Background
Chronic neuropathic pain is a frequent and sometimes devastating sequel of nerve injury. A major cause of this type of pain is axotomy-triggered alteration in the expression of genes in corresponding dorsal root ganglion (DRG) neurons. Gene regulation directly alters the phenotype of neurons that have been axotomized, and in addition may affect neighboring uninjured DRG neurons as well as spinal cord neurons, and glia in both the cord and DRGs. One consequence is the development of electrical hyperexcitability and consequent ectopic neural discharge in affected neurons. Another is altered levels of neurotransmitters, receptors and neuro-immune mediators. These changes can lead to pain hypersensibility through abnormal signal processing in the peripheral and central nervous system [
1].
Identification of the genes involved in pain hypersensibility has been the objective of a number of prior studies. This work has revealed that many hundreds of genes are significantly regulated in DRGs in neuropathic pain models (e.g. [
2‐
11]). Only a small fraction of these, however, is likely to contribute to neuronal hyperexcitability and pain. Most are presumably involved in other nerve injury-triggered phenomena such as inflammation, tissue repair, nerve regeneration and apoptosis. Strategies are needed for vetting which changes are causally related to neuropathic pain, and which are not [
12]. Here we have implemented such a strategy.
Our approach, correlational analysis, is based on the fact that pain phenotype is highly variable among individuals [
13,
14]. Correlational analysis begins with quantification of pain phenotype on the one hand, and associated change in gene expression (or any other neurobiological parameter) on the other, in a variety of mouse strains. Then one searches for genes in which the degree of regulation correlates with the degree of altered pain phenotype. If up- or down-regulation of a particular gene transcript is an important pain substrate, then the degree of regulation should correlate across strains with the intensity of the pain.
It is not expected that correlational analysis, alone, will identify individual pain genes. Correlations can arise by chance, particularly considering practical limits on the number of mouse strains that can be analyzed, and the large numbers of transcripts that need to be examined simultaneously. Nonetheless, this approach can deliver an enriched pool of candidates for secondary analysis in mice, and ultimately for confirmation in human pain cohorts. Here we correlated neuropathic pain phenotype with nerve injury-evoked changes in gene expression in DRGs in five inbred mouse strains. First we report a limited test of the idea in which a small number of candidate genes previously implicated in neuropathic pain was hybridized in situ to DRG sections from the five mouse strains. The degree of regulation found was correlated against pain behavior. We then report results of a genome-wide correlational analysis based on regulation data obtained from Affymetrix expression microarrays supported by TaqMan.
Discussion
Primary sensory neurons in DRGs undergo a massive change in gene expression in experimental models of neuropathic pain, changes that undoubtedly contribute to altered somatosensory signal processing in the event of neuropathy [
2‐
11]. In our material, >11% of all expressed genes, 2552 on average, were found to be significantly up- or down-regulated in the L5DRG 3 days following transection of the L5 spinal nerve. This number is larger than prior estimates probably because of the proximal location of the lesion, and the improved protocol for detecting change. In common with prior estimates, however, it highlights a fundamental challenge. The sheer number of regulated genes means that additional methods are needed to identify which individual transcripts play an important role in pain sensation versus other processes induced by axotomy [
12]. Using correlational analysis we considerably shortened the list of candidates. In contrast, we found that very few genes were regulated in the L4DRG following transection of the L5 spinal nerve. This is surprising in light of functional studies that have argued that abnormal activity in "uninjured" L4DRG nociceptors may be important for the development of tactile allodynia in the SNL model [
19]. Perhaps the relevant changes occur in sensory signaling processes not related to altered gene expression in the L4DRG. The observations on the L4DRG confirm that the massive regulation seen in the L5DRG was in fact related to axotomy.
We recognize that the shortened gene list still includes many transcripts unrelated to neuropathic pain. Assuming random assortment, 5% of transcripts are expected to correlate with pain phenotype at p ≤ 0.05 by chance alone. However, despite residual contamination transcripts whose regulation is functionally related to pain behavior are expected to be retained at higher frequency than pain-neutral transcripts, yielding significant enrichment. At the most optimistic, assuming that all functionally relevant genes were captured by our correlation procedure, at least 95% enrichment is expected. Thus, while the lists of correlated transcripts remain too long to permit direct selection of candidate pain genes, they constitute a valuable and relatively bias-free look-up table, a tool for screening candidates derived from other approaches. The list, however, is clearly not complete. In addition to pain-correlated genes missed, there is no inherent biological contradiction to the possibility that some genes are functionally related to pain phenotype but that their regulation does not contribute importantly to strain differences.
