All procedures were approved by the Canton of Vaud’s Committee on Animal Experimentation (Switzerland), in accordance with Swiss Federal Law on Animal Welfare and International Association for the Study of Pain guidelines .
The spared nerve injury (SNI) model of neuropathic pain was previously described in rats [32, 52] and mice . Briefly, adult C57BL/6 J mice (Charles River, L’Arbresle, France) were anesthetized with 1.5% isoflurane and after exposure of the sciatic nerve, the common peroneal and tibial nerves were ligated together with a 6.0 silk suture (Ethicon, Johnson and Johnson AG, Zug, Switzerland) and transected. In the SNI variant (SNIv(cp,t))  the ligation and transection were performed on the sural nerve, leaving the common peroneal and tibial nerves intact. The incision was closed in distinct layers (muscle and skin). Sham surgery was performed similarly except for the nerve ligation and transection.
Spinal nerve ligation (SNL) surgery was adapted from the procedure described by Kim and Chung , and transposed to mice. Briefly, after skin and muscle incision the L5 transverse process of vertebra was exposed and carefully removed. The L4 and L5 spinal nerves were exposed and the L5 spinal nerve was tightly ligated. The incision was closed in distinct layers (muscle and skin).
Briefly, mice were terminally anesthetized with sodium pentobarbital (Esconarkon; Streuli Pharma AG, Uznach, Switzerland) and the biceps femoris muscle of the left thigh was incised. The genus descendes artery was used as a reference for the muscle incision, which lead to the exposure of the sciatic nerve and the trifurcation of the peripheral branches: common peroneal, tibial and sural nerves. The sciatic nerve was followed in the rostral direction, removing muscle tissue until reaching the vertebral column. Vertebral lamina, pedicles and spinous processes were trimmed away to expose the spinal cord and DRG. For the nomenclature of DRG, refer to Figure 3.
Quantitative real-time reverse transcription PCR (qRT-PCR)
Ipsilateral DRG were rapidly dissected and collected in RNAlater solution (Qiagen, Basel, Switzerland). For SNI, 2 series of mice were used, one where the L4 and L5 were pooled together, as usually done (4 DRG pooled from 2 mice per sample), and one series where L3, L4 and L5 were dissected separately (8 DRG pooled from 8 mice per sample). For SNL, L4 and L5 DRG were consistently separated (2 DRG pooled from 2 mice per sample) as they represented non-injured and injured neurons, respectively. For all conditions tested, n = 3–4 / sample were used. mRNA was extracted and purified using a RNeasy Plus Mini Kit (Qiagen) and quantified using a RNA 6000 Nano Assay (Agilent Technologies AG, Basel, Switzerland). A total of 600 ng of RNA was reverse-transcribed for each sample using Omniscript Reverse Transcriptase Kit (Qiagen). Primer sequences and working concentrations for Navs α-subunits, ATF3 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) can be found in Table 1.
We used GAPDH as a reference gene to normalize Navs mRNA expression since it is stable between sham and SNI conditions (M-value = 0.30), taking into account the efficiencies of qPCR reaction. Gene-specific mRNA analyses were performed using the iQ SYBR-green Supermix (BioRad, Reinach, Switzerland) and the iQ5 real-time PCR detection system (BioRad). Only reactions with appropriate amplification and melting curves were analyzed. All samples were run in triplicate. Normalized transcripts were then expressed as a ratio of the level in SNI and SNL models to that in sham-operated mice. The bar graphs in Figures 2A, B and C represent these ratios for each isoform. Each qPCR product was sequenced to confirm the specificity of amplification. Briefly, qPCR products were first loaded on a low-melt agarose gel to confirm the size of the amplicon. Amplicons were then subcloned in a pGEM-T Vector System (Promega, Madison, WI, USA), and sent for sequencing using T7 promoter (Fasteris, Geneva, Switzerland). All qPCR products were validated as specific for each of the Navs tested using the primers in Table 1.
One week after sham SNI or SNIv(cp,t) surgery, animals were terminally anesthetized with sodium pentobarbital (Esconarkon), and then transcardially perfused with saline solution, directly followed by paraformaldehyde 4% diluted in phosphate buffered saline (PBS). The L3 to L5 DRG were dissected, post-fixed at 4°C for 90 min and then transferred in sucrose solution (20% sucrose in PBS) overnight. The following day, tissues were mounted in cryoembedding fluid (Tissue-Tek; Sakura Finetek, Zoeterwoude, Holland), frozen, cryosectioned in 12 μm-thick sections and thaw-mounted onto slides.
The rabbit anti-ATF3 antibody (1:300, Santa Cruz Biotechnology, Heidelberg, Germany) was used as the nuclear marker of injured neurons, and the goat anti-HuD antibody (Elav like proteins, 1:50, Santa Cruz Biotechnology) was used as the marker of total neuron numbers. Secondary antibodies were as follows: Cy3-conjugated anti-rabbit (1:400, Jackson ImmunoResearch, Suffolk, UK) for ATF3, and AlexaFluor 488-conjugated anti-goat (Molecular Probes, Basel, Switzerland) for HuD. Standard protocols for fluorescent immunohistochemistry were used. Sections of DRG were blocked for 30 min at room temperature (RT) with normal horse serum (NHS) 10% and PBS 1X-Triton X-100 0.3%. Primary antibodies were diluted in NHS 5% and PBS 1X-Triton X-100 0.1%, and incubated on sections overnight at 4°C. Slides were washed in PBS 1X and then incubated for 90 min at RT with the corresponding secondary antibody diluted in NHS 1% and PBS 1X-Triton X-100 0.1%. Slides were washed in PBS 1X and mounted in Mowiol medium (Calbiochem, Merck Millipore, Darmstadt, Germany).
Fluorescence was detected using an epifluorescence microscope (AxioVision, Carl Zeiss, Feldbach, Switzerland). Images were taken at 20× magnification, with the same parameters used for all experimental conditions. The complete DRG images were reconstructed by juxtaposing the different images using Photoshop CS4 software (11.0, Sun Microsystems, Redwood City, CA). Mean cell counts from each DRG were the average of 4 to 7 sections. Each first section was selected randomly, and the following ones were chosen every 72 m from the series of consecutive cut sections. Four animals were analyzed per condition. The percentage of injured neurons was expressed as the number of ATF3-IR neurons over the total number of cells (HuD-IR neurons). It should be noted that the percentage of ATF3-positive cells was probably a slight under-estimation of the actual proportion of injured cells because it represented the ratio of ATF3-positive cells over HuD-positive cells, which were counted independently to the presence or absence of the nucleus (one cell might have been counted twice in different stack).
For the expression of Navs after SNI, normalized transcripts were compared between sham surgeries versus SNI conditions using bilateral, unpaired Student’s t test. We used a two-way ANOVA with independent measures for the analysis of ATF3 expression and Navs (each of them separately) in the L3, L4 and L5 DRG independently; one variable being the treatment (SNI or SNL), the other being the DRG. We used post hoc Bonferroni tests to assess whether the treatment (SNI or SNL) had a significant effect on the expression of each Nav isoform. GraphPad Prism (version 5.01) was used to calculate statistics. This software does not provide exact p-values for post hoc Bonferroni tests and stars are shown for significance (*p < 0.05, **p < 0.01 and ***p < 0.001).