting probe intensities on any one array using probe information only on that array. Normalization is the process of removing nonbiologic variability between arrays from the analysis. The software breaks the array into regions and within each grid determines a signal-to- noise value for that grid. For each probe, the background/noise adjustment takes the weighted average of the grid background/noise values, with weights dependent on distance from the probe location to the center of each grid. The robust multichip average (RMA) method (Irizarry et al., 2003
) fits a robust model to the log transformed raw data using a quantile approach with a median polish algorithm of probe and chip effects to predict chip expression with a robust multichip average. Differences in gene expression were analyzed using repeated measures analysis of variance. To correct for multiple testing, we applied the false discovery rate (Reiner, Yekutieli, & Benjamini, 2003
) with an a priori alpha of .05. Thus, only genes that were significantly different at p
< .05 after correction were considered for further analysis.
qPCR Verification of Identified Target Gene
The identification of significant changes in expression of the giant axonal neuropathy 1 (Gan1) gene through microarray was validated using a quantitative real-time qPCR technique. This approach uses the PCR to detect, amplify, and quantify the absolute number of copies of a specific DNA sequence in a sample relative to a normalized DNA sample. Two cohorts of 8 mice (drug group n = 4 and vehicle group n = 4 per cohort, see ) were administered weight-based doses of stavudine or physiologic saline vehicle on Day 0 for RNA extraction on Days 1 and 3 after drug treatment. Harvested RNA was reverse transcribed using the Superscript II reverse transcriptase and random primers (Invitrogen). We performed 40 cycles of qPCR using the Lightcycler 480 SYBR Green I Master Mix (Roche Applied Science, Indianapolis, IN). Because fluorescence dye is used to monitor the double-stranded DNA, plotting fluorescence as a function of temperature as the thermal cycler heats through the dissociation temperature of the DNA produces a DNA melting curve. The position and shape of this DNA melting curve can be used to differentiate amplified DNA sequences separated by less than 2°C in melting temperature. The relative abundance of each transcript was computed using Roche Lightcycler 480 Relative Quant software (Roche Applied Science). The following Gan1-specific amplification primers, which span at least two intron/exon boundaries, were selected: forward: 5′-ATG CCC ACT GAA AGA GAG GTT-3′; reverse: 5′-TGG CAG GGA TGC ATA GGT TCT GAT-3′ (Integrated DNA Technologies, Coralville, IA). We used the β-actin gene as the reference gene with these specific primers: forward: 5′-TGT GGT GCC AGA TCT TCT CCA TGT-3′; reverse: 5′-TGT GGT GCC AGA TCT TCT CCA TCT-3′ (Integrated DNA Technologies).
Western Blot Analysis of Gene Protein Product
To determine whether changes to the Gan1 gene result in changes to the expressed level of its protein product, gigaxonin, 9 mice received stavudine or physiologic saline via tail vein on Day 0 for tissue harvesting on Days 1 and 3 after drug treatment (see for cohort breakdown).
Euthanization and spinal cord removal and preparation followed the same procedure as described above. The spinal dorsal horn tissue was mechanically homogenized in lysis buffer with protease and phosphatase inhibitors (Tris-buffered saline [pH = 8.0] plus 10% glycerol, 0.1% Triton X-100, protease and phosphatase pellets [Roche Applied Science]). SDS (sodium dodecyl sulfate), solubilized tissue extracts were incubated at 100°C for 5 min, fractionated on 4–12% NuPAGE bis-tris gels (Invitrogen) and transferred to a nitrocellulose membrane. After blocking in nonfat dried milk, membranes were incubated overnight at 4°C with a primary antibody to gigaxonin (anti-gigaxonin 1:500, Santa Cruz Biotechnology, Santa Cruz, CA) followed by incubation with a peroxidase-conjugated secondary antibody (antirabbit IgG 1:4000, Amersham Pharmacia Biotech, Piscataway, NJ) and visualized by chemiluminescence (Amersham Pharmacia Biotech). Blots were quantified by scanning autoradiographs into ImageJ (NIH, Bethesda, MD, version 1.62) imaging software to determine the optical density of each band. After processing, to standardize samples for protein loading quantities, the blots were stripped and reprobed with a primary antibody to actin (anti-actin 1:500, Sigma Aldrich) followed by incubation with a peroxidase-conjugated secondary antibody (anti-rabbit IgG 1:4000, Amersham Pharmacia Biotech) and visualized by chemiluminescence (Amersham Pharmacia Biotech). The blots were quantified again by scanning autoradiographs into ImageJ (NIH) imaging software to determine the optical density of each band.
The behavioral data, expressed as mean gram force ± SEM, were analyzed using one-way analysis of variance (ANOVA). The microarray data (reported as percentage change) were analyzed using repeated measures ANOVA with false discovery rate correction to control multiple testing error. Post hoc testing was done using Tukey’s honestly significant difference (HSD). The qPCR and Western blot data (reported as percentage change) were analyzed using repeated measures ANOVA with post hoc testing by Tukey’s HSD.