Might the gene lists be further enriched by increasing the stringency of the selection criterion from p ≤ 0.05 to, say, p ≤ 0.01? While not impossible, given the intrinsic noise present in the phenotyping process we believe that this step would be more likely to eliminate functionally significant transcripts. This is because within the enriched list, genes with the highest correlation coefficients are not much more likely to be functionally related to pain than those with lesser, but still high correlation coefficients. Further vetting requires additional biological information. For example, there is a clear benefit to increasing the number of data points used to generate correlations. We accomplished this in three of the five strains studied by collecting phenotypic data (on tactile allodynia) from individual mice rather than relying on strain means. This resulted in a reduction in the number of genes with significant correlation coefficients, presumably by removing more false positive results than true positive results. Related approaches are to increase the number of mouse strains studied and the behavioral diversity among them.
Two other enrichment approaches have been attempted in the past. In one, lists of regulated transcripts were assembled using a variety of different rodent models of painful neuropathy [
11,
20,
21]. As expected, for each model hundreds of transcripts were significantly regulated in the axotomized DRGs. Lists were shortened by identifying genes similarly regulated in more than one model. This strategy is likely to reduce random noise, but it is also likely to capture regulated genes that are not related to pain. Transcripts related to regeneration and apoptosis, for example, would all be positively selected for. The approach also makes the risky assumption that different pain models (diagnoses) share the same underlying pathophysiology. If this is not the case then pain-related genes would be systematically excluded.
A second approach compared genes regulated in the DRG following nerve injury in two rat strains that had a consistent difference in pain phenotype in the SNL neuropathy model [
4]. Although closer in concept to our study, the rat strains used were not congenic, or even particularly close in genetic background. For this reason observed differences in gene regulation may well have reflected pain-neutral differences in genetic background. Correlations based on only two strains provide zero degrees of freedom. We used five strains, and alternatively 26 individual mice (from three strains) to generate correlations. In this context it is worth noting that not only does baseline gene expression level within a particular tissue (e.g. DRG) vary among mouse strains, but it also varies among tissues and among cell types within tissues [
22,
23]. Our in situ hybridization data affirm this to be the case with respect to DRG neurons of varying size, at least for some transcripts. An example is up-regulation of
Trpv1 in CBA mice and down-regulation in B6 mice, and the de novo appearance of
Trpv1 in DRG cells of medium and large size, but only in CBA mice.
An intrinsic limitation of microarray based studies, including ours, is that massive multiple hypothesis testing undermines the ability to establish statistical significance for any given transcript identified. This limitation is only partly eased using FDR analysis [
24]. Ultimately, our enriched list of pain-related genes needs to be subjected to secondary screens. In situ hybridization is one example. Although this method is too resource intensive to be used to screen large numbers of transcripts, it represents an independent implementation of correlational analysis. The overall pattern of gene regulation seen in the arrays was reiterated in all six candidate genes subjected to in situ analysis. However, the across-strains pattern was reproduced in only two of the six. There are several possible reasons for this difference between microarray and in situ measurements. Microarray (and TaqMan) analysis integrates over the entire ganglion, including neurons, glia and other resident cells, and includes cells with low expression levels. The cellular source of the mRNA is not identified. Likewise, the method does not take into account the possibility of up-regulation in one cell population balanced by down-regulation in another. Finally, stable expression levels in many cells may mask significant regulation in a small but important subpopulation.
Interestingly, the degree of regulation correlated significantly with pain phenotype for two of the six candidate genes studied with in situ hybridization (33%;
Scn11a and
Trpm8). This yield was much higher than for the array analysis which considered all regulated genes (p < 0.001). This outcome, which presumably reflects the additional information that underlay the choice of the six candidates, lends validation to the correlational approach. Regulation of the Na
+ channel α subunit
Scn11a correlated with levels of spontaneous pain behavior, and regulation of the cool receptor
Trpm8 correlated with heat hypersensibility. For
Scn11a, expression was reduced in all strains, with the degree of reduction minimal in animals that exhibited a high level of ongoing pain behavior and maximal in strains with minimal ongoing pain. A good deal of evidence links Na
+ channels, including
Scn11a, to the emergence of ectopic afferent hyperexcitability after nerve injury and consequent spontaneous firing and spontaneous neuropathic pain [
1,
16,
25]. The observed correlation is therefore consistent with pain-protected mouse strains showing the greatest level of
Scn11a down-regulation. We note, however, that mice with null mutation of
Scn11a continue to show tactile allodynia in neuropathy models [
26,
27]. These knockout mice have not been checked for autotomy.
For
Trpm8, down-regulation occasioned reduced heat allodynia.
Trpm8 functions as a cold transducer in primary afferent nociceptors [
28]. The relation to thermal sensation is obvious, but the link to altered heat response requires further investigation. Cold allodynia in the SNL model has not been compared in our series of mouse strains. In intact mice, however, sensitivity to cold is genetically correlated with sensitivity to heat [
29]. We stress that failure to find a significant correlation between gene regulation and pain phenotype in the other four transcripts tested using in situ hybridization does not mean that the corresponding genes, or their regulation following nerve injury, is not important for pain phenotype. Each of these genes may well play an essential role in pain physiology. However, the result indicates that their degree of regulation does not contribute much to across strain variability, at least in the five strains examined.
Methods
Animals
We used young adult males (8–19 weeks of age) of five different inbred mouse strains obtained from The Jackson Laboratory (Bar Harbor, ME, USA). The strain were: AKR/J (AKR), CBA/J (CBA), C3H/HeJ (C3H), C57BL/6J (B6 or C57) and C58/J (C58). Experiments were carried out at two different venues, the Karolinska Institutet (KI, Stockholm) and Sanofi-Aventis Deutschland GmBH (S-A, Frankfurt). Both before and after experimental surgery mice were maintained in groups in transparent plastic cages bedded with wood shavings (Lantmännen, Kimstad, Sweden at KI) or Lignocel HBK 1500/3000 (JRS, Resenberg, Germany at S-A). Water and food pellets (RM3 pellets, Special Diets Services, Witham, UK at KI; product R/M-H 10 mm, Ssniff, Soest, Germany at S-A) were available ad libitum. At both venues the light:dark cycle was 12:12 with lights on at 06:00 AM. All procedures were carried out with the approval of the Institutional Animal Care and Use Committee of KI or in accordance the German Animal Protection Law as overseen by the local Ethics Committee for Research on Laboratory Animals at S-A.
Experiments were based on the SNL model of neuropathic pain [
30] with slight modifications. For in situ hybridization analysis (KI) we used a total of 40 mice, 8 from each of the five strains. For each strain 6 mice were sacrificed 3 dpo and 2 were intact controls. For TaqMan and microarray analysis (S-A) we used a total of 144 mice. Observations were made at two postoperative time points, 1 dpo for C58 and C3H mice and 3 dpo for all five strains. In each group 10 mice underwent SNL surgery and 8 mice underwent sham surgery. An additional group of 18 B6 mice (10 SNL and 8 sham) were used for "step 1" TaqMan pilot observations (see below). In all five strains measurements of tactile and heat allodynia in the SNL model of neuropathic pain, and of spontaneous pain in the neuroma model of neuropathic pain, were based on our observations on other individuals of these same strains [
13,
15]. The reason is that we wanted to track pain behavior in the strains beyond 3 dpo. However, in three stains we made behavioral measurements of tactile allodynia 2 dpo on the actual, individual mice that were sacrificed on the following day for use in the microarray analysis.
Surgical procedures
Mice were anaesthetized with 350 mg/kg i.p. chloral hydrate (KI) or 1.5–3% isoflurane in a 1:2 mixture of O2 and N2O (S-A). Lower back skin was shaved and wiped with 70% ethanol. Using sterilized instruments an incision was made through the skin and the paraspinal muscles were separated from the spinous processes at the L5-L6 levels. The L6 transverse process was then removed exposing the L5 spinal nerve. Finally, the spinal nerve was transected 3–5 mm distal to the ganglion, without ligation, and a 3–4 mm segment was removed from the distal nerve stump. Care was taken throughout to avoid damage to the adjacent L4 and L6 nerves. To obviate the risk of severing the wrong spinal nerve, at terminal dissection the nerve identified as L5 was dissected into the lumbosacral plexus to insure that it indeed made a major contribution to the sciatic trunk. Sham operated control mice underwent the identical surgical exposures but nerves were not transected. In all animals muscle and skin incisions were closed in layers using surgical sutures and stainless steel clips (Stoelting, Wood Dale, Illinois, US). Mice were given buprenorphine as post-operative analgesic (0.06 mg/kg subcu., S-A) and following recovery from the anesthetic they were returned to their home cage.
Behavioral testing
We obtained data on tactile allodynia, heat allodynia and spontaneous pain in the neuroma model on independent groups of mice of the same five strains used to assess gene regulation. Methods and results were reported previously, with key values presented in Table
1[
13,
15]. Briefly, SNL surgery was carried out as described above. Tactile allodynia was assessed using von Frey monofilaments and heat allodynia was assessed by measuring withdrawal latency to a light beam projected from below onto the plantar hindpaw. Autotomy, the behavioral endpoint in the neuroma model of spontaneous neuropathic pain [
31,
32], was scored at weekly intervals for 5 weeks following transection of the sciatic and saphenous nerves. The scoring protocol was according to Wall et al. [
31]. Briefly, one point was registered for loss of one or more toe nails and an additional point was scored for injury to the distal and the proximal half of each toe for a total maximal score of 11. For calculation of the Spearman rank correlation statistic (r
s) behavioral scores were converted to strain ranks (1 to 5) where 1 represents the strain with the least nerve injury-induced behavioral change and 5 represents the strain with the largest change.
In addition, in the AKR, B6 and CBA strains, we obtained behavioral data on tactile allodynia from the individual mice used for the array analysis. Two days after SNL surgery the mice were placed separately in a 10 × 10 cm enclosure. Animals were habituated to the test environment for at least 30 minutes and then cutaneous sensitivity of the lateral plantar aspect of the hind paw was determined using the automated dynamic plantar aesthesiometer (Ugo Basile, Comerio, Italy). In this apparatus a metal filament 0.5 mm in diameter automatically rises from below the wire grid floor and applies a force that increases linearly from 0 to 5 g in 10s. The force that elicited paw withdrawal by the mouse, registered automatically by the apparatus, was taken as the animal's response threshold for a given trial. Each paw was tested four times over a period of 90–120 min with at least 30 min rest between tests. Tactile (hyper) sensitivity was defined as the average of the four measurements. For correlational analysis, the degree of tactile allodynia was calculated as the natural logarithm (ln) of threshold measurements for (SNL contralateral paw/SNL ipsilateral paw) - ln (mean of sham group contralateral paw/mean of sham group ipsilateral paw).
In situ hybridization
Three days postoperative mice were deeply anaesthetized with chloral hydrate (350 mg/kg i.p.) and perfused transcardially with 20 mL Tyrode's solution. Identification of the L5 spinal nerve and DRG was confirmed by a carbon mark left during surgery. DRGs L4 and L5 were quickly dissected out, snap frozen on dry ice and stored at -70°C for up to 6 days until sectioning. For operated mice DRGs were taken from the side of the SNL injury while for intact controls they were taken from either side. DRGs of a given level from operated and control animals of each strain were embedded side by side. Further, DRG pairs of this sort from all five strains were embedded together in single blocks for simultaneous processing using Tissue-Tek™ OCT compound (Sakura, Zoeterwounde, Netherlands). Blocks, each containing 10 DRGs, were cryo-sectioned at 12 μm and thaw-mounted onto Super Frost/Plus slides (Menzel GmbH, Braunschweig, Germany). This multiple embedding protocol ensured that sets of experimental and control ganglia from each mouse strain were processed in parallel under identical conditions.
Oligonucleotide probes were synthesized complementary to the mouse mRNA sequences of the Na
+ channel α-subunit Na
v 1.8 [GenBank: NM_009134], (base 2980–3021), the Na
+ channel α-subunit Na
v1.9 [GenBank: NM_011887], (base 3429–3470), the ATP receptor P2X
3 [GenBank: NM_145526], (base 1123–1164), the heat- and capsaicin-sensitive receptor TRPV1 [GenBank: NM_001001445], (base 531–572), the noxious cold-sensitive receptor TRPA1 [GenBank: NM_177781], (base 739–780) and the cooling- and menthol-sensitive receptor TRPM8 [GenBank: NM_134252], (base 809–850); Cybergene, Huddinge, Sweden). Procedures for in situ hybridization were as previously described [
33], but with slight modifications. Briefly, oligonucleotides were 3'-end labelled with (
33P)-dATP using terminal deoxyribonucleotidyl transferase (TdT). Sections were hybridized for 16–20 hours at 42°C in a humidified chamber with ~7 μL probe (~1.5 × 10
5 cpm/μL) made to approximately 200 μL solution for each slide, in a mixture of 4× SSC (standard saline citrate) buffer (1× SSC = 0.15 M NaCl, 0.015 M sodium citrate), 50% formamide, 1× Denhardt's solution (0.02% each of polyvinyl-pyrrolidone, bovine serum albumin and Ficoll), 1% sarcosyl, 0.02 M phosphate buffer (pH 7.0), 10% dextran sulphate, 500 mg/mL heat denaturated salmon sperm DNA and 200 mM dithiothreitol. Slides were then rinsed 5× 15 minutes at 60°C in 1× SSC and the last rinse was allowed to cool to room temperature. Slides were dipped in distilled water, dehydrated through graded ethanols (70%, 90%, 99.5%), air-dried, dipped in photographic emulsion (Kodak NTB2, diluted 1:1 in distilled water) and exposed at 4°C for 2–4 weeks. The slides were then developed, fixed and coverslipped using Pertex mounting medium (CellPath Plc, Hemel Hempstead, UK).
Quantification of in situ hybridization
Quantitative analyses of mRNA hybridization signals used a Nikon E600 microscope equipped with a darkfield condenser and a Nikon DXM 1200 digital camera. Images were captured under darkfield and brightfield illumination to facilitate identification of cell borders. Using Easy Image software (EI 3000, v.3000; Tekno Optik AB, Huddinge, Sweden) the darkfield image was thresholded based on pixel intensity. This mask was overlaid on the bright field image. The circumference of neuronal profiles in the field of view was outlined with an on-screen cursor, excluding profiles that had an area < 150 μm2. For each neuronal profile outlined, EI3000 extracted area and labelling intensity, which were used for subsequent calculation of the signal/noise (S/N) ratio. S/N ratio was based on comparison of mean pixel intensity within the perimeter of the cell and the background. Additional fields were sampled until a minimum of 150 neuronal profiles were evaluated per animal for each type of mRNA hybridization. Neuronal profiles were categorized by size as follows: small (150 ≤ 300 μm2), medium (300 ≤ 600 μm2) and large (> 600 μm2). Neuronal profiles with S/N ratio ≥ 5 were counted as "labelled" and the proportion of such labelled cells was calculated.
Detailed morphometric data will be given elsewhere. Here we focus on regulation. Change in expression (i.e. "regulation") for each marker examined, in each strain, was the ratio of the proportion of cells labelled comparing operated and control DRGs from matched pairs. All analyses were performed blind.
RNA extraction for TaqMan and microarray analysis
One or three days after the surgery mice were killed by CO2 inhalation and the lumbar spine was divided longitudinally. Identification of the L5 spinal nerve and DRG was confirmed by dissection proximally into the sciatic nerve. L4 and L5 DRGs were removed into PBS, frozen on dry-ice and stored at -70°C. After DRGs from all groups were collected, total RNA was isolated from the individual ganglia using the PicoPure RNA Isolation Kit following the manufacturer's protocol (Arcturus Bioscience Inc, Mountain View, CA, USA). Average RNA yield was ca. 500 ng/DRG. The high quality of the RNA extracted was verified using the Agilent RNA 6000 Pico Kit (Agilent Technologies, Santa Clara, CA, USA).
cDNA synthesis and TaqMan analysis
cDNA synthesis was performed using the Applera Ltd. (Norwalk, CT, USA) Reverse Transcriptase Kit (N8080234) and RNase Inhibitor (N8080119). cDNA synthesis was performed from 90 ng RNA samples using 2.2 μL 10× RT buffer, 4.84 μL MgCl2 (25 mM), 4.4 μL dNTPs (10 mM), 1.1 μL random hexamers (50 μM), 1.1 μL oligo(dT)16 (50 μM), 0.44 μL RNase Inhibitor (20 U/μL), 0.55 μL Multiscribe (50 U/μL) in a total volume of 22 μL (final cDNA concentration: 4 ng/μl). Samples were incubated at 25°C for 10 minutes and 42°C for 60 minutes. The reaction was stopped by heating to 95°C for 5 minutes. TaqMan reactions were performed using Applera TaqMan Universal PCR Master Mix (4305719) and TaqMan Rodent GAPDH Control Reagents (4308313). Oligonucleotides were synthesized by Operon Biotechnologies GmbH (Köln, Germany). For Scn11a the following sequences were selected: target forward primer: GAGGAATGTGCCGCTGTCA, target reverse primer: CTTTCAGCTTCAGCTTGATCATCTT. The target probe was labeled with FAM: CCATGTGTCTCCGGTAGGCCCTCTG. For Gal the sequences were: target forward primer: CATGCCATTGACAACCACAGA, target reverse primer: TCCTTTCCTCCACCTCCAGTT, target probe labeled with FAM: CCCTCTTGCCTGTGAGGCCATGCT. PCR was performed using the ABI Prism 7900 (Applera) under the following PCR conditions: 2 minutes at 50°C, 10 minutes at 95°C, 40 cycles with 95°C for 15 s and 1 minute at 60°C. PCR was set up as a multiplex PCR using 0.125 μL target probe (50 μM), 0.45 μL target forward primer (50 μM), 0.45 μL target reverse primer (50 μM), 12.5 μL TaqMan 2× PCR Master Mix, 0.25 μL each of primers and probes (TaqMan Rodent GAPDH Control Reagents), 2.5 μL or 1.25 μl cDNA sample in a total reaction volume of 25 μl.
Microarray expression profiling
First-strand cDNA synthesis was performed using 500 ng total RNA with a 100 pM T7-(dT)24 oligomer (GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-dT24) according to Baugh et al. [
34] and SuperScript II reverse transcriptase following the manufacturer's instructions. Double-stranded cDNA was synthesized and then extracted using phenol-chloroform followed by an ethanol precipitation step. An in vitro transcription reaction was performed with the double stranded cDNA sample using the BioArray High Yield RNA Transcription Labeling kit (Enzo Life Sciences, Farmingdale, NY, USA) according to the manufacturer's instructions. Transcription reactions were incubated at 37°C for 16 h. cRNA was purified using the RNeasy Mini kit protocol for RNA cleanup (Qiagen GmbH, Hilden, Germany) and quantified spectrophotometrically. Mice were excluded if < 12 μg of cRNA was obtained. The biotin-labeled cRNA was fragmented using a RNA fragmentation buffer (200 mM Tris-acetate, 500 mM KOAc, 150 mM MgOAc, pH 8.1). Hybridization and staining on mouse MG430_2 GeneChipsTM (Affymetrix Inc., Santa Clara, CA, USA) was performed according to the manufacturer's instructions.
The microarrays were scanned using a GeneChip 3000 Scanner. The scanned data were analyzed using Resolver v5.1 expression data analysis software (Rosetta Biosoftware, Seattle, WA, USA). Before computing correlations, the array data were filtered for significant expression. The criterion for expression for each gene (i.e. for each transcript included on the array) was a p-value < 0.001 for signal intensity vs. noise level in at least 5 L5DRGs, considering all 5 mouse strains together. Only genes that were expressed above this level were included in the correlation analysis. Means are given ± the standard deviation (SD) unless otherwise indicated.
Correlational analysis
In situ hybridization data
For each target gene the proportion of mRNA-positive neuronal profiles in the 4 DRGs from naive mice and the 4–6 DRGs from nerve injured mice was calculated. The Mann-Whitney U test was used to evaluate the significance of changes for each gene by comparing proportions in the naïve and operated mice (Statistica, StatSoft Scandinavia AB, Uppsala, Sweden). In addition, values from the individual naive mice were averaged to yield a baseline expression level for each gene in each strain (mean percent ± SEM). Change in expression of a given transcript for each nerve injured mouse was calculated by subtracting the mouse's individual expression value from the mean of the naïve mice in the same strain. The values for change in expression were then averaged for each strain and compared across strains using one-way ANOVA. Average change in expression values for all replicate pairs in each of the five strain were converted to ranks (1 to 5) where 1 represents the strain with the smallest magnitude of down-regulation in expression and 5 represents the strain with the largest. Finally, strain average values for regulation were plotted against the strain-specific score for pain sensibility to yield Pearson (parametric) and Spearman (rank) correlation coefficients (rP and rS ; n = 5 data points per plot).
Microarray data – Mouse strains based analysis
For each of the five strains L5 nerve injury-induced change in expression for each gene (fold up- or down-regulated) in the L5 and the L4DRGs was calculated. The calculation, log of the ratio of SNL vs. sham expression based on intensity values merged over all animals in each strain, was implemented by the corresponding algorithm in Resolver v5.1. Only transcripts with fold regulation ≥ 1.5 in at least three strains, or ≥ 2 in at least one strain were included for further analysis. Overall, 3927 transcripts met this criterion (3 dpo). For these transcripts, gene expression data for all individuals in each of the five mouse strains was averaged, and the result plotted against strain-specific pain phenotype and r
P and r
S were calculated (n = 5 data points per plot). Plots were also made using raw signal intensities from DRGs of operated mice. Pain phenotype was based on the values given in Table
1.
Microarray data – Individual mouse analysis
This analysis was carried out for the three strains in which we assessed tactile allodynia in individual mice (AKR, B6 and CBA). Rather than averaging over all SNL operated animals in a given strain, fold regulation of gene expression in the ipsilateral L5 and L4 DRGs was calculated for each DRG, in each individual mouse, in each strain. Specifically, we calculated the log of the ratio of expression intensity in individual SNL operated DRGs vs. the average expression intensity for the corresponding DRGs in the sham operated mice of the same strain (Resolver v5.1). Genes with fold regulation ≥ 1.5 in at least 60% of all SNL operated mice in the three strains, or ≥ 2 in at least 20% of these mice, were used for correlational analysis. 3963 transcripts met this criterion. Data on the expression of individual genes in each mouse were plotted against the behavioral results of the individual mice and rp was calculated (n = 26 data points per plot). Plots were also made using raw signal intensities from DRGs of operated mice.
Statistical formalism calls for p-values associated with calculated correlation coefficients r
p and r
s to be corrected for multiple testing. As in microarray expression analyses the very large number of comparisons made undermines the ability to declare any given comparison statistically significant. However, the list of genes with high r
p and r
s is nonetheless expected to be enriched in candidates whose regulation is related to pain phenotype. The evaluation of enrichment can be enhanced by considering false discovery rates (FDRs) [
35]. FDR analysis considers the distribution of r-values and asks whether the number of high values exceeds that expected at random. It then assigns a q-value which indicates the expected number of false discoveries in the set of genes with large r values. The absolute value of r
P was used in the FDR analysis. Computations were done in the R environment using the packages "stats" and "fdrtool" [
36]. A value of q < 0.05 was considered to be significant. Lists of correlated genes were further analyzed according to functional category, both raw categorical distributions and enrichment with respect to the entire mouse genome, using the Gene Ontology Tree Machine (GOTM, [
37]) and Ingenuity Pathway Analysis (IPA, Ingenuity Systems, Redwood City, CA, USA; [
38]).
Competing interests
Some of the authors are employees of the commercial firm Sanofi-Aventis. Sanofi-Aventis has intellectual property rights in some of the results published in this paper, and related results not included in the paper. The authors who are employees do not have a significant direct conflict of interests with regard to this paper. However, they would potentially benefit from the success of the company and in this sense might be considered to have an indirect conflict of interests.
Authors' contributions
Authors MG, SJ, CMW, AMS, HCS, DDP, and JT carried out the surgery, phenotyping hybridization and data analysis associated with the microarray studies. AKP, XJX, ZWH and KF carried out the corresponding tasks associated with the in situ hybridization work. AKP, MG, AD, KF and MD conceived of the project and managed it, carried out statistical, genomic and literature reviews, and drafted the text, which was approved by all authors